-------
Human

1.0E-05 -|	

Animal Bioassays

ON

rj
si



>}
%•

?5'

si

"3

O

>)
o

53

a
s
Si-

§¦
>}

. o

c>
o

53

Si

a

o

2!

H|

H S.

tn , 1 0E-08

CO	«

o

X	51

^	t

a	s=

£	£ 1.0E-09

1.0E-10

1.0E-11

1.0E-12

CO

o
o

CN
~TO

cd

CO

o
o

CM
"TO
CD

o
o

CNI

CD
TO

CO
O
o
CM

"TO

CD

— — =5

TO

OQ

_TO
<

00

o
o

CM
"TO
CD
N

_o

TO

E

00

-Q
TO

CO

o
o

CN

r^-~
o
o

Cn|

"to
cd
_a>
"a5

CO
r--

cd

TO

CD

CN
O
O
CN

c&

CN
00
CT>

CO
CO
CT>

P"	

CO

O

CD

CD

o



t—

CN

"to

"to

"to



4—'



CD

CD

CD

~

O

jE



Q_

00



TO





o





CD





Q



O
O
CN

CM
O
O
CN

CO
I-—
CT>

00


CO
O
O
CN

O
O
CM

CO
O
O
CN

O
O
O
CN

00
O
O
CN

O
O
CN

TO

-t—j



o
>

TO

a>

TO

CD

a>
i=!

+-i ' m

TO
CD
O
TO

TO

O

CD



h-

r--

h-

o

o

o

CT)

o

o

o

rr~

CN

CN

CN

"to

"to

"to

"to









(D

CD

CD



U

3

—

TO

£=

J=

CD

=5

TO

U

CD



ll





TO
LO

CD
CD

TO

CD

CD
CD

m

TO
>

CO
r---
CD

TO

-t—'

CD
TO
_Q
o
o

o
o
o

CN

"TO

-i—'

CD
CD
O
tl
TO

LO

CD
CT)

CD
O
CD

CO

Figure 4-4. Candidate RfD array.

LO
O
O
CN

"TO
CD
C
O
4=±
O

o

LO

CT>
CD

TO
0)

CD

00

~	UFL (LOAEL-to-NOAEL)
Q UfH (human variability)

~	UFA (animal-to-human)

*	RfD

~	Point of Departure

-------
[This page intentionally left blank.]

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

5. CANCER ASSESSMENT

5.1. QUALITATIVE WEIGHT-OF-EVIDENCE CARCINOGEN CLASSIFICATION
FOR 2,3,7,8-TETRACHLORODIBENZO-jp-DIOXIN (TCDD)

5.1.1. Summary of National Academy of Sciences (NAS) Comments on the Qualitative
Weight-of-Evidence Carcinogen Classification for 2,3,7,8-Tetrachlorodibenzo-
jp-Dioxin (TCDD)

In its charge, the National Academy of Sciences (NAS) was requested to comment
specifically on U.S. Environmental Protection Agency (EPA)'s conclusion that
2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) is best characterized as "carcinogenic to humans."
While indicating that distinction between the categories of "carcinogenic to humans" and "likely
to be carcinogenic to humans" is ".. .based more on semantics than on science..." (NAS, 2006,
198441. p. 141) and recommending that EPA "... spend its energies and resources on more
carefully delineating the assumptions used in quantitative risk estimates for TCDD..." (NAS,
2006, 198441. p. 141) rather than on the qualitative cancer descriptor for TCDD, the NAS
provided the following comments:

.. .the classification of dioxin as "carcinogenic to humans" versus "likely to be
carcinogenic to humans" depends greatly on the definition and interpretation of
the specific criteria used for classification, with the explicit recognition that the
true weight of evidence lies on a continuum with no bright line that easily
distinguishes between these two categories. The committee agreed that, although
the weight of epidemiological evidence that dioxin is a human carcinogen is not
strong, the human data available from occupational cohorts are consistent with a
modest positive association between relatively high body burdens of dioxin and
increased mortality from all cancers. Positive animal studies and mechanistic data
provide additional support for classification of dioxin as a human carcinogen.

However, the committee was split on whether the weight of evidence met all the
necessary criteria described in the cancer guidelines for classification of dioxin as
"carcinogenic to humans." EPA should summarize its rationale for concluding
that dioxin satisfies the criteria set out in the most recent cancer guidelines for
designation as either "carcinogenic to humans" or "likely to be carcinogenic to
humans (NAS, 2006, 198441. p. 140).

If EPA continues to designate dioxin as "carcinogenic to humans," it should
explain whether this conclusion reflects a finding that there is a strong association
between dioxin exposure and human cancer or between dioxin exposure and a key
precursor event of dioxin's mode of action (presumably AhR binding). If EPA's
finding reflects the latter association, EPA should explain why that end point

This document is a draft for review purposes only and does not constitute Agency policy.

5-1 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

(e.g., AhR binding) represents a "key precursor event (NAS, 2006, 198441, p.

141).

5.1.2. EPA's Response to the NAS Comments on the Qualitative Weight-of-Evidence
Carcinogen Classification for TCDD

A cancer descriptor is used to express the conclusion of the weight of evidence regarding
the carcinogenic hazard potential of a compound. EPA agrees with the NAS committee that
cancer descriptors represent points along a continuum of evidence. Relatedly, EPA
acknowledges that there are gradations and borderline situations that cannot be communicated
through a descriptor and are best clarified by a full weight of evidence narrative.

The 2003 Reassessment contains a detailed discussion of TCDD carcinogenicity in both
humans (Part II, Chapter 7a; 8) and animals (Part II, Chapter 6; 8) as well as an overall summary
of TCDD carcinogenicity (Part III, Chapter 2.2.1). Since the release of the 2003 Reassessment,
the database pertaining to TCDD carcinogenicity has been strengthened and expanded by
numerous publications (U.S. EPA, 2008, 5192611 including a new chronic bioassay in female
rats (NTP, 2006, 543749) and several new follow-up epidemiological investigations (see
Section 2.4.1 and references therein). Many of these studies have been published subsequent to
the NAS review. These new data are summarized and evaluated in Section 2.4 of this document.

As noted by the NAS, the 2003 Reassessment was released prior to EPA's publication of
the U.S. EPA Guidelines for Carcinogen Risk Assessment ("2005 Cancer Guidelines"; U.S. EPA,
2005, 086237). Using EPA's guidance at the time of its release (U.S. EPA, 1996, 198087). the
2003 Reassessment determined that the available evidence was sufficient to classify TCDD as a
"human carcinogen." The 1996 guidance suggested "human carcinogen" to be an appropriate
descriptor of carcinogenic potential when there is an absence of conclusive epidemiologic
evidence to clearly establish a cause-and-effect relationship between human exposure and
cancer, but there are compelling carcinogenicity data in animals and mechanistic information in
animals and humans demonstrating similar modes of carcinogenic action.

The 2005 Cancer Guidelines (U.S. EPA, 2005, 086237) are intended to promote greater
use of the increasing scientific understanding of the mechanisms that underlie the carcinogenic
process. The 2005 Cancer Guidelines expand upon earlier guidance applied in the 2003
Reassessment and encourage the use of chemical- and site-specific data versus default options,
the consideration of mode of action information and understanding of biological changes, fuller

This document is a draft for review purposes only and does not constitute Agency policy.

5-2 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

characterization of carcinogenic potential, and consideration of differences in susceptibility. The
2005 Cancer Guidelines also emphasize the importance of weighing all of the available evidence
in reaching conclusions about the human carcinogenic potential of an agent. As noted above,
additional information on TCDD carcinogenicity has been published since the release of the
2003 Reassessment. This information has expanded the TCDD database and provided additional
support for conclusions made in the 2003 Reassessment regarding the carcinogenic potential of
TCDD.

Under the 2005 Cancer Guidelines (U.S. EPA, 2005, 086237). TCDD is characterized as
carcinogenic to humans, based on the available data as of 2009. The 2005 Cancer Guidelines
indicate that this descriptor is appropriate when there is convincing epidemiologic evidence of a
causal association between human exposure and cancer or when all of the following conditions
are met (a) there is strong evidence of an association between human exposure and either cancer
or the key precursor events of the agent's mode of action, but not enough for a causal
association, and (b) there is extensive evidence of carcinogenicity in animals, and (c) the mode(s)
of carcinogenic action and associated key precursor events have been identified in animals, and
(d) there is strong evidence that the key precursor events that precede the cancer response in
animals are anticipated to occur in humans and progress to tumors, based on available biological
information.

As noted above, the NAS commented that EPA should ".. .explain whether this
conclusion reflects a finding that there is a strong association between dioxin exposure and
human cancer or between dioxin exposure and a key precursor event of dioxin's mode of action
(presumably AhR binding)" (NAS, 2006, 198441). When evaluating the carcinogenic potential
of a compound, EPA employs a weight of evidence approach in which all available information
is evaluated and considered in reaching a conclusion. The following sections provide a summary
of EPA's weight of evidence evaluation for TCDD.

5.1.2.1. Summary Evaluation of Epidemiologic Evidence of TCDD and Cancer

The available occupational epidemiologic studies provide convincing evidence of an
association between TCDD exposure and all cancer mortality. Among the strongest of these are
the studies of over 5,000 U.S. chemical manufacturing workers (the National Institute for
Occupational Safety and Health [NIOSH] cohort) (Ay 1 ward et al., 1997, 594365; Cheng et al.,

This document is a draft for review purposes only and does not constitute Agency policy.

5-3 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

2006, 523122; Collins et al., 2009, 197627; Fingerhut et al., 1991, 197301; Steenland et al.,
1999, 197437; Steenland et al., 2001, 198589); a study of nearly 2,500 German workers involved
in the production of phenoxy herbicides and chlorophenols (the Hamburg cohort) (Becher et al.,

1996,	197121; Becher et al., 1998, 197173; Flesch-Janys et al., 1995, 197261; Flesch-Janys et
al., 1998, 197339; Manz et al., 1991, 199061; Nagel et al., 1994, 594369); a study of more than
2,000 Dutch workers in two plants involved in the synthesis and formulation of phenoxy
herbicides and chlorophenols (the Dutch cohort) (Bueno et al., 1993, 196993; Hooiveld et al.,
1998, 197829); a smaller study of roughly 250 workers involved in a chemical accident cleanup
(the BASF cohort) ed in a chemical accident cleanup (the BASF cohort) (Ott and Zober, 1996,
198101; Thiess et al., 1982, 064999; Zober et al., 1990, 197604); and an international study of
more than 18,000 workers exposed to phenoxy herbicides and chlorophenols (Kogevinas et al.,

1997,	198598; Saracci et al., 1991, 199190) including newer studies of smaller subsets of these
workers (McBride, 2009, 198490; McBride et al., 2009, 197296; t' Mannetje et al., 2005,

197593). The findings from these studies have been thoroughly described either in the 2003
Reassessment or in Section 2.4.1 of this document.

As noted in Section 2.4, there are considerable challenges inherent in addressing potential
sources of confounding from smoking and co-exposure to other carcinogens, (which could
produce inflated or spurious associations), the healthy worker effect, (which could result in
attenuated effects through comparison with a referent background with an inappropriately high
background risk), and quantifying exposure to the populations included in many of these
retrospective studies. The more recent studies of these cohorts have made significant advances
in reducing the potential for bias from the healthy worker effect through use of internal cohort
analyses and/or controlling for potential confounders through statistical adjustment, restriction,
and use of internal comparisons. Although some exposure assessment uncertainties remain,
some of these studies have also collected individual-level TCDD exposure estimates that allow
quantification of effective dose necessary for dose-response modeling. Overall, the occupational
data provide consistent support for an association between exposure to TCDD and increased
cancer mortality.

Additional epidemiologic evidence supporting an association between TCDD exposure
and cancer comes from studies investigating the morbidity and mortality of residents exposed to
TCDD following an accidental release from a chemical plant near Seveso, Italy (the Seveso

This document is a draft for review purposes only and does not constitute Agency policy.

5-4 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

cohort) (Bertazzi et al., 1989, 197013; Bertazzi et al., 1993, 192445; Bertazzi et al., 1997,

197097; Bertazzi et al., 2001, 197005; Consonni et al., 2008, 524825; Pesatori et al., 1998,
523076; Pesatori et al., 2003, 197001; Warner et al., 2002, 1974891 Pesatori et al. (2003,
197001) and Consonni et al. (2008, 524825) were not available at the time the 2003
Reassessment was released. Among individuals with relatively high exposure at Seveso
(Zones A and B combined), all-cancer mortality in the 20-year post-accident period and all-
cancer incidence in the 15-year post-accident period failed to exhibit significant departures from
the expected 197001). However, an increased risk of all-cancer mortality was noted among men
15-20 years after first exposure; not only is the association similar in magnitude to other studies
(relative risk [RR] = 1.3; 95% confidence interval [CI] = 1.0-1.7) but also emphasizes the
importance of consideration of latency (Bertazzi et al., 2001, 197005). Furthermore, associations
between TCDD and some specific cancer sites were detected in this cohort, including increased
incidence (based on 15 years of follow-up) and mortality (based on 20 years follow-up) from
lymphatic and hematopoietic neoplasms in both males and females from Zones A and B
(Consonni et al., 2008, 524825). This excess was primarily due to non-Hodgkin's lymphoma.
Additionally, there was an increase in lung and rectal cancer mortality in men (Bertazzi et al.,

2001,	197005) and limited evidence of increased liver cancer incidence in women based on the

15-year follow-up study (Bertazzi et al., 1993, 192445). In a separate analysis of 981 women in
Zone A, breast cancer incidence (n = 15) was associated (a 2-fold increase for a 10-fold increase
in serum TCDD) with TCDD measurements first collected in 1976 and 1977 (Warner et al.,

2002,	197489). The authors also reported a 2-3-fold increase in all cancer incidence (n = 21) for
the two upper quartiles of TCDD exposure.

Overall, the newer studies of the Seveso cohort have reported significant increases in
cancer incidence and elevations in cancer mortality that were not evident in earlier studies of this
cohort. While these studies demonstrate an association between TCDD exposure and different
types of cancer, one of the main limitations is the small number of cancer cases to assess
site-specific associations with TCDD exposure. Ongoing studies in that cohort should help
further elucidate potential risk for specific cancer types (and other endpoints) associated with
TCDD exposures among this population.

This document is a draft for review purposes only and does not constitute Agency policy.

5-5 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

5.1.2.1.1. Evidence for causality.

The evidence for causality for cancer from the human studies is briefly summarized in the
paragraphs that follow and is based on recommendations from the 2005 Cancer Guidelines. It
should be noted that there are methodological limitations of the epidemiologic studies that may
temper some of the conclusions regarding causality. These limitations include limited statistical
power, exposure assessment uncertainty, and lack of control of confounders (e.g., dioxin-like
compounds and smoking) in some studies. There also is additional uncertainty in the evidence
for causality due to the lack of organ specificity in TCDD associated cancers, as the most
consistent results occurred for all-cancer mortality; however, this would be consistent with a
hypothesized carcinogenic mode of action of TCDD as a promoter. Despite these uncertainties,
many of the more recent studies have greatly improved exposure assessments compared to
earlier studies of the same cohorts and have addressed the potential for confounding and other
types of biases.

Temporality—exposure must precede the effect for causal inference. Given the long
induction period for many types of cancers, exposure should precede the effect with a sufficient
latency (i.e., typically 15-20 years for environmental carcinogens). In all the occupational
studies reviewed (with the exception of (McBride, 2009, 198490)1 TCDD exposure has
preceded the effect with sufficient latency to be considered causally associated. In the studies of
the Seveso cohort, the follow-up exposure period has now reached 20 years, a latency sufficient
to address some carcinogenic endpoints. Since most of the studies are based on occupational
exposures or accidental releases into the environment, temporality is more readily established
due to the obvious determination of the specific exposure windows prior to disease onset.

Strength of Association—refers to the magnitude of measures of association such as the
ratio of incidence or mortality (e.g., standardized mortality ratio [SMRs], standardized incidence
ratios, RRs, or odds ratios) in addition to statistical significance considerations. Effect estimates
that are large in magnitude are less likely to be due to chance, bias, or confounding. Reports of
modest risk, however, do not preclude a causal association and may reflect an agent of lower
potency, lower levels of exposure or attenuation due to nondifferential exposure
misclassification. The four occupational cohorts with the highest exposures (NIOSH, Hamburg,
Dutch, and BASF) consistently showed statistically significant, although moderate, elevations in
cancer mortality. When the data were combined, the SMR for all four subcohorts was 1.4

This document is a draft for review purposes only and does not constitute Agency policy.

5-6 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

[95% CI = 1.2-1.6] (IARC, 1997, 537123). Based on findings from the International Agency for
Research on Cancer (IARC) Working Group, increases in all cancer (combined) mortality of the
magnitude reported for TCDD have rarely been found in occupational cohort studies (IARC,
1997, 537123). Although these estimates are higher than the all-cancer mortality results among
Seveso men(RR= 1.1; 95% CI = 1.0-1.3), they are comparable to the risk estimated in this
population (RR = 1.3; 95% CI = 1.0-1.7) 15-20 years after first exposure.31 These consistent
results comparable in magnitude from the occupational cohorts and Seveso population are not
likely due to chance.

The occupational cohort studies also show an increased risk for lung cancer in the
previously mentioned four subcohorts. The relative risk for lung cancer in the combined highly
exposed subcohorts was estimated to be 1.4 (95% CI = 1.1-1.7) (IARC, 1997). This is
consistent with the lung cancer mortality findings for the highest exposed group of men in
Seveso (RR = 1.3; 95% CI = 1.0-1.7). Additionally, there was an increase in rectal cancer
mortality in the Seveso cohort (RR = 2.4; 95% CI = 1.2-4.6) (Bertazzi et al., 2001, 197005) with
a corresponding increase in incidence. Consistent relative risks of more than two were also
detected for rectal cancer in the Hamburg and New Zealand cohorts, but increased risks were not
found in the other cohorts. Although there was limited evidence of increased incidence or
morality from hepatobiliary cancers across the cohorts, liver cancer incidence was elevated in the
15-year post accident period among women in the Seveso cohort (RR = 2.4; 95% CI = 1.1-5.1,
(Warner et al., 2002, 197489)). An association in this population was also detected for between
breast cancer incidence (RR = 2.1; 95% CI = 1.0-4.6) and serum TCDD levels (per a 10-fold
increase in serum TCDD). Although findings were based on small numbers, three- and four-fold
increased risks of soft tissue sarcoma were detected among the NIOSH (Collins et al., 2009,
197627) and New Zealand cohorts (McBride, 2009, 198490). No other cases of this very rare
cancer were detected in the exposed populations from the other cohorts.

31In addition to consideration of statistical significance to address the possibility of random variability (i.e., chance),
many other factors are important to consider when assessing causality using a weight of evidence determination. As
noted in the EPA's Cancer Guidelines, a number of factors besides statistical significance are relevant for assessing
evidence of adverse health effects based on human data. These include strength of association, temporality,
biological gradient (i.e., dose-response concordance), biological plausibility, etc.). In analyzing the body of
information in the literature, the consistency of the magnitude of reported risk estimates (across different studies) is
considered when addressing causality; rather than relying solely on statistical significance.

This document is a draft for review purposes only and does not constitute Agency policy.

5-7 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

Elevated risk of lymphohemopoietic cancer mortality was noted among the Seveso cohort
(RR = 1.7; 95% CI = 1.2, 2.5) (Consonni et al., 2008, 524825). Increased SMRs for
lymphohemopoietic cancer comparable in magnitude (range: 1.6-2.2) were also detected among
the Hamburg and New Zealand occupational cohorts, but limited evidence (range: 1.0 to 1.2) of
increased mortality was found in the BASF, NIOSH and Ranch Hands employees (Akhtar et al.,
2004, 197141: Ott and Zober, 1996, 198101: Steenland et al., 1999, J97437). Most of the
lymphohemopoietic cancer mortality risk was reportedly due to non-Hodgkin's lymphoma in
most of the cohorts. Relative risks for non-Hodgkin's lymphoma among TCDD exposed
populations from the NIOSH, Hamburg, New Zealand, Dutch, and Seveso cohorts ranged from
1.2 to 3.8. Although statistical power was limited in most of these studies, relative risks
exceeded 3.0 for non-Hodgkin's lymphoma in three of these cohorts (Consonni et al., 2008,
524825: Flesch-Janys et al., 1998, 197339: Hooiveld et al., 1998, 197829V

Consistency—the observation of the same site-specific effect across several independent
study populations strengthens an inference of causality. Despite differences across occupational
cohorts, most studies have consistently reported increases in all-cancer mortality with TCDD
exposure. Several of these studies have also reported increases in lung cancer related to TCDD
exposure. As noted above, there is also suggestive evidence of an increased risk in all-cancer
and lung cancer mortality among the Seveso cohort consistent in magnitude to the occupational
cohorts. Elevated risk of lymphohemopoietic cancer mortality consistent in magnitude
(range: 1.6-2.2) was also detected among the Seveso, Hamburg and New Zealand cohorts. An
increased risk for non-Hodgkin's lymphoma was found in two of the occupational cohorts as
well as in the Seveso cohort, although the relative risks largely did not achieve statistical
significance. Among those studies detecting an association, consistent two-fold relative risks
were found for rectal cancer (Bertazzi et al., 2001, 197005: Flesch-Janys et al., 1998, 197339:
McBride, 2009, 198490) and relative risks in excess of three were detected for soft tissue
sarcoma (Collins et al., 2009, 197627: McBride, 2009, 198490).

Biological Gradient—refers to the presence of a dose-response and/or duration-response
between a health outcome and exposure of interest. Several of the occupational cohort studies
(Flesch-Janys et al., 1998, 197339: Manz et al., 1991, 199061: Michalek and Pavuk, 2008,
199573: Ott and Zober, 1996, 198101: Steenland et al., 1999, 197437) found evidence of a
dose-response relationship for all cancers and various TCDD exposure measures. The SMR

This document is a draft for review purposes only and does not constitute Agency policy.

5-8 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

analyses based on internal comparisons within the occupational cohorts show a biological
gradient by comparing highly TCDD exposed workers to low or unexposed workers. A
biological gradient was also demonstrated in the Seveso cohort by comparing highly exposed
individuals (Zones A and B) to individuals in lower exposure zones (Zones C and R). Warner et
al. (2002, 197489) also reported evidence of a dose-response trend for breast cancer and
increasing TCDD exposures.

Biological Plausibility—refers to the observed effect having some biological link to the
exposure. Most evidence suggests that toxic effects of TCDD are mediated by interaction with
the aryl hydrocarbon receptor (AhR). AhR is a highly conserved protein among mammals,
including humans (Fujii-Kuriyama et al., 1995, 543727; Harper et al., 2002, 198124; Nebert et
al., 1991, 5437281 Several hypothesized modes of action have been presented for TCDD-
induced tumors in rodents, all involving AhR activation. The available evidence does not
preclude the relevance of these hypothesized modes of action to humans.

Specificity—as originally intended, refers to increased inference of causation if a single
site effect, as opposed to multiple effects, is observed and associated with exposure. Based on
current biological understanding, this is now considered one of the weaker guidelines for
causality. As stated in the 2005 Cancer Guidelines, given the current understanding that many
agents cause cancer at multiple sites, and cancers have multiple causes, the absence of specificity
does not detract from evidence for a causal effect. Given that the most consistent findings
associating TCDD and cancer are for all-cause cancer mortality, epidemiological evidence
suggests that TCDD lacks specificity for particular tumor sites. A key event in TCDD's mode of
action is binding to and activating AhR; however, downstream events leading to tumor formation
are uncertain and may likely be tissue specific. Given that the AhR is highly conserved among
species and is expressed in various human tissues, the lack of tumor site specificity does not
preclude a determination of causality.

In summary, EPA finds the available epidemiological information provides strong
evidence of an association between TCDD exposure and human cancer that cannot be reasonably
attributed to chance or confounding and other types of bias, and with a demonstration of
temporality, strength of association, consistency, biological plausibility, and a biological
gradient. Additional evidence from animal studies and from mechanistic studies (described
below) provides additional support for the classification of TCDD as carcinogenic to humans.

This document is a draft for review purposes only and does not constitute Agency policy.

5-9 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

5.1.2.2.	Summary of Evidence for TCDD Carcinogenicity in Experimental Animals

An extensive database on the carcinogenicity of TCDD in experimental animals is
described in detail in Part II, Chapter 6 of the 2003 Reassessment. There is substantial evidence
that TCDD is carcinogenic in experimental animals based on long-term bioassays conducted in
both sexes of rats and mice (Kociba et al„ 1978, 001818; NTP, 1982, 594255; NTP, 2006,
543749) and in male hamsters (Rao et al., 1988, 199032). Additionally, National Toxicology
Program (NTP, 2006, 543749) has completed a new chronic bioassay in female Sprague Davvley
rats. These studies are summarized in Section 2.4.2 of this document. All studies have produced
positive results, with TCDD increasing the incidence of tumors at sites distant from the site of
treatment and at doses well below the maximum tolerated dose. In both sexes of rodents, when
administered by different routes and at low doses, TCDD caused tumors at multiple sites; tumors
were observed in liver, lung, lymphatic system, soft tissue, nasal turbinates, hard palate, thyroid,
adrenal, pancreas, and tongue. The most consistent and best characterized carcinogenic
responses to TCDD are in the rodent liver, lung, and thyroid (discussed below in
Section 5.1.2.3).

5.1.2.3.	TCDD Mode of Action

The 2005 Cancer Guidelines defines the term "mode of action" as "a sequence of key
events and processes, starting with interaction of an agent with a cell, proceeding through
operational and anatomical changes, and resulting in cancer formation." A "key event" is an
empirically observable precursor step that is itself a necessary element of the mode of action or is
a biologically based marker for such an element. Mode of action is contrasted with "mechanism
of action," which implies a more detailed understanding and description of events, often at the
molecular level. In the case of TCDD, the terms 'mechanism of action' and 'mode of action' are
often used interchangeably in the scientific literature in reference to TCDD's interaction with the
AhR. A thorough discussion of TCDD's interaction with the AhR can be found in the 2003
Reassessment (Part II, Chapter 2; Part III, Chapter 3), and is summarized below (see
Section 5.1.2.3.1).

Most evidence suggests that the majority of toxic effects of TCDD are mediated by
interaction with the AhR. EPA considers interaction with the AhR to be a necessary, but not
sufficient, event in TCDD carcinogenesis. The sequence of key events following binding of

This document is a draft for review purposes only and does not constitute Agency policy.

5-10 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

TCDD to the AhR and that ultimately leads to the development of cancer is unknown.

Therefore, in the strictest sense, TCDD's interaction with the AhR does not constitute a "mode
of action" as defined by the 2005 Cancer Guidelines because information about the progression
of necessary events is lacking. However, AhR binding and activation by TCDD is considered to
be a key event in TCDD carcinogenesis.

5.1.2.3.1. The aryl hydrocarbon receptor (AhR).

While substantial evidence suggests that most toxic effects of TCDD are mediated by
interaction with the AhR, less is known about the complex responses that result in tumor
formation. Nonetheless, a picture is emerging wherein TCDD is considered a
"receptor-mediated carcinogen" in laboratory animals (see Figure 5-1), acting in a manner
similar to peroxisome proliterators, phorbol esters, or estrogen (Woods et al., 2007, 543735).

TCDD activates the AhR, a member of the basic helix-loop-helix, Per-Arnt-Sim
(bHLH-PAS) family of transcription factors. AhR is present in most cell types and in the
inactivated state is cytosolic and exists in a complex with chaperone proteins, such as heat shock
protein 90 (Hsp90). Binding of TCDD to AhR leads to nuclear translocation and
heterodimerization with its partner protein Arnt, another bHLH-PAS family member. The
AhR: Arnt heterodimer binds to specific cognate DNA sequence elements known as
dioxin/xenobiotic response elements (DRE/XRE) present in the regulatory region of specific
genes. Binding of the AhR: Arnt heterodimer to these elements, and subsequent recruitment of
tissue specific transcriptional coactivator complexes, leads to increased transcription of specific
genes, known as "target genes." There is a battery of genes affected in this manner and targets
include certain xenobiotic-metabolizing enzymes, such as cytochrome P450 (CYP)l Al,

CYP1A2, CYP2B1, and UDP-glucuronosyltransferase (UGT)1A6 (reviewed in Schwartz and
Appel, 2005, 543737). In addition, genes affected by the TC D D/Ah R-com pi ex code for both
inhibitory and stimulatory growth factors; their gene products affect cellular growth,
differentiation and homeostasis and have been shown to contribute to carcinogenicity as well as
other forms of toxicity (reviewed in Popp et al., 2006, 197074).

Detailed molecular biology research has been performed to identify the extent of the
genes regulated by AhR (Woods et al., 2007, 543735): however a complex and still ill-defined
profile remains. The basic physiology of AhR signaling is still poorly understood, despite being

This document is a draft for review purposes only and does not constitute Agency policy.

5-11 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

highly conserved among vertebrate species (reviewed in Hahn, 2002, 099302). In fact, it is now
known that the AhR recognizes a large number of chemical structures, including nonaromatic
and nonhalogenated compounds (Denison and Nagy, 2003, 197226). which supports the
biological role of the AhR as a receptor that helps regulate the expression of genes necessary for
biotransformation of environmental chemicals (i.e., CYP1A1). However, the endogenous
physiological role of AhR is complicated, as evidenced by the numerous studies examining AhR
null (ArH -/-) mice, which demonstrate alterations in the liver, immue system, ovary, heart and
other organs (reviewed in Hahn, 2009, 477460). The endogenous function of AhR remains
unknown.

Given that the AhR is expressed in most tissues (Dolwick et al., 1993, 543762) with
tissue-specificity in terms of level of expression and the profile of target genes, there is
substantial complexity and difficulty associating TCDD-mediated transcription of specific target
genes and tissue-specific toxic responses, including cancer. It is important to note that the extent
of the response of individual TCDD target genes does not correlate with site-specific
tumorigenicity. For example, while TCDD is ineffective as a tumor promoter in ovariectomized
rats and does not stimulate liver cell proliferation in these animals, it is still capable of inducing
CYP1A2 in roughly the same magnitude as in the intact female rats (Lucier, 1991, 198691).
Similarly, CYP1 Al induction by TCDD is very similar in male and female rats even though
males are almost completely resistant to TCDD carcinogenicity (Wyde et al., 2002, 197009).

Some of AhR's effects on gene expression may be the result of interaction with other
transcription factors (such as the retinoblastoma protein(Ge and El fieri nk, 1998, 197702). NF-kB
(Tian et al., 1999, 198378) or with the tyrosine kinase c-Src (Blankenship and Matsumura, 1997,
543751) rather than via direct interaction with DNA. By far the most extensive studies involving
cross-talk between AhR and another transcription factor are those involving the estrogen receptor
alpha (ERa). The anti-estrogenic properties of TCDD have been well-documented, beginning
with the observations that TCDD repressed estradiol function in rat uterus and liver. The
AhR-ERa cross-talk can be manifested at several levels including direct protein interaction,
association of the receptors with the other's response element and altered metabolism of estradiol
by AhR ligand (Takemoto et al., 2004, 543753). The interactions between AhR/Arnt- and
estrogen receptor-dependent signaling pathways, which mediate anti-estrogenic effects of
dioxins and dioxin-like polychlorinated biphenyls (PCBs; Bock, 1994, 543755). is probably

This document is a draft for review purposes only and does not constitute Agency policy.

5-12 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

causal for the well-documented gender-specificity of the carcinogenic effects of these agents
(e.g., hepatocarcinogenicity of TCCD in female as opposed to male rats) (Lucier, 1991, 198691).
In addition, cross-talk between AhR/Arnt and other nuclear receptors, their coactivators, and
corepressors, has been described. In fact, cross-talk has been reported for AhR and numerous
signaling pathways involved in a broad range of physiological processes. The molecular
mechanisms by which the AhR interferes with these signaling networks are multifaceted and
occur at multiple levels of regulation (many beyond transcriptional control)
(Haarmann-Stemmann et al., 2009, 197874). It remains unknown how any of these molecular
pathways involving AhR signaling are linked to TCDD-mediated carcinogenesis.

Pertinent to human risk assessment, there are wide inter- and intraspecies differences in
the toxicological responses to TCDD (Etna et al., 1994, 197313; Poland and Glover, 1990,
543759; Poland et al., 1994, 198439) some of which can be explained by polymorphisms in
AhR. For instance, there is a 10-fold difference in susceptibility to TCDD-induced toxicity
between the TCDD-sensitive C57BL/6 and the TCDD-resistant DBA/2 strains of mice (Poland
and Glover, 1980, 543761) that can be explained by polymorphic variations in the ligand-binding
domain and in the C-terminal region of the AhR molecule of each strain (Dolwick et al., 1993,
543762). Depending on the system examined, the estimated affinity of binding of TCDD (and
related compounds) to the human AhR is about 10-fold lower than that observed to the AhR
from "responsive" rodent species and is comparable to that observed to the AhR from
"nonresponsive" mouse strains (Ramadoss and Perdevv, 2004, 198824). This reduced affinity is
due, in part, to a single amino acid substitution within the ligand binding domain of the human
and "nonresponsive" mouse AhRs (Ramadoss and Perdevv, 2004, 198824). Although the affinity
of binding of TCDD and related compounds to the human AhR is reduced compared with rodent
AhRs, the qualitative and quantitative rank-order potency of these chemicals is similar. The
considerable tissue and species variability in response to TCDD cannot be ascribed solely to
polymorphisms of the AhR gene (Geyer et al., 1997, 543768; Pohjanvirta and Tuomisto, 1994,
543767). further complicating this key event in TCDD-mediated carcinogenesis.

5.1.2.3.1.1. Other AhR considerations.

In addition to the potent agonist TCDD, there are many other exogenous ligands for the
AhR, including certain polycyclic aromatic hydrocarbons, polychlorinated dibenzofurans, and

This document is a draft for review purposes only and does not constitute Agency policy.

5-13 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

PCBs (Bock, 1994, 543755). Several natural and endogenous compounds are also regulators of
AhR (Chiaro et al., 2008, 543771). The classes of endogenous compounds that have been shown
to induce CYP1 and/or activate AhR include: (a) tryptophan metabolites, other indole-containing
molecules, and phenylethylamines (Gielen and Nebert, 1971, 543775); (b) tetrapyrroles such as
bilirubin and biliverdin; (c) sterols such as 7-ketocholesterol and the horse steroid equilenin;
(d) fatty acid metabolites, including at least six different prostaglandins (Seidel et al., 2001,
543776) and lipoxin A4; and (e) the ubiquitous second messenger cAMP (reviewed in McMillan
and Bradfield (2007, 543777) and Barouki et al. (2007, 543778)). Several of these endogenous
and exogeous compounds, including bilirubin, biliverdin, and P-naphthoflavone, that also bind to
the AhR are not carcinogenic in rodent models, therefore, some other key precursor event(s)
need to be identified. Further, the existence of multiple ligands with varying affinity and
responses suggests that "selective receptor modulators" (or SRMs) of the AhR exist. SRMs are
ligands for a receptor that, upon binding, elicit a conformational change in the receptor that
results in differential recruitment of coregulatory molecules to the target gene promoter region,
thereby imparting a different biological activity relative to the prototypical ligand. This
phenomenon has been most studied for nuclear receptors such as the ERa with the classic
example being tamoxifen, which has estrogen-like activity in the uterus but anti-estrogen-like
effects in the breast. Thus, the relative abilities of compounds to stimulate gene expression or
other effects vary in promoter- and cell type-specific manners. It is now apparent that SRMs
exist for the AhR as well (SAhRMs, Fretland et al., 2004, 197357). For example,
6-methyl-l,3,8-trichlorodibenzofuran (6-MCDF), a SAhRM whose structure is similar to that of
TCDD, can induce CYP1 Al gene expression in liver but does not lead to the toxic responses
associated with TCDD (Fritz et al., 2009, 594372). The existence of SAhRMs further
complicates the role of TCDD binding to AhR as a key event in TCDD-mediated
carcinogenicity, and suggests that additional information is necessary to elucidate the
carcinogenic mode of action of TCDD.

TCDD may have dose-dependent modes of action. It has been demonstrated that
AhR-deficient (AhR-/-) mice show no signs of toxicity at doses of TCDD approximating the
lethal dose eliciting 50% response (LD50) dose (200 (J,g/kg) in AhR +/+ mice (Fernandez -
Salguero et al., 1996, 197650). However, a single high exposure of 2,000 (.ig/kg to
AhR-deficient mice produced several minor lesions including scattered necrosis and vasculitis in

This document is a draft for review purposes only and does not constitute Agency policy.

5-14 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

the liver and lungs. These data suggest that a pathway leading to toxicity exists, albeit at very
high doses, that is independent of the AhR. However, these data also indicate that, at least in
mice, the major in vivo effects of TCDD are mediated through the AhR. The finding of
carcinogenicity in hamsters (Rao et al., 1988, 199032) is of special interest since hamsters have
been found to be relatively resistant to the lethal effects of TCDD (Henck et al., 1981, 543779;
Olson et al., 1980, 197976). To date, there have been no chronic bioassay studies of TCDD
carcinogenicity in AhR-deficient transgenic animals.

There are additional insights into the complexity of TCDD's mechanism of action
involving AhR. Some biochemical responses to TCDD treatment in isolated cells have been
reported in cells lacking Arnt, in cells expressing a mutated Arnt protein and in cells with highly
reduced levels of AhR (Kolluri et al., 1999, 548721; Puga et al., 1992, 543784). implying either
a non nuclear role of the AhR in mediating these events or an AhR-independent process.

Additionally, recent studies have linked AhR activation in the absence of exogenous
ligand to a multitude of biological effects, ranging from control of mammary tumorigenesis to
regulation of autoimmunity (Hahn et al., 2009, 548725). Finally, constitutively activated AhR in
rodents has been shown to induce stomach tumors (Andersson et al., 2002, 197101). This
indicates that AhR activation alone (i.e., in the absence of ligand) is sufficient to induce tumors.

5.1.2.3.2. TCDD as a tumor promoter.

The role of TCDD as a tumor promoter is discussed in the 2003 Reassessment (Part II,
Chapter 6). The following is a brief summary of the information regarding TCDD as a tumor
promoter.

Numerous studies have examined the tumor promoting potential of TCDD. Using the
traditional two-stage initiation-promotion study design in the liver, studies have demonstrated
that TCDD is a dose- and duration-dependent liver tumor promoter (Dragan and Schrenk, 2000,
197243; Maronpot et al., 1993, 198386; Pitot et al., 1980, 197885; Teeguarden et al., 1999,
198274; Walker et al., 2000, 198733)(Walker et al., 1998). TCDD has also tested positive for
tumor promoting ability in the two-stage models of mouse skin tumorigenesis (Dragan and
Schrenk, 2000, 197243; I ARC, 1997, 537123). and in the lung (Anderson et al., 1991, 201761;
Beebe et al., 1995, 548754). Overall, the data demonstrate that TCDD is a tumor promoter and
potentially harbors only weak initiating activity.

This document is a draft for review purposes only and does not constitute Agency policy.

5-15 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

TCDD is typically designated as a nongenotoxic and nonmutagenic carcinogen because it
does not damage DNA directly through the formation of DNA adducts, is negative in most
short-term assays for genotoxicity, and is a potent tumor promoter and a weak initiator or
noninitiator in multistage models for chemical carcinogenesis (Clark et al., 1991, 594378;
Flodstrom and Ahlborg, 1991, 548728; Graham et al., 1988, 594375; Lucier, 1991, 198691; Pi tot
et al., 1980, 197885; Poland et al., 1982, 199756). However, mechanisms have been proposed
that support the possibility that TCDD might be indirectly genotoxic, either through the
induction of oxidative stress or by altering the DNA-damaging potential of exogenous and
endogenous compounds, such as estrogens. In addition, there have been numerous reports
demonstrating TCDD-induced modifications of growth factor signaling pathways and cytokines
in experimental animals and cell culture systems. Some of the altered signaling pathways
include those for epidermal growth factor, transforming growth factor alpha, glucocorticoids,
estrogen, tumor necrosis factor-alpha, interleukin 1-beta, plasminogen inactivating factor-2, and
gastrin. Many of these pathways are involved in cell homeostasis, proliferation, and
differentiation and provide plausible mechanisms responsible for the carcinogenic actions of
TCDD. Unfortunately, information on the etiology of the different tumor types is lacking to
equivocally link tumor promotion or indirect genotoxic action of TCDD to a specific mechanism
or mode of TCDD carcinogenesis.

5.1.2.3.3. Hypothesized modes of action of TCDD in rodents.

TCDD has been shown to consistently induce multiple tumors in both sexes in several
rodent species. These tumors are observed in various tissues, including (but not limited to):
liver, lung, thyroid, lymphatic system, soft tissue, nasal turbinates, hard palate, adrenal, pancreas,
and tongue. While the mode of action of TCDD in producing cancer has not been elucidated for
any tumor type, the best characterized carcinogenic actions of TCDD are in rodent liver, lung,
and thyroid. The hypothesized mode of action for each of these three tumor types is briefly
discussed below and is described in Figure 5-2. The hypothesized sequence of events following
TCDD interaction with the AhR is markedly different for each of these three tumor types. No
detailed hypothesized mode of action information exists for any of the other reported tumor
types. Further, no single definitive mode of action of TCDD-mediated carcinogenicity has been
identified.

This document is a draft for review purposes only and does not constitute Agency policy.

5-16 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

5.1.2.3.3.1.	Liver tumors.

The mode of action of TCDD in producing liver cancer in rodents has not been
elucidated. One hypothesized mode of carcinogenic action of TCDD in the liver is mediated
through hepatotoxicity. Generically speaking, TCDD activation of the AhR leads to a variety of
changes in gene expression, which then lead to hepatotoxicity, followed by compensatory
regenerative cellular proliferation and subsequent tumor development (see Figure 5-2). The
details of the mechanism of TCDD-induced hepatotoxicity have not been fully determined but
both CYP induction and oxidative stress have been postulated to be involved (Maronpot et al.,
1993, 198386; Viluksela et al., 2000, 198968). The enhanced cell proliferation arising from
either altered gene expression or hepatotoxicity, or both, may lead to the promotion of
hepatocellular tumors (Whysner and Williams, 1996, 197556). The sensitivity of female rat liver
to TCDD, which apparently does not extend to the mouse, depends on ovarian hormones (Lucier,
1991, 198691; Wyde et al., 2001, 198575). This sensitivity has been ascribed to induction of
estradiol metabolizing enzymes (Graham et al., 1988, 594375) and is hypothesized to lead either
to generation of reactive metabolites of endogenous estrogen or to active oxygen species of
estrogens. Oxidative DNA damage has been implicated in liver tumor promotion (Umemura et
al., 1999. 198001).

A dose-response relationship exists for TCDD-mediated hepatotoxicity, and this parallels
the dose-response relationship for tumor formation (or formation of foci of cellular alteration as a
surrogate of tumor formation). However, the dose-response relationship for other
TCCD-induced responses such as enhanced gene expression is different from the dose-response
for tumor formation in terms of both efficacy and potency (see Popp et al. (2006, 197074) for
review). It is important to note that differences in potency between events (i.e., gene expression
versus cell proliferation) does not necessary imply alternative mechanisms of action.

5.1.2.3.3.2.	Luns tumors.

The mode of action of TCDD in producing lung cancer in rodents (predominantly
keratinizing squamous cell carcinoma, (Larsen, 2006, 548744)) has not been elucidated. One
hypothesized mechanism of the carcinogenic action of TCDD in the lung involves disruption of
retinoid homeostasis in the liver (see Figure 5-2). Retinoic acids and their corresponding nuclear
receptors, the retinoic acid receptors (RARs) and the retinoid X receptors (RXRs), work together

This document is a draft for review purposes only and does not constitute Agency policy.

5-17 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

to regulate cell growth, differentiation, and apoptosis. It is hypothesized that TCDD, through
activation of the AhR, can affect parts of the complex retinoid system and/or other signaling
systems regulated by, and/or cross-talking with, the retinoid system (reviewed in (Nilsson and
Hakansson, 2002, 548746)). These effects are then hypothesized to lead to lung tumor
development; however, the mechanisms underlying this hypothesis are not well-defined.
Pulmonary squamous proliferative lesions have been reported following oral exposure to TCDD
in rats (Tritscher et al., 2000, 197265). In general, squamous metaplasia with some
inflammation is associated with significant forms of injury via inhalation of toxic compounds but
is also seen with vitamin A deficiency (Tritscher et al., 2000, 197265) and gives some credence
to this hypothesis.

Another hypothesized mechanism for the carcinogenic action of TCDD in the lung is
through induction of metabolic enzymes. Through activation of AhR and subsequent induction
of metabolizing enzymes (such as CYP1 Al), TCDD may enhance bioactivation of other
carcinogens in lung (Tritscher et al., 2000, 197265). There have been few studies to support this
hypothesis; however, in a long-term continuous-application study of carcinogenesis using
airborne particulate extract (APE), squamous cell carcinoma occurred in 8 of 17 AhR+/+ mice
(47%) while no tumors were found in AhR-/- mice (Matsumoto et al., 2007, 548748). In
addition CYP1 Al was induced in AhR+/+ mice but not in AhR-/- mice in this study. These
results suggest that AhR plays a significant role in APE-induced carcinogenesis in AhR+/+ mice
and CYP1A1 activation of carcinogenic polycyclic aromatic hydrocarbons (the primary
carcinogenic component of APE) is also of importance.

5.1.2.3.3.3. Thyroid tumors.

The mode of action of TCDD in producing thyroid cancer in rodents has not been
elucidated. It is hypothesized that TCDD increases the incidence of thyroid tumors through an
extrathyroidal mechanism (see Figure 5-2). The prevailing hypothesis for the induction of
thyroid tumors by TCDD involves the disruption of thyroid hormone homeostasis via induction
of Phase II enzymes UGTs in the liver (reviewed in Brouwer et al., 1998, 201801) by an
AhR-dependent transcriptional mechanism (Bock et al., 1998, 548752; Nebert et al., 1990,
548756). This induction of hepatic UGT results in increased conjugation and elimination of
thyroxine (T4), leading to reduced serum T4 concentrations. T4 synthesis is controlled by the

This document is a draft for review purposes only and does not constitute Agency policy.

5-18 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

thyroid stimulating hormone (TSH) which is under negative and positive regulation from the
hypothalamus, pituitary, and thyroid via thyrotrophin-releasing hormone, TSH, T4, and
triiodothyronine. Consequently, the reduced serum T4 concentrations lead to a decrease in the
negative feedback inhibition on the pituitary gland. This would then lead to a rise in secreted
TSH and stimulation of the thyroid. The persistent induction of UGT by TCDD and the
subsequent prolonged stimulation of the thyroid could result in thyroid follicular cell hyperplasia
and hypertrophy of the thyroid, thereby increasing the risk of progression to neoplasia. Increases
in blood TSH levels are consistent with prolonged stimulation of the thyroid and may represent
an early stage in the induction of thyroid tumors identified in animal bioassays. Statistically
significant increases in neonatal blood TSH levels have been recently been reported in children
born to TCDD-exposed mothers in the Seveso cohort (Baccarelli et al., 2008, 197059. discussed
in Section 2.4.1.1.1.4.4). Support for this hypothesis comes from several studies showing that
TCDD decreases serum total thyroxine and free thyroxine concentrations in rats following both
single dose and repeated dose exposures (Bastomsky, 1977, 548760; Brouwer et al., 1998,
201801; Pohjanvirta et al., 1989, 548766; Potter et al., 1983, 548769; Potter et al., 1986, 548771;
Sewall et al., 1995, 198145; Van Birgelen et al., 1995, 198052). Further support comes from
studies of transgenic animals in which TCDD exposure resulted in a marked reduction of total
thyroxin and free T4 levels in the serum of AhR+/- mice but not AhR-/- mice (Nishimura et al.,
2005, 197860). Additionally, gene expression of UGT 1A6, CYP1A1, and CYP1A2 in the liver
was markedly induced by TCDD in AhR+/- but not AhR-/- mice (Nishimura et al., 2005,

197860).

5.1.2.3.4. Summary of TCDD mode of action in rodents.

Overall, there are inadequate data to support the conclusion that any of the particular
mode of action hypotheses described above is operant in TCDD-induced carcinogenesis.
However, the wealth of scientific evidence available indicates that most, if not all, of the
biological and toxic effects of TCDD are mediated by the AhR. Although the receptor may be
necessary for the occurrence of these events, it is not sufficient because other proteins and
conditions are known to affect the activity of the receptor and its ability to alter gene expression
or to induce other effects. Certain studies could be interpreted to indicate AhR-independent
mechanisms, although these studies have not clearly ruled out involvement of the AhR. The

This document is a draft for review purposes only and does not constitute Agency policy.

5-19 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

only consistent, but limited, evidence for TCDD-induced effects that do not involve the AhR
comes from studies using AhR-deficient transgenic animals. Here however, only minor effects
occurred following treatment with extremely high doses of TCDD. Thus, a toxic response to
TCDD has AhR interaction as a key event, but there are various species-, cell-, development-,
gender-, and disease-dependent differences in the cellular milieu that can affect the nature and
extent of the response observed.

The findings that many AhR-modulated effects are regulated with distinct specificity
supports the understanding that the molecular and cellular pathways leading to any particular
toxic event are extremely complex. Precise dissection of these events represents a considerable
challenge, especially in that a toxic response may depend on timely modulation of several genes
rather than of just one particular gene, and possibly modulation of these genes in several rather
than just one cell type or tissue.

While a defined mechanism at the molecular level or a defined mode of action for
TCDD-induced carcinogenicity is lacking, EPA concludes the following

•	interaction with the AhR is a necessary early event in TCDD carcinogenicity in
experimental animals.

•	through interaction with the AhR, TCDD modifies one or more of a number of cellular
processes, such as induction of enzymes, changes in growth factor and/or hormone
regulation, and/or alterations in cellular proliferation and differentiation.

•	AhR activation is anticipated to occur in humans and may progress to tumors. AhR is
present in human cells and tissues, studies using human cells are consistent with the
hypothesis that the AhR mediates TCDD toxicity and no data exist to suggest that the
biological effects of AhR activation by TCDD are precluded in humans.

•	non-AhR mediated carcinogenic effects of TCDD are possible.

5.1.3. Summary of the Qualitative Weight of Evidence Classification for TCDD

Under the 2005 Cancer Guidelines (U.S. EPA, 2005, 086237). TCDD is characterized as
carcinogenic to humans, based on the available data as of 2009. This conclusion is based on

•	Multiple occupational epidemiologic studies showing strong evidence of an association
between TCDD exposure and increased mortality from all cancers.

•	Epidemiological studies showing an association between TCDD exposure and certain
cancers in individuals accidentally exposed to TCDD in Seveso, Italy.

This document is a draft for review purposes only and does not constitute Agency policy.

5-20 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

•	Extensive evidence of carcinogenicity at multiple tumor sites in both sexes of multiple
species of experimental animals.

•	General scientific consensus that the mode of TCDD's carcinogenic action in animals
involves AhR-dependent key precursor events and proceeds through modification of one
or more of a number of cellular processes, such as induction of enzymes, changes in
growth factor and/or hormone regulation, and/or alterations in cellular proliferation and
differentiation.

•	The human AhR and rodent AhR are similar in structure and function and human and
rodent tissue and organ cultures respond to TCDD in a similar manner and at similar
concentrations.

•	General scientific consensus that AhR activation is anticipated to occur in humans and
may progress to cancers.

5.2. QUANTITATIVE CANCER ASSESSMENT

5.2.1. Summary of NAS Comments on Cancer Dose-Response Modeling

5.2.1.1. Choice of Response Level and Characterization of the Statistical Confidence Around
Low Dose Model Predictions

The NAS commented on the low dose model predictions in the 2003 Reassessment,

including EPA's development of ED0i (effective dose eliciting x percent response) estimates for

numerous study/endpoint combinations. The committee also suggested that EPA had not

appropriately characterized the statistical confidence around such model predictions in the low-

response region of the model.

The committee concludes that EPA did not adequately justify the use of the 1%
response level (the ED0i) as the POD for analyzing epidemiological or animal
bioassay data for both cancer and noncancer effects. The committee recommends
that EPA more explicitly address the importance of the selection of the POD and
its impact on risk estimates by calculating risk estimates using alternative
assumptions (e.g., the ED0s) (NAS, 2006, 198441. p. 18)

It is critical that the model used for determining a POD fits the data well,
especially at the lower end of the observed responses. Whenever feasible,
mechanistic and statistical information should be used to estimate the shape of the
dose-response curve at lower doses. At a minimum, EPA should use rigorous
statistical methods to assess model fit, and to control and reduce the uncertainty of
the POD caused by a poorly fitted model. The overall quality of the study design
is also a critical element in deciding which data sets to use for quantitative
modeling (NAS, 2006, 198441. p. 18).

EPA should ... assess goodness-of-fit of dose-response models for data sets and
provide both upper and lower bounds on central estimates for all statistical

This document is a draft for review purposes only and does not constitute Agency policy.

5-21 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

estimates. When quantitation is not possible, EPA should clearly state it and
explain what would be required to achieve quantitation (NAS, 2006, 198441. p.
10).

The NAS also suggested that EPA report information describing the adequacy of dose-
response model fits, particularly in the low-response region. For those cases where biostatistical
modeling was not possible, the NAS recommended that EPA identify the reasons.

The Reassessment should also explicitly address the importance of statistical
assessment of model fit at the lower end and the difficulties in such assessments,
particularly when using summary data from the literature instead of the raw data,
although estimates of the impacts of different choices of models would provide
valuable information about the role of this uncertainty in driving the risk estimates
(NAS, 2006, 198441. p. 73).

5.2.1.2. Model Forms for Predicting Cancer Risks Below the Point of Departure (POD)

The NAS focused much of its review on EPA's derivation of a cancer slope factor.
Specifically, the NAS commented extensively on the selection of the appropriate point of
departure (POD) and the extrapolation of dose response modeling below the POD.

The NAS questioned EPA's choice of a linear, nonthreshold model for extrapolating risk
associated with exposure levels below the POD, concluding that the current scientific evidence
was sufficient to justify the use of nonlinear methods when extrapolating below the POD for
TCDD carcinogenicity. The committee further recommended that EPA include a nonlinear
model for low dose cancer risk estimates as a comparison to the results from the linear model.

The committee concludes that EPA's decision to rely solely on a default linear
model lacked adequate scientific support. The report recommends that EPA
provide risk estimates using both nonlinear and linear methods to extrapolate
below PODs(NAS, 2006, 198441. p. 5).

After reviewing EPA's 2003 Reassessment and additional scientific data
published since completion of the Reassessment, the committee unanimously
agreed that the current weight of scientific evidence on the carcinogenicity of
dioxin is adequate to justify the use of nonlinear methods consistent with a
receptor-mediated response to extrapolate below the POD. The committee points
out that data from NTP released after EPA generated the 2003 Reassessment
provide the most extensive information collected to date about TCDD
carcinogenicity in test animals, and the committee found the NTP results to be

This document is a draft for review purposes only and does not constitute Agency policy.

5-22 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

compelling. The committee concludes that EPA should reevaluate how it models
the dose-response relationships for TCDD... (NAS, 2006, 198441. p. 16).

Because EPA's assumption of linearity at doses below the 1% excess risk level
for carcinogenic effects of TCDD, other dioxins, and DLCs is central to the
ultimate determination of regulatory values, it is important to critically address the
available scientific evidence on the most plausible shape of the dose-response
relationship at doses below the POD (LED0i). On the basis of a review of the
literature, including the detailed review prepared by EPA and presented in Part II
of EPA's Dioxin Risk Assessment and new literature available since the last EPA
review, the committee concludes that, although it is not possible to scientifically
prove the absence of linearity at low doses, the scientific evidence, based largely
on mode of action, is adequate to favor the use of a nonlinear model that would
include a threshold response over the use of the default linear assumption (NAS,
2006, 198441. p. 122).

On the whole, the committee concluded that the empirical evidence supports a
nonlinear dose-response below the ED0i, while acknowledging that the possibility
of a linear response cannot be completely ruled out. The Reassessment
emphasizes the lack of such nonlinear models, hence its adoption of the approach
of linear extrapolation below the POD level. Although this approach remains
consistent with the cancer guidelines (U.S. EPA, 2005, 086237; see also
Appendix B), EPA should acknowledge the qualitative evidence of nonlinear dose
response in a more balanced way, continue to fill in the quantitative data gaps,
and look for opportunities to incorporate mechanistic information as it becomes
available. The committee recommends adopting both linear and nonlinear
methods of risk characterization to account for the uncertainty of dose-response
relationship shape below ED0i (NAS, 2006, 198441. p. 72).

5.2.2. Overview of EPA Response to NAS Comments on Cancer Dose-Response Modeling

EPA agrees with the NAS that the approaches to cancer dose-response modeling for
TCDD should be clearly communicated and justified. Furthermore, due to the abundance of new
information on TCDD carcinogenicity published since the 2003 Reassessment, EPA has
reevaluated the cancer dose-response modeling for TCDD presented in the 2003 Reassessment.
As detailed below in Section 5.2.3, EPA has conducted an updated cancer dose-response
assessment for TCDD that incorporates key NAS recommendations discussed in this document,
reflects the current state-of-the science in cancer dose-response modeling and integrates new
TCDD carcinogenic information. Detailed responses to the NAS comments summarized above
are found in Section 5.2.3.3.

This document is a draft for review purposes only and does not constitute Agency policy.

5-23 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

The 2003 Reassessment presents an extensive dose-response assessment of TCDD and
provides a comprehensive summary of dose-response relationships. The analyses and
discussions synthesized a considerable breadth of data and model types, highlighting the
strengths and weaknesses of the then-available scientific information. Modeling included both
administered dose and steady state body burden dose metrics, taking into account variation in
half-lives of TCDD across species. These body burden calculations used a simple one-
compartment kinetic model based on the assumption of a first-order decrease in the levels of
administered dose as a function of time. An excess risk of 1% was chosen to model the cancer
data, but comparative results were also shown for 5% and 10% excess risk (see Table 8-2 of the
2003 Reassessment). Dose response was also explored thoroughly for a number of in vitro and
biochemical endpoints in addition to the in vivo data analyses, and ranges of these values were
presented (see Figures 8-1, 8-2 and 8-3 of the 2003 Reassessment). Thus, the 2003
Reassessment provides an initial evaluation of the carcinogenic database for TCDD and serves as
the foundation for the analyses presented below.

5.2.3. Updated Cancer Dose-Response Modeling for Derivation of Oral Slope Factor

The following sections describe the dose-response analysis of the cancer data from
epidemiologic cohort studies (see Section 2.4.1 and Table 2-4) and rodent bioassays (see
Section 2.4.2 and Table 2-6), concluding with the derivation of oral slope factors for TCDD
based on epidemiologic data (see Section 5.2.3.1) and rodent bioassay data (see Section 5.2.3.2).

5.2.3.1. Dose-Response Modeling Based on Epidemiologic Cohort Data

The 2003 Reassessment included dose-response analyses and the development of oral
slope factors from the following three occupational cohorts: the NIOSH cohort, the Hamburg
cohort, and the BASF cohort. In this document, EPA determined that specific studies from each
of these cohorts (Becher et al., 1998, 197173; Ott and Zober, 1996, 198408; Steenland et al.,
2001, 198589) met the epidemiologic study inclusion criteria (see Section 2.3.1 and
Section 2.4.1). In Section 5.2.3.1.1, the oral slope factors derived from these studies in the 2003
Reassessment are reviewed. Another study that met the current epidemiologic study inclusion
criteria (Warner et al., 2002, 197489) was also briefly discussed in the 2003 Reassessment, but
an oral slope factor was not derived from that study. In Section 5.2.3.1.2.2, EPA discusses its

This document is a draft for review purposes only and does not constitute Agency policy.

5-24 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

unsuccessful attempt to use the categorical results published by (Warner et al., 2002, 197489) to
develop an oral cancer risk estimate.

Since the publication of the 2003 Reassessment, additional cancer epidemiologic studies
based on these cohorts have been published in the peer-reviewed literature. Of these, Collins et
al. (2009, 197627) and Cheng et al. (2006, 523122) met the epidemiologic study inclusion
criteria (see Section 2.3.1 and Section 2.4.1). In Section 5.2.3.1.2, EPA evaluates the suitability
of deriving an oral slope factor from the Cheng et al. (2006, 523122) study and derives oral slope
factor estimates. Although the Collins et al. (2009, 197627) study met the study inclusion
criteria, EPA could not derive an oral slope factor from that study. In Section 5.2.3.1.2.3, EPA
discusses why an oral cancer risk estimate was not developed using the positive results for the
soft-tissue sarcoma mortality published by Collins et al. (2009, 197627).

5.2.3.1.1. Evaluation of Epidemiologic Studies Used in the 2003 Reassessment for OSF
Derivation.

In the 2003 Reassessment, EPA reported dose-response modeling results for three
epidemiologic studies of human occupational cohorts: the NIOSH cohort with data published by
Steenland et al. (2001, 198589); the Hamburg cohort with data published by Becher et al. (1998,
197173); and the BASF cohort with data published by Ott and Zober (1996, 198408). Each of
these studies is summarized in Section 2.4.1 of this document and in the 2003 Reassessment
(Part II, Chapter 8; Part III, Chapter 5). Furthermore, EPA has evaluated the suitability of these
studies for use in TCDD dose-response modeling and concluded that each of these studies meet
the inclusion criteria for epidemiology studies presented in Section 2.3.1.

Each of these studies reports all cancer mortality as an outcome. Steenland et al. (2001,
198589) and Becher et al. (1998, 197173) analyzed cohorts of primarily male workers who
experienced occupational exposures to TCDD over long periods of time, while Ott and Zober
(1996, 198408) studied a cohort of primarily male workers who were exposed to high TCDD
concentrations at a single point in time due to an industrial accident.

The authors of all three of these studies measured, and then back-extrapolated, TCDD
levels in a subset of workers to estimate exposures during employment and then the authors used
this information to estimate exposures in the remainder of the cohort. These measured TCDD
samples generally were collected decades after the last known occupational exposure. In each

This document is a draft for review purposes only and does not constitute Agency policy.

5-25 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

study, the authors relied on TCDD measures in the cohort to back-calculate serum lipid or body
fat levels of TCDD using a simple one-compartment kinetic model based on the assumption of a
first-order decrease in the levels of exposure dose as a function of time. The assumed half-life of
TCDD used in the models varied from study to study. None of the studies sampled TCDD levels
from the entire cohort; for example, Ott and Zober collected samples from 138/243 workers
(57% of the cohort), which was the highest percentage of workers sampled among the three
studies. Steenland et al. (2001, 198589) and Becher et al. (1998, 197173) used the measured and
back-extrapolated TCDD concentrations to estimate the exposures that were associated with
various occupations within the cohort, and subsequently used this information to develop
exposure matrices (i.e., the TCDD load per unit time for an occupation) that then could be used
to estimate the cumulative dioxin dose for each cohort member. Ott and Zober (1996, 198408)
used regression procedures with data on time spent at various occupational tasks to estimate
TCDD levels for all members of the cohort. Following the estimation of worker exposures in
each of these three studies, the studies' authors divided these cohorts into exposure subgroups
based on the estimated TCDD levels.

In the 2003 Reassessment, EPA identified a POD based on a 1% response in cancer
mortality (ED0i) for the Steenland et al. (2001, 198589). and the Ott and Zober (1996, 198408)
studies. EPA extrapolated from this POD to lower doses using a straight line drawn from the
POD to the origin—zero incremental dose, zero incremental response—to give a probability of
extra risk. Because there was insufficient evidence to support an assumption of nonlinearity,
EPA chose to develop these models using a linear model.

5.2.3.1.1.1. Steenland et al. (2001,19S5S9).

Steenland et al. (2001, 198589) developed dose-response models based on TCDD
exposures and all cancer mortalities from eight plants in the NIOSH cohort (see Section
2.4.1.1.1.1.3 for study details). Serum lipid levels of TCDD in 1988 were measured in
193 workers at one of these plants. Steenland and coauthors relied on a first-order kinetic model
(assuming a constant 8.7 year half-life) to back-extrapolate to serum TCDD levels at the time of
the last occupational exposure. The study authors assigned exposure estimates to each of the
3,538 workers in the cohort based on a job-exposure matrix. This matrix was based on (1) an
estimated level of contact with TCDD, (2) the degree of TCDD contamination of the products

This document is a draft for review purposes only and does not constitute Agency policy.

5-26 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

the workers produced, and (3) the fraction of a workday during which the worker likely
contacted the TCDD-contaminated products. They then estimated each worker's serum TCDD
levels as an area under the concentration curve (AUC) for lipid-adjusted serum levels over time.
The mortality analysis was conducted on 256 cancer decedents.

Several different dose-response models were fit to these data to provide estimates of fatal
cancer risk. The best-fitting model was a Cox regression exposure-response model using the
log(AUC) of TCDD lipid concentration (ppt-year) lagged by 15 years as the exposure metric.
Steenland and colleagues also developed a piecewise linear regression model with no lag, in
which two separate linear slopes were estimated. This analysis assumed a background exposure
of 0.5 pg/kg-day. The lipid concentrations were converted to body burdens by dividing by 4.
The central tendency estimate and lower bound ED0iS from the piecewise linear model and their
associated cancer slope factors for the most sensitive endpoint (male cancer mortality) are
presented in Table 5-1.

5.2.3.1.1.2. Becheretal. (1998.197173).

Based on the Hamburg cohort, Becher et al. (1998, 197173) reported a dose-response
analysis for all fatal cancers combined (see Section 2.4.1.1.1.3.4 for study details). The mortality
analysis was conducted in 1992 on 124 cancer decedents. The exposure variable in the study
was the integrated blood levels for TCDD concentration over time (AUC, ng/kg-years), as
estimated by Flesch-Janys et al. (1998, 197339); these were converted to body burdens by
dividing by 4. Estimates of the half-life of TCDD, based on the sample of 48 individuals with
repeated measures, were incorporated into a model that back-extrapolated TCDD exposures to
the end of the employment after accounting for the workers' ages and body fat percentages.

These estimated exposure measures were then applied to the entire cohort, which consisted of all
1,189 regular male employees who were employed for at least 3 months between 1952 and 1984
at the Boehringer chemical plant in Hamburg, Germany.

Becher et al. (1998, 197173) used a Cox regression approach for the dose-response
modeling and developed three models: a multiplicative model, an additive model, and a power

This document is a draft for review purposes only and does not constitute Agency policy.

5-27 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

model.32 The response variable in each model was the SMR for total cancer mortality. The
models were calculated with lag times of up to 20 years. The multiplicative model provided the
best fit; however, the study authors judged the fits for all three models to be acceptable. The
model results were used to calculate unit risk estimates derived as the risk of cancer death
through age 70 given a daily dose of 1 pg/kg body weight of TCDD minus the risk given no
exposure to TCDD. These calculations were based on background German cancer mortality
rates. The model results were used to calculate cancer risk estimates. The lower bound
estimates on the dose were not available for models published by Becher et al. due to the absence
of statistical parameter measures. The central tendency estimate ED0is from the three statistical
models and their associated cancer slope factors are presented in Table 5-1.

5.2.3.1.1.3. Ott and Zober (1996.198408).

In the 2003 Reassessment, EPA also developed a dose-response analysis based on a study
reported by Ott and Zober (1996, 198408) for cancer incidence and mortality experienced by
243 men, who were exposed to TCDD in 1953 during an accident at the BASF plant in Germany
(see Section 2.4.1.2.1.2.1 for study details). The cohort was followed through 1992. TCDD
blood lipid levels were available for 138 of these men 30 years after the accident. These levels
were back-extrapolated and used to estimate the AUC for TCDD. Body burdens (ng/kg) were
estimated by dividing AUC by 4, and steady-state body burdens were estimated assuming a
constant half-life of approximately 7.1 years.33 Ott and Zober (1996, 198408) used Cox
proportional hazard approaches to estimate both cancer incidence and cancer mortality risk per

32The "multiplicative model" set relative risk (RR) equal to exp(fid), where the dose d is the AUC. The "additive
model" set RR = 1 +fid. and the "power model" set RR = exp(/? log (kd+\)). The values /; and k are estimated
parameters.

33Based on the initial body burden (B0) EPA estimated the body burden at time t using the following formula:

k t

B(t) = B0e c . where ke is an elimination constant equal to ln(2)/(half-life in years). This implies that the AUC at

Br -k T

time T after initial exposure is A UC = — (1 -e e ). T in this case was 39 years (time from the accident in 1953 to

K

the follow-up in 1992). Dividing by a lifetime of 71 years (mean age in 1954, 33 years, plus 38 years from 1954 to
the followup in 1992) yields the lifetime mean body burden as:

D	i

B =	°—(\-e<> \. In the 2003 Reassessment, EPA converted the steady-state body burden to units of equivalent

mean _, , ^	'

i\ke

1	k T

initial dose by dividing by the constant	(l - e e ). With the given values for half-life and T, that constant is

l\ke

0.1411 and l/(the constant) is 7.09.

This document is a draft for review purposes only and does not constitute Agency policy.

5-28 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

unit TCDD dose.34 Ott and Zober reported conditional risk ratios for cancer mortality that were
slightly larger than the conditional risk ratios for cancer incidence, which is counter-intuitive.
The risk of cancer mortality would be expected to be greater than the risk of cancer incidence.
The conditional risk ratio (and 95%CI) for all cancer mortality (1.22; 1.00-1.50) exceeded the
conditional risk ratio for all incident cancer cases (1.11; 0.91-1.35). Similarly, the conditional
risk ratios for digestive cancer mortality (1.46; 1.13-1.89) and respiratory cancer mortality (1.09;
0.70-1.68) were also both larger than the conditional risk ratios for all digestive cancers (1.39;
1.07-1.69) and all respiratory cancers (1.02; 0.65-1.59). As expected, in this cohort, incident
cases exceeded cancer mortality for total cancers (47 vs. 31), digestive cancers (12 vs. 11) and
respiratory cancers (13 vs. 11). Ott and Zober also reported that conditional risks for mortality
for all cancer and lung cancer associated with cigarette smoking were also higher than the
respective incidence risks. In their Cox regression analysis, Becher et al. (1998, 197173) also
report that the regression coefficient for total cancer mortality (0.0096) was slightly larger than
the regression coefficient for total cancer incidence (0.0089). The finding of Ott and Zober and
Becher et al. that the risk of cancer mortality is greater than cancer incidence is possibly due to a
systematic difference in the reference population for incidence vs. the reference population for
mortality. The central tendency estimate and lower bound ED0iS from the modeling and their
associated cancer slope factors are presented in Table 5-1.

5.2.3.1.2. Evaluation of Other Epidemiologic Studies Considered for OSF Derivation.

Three additional epidemiologic studies that met the study inclusion criteria (see
Section 2.3) for use in dose response modeling as set forth in this document are evaluated in this
section for the estimation of cancer risk estimates. These studies were either published after the
Reassessment (Cheng et al. (2006, 523122) and Collins et al., (1996, 197637)). or not used to
derive an OSF in the Reassessment (Warner et al., 2002, 197489). Each study is summarized in
Section 2.4.1.

34The model from Ott and Zober has risk proportional to d"''c with [i = ln(1.22). The corresponding slope for the

mean (steady-state) body burden is 7.0851 * log(1.22) * 0.001 (the 0.001 converts nanograms to micrograms).

This document is a draft for review purposes only and does not constitute Agency policy.

5-29 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

5.2.3.1.2.1. Chens et al. (2006. 523122).

As discussed in Section 2.4.1.1.1.1.4, Cheng et al. (2006, 523122)) analyzed the
relationship between TCDD dose and all cancer mortality for the same subset of NIOSH workers
as analyzed previously by Steenland et al. (2001, 198589). In contrast to Steenland et al., Cheng
et al. (2006, 523122) used the "concentration- and age-dependent elimination model"
(concentration- and age-dependent elimination [CADM], discussed in Section 3.3; see also
Ay 1 ward et al. (2005, 197114)), rather than a constant 8.7-year half-life, and calculated serum-
derived TCDD estimates for use in dose-response analysis. An important feature of CADM is
that it incorporates concentration- and age-dependent elimination of TCDD from the body,
meaning that the effective half-life of TCDD elimination varies based on exposure history, body
burden, and age of the exposed individuals. As discussed in Section 3.3, the use of the CADM
model to simulate TCDD kinetics in the NIOSH cohort results in time-integrated body burden
estimates four to five times greater than those obtained with the simple first-order model, with
smaller differences between the two methods at lower exposures.

Following the estimation of dose using the CADM-derived AUC values, Cheng and
colleagues (Cheng et al., (2006, 523122); the "Cheng analysis") derived dose-response estimates
for the NIOSH cohort using linear Cox regression for both lagged and un-lagged exposure on
various subsets of the data (high-exposures trimmed). The results for the lagged-exposure
analysis are summarized in Table 5-2. For comparison, the Cox regression coefficient from the
analysis conducted by Steenland et al. (2001, 198589), which relied on a first-order elimination
rate model assuming a constant 8.7-year half-life, is also shown in the table. As in Steenland et
al. (2001,198589),35 Cheng et al. (2006, 523122) found a much stronger relationship between
cancer mortality and exposure metrics lagged 15 years compared to the relationships for
unlagged exposures. Cheng et al. (2006, 523122) also noted that the dose-response relationship
plateaued above the 95th percentile of exposure. For exposures lagged 15 years, the regression
coefficient (P) of the linear slope derived by Cheng et al. (2006, 523122) was 3.3 x 10~6 per
ppt-year lipid-adjusted serum TCDD, with a standard error of 1.4 x 10 6 (Table III of Cheng et
al. (2006, 523122)). The upper 5% of the exposure range (individuals >252,950 ppt-year lipid
adjusted serum TCDD) was excluded in estimating this slope. Because this exclusion reduces

35 Lagged exposures modeled only for log-transformed serum concentrations, not for untransformed serum
concentrations in the piece-wise linear model.

This document is a draft for review purposes only and does not constitute Agency policy.

5-30 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

the upper portion of the response where the slope is shallow36, this likely better represents the
slope in the region of the curve where the fatal cancer risk is increasing with dose, which is the
equivalent of dropping the highest dose in an animal bioassay or using a piece-wise linear model
as in Steenland et al. (2001, 198589).

To develop cancer risks for TCDD, EPA used the modeling results of the Cheng analysis,
with conversion to oral intake using the Emond human PBPK model as follows. The slope (ft)
from the Cheng analysis is the slope of the linear relationship between the natural logarithm of
the rate ratio (RR) and the cumulative fat TCDD concentration (fat-AUC). Conceptually, the
slope (P) is similar to an OSF, except that it is expressed in terms of fat-AUC rather than intake.
Also, the slope represents the incremental increase in cancer mortality (expressed as an RR)
above the background TCDD exposure experienced by the NIOSH cohort rather than above zero.
Using the upper 95% bound on (J> and assuming that the slope is the same below the NIOSH
cohort background exposure level (approximately 5 ppt/yr TCDD fat concentration), EPA
calculated risk-specific doses (as daily oral intakes) for TCDD for risk levels of concern to EPA.
The risk-specific doses were estimated from the Emond human PBPK model for the lifetime-
average TCDD fat concentrations corresponding to the fat-AUC predicted by the Cheng et al.
model for each of the risk levels of concern. The steps in this computation are as follows:

•	Background cancer mortality risk estimate (Ro). EPA used an R0 of 0.112 as reported by
Cheng et al. (2006, 523 122)37

•	Total cancer mortality risk in the exposed group associated with a specified (extra) risk
level (RL) of fatal cancer (TRw). A TRm associated with any given extra risk level (e.g.,
0.01, 1 x 10"6) can be calculated using the following relationship for extra risk:

ER = TRrl ~ Rq	(Eq 5_^

\-R0

36	Steenland et al. (2001, 1985891: Steenland and Deddens (2003, 1985871 found a slightly negative slope for the
higher exposures, stating that the phenomenon could be a result of exposure misclassification, depletion of
susceptible individuals or saturation of receptor-mediated processes.

37	In Table IV, Cheng et al. (2006, 5231221 report two estimates of background fatal cancer risk. R0, for males aged
75 years: 0.112 and 0.124. A R0 estimate of 12.4% was used by Steenland et al. (2001, 198589). and 11.2%, as
estimated for all males in the 1999-2001 Surveillance Epidemiology and End Result data set. EPA chose to use the
more recent estimate of 11.2% for the purpose of predicting risk in the current U.S. population.

This document is a draft for review purposes only and does not constitute Agency policy.

5-31 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

• Incremental cancer mortality risk in the exposed population based on a given extra risk

(Rn). Rd, is calculated as the difference between the total risk and background risk and
expressed in terms of RL and R0 by combining Equations 5-2 and 5-1.

Rd = TRrl-R0	(Eq. 5-2)

RD = RLx (\-R0)	(Eq. 5-3)

•	Cumulative TCDD concentration in the fat compartment for a given extra risk (AUCrt
AUCrl is then calculated by taking the natural logarithm of Equation 3 from Cheng et al.
(2006, 523122). rearranging and substituting for RR38 (RR = [RD + R0]/Ro):

AUCrl = In((Rd + R0)IRo)lp*	(Eq. 5-4)

where /?* is the central-tendency regression slope or the 95% upper bound (figs)
determined by summing the regression coefficient (P) and the product of 1.96 and the
standard error of the regression coefficient, yielding an estimate of 6.0 x 10 6 per
ppt-year lipid adjusted serum TCDD, as follows:

P95 = p + \.96*SE	(Eq. 5-5)

•	Continuous daily TCDD intake associated with a given extra risk TDrt 1. Because the fat
concentrations generated by CADM are not linear with oral exposure at higher doses, a
single oral slope factor to be used for all risk levels cannot be obtained; the response is
approximately linear with fat concentrations and oral intake at lower doses. Instead, a
risk-specific DRL must be estimated by converting the respective AUCrl to the
corresponding lifetime daily intake, using an appropriate human toxicokinetic model.
EPA has chosen to use the Emond human PBPK model for this purpose because the
CADM configuration does not facilitate this process and so that the dose conversions are
consistent with those used in the derivation of the RfD. A DRL is obtained from the
Emond model by finding the average lifetime daily intake corresponding to the AUCrl in
the fat compartment.39

Note that there are two nonlinear steps in the estimation of risk-specific doses from the
Cheng et al. model. First, fat-AUC {AUCrl) and the incremental cancer mortality risk (RD) do

38	As defined by Cheng et al. (2006, 523122. p. 1063).

39	Although the NIOSH cohort exposures are reported as LASC, they are treated as fat concentrations in the Cheng
analysis because fat in all tissues are modeled as one compartment (hence equal) in CADM. The translation to oral
intake in the Emond model is from the fat compartment, rather than the serum compartment, even though the serum
and fat compartments are not equivalent, because the regression slope ((3) in the Cheng analysis is in terms of the
(equivalent) fat compartment.

This document is a draft for review purposes only and does not constitute Agency policy.

5-32 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

not have a linear relationship (see Eq. 5-5); however, the relationship becomes virtually linear
below an incremental risk of 10"3 (see Table 5-3). Second, TCDD fat concentration is not linear
with oral intake in the Emond human PBPK model (see Section 3); this relationship also is close
to linear below the 10"5 risk level. The resulting predicted cancer-mortality risk is approximately
linear with daily oral intake at low doses. Table 5-3 shows the AUCrl and DRL based on the
95% upper-bound regression slope {figs) from the Cheng analysis for a number of risk levels of
interest to the EPA. For comparative purposes, EPA has also shown the equivalent oral slope
factor (RL +Drl) for those same risk levels. Table 5-4 also shows analogous results based on
the best estimate of regression coefficient (fi = 3,3 x 10 6) for total fatal cancers from the Cheng
analysis.

5.2.3.1.2.2. Warner et al. (2002.197489).

Warner et al.(2002, 197489) is a study of 981 females exposed to elevated TCDD levels
following the Seveso accident of 1976 (see Section 2.4.1.1.1.4.2 for study details). The TCDD
exposure pattern involving a single period of elevated TCDD exposures followed by an extended
period of lower ambient level TCDD exposures and elimination is similar to that of the BASF
cohort (Ott and Zober, 1996, 198408). TCDD levels, measured or estimated in blood lipids
shortly after the time of the accident, were available for all women. These women were divided
into four exposure groups of <20, 20-44, 44.1-100, and >100 ppt. In this cohort, 21 total
cancers have been observed; 15 of these were breast cancer cases and 3 were thyroid cancer
cases. Cox proportional hazards modeling showed that the hazard ratio for breast cancer
associated with a 10-fold increase in serum TCDD levels (logio (TCDD)) was significantly
increased to 2.1 (95% CI = 1.0-4.6). Rate ratios (95% CI) for cancer incidence in these 4 groups
were 1.0, 1.0 (0.2-5.5), 2.2 (0.5-10.8) and 2.5 (0.5-11.8). Using a Cox proportional hazards
model and assuming continuous exposure, the rate ratio was 1.7 (0.9-3.4) for each 10-fold
increase in serum TCDD; that is, a logio transformation of the exposure estimates in their
analysis was presented. They reported a test for trend ofp = 0.09.

EPA attempted to estimate an ED0i from the modeled results of Warner et al. (2002,
197489) from the statistically significant hazard ratio for breast cancer. However, EPA had to
estimate the slope of the tangent to the log-linear relationship. Because the exponentiated slope
of a log-dose linear relationship is not constant but varies with dose, and because the lowest

This document is a draft for review purposes only and does not constitute Agency policy.

5-33 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

exposure measure was well-above the 1% response region of interest, EPA could not confidently
estimate the tangent to the log-dose linear relationship. The transformation of the logio dose
units to linear units of TCDD yielded an implausibly low EDoi and correspondingly high cancer
risk that was inconsistent with a visual inspection of the untransformed plot. EPA was not
confident in these values for health risk assessment because of uncertainties in the transformation
in the low response region of the original model. Thus, EPA did not derive an ED0i or oral slope
factor for this study.

5.2.3.1.2.3. Collins et al. (2009.197627).

Collins et al. (2009, 197627s) investigated the relationship between serum TCDD levels
and mortality rates in the NIOSH cohort (see Section 2.4.1.1.1.1.5 for study details). The
investigators completed an extensive dioxin serum evaluation of workers employed by the Dow
Chemical plant in Midland, Michigan, that made 2,4,5-trichlorophenol (TCP) from 1942 to 1979
and 2,4,5-T from 1948 to 1982. Collins et al. (2009, 197627) developed historical TCDD
exposure estimates for all 1,615 workers using serum samples from 280 former workers that
were collected during 2004-2005. Investigators calculated a cumulative measure of exposure
using a simple one-compartment first-order pharmacokinetic model and elimination rates as
estimated from the BASF cohort (Flesch-Janys et al., 1996, 197351). The follow-up interval for
these workers covered the period between 1942 and 2003. Thus, the study included 10 more
years of follow-up than earlier investigations of the entire NIOSH cohort. A key limitation of
this study is that the derivation of the SMRs and slope parameters did not include a lag period,
unlike other analyses of the NIOSH cohort (e.g., Cheng et al., 2006, 523122; Steenland et al.,
2001, 198589).

Although results were largely negative, statistically significant mortality in the cohort
was found for soft-tissue sarcoma (SMR = 4.1, 95% CI = 1.1-10.5), based on only four deaths.
A regression coefficient of 0.05872 (standard error not reported), and a hazard ratio of 1.060
(95% CI = 1.017 to 1.106) were reported by Collins et al. (2009, 197627) for soft-tissue sarcoma.
Although it met the dose-response study criteria, EPA could not calculate an upper bound on the
regression coefficient because the standard error was not given. In addition, EPA was unable to
estimate an extra-risk value because the reference population response was not specified. Thus,
EPA did not derive an ED0i or oral slope factor for this study.

This document is a draft for review purposes only and does not constitute Agency policy.

5-34 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

5.2.3.2. Dose-Response Modeling Based on Animal Bioassay Data

Figure 5-3 provides a summary of the process EPA has utilized to select and identify
candidate TCDD OSFs from key animal bioassays that were identified in Section 2.4.3 of this
document. For each in vivo animal cancer study that qualified for TCDD dose-response
assessment using the study inclusion criteria specified in Section 2.3.2, EPA first selected the
species/sex/tumor data set combinations that had been characterized as having statistically
significant increases in tumor incidence by either a pair-wise test between the treated group and
the controls or by a trend test showing increases in tumors with increases in dose. Next, EPA
used the Emond animal kinetic model discussed in Section 3 to estimate blood concentrations
corresponding to each study's average daily administered doses for use in dose response
modeling. Benchmark dose lower confidence bounds (BMDLois) were then estimated for the
blood concentrations by (1) using the multistage cancer model for each species/sex/tumor
combination within each study and (2) using a Bayesian Markov Chain Monte Carlo framework
that assumes independence of tumors, modeling all tumors together for each species/sex
combination within each study. The final selected models were subjected to goodness-of-fit tests
and visual inspections of fit to the raw data. Thus, for each sex/species combination within each
study, this process generated a BMDLoi for each single tumor type and another BMDLoi for the
combined tumors. Finally, using the Emond human kinetic model discussed in Section 3, human
equivalent doses (BMDLreds) were then estimated for each of the BMDLois and, using a linear
extrapolation, OSFs were calculated by OSF = 0.01/BMDLHed- The highest OSF for a
species/sex combination for either a single tumor type or all combined tumors was selected as a
candidate OSF for TCDD cancer assessment. These steps in Figure 5-3 are further described in
detail in the following sections.

5.2.3.2.1. Selection of key data sets.

Based on the study selection criteria outlined in Section 2.3.2 (see Figure 2-3), EPA
selected five animal bioassays for use in the cancer dose-response assessment for TCDD (see
Table 2-6 and Section 2.4.2 for detailed study descriptions). Four of these studies (Delia et al.,
1987, 197405; Kociba et al., 1978, 001818; NTP, 1982, 594255; Toth et al., 1979, 197109). were
evaluated in the 2003 Reassessment, while one study (NTP, 2006, 543749) was published after
the 2003 Reassessment was released. The NTP (2006, 543749) study was specifically called out

This document is a draft for review purposes only and does not constitute Agency policy.

5-35 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

by the NAS (2006, 198441)) report as cancer bioassay data that EPA should evaluate prior to
completing its TCDD dose-response assessment. As discussed below, EPA has chosen to
conduct dose-response modeling for a number of tumor types from each of the sex/species
combinations in these studies in order to maximize the amount of information available to
support OSF derivation. Because tumors occurred in multiple sites in the exposed animals, each
tumor type was considered separately (individual tumor models) and were also combined into
composite tumor incidence dose estimates (multiple tumor models).

The tumor incidence tables for these five bioassays are shown in Tables 5-5 through 5-14
(see Section 2.4.2 for details of these studies). The data in these tables are summarized from
each study's reference publication and are the species/sex/tumor incidence data used for TCDD
dose-response modeling in this report. EPA selected the animal bioassay data sets in Tables 5-5
through 5-14 because they had been characterized by the study authors as having statistically
significant increases in tumor incidence by either a pair-wise test between at least one treated
group and the controls or by a trend test showing increases in tumors with increases in dose. An
exception was made for cases where statistical significance was found in only one dose group
that was not the highest dose group, and there were zero responses in every other dose group
including controls; these datasets were not modeled. For example, in NTP (2006, 543749), EPA
notes that while the uterine tumors were statistically significant at 46 ng/kg using a pair-wise
test, there were no uterine tumors in any other dose group, including the control and high dose
groups, and the trend test was not significant; EPA excluded this tumor type from the analysis.
In addition, datasets with combined tumors for the same site were given priority over subsets of
tumors for that site. For example, in the NTP (1982, 594255) study on female mice, data on
combined hepatocellular adenomas or carcinomas were modeled, but not data on hepatocellular
adenomas alone (not statistically significant) or on hepatocellular carcinomas alone (statistically
significant trend and high dose group). In the case of the Kociba et al. (1978, 001818) female rat
combined hepatocellular adenomas and carcinomas only, EPA used data from a reanalysis of the
pathology slides that was published by Goodman and Sauer (1992, 197667); because the study
authors did not statistically analyze the revised tumor incidence data from their reanalysis, EPA
applied a Fischer's Exact Test to evaluate the statistical significance of those data. In the case of
the NTP (2006, 543749) study only, information was available regarding the length of time that
the animals stayed on test (105 weeks); animals who died within the first year were censored

This document is a draft for review purposes only and does not constitute Agency policy.

5-36 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

from analysis in this document because animals who died within the first year were not
considered to have been alive long enough to develop tumors. Therefore, those animals were not
included in the denominators in Table 5-11. These adjusted incidence data were used in the
analysis of tumor dose-response for NTP (2006, 543749) in this document. The tumor incidence
data in Tables 5-5 through 5-14 include

•	nasal, tongue and adrenal tumors in males (Table 5-5), and liver, nasal and lung tumors in
females from the Kociba et al. (1978, 001818) 2-year study of Sprague-Dawley rats
(Table 5-6),

•	subcutaneous tissue, liver, adrenal and thyroid tumors in females (Table 5-7) and liver,
thyroid and adrenal tumors in males (Table 5-8), from the NTP (1982, 594255) 2-year
study of Osborne-Mendel rats,

•	subcutaneous tissue, hematopoietic system, liver and thyroid tumors in females
(Table 5-9), and lung and liver tumors in males, from the NTP (1982, 594255) 2-year
study of B6C3Fi mice (Table 5-10),

•	liver, oral mucosa, pancreas and lung tumors in females from the NTP (2006, 543749) 2-
year study of Sprague-Dawley rats (Table 5-11),

•	liver tumors in males from the Toth et al. (1979, 197109) 1 -year study of Swiss/H/Riop
mice (Table 5-12), and

•	liver tumors in males (Table 5-13) and females from the Delia Porta et al. (1987, 197405)
52-week study of B6C3Fimice (Table 5-14).

For each cancer endpoint, the reported (administered) doses from each study were converted,
where necessary, to average daily doses in ng/kg-day (e.g., doses administered 5 days/week were
adjusted by multiplying by 5 and dividing by 7 to get average daily doses). These doses were
then subjected to kinetic modeling to generate blood concentrations for use in TCDD dose-
response modeling.

5.2.3.2.2. Dose adjustment and extrapolation methods for selected data sets.

5.2.3.2.2.1. Dose metric estimation for dose-response modelin2.

Tables 5-5 through 5-14 show the blood concentrations that were used in TCDD dose-
response modeling of the animal bioassay data. Based on kinetic analysis (see Section 3), a
choice for whole blood concentration of TCDD was made for the purpose of dose extrapolation
between animals and humans. In order to estimate blood concentrations for each study selected,

This document is a draft for review purposes only and does not constitute Agency policy.

5-37 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

the Emond PBPK model was run using ACSLX® software, version 2.5.0.6 (see Section 3).
Depending on the selected study, either rat or mouse versions of the model were used. In each
case, the simulation was performed using the exposure and observation durations, the body
weights, and the adjusted doses from the original studies. Details of PBPK model input
parameters are given for each study's m-file in Appendix C.2. In the case of Toth et al. (1979,
197109) study, which dosed the animals for a year and then followed up for the lifetime of the
animal, only the one-year simulation was performed. The m-files were used to run the
appropriate PBPK model to estimate time-averaged, maximum, and terminal (end of exposure)
blood concentration (see Appendix C.3). Other model simulated dose metrics such as
concentrations for liver, fat, Ah-receptor bound in liver, body burden, and the time at which the
maximum concentration was reached for each dose metric are also reported for illustrative
purposes in Appendix C.3. The complete results for each study modeled are shown in
Appendix C.3.

5.2.3.2.2.2. Calculation of human equivalent doses (HEDs).

Human equivalent doses (ng/kg-day), corresponding to each BMDL (ng/kg) were
calculated using the Emond human PBPK model (see Section 3) and are denoted as BMDLreds.
The Emond human PBPK model was run for 70 years assuming a constant daily dose starting
from birth. The model concentrations were averaged over both the entire 70 year lifetime
(lifetime average) and over the five years surrounding the peak concentration (five-year average)
(see Section 3.3.1, describing first order body burden estimation). The human equivalent doses
were estimated by adjusting the daily dose model input until the time-averaged whole blood
concentration matched the associated alternative dose BMDL (derived earlier from animal PBPK
model). For animal studies which lasted longer than 540 days, the lifetime average was used; for
studies lasting less than 540 days, the five year average was used. The process was iterative and
continued until the modeled human concentration was within 1% of the BMDL. In general,
however, the concentrations matched to within 0.1%.

This document is a draft for review purposes only and does not constitute Agency policy.

5-38 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

5.2.3.2.3. Dose-response modeling approaches for rodent bioassays.

5.2.3.2.3.1. Modelin2 of individual tumors.

EPA's BMDS Software, version 2.1 was used to estimate the BMDLois for each of the
species/sex/tumor combinations, using the blood concentrations and incidence data shown in
Tables 5-5 through 5-14. Each data set was modeled using the multistage cancer model, and a
BMDLoi in blood concentration was estimated. The multistage model has been used by EPA in
the majority of its quantitative cancer assessments because it is statistically robust and able to
provide good fits to a wide range of dose-response patterns. It is also consistent with the
multistage nature of the carcinogenic process. The mathematical form of the multistage model is

P(d) = 1 - exp[-(^0 + q\d + q2d2 + ... + qgf)]	(Eq. 5-6)

where

P(d) = lifetime excess risk (probability) of cancer at dose d

q, = parameters estimated in fitting the model, i = 1, ...,k.

To estimate the BMDoiS and BMDLois, BMDS was run with all parameters set to their
defaults; up to three degrees of freedom were specified for the dichotomous, multistage cancer
model; and a 95% confidence level. A 1% extra risk benchmark response (BMR) was used for
each tumor type, as this response level was judged to be sufficiently close to the observed
responses (see Section 5.2.3.2.6.11 for an expanded discussion). The BMDLoi (ng/kg) was then
converted to a BMDLred (ng/kg-day) using the Emond human model, and an OSF in units of
(mg/kg-day)-1 was calculated by, OSF = 0.01/BMDLHed x106. Because of the nonlinearity of
blood concentration and ingested dose in the Emond Human PBPK model, the cancer risk is only
approximately linear with the TCDD blood concentration and low TCDD oral ingestion doses,
but is not linear with ingested TCDD at higher doses.40 Thus, to use these estimates in human
health risk assessment, risk-specific TCDD oral intake levels corresponding to the target risk
levels should be calculated, using a procedure similar to that for the slope factors based on
epidemiologic data (see Table 5-3). In the following sections, results are presented for the

40 This situation is analogous to that for the cancer risk modeling of epidemiologic data from the Cheng et al. (2006)
analysis in Section 5.2.3.1.2.1.

This document is a draft for review purposes only and does not constitute Agency policy.

5-39 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

models that provided the best overall fit to the data as judged by comparison of likelihood ratios
for models that had an acceptable fit (chi-squared goodness of fit statisticp > 0.05).

5.2.3.2.3.2. Multiple tumor (Bavesian) models.

Statistically significant increased tumor incidences were observed at multiple sites in
male and/or female rats (Kociba et al., 1978, 001818; NTP, 1982, 594255; NTP, 2006, 543749)
and male and female mice (NTP, 1982, 594255) following oral exposures to TCDD. With this
multiplicity of tumors, the concern is that a potency or risk estimate based solely on one tumor
site (e.g., the most sensitive site) may underestimate the overall cancer risk associated with
exposure to this chemical. Relevant approaches in the 2005 Cancer Guidelines (U.S. EPA, 2005,
086237) for characterizing total risk include the following: (1) analyze the incidence of tumor-
bearing animals, or (2) combine the potencies associated with significantly elevated tumors at
each site. The NRC (1994, 006424) concluded that an approach based on counts of animals with
one or more tumors (tumor-bearing animals) would tend to underestimate overall risk when
tumor types occur independently, and thus an approach based on combining the risk estimates
from each separate tumor type should be used. On independence of tumors, NRC (1994,

006424) stated "... a general assumption of statistical independence of tumor-type occurrences
within animals is not likely to introduce substantial error in assessing carcinogenic potency."

Because potencies are typically upper bound estimates, summing such upper bound
estimates across tumor sites is likely to overstate the overall risk. Therefore, following the
recommendations of the NRC (1994, 006424) and the 2005 Cancer Guidelines (U.S. EPA, 2005,
086237). a statistically valid upper bound on combined risk was derived, assuming
independence, in order to gain some understanding of the overall risk resulting from tumors
occurring at multiple sites. In the case of TCDD, tumors are thought to be independent across
the sites found in these three studies because: (1) they are in different organs and tissues,
specifically liver, lung, thyroid, subcutaneous tissue, oral cavity, tongue, pancreas, adrenal cortex
and the hematopoietic system; (2) different kinds of tumors were found, even within the same
organ (e.g., both cholangiocarcinomas and hepatocellular adenomas were found in female rat
livers in NTP (2006, 543749); and (3) the tumors found in these studies were not progressive
(i.e., they did not metastasize to other sites in the body). It is important to note that this estimate

This document is a draft for review purposes only and does not constitute Agency policy.

5-40 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

of overall potency describes the risk of developing tumors at any combination of the sites and is
not the risk of developing tumors at all sites simultaneously.

For modeling individual tumor data, the multistage model is specified as shown in the
previous section (see Eq. 5-6). Under the assumption of independence, the model for the
combined (or composite) tumor risk is still multistage, with a functional form that has the sum of
stage-specific multistage coefficients as the corresponding multistage coefficient.

Pc(d) = 1 - exp[-(Lq0i + dLqn + cfl.q2i + ... + d"I.qm)\, for i = 1,..., k, (Eq. 5-7)
where k = total number of sites.

The resulting equation for fixed extra risk (BMR) is polynomial in dose (when logarithms
of both sides are taken) and can be solved in a straightforward manner for the combined BMD.
However, the current version of BMDS cannot estimate confidence bounds for this combined
BMD.

Therefore, a Bayesian approach to finding confidence bounds on the combined BMD was
implemented using Win BUGS (Spiegelhalter et al., 2003, 594261). Win BUGS software is freely
available and implements Markov Chain Monte Carlo (MCMC) computations. Use of
WinBUGS has been demonstrated for derivation of a distribution of BMDs for a single
multistage model (Kopylev et al., 2007, 194860) and is easily generalized (Kopylev et al., 2009,
198071) to derive the distribution of BMDs for the combined tumor load, following the NRC
(1994, 006424) methodology described above. The advantage of a Bayesian approach is that it
produces a distribution of BMDs that allows better characterization of statistical uncertainty. For
the current analysis, a diffuse (high variance or low tolerance) Gaussian prior restricted to be
nonnegative was used. The posterior distribution was based on three simulation chains with
50,000 burn-in (i.e., the initial 50,000 iterations were dropped) and a thinning rate of 20,
resulting in 150,000 interactions total. The median and 5th percentile of the posterior distribution
provided the BMD0i (central estimate) and BMDLm (lower bound) for combined tumor load,
respectively.

The methodology above was applied to the statistically significant dose-response data
from Kociba et al. (1978, 001818). NTP (1982, 594255). and NTP (2006, 543749) (see

This document is a draft for review purposes only and does not constitute Agency policy.

5-41 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

Section 2.3.2 for data set selection criteria).41 As with the risk estimates generated for individual
tumor sites, the combined analysis used the internal dose metric, whole blood concentration (see
Section 3). For the combined tumors for each sex/species combination, a BMDLoi in blood
concentrations was estimated. The BMDLoi (ng/kg) was then converted to a BMDLred
(ng/kg-day) using the Emond human model, and an OSF in units of (mg/kg-day) 1 was
calculated by, OSF = 0.01/BMDLHed x 106. Because of the nonlinearity of blood concentration
and ingested dose in the Emond Human PBPK model, the cancer risk is linear only with the
TCDD blood concentration and low TCDD oral ingestion doses, but is not linear with ingested
TCDD at higher doses; a single OSF cannot represent the entire range of risks for oral ingestion.
Thus, to use these estimates in human health risk assessment, risk-specific TCDD oral intake
levels corresponding to the target risk levels should be calculated using a procedure similar to
that for the slope factors based on epidemiologic data (see Table 5-3).

5.2.3.2.4. Results of dose-response modeling for rodent bioassays.

Table 5-15 presents the benchmark dose modeling results for both the individual tumors
and the combined tumors based on TCDD blood concentrations. The ^-values in the table are
for a chi-square goodness of fit statistic with significance ofp > 0.05. Goodness of fit was
acceptable at/? > 0.05 for all models. The difference in log likelihood (dLL) statistic documents
the difference in log likelihoods between stages of the models in cases where the stage is
above 1; it shows the difference between the stage in the table and the lower stage. For example,
for the NTP (2006, 543749) liver cholangiocarcinomas, twice the difference of 2.92 would be
>3.84, the test statistic from the assumed chi-square distribution,42 withp = 0.95, justifying the
choice of 3 stages over 2 stages. The best fitting multistage models include: a 1-stage (linear)
model for all of the individual tumor data sets from Kociba et al. (1978, 0018181 NTP (1982,
594255). and Toth et al. (1979, 197109). for liver carcinomas in females in Delia Porta et al.
(1987, 197405). as well as for the pancreatic and oral mucosa tumors in NTP (2006, 543749); a

41 Because only one turmor site was statistically significantly elevated in both the Delia Porta et al. (1987, 1974051
and Toth et al. (1979, 1971091 (i.e., only increased incideces of liver tumors were statistically significant elevated in
both studies), a multi-tumor analysis was not conducted.

42The chi-square distribution with 1 degree of freedom is the correct distribution only under standard conditions
(e.g., no boundary parameters in null hypothesis). Thus, the correct distribution for the situation where the
parameter of interest is on the boundary, as happens with testing for the order of the multistage model, and, possibly
nuisance parameters (estimated parameters of the model), is very difficult to derive (Self and Liang. 1987, 594398).
Therefore the p-valuc of chi-square with one degree of freedom is used as the best available choice.

This document is a draft for review purposes only and does not constitute Agency policy.

5-42 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

2-stage model for the lung tumors in NTP (2006, 543749) and for liver carcinomas in males from
Delia Porta et al. (1987, 197405); and a 3-stage model for the liver cholangiocarcinoma and liver
adenoma data sets from NTP (2006, 543749). The multi-stage model fit was not significant (p >
0.1) in the NTP (1982, 594255) study for lung tumors in the male mouse (p = 0.09), adrenal
cortex (p = 0.06)and thyroid follicular cell adenomas (p = 0.06) in male rats, and subcutaneous
tissue in female mice (p = 0.09), and was also not significant for liver carcinomas (p = 0.019) in
female mice in Delia Porta et al. (1987, 197405). For the Toth et al. (1979, 197109) liver
tumors, the model fit to all of the data was poor, and the highest dose group was dropped in order
to achieve an acceptable fit (p = 0.29). The BMDois and BMDLois (ng/kg) were estimated from
these multistage models for the individual tumors. BMDois and BMDLois (ng/kg) were also
provided in Table 5-15 for the combined tumors for each sex/species combination within a study.
These were estimated from the distributions of BMDois produced by the Bayesian MCMC
simulation (see Section 5.2.3.1.2.3.2). The BMDois and BMDLois (ng/kg) for the combined
tumors in Table 5-15 are the mean and lower 95% percentile values from these distributions,
respectively.

5.2.3.2.4.1. Individual tumor models.

Table 5-16 shows the BMDLredS (ng/kg-day) that were estimated from the BMDLois in
Table 5-15 using the Emond human model (see Section 5.2.3.1.2.2.2) and the OSFs calculated
by, OSF = 0.01/BMDLhed x 106 to convert the OSF to (mg/kg-day)-1 units. BMDS results,
details of the model fits and dose-response graphics for all endpoints are shown in Appendix F.
Although only the blood concentration results are presented in this section, for comparison
purposes, Appendix F also provides modeling results for the studies' administered average daily
doses. Table 5-16 lists the OSFs in decreasing value. It can be seen that liver tumors in male
mice yield the highest slope factors; OSF values are 5.9 xlO6 and 5.2 xlO6 per mg/kg-day in
NTP (1982, 594255) and Toth et al. (1979, 197109). respectively. The OSFs for the new NTP
(2006, 543749) study in female rats are two orders of magnitude lower, ranging from 1.8 xlO4 to
1.8 xlO5 per mg/kg-day, representing the lowest OSFs for TCDD from the individual tumor
models.

This document is a draft for review purposes only and does not constitute Agency policy.

5-43 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

5.2.3.2.4.2. Multiple tumor (Bavesian) models.

Table 5-17 shows the BMDLHedS (mg/kg-day) that were estimated from the BMDL0iS in
Table 5-15 using the Emond human model (see Section 5.2.3.1.2.2.2) and the OSFs calculated
by, OSF = 0.01/BMDLred x 106 to convert the OSF to (mg/kg-day)-1 units. Table 5-17 lists the
OSFs in decreasing value. It can be seen that the combined liver and lung tumors in male mice
yield the highest OSF value of 9.4 xlO6 per mg/kg-day from NTP (1982, 594255). and the
combined adrenal, tongue and nasal tumors in male rats yield the lowest OSF value of 3.2 x 105
from Kociba et al. (1978, 001818). The OSF for the combined liver, oral mucosa, lung, and
pancreatic tumors in female rats from the newer NTP (2006, 543749) study is 4.4 xlO5.

5.2.3.2.5. Summary evaluation of slope factor estimates from rodent bioassays.

To estimate a range of candidate TCDD OSFs from the animal data, dose-response
modeling of the five chronic rodent bioassays identified in Section 2.4.3 was conducted. Dose-
response modeling was performed using whole blood concentrations, and BMDLred values
(ng/kg-day) were derived for the 28 species/sex/endpoint data sets individually (see Table 5-16)
and for seven species/sex combined tumor data sets (see Table 5-17).

The highest OSFs that have been derived for these animal cancer bioassays using the
multistage models are from the multiple tumor analyses for NTP (1982, 594255; 2006, 543749)
and Kociba et al. (1978, 001818). presented in Table 5-17, and from the individual tumor
analyses for Toth et al. (1979, 197109) liver tumors and Delia Porta et al. (1987, 197405) liver
carcinomas in male mice, presented in Table 5-16. The most sensitive species and sex is male
mice, for which the estimated BMDLred for combined tumors is 1.1 x 10"3 ng/kg-day. This
result, which is derived under the assumption that multiple tumor types occur independently in
the exposed animals, is, as expected, lower than the BMDLred for the most sensitive individual
tumor.

Based on these results, EPA believes that a credible value for the BMDLred derived from
the animal studies lies in the range shown in Table 5-17 between 3.1 x 10"2 and
1.1 x 10 3 ng/kg-day. These values, which correspond to oral slope factor values of 3.2 x io5
and 9.4 x io6 per mg/kg-day, respectively, encompass the range at which elevated cancer risks
can be detected for the most sensitive species, sex, and endpoints in the animal bioassay data.

This document is a draft for review purposes only and does not constitute Agency policy.

5-44 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

As noted above in Sections 5.2.3.1.2.2 and 5.2.3.1.2.3, the cancer mortality risk is strictly
linear only with TCDD blood concentration, such that a single OSF cannot represent the entire
range of risks for oral ingestion. The OSFs shown in Tables 5-16 and 5-17 are based on HEDs
corresponding to the BMDLoi, which are most representative of lower human exposure levels,
including ambient exposures. For higher exposures, the risks increase at a slower rate with
increasing dose and the corresponding OSFs are lower; in those cases, risk-specific doses can be
calculated as previously described (see Section 5.2.3.2.3.2).

5.2.3.2.6. Qualitative uncertainties in slope factor estimates from rodent bioassays.

This section presents a qualitative discussion of the uncertainties associated with
calculating the OSF for TCDD from chronic animal bioassay data. Discussions on the feasibility
of conducting a quantitative uncertainty analysis for TCDD using dose-response methods are
provided in Section 6.4.2 of this document.

5.2.3.2.6.1. Quality of studies relied upon for determining POD.

EPA considers the overall quality and breadth of the studies used for the cancer dose-
response analysis to be excellent. All of the studies were published in the peer-reviewed
literature, and two of them were conducted by NTP (1982, 594255; 2006, 5437491
Kociba et al. (1978, 001818). Delia Porta et al. (1987, 197405) and Toth et al. (1979, 197109)
are older studies, but appear to have been conducted according to good laboratory practice
standards. The control and dose group sample sizes were relatively large, -40-50 animals or
more per group for all of the studies. All five studies exposed the test animals via the oral route
to TCDD alone, as was stipulated in EPA's study inclusion criteria. Collectively, these five
studies reported development of numerous cancer endpoints (tumors) in both sexes in two strains
of rats (Sprague-Dawley and Osborne-Mendel) and two strains of mice (i.e., B6C3Fi,
Swiss/H/Riop). The overall high quality of these studies and the strong, positive association
between TCDD exposure and cancer suggests that study quality is not a major contributing factor
to uncertainty in the risk estimates.

This document is a draft for review purposes only and does not constitute Agency policy.

5-45 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

5.2.3.2.6.2.	Interpretation of results from studies relied upon for determining POD.

As discussed in Section 3.4.3.2.1, questions arose about the interpretation of liver tumor
responses in female rats in the Kociba et al. (1978, 001818) study. Three re-evaluations of the
slides have been reported (Goodman and Sauer, 1992, 197667; Kociba et al., 1978, 001818;
Squire, 1980, 594272). The decision to use the Goodman and Sauer (1992, 197667) evaluation
was based on their use of the most current tumor classification procedures. The incidence of
hepatocellular adenomas and carcinomas (individually and combined), however, did vary
(sometimes widely) for each dose group across the three evaluations. Although the state-of-the-
science is reflected in the Goodman and Sauer analysis, there is some uncertainty in the
interpretation of any post-hoc analysis. No issues have arisen with regard to the interpretation of
the NTP (1982, 594255; 2006, 543749). Delia Porta et al. (1987, 197405) or Toth et al. (1979,
197109) tumor identification and classification.

5.2.3.2.6.3.	Consistency of results across chronic rodent bioassavs.

The existence of five high-quality chronic bioassays for TCDD increases confidence and
reduces uncertainty in the cancer OSFs. Considered together, these studies tested two species
and both sexes of mice and rats, and a wide range of well-characterized tumor types. All five
studies were consistent in observing increases (at some dose level) in rates of liver tumors (in
both species and sexes). While tumors at other sites were observed (and those sites varied across
study, species, and sex), the liver tumors were consistently the most sensitive indicators of
carcinogenic response (with respect to BMDLred estimates). Lung tumors were also
consistently observed across three of the studies, in male mice in the NTP (1982, 594255) study
and in female rats in Kociba et al. (1978, 001818) and NTP (2006, 543749). As discussed above,
the two most sensitive single-tumor endpoints as judged by BMDL0i values were associated with
elevated liver tumor risks, followed by lung, lymphoma or leukemia, thyroid and adrenal
cancers. The consistency of tumor types and sensitivities across endpoints and studies lends
confidence to the multistage modeling results.

5.2.3.2.6.4.	Human relevance of rodent tumor data.

There is some concordance in the tumor responses observed in the rodent test species and
humans, however, the most sensitive tumor site in the animals, the liver, has not been associated

This document is a draft for review purposes only and does not constitute Agency policy.

5-46 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

with cancer from TCDD exposures in humans. On the other hand, lung cancer and leukemia are
found both in the animal studies and in epidemiologic studies of exposed workers. The
consistency across sex, species, and strains in the animal studies suggests that the occurrence of
several of these tumors, in particular, liver and lung tumors is not an idiosyncratic response of a
particular combination of species, strain, or sex. As discussed in Section 5.2.1, the likely AhR
related carcinogenic mechanism is credible for humans as well as for rodent species.

5.2.3.2.6.5.	Relevance of rodent exposure scenario.

Three of the five chronic rodent bioassays exposed the test animals for ~2 years, the
majority of their lifespans. T oth et al. (1979, 197109) exposed the animals only for one year, but
they were kept on the study for a second year before they were evaluated for cancer. The Delia
Porta et al. (1987, 197405) study also exposed the test animals for one year, and a dosing error
occurred during the study. At ages 31 to 39 weeks, 41 male mice and 32 female mice in the
2,500 ng/kg BW dose group were mistakenly administered a single dose of 25,000 ng/kg BW
TCDD. TCDD treatment for the 2,500 ng/kg BW dose group was halted for 5 weeks (beginning
the week after the 25,000 ng/kg BW dose was administered in error) and resumed until exposure
was terminated at 57 weeks. Thus, the large single dose and subsequent period without TCDD
exposure confounds the dose-response relationship for this study. In general, these lifetime
bioassays in animals have long been used by EPA to assess potential lifetime exposures and
effects in humans. However, in the case of TCDD, the half life of TCDD in the body for rats,
mice, and humans is very different (see Section 3). Thus, there is a significant amount of
uncertainty in the use of rat and mouse data to develop OSFs for human cancer risk assessment
of TCDD.

5.2.3.2.6.6.	Impact of background TCDD exposures.

It is known that TCDD has been found in the feed used in animal bioassays, and that this
is a confounding factor, particularly in older studies. The effect of TCDD in the diets of test
species has the potential to be quite significant given the low levels of TCDD at which adverse
effects have been observed. Insofar as that is an issue, the risks associated with TCDD
exposures in the animal bioassays, and therefore the OSFs, would be biased high, which could be
the case for the NTP (1982, 594255). Delia Porta et al. (1987, 197405). Kociba et al. (1978,

This document is a draft for review purposes only and does not constitute Agency policy.

5-47 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

001818) and Toth et al. (1979, 197109) studies. The impact of this issue is that the newer study,
NTP (2006, 543749), accounted for TCDD exposures in the animal feed. Thus, there is likely to
be less uncertainty in the TCDD dose-response information presented in NTP (1982, 594255;
2006, 543749) than in the other four studies conducted before 1990.

5.2.3.2.6.7.	Choice of endpoint for POD derivation.

As noted above, the liver tumor PODs represent the most sensitive single-tumor endpoint
across the five cancer bioassays. Thus, the liver cancer endpoints must be seriously considered
for derivation of a TCDD OSF. As discussed in the previous section, EPA has also developed
Bayesian dose-response estimates for combined tumors, which yield BMDLoi values slightly
lower than those for any individual tumor type. Although it is the most conservative choice to
select the lowest combined tumor POD for OSF derivation, there are uncertainties associated
with the multiple tumor analysis. The assumption of independence of tumors across sites is
reasonable, particularly since the tumors from TCDD do not metastasize. However, the
independence assumption lacks hard evidence and needs further laboratory confirmation.

5.2.3.2.6.8.	Choice of animal-to-human extrapolation method.

The analyses presented here have used the Emond human kinetic model for extrapolating
dose from animals to humans (as discussed in Section 3.4.2). The rationale for this choice is that
the blood concentration metric most accurately reflects the concentration of TCDD in the various
tissues. As discussed in Section 3.4.3.2.4, use of the blood concentration dose metric results in
critical dose estimates (HEDs) that are considerably lower (10- to more than 100-fold) than those
derived based on administered dose. This does not reflect bias in the blood-based measure;
rather it is a reflection of the nonlinear biokinetics of TCDD in the body. EPA has also explored
the impacts of using other dose metrics, including AhR-bound TCDD concentration calculated
based on the Emond model. As discussed in Section 3.4.3.2.6.2, this also results in HED
estimates much lower than those obtained based on administered dose.

5.2.3.2.6.9.	Choice of model for POD and model uncertainty for POD derivation.

The bioassay-based cancer dose-response assessment in this section has used the
multistage model which is the standard model choice for such assessments and has been the basis

This document is a draft for review purposes only and does not constitute Agency policy.

5-48 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

for most of EPA's cancer risk assessments. The multistage model is the standard because it is
the only available model form that allows for low-dose linearity while accommodating
curvilinearity at higher doses and can be readily implemented.

There is some model choice uncertainty associated with instances of lack of fit. When
the multistage model does not adequately describe the observed pattern of responses (typically
determined by examining the p-walue for lack of fit), a decision must be made about possible
adjustments, including the dropping of higher dose groups thought to be less relevant to the
estimation of low-dose slopes. In this analysis, poorer fits (p-values less than 0.10) were
observed in five cases, four from NTP (1982, 594255) and one from Delia Porta et al. (1987,
197405) (see Table 5-15). The lowest BMDLoi associated a low p-w alue (p = 0.09) was for the
lung tumors in the NTP (1982, 594255) male mouse, the third lowest POD behind the liver
PODs in the individual tumor data sets. The other instances were for adrenal cortex and thyroid
follicular cell adenomas in male rats and for subcutaneous tissue in female mice in the NTP
(1982, 594255) study and for liver carcinomas in female mice in Delia Porta et al. (1987,
197405). In those instances, the/>-values were 0.06, 0.06, 0.09, and 0.019, respectively. These
poorly fit data sets provide OSF estimates that are uncertain and also contribute to uncertainty in
the combined tumor PODs from NTP (1982, 594255). The lowest BMDLoi in the combined
tumors is for the male mice combined liver and lung tumors, thus estimates from this sex/species
combination from NTP (1982, 594255) is highly uncertain and impacts its choice as a POD.

5.2.3.2.6.10. Statistical uncertainty in model fits.

Every model fit to a data set is associated with some inherent statistical uncertainty. For
this reason, bounds were calculated and used for OSF derivation (e.g., lower bounds on
benchmark doses, in this case the BMDLois). Those bounds account for uncertainties associated
with finite samples of test animals, both in terms of the number of dose groups and of the
number of animals per dose group. Valid and accepted statistical procedures have been applied
to ascertain the impact of those limitations on the estimates of interest. That being the case, the
statistical uncertainties associated with finite samples have been adequately addressed.

This document is a draft for review purposes only and does not constitute Agency policy.

5-49 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

5.2.3.2.6.11. Choice of risk level for POD derivation.

The BMR level that has been used for the POD in deriving the cancer OSF is one percent
extra risk. A single BMR was chosen for consistency across studies. Also, a BMR of 1% was
judged to be near the range of the observations. For the TCDD animal cancer bioassay data,
although many of the first positive tumor incidence responses (relative to controls) are closer to
10% (some higher), some are as low as 2%. Furthermore, most of the BMDoi values are within a
factor of 3 of the lowest tested dose, and the BMDLoi values are generally less than a factor of 2
below the BMD. Table 5-18 presents a comparison of BMDs, BMDLs and slope factors for 1%,
5% and 10% BMRs from the multi-tumor analyses of NTP (1982, 594255; 2006, 543749) and
Kociba et al. (1978, 001818) and for selected single tumor data sets from Toth et al. (1979,
197109) and Delia Porta et al. (1987, 197405). In Table 5-18, the choice of BMR has little or no
impact on the slope factors based on TCDD blood concentration for the combined or single
tumor incidences selected as representative of each study.43 In contrast, Table 5-19 presents a
comparison of Human Equivalent Dose BMDs, BMDLs and slope factors for 1, 5, and 10%
BMRs from these same datasets. Table 5-19 shows that, when converting the blood
concentration to the equivalent HED, a 2-fold to 4-fold decrease in the OSF is obtained when
using a BMR of 10% rather than 1%. This result is a consequence of the nonlinearity in the
Emond PBPK model at higher doses, where dose-dependent elimination of TCDD in the liver
results in a less-than-proportional increase in blood concentration relative to oral intake. At
lower exposure levels, blood concentration is proportional to oral intake. Therefore, EPA has
chosen the lower BMR of 1% as more representative of the low-dose risk.

5.2.3.3. EPA 'v Response to the NAS Comments on Choice of Response Level and

Characterization of the Statistical Confidence Around Low Dose Model Predictions

The NAS was concerned with the statistical power to determine the shape of the dose
response curve at low doses, well below observed dose-response information. EPA shares this
concern in that the shape of the dose-response curve in the low-dose region cannot be determined
with confidence when based on higher dose information.

43 This will generally be the case for multistage model fits with lst-degree coefficients greater than zero because the
response at the BMDL is virtually linear at BMRs of 10% or less. For model fits dominated by higher-order
coefficients, linearity of response at the BMDL begins at lower BMRs.

This document is a draft for review purposes only and does not constitute Agency policy.

5-50 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

When tumor data are used for dose-response modeling, a POD is obtained from the
modeled tumor incidences. When assessing carcinogenicity using a linear extrapolation
approach from a POD, a balance must be struck between staying within the range of the
observations and obtaining a representative estimate of the low-dose slope. Traditional cancer
bioassays, with approximately 50 animals per group, can typically support modeling down to an
increased incidence of 1-10%; epidemiologic studies, with larger sample sizes, below 1%. For
the TCDD animal cancer bioassay data, most of the low-dose tumor incidence responses are
under 10% (relative to controls), with some as low as 2%. For comparison purposes, BMDs,
BMDLs and OSFs from the animal cancer bioassay benchmark dose modeling assuming 1, 5,
and 10%) extra risk are shown in units of blood concentrations and human equivalent doses in
Tables 5-18 and 5-19, respectively. After evaluating the magnitude of the uncertainty in
BMDLoiS against the impact of using BMDLi0s, EPA has chosen to use a 1% BMR in all cases,
determining that the uncertainty bounds on the BMDLoi values are reasonable.

In the analysis of the animal cancer bioassays presented in this document, the multistage
cancer model was applied with a linear dose extrapolation to zero. EPA used a 1% excess risk
estimate, i.e., a BMDLoi, as the POD for development of candidate TCDD cancer oral slope
factors using a Bayesian multitumor approach (see Section 5.2.3.2. The advantage of a Bayesian
approach is that it produces a distribution of BMDs that allows better characterization of
statistical uncertainty.

Central tendency slope estimates and upper bound oral slope factor estimates are part of
the standard BMDS multistage cancer model and are included in each output file for the animal
bioassay single tumor analyses in Appendix F. Central tendency BMDs are also reported for the
results of the animal bioassay multitumor analysis (see Table 5-15). Central tendency slope
estimates are given for all the qualifying epidemiological studies as well (see Tables 5-1 and
5-4), where possible.

5.2.3.4. EPA 'v Response to the NAS Comments on Model Forms for Predicting Cancer Risks
Below the POD

The NAS offered extensive comments on the cancer dose-response modeling in the 2003
Reassessment. Although epidemiologic and rodent bioassay data are useful for the evaluation of
the dose-response curve within the range of the observed response data, they have traditionally

This document is a draft for review purposes only and does not constitute Agency policy.

5-51 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

not been useful sources of information for identifying a threshold or for estimating the shape of
the dose-response curve below the POD. Rather, mechanistic toxicological data have been the
evidentiary sources of choice for those types of analyses. As noted above, any quantitative
estimation of carcinogenic risk associated with TCDD exposure requires low-dose extrapolation
of experimental data. Unfortunately, the shape of the dose-response curve in the low dose region
is unknown.

Several of the analyses of epidemiological cohort data evaluated the fit of different dose-
response models to the data. Log-dose models accentuate the importance of low-dose low-
magnitude responses and can yield implausible results. The most relevant models used in these
studies are the untransformed-dose Cox regression models. Better results have been obtained in
the cohort analyses when the flattening of the hazard-ratio curve is taken into account. The latter
has been modeled explicitly by Steenland et al. (2001, 198589). who use a piece wise linear
model and implicitly by Cheng et al. (2006, 523122). who drop out a percentage of the high-dose
response data and fit a linear model to the remainder. Importantly, the analyses of the
epidemiologic cohorts presented in Section 5.2.3.1 are limited to evaluation and reanalyses of
published data as reported by the study authors. EPA does not have access to the raw data from
these epidemiologic studies and, therefore, could not conduct de novo analyses.

5.2.3.4.1. Choice of extrapolation approach
5.2.3.4.1.1. TCDD and receptor theory.

TCDD is considered to be a receptor-mediated carcinogen in animals. Nearly all TCDD
experimental data are consistent with the hypothesis that the binding of TCDD to the AhR is the
first step in a series of biochemical, cellular, and tissue changes that ultimately lead to toxic
responses observed in both experimental animals and humans (Part II, Chapter 2 of the 2003
Reassessment). Ligand-receptor binding, like any bimolecular interaction, obeys the law of mass
action as originally formulated by A.J. Clark (Limbird, 1996, 594276). The law of mass action
predicts the fractional receptor occupancy at equilibrium as a function of ligand concentration.
Fractional occupancy (Y) is defined as the fraction of all receptors that are bound to ligand:

y _ [TCDD - AhR ] _ [TCDD - AhR ] _ [TCDD

[AhR ]TOT [AhR ] + [TCDD - AhR ] [TCDD ] + Kd	^ 5_g)

This document is a draft for review purposes only and does not constitute Agency policy.

5-52 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

where [TCDD] is the concentration of the ligand, [AhR] is the concentration of the receptor and
[TCDD-AhR] is the amount of liganded receptor. The equilibrium dissociation constant Kd
describes the affinity of the interaction and is the concentration of TCDD that results in 50%
receptor occupancy. This simple equation defines a rectangular hyperbola, which is the
characteristic shape of the vast majority of biological dose-response relationships.

In certain cases, no response occurs even when there is some receptor occupancy. This
suggests that there may be a threshold phenomenon that reflects the biological "inertia" of the
response (Ariens et al., 1960, 594279). In other cases, a maximal response occurs well before all
receptors are occupied, a phenomenon that reflects receptor "reserve" (Stephenson, 1956,
594280). Therefore, the law of mass action cannot by itself fully explain the effect or response
observed after TCDD interacts with AhR. The ligand-receptor complex is associated with a
signal transduction or effector system. In the case of the AhR, this effector system can be
considered to be the transcriptional machinery itself. The key feature of this formulation is that a
response is proportional, or a function of, the number of receptors occupied.

Furthermore, for a ligand such as TCDD that elicits multiple receptor-mediated effects,
one cannot assume that the binding-response relationship for a simple effect (such as enzyme
induction) will necessarily be identical to that for a different and more complex effect (such as
cancer). The cellular cascades of events leading to different complex responses (e.g., altered
immune function, developmental effects, or cancer) are different, and other rate-limiting events
likely influence the final biological outcome resulting in different dose-response curves. Thus,
even though TCDD binding to AhR is assumed to be the initial event leading to a spectrum of
biological responses, TCDD-AhR binding data may not always correlate with the dose-response
relationship observed for particular effects.

A receptor-based mechanism would predict that, except in cases where the concentration
of TCDD is already high (i.e., [TCDD]~Kd), incremental exposure to TCDD will lead to some
increase in the fractional occupancy of AhR. However, as discussed above, it cannot be assumed
that an increase in receptor occupancy will necessarily elicit a proportional increase in all
biological response(s), because numerous molecular events contributing to the biological
endpoint are integrated into the overall response. That is, the final biological response could be
considered as an integration of a series of interdependent dose-response curves with each curve

dependent on the molecular dosimetry for each particular step. Dose-response relationships that

This document is a draft for review purposes only and does not constitute Agency policy.

5-53 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

will be specific for each endpoint must be considered when using mathematical models to
estimate the risk associated with exposure to TCDD. It remains a challenge to develop models
that incorporate all the complexities associated with each biological response as the modes of
action for various toxicological endpoints appear to vary greatly. For TCDD, extensive
experimental data from studies using animal and human tissues indicate that cell- or tissue-
specific factors determine the quantitative relationship between receptor occupancy and the
ultimate biological response. This would suggest that the parameters for each mathematical
model might only apply to a single biological response within a given tissue and species, making
extrapolation to other systems challenging.

5.2.3.4.1.2. Low-dose extrapolation: threshold or no threshold?

As indicated in the 2005 Cancer Guidelines,44 toxicity reference values for human
noncancer endpoints have historically been estimated based on a no-observed-adverse-effect
level (NOAEL) or lowest-observed-adverse-effect level (LOAEL) from animal bioassay studies.
This terminology suggests a biological population threshold beneath which no harm is
anticipated. Reference values such as the oral reference dose (RfD) or inhalation reference
concentration (RfC) are derived by applying uncertainty factors (UFs) to a POD. Depending on
the nature of available data and modeling choice, a POD can be selected from values other than
an NOAEL or LOAEL, such as an EDX, or a benchmark dose (BMD) or its BMDL. An RfD is
described as "likely to be without appreciable risk" but the probabilistic language has not as yet
been operationalized. There is no quantitative definition of "appreciable" and no mechanism to
compute risk as a function of dose, so as to ascertain that the risk is indeed not appreciable. The
risk at the RfD is not calculated, and it cannot be calculated within the current UF framework.
Instead, a hazard quotient is computed as the ratio of a given exposure to the RfD, or a margin of
exposure is estimated as the ratio of the POD to the human exposure level.

Cancer endpoints are predominantly thought to have no population biological threshold.
Although the terminology "threshold/nonthreshold" is still common in cancer dose-response

44As stated in the 2005 Cancer Guidelines (U.S. EPA. 2005, 086237): "For effects other than cancer, reference
values have been described as being based on the assumption of biological thresholds. The Agency's more current
guidelines for these effects (U.S. EPA. 1996, 594399: U.S. EPA. 1998, 0300211 however, do not use this
assumption, citing the difficulty of empirically distinguishing a true threshold from a dose-response curve that is
nonlinear at low doses."

This document is a draft for review purposes only and does not constitute Agency policy.

5-54 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

discussions, the 2005 Cancer Guidelines propose a different terminology, whereby "nonlinear
models" are those whose dose-response slope is zero at or above zero. In the natural language,
and indeed in data analysis, it is difficult to distinguish the following situations:

•	The response approaches zero as dose goes to zero, versus

•	The response slope goes to zero as dose goes to zero (nonlinear model).

This use of "nonlinear" is acknowledged to be idiosyncratic.45 The NAS review (NAS,
2006, 19844H does not consistently apply the terminology from the 2005 Cancer Guidelines, nor
does it consistently distinguish the above two circumstances: ".. .the observed data are more
consistent with a sublinear response that approaches zero at low doses rather than a linear dose
response" (NAS, 2006, 198441). The point of a nonlinear model in the sense of the 2005 Cancer
Guidelines is that the response slope approaches zero. Both linear and nonlinear responses
approach zero at low dose (in the absence of background). Since the terms "linear," "sublinear,"
and "nonlinear" invite confusion in this context, the following terminology is used in this
document:

Threshold Model. There is some threshold dose T > 0 such that the probability of
response for any dose less than or equal to T is zero, and the probability is nonzero for
any dose greater than T.

Linear/ Linear above Threshold Model. For the linear model, the probability of response
is proportional to the dose. For the linear over threshold model, the probability of
response is zero for a dose below the threshold, and it is proportional to the excess dose
over the threshold otherwise. Note that under the EPA cancer guidelines, the linear
above threshold model is classified as a nonlinear model.

Nonlinear Model. Any model that is not linear.

Supralinear/ Supralinear above Threshold Model. For the supralinear model, the slope of
the probability of response decreases as dose increases; in other words, the second
derivative of the response curve is negative. For the supralinear above threshold model,

45

From the 2005 Cancer Guidelines (U.S. EPA. 2005, 086237): "The term 'nonlinear' is used here in a narrower
sense than its usual meaning in the field of mathematical modeling. In these cancer guidelines, the term 'nonlinear'
refers to threshold models (which show no response over a range of low doses that include zero) and some
nonthreshold models (e.g., a quadratic model, which shows some response at all doses above zero). In these cancer

guidelines, a nonlinear model is one whose slope is zero at (and perhaps above) a dose of zero	Use of nonlinear

approaches does not imply a biological threshold dose below which the response is zero."

This document is a draft for review purposes only and does not constitute Agency policy.

5-55 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

the second derivative is negative above the threshold, and the response probability is zero
below the threshold.

Sublinear/Sublinear above Threshold Model. For the sublinear model, the slope of the
probability of response increases as dose increases; in other words, the second derivative
of the response curve is positive. For the sublinear above threshold model, the second
derivative is positive above the threshold, and the response probability is zero below the
threshold.

Zero Slope at Zero Model. The slope of the response curve is zero at or above dose zero.

All of these models may be understood in an individual or population sense. According
to the 2005 Cancer Guidelines, the trigger for applying the basic RfD methodology for cancer
endpoints is sufficient evidence for the "zero slope at zero" model for the population. By
definition, any sublinear, supralinear, or linear model above the threshold i s a zero slope at zero
("ZS@Z") model.

The relation between individual and population models is not immediately evident.

Figure 5-4 shows dose-response curves of the probability of response vs. dose for different
models dose-response shapes. The left panel in Figure 5-4 shows a supralinear dose-response
curve; the rate of increase of the response probability goes down as dose increases, or in the strict
mathematical sense, the second derivative is negative. The middle panel shows a sublinear dose-
response curve; the second derivative is positive. In this case the slope at zero is zero (ZS@Z).
However, sublinearity, in the strict mathematical sense, by itself does not imply that the slope at
zero is zero. The probit dose-response model shown in the right graph is sublinear and has
positive slope at zero (the log-probit model is zero slope at zero).

If individuals in a population have different dose-response curves, then the population
dose-response curve is obtained by averaging all these dose-response curves over the population.
The shape of the population dose-response curve will generally be quite different from the
individual curves. Figure 5-5 is a simple depiction of the relationship of individual vs.
population dose response. The left panel in Figure 5-5 shows dose-response curves for seven
individuals, each with a supralinear dose-response curve above individual-specific thresholds.
Averaging these curves gives the dashed dose-response curve, which is nearly linear. The graph
on the right is similar, except that the individual dose-response curves are linear above individual
thresholds. The population curve is quadratic and zero slope at zero applies.

This document is a draft for review purposes only and does not constitute Agency policy.

5-56 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

Of course these are not the only possibilities; in general, the population dose-response
curve depends on (1) the distribution of individual thresholds in the neighborhood of zero, (2) the
dose-response curve for each individual, and (3) the dose metric. Under EPA's Cancer
Guidelines, the zero-slope-at-zero criterion applies strictly to ingested dose, but the other two
factors (distribution of individual thresholds and dose-response curve for each individual) need
to be established before a zero slope at zero dose can be established. Otherwise the default linear
extrapolation to zero approach applies.

On the nature or the distribution of individual thresholds, often referred to as the
population tolerance distribution, there is ongoing debate as to how receptor kinetics influence
the shape of that distribution. Even within an individual, there is a lack of consensus as to
whether receptor kinetics confer linear or sublinear attributes to downstream events, or whether
receptor kinetics, themselves, are linear, sublinear, or supralinear. Whatever the nature of the
form of receptor kinetics, it may have little or no influence on the ultimate population response.
The kinetics of receptors is in the domain of the individual, rather than the population. As
described previously, receptor kinetics are governed by the law of mass action, which leads to a
low-dose proportional response model, generally modeled by some form of Hill function, the
low-dose linear form being Michaelis-Menten kinetics. There is no a priori reason to believe
that the shape of the dose-response curve in an individual has any relationship to the shape of the
population response, particularly for quantal endpoints. Lutz and Gay lor (2008, 594297) present
an argument for considering the population response in terms of the more traditional tolerance
distribution, which is likely the result of more variable factors than the shape of receptor kinetics.
Perhaps more to the point, receptor activation is only the first of many events in the path to the
apical event (a tumor in this example). Because there are undoubtedly numerous additional
downstream events that must occur before the apical effect is observed, there are many
opportunities for interindividual variability to become manifest in the tolerance distribution.

Even at the first step, a more likely contributor to interindividual variability than the shape of the
response is the dose resulting in the response, as measured by the ED50 (Km in the Michaelis-
Menten formulation), which shifts the response curve. Factors that influence shifts in response
curves are generally modeled as normal or log-normal distributions and may confer a log-normal
shape on the population tolerance distribution, particularly if there are a number of dependent
sequential steps or distinct subpopulations (Hattis and Burmaster, 1994, 594301; Hattis et al.,

This document is a draft for review purposes only and does not constitute Agency policy.

5-57 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

1999, 594299; Lutz, 1999, 594298), although other distributions could be equally likely (Crump
etal., 2010. 380192).

To see how the discussion over threshold/nonthreshold might play out for TCDD,
consider the equilibrium dissociation constant Kd for TCDD, which measures the binding affinity
of TCDD to the AhR. Lower values indicate higher binding affinity and (other things being
equal) greater risk. For Han/Wistar rats, the value Kd = 3.9 is reported (standard deviation not
given); human values are reported as Kd = 9.6 ±7.8 (0.3 - 38.8 with 15 of 67 donors without
detectable binding) (Connor and Ay 1 ward, 2006, 197632). If AhR binding is the rate-limiting
step for carcinogenesis, then the majority of a human population may be less susceptible than
Han/Wistar rats, whereas a population threshold, if it exists, might still be well below the
Han/Wistar rat threshold, given the large variability in the human Kd estimate (see also Section
6.4.2.9). The NAS contends that an AhR-mediated mode of action indicates a threshold dose-
response relation (NAS, 2006, 198441). Presumably, the value of the threshold, if it exists,
depends on the AhR binding affinity. Arguing for a population threshold in this case requires
two types of information:

1.	The distribution of the individual thresholds induced by, among other things, the
individual Rvalues; and

2.	The dose-response function for values above the threshold induced by Kd.

Without this information, the shape of the population dose-response curve cannot be
determined with any confidence and the default linear relationship applies; response probability
is modeled as a linear function of dose, for dose near zero. However, from the 2005 Cancer
Guidelines: "When adequate data on mode of action provide sufficient evidence to support a
nonlinear mode of action for the general population (emphasis added) and/or any subpopulations
of concern, a different approach—a reference dose/reference concentration that assumes that
nonlinearity—is used." In current terminology, the reference dose methodology applies if there
is sufficient evidence supporting a "zero slope at zero" model; otherwise, the linear nonthreshold
model applies by default.

In principle, the choice between the above models could fall within the purview of dose-
response modeling. However, standard statistical methods encounter well-known difficulties in

This document is a draft for review purposes only and does not constitute Agency policy.

5-58 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

detecting thresholds. Without going into detail, suffice to say that the maximum likelihood
estimate of response probability when no responses are observed in a finite sample is always
zero. That said, some researchers have attempted to identify thresholds (Aylward et al., 2003,
594305; Mackie et al., 2003, 594303) or nonlinearity (Hoel and Portier, 1994, 198741) by means
of parameter estimation of appropriate models. A review of 344 rodent bioassays on 315
chemicals led to the following conclusion by Hoel and Portier (1994, 198741):

We have also found that the oft-held belief that genotoxic compounds typically
follow a linear dose-response pattern and that nongenotoxic compounds follow a
nonlinear or threshold dose response pattern is not supported by the data. In fact
we find the opposite with genotoxic compounds differing from linearity more
often than nongenotoxic compounds.

The choice between a linear and "zero slope at zero" model in current practice does not
fall under dose-response model fitting, it is made on the basis of a structured narrative as set
forth in the 2005 Cancer Guidelines (U.S. EPA, 2005, 086237):

In the absence of sufficiently, scientifically justifiable mode of action information,
EPA generally takes public health-protective, default positions regarding the
interpretation of toxicologic and epidemiologic data: animal tumor findings are
judged to be relevant to humans, and cancer risks are assumed to conform with
low dose linearity. ... The linear approach is used when: (1) there is an absence of
sufficient information on modes of action or (2) the mode of action information
indicates that the dose-response curve at low dose is or is expected to be linear.

Where alternative approaches have significant biological support, and no
scientific consensus favors a single approach, an assessment may present results
using alternative approaches. A nonlinear approach can be used to develop a
reference dose or a reference concentration.

5.2.3.4.1.3. Extrapolation method.

The 2005 Cancer Guidelines (U.S. EPA, 2005, 086237) emphasize that the method used
to characterize and quantify cancer risk from a chemical is determined by what is known about
the MOA of the carcinogen and the shape of the cancer dose-response curve.

The NAS was critical of EPA's decision to apply linear low-dose extrapolation for
TCDD cancer assessment in the 2003 Reassessment and encouraged the use of a nonlinear
approach. The 2005 Cancer Guidelines state that a nonlinear approach should be used when

This document is a draft for review purposes only and does not constitute Agency policy.

5-59 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

"there are sufficient data to ascertain the mode of action and conclude that it is not linear at low
doses and the agent does not demonstrate mutagenic or other activity consistent with linearity at
low doses."

Receptor modeling theory (as outlined in the 2003 Reassessment, Part II, Chapter 8)
indicates that exogenous compounds which operate through receptor binding mechanisms, such
as TCDD, will follow a linear dose-response binding in the 1-10% receptor occupancy region.
This theory has been supported by empirical findings and suggests that the proximal biochemical
effects (such as enzyme induction) and transcriptional reactions for TCDD may also follow
linear dose-response kinetics. More distal toxic effects could take any one of multiple forms
(i.e., linear, sublinear, supralinear or threshold) depending on (1) the toxic mechanism;
(2) location on the dose-response curve; and (3) interactions with other processes such as
intracellular protein binding and cofactor induction/repression.

In the case of TCDD, many adverse effects experienced at low exposure levels have too
much data variability to distinguish on a statistical basis (goodness-of-fit) between dose-response
curve options, and whether the dose-response is linear, sublinear or supralinear. For tumor
responses, with the exception of squamous cell carcinoma of the oral mucosa and adenomas or
carcinomas of the pancreas, which were fit with a linear multistage model, the tumor endpoints
in the NTP (2006, 543749) study using female Sprague-Dawley (S-D) rats are all best fit with a
sublinear model (i.e., the multistage model fits to tumor incidence data were second or third
degree; see Table 5-15 and Appendix F). For all tumor incidence data from three of the other
cancer bioassays that met the study inclusion criteria (Kociba et al., 1978, 001818; NTP, 1982,
594255; Toth et al., 1979, 197109). the multistage model fit was linear (first degree), when based
on either administered dose or modeled blood concentrations (see Appendix F). For Delia Porta
et al. (1987, 197405). the female liver carcinomas were linear (first degree), but the female liver
adenomas and the male liver carcinomas were best modeled using a second degree model (see
Table 5-15).

Another issue of potential importance when evaluating the shape of the dose-response
curve for low dose effects is the concept of "interacting background." The concept of interacting
background refers to a pathological process in the exposed population that shares a causal
intermediate with the toxicant being evaluated. On this issue, a recent NAS committee (NAS,
2009, 594307) contended that

This document is a draft for review purposes only and does not constitute Agency policy.

5-60 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

.. .the current EPA practice of determining "nonlinear" MO As does not account
for mechanistic factors that can create linearity at low dose. The dose-response
relationship can be linear at a low dose when an exposure contributes to an
existing disease process Crump et al., 1976, 003192; Lutz, 1990, 000399. Effects
of exposures that add to background processes and background endogenous and
exogenous exposures can lack a threshold if a baseline level of dysfunction occurs
without the toxicant and the toxicant adds to or augments the background process.

Thus, even small doses may have a relevant biologic effect. That may be difficult
to measure because of background noise in the system but may be addressed
through dose-response modeling procedures. Human variability with respect to
the individual thresholds for a nongenotoxic cancer mechanism can result in
linear dose-response relationships in the population (Lutz, 2001, 053426; NAS,
2009, 594307.

AhR activation could be considered a causal intermediate in several disease processes.
Recent studies have linked AhR activation in the absence of exogenous ligand to a multitude of
biological effects, ranging from control of mammary tumorigenesis to regulation of
autoimmunity (reviewed in Hahn et al., 2009, 548725). While the level of background activation
of AhR by endogenous compounds (or exogenous compounds other than TCDD) in the human
population is unknown, given the ubiquitous nature of several of the known endogenous and
exogenous AhR ligands, it is reasonable to assume that a certain baseline level of AhR activation
exists in the population. The degree to which TCDD exposure augments this baseline level of
AhR activation is unknown.

The 2005 Cancer Guidelines (U.S. EPA, 2005, 086237) recommend that the method used
to characterize and quantify cancer risk from a chemical be determined by what is known about
the mode of action of the compound and the shape of the cancer dose-response curve. The linear
approach is used if there is sufficient evidence supporting linearity or if the mode of action is not
understood (U.S. EPA, 2005, 086237). In the case of TCDD, (1) the mode of action of TCDD-
induced carcinogenesis beyond potential AhR activation is unknown; (2) information is lacking
to determine the shape of the dose-response curves at low doses for various adverse endpoints
(including cancer) in humans or experimental animals; (3) there is undoubtedly a certain level of
interacting background (i.e., AhR activation by endogenous ligands) in the human population;
(4) many of the rodent cancer dose-response relationships (Kociba et al., 1978, 001818; NTP,
1982, 594255; Toth et al., 1979, 197109) are consistent with low-dose linearity (first degree
multistage model fit) when based on either administered dose or modeled blood concentrations;

This document is a draft for review purposes only and does not constitute Agency policy.

5-61 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

and (5) higher human interindividual variability compared to experimental rodents will tend to
shift the shape of the dose-response towards linear (relative to rodents). None of these
suggestions of linearity, however, is conclusive (see next section for additional detail). The true
shape of the dose-response curve remains unknown. Therefore, in the absence of sufficient
evidence to the contrary or evidence to support nonlinearity, to estimate human carcinogenic risk
associated with TCDD exposure EPA assumed a linear low-dose extrapolation approach.

5.2.3.4.1.4. Discussion of low-dose linearity.

Any quantitative estimation of carcinogenic risk associated with TCDD exposure requires
low-dose extrapolation of high dose experimental and epidemiologic data. Unfortunately,
despite the availability of the extensive database on the biological effects of TCDD, the shape of
the dose-response curve in the low-dose region is not known. This situation is not unique to
TCDD. For most carcinogens the available biological data do not provide sufficient mechanistic
information to determine the shape of the dose-response relationship at doses below the levels
where direct experimental or epidemiologic data are available. EPA's Guidelines for Carcinogen
Risk Assessment (2005, 086237) recognize this situation and describe approaches the Agency
uses for dose response assessment in cancer risk assessments depending on the available
scientific database. EPA's basic approach makes a distinction between "low-dose linear" and
"nonlinear" dose response patterns. This distinction is important to understand as it addresses
the potential response at low dose, not the empirical pattern of response seen in the available
(often high dose) tumor data. To put matters simply, under a low-dose-linear model, the
estimated risk due to the carcinogen exposure is approximately proportional to the dose received
(at low dose). In mathematical terms, a low-dose-linear model is one whose slope is greater than
zero at a dose of zero (U.S. EPA, 2005, 086237; footnote, p. 1-11). Importantly, a low-dose-
linear model need not be linear at higher doses, and this is consistent with upward curving
responses (e.g., linear-quadratic) and downward curving (plateauing) responses that may be seen
various cancer studies. In EPA's terminology a "nonlinear" dose-response, refers to situations
where there is not a linear component in the response at low-dose. In this context, a "nonlinear
model" is one whose slope is zero at (and perhaps above) a dose of zero (ibid). Nonlinear
response patterns can include threshold models where there is no response below a defined dose

This document is a draft for review purposes only and does not constitute Agency policy.

5-62 DRAFT—DO NOT CITE OR QUOTE

-------
1	level, or other patterns where response at low dose otherwise decreases rapidly as compared to a

2	low-dose-linear model.

3	As stated in the previous section, the low-dose linear approach for the TCDD

4	carcinogenicity assessment in this document is based on EPA's scientific baseline inference

5	("default") regarding dose-response modeling. EPA believes that the mode of action is not

6	known, so is using the default linear extrapolation approach specified by EPA's cancer

7	guidelines.

8	Nonetheless, there are biological data on TCDD that help inform the appropriateness of

9	low-dose-linear risk extrapolation for this compound. Furthermore, there is utility in

10	summarizing scientific reasoning that supports the approach of low-dose linearity as an

11	appropriate scientific baseline inference ("default") for carcinogen risk assessment.

12	The issues pertaining to low-dose linearity were discussed in the report of a recent state-

13	of-the-science workshop on issues in low-dose risk extrapolation held by U.S. EPA and Johns

14	Hopkins Risk Science and Public Policy Institute in 2007 (White et al., 2009, 622764). This

15	report states:

16

17	The complex molecular and cellular events that underlie the actions of agents that

18	lead to cancer and noncancer outcomes are likely to be both linear and nonlinear.

19	At the human population level, however, biological and statistical attributes tend

20	to smooth and linearize the dose-response relationship, obscuring thresholds that

21	might exist for individuals. Most notable of these attributes are population

22	variability, additivity to preexisting disease or disease processes, and background

23	exposure-induced disease processes; measurement error also undoubtedly

24	contributes to this phenomenon. The linear appearance of the population-level

25	dose-response function does not presume that the dose-response relationship is

26	necessarily linear for individuals (Lutz, 1990, 000399; 2001, 053426; Lutz et al.,

27	2005, 087763). but may reflect a distribution of individual thresholds. These

28	attributes are likely to explain, at least in part, why exposure-response models of

29	the relationship between cancer or noncancer health effects and exposure to

30	environmental toxicants with relatively robust human health effects databases at

31	ambient concentrations (e.g., ozone and particulate matter air pollution, lead,

32	secondhand tobacco smoke, radiation) do not exhibit evident thresholds, even

33	though the MO As include nonlinear processes for key events NRC (2005);

34	U.S. EPA (2006, 088089; 2006, 157071; 2006, 090110); U.S. DHHS (2004,

35	056384).

36

This document is a draft for review purposes only and does not constitute Agency policy.

5-63 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

Original arguments in favor of low-dose linearity for carcinogen risk assessment
(including for ionizing radiation, as developed from human data) are based on the occurrence of
damage (often termed "hits") to DNA and the inference that resulting mutations would
contribute to cancer development. These arguments envisioned direct damage to DNA;
however, based on subsequent advances in mechanistic understanding, damage to DNA by
"secondary" reactive molecules (not just direct hits to DNA by radiation or other agents) is also
considered to play a major role. TCDD is not thought to produce DNA damage directly.
However, DNA damage may result subsequent to increased formation of reactive molecules
(reactive oxygen species (ROS) and metabolites of endogenous compounds). Thus, the presence
of low-dose linearity by this pathway would depend on whether such reactive molecules were
produced at low dose and whether that increased formation was proportional to dose. If that
were the case for TCDD, which is still unknown, arguments in favor of low-dose linearity
remain similar to those for direct-acting agents.

The kinetics of ligand receptor binding, and then the attachment of a receptor/ligand
complex to a promoter region of DNA are biochemical processes where low-dose linearity can
occur. Simple receptor binding interactions are often modeled using Michaelis-Menten
relationships which are linear at low dose. Thus, the early key events in a process of a receptor-
mediated toxicity pathway may often be expected to be low-dose linear. However, as in any
toxicity process, the ultimate shape of the dose-response relationship for an apical46 toxicity
endpoint will depend on all the processes involved, not just receptor kinetics. These issues were
considered by NRC (NAS, 2009, 594307) which included as an indication for non-threshold
dose response: "The fact that in receptor-mediated events, even at very low doses a chemical can
occupy receptor sites and theoretically perturb cell functions (such as signal transduction or gene
expression) or predispose the cell to other toxicants that bind to or modulate the receptor systems
(such as organochlorines and the aryl hydrocarbon receptor or endocrine disruptors and
hormonal binding sites)." The role of these factors for TCDD has not been fully elucidated.

Two other factors supporting low-dose linearity discussed in the workshop described by
White et al. (2009, 622764) are additivity to background processes (dose additivity) and the
magnitude of human heterogeneity.

46 An apical endpoint is an observable outcome in a whole organism, such as a clinical sign or pathologic state, that
is indicative of a disease state that can result from exposure to a toxicant (NAS, 2007).

This document is a draft for review purposes only and does not constitute Agency policy.

5-64 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Concerning dose additi vity. Crump et al. (1976, 003192) argued in the context of a
carcinogenic response that if the carcinogenic process resulting from exposure to an exogenous
agent (e.g., TCDD) is already operant in causing background responses, then the effect of the
exposure is to augment this process in a dose-additive fashion. The additional response caused
by the exposure is expected to increase approximately linearly with exposure at low exposures
(i.e., be low-dose linear). The NRC Science and Decisions report (NAS, 2009, 594307)
examined the issue of additivity to background, in particular calling attention to a need for
systematic consideration of endogenous processes related to disease development as well as
additivity to other exogenous exposures.47 While the baseline activity (unexposed to exogenous
agents) of AhR is not well understood, the effects of exogenous agents need to be considered in
terms of how they add on to or modulate baseline physiological processes instead of considering
TCDD or other exogenous ligands to be "acting in a vacuum."

The issue of human heterogeneity relative to the rodents used in bioassays has been
discussed at length in the literature and will not be repeated here (see also relevant text in
Section 5.2.3.4.1.3). However, as discussed by NAS (2009, 594307). even in situations where
processes thought to be nonlinear are precursors to the development of cancer in test animals, a
different situation may result in humans: "However, given the high prevalence of those
background processes, and given the multitude of chemical exposure and high variability in
human susceptibility, the results may still be manifested as low-dose linear dose-response
relationships in the human population." The population dose-response will be influenced by
heterogeneities in the population that affect internal dose as well as response. First, even if there
is strong curvilinearity in the dose-response curve in the dose range of relevance to human
exposures, there may be large differences across individuals in the doses at which transitions in
the shape of the dose-response curve occur. Greater variability in response to exposures would
be anticipated in heterogeneous populations than in inbred laboratory species under controlled
conditions (due to, e.g., genetic variability, disease status, age, and nutrition). The effect of
increased heterogeneity will be a broadening of the dose-response curve (i.e., less rapid fall-off
of response with decreasing dose) in diverse human populations and, accordingly, a greater

47 It may be noted that when there are multiple exogenous exposures, it may be difficult to ascertain which exposure
came first. However, the point is that if a combination of endogenous and exogenous factors is operative in causing
biological response, then an additional small, dose additive, exposure can be predicted to cause a proportionate
change in response.

This document is a draft for review purposes only and does not constitute Agency policy.

5-65 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

potential for risks from low-dose exposures (Lutz et al., 2005, 087763; Zeise et al., 1987,
060867). The degree to which heterogeneity must be increased to "linearize" sublinear
responses of varying degrees has not yet been established.

Interpreting the shape of animal bioassay dose-response model fits always involves
assumptions about the shape of the response in the unobserved range (i.e., low dose). Cancer
bioassays can provide relatively little information on actual dose-response patterns below the
point of departure. However, it is generally not possible to either exclude or affirm low-dose
linear components statistically based upon empirical modeling of the dose-response data.48
Dose-response modeling can, however, be useful in describing the size of a linear component in
the response that is compatible with study data. As an example, NRC (NAS, 2006, 198441)
advised EPA to examine the results of the NTP (2006, 543749) study as indicating nonlinearity
of the observed tumor response. Among the tumors seen in the NTP bioassay, the dose-response
shape for cholangiosarcoma is notably curvilinear in the dose range of the observed tumor
response. Figure 5-6 shows the multistage modeling of the cholangiosarcoma data from the NTP
bioassay. The BMDL is calculated at an extra risk of 0.01. Even though the MLE dose response
is nonlinear (lst-degree coefficient is zero), the dose-response curve pertaining to the statistical
upper bound on risk (calculated here as the 95% lower confidence bound on dose) is
approximately linear below the 0.01 benchmark level and roughly superposes on the EPA default
linear extrapolation (see Figure 5-6B). For the oral squamous cell carcinoma (SCC) tumor data
(plot not shown), the MLE dose-response curve itself displays low-dose linearity (lst-degree
coefficient is greater than zero) and the EPA low-dose linear extrapolation is indistinguishable
from the upper bound curve. These observations are consistent with the findings of
Subramaniam et al. (2006), that for the large majority of chemicals, straight line extrapolation of
risk from the BMDL provides slope factor values very similar to those obtained by using an
upper bound on the multistage model risk estimate. Furthermore, in this assessment, EPA has
chosen to derive oral slope factors based on combined tumor incidence whenever possible,
modeling them under an assumption of independence. A Bayesian analysis is used in this
document to develop PODs based on combined tumor risk across the significantly elevated
tumor types observed in this bioassay (see Section 5.2.3.2.3.2). As a result of this analysis, the

48 EPA policy is to allow for low-dose linearity in the modeling of tumors if a non-linear MOA has not been
established.

This document is a draft for review purposes only and does not constitute Agency policy.

5-66 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

central estimate for the composite dose-response curve shows little curvilinearity and the MLE
dose-response curve is substantially linear below a 0.1 extra risk level (see Figure 5-7A and
5-7B; see also Section 5.2.3.2.6.11).

The results here provide a comparison of EPA's linear (straight line) dose-response
estimates with the degree of linearity seen in the fitted dose-response curves and the statistical
upper bounds on these curves. To do this the fitted model needs to allow for the possibility of
both curvilinearity at high dose and linearity at low dose. The multistage model has these
properties, which is among its advantages for application in carcinogen risk assessment. Most
other models commonly used to fit data in the observed range do not have this property.49

One other issue relative to the determination of linearity arises in the visual interpretation
of dose-response plots. The common practice of plotting receptor kinetics data on semi-
logarithmic plots for scale convenience has unfortunately led to difficulties in the interpretation
of the shape of these relationships. An example is presented using the modeling study of Kohn
and Mel nick (2002, 199104). which was cited by NRC (NAS, 2006, 198441) in its review of
EPA's dioxin assessment as an example of nonlinear behavior at low dose: "Response is a
function of the number of occupied and activated receptors, which typically exhibit steep dose-
response relationships. For example, Kohn and Mel nick (2002, 199104) modeled the shape of
the dose-response relationship for receptor-mediated responses, using the estrogen receptor and
various xenoestrogens as a model receptor and ligands, respectively. The model included a
variety of assumptions with regard to receptor number, ligand binding affinity, and partial
agonist activities, yet in every instance clear sublinear responses were observed at low doses."
However, as shown in Figure 5-8, the apparent strong upward curvature of the low-dose
relationship is no longer seen when the results are plotted on an arithmetic scale. Instead, the
system may be seen as providing an example of close to linear behavior in the low-dose region.

49 The standard Hill models do not: A Hill model is only linear at low dose when the Hill parameter is equal to 1
(and in that case the Hill model is linear over the full dose range until the high dose region of "saturation" where the
km parameter results in downward curvature). Thus, while the Hill model is a valuable tool for fitting data in the
observed experimental range, it is not helpful in illustrating the potential for low-dose response. However, some
have considered a dose-additive version of the Hill model which would allow for low-dose linearity.

This document is a draft for review purposes only and does not constitute Agency policy.

5-67 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

5.2.3.4.1.5. Consideration of nonlinear methods.

While the 2005 Cancer Guidelines deem linear extrapolation to be most appropriate for
TCDD, EPA has carefully considered the NAS recommendation to provide risk estimates using
both linear and nonlinear methods.

The 2005 Cancer Guidelines state

For cases where the tumors arise through a nonlinear mode of action, an oral
reference dose or an inhalation reference concentration, or both, should be
developed in accordance with EPA's established practice for developing such
values ... This approach expands the past focus of such reference values
(previously reserved for effects other than cancer) to include carcinogenic effects
determined to have a nonlinear mode of action.

In this section, EPA presents two illustrative examples of RfD development for
carcinogenic effects of TCDD. Each of these examples focuses on data derived from animal
bioassays as described in Section 2.4.2.

5.2.3.4.1.5.1. Illustrative RfDs based on tumorigenesis in experimental animals.

TCDD has been shown to be a multisite carcinogen in both sexes of several species of
experimental animals. It also has been shown to be carcinogenic to humans. Most of the
available quantitative human epidemiologic data related to TCDD carcinogenesis are for all
cancer mortality. Mortality is a frank effect and is generally considered to be inappropriate for
RfD development, therefore, the illustrative example below utilizes available evidence from
experimental animals. Table 5-20 presents candidate PODs and RfDs for TCDD carcinogenicity
based on combined tumor responses from the animal bioassays described in Section 2.4.2. The
PODs from the NTP (2006, 549255; 2006, 543749) and Kociba et al. (1978, 0018m animal
studies were derived from Bayesian multitumor dose-response modeling (as described in
Section 5.2.3.2, Table 5-17) using a BMR of 1%. Because only TCDD-induced liver tumors
were reported by Toth et al. (1979, 197109). the BMR of 1% (POD) from that study was
generated using a first degree linear multistage model (see Table 5-15). TCDD-induced liver
tumors were reported by Delia Porta et al. (1987, 197405). with the male mouse producing the
lowest BMR of 1% (POD) using a second degree linear multistage model (see Table 5-15).
Following BMD modeling, BMDLhedS were then estimated (see Tables 5-16 and 5-17) using the

This document is a draft for review purposes only and does not constitute Agency policy.

5-68 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

TCDD whole-blood-concentration dose metric from the Emond model as described in Section 3.
The illustrative RfDs were derived by dividing the BMDLredS by appropriate uncertainty
factors. In each instance, a total UF of 30 was applied, comprising factors of 3 for the
toxicodynamic component of the interspecies extrapolation factor (UFA) and a factor of 10 for
human interindividual variability (UFH).

As shown in Table 5-20, the illustrative RfDs for TCDD-induced tumors range from
3.6E-1 1 for liver and lung tumors in male mice (NTP, 1982, 594255) to 1.0E-9 for adrenal
cortex, tongue and nasal/palate tumors in male rats (Kociba et al., 1978, 001818). This
illustrative RfD range for TCDD tumorigenesis falls within the range of candidate RfDs for
noncancer TCDD effects presented in Table 4-5.

5.2.3.4.1.5.2. Illustrative RfDs based on hypothesized key events in TCDD's MO As for liver
and lung tumors.

As described in Section 5.1, most evidence suggests that the majority of toxic effects of
TCDD are mediated by interaction with the AhR. EPA considers interaction with the AhR to be
a necessary, but not sufficient, event in TCDD carcinogenesis. The sequence of key events
following binding of TCDD to the AhR and that ultimately leads to the development of cancer is
unknown. While the mode of action of TCDD in producing cancer has not been elucidated for
any tumor type, the best characterized carcinogenic actions of TCDD are in rodent liver, lung,
and thyroid. The hypothesized sequence of events following TCDD interaction with the AhR is
markedly different for each of these three tumor types. Additionally, no detailed hypothesized
mode of action information exists for any of the other reported tumor types.

The endpoints selected for this illustration were evaluated to provide insight into the
quantitative relationships between tumor development and precursor events in TCDD-induced
carcinogenesis. The endpoints described below may or may not be biologically adverse in
themselves; the intent herein was to consider TCDD-induced biochemical and cellular changes
that could lead to subsequent tumor development.

In the following exercise, illustrative RfDs were derived for key events in TCDD's
hypothesized modes of action in the liver and lung. No appropriate dose-response data were
identified for key events in TCDD's hypothesized MOA for thyroid tumors in a

This document is a draft for review purposes only and does not constitute Agency policy.

5-69 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

sex/species/strain that has been shown to develop thyroid tumors (i.e., female B6C3F1 mice and
male and female Osborne-Mendel rats (NTP, 1982, 594255)).

As this is an illustrative exercise only, only studies that were originally identified in
Section 2 for potential noncancer dose-response modeling were evaluated here (see Section 2.4.2
for study details). There may be additional studies available in the literature that would further
inform the dose-response assessment of these endpoints.

Additionally, for animal model consistency, only results from studies conducted in
female S-D rats are presented here. The majority of the available information on TCDD
carcinogenicity (and TCDD carcinogenic precursor events) comes from studies conducted in
female S-D rats and the most recent TCDD carcinogenicity study was conducted in female S-D
rats (NTP, 2006, 197605). While both Kociba et al. (1978, 001818) and NTP (2006, 543749)
have conducted TCDD carcinogenicity studies in female S-D rats, different substrains were used;
this difference in substrain may have resulted in the different carcinogenic responses reported
from the two studies. While the carcinogenicity of TCDD in female S-D rats has been well
characterized, this animal model does not exhibit the full suite of tumor responses reported for
TCDD (for instance, female S-D rats have not been shown to develop thyroid tumors).
Additionally, the most sensitive single tumor response in female S-D rats from NTP (2006,
543749) is squamous cell carcinoma of the oral mucosa (see Section 5.2.3.2), a tumor type for
which no mode of action information exists. Therefore, the illustrative RfDs described below
may not be protective against all tumor types.

For each endpoint, PODs for illustrative cancer RfD development were identified as
described for the noncancer RfD derivation in Section 4. Briefly, for the endpoints identified
below, the NOAELreds and/or LOAELreds were determined based on EPA analysis of the
original data presented by the study author (see Section 2.4.2 for details) and by application of
the Emond PBPK models as described in Section 3.3.4. BMDLreds were determined as
described in Section 4.2 for all data sets amenable to BMD modeling. Modeling outputs for the
endpoints are presented in Appendices E and G as noted in Table 5-21. The illustrative RfDs
were derived by dividing the POD by appropriate uncertainty factors as indicated in Table 5-21.
5.2.3.4.1.5.2.1. Liver tumors.

Figure 5-9 presents one hypothesized mode of action for TCDD-induced liver tumors in
rats. TCDD activation of the AhR leads to a variety of changes in gene expression, including

This document is a draft for review purposes only and does not constitute Agency policy.

5-70 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

increased CYP1A1 mRNA and subsequent increases in CYP1A1 activity. These alterations in
gene expression are hypothesized to lead to hepatotoxicity, followed by compensatory
regenerative cellular proliferation and subsequent tumor development. The details of the
mechanism of TCDD-induced hepatotoxicity have not been fully determined but both CYP
induction and oxidative stress have been postulated to be involved (Maronpot et al., 1993,
198386; Viluksela et al., 2000, 198968). Additionally, oxidative DNA damage has been
implicated in liver tumor promotion (Umemura et al., 1999, 1980011 The enhanced cell
proliferation arising from either altered gene expression or hepatotoxicity, or both, could be the
principal factor leading to promotion of hepatocellular tumors (Whysner and Williams, 1996,
197556).

A dose-response relationship exists for TCDD-mediated hepatotoxicity, and this parallels
the dose-response relationship for tumor formation (or formation of foci of cellular alteration as a
surrogate of tumor formation). However, the dose-response relationship for other
TCCD-induced responses such as enhanced gene expression is different from the dose-response
for tumor formation in terms of both efficacy and potency (see Popp et al. (2006, 197074) for
review).

A representative endpoint for each of the hypothesized key events following AhR
activation for TCDD-induced liver tumors was identified and is shown in Figure 5-9. Illustrative
RfDs based on each representative endpoint are shown in Table 5-21.

5.2.3.4.1.5.2.2. Lung tumors.

Far less is known about TCDD's mode of action in the lung. Figure 5-10 presents two
hypothesized modes of action for TCDD-induced lung tumors in rats. The first hypothesized
mode of action of TCDD in the lung involves disruption of retinoid homeostasis in the liver.
Retinoic acids and their corresponding nuclear receptors, the RARs and the RXRs, work together
to regulate cell growth, differentiation, and apoptosis. It is hypothesized that TCDD, through
activation of the AhR, can affect parts of the complex retinoid system and/or other signaling
systems regulated by, and/or cross-talking with, the retinoid system (reviewed in (Nilsson and
Hakansson, 2002, 548746)). These effects are then hypothesized to lead to lung tumor
development, however the mechanisms underlying this hypothesis are not well-defined. The
second hypothesized mechanism for the carcinogenic action of TCDD in the lung is through

This document is a draft for review purposes only and does not constitute Agency policy.

5-71 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

induction of metabolic enzymes. Through activation of AhR and subsequent induction of
metabolizing enzymes (such as CYP1A1), TCDD may enhance bioactivation of other
carcinogens in lung (Tritscher et al., 2000, 197265). However, there are few studies to support
this hypothesis.

Representative endpoints could only be identified for two of the hypothesized key events
following AhR activation for TCDD-induced lung tumors. These endpoints are presented in
Figure 5-10. Illustrative RfDs based on each of these two representative endpoints are shown in
Table 5-21. There is insufficient information to form any conclusions on the quantitative
progression to tumorigenicity or on the relative protection afforded by preventing the key events
shown.

5.2.3.4.1.5.2.3. Limitations of illustrative RfDs based on hypothesized key events in TCDD's
MO As for liver and lung tumors.

A trend for increasing RfD values that follows the progression of endpoints towards the
production of tumors is evident. However, there are a number of factors that prevent making
strong conclusions based on this exercise. These limitations include the following

•	This example addresses only two tumor types in one species, strain and sex (female S-D
rats), with little information available on the hypothesized mode of action for lung
tumors. No mode of action information is available for the most sensitive tumor type in
this animal model (squamous cell carcinoma of the oral mucosa). Therefore, it is
possible that the illustrative RfDs presented in this example would not be protective
against all tumor types in female S-D rats. Importantly, other animal models have been
shown to be more sensitive to TCDD-induced carcinogenesis based on combined tumor
analysis (see Section 5.2.3.2); an RfD based on tumorigenesis in this animal model may
not be protective against tumorigenesis in other, more sensitive, animal models (or, by
extension, in humans).

•	Several of the BMDLs are based on poorly-fitting models, such that the RfD is based on
a LOAEL (or LOEL), which is not a particularly good measure for comparison across
endpoints (e.g., LOAELs are dependent on dose spacing in bioassays). Furthermore, the
hepatotoxicity BMDL based on a dichotomous 10% BMR, is not directly comparable to
all the other BMDLs based on a continuous 1 standard-deviation BMR (Crump, 2002,
035681). In addition, as the earlier effects (CYP induction, cellular proliferation) are not
considered to be necessarily adverse in themselves, the BMR of 1 standard-deviation
from the mean may not be the best choice for determining a POD based on biological
signficance. The use of the 1 standard-deviation BMR for the illustrative examples is
primarily for comparison on an equal-magnitude-of-response basis across endpoints.

This document is a draft for review purposes only and does not constitute Agency policy.

5-72 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

•	The endpoints selected as representative of each hypothesized key event may not be the
most appropriate choices. These particular endpoints were chosen because they were the
most sensitive indicator (i.e., lowest POD) from the available data or were the only
available choice based on a lack of data for other effects related to the hypothesized key
event.

•	The optimum timing of these events may not be reflected in the endpoints selected.
Almost certainly, changes in gene expression are early events, such that a single
exposure should be relevant, as in the mRNA changes reported after a single TCDD
exposure (Vanden Heuvel et al., 1994, 594318). although it is not known whether the
magnitude of these changes would be altered after longer-term exposure, or whether
longer-term exposure would be more relevant to downstream events. Similarly, single
exposures for induction of CYP enzymes would seem to relevant as a measure of the
immediate effect, but it may be longer-term repeated CYP activity that is important for
longer-term downstream events; Table 5-21 shows a nominal order-of-magnitude
difference in effect levels for similar effect magnitudes (ca. 20-fold) from single
exposures (Kitchin and Woods, 1979, 198750) and long-term exposures (53-weeks;
NTP, 2006, 543749). The relevant exposure durations for oxidative stress and later
effects are longer term, so a measurement of oxidative stress at 90-days in a rodent may
be appropriate; Wyde et al. (2001, 198575) suggest that induction of 8-oxo-dG DNA
adducts are a result of longer-term oxidative stress because of the lack of effect of single
exposures. Hepatotoxicity and hepatocellular proliferation events would appear at
successively later times, but the effective exposure levels would depend heavily on the
endpoints chosen to represent those events and the time at which they were measured.
The toxic hepatopathy endpoint reported in NTP (2006, 543749). is a general measure of
mild to moderate liver toxicity, but is measured only at the end of the study when tumors
have already appeared. Hepatocyte hypertrophy, measured at 31 weeks may be more
duration-relevant, but may not indicate actual hepatocellular toxicity.

•	The lowest of the tested doses may well be much higher, given that all animal diets are
contaminated to a certain extent by TCDD, resulting in initial TCDD body burdens in all
animals. Vanden Heuvel et al. (1994, 594318) reported TCDD liver concentrations in
control animals almost as high as for the low-dose group, which could equate to a
significant increase in the actual exposure experienced by the low-dose group. A similar
effect on the low-dose group (0.45 ng/kg) in Kitchin and Woods (1979, 198750) is
possible, although they did not report control animal tissue concentrations. Higher
exposure levels or longer-term exposures would not be affected to the same degree, as
administered TCDD levels would likely be large compared to initial body burden or low-
level feed stock exposure.

Given the limitations described above, establishing an unambiguous progression of
effects is extremely problematic given the lack of sufficient data. Identifying a RfD that could
be considered to be protective against tumorigenesis in humans based on these data and models
is subject not only to the determination of effective low doses for the RfDs in Table 5-21 but also

This document is a draft for review purposes only and does not constitute Agency policy.

5-73 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

to the determination of effective exposures that could be considered to be protective of all other
tumor types in female S-D rats as well as all other animal models. The latter would entail
identifying precursors that are sufficient in themselves for progression to tumorigenesis for all
tumor types. Given the disparate sequence of hypothesized key events following TCDD-induced
AhR activation for the tumor types for which some information is available, AhR
binding/activation is the only key event that is likely to be shared across tumor types. No
appropriate quantitative data on AhRbinding/activation by TCDD in relevant animal models
were located; therefore, an illustrative RfD based on TCDD AhR activation could not be
developed.

Simon et al. (2009, 594321) present a similar analysis for the liver tumors observed in the

NTP (2006, 		_|) study, showing a progression of effects from early biochemical events to

irreversible liver toxicity, culminating in tumorigenesis. While illustrative of the putative tumor-
promoting MOA for TCDD, the limitations of using such an approach within the context of an
assessment of the overall carcinogenic risk of TCDD as detailed above still apply. Simon and
colleagues also present RfDs for liver tumors and several precursor endpoints. All the RfDs
presented in Simon et al. (2009, 594321) are essentially equivalent and are 1 to 3 orders of
magnitude higher than the RfDs for equivalent endpoints presented in Table 5-21. These
discrepancies are partly due to the fact that the Emond PBPK models (Emond et al., 2004,
197315; Emond et al., 2005, 197317; Emond et al., 2006, 197316; see also Section 3.3.4) used in
this document predicts lower TCDD intakes for similar tissue concentrations than the CADM
kinetic model (Ay 1 ward et al., 2005, 197014; Carrier et al., 1995, 197618) used by Simon and
colleagues. However, a larger contributor to these discrepancies is the use of a chemical-specific
adjustment factor (CSAF) of 0.1 for the toxicodynamic component of the interspecies
uncertainty factor by Simon et al. (2009, 594321). while EPA used an uncertainty factor of 3.
EPA does not find that the in vitro evidence presented by Simon et al. in support of a CSAF of
0.1 for interspecies toxicodynamics meets the burden of proof necessary for a reduction in this
uncertainty factor.

This document is a draft for review purposes only and does not constitute Agency policy.

5-74 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

5.3. DERIVATION OF THE TCDD ORAL SLOPE FACTOR AND CANCER RISK
ESTIMATES

EPA was able to derive candidate OSFs for all cancer mortality from human
epidemiologic studies as well as for individual and combined tumor incidence from rodent
cancer bioassays. Each of these studies was selected for TCDD dose-response modeling using
the study inclusion criteria outlined in Section 2. The derivation of these OSFs can be found for
the epidemiologic data in Section 5.2.3.1 and for the rodent bioassay data in Section 5.2.3.2.

The OSFs based on epidemiologic studies from three cohorts ranged from 3.75 x io5 to
2.5 x 106 per mg/kg-day (see Tables 5-1 and 5-3). For the animal data, OSFs based on
individual tumors were developed for 28 study/sex/endpoint combinations, and the results ranged
from 1.8 x 104 to 5.8 x 106 per mg/kg-day (see Table 5-16). The OSFs based on combined
tumors were developed for 7 study/sex combinations, and the results ranged from 3.2 x 105 to
9.4 x io6 per mg/kg-day (see Table 5-17). Figure 5-11 demonstrates the range of these OSFs in
units of per mg/kg-day. The human study OSFs are shown at the far left of the figure, and the
rodent endpoints are arranged by species to the right. For comparison with the other studies, the
OSF from Cheng et al. (2006, 523122) is based on a 1 10~6 risk level (Table 5-3).

As recommended by expert panelists at EPA's 2009 Dioxin Workshop (U.S. EPA, 2009,
522927) and in the 2005 Cancer Guidelines (U.S. EPA, 2005, 086237). EPA has chosen to give
higher consideration to the human epidemiologic data rather than the animal bioassay data in
developing an OSF for TCDD. Candidate OSFs derived from the human data are consistent with
the animal bioassay OSFs; specifically, the human OSFs fall within the same range as the animal
bioassay OSFs. Because all the human and animal studies were considered to be of high quality
and yielded similar ranges of OSFs, EPA has chosen to rely on the epidemiologic data for OSF
derivation.

The strengths and limitations of the five epidemiological studies meeting the inclusion
criteria for cancer dose-response modeling are summarized in Table 5-22. Among the human
studies, the occupational TCDD exposures in the NIOSH and Hamburg cohorts are assumed to
be reasonably constant over the duration of occupational exposure. In contrast, the TCDD
exposure patterns in the Seveso and BASF cohorts are associated with industrial accidents; as a
consequence, the exposure patterns are acute, high dose followed by low-level background
exposure. Such exposure patterns similar to those experienced by the BASF and Seveso cohorts

This document is a draft for review purposes only and does not constitute Agency policy.

5-75 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

have been shown to yield higher estimates of risk when compared to constant exposure scenarios
with similar total exposure magnitudes (Kim et al., 2003, 199146; Murdoch and Krewski, 1988,
548718; Murdoch et al., 1992, 548719). Thus, EPA has judged that the NIOSH and Hamburg
cohort response data are more relevant than the BASF and Seveso data for assessing cancer risks
from continuous ambient TCDD exposure in the general population.

The NIOSH (Cheng et al., 2006, 523122; Steenland et al., 2001, 198589) and Hamburg
(Becher et al., 1998, 197173s) cohort studies report cumulative TCDD levels in the serum for
cohort members. The most significant difference among the Cheng et al. (2006, 523122)
analysis and those of Steenland et al. (2001, 198589) and Becher et al. (1998, 197173) is the
method used to back-extrapolate exposure concentrations based on serum TCDD measurements.
Steenland et al. (2001, 198589) and Becher et al. (1998, 197173) back-extrapolated exposures
and body burdens using a first-order model with a constant half-life. In contrast, Cheng et al.
(2006, 523122) back-extrapolated body burdens using a kinetic modeling approach that
incorporated concentration- and age-dependent elimination kinetics.

Although all three of these are high-quality studies, the kinetic modeling used by Cheng
et al. (2006, 523122) is judged to better reflect TCDD pharmacokinetics, as currently
understood, than the first-order models used by Steenland et al. (2001, 198589) and Becher et al.
(1998, 197173). EPA believes that the representation of physiological processes provided by
Cheng et al. (2006, 523122) is more realistic than the assumption of simple first-order kinetics
and this outweighs the attendant modeling uncertainties. Furthermore, the use of kinetic
modeling is consistent with recommendations both by the NAS and the Dioxin Workshop panel.

However, as discussed in Section 3.3.2, the kinetic model that they employed does have
certain limitations, including the fact that it has been calibrated based on a relatively small
number of human subjects. In addition, their kinetic model does not allow body mass index
(BMI; and hence fat content) to vary with age, which may bias the model results. Nonetheless,
EPA prefers the increased technical sophistication of the dose estimates used in the cancer
mortality risk estimates derived from Cheng et al. (2006, 523122) to those derived from
Steenland et al. (2001, 198589).

EPA, therefore, has decided to use the results of the Cheng et al. (2006, 523122)
study for derivation of the TCDD OSF based on total cancer mortality as calculated by
EPA using data and models from the Cheng et al. (2006, 523122) study as described in

This document is a draft for review purposes only and does not constitute Agency policy.

5-76 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

Section 5.2.3.1.2. Although the OSF is only strictly defined for exposures above the
background exposure experienced by the NIOSH cohort, which was assumed to be 0.5
pg/kg-day TCDD, or 5 pg/kg-day total TEQ, EPA assumes that the slope (risk vs. blood
concentration) is the same below those background exposure levels as it is above. Table 5-3
shows the oral slope factors at specific target risk levels (OSFRLs) which range from
1.1 x 10s to 1.3 x 106 per (mg/kg-day). EPA recommends the use of an OSF of 1 x 106 per

	£		q

(mg/kg-day) when the target risk range is 10 to 10 . Although EPA prefers the human
data, EPA also presents a number of OSFs derived from rodent bioassays. Most of these
animal studies are of note, because in general they were well-designed and conducted. In
particular, the NTP (2006, 543749) study was recently conducted and represents the most
comprehensive evaluation of TCDD chronic rodent toxicity to date.

5.3.1. Uncertainty in Estimation of Oral Slope Factors from Human Studies

A fair amount of uncertainty is associated with the estimation of slope factor values and
cancer risk specific doses for TCDD based on the epidemiological studies. In some instances,
the influence of a given factor is theoretically amenable to analysis, but such investigation is
limited by the availability of sufficiently detailed data to support such an analysis. In other
cases, only very broad ranges can be placed on the uncertainty associated with a given feature of
the analysis, or uncertainties must be discussed qualitatively.

The following four sources of uncertainty are addressed in this section: uncertainty in
exposure estimates in the epidemiologic studies (see Section 5.3.1.1), uncertainty in the shape of
the dose-response curve (see Section 5.3.1.2), uncertainty in extrapolating risks below exposure
levels in the reference population (see Section 5.3.1.3), uncertainty in cancer risk estimates
arising from background DLC exposure (see Section 5.3.1.4) and uncertainty in cancer risk
estimates arising from occupational coexposures to DLCs (see Section 5.3.1.5). Section 5.3.2
explores other sources of uncertainty in the epidemiologic risk estimates including the use of
cancer mortality rather than cancer incidence data in the derivation of the oral slope factor,
possible influences of inter-individual variability in TCDD kinetics, and exposures to other
occupational carcinogens.

This document is a draft for review purposes only and does not constitute Agency policy.

5-77 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

5.3.1.1. Uncertainty in Exposure Estimation

The major technical challenge within each of the epidemiological studies was developing
relevant and precise estimates of exposure. While Warner et al.(2002, 197489) collected blood
samples relatively close to the time of the Sevesso accident and could reasonably estimate peak
exposures based on these collected samples, in the case of the Becher et al. (1998, 197173). Ott
and Zober (1996, 198408). Steenland et al. (2001, 198589). and Cheng et al. (2006, 523122)
studies, the major exposure issues included the following

•	Selecting (an) appropriate dose metric(s) for dose-response modeling,

•	Estimating serum TCDD levels for the entire cohort based on measurements from a
smaller number of the subjects in the cohort collected long after the occupational
exposures had occurred, and then assigning exposures to the remaining members of the
cohort based on qualitative job classifications.

•	Estimating time-weighted average tissue doses (e.g., lipid-average serum concentration
over time) based on single samples taken at one point in time. (Except for the Becher et
al. (1998, 197173) analysis where one of the study strengths was their estimate of TCDD
half life, which utilized repeated measurements from a subset of their cohort).

In the Becher et al. (1998, 197173), Steenland et al. (2001, 198589), and Cheng et al.
(2006, 523122) studies, dose-response modeling was performed using ppt-years 1 ipid-adjusted
serum concentration as the primary dose metric for TCDD; serum TCDD was the only direct
measurement of exposure or dose that was available. In addition, as discussed in Section 3.3.4,
serum concentration is a reasonable index of total tissue concentration (target organ dose), and
lipid-adjusted serum concentration provides a reasonable index of TCDD in the fatty components
of tissues. Ott and Zober (1996, 198408) used ng/kg body weight at the time of the accident as
the primary dose metric, and U.S. EPA (2003, 537122) later converted these to units of ppt-years
lipid-adjusted serum concentration.

The decision to use cumulative serum concentrations (ppt-years) as the primary dose
metric for carcinogenicity is based on the understanding that time weighted concentrations (over
a chronic exposure period) are the most appropriate dose measures for cancer risk assessment.
This may not be strictly true if cancer induction by TCDD is considered to be a "threshold
process." However, as discussed in Section 5.2, there are reasonable grounds to believe that the

This document is a draft for review purposes only and does not constitute Agency policy.

5-78 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

assumption of low-dose linearity is reasonable for TCDD, especially when calculating
population risks where the effects of interindividual variability must be taken into account.

In addition to the issue of low-dose thresholds, the rationale for using cumulative dose
metrics also can fail at high doses if the adverse response in question involves a step that is
saturable (e.g., where there is a maximum level of response that cannot be exceeded owing to a
rate-limited process). There is some evidence for such a phenomenon in the NIOSH cohort
where cancer risks in the highest exposure group (>50,000 ppt-years) appear to saturate, and the
response decreases at this level (Steenland et al., 2001, 198589). Steenland et al. (2001, 198589)
suggest that the apparent saturation of dose-response in this cohort may be due, at least partially,
to exposure misclassification among the highest exposed individuals, rather than to an actual
reduction in response per unit exposure.

The uncertainty associated with differences in the exposure patterns is important to
consider across the five epidemiologic studies. Steenland et al. (2001, 198589). Cheng et al.
(2006, 523122). and Becher et al. (1998, 197173) studied cohorts exposed to elevated TCDD
levels over a long period of time, while Ott and Zober (1996, 198408) and Warner et al.(2002,
197489) studied cohorts exposed to TCDD levels significantly above background at one point in
time but the exposures and likely the TCDD body burdens declined significantly following these
periods of elevated exposure. Both these chronic and acute exposures can be analyzed in terms
of cumulative exposure to TCDD. Use of such a metric requires an assumption that the "actual"
cancer potency associated with a cumulative dose where much of the dose is received at a single
point in time and then gradually eliminated would be similar to the cancer potency of the same
cumulative dose received over a longer period of time and also gradually eliminated. While EPA
believes that such an assumption is not unreasonable, the experiment of Kim et al. (2003,
199146). which showed statistically significant increase in liver effects due to a peak TCDD
dose when compared to chronically-dosed Sprague-Dawley rats administered the same levels of
TCDD when measured as a cumulative dose, suggests that additional analyses of cumulative and
peak TCDD dose measures may need to be conducted.

There are uncertainties associated with the approaches used to estimate TCDD exposures
in the members of the occupational epidemiologic studies for which no measurement data were
available. To impute TCDD levels for workers without measured samples, all four occupational
epidemiologic studies matched workers for whom measured TCDD samples had never been

This document is a draft for review purposes only and does not constitute Agency policy.

5-79 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

reported to workers with measured TCDD levels based on job histories. The NIOSH cohort is
used to illustrate some of the uncertainties. In the NIOSH cohort, the subset of workers (roughly
5% of the total cohort) with blood serum data comprised surviving members of the cohort (in
1988), and therefore, their age distribution likely differed from the rest of the cohort. For each
worker in this subset, the following data were available: (1) job classification information,
(2) employment history, and (3) serum TCDD measures. All of the workers in this subset were
employed at a single plant where the work histories were less detailed than at other plants, and
many of the workers at this plant had the same job title and were employed during the same
calendar period. There is an assumption that workers with same job title and work history were
exposed to the same TCDD levels within a plant and across plants; this obviously does not
account for exposure heterogeneity.

Both Steenland et al. (2001, 198589) and Cheng et al. (2006, 523122) addressed the
potential for exposure measurement error in TCDD estimates and possible exposure
misclassification. For the highest exposure workers, Steenland et al. (2001, 198589) and Cheng
et al. (2006, 523122) found weak, "noisy," and/or negative exposure-response relationships.
Steenland et al. (2001, 198589) suggests that possible explanations for this observation include
the saturation of effects at the upper end of the dose-response curve, instability of the TCDD
exposure estimates based on the limited number of highly exposed individuals, and the increased
probability of exposure misclassification for workers whose job histories indicate the highest
exposures. As Steenland et al. (2001, 198589) reported, some of the highest exposures might
have been inaccurately estimated because they occurred in workers exposed to short-term, high-
dose exposures during spill clean-up. Cheng et al. (2006, 523122) used sensitivity analyses to
examine this measurement error issue and evaluated the potential for exposure misclassification
by using ln-transformed TCDD ppt-years. The authors removed all observations with exposures
within the lower and upper 1, 2.5, or 5th percentiles of the TCDD ppt-year distribution and also
removed observations within just the upper 1, 2.5, or 5th percentile of TCDD ppt-years. These
sensitivity analyses yielded results similar to those reported in the primary analysis. An
additional concern is that exposure errors might distort the exposure distribution in the
population, which generally spreads the response out over a wider dose range. This serves to
increase the variance of the regression model, altering both the POD and the corresponding OSF.

This document is a draft for review purposes only and does not constitute Agency policy.

5-80 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

Becher et al. (1998, 197173) only considered workers from a single plant but their
analysis included workers employed in five different job locations within the plant. The
influence of worker location on slope factor estimates does not appear to be further explored and
may represent a source of uncertainty.

To estimate long-term body burden metrics from the serum TCDD measurements,
Steenland et al. (2001, 198589) employed simple first order kinetic elimination rate model with a
half-life of 8.7 years. Limitations of this approach include (1) the average elimination half-life
among the study subjects may not be 8.7 years given differences between the study population
and the Ranch Hand population from which the value was estimated, (2) use of a single-value
estimate fails to take into account the inherent variability in elimination half life among the
individual workers, and (3) it fails to take into account variations in elimination kinetics
throughout the lifetime of the exposed worker due to change in body fat, age, etc. The impact of
these potential sources of bias on the estimates of time-integrated body burden cannot be
quantitatively assessed. However, Steenland et al. (2001, 198589) noted that modest changes in
elimination half-life (to 7.1 years) had only a very small impact on risk estimates.

Cheng et al. (2006, 523122) estimated past body burdens using the CADM approach
(described in Section 3) (Aylward et al., 2005a, b) rather than a half-life estimate. As noted
above, the incorporation of concentration- and age-dependent elimination into this approach has
significant advantages over the use of a constant elimination half-life. However, as discussed in
Section 3.3, the CADM has only been subject to limited testing against human validation data
sets, so the degree to which its advantages are realized in practice cannot be easily assessed.

There are no available human data in the low dose region, the region of interest to this
assessment, to compare with the CADM (or Emond) model predictions.

Becher et al. (1998, 197173) developed half life estimates based on multiple TCDD
blood measures in 48 individuals from this cohort. These half life estimates were then used to
back calculate TCDD concentrations at the end of each worker's employment, accounting for
age and percentage of body fat. This cohort-specific information may provide a better exposure
estimate than Steenland et al. (2001, 198589) or Ott and Zober (1996, 198408) who used similar
kinetic approaches. However, the comparison of the accuracy of the exposure estimates across
the cohorts is not easily assessed. There are several assumptions and important uncertainties
involved in modeling TCDD exposures in these cohorts. The study authors have invoked

This document is a draft for review purposes only and does not constitute Agency policy.

5-81 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

different kinetic assumptions when extrapolating measured levels of TCDD in sera backward in
time to estimate higher chronic or peak dosage (i.e., there is uncertainty in these back-
calculations that includes assumptions regarding elimination kinetics). There is also uncertainty
in applying such estimates to other members of the cohort based on similar characteristics (e.g.,
job category).

5.3.1.2. Uncertainty in Shape of the Dose-Response Curve

Another source of uncertainty is the nature of the dose-response curve in the low dose
region of interest for risk assessment for environmental exposures (e.g., <1 pg/kg-day). The
epidemiologic data are based on occupational studies in which exposures were often several
orders of magnitude higher than environmental exposures. In these studies, data from the low
dose region are quite sparse, and only one study examined uncertainty due to the low dose
region. Steenland and Deddens (2003, 198587) attempted to analyze this region specifically by
fitting threshold curves to the NIOSH data in which there was no extra risk from exposure until
some specific level. However, this model did not fit as well as models without a threshold. In
general, the usual assumption of linearity in the low dose region seems reasonable when using
epidemiologic data given the lack of data in this region that precludes the rejection of linearity.

There is uncertainty in the extrapolation of the OSF to the low dose region (e.g.,
<5 pg/kg-day). EPA developed the cancer assessment in this document assuming the slope in the
low-dose region of the dose-response curve is linear; the decision was made due to the lack of
sufficient evidence to support an assumption of nonlinearity as outlined in the EPA's Cancer
Guidelines (U.S. EPA, 2005, 086237). Similarly, there is uncertainty as to whether a threshold
exists for TCDD-induced toxicity leading to tumorigenesis and the dose associated with such a
threshold, if it exists, is unknown. EPA chose to model this dose-response without a threshold
because there is insufficient evidence to support an assumption of a threshold.

It also is noteworthy that the shapes of the exposure-response in several of these studies,
based on the published statistical models, is indicative of a response that tends to tail off or
"plateau" at high cumulative exposures to TCDD. This phenomenon has been seen in many
studies of occupational carcinogens, and may reflect a number of things including exhaustion of
people susceptible to cancer, saturation of biological pathways which are part of the pathway to

This document is a draft for review purposes only and does not constitute Agency policy.

5-82 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

cancer, and increased error measurement of dose at high levels biasing dose-response towards
the null (Stayner et al., 2003, 054922).

5.3.1.3. Uncertainty in Extrapolating Risks below Reference Population Exposure Levels

Another source of uncertainty in using human epidemiologic data is due to the lack of
completely unexposed populations; there are no human populations that have zero dioxin
exposure. The cancer exposure responses modeled in all epidemiologic cohorts, whether
primarily exposed via occupational or environmental exposures, can be evaluated with
confidence only above the lowest exposed group (i.e., the reference population). There are
substantial uncertainties associated with estimating cancer risks from background exposures of
TCDD and DLCs because these risks are aggregated in the overall background risk of the
referent population, to which outcomes of cohort subjects experiencing higher dioxin exposures
are compared. Therefore, the risk modeled from the epidemiologic data is unavoidably the
incremental risk above a background exposure to dioxins in the general environment (assumed to
be primarily from food intake). Typically, serum TCDD levels in the general populations in the
geographic locations and times at which the epidemiologic studies were undertaken have been
reported to be on the order of 5 to 20 ppt (Mocarelli et al., 1991, 199600)(WHO. 1998; Pin sky
and Lorber, 1998). Hence, the extra risks should be considered as those incurred by added
exposure above these background exposures, which then introduces uncertainty associated in the
cancer slope factor estimate at exposures below background levels. EPA assumes that the slope
of the risk curve below the background exposure experienced by the epidemiologic study cohorts
is the same as the (modeled) slope above those background exposure levels; data do not exist to
test this assumption.

Also, background TCDD/DLC exposures experienced by the epidemiologic study cohorts
have been estimated to be much larger (5 to 10-fold) than current background levels. Lorber et
al. (2009, 543766) estimate that current U.S. intake rates are roughly 0.58 pg TEQ/kg-day at the
50th percentile and suggest that human TEQ ingestion exposures likely peaked in the 1970's.
Steenland et al. (2001, 198589). presumably based in part on WHO (1998), estimated
background intake rates to be 5 pg TEQ/kg-day for the NIOSH cohort. As a result, the
assessment of cancer mortality risk at current background exposure levels is also subject to
extrapolation uncertainty.

This document is a draft for review purposes only and does not constitute Agency policy.

5-83 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

5.3.1.4. Uncertainty in Cancer Risk Estimates Arising from Background DLC Exposure

None of the slope factors presented in this document, whether based on epidemiologic
studies or animal bioassays, takes into account the impact of background exposure to DLCs.
Background DLC exposure can be estimated for only one of the animal cancer bioassays NTP
(2006, 543749). Background TCDD and DLC exposure for the rats in the NTP (2006, 543749)
does not appear to have been significant, with respect to the magnitude of administered doses
(see Section 5.3.2.1). However, given the trend towards lower exposures to TCDD in recent
years, the TCDD/DLC exposure may have been much higher in the older studies (e.g., Kociba et
al., 1978, 001818; NTP, 1982,	1; Toth et al., 1979, •). The impact of background

TCDD/DLC exposure on the cancer risk modeling of any of the bioassay data would be to
increase the dose term associated with each response; consequently, increasing the magnitude of
the BMDL, with a proportional reduction in the magnitude of the slope factor, although the
effect would probably be small (see Section 5.3.2.1). Note that the shift in dose increases the
estimated low doses proportionately more than the higher doses, potentially obscuring the
relationship between dose and response in the low dose region.

Background dioxin exposure for the epidemiologic cohorts, however, could have been
substantial with respect to the TCDD exposures in the reference populations used in the
modeling. As an example, the background dioxin intake the NIOSH cohort, which is the basis
for the oral slope factor described previously in this section (5.3), was estimated to be
0.5 pg/kg-day for TCDD and 10 times higher (5 pg/kg-day) for total TEQ (Steenland et al., 2001,
197433)(WHO. 1998). WHO (1998) estimated that TCDD comprised only about 5 to 10% of
total TEQ from exposure to DLCs in food, based on DLC exposure estimates and TEFs available
at that time. Eskenazi et al. (2004, 197160) estimated that TCDD was 20% of total TEQ in the
serum of the reference population in the Seveso Women's Health Study from measurements
taken in 1976. Based on more recent estimates (Lorber et al., 2009, 543766). TCDD is about
10% of total TEQ in human serum in the United States. Steenland et al. (2001, 198589) assumed
a (cumulating) background exposure of 5-6 ppt TCDD and 50 ppt total TEQ per year in serum
for their analysis of the NIOSH cohort cancer mortality response. The resulting cumulative
background exposures, particularly for total TEQ, are large compared to the lower cumulative
occupational exposures over the life-time of the cohort (birth to death or end of follow-up).

This document is a draft for review purposes only and does not constitute Agency policy.

5-84 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

Crump et al. (2003, 197384). based on Steenland et al. (2001, 198589), assumed a cumulative
background serum concentration of 3,000 ppt-year for total TEQ (50 ppt/year x 60 years), which
is much larger than the lower NIOSH cohort occupational TCDD exposures. The latter, when
grouped in cumulative TCDD serum-concentration septiles Steenland et al. (2001, 197433),
range from 260 to 850 ppt-yr in the first few septiles. Conceivably, the much larger background
exposure could have a somewhat larger effect on the slope factor than for the relatively lower
background exposure in the animal bioassays. Because the Cheng et al. (2006, 523122)
modeling does not account for background TEQ, the resulting slope factor is biased high. None
of the published analyses of the NIOSH cohort data (Cheng et al., 2006, 523122; Crump et al.,
2003, 197384; Steenland et al., 2001, 198589) present an analysis that addresses the effect of
background TEQ exposure on the modeled risk.50 Given the data and modeling results currently
available, the EPA could not find an approach for expressing the quantitative impact with any
accuracy or confidence.

5.3.1.5. Uncertainty in Cancer Risk Estimates Arising from Occupational DLC Coexposures

The slope factor estimates are based on an assumption that occupational exposure was
entirely to TCDD, with no explicit consideration of the risk attributable to occupational DLCs.
Because TCDD typically occurs as a component of a mixture with other DLCs that are assumed
to affect cancer risk through dose addition, the assumption that the exposures are entirely TCDD
could lead to a positive bias in the slope factor estimates derived from these epidemiologic
studies, if the estimates are confounded by other exposures to DLCs and the TEQ dose is larger
than the fraction accounted for by TCDD alone. The magnitude of the potential bias can be
estimated in a general way through the estimation of risks for plausible mixtures of DLCs and
TCDD exposures in the cohort with the same composition as the Steenland et al. (2001, 198589)
and Cheng et al. (2006, 523122) studies, but the detailed data required to perform such an
analysis on the NIOSH cohort are not available. In addition to the slope factor estimated for
TCDD, Becher et al. (1998, 197173) also evaluated the slope based on TEQs. They found a
dose-response effect for TCDD but not for TEQ (excluding TCDD) which suggests that
confounding by DLCs did not occur.

50 Steenland et al. (2001, 1974331 present a TEQ analysis but for a scenario where total TEQ is 10 times the TCDD
exposure for both background and occupational exposure.

This document is a draft for review purposes only and does not constitute Agency policy.

5-85 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

5.3.2. Other Sources of Uncertainty in Risk Estimates from the Epidemiological Studies

Other aspects of the Steenland et al. (2001, 198589). Cheng et al. (2006, 523122). Becher
et al. (1998, 197173). and Ott and Zober (1996, 198408) studies that are not directly associated
with TCDD or DLCs may contribute uncertainty to the cancer slope factor estimates. This
section lists several of these and discusses their potential directional bias in slope. General issues
associated with potential confounding effects also were discussed in the 2003 Reassessment
(U.S. EPA, 2003, 537122).

All of the studies that meet the criteria (with the exception of Warner et al., 2002,
189) measure cancer mortality rather than cancer incidence. This likely biases the slope
factor downward relative to a slope calculated for cancer incidence, the typical basis of EPA
cancer slope factors. In the NIOSH cohort, roughly one-third of the fatal cancers were identified
as lung cancer. Because of the high case mortality rate associated with lung cancer during the
period of cohort evaluation (e.g., the 5-year relative survival rates for lung cancer were less than
10% before 1973 and were less thanl5% before 1995 (Horner et al., 2009), the slope factor
estimated for cancer mortality might not be much lower than that calculated for cancer incidence.
This assumes that the outcome of a cancer incident (i.e., cancer mortality) is independent of
occupational TCDD exposure levels. Estimation of cancer incidence in the general population
associated with TCDD exposure would require assumptions related to the relative survival and
age-specific cancer risks in the exposed population compared to the NIOSH cohort or the
Hamburg cohort; insufficient data are available to support such an analysis.

The routes of TCDD exposures in the occupational cohorts include dermal and inhalation
exposures (Steenland et al., 1999, 197437). the U.S. population is assumed to be primarily
exposed through the intake of TCDD and DLCs in foods). Given the persitence of TCDD in the
body, differences in exposure routes may not be significant, but route-specific effects can not be
precluded. The directional bias on the slope factor that is associated with this uncertainty is not
known.

Occupational exposures to other carcinogens could lead to uncertainty in the slope factor.
For example, in addition to TCDD, the Hamburg cohort was also exposed to
hexachlorocyclohexane (HCH), which IARC classified as possibly carcinogenic to humans, and
lindane, which EPA (2001) stated had "suggestive evidence of carcinogenicity, but not sufficient
to assess human carcinogenic potential." While such co-exposures would not bias the exposure

This document is a draft for review purposes only and does not constitute Agency policy.

5-86 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

metric (i.e., not dose additive), to the extent that they were correlated with TCDD exposure, the
cancer mortality risk attributed to TCDD would be overestimated,biasing the slope factor high
because all cancers are attributed to TCDD. To examine this, Cheng et al. (2006, 523122)
assessed the impact of possible confounding by conducting excluding individual plants in the
modeling. If the estimated cancer risks as a function of exposure did not change too much when
specific facilities were left out, then confounding was deemed unlikely. Cheng et al. (2006,
523122) likewise found little variation in risks based on these analyses.

There is adequate evidence to believe age, gender, and body fat content all can have a
significant impact on elimination kinetics and consequent cancer risks associated with TCDD
exposure (U.S. EPA, 2003, 537122). While the authors evaluating the Hamburg cohort
accounted for such impacts in their kinetic analysis, interindividual kinetic differences were not
considered in evaluations of other cohorts.

There may be gender differences that affect susceptibility to TCDD exposure. The
cohorts analyzed by Steenland et al. (2001, 198589). Cheng et al. (2006, 523122). Ott and Zober
(1996, 198408) and Becher et al. (1998, 197173) were comprised almost exclusively of men.

This precluded systematically addressing differences between males and females in these studies.
Further, because EPA could not develop an estimate from the Warner et al. (2002, 197489)
cohort, none of the studies analyzed here for cancer dose-response contained a significant
percentage of women. Thus, the generalizability of the slope factor estimates to women is
uncertain.

Finally, of these cancer cohorts only the Seveso cohort included children. The unique
sensitivities of infants, toddlers, and children cannot be addressed based on information in the
occupational cohorts, although the increases in cancer risk in the Seveso cohort, to date, appear
to be modest. Aside from differences in exposure patterns and body fat content, the unique
developmental status of children may result in a substantially different profile of cancer risks
(and magnitudes of those risks) than can be addressed by simply compensating on the basis of
differences in body weight, food intake, etc. Further, because EPA could not develop an
estimate from the Warner et al. (2002, 197489) cohort, none of the studies for cancer dose-
response analyzed contained a significant percentage of women. Thus, the generalizability of the
slope factor estimates to women and children is uncertain.

This document is a draft for review purposes only and does not constitute Agency policy.

5-87 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

A number of other factors are routinely evaluated in cancer epidemiology studies, but
appear likely to have little impact on the direction of the slope factor; however, they likely
increase overall variability either in the dose or response. These include smoking and lifestyle
factors. Intraindividual variation in TCDD kinetics and susceptibility also could affect the
relationship between exposure and cancer risk. In each of these cases, it is difficult to determine
the directional bias these factors introduce into the derivation of the slope factor, unless
somehow they are correlated with with occupational dioxin exposures.

5.3.2.1. Effect of Added Background TEQ on TCDD Dose-Response

A source of uncertainty for TCDD dose-response modeling is the impact that background
exposures of TCDD and other DLCs might have on the modeling output. As mentioned
previously in Text Box 4-1, NTP (2006, 543749) presented measurements of TCDD in the fat of
control animals. To study the potential impact of background TCDD and total TEQ on the
cancer dose-response modeling for the NTP (2006, 197605) study, EPA has estimated
background levels of TCDD and TEQ (based on total TCDD, PeCDF and PCB126) from the
mixture study to serve as surrogates for background exposures in the TCDD-only study (limit of
detection too high for control level measurements). Background doses were estimated as:

Chemicalt(B) = Chem^ifa^Mc)/™'l x(Eq 5.9)
TCDD\fatTCDD J

where

Chemicali(B)

Chemicali(fatMc)

TCDD(fatxcDD)

DosexcDD

TEF;

= estimate of background exposure to Chemical i in ng/kg units of TCDD
blood concentrations at 105 weeks, for i = TCDD, PeCDF and PCB126.

= mean pg/g of Chemical i in the fat tissues of the control animals at
105 weeks in mixtures study (NTP, 2006, 		_j).

= mean pg/g of TCDD in the fat tissues of the 3 ng/kg dose group at
105 weeks in the TCDD study (NTP, 2006, 197605).

= 2.56 ng/kg TCDD blood concentration for the 3 ng/kg dose group in the
TCDD study (from the Emond rat PBPK modeling of NTP, 2006,
197605)

= Toxicity Equivalence Factor for Chemical i (from Van den berg et al.
(2006, 543769)).

This document is a draft for review purposes only and does not constitute Agency policy.

5-88 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

Assuming simple proportionality of blood TCDD concentrations between controls and
low-dose (3 ng/kg) animals, the TEF-adjusted ratio of each congener (Chemical z) in control-
animal fat to low-dose-animal fat is multiplied by the modeled TCDD blood concentration for
the low-dose animals to obtain an equivalent background exposure in the dose metric (ng/kg
whole blood) used to calculate all the OSFs in this assessment. For total TEQ, the estimates
across the three congeners are summed. The total TEQ estimates are biased somewhat high
because they are based on terminal (2-year) measurements rather than representing lifetime
averages. Background exposures are then added to the modeled TCDD blood concentrations for
several different background exposure scenarios (see Table 5-23) prior to conducting
Benchmark-Dose (BMD) modeling.

BMD modeling was conducted for the cholangiocarcinoma endpoint in the TCDD study
(NTP, 2006, 197605). This was done for scenarios that added the following estimated TCDD or
TEQ background doses to the TCDD study doses: background TCDD only, total estimated TEQ,
twice the total TEQ and ten times the background TCDD (see Table 5-23). These doses may
bound the potential background exposures as TCDD has been thought to represent about 10% of
all TEQs at environmental levels (WHO, 1998). Table 5-24 shows that, as expected, adding to
the exposure term increases the BMDL (and decreases the OSF) and also shifts the shape of the
dose-response slope slightly towards sublinear (see Appendix I). However, at these background
exposure levels relative to the administered dose levels, there is very little quantitative impact on
the cancer dose-response modeling for the NTP (2006, 197605) study. Even with the most
extreme assumption that background TCDD is only 10% of total background TEQ, the BMDL
changes by only 12%. Assuming that background exposures were higher for older studies (e.g.,
Kociba et al., 1978, 001818; NTP, 1982, 594255), the impact would be somewhat higher, but
unless the background exposures were substantially higher than the lower tested doses (ca.
1-10 ng/kg-day), a significant change in the dose-response modeling results would not be
expected.51

However, as discussed previously, background TEQ exposures were likely very high
with respect to the lower occupational TCDD exposure levels as reported in the epidemiologic
studies. Table 5-25 shows the relative increase in exposure levels (as cumulative serum TCDD

51 Note that the situation is different for single-exposure studies where accumulated body burden from background
exposures could be higher than the lowest administered dose (see Tex Box 4.1 in Section 4.4).

This document is a draft for review purposes only and does not constitute Agency policy.

5-89 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

concentrations) for the NIOSH cohort septiles assuming that total background TEQ is 10 times
background TCDD and that 50 ppt TEQ per year is accumulated in serum (Crump et al., 2003,

84; Steenland et al., 2001, 198589). Although definitive quantitative analyses have not yet
been published or designed, the impact on modeled TCDD risk from these studies could be
substantial. The expectation for the direction of the effect would be the same as for the animal
bioassays; adding to the exposure magnitude without changing the response would decrease the
unit risk.

5.3.3. Approaches to Combining Estimates from Different Epidemiologic Studies

Meta-analyses and pooled analyses are two common approaches for combining
epidemiologic study data. Meta-analyses are a useful way to combine epidemiologic data from
different studies and derive a common estimate of effect, particularly when there are a large
number of comparable studies that are fairly homogenous as to make them possible to combine.
A meta-analysis often involves a weighted average of effect measures, dose-response
coefficients, orEDoiS.

Unlike a meta-analysis, a pooled analysis combines the original exposure and health
outcome data across multiple studies, enabling a fit of new models to the data which were not
used in the original publications. Whereas a pooled analysis of the four different cohorts
considered here would be useful to explore the functional form and fit of models (either
statistical or multistage) across all four cohorts, this would entail a lengthy undertaking and is not
being contemplated here, due in part to concerns about the confidence in the results of such an
undertaking.

5.3.3.1. The Crump et al. (2003,197384) Meta-analysis

Crump et al. (2003, 197384) published a meta-analysis that incorporated data from the
three studies EPA used in the quantitative dose-response modeling presented in the 2003
Reassessment (U.S. EPA, 2003, 537122). These three study populations were the NIOSH
(Steenland et al., 2001, 197433). the Hamburg (Becher et al., 1998, 197173). and the BASF (Ott
and Zober, 1996, 198408) cohorts. The data for the NIOSH study included six additional years
of follow-up and improved TCDD exposure estimates that had not been applied to EPA's dose-
response modeling in the 2003 Reassessment. This study examined the relationship between

This document is a draft for review purposes only and does not constitute Agency policy.

5-90 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

TCDD exposure and all-cancer mortality. SMR statistics that had been used in all three studies
were applied.

The Crump et al. (2003, 197384) analysis was based on published data, and therefore,
selection of the dose metric was limited to how aggregated data had been presented in the
publications. For the NIOSH component of the analysis, the exposure data were based on
worker-specific data and specific processes performed at each plant (Steenland et al., 2001,
i ). The previous approach assigned workers that had broad categories of exposure
duration with the same cumulative serum level, and did not take into account the particular plant
or the job assignment within the plant. The Crump et al. (2003, 197384) approach did take into
account when exposure occurred in relation to the follow-up interval. The TCDD exposure
metric used was a cumulative serum lipid concentration (CSLC). For the Hamburg cohort,
Crump et al. (2003, 197384) used an average value from the exposure ranges provided in Flesch-
Janys et al. (1998, 197339). For the BASF cohort, arithmetic averages for the dose categories
were converted to TCDD CSLC intakes by dividing them by 0.25 (average body fat of 25%) and
a decay rate that corresponded to a half-life of 7 years.

The outcome variable for the dose-response modeling was all cancer mortality, and
CSLC was the independent variable. Crump et al. (2003, 197384) performed a series of trend
tests to determine the lowest dose for which a statistically significant trend in SMR could be
shown and all other lower doses. These tests also examined the highest dose in which there was
no statistically significant trend using data from this dose and all other lower doses. Estimates of
EDio, EDos, and ED0i for TEQ with respect to the lifetime probability of dying from cancer were
calculated. This calculation assumed a first-order elimination process with a half-life of
7.6 years, a 50% systemic uptake of ingested dioxin, that dioxin concentration in serum lipid is a
suitable measure for dioxin concentration in all lipid, and that all dioxin is sequestered in lipid
(which comprises 25% of body weight). Age-specific mortality rates in the presence of dioxin
exposure were then generated. Life-table methodology was used to calculate lifetime risks of
cancer mortality.

Based on the modeling results, the hypothesis of a baseline SMR of 1.0 was rejected, and
the linear model produced an SMR estimate of 1.17 (95% CI = 1.04-1.30) from these studies.
The dose-response curves for the three studies were not homogeneous. Namely, the points from
the BASF cohort fell below the predicted curve. Because the heterogeneity was not judged to be

This document is a draft for review purposes only and does not constitute Agency policy.

5-91 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

extreme by different statistical tests, however, the investigators used a common model in a
combined analysis of the data from the three studies. The linear model provided an adequate fit
of the data, and the slope associated with CSLC-ppt was 6.3 x io~6 (95% CI = 8.8 x io~7 to
1.3 x 10~5). Based on goodness of fit analysis, the preferred estimate of ED0i was 45 pg/kg-day,
which was six times higher than the estimate of 7.7 pg/kg-day derived by Steenland et al. (2001,
198589V

5.3.3.2. EPA 'v Decision Not to Conduct a Meta-analysis

From a statistical perspective, meta-analyses may not be very reliable when applied to a
small number of studies. Crump et al. (2003, 197384) used only three studies. Had EPA
undertaken a meta-analysis for the studies that met its criteria, most of the weight would come
from the two large studies on the NIOSH and Hamburg cohorts. However, such an analysis
relies on an assumption of a normally distributed between-study effect. This normality
assumption cannot be assessed with only three observations, yet the meta-analysis estimate is
highly sensitive to this distributional assumption (Higgins et al., 2009, 594339). Because of this
limitation and the imprecision of the between-study variance estimate, statisticians often
recommend forgoing meta-analysis in favor of discussing the individual studies when few
studies are available (Cox, 2006, 594342; Higgins et al., 2009, 594339). Based on these
considerations, EPA decided not to undertake a meta-analysis in this document.

As noted previously. Crump et al. (2003, 197384) has conducted a meta-analysis of the
three cohorts considered here, i.e., the NIOSH, Hamburg, and BASF cohorts. However, Crump
et al. modeled SMR data in which the cohorts were compared to the general population, rather
than on internal exposure-response analyses as relied upon in this document. Their analysis
included a total of 15 different SMRs from the three studies. A prior analysis of the dose-
responses by Becher et al. (1998, 197173) was used (i.e., the categorical SMR analysis by
Flesch-Janys et al. (1998, 197339)). Additionally, a prior analysis of the NIOSH cohort
(Steenland et al., 1999, 197437) in which SMRs were calculated was used. Crump et al. (2003,

	84) found that a linear dose-response gave a good fit to the data, and used that for deriving

an EDoi. However, they found that a supra-linear dose-response provided a better fit to the data,
but rejected the supra-linear model (a power model) because of an infinite slope at zero dose. In
the original publications by Becher et al. (1998, 197173) and Steenland et al. (2001, 198589).

This document is a draft for review purposes only and does not constitute Agency policy.

5-92 DRAFT—DO NOT CITE OR QUOTE

-------
1	both observed a supra-linear dose-response trend. Crump et al. (2003, 197384) concluded that

2	the EDoi was 45 pg/kg-day, six times higher than the ED0i of 7.7 pg/kg-day calculated by

3	Steenland et al. (2001, 198589) using the same dietary units (pg/kg-day).

This document is a draft for review purposes only and does not constitute Agency policy.

5-93 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

Table 5-1. Cancer slope factors calculated from Becher et al. (1998,197173),
Steenland et al. (2001,197433) and Ott and Zober (1996,198408) from 2003
Reassessment Table 5-4

Study

EDoi (LEDoi)
(ng/kg)

Cancer slope factor per
ng/kg-day above
background3 (UCL)

Hamburg cohort
Power model

Becher et al. (1998. 197173)

6 (N.A.)

5.1 (N.A.)

Hamburg cohort

Additive model

Becher et al. (1998. 197173)

18.2 (N.A.)

1.6 (N.A.)

Hamburg cohort
Multiplicative model
Becher et al. (1998. 197173)

32.2 (N.A.)

0.89 (N.A.)

NIOSH cohort
Piecewise linear model
Steenland et al. (2001. 198589)

18.6 (11.5)

1.5 (2.5)

BASF cohort, from Ott and Zober
(1996. 198408). multiplicative

50.9 (25.0)

0.57 (1.2)

aAssumes 25% of body weight is lipid; in humans 80% of dioxin dose is absorbed from the normal
diet; the TCDD half-life is 7.1 years in humans. Background all cancer mortality rate calculated
through lifetable analysis to 75 years. Summary results are for male all cancer risk, because the
male lifetime (to 75 years) all cancer risk is greater than for females, leading to correspondingly
higher cancer slope factors. As detailed in Part III, Chapter 8, RelRisk(ED0i) = 0.99 +

0.01/Risk(0 dose)- Based on the manner in which the dose-response data were calculated using Cox
regression rate ratio analyses, risks are given as cancer slope factors for 1 pg/kg-day above
background, assumed 5 ppt TCDD in lipid.

UCL = upper confidence limit.

Source: U.S. EPA (U.S. EPA. 2003, 537122: Part III. Chapter 5. Table 5-4)

This document is a draft for review purposes only and does not constitute Agency policy.

5-94 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table 5-2. Cox regression coefficients and incremental cancer-mortality risk

2	for NIOSH cohort data

3

Model

Cox regression
coefficient estimate
(ppt-year)"1

Incremental risk3

Steenland et al.(2001. 197433) (untagged exposures)

Piecewise linear

1.5 x 10 5

7.0 x 10 4

Cheng et al. (Cheng et al.. 2006. 523122) (exposures lagged 15 vears)

Linear, lower 95% of observations

3.3 x 10 6b

1.2 x 10 4

Linear, full data

1.7 x I0~8c

6.3 x 10~7

4

5	aCompared to internal reference population (lowest exposure group),with a cancer mortality rate of 0.214; assumes

6	background exposure of 5 ppt per year serum-lipid TCDD concentration.

7	hp < 0.05.

8	cp< 0.05.

9	dNot statistically significant (p > 0.05).

10

11	Source: Cheng et al. (2006, 523122: Table IV).

This document is a draft for review purposes only and does not constitute Agency policy.

5-95 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

Table 5-3. Comparison of fat concentrations, risk specific dose estimates and
associated oral slope factors based on upper 95th percentile estimate of
regression coefficient3 of all fatal cancers reported by Cheng et al. (2006,
523122) for selected risk levels

Risk level
(RL)

AUCrl
(ppt-yr)

FATrL
(ng/kg)

Risk specific doseb
(Drl) (ng/kg-day)

Equivalent oral slope
factors (OSFrl) per
(mg/kg-day)

1 x 10~2

1.262 x 104

1.803 x 102

8.79 x 10~2

1.1 x 105

5 x 10~3

6.432 x 103

9.189 x 101

3.14 x 10~2

1.6 x 105

1 x 10~3

1.307 x 103

1.867 x 101

2.88 x 10~3

3.5 x 105

5 x 10~4

6.546 x 102

9.352 x 10°

9.56 x 10~4

5.2 x 105

1 x 10~4

1.311 x io2

1.873 x 10°

1.29 x 10~4

7.8 x 105

5 x 10~5

6.558 x 101

9.368 x 10"1

5.52 x 10~5

9.1 x io5

1 x 10~5

1.312 x 101

1.874 x 10_1

8.94 x 10~6

1.1 x io6

5 x 10~6

6.559 x 10°

9.370 x 10"2

4.25 x 10~6

1.2 x 106

1 x 10~6

1.312 x 10°

1.874 x 10~2

8.08 x 10~7

1.2 x 106

5 x 10~7

6.559 x 10"1

9.370 x 10"3

4.00 x 10~7

1.3 x 106

1 x 10~7

1.312 x 10_1

1.874 x 10~3

7.92 x 10~8

1.3 x 106

a Based on regression coefficient of Cheng et al. (2006, 523122. Table III), excluding observations in the upper 5%
range of the exposures; where reported (3 = 3.3 x 10 6 ppt-years and standard error = 1.4 x 10 6. Upper 95th
percentile estimate of regression coefficient (P95) calculated to be 6.04 x 10 6 = (3.3 x 10 6) + 1.96 x (1.4 x 10 6);
background cancer mortality risk is assumed to be 0.112 as reported by Cheng et al. (2006, 5231221.
bTo calculate the extra cancer risk (ER) and OSF for any TCDD daily oral intake (D):

5.	For D in ng/kg-d, look up the corresponding fat concentration (ng/kg = ppt) from the conversion chart
(nongestational lifetime dose metrics) in Appendix C.4.1.

6.	Calculate the AUC in ppt-yrs by multiplying the fat concentration by 70 years.

7.	Calculate Extra Risk (ER) using the following equation:

ER = [exp(AUC x 6.04E-6) x 0.112 - 0.112] - 0.888.

8.	Calculate the OSF (mg/kg-d)"1 = 1E6 x (ER D).

Example for risk at the RfD: D = 7 x 10"4 ng/kg-d; fat concentration = 6.93 ng/kg;

AUC = 70 years x 6.93 ppt = 485 ppt-year;

ER = exp(485 ppt-year x 6.04E-6 (ppt-yr)"1) x 0.112 - 0.112) - 0.888 = 3.7 x 10"4
OSF = 1E6 ng/mg x (3.7 x io~4 + 7 x io-4 ng/kg-d) = 5.3 x 105 (mg/kg-d)"1.

This document is a draft for review purposes only and does not constitute Agency policy.

5-96 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table 5-4. Comparison of fat concentrations, risk specific dose estimates and

2	associated central tendency slope estimates based on best estimate of

3	regression coefficient3 of all fatal cancers reported by Cheng et al. (2006,

4	523122) for selected risk levels

5

Risk level
(RL)

AUCrl,
(ppt-yr)

FATrL
(ng/kg)

Risk specific dose
(Drl) (ng/kg-day)

Central
tendency slope

estimates
(mg/kg-day)1

1 x 1(T2

2.312 x 104

3.303 x 102

2.21 x KT1

4.5 x 104

1 x l(T3

2.393x 103

3.419 x 101

6.97 x 10~3

1.4 x 105

1 x l(T4

2.402 x 102

3.431 x 10°

2.74 x 10~4

3.7 x 105

1 x l(T5

2.403 x 101

3.432 x 10_1

1. 74 x 10~5

5.7 x 105

1 x l(T6

2.403 x 10°

3.432 x 10~2

1.50 x 10~6

6.7 x 105

1 x l(T7

2.403 x 10_1

3.432 x 10~3

1.46 x 10~7

7.0 x 105

6

7	'Based on regression coefficient of Cheng et al (2006, 523122: Table III) excluding observations in the upper 5%

8	range (>252,950 ppt-year lipid adjusted serum TCDD) of the exposures; where reported (3 = 3.3 x 10 6 ppt-years;

9	background cancer mortality risk is assumed to be 0.112 as reported by Cheng et al. (2006, 5231221.

10

11

12	Table 5-5. Kociba et al. (1978, 001818) male rat tumor incidence data3 and

13	blood concentrations for dose-response modeling

14

Morphology: topography

Vehicle control
(ng/kg)

Low dose
(ng/kg)

Medium dose
(ng/kg)

High dose
(ng/kg)

0

1.56

7.16

38.72

Stratified squamous cell
carcinoma of hard palate or nasal
turbinates

0/85

0/50

0/50

4/5 0b

Stratified squamous cell
carcinoma of tongue

0/85

1/50

1/50

3/50b

Adenoma of adrenal cortex

0/85

0/50

2/50

5/5 0b

15

16	"Source: Kociba et al.(1978, 001818: Table 4).

17	Statistically significant by Fischer Exact Test (p < 0.05).

This document is a draft for review purposes only and does not constitute Agency policy.

5-97 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table 5-6. Kociba et al. (1978, 001818) female rat tumor incidence data3 and

2	blood concentrations for dose-response modeling

3

Morphology: topography

Vehicle control
(ng/kg)

Low dose
(ng/kg)

Medium dose
(ng/kg)

High dose
(ng/kg)

0

1.55

7.15

38.56

Hepatocellular adenoma(s) or
carcinoma(s)

2/86

1/50

9/5 0a

18/45b

Stratified squamous cell
carcinoma of hard palate or
nasal turbinates

0/86

0/50

1/50

4/49b

Keratinizing squamous cell
carcinoma of lung

0/86

0/50

0/50

7/49b

4

5	"Source: Kociba et al. (1978, 001818: Table 5). Incidence for Hepatocellular adenomas or carcinomas is from

6	Goodman and Sauer (Goodman and Sauer. 1992, 197667: Table 1); EPA calculated statistical significance as the

7	study authors did not provide this.

8	Statistically significant by Fischer Exact Test (p < 0.05).

9

10

11	Table 5-7. NTP (1982, 594255) female rat tumor incidence data" and blood

12	concentrations for dose-response modeling

13

Morphology: topography

Vehicle control
(ng/kg)

Low dose
(ng/kg)

Medium dose
(ng/kg)

High dose
(ng/kg)

0

1.96

5.69

29.75

Subcutaneous tissue: fibrosarcoma

0/75

2/50

3/50

4/49b

Liver: neoplastic nodule or
hepatocellular carcinoma

5/75c

1/49

3/50

14/49b

Adrenal: cortical adenoma, or
carcinoma or adenoma, NOS

1l/73c

9/49

5/49

14/46b

Thyroid: follicular-cell adenoma

3/73c

2/45

1/49

6/47

14

15	"Source: NTP (1982, 594255: Table 10).

16	Statistically significant by Fischer Exact Test (p < 0.05).

17	Statistically significant trend by Chochran-Armitage test (p < 0.05).

This document is a draft for review purposes only and does not constitute Agency policy.

5-98 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table 5-8. NTP (1982, 594255) male rat tumor incidence data3 and blood

2	concentrations for dose-response modeling

3

Morphology: topography

Vehicle control
(ng/kg)

Low dose
(ng/kg)

Medium dose
(ng/kg)

High dose
(ng/kg)

0

1.96

5.70

29.87

Liver: neoplastic nodule or
hepatocellular carcinoma

0/74b

0/50

0/50

3/50

Thyroid: follicular-cell adenoma or
carcinoma

l/69b

5/48c

8/5 0C

1 l/50c

Adrenal cortex: adenoma

6/72

9/50

12/49b

9/49

4

5	"Source: NTP(1982, 594255: Table 9).

6	Statistically significant trend by Chochran-Armitage test (p < 0.05).

7	Statistically significant by Fischer Exact Test (p < 0.05).

8

9

10	Table 5-9. NTP (1982, 594255) female mouse tumor incidence data3 and

11	blood concentrations for dose-response modeling

12

Morphology: topography

Vehicle control
(ng/kg)

Low dose
(ng/kg)

Medium dose
(ng/kg)

High dose
(ng/kg)

0

1.95

5.84

32.06

Subcutaneous tissue: fibrosarcoma

l/74b

1/50

1/48

5/47c

Hematopoietic system: lymphoma
or leukemia

18/74b

12/50

13/48

20/47c

Liver: hepatocellular adenoma or
carcinoma

3/73b

6/50

6/48

1 l/47c

Thyroid: follicular-cell adenoma

0/69b

3/50

1/47

5/46c

13

14	"Source: NTP (1982, 594255: Table 15).

15	Statistically significant trend by Chochran-Armitage test (p < 0.05).

16	Statistically significant by Fischer Exact Test (p < 0.05).

This document is a draft for review purposes only and does not constitute Agency policy.

5-99 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table 5-10. NTP (1982, 594255) male mouse tumor incidence data3 and

2	blood concentrations for dose-response modeling

3

Morphology: topography

Vehicle control
(ng/kg)

Low dose
(ng/kg)

Medium
dose (ng/kg)

High dose
(ng/kg)

0

0.77

2.27

11.24

Lung: alveolar/bronchiolar adenoma
or carcinoma

10/7 lb

2/48

4/48

13/50

Liver: hepatocellular adenoma or
carcinoma

15/73b

12/49

13/49

27/50c

4

5	"Source: NTP (1982, 594255: Table 14).

6	Statistically significant trend by Chochran-Armitage test (p < 0.05).

7	Statistically significant by Fischer Exact Test (p < 0.05).

8

9

10	Table 5-11. NTP (2006,197605) female rat tumor incidence data" and blood

11	concentrations for dose-response modelingb

12

System:

morphology:

topography

Vehicle
control
(ng/kg)

Low
dose
(ng/kg)

Low-med
dose (ng/kg)

Median

dose
(ng/kg)

Med-high
dose (ng/kg)

High dose
(ng/kg)

0

2.56

5.69

9.79

16.57

29.70

Liver:

cholangiocarcinoma

0/49c

0/48

0/46

1/50

4/49

25/53c

Liver:

hepatocellular
adenoma

0/49c

0/48

0/46

0/50

1/49

13/53c

Oral mucosa:
squamous cell
carcinoma

l/49c

2/48

1/46

0/50

4/49

10/5 3c

Pancreas: adenoma
or carcinoma

0/48c

0/48

0/46

0/50

0/48

3/51

Lung: cystic
keratinizing
epithelioma

0/49c

0/48

0/46

0/49

0/49

9/52c

13

14	'Source: NTP (2006. 197605: Table A3a).

15	incidence adjusted for animals <365 days on study.

16	Statistically significant by Poly-3 Test (p < 0.05).

17

18

This document is a draft for review purposes only and does not constitute Agency policy.

5-100 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table 5-12. Toth et al. (1979,197109) male mouse tumor incidence data3 and

2	blood concentrations for dose-response modeling

3

Morphology: topography

Vehicle control
(ng/kg)

Low dose
(ng/kg)

Medium dose
(ng/kg)

High dose
(ng/kg)

0

0.57

14.21

91.21

Liver tumors

7/38

13/44

21/44b

13/43

4

5	"Source: Toth et al. (1979, 197109: Table 1).

6	Statistically significant by Chi2 Test (p < 0.01).

7

8

9	Table 5-13. Delia Porta et al. (1987,197405) male mouse tumor incidence

10	data3 and blood concentrations for dose-response modeling

11

Morphology: topography

Vehicle control
(ng/kg)

Low dose
(ng/kg)

High dose
(ng/kg)

0

38.00

67.77

Hepatocellular carcinoma

5/43

15/5lb

33/50b

12

13	"Source: Delia Porta et al. (1987, 197405: Table 4).

14	Statistically significant by Chi2 Test (p < 0.05).

15

16

17	Table 5-14. Delia Porta et al. (1987,197405) female mouse tumor incidence

18	data3 and blood concentrations for dose-response modeling

19

Morphology: topography

Vehicle control
(ng/kg)

Low dose
(ng/kg)

High dose
(ng/kg)

0

37.59

66.97

Hepatocellular adenoma

2/49

4/42b

1l/48b

Hepatocellular carcinoma

1/49

12/42b

9/4 8b

20

21	"Source: Delia Porta et al. (1987, 197405: Table 4).

22	Statistically significant by Chi2 Test (p < 0.05).

This document is a draft for review purposes only and does not constitute Agency policy.

5-101 DRAFT—DO NOT CITE OR QUOTE

-------
Table 5-15. Comparison of multi-stage modeling results across cancer bioassays using blood concentrations

Study

Species

Sex

Morphology: topography

Multi-stage modeling:3
stage, GoF />-valuc, LL difference

BMDoi
(ng/kg)

BMDLoi

(ng/kg)

Delia
Porta et
al.

(1987,
197405)

Mouse

Male

Hepatocellular carcinoma

2,/? = 0.52

7.14

1.17

Female

Hepatocellular adenoma

2,/? = 0.86

14.49

2.34

Hepatocellular carcinoma

\.p = 0.019

2.30

1.54

Kociba
et al.

(1978,

001818)

Rat

Male

Stratified squamous cell carcinoma of hard palate or nasal turbinates

\,p = 0.81

5.76

2.79

Stratified squamous cell carcinoma of tongue

\,p = 0.47

6.09

2.60

Adenoma of adrenal cortex

l,p= 0.78

3.25

1.85

Combined tumors Bayesian analysis



1.57

0.96

Female

Hepatocellular adenoma(s) or carcinoma(s)

l,p = 0.24

0.70

0.50

Stratified squamous cell carcinoma of hard palate or nasal turbinates

l,p = 0.97

4.51

2.34

Keratinizing squamous cell carcinoma of lung

l,p = 0.63

3.14

1.79

Combined tumors Bayesian analysis



0.51

0.37

NTP

(1982,

594255)

Rat

Female

Subcutaneous tissue: fibrosarcoma

\,p = 0.18

3.13

1.38

Liver: neoplastic nodule or hepatocellular carcinoma

\,p = 0.22

1.17

0.74



Adrenal: cortical adenoma, or carcinoma or adenoma, NOS

\,p = 0.34

1.61

0.81

Thyroid: follicular-cell adenoma

\,p = 0.57

3.38

1.55

Combined tumors Bayesian analysis



0.46

0.31

Male

Liver: neoplastic nodule or hepatocellular carcinoma

\,p = 0.85

6.14

2.70

Thyroid: follicular-cell adenoma or carcinoma

\,p = 0.06

1.21

0.70

Adrenal cortex: adenoma

\,p = 0.06

3.98

1.22

Combined tumors Bayesian analysis



0.74

0.44

-------
Table 5-15. Comparison of multi-stage modeling results across cancer bioassays using blood concentrations (continued)

Study

Specie
s

Sex

Morphology: topography

Multi-stage modeling:3
stage, GoF />-value, LL difference

BMDoi
(ng/kg)

BMDLoi

(ng/kg)

NTP

(1982,

594255)

Mouse

Female

Subcutaneous tissue: fibrosarcoma

l,p = 0.93

3.40

1.69

Hematopoietic system: lymphoma or leukemia

l,p = 0.98

1.14

0.61

cont.

Liver: hepatocellular adenoma or carcinoma

l,p = 0.34

1.49

0.83

Thyroid: follicular-cell adenoma

1 ,p = 0.09, no improvement with
higher orders

3.05

1.44

Combined tumors Bayesian analysis



0.44

0.29

Male

Lung: alveolar/bronchiolar adenoma or carcinoma

l,p = 0.09

2.53

0.41

Liver: hepatocellular adenoma or carcinoma

l,p = 0.93

0.21

0.14

Combined tumors Bayesian analysis



0.16

0.11

NTP

(2006,

197605)

Rat

Female

Liver: cholangiocarcinoma

3,p = 0.99, dLL = 2.93

7.57

4.13

Liver: hepatocellular adenoma

3,p = 0.93, dLL = 2.10

10.22

6.53



Oral mucosa: squamous cell carcinoma

l,p = 0.27

2.20

1.39

Pancreas: adenoma or carcinoma

l,p= 0.64

10.52

4.63

Lung: cystic keratinizing epithelioma

2,p = 0.51, dLL = 3.55

8.30

5.24

Combined tumors Bayesian analysis



1.18

0.78

Toth et
al.

(1979,
197109)

Mouse

Male

Liver: tumors

l,p = 0.29

0.37

0.21

*Analysis uses a chi-square goodness of fit statistic for differences in the log likelihoods (p > 0.05).

-------
1	Table 5-16. Individual tumor points of departure and slope factors using

2	blood concentrations

3

Study

Tumor Site (Sex/Species)

BMDLhed
(ng/kg-day)

OSF
(per mg/kg-day)

NTP (1982. 594255s)

Liver: adenoma or carcinoma (male mice)

1.7E-03

5.8E+6

Tot.het.al. (1979. 197109^

Liver tumors (male mice)

1.9E-03

5.2E+6

NTP. (1982. 594255V

Lung: adenoma or carcinoma (male mice)

8.7E-03

1.1E+6

Kocibaetal. (1978. 001818)

Liver: adenoma or carcinoma (female rats)

1.2E-02

8.6E+5

NTP (1982. 594255s)

Hematopoietic: lymphoma or leukemia (female
mice)

1.6E-02

6.4E+5

NTP (1982. 594255V

Thyroid: follicular cell adenoma (male rats)

1.9E-02

5.2E+5

NTP (1982. 594255s)

Liver: neoplastic nodule or hepatocellular
carcinoma (female rats)

2.1E-02

4.8E+5

NTP (1982. 594255s)

Adrenal: cortical adenoma or carcinoma or
adenoma, NOS (female rats)

2.4E-02

4.1E+5

NTP (1982. 594255s)

Liver: adenoma or carcinoma (female mice)

2.5E-02

4.0E+5

Delia Porta et al. (1987. 197405s)

Hepatocellular carcinoma (male mice)

3.1E-02

3.2E+5

NTP (1982. 594255V

Adrenal cortex: adenoma (male rats)

4.5E-02

2.2E+5

Delia Porta et al. (1987,
197405V

Hepatocellular carcinoma (female mice)

4.9E-02

2.0E+5

NTP (1982. 594255s)

Subcutaneous fibrosarcoma (female rats)

5.4E-02

1.8E+5

NTP (2006. 197605s)

Oral mucosa: squamous cell carcinoma (female
rats)

5.5E-02

1.8E+5

NTP (1982. 594255V

Thyroid: adenoma (female mice)

5.7E-02

1.7E+5

NTP (1982. 594255s)

Thyroid: follicular cell adenoma (female rats)

6.5E-02

1.5E+5

NTP (1982. 594255s)

Subcutaneous fibrosarcoma (female mice)

7.4E-02

1.4E+5

Kociba et al. (1978. 001818s)

Lung: carcinoma (female rats)

8.0E-02

1.2E+5

Kociba et al. (1978. 001818s)

Adenoma of adrenal cortex (male rats)

8.5E-02

1.2E+5

Delia Porta et al. (1987. 197405s)

Hepatocellular adenoma (female mice)

9.4E-02

1.1E+5

Kociba et al. (1978. 001818s)

Nasal/Palate: carcinoma (female rats)

1.2E-01

8.2E+4

Kociba et al. (1978. 001818s)

Tongue: carcinoma (male rats)

1.4E-01

7.0E+4

NTP (1982. 594255s)

Liver: neoplastic nodule or hepatocellular
carcinoma (male rats)

1.5E-01

6.6E+4

Kociba et al. (1978. 001818s)

Nasal/Palate: carcinoma (male rats)

1.6E-01

6.3E+4

NTP C2006. 197605s)

Liver: cholangiocarcinoma (female rats)

2.9E-01

3.5E+4

NTP (2006. 197605s)

Pancreas: adenoma or carcinoma (female rats)

3.4E-01

2.9E+4

NTP (2006. 197605s)

Lung: cystic keratinzing epithelioma (female rats)

4.1E-01

2.4E+4

NTP (2006. 197605s)

Liver: hepatocellular adenoma (female rats)

5.6E-01

1.8E+4

This document is a draft for review purposes only and does not constitute Agency policy.

5-104 DRAFT—DO NOT CITE OR QUOTE

-------
1

2	Table 5-17. Multiple tumor points of departure and slope factors using blood

3	concentrations

4

Study

Sex/species: tumor sites

BMDLhed
(ng/kg-day)

OSF
(per mg/kg-day)

NTP (1982. 594255s)

Male mice: liver adenoma and carcinoma, lung

1.1E-03

9.4E+6

NTP (1982. 594255s)

Female mice: liver adenoma and carcinoma, thyroid
adenoma, subcutaneous fibrosarcoma, all lymphomas

5.3E-03

1.9E+6

NTP (1982. 594255s)

Female rats: liver neoplasitc nodules, liver adenoma
and carcinoma, thyroid follicular cell adenoma,adrenal
cortex adenoma or carcinoma

5.7E-03

1.8E+6

Kociba et al. (1978,

001818s)

Female rats: liver adenoma carcinoma, oral cavity,
lung

7.3E-03

1.4E+6

NTP (1982. 594255s)

Male rats: thyroid follicular cell adenoma, adrenal
cortex adenoma

9.6E-03

1.0E+6

NTP (2006. 197605s)

Female rats: liver cholangiocarcinoma, hepatocellular
adenoma, oral mucosa squamous cell carcinoma, lung
cystic keratinizing epithelioma, pancreas adenoma,
carcinoma

2.3E-02

4.4E+5

Kociba et al. (1978,

001818s)

Male rats: adrenal cortex adenoma, tongue carcinoma,
nasal/palate carcinoma

3.1E-02

3.2E+5

5

This document is a draft for review purposes only and does not constitute Agency policy.

5-105 DRAFT—DO NOT CITE OR QUOTE

-------
Table 5-18. Comparison of cancer BMDs, BMDLs, and slope factors for combined or selected individual tumors
for 1, 5, and 10% extra risk

Study

Species

Sex

BMDoi
(ng/kg)

BMDLoi
(ng/kg)

SF01
(ng/kg) 1

bmd05

(ng/kg)

bmdl05

(ng/kg)

sf05

(ng/kg) 1

BMD10
(ng/kg)

BMDL10
(ng/kg)

SF10
(ng/kg) 1

Kociba

(1978,

001818V

Rat

Female

4.9E-01

3.8E-01

2.7E-02

2.5E+00

1.9E+00

2.7E-02

4.9E+00

3.8E+00

2.7E-02

Male

1.5E+00

9.6E-01

1.0E-02

7.2E+00

4.8E+00

1.0E-02

1.5E+01

9.6E+00

1.0E-02

NTP

(1982,

594255V

Rat

Female

4.4E-01

3.2E-01

3.2E-02

2.2E+00

1.6E+00

3.2E-02

4.4E+00

3.2E+00

3.2E-02

Male

6.9E-01

4.5E-01

2.2E-02

3.5E+00

2.2E+00

2.2E-02

6.9E+00

4.5E+00

2.2E-02



Mouse

Female

4.3E-01

3.0E-01

3.4E-02

2.1E+00

1.5E+00

3.4E-02

4.3E+00

3.0E+00

3.4E-02

Male

1.5E-01

1.1E-01

9.4E-02

7.7E-01

5.4E-01

9.4E-02

1.5E+00

1.1E+00

9.4E-02

NTP

(2006,

197605V

Rat

Female

1.1E+00

7.8E-01

1.3E-02

4.8E+00

3.6E+00

1.4E-02

8.2E+00

6.6E+00

1.5E-02

Delia Porta
et al.

(1987,

197405V

Mouse

Male

7.1E+00

1.2E+00

8.5E-03

1.4E+01

5.0E+00

1.0E-02

2.0E+01

9.7E+00

1.0E-02

Female

2.3E+00

1.5E+00

6.5E-03

1.0E+01

6.8E+00

7.3E-03

2.1E+01

1.4E+01

7.1E-03

Toth et al.,
(1979

197109V

Mouse

Male

3.7E-01

2.1E-01

4.8E-02

1.9E+00

1.1E+00

4.7E-02

3.9E+00

2.2E+00

4.6E-02

aCombined tumors, Bayesian analysis.
bHepatocellular carcinomas for both males and females,
hepatocellular carcinomas.

TCDD blood concentrations from Emond rodent PBPK models.
SF = BMR BMDLbmr, where BMR = 0.01, 0.05, or 0.10.

-------
Table 5-19. TCDD human-equivalent dose (HED) BMDs, BMDLs, and oral slope factors (OSF) for 1, 5, and 10%
extra risk

Study

Species

Sex

BMDoi
(ng/kg-d)

BMDLoi
(ng/kg-d)

OSF01
(ng/kg-d) 1

bmd05

(ng/kg-d)

bmdl05

(ng/kg-d)

osf05

(ng/kg-d) 1

BMD10
(ng/kg-d)

BMDL10
(ng/kg-d)

OSF10
(ng/kg-d) 1

Kociba

(1978,

001818V

Rat

Female

1.1E-02

7.4E-03

1.4E+00

1.3E-01

8.6E-02

5.8E-01

3.8E-01

2.59E-01

4.0E-01

Male

5.9E-02

3.1E-02

3.3E-01

6.6E-01

3.6E-01

1.4E-01

1.8E+00

9.7E-01

1.0E-01

NTP

(1982,

594255V

Rat

Female

9.7E-03

5.8E-03

1.7E+00

1.1E-01

6.6E-02

7.6E-01

3.2E-01

1.9E-01

5.2E-01

Male

1.9E-02

9.7E-03

1.0E+00

2.2E-01

1.1E-01

4.5E-01

6.2E-01

3.3E-01

3.1E-01



Mouse

Female

9.1E-03

5.4E-03

1.9E+00

1.1E-01

6.0E-02

8.3E-01

3.0E-01

1.8E-01

5.7E-01

Male

1.9E-03

1.2E-03

8.3E+00

2.2E-02

1.3E-02

3.8E+00

6.4E-02

3.8E-02

2.7E+00

NTP

(2006,

197605V

Rat

Female

4.1E-02

2.3E-02

4.4E-01

3.6E-01

2.4E-01

2.1E-01

7.9E-01

5.7E-01

1.8E-01

Delia Porta
et al.

(1987,

197405V

Mouse

Male

5.2E-01

3.1E-02

3.2E-01

1.7E+00

3.8E-01

1.3E-01

2.8E+00

1.0E+00

1.0E-01

Female

9.2E-02

4.9E-02

2.0E-01

1.1 E+00

6.0E-01

8.3E-02

2.9E+00

1.7E+00

5.9E-02

Toth et al.
(1979,

197109V

Mouse

Male

5.1E-03

1.9E-03

5.3 E+00

6.7E-02

2.7E-02

1.9E+00

2.0E-01

8.5E-02

1.2 E+00

2! ^

H S.
W K-

oy

o

c
o

H
W

aCombined tumors, Bayesian analysis.
bHepatocellular carcinomas for both males and females,
hepatocellular carcinomas.

HEDs from Emond human PBPK model corresponding to blood concentration BMDs and BMDLs in Table F3-1.
OSF = BMR BMDLbmr, where BMR = 0.01, 0.05, or 0.10.

-------
Table 5-20. Illustrative RfDs based on tumorigenesis in experimental animals

Study

Species, strain
(sex)

Protocol

Endpoint

BMDLhed3
(ng/kg-day)

RID1'

(mg/kg-day)

NTP (1982,
594255s)

Mouse, B6C3F1,
male

2-year gavage;
n = 50

Liver adenoma and carcinoma, lung

1.1E-3

3.6E-11

Toth et al.
(1979. 197109s)

Mouse, Swiss/
H/Riop, male

1-year gavage
(1-year average);
n = 38-44

Liver tumors

1.9E-3

6.3E-11

NTP (1982,
594255s)

Mouse, B6C3F1,
female

2-year gavage;
n = 50

Liver adenoma and carcinoma, thyroid adenoma,
subcutaneous fibrosarcoma, all lymphomas

5.3E-3

1.7E-10

NTP (1982,

594255s)

Rat, Osborne-
Mendel, female

2-year gavage;
n = 50

Liver neoplasitc nodules, thyroid follicular cell adenoma,
liver adenoma and carcinoma, adrenal cortex adenoma or
carcinoma

5.7E-3

1.9E-10

Kociba et al.
(1978. 001818s)

Rat, S-D, female

2-year dietary;
n = 50

Liver adenoma carcinoma, oral cavity, lung

7.3E-3

2.4E-10

NTP (1982,
594255s)

Rat, Osborne-
Mendel, male

2-year gavage;
n = 50

Thyroid follicular cell adenoma, adrenal cortex adenoma

9.6E-3

3.2E-10

Delia Porta et al.
(1987. 197405s)

Mouse, B6C3F1,
male

1-year gavage;
n = 40-50

Hepatocellular carcinoma

3.1E-02

1.0E-9

NTP (2006,

197605s)

Rat, S-D, female

2-year gavage;
n = 53

Liver cholangiocarcinoma, hepatocellular adenoma, oral
mucosa squamous cell carcinoma, lung cystic keratinizing
epithelioma, pancreas adenoma, carcinoma

3.1E-2

1.0E-9

Kociba et al.
(1978. 001818s)

Rat, S-D, male

2-year dietary;
n = 50

Adrenal cortex adenoma, tongue carcinoma, nasal/palate
carcinoma

3.1E-2

1.0E-9

H J

H S.
W K-

oy

o

c
o

H
W

aBMR = 0.01.

bUF = 30; UFa = 3, UF„ = 10.

-------
Table 5-21. Illustrative RfDs based on hypothesized key events in TCDD's MOAs for liver and lung tumors

Key event

Endpoint and exposure duration

NO(A)ELhed
(ng/kg-day)

LO(A)ELhed
(ng/kg-day)

BMDLhed3
(ng/kg-day)

RID1'

(mg/kg-day)

Study

Liver tumors

Changes in gene expression

CYP1A1 mRNA,
1 day

1.8E-05

3.4E-04

2.3E-030
(Appendix H)

6E-13d'e

Vanden Heuvel et al.
(1994.594318)

Changes in gene expression

Benzo(a)pyrene hydroxylase (BPH)
activity (CYP1A1), 1 day

9.2E-04

6.0E-03

4.6E-04cd

(Appendix H)

2E-llde

Kitchin and Woods
(1979. 198750)

EROD (CYP1A1), 53 weeks

none

1.4E-01

9.5E-03C

(Appendix H)

3E-10e

NTP (2006. 197605)

Oxidative stress

DNA single-strand breaks,
90 days

none

3.3E-02

2.2E-02C

(Appendix H)

7E-10e

Hassoun et al. (2000,

197431)

TBARS, 90 days

—

—

4.4E-02

(Appendix H)

2E-09e

Hassoun et al. (2000,

197431)

Cytochrome C reductase, 90 days

—

—

8.8E-02

(Appendix H)

3E-09e

Hassoun et al. (2000,

197431)

Hepatotoxicity

Toxic hepatopathy,
2 years

none

1.4E-01

1.8E-010
(Appendix E)

5E-09f

NTP (2006. 197605)

Hepatocyte hypertrophy, 31 weeks

9.3E-02

3.3E-01

8.8E-03

(Appendix E)

3E-10e

NTP (2006. 197605)

Hepatocellular proliferation

Labeling index,
31 weeks

none

1.4E-01

6.6E-020

(Appendix H)

2E-09e

NTP (2006. 197605)

Lung tumors

Metabolic enzyme induction

EROD (CYP1A1), 53 weeks

none

1.4E-01

2.9E-040

(Appendix H)

1E-Ile

NTP (2006. 197605)

Retinoid homeostatsis

Hepatic retinol and retinyl
palmitate, 90 days

none

1.1E+00

1.7E-010

(Appendix E)

6E-09e

Van Birgelen et al.
(1995. 198052)

aBMR for continuous endpoints—1 standard deviation; for quantal endpoints—10%.

bBolded NOAEL, LOAEL, or BMDL is selected POD; poorly-fitting BMDLs above the LOAEL not used.

°Poor BMD model fit or no good model fit.

dCouldbe higher depending on the effect of background exposureoor (see Section 5.3.2.1).

eUF = 30; UFa = 3; UFH = 10.

fUF = 300; UFA = 3; UFH = 10; UFL = 10.

-------
1	Table 5-22. Comparison of principal epidemiological studies

2

Strengths

Weaknesses

Study

Cumulative TCDD levels in the serum were
estimated on an individual-level basis for
the 3,538 workers.

Evaluated effect of lag periods (0 and 15
years).

Measured and back-extrapolated TCDD
concentrations to refine and quantify job
exposure matrices, which were then used to
estimate dioxin cumulative dose for each
member of their entire cohort.

Internal cohort comparisons (Cox regression
model).

Background exposure estimated.

•	Exposure to other chlorinated hydrocarbons
(dioxin like compounds).

•	Extrapolation of dose from a small subset
(roughly 5%, n = 170) of the cohort.

•	Serum fat or body fat levels of TCDD were
back-calculated using a simple first-order
model. Half-life of TCDD is variable but
simulated as a constant. Changes in the lipid
fraction of body weight or presence/absence
of genetic differences in humans that alter the
distribution and metabolism of TCDD were
not considered.

•	Serum lipid levels of TCDD in 1988 were
measured only at one of the eight plants in
the study. No follow-up measures. The
estimates of dose are based on blood samples
taken decades after exposure.

NIOSH cohort
Steenland et al.
(2001. 197433)

Cumulative TCDD levels in the serum were
estimated on an individual-level basis for
the 3,538 workers.

TCCD dose estimates were simulated with a
kinetic model that included considerations
of exposure intensity and age-dependent
body weight and fat levels.

Evaluated effect of lag periods (0 and 15
years).

Background exposure estimated.

Stratified risk estimates for smoking and
nonsmoking.

Race and age adjustments.

Internal cohort noted an inverse-dose
response for high-exposure groups and thus
excluded the data resulting in stronger
associations.

•	Extrapolation of dose from a small subset
(roughly 5%, n = 170) of the cohort.

•	The authors reported the CADM model
provided an improved fit over the one-
compartmental model, but no evidence was
reported regarding any formal test of
statistical significance.

•	Serum lipid levels of TCDD in 1988 were
measured only at one of the eight plants in
the study. No follow-up measures. The
estimates of dose are based on blood samples
taken decades after exposure.

•	Exposure to other chlorinated hydrocarbons
(dioxin like compounds).

•	No consideration for recent exposures to
TCDD, changes in the lipid fraction of body
weight or presence/absence of genetic
differences in humans that alter the
distribution and metabolism of TCDD could
cause misclassification.

NIOSH cohort
Cheng et al.
(2006. 523122)

This document is a draft for review purposes only and does not constitute Agency policy.

5-110 DRAFT—DO NOT CITE OR QUOTE

-------
1

Table 5-22 Comparison of principal epidemiological studies (continued)

Strengths

Weaknesses

Study

•	Repeated TCDD measures in serum in 48
individuals. Used to estimate half-life for
study cohort. Took into account the age
and body fat percentage of the workers.
Measured and back-extrapolated TCDD
concentrations to quantify exposures for the
remaining cohort members using 5 different
working areas of the plant

•	Evaluated effect of lag periods up to 20
years.

•	Multiple statistical models used to evaluate
fatal cancer slope estimates.

•	Background exposure estimated.

•	Exposure to other chlorinated hydrocarbons
(dioxin like compounds), HCH, and lindane.

•	Extrapolation of dose from a small subset
(roughly 4%, n = 1,189) of the cohort.

•	Serum fat or body fat levels of TCDD were
back-calculated using a simple first-order
model. Presence/absence of genetic
differences in humans that alter the
distribution and metabolism of TCDD were
not considered.

•	Serum lipid levels of TCDD for only 275
workers.

Becher et al.
(1998. 197173);
Hamburg
Cohort

•	Both internal and external analyses.

•	Adjustment for age, BMI, and smoking.

•	Both cancer incidence and cancer mortality
data available, although results somewhat
discordant, with steeper dose-response seen
for cancer mortality.

•	Acute dose due to accident may not be
comparable to chronic dose accumulated over
a long time, as in most environmental
exposures.

•	Relatively small number of cancer deaths
compared to NIOSH and Hamburg cohorts
(n = 31).

•	Serum TCDD levels measured 30 years after
accident, requiring extrapolation back in time
to estimate cumulative dose over time.

•	Serum TCDD levels measured only on a
sample of the cohort (138 out of 243),
requiring assumptions about similarities in
exposure scenario for other workers to
estimate their exposure

Ott and Zober
(1996. 198408)

This document is a draft for review purposes only and does not constitute Agency policy.

5-111 DRAFT—DO NOT CITE OR QUOTE

-------
Table 5-22 Comparison of principal epidemiological studies (continued)

Strengths

Weaknesses

Study

•	TCDD levels measured in all 891 members
of this female cohort.

•	Most TCDD measurements based on
observed levels in stored serum at the time
of the accident in 1976, no extrapolation
needed to estimate past levels.

•	Internal analyses.

•	Evaluates female cancer incidence, other
studies evaluate male cancer mortality.

•	Presumed adjustment for age and potential
breast cancer confounders (15 of 21 cancers
were breast cancer).

•	Acute dose due to accident may not be
comparable to chronic dose accumulated over
a long time, which is typical of most
environmental exposures.

•	Did not evaluate different lag periods.

•	Not clear if any adjustment for confounders.

•	Small number of cancers (n = 21).

•	Doses known in 1976, require assumptions
about excretion over time to estimate
cumulative dose (9 year half life assumed),
presumed metric of primary interest. No
more recent TCDD concentration data used.

•	Reported logio transformation of the exposure
estimates in their regression analysis.

Warner et al.
(2002. 197489)

1

2

3	Table 5-23. Added background TEQ exposures to blood TCDD/TEQ

4	concentrations in ratsa

5

Background TEQ added

None

Est. TCDD onlyb

Est. TEQC

2x Est. TEQd

10x Est. TCDDe

0

0.064

0.19

0.38

0.64

2.56

2.62

2.75

2.94

3.20

5.69

5.75

5.88

6.07

6.33

9.79

9.85

9.98

10.1

10.5

16.6

16.7

16.8

17.0

17.2

29.7

29.8

29.9

30.1

30.3

6

7	"Background exposures estimated from NTP (2006, 543749); rat TCDD concentrations from NTP (2006, 197605)').

8	' Estimated from TCDD fat concentration measurements in NTP (2006, 543749).

9	"Estimated from combined TCDD. PeCDF. and PCB-126 fat concentration measurements in NTP (2006, 543749).

10	dAssumes that measured congeners comprise 50% of actual TEQ exposure.

11	"Assumes that TCDD comprises 10% of total background TEQ exposure.

This document is a draft for review purposes only and does not constitute Agency policy.

5-112 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table 5-24. Effect of added background TEQ exposure on BMDL0i for

2	cholangiocarcinomas in rats (NTP, 2006,197605)

3

Background TEQa

Added exposure
(ng/kg blood TEQ)

BMDL0ib
(ng kg blood)

Nonec

0

4.14

Est. TCDD only

0.064

4.19

Est. TEQ

0.19

4.30

2x Est. TEQ

0.38

4.45

10x Est. TCDD

0.64

4.65

4

5	Scenarios as in Table 5-20.

6	'Multistage model results from BMDS version 2.1.1 (see Appendix I for modeling details).

7	°Same result as for the single tumor modeling presented previously in this section.

8

9

10	Table 5-25. NIOSH cohort septile data with added TEQ background3

11

Septile

TCDD serum level

(ppt-yr)

TCDD + background TEQ
(ppt-yr)

Relative increase
(%)

1

260

2,960

1,040

2

402

3,102

770

3

853

3,553

320

4

1,895

4,595

140

5

4,420

7,120

60

6

12,125

14,825

20

7

59,838

62,538

5

12

13	'Septile data from Steenland et al. (2001, 1974331: cumulative background TEQ estimate from Crump et al.

14	(2003, 1973841: both based on estimates by WHO (1998).

This document is a draft for review purposes only and does not constitute Agency policy.

5-113 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

Cell-type specific
factors (kinases,
cofactors etc.)

TCDD



AHR

Post-translational
modification

Binding to other
transcription factors
(Rb, ERa)

Altered
Cell Cycle
Regulation

Altered
Metabolism

Figure 5-1. Mechanism of altered gene expression by AhR. The regulation of
gene expression by TCDD in mammalian cells requires binding of the xenobiotic
to the aryl hydrocarbon receptor (AhR). The AhR is part of a multi-protein
complex that includes heat shock proteins and various kinases and other post-
translational modifying factors. Upon ligand binding, the AhR heterodimerizes
with the aryl hydrocarbon receptor nuclear translocator (Arnt) and binds to dioxin
response elements (DREs) found in target genes. Alternatives to DRE-dependent
gene expression exist whereby the AhR complex associates with other
transcription factors and results in a cross-talk between these systems. The
culmination of regulation of AhR targets genes (both increases ad decreases in
transcription) results in an alteration in cellular phenotypes, including changes in
intracellular metabolism and changes in cell cycle regulation.

This document is a draft for review purposes only and does not constitute Agency policy.

5-114 DRAFT—DO NOT CITE OR QUOTE

-------
Liver

Lung

Thyroid

TCDD

I

AhR

TCDD

J

AhR

I

I

Co-carcinogens

Changes in
Gene Expression

M

Oxidative
Stress

Hepatotoxicity 		

Cellular
Proliferation

~

Adenoma

and
Carcinoma

Changes in
Gene£flflgg>ion

Metabolic

Retinoid	Enzymes

Homeostasis	(Cyps, COX2)

¦		 Toxicity 	

'*• Proliferation

~

Adenoma

and
Carcinoma

Liver

Decreased T4

i

Increased-fTSH

	|	>

Proliferation

Thyroid

Adenoma

and
Carcinoma

Figure 5-2. TCDD's hypothesized modes of action in site-specific carcinogenesis. See text for details. In each
instance, the solid arrows depict pathways that are well-established and are associated with low uncertainty. The
dashed arrows represent connections that are less established and are associated with higher uncertainty.

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

Figure 5-3. EPA's process to select and identify candidate OSFs from key
animal bioassays for use in the cancer risk assessment of TCDD.

For each cancer study that qualified for TCDD dose-response assessment using the study inclusion criteria,
EPA first selected the species/sex/tumor combinations with statistically significant increases in tumor
incidence by either a pair-wise test between the treated group and the controls or by a trend test showing
increases in tumors with increases in dose. Next, EPA used an animal kinetic model to estimate blood
concentrations corresponding to the study average daily administered doses for use in dose response
modeling. BMDL0i's were then estimated for the blood concentrations by, (1) using the linearized
multistage model for each species/sex/tumor combination within each study, and (2) using the linearized
multistage model within a Bayesian Markov Chain Monte Carlo framework that assumes independence of
tumors and modeling all tumors together for each species/sex combination within each study. Using the
human kinetic model, human equivalent doses (BMDLheDs) were then estimated for each of the BMDL„iS
and oral slope factors were calculated by OSF = 0.01/BMDLhed- The lowest OSF for a species/sex
combination for either a single tumor type or all tumors combined was selected as a candidate OSF for
TCDD risk assessment.

This document is a draft for review purposes only and does not constitute Agency policy.

5-116 DRAFT—DO NOT CITE OR QUOTE

-------
Supralinear(Weibull): Pr(d)=l-exp(-dA0.7)

0.2	0.4	0.6	0.8	1

Dose

Dose

Probit model: Sublinear, NOT ZS@S:



0.16



0.14-

(LI



m



C

0.12

O

&



n



(L)



u

HH

0.1

o



£





0.08

IB



CO



o

0.06

t,



P-.





0.04

0	0.2	0.4	0.6	0.8	1

Dose

1

2

3	Figure 5-4. Dose-response model shape

This document is a draft for review purposes only and does not constitute Agency policy.

5-117 DRAFT—DO NOT CITE OR QUOTE

-------
Individuals with supralinear above threshold, and population DR curve

1

2

3

4

5

0.04

tu

H

s

Q

a

to

u 0.03

£ 002
¦

"S

p 0.01

H

Ph

0	0.02	0.04	0.06	O.0B	0.1

Individuals with linear above threshold, and population DR curve

0.1

u
n

5
~

6
ai
a
u

0.08

0.06

^ 0.04

¦a

-D
~
I-

0.02

Dose

Figure 5-5. Comparison of individual and population dose-response curves;
a simple illustration.

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

5-118 DRAFT—DO NOT CITE OR QUOTE

-------
A. Full response range

Multistage Cancer Model with 0.95 Confidence Level

dose

09:48 04/22 2010

B. Low-dose region

Cholangiosarcoma low dose

Blood Cone (ng/kg)

Figure 5-6. Multistage benchmark dose modeling of NTP (2006, 197605)
cholangiosarcoma data.

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

5-119 DRAFT—DO NOT CITE OR QUOTE

-------
A. Full response range

Composite Risk of Dioxin in NTP (2006a) female rats

EXPERIMENTAL DOSE RANGE

B. Low-dose region

Composite Risk of Dioxin in NTP (2006a) female rats around BMD01

DOSE

Figure 5-7. Multistage benchmark dose modeling of NTP (2006, 197605)
combined tumor data.

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

5-120 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

A.

Kohn and Melnick (2002) Figure 5 on log Scale (KR.X=3300)

B.

Kohn and Melnick (2002) Figure 5 on Arithmetic Scale (KR.X=3300)

i	r

20	30

Concentration X in nM

5.0	10.0

log 10 Concentration X in nM

Figure 5-8. Estrogen receptor-mediated response-modeling plot from Kohn
and Melnick (2002,199104): low-dose region shown.

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

5-121 DRAFT—DO NOT CITE OR QUOTE

-------
Liver

TCDD

1

AhR

I

Changes in

gene ex

oression

Oxidative
stress

Hepatotoxicity***'

^•....^Hepatocellular
proliferation

CYP1A1 mRNA, 1 day
Vanden Heuvel etal., 1994

EROD (CYP1A1), 53 weeks
NTP, 2006

TBARS, 90 days
Hassoun et al., 2000

Toxic hepatopathy, 2 years
NTP, 2006

Labeling index, 31 weeks
NTP, 2006

Adenoma

and
carcinoma

Figure 5-9. Representative endpoints for each of the hypothesized key events
following AhR activation for TCDD-induced liver tumors.

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

5-122 DRAFT—DO NOT CITE OR QUOTE

-------
Lung

1

2

3

4

5

Hepatic retinol, 90 days
Van Birgelen et al., 1995a

TCDD

I

AhR

Cocarcinogens

Changes in

gene ex

— Retinoid
homeostatsis

pression

Metabolic

enzymes
(Cyps, COX2)

•~Tox

city*	

EROD (CYP1A1), 53 weeks
NTP, 2006

'^Proliferation

Adenoma

and
carcinoma

Figure 5-10. Representative endpoints for two hypothesized key events
following AhR activation for TCDD-induced lung tumors.

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

5-123 DRAFT—DO NOT CITE OR QUOTE

-------
Cancer Slope Factors for 2,3,7,8-TCDD

to

3?
>!

§•
ss



>}
&

?5'

^3

ss

*3

o

O
£

§•
>j

s
. o

ri

o
s



O

o

z;
o^

H S.

w .s-
o *
p

o
c
o

H

W

1.0E+07

Human

u
ro

8 "D
a. '

O 0>

§ !
O 
cm"

1.0E+06

1.0E+05

Mouse

«#

n>

*V		„o,	4>





<£>

o* ^

¦c^	-C^"

r#	#

O	0^	<2>N

c#1 >

£> ^
*> N

^ ->v

_oJJ



.eT1 ^ \S?	\g?	sg^ " sg?	\S?	vS?

^ <# <# <# <# <# <# <# <# <#
^	riV	^	_v

£* S

&

K

ST

K-

rV0 -£



&

> *><



;S>



f^**1

r>

9f

<* 

&

v

jy

/•

>v

Sf

9f

Figure 5-11. Candidate oral slope factor array.

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

6. FEASIBILITY OF QUANTITATIVE UNCERTAINTY ANALYSIS
FROM NAS EVALUATION OF THE 2003 REASSESSMENT

6.1. INTRODUCTION

This section focuses on the third area for improvement in the 2003 Reassessment that was
identified by the National Academy of Sciences (NAS) review committee (NAS, 2006, 198441),
i.e., improving transparency, thoroughness, and clarity in quantitative uncertainty analysis.
Although the NAS committee summarized the shortfalls in the 2003 Reassessment categorically,
the elaborations within their report often contain the qualification "if possible" and do not take a
position with regard to the feasibility of many of its suggestions. With appreciation for the
extent of information available for dioxin, the goal of this section is to circumscribe the
feasibility of a data-driven quantitative uncertainty analysis for TCDD dose-response
assessment. Following brief highlights of the evolution of quantitative uncertainty analysis for
such applications, this section lays out definitions of key terms, reviews EPA's position
regarding cancer and noncancer endpoints, summarizes the NAS critique, and evaluates the
feasibility of quantitative uncertainty analysis for TCDD within the framework of EPA's
noncancer RfD and cancer slope factor dose-response methodologies.

6.1.1. Historical Context for Quantitative Uncertainty Analysis

The basic methods of probabilistic risk assessment (PRA) were developed in the
aerospace program in the 1960s, and they found their first full-scale application in the
U.S. Nuclear Regulatory Commission's (U.S. NRC's) Reactor Safety Study of1975—including
accident consequence analysis and uncertainty analysis (U.S. NRC, 1975, 543729). This study,
commonly referred to as the Rasmussen Report after its lead author, is considered to be the first
modern PRA. In the aftermath of the 1979 Three Mile Island accident, a new generation of
PRAs appeared in which some of the methodological problems of the 1975 study were avoided.
These advances were reflected in the Commission's Fault Tree Handbook (U.S. NRC, 1981,
543730) and PRA guide (U.S. NRC, 1983, 543732), which shored up and standardized much of
the risk assessment methodology. An extensive chapter of the latter was devoted to uncertainty
and sensitivity analysis. These documents formed the basis for standards and guidelines

This document is a draft for review purposes only and does not constitute Agency policy.

6-1 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

established by other agencies, including the U.S. Department of Energy (U.S. DOE, 1992,
543733) and National Aeronautics and Space Administration (NASA, 2002, 543734).

In 1991, a set of U.S. NRC studies known as NUREG 1150 used structured expert
judgment to quantify uncertainty and set new standards for uncertainty analysis, in particular
with regard to expert elicitation (U.S. NRC, 1991, 543736). This was followed by a joint
U.S.-European Union (EU) program for quantifying uncertainty in accident consequence models.
Expert judgment methods were further elaborated in those evaluations, as well as screening,
dependence modeling and sensitivity analysis (EC, 2009, 543738). Studies building off of this
work have performed a large-scale uncertainty analysis of European consequence models and
provided extensive guidance on identifying important variables; selecting, interviewing and
combining experts; propagating uncertainty; inferring distributions on model parameters; and
communicating results, as documented by Goossens et al. (1996, 548727; 1997, 543752; 1998,
548726; 2001, 548730; 2001, 548731; 2001, 548732; 2001, 548735; 2001, 548737; 2001,
548738; 2001, 548734) and others (Brown et al., 1997, 543739; Harper et al., 1995, 202317;
2002, 198124).

The National Research Council (NRC) has been a persistent voice in urging the
government to enhance its risk assessment methodology beginning with its report on risk
assessment in the federal government (NRC, 1983, 194806). The Council"s 1989 report.
Improving Risk Communication, inveighed against minimizing the existence of uncertainty and
noted the importance of considering the distribution of exposure and sensitivities in a population
(NRC, 1989, 000858). The issue of uncertainty was a clear concern in subsequent reports,
including those assessing human exposure to airborne pollutants (NRC, 1991, 037823). Building
on its evaluation of Issues in Risk Assessment (NRC, 1993, 078637). the landmark study Science
and Judgment in Risk Assessment (NRC, 1994, 006424) gathered many of these themes in a plea
for quantitative uncertainty analysis as "the only way to combat the false sense of certainty
which is caused by a refusal to acknowledge and (attempt to) quantify the uncertainty in risk
predictions." A subsequent report, Estimating the Public Health Benefits of Proposed Air
Pollution Regulations (NRC, 2002, 035312). identified three barriers to the broad acceptance of
recent EPA health benefit analyses: (1) the large amount of uncertainty inherent in these
analyses, (2) the manner in which EPA deals with this uncertainty, and (3) "... projected health
benefits are often reported as absolute numbers of avoided death or adverse health outcomes

This document is a draft for review purposes only and does not constitute Agency policy.

6-2 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

without a context of population size or total numbers of outcomes." The Council encouraged
EPA to "explore alternative options for incorporating expert judgment into its probabilistic
uncertainty analyses."

In an early 2009 report, Science and Decisions: Advancing Risk Assessment, the NRC
committee on improving risk analysis encouraged EPA to harmonize approaches for cancer and
noncancer dose-response assessment (NRC, 2009, 194810). which involves uncertainty issues
discussed in this section. Even more recently, EPA released a draft white paper, Using
Probabilistic Methods to Enhance the Role of Risk Analysis in Decision Making (U. S. EPA,
2009, 522927). Although not focused specifically on quantitative uncertainty analysis, there is
overlap with the issues treated here, and relevant insights are anticipated from ongoing efforts in
this area.

6.1.2. Definition of Terms

For purposes of this study, the following definitions are adopted:52

Uncertainty Characterization. This consists of a Structured Uncertainty Narrative and, if
the uncertainty is supported by quantitative models, Quantitative Uncertainty Analysis.

Structured Uncertainty Narrative. This identifies the assumptions conditional on which
uncertainty is to be characterized and delineates the type of arguments with supporting
evidence that buttress these assumptions.

Quantitative Uncertainty Analysis. This is a quantification of the uncertainty attending
the use of quantitative models. It applies to a mathematical model of physical
phenomena, some of whose parameter values are not known with certainty. A joint
distribution is assigned to uncertain model parameters and propagated through the model
to yield a joint distribution over the model output. Thus, a quantitative uncertainty
analysis always has a joint distribution over model outputs as its result.

Joint Distribution/Marginal Distribution. For a set of uncertain quantities, a joint
distribution is an assignment of probabilities (or probability densities) for each possible
combination of values of these quantities. Each uncertain quantity has a marginal
distribution, that is, an assignment of probabilities (or probability densities) to each
possible value of that quantity. Assigning a marginal distribution to each quantity is not
equivalent to assigning a joint distribution to the set of quantities, unless the quantities
are independent; in this case the joint distribution is just the product of the margins.

52Many of these definitions are standard terms in probability and statistics, as described in Saltelli et al. (2000,
543756). Cox (2006, 5943421. Kurowicka and Cooke (2006, 543758). and NRC (2007, 543748); some are reflected
in current Agency practice (U.S. EPA. 2009, 522927).

This document is a draft for review purposes only and does not constitute Agency policy.

6-3 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

Qualitative/Informal Uncertainty Analysis. This assembles the arguments and evidence
and provides an assessment of their plausibility in terms of verbal modifiers. The
meaning of verbal modifiers such as "likely/unlikely" or "plausible/implausible" in the
natural language53 is indeterminate and context dependent. The way in which these
qualifiers combine in the natural language requires critical attention from a quantitative
viewpoint. (For example, if A is likely and B is likely and C is likely, is A and B and C
likely?) It is sometimes claimed that the probability formalism does not capture the way
people reason with uncertainty, and many alternatives have been proposed.54

This is not the place to discuss foundational issues, except to remark that the practitioner
wishing to depart from the standard probability formalism should carefully explore the
whole range of alternatives and critically examine the operational meaning of the
primitive notions.

Sensitivity Analysis. If a quantitative model uses "nominal values" (approximations of the
real values) for various input parameters, a sensitivity analysis is performed by choosing
different values for these parameters and re-running the model to assess the impact of
changes in these parameters on model output. Applicable methods include one- and
two-at-a-time methods, design of experiments and Morris's method (Saltelli et al., 2000,
543756). They aim at estimating first- and perhaps higher-order effects with a minimal
number of model runs, by systematically varying the nominal values. In large
uncertainty analyses, sensitivity analysis is used to screen variables for in-depth
uncertainty quantification, and thus is part of a quantitative uncertainty analysis
(Kurowicka and Cooke, 2006, 543758). As a note, the NAS committee report (NRC,
2006) does not distinguish between uncertainty and sensitivity analysis. In fields which
have not developed a tradition in uncertainty quantification, the spread of values
generated by a sensitivity analysis is sometimes presented as a representation of
uncertainty (Murphy et al., 2004, 543741). The question of whether this is or is not the
case is moot so long as the uncertainty on model input parameters is not quantified.
Systematically varying input values is not the same as sampling input parameter values
from their uncertainty distributions. In any event, a systematic approach to parameter
variation is essential; simply choosing a few values of interest and generating different
output is of limited scientific benefit and inevitably raises questions of selection bias.

That said, if alternative values are commonly used and therefore recommend themselves,
then running these through the models can help sensitize users to parameter variations
and their impacts on model outputs.

53Natural language denotes any discourse in which the meaning of the words is not formalized; rather, these words
are just "as they come in off the street" with whatever meaning a participant may give them.

54Before the advent of personal computers, various shorthand techniques were developed for computing system risk.
In control theory, schemes of 'interval probabilities' were proposed which could be propagated through a system to
yield bounds on system reliability. Whereas these bounds originally reflected accuracy of shorthand approximations
of complex formulae, their offspring have been proposed as quantifications of uncertainty. Alternative notions of
uncertainty are also proposed with the goal of simplifying the assessment and computational burden or capturing
putative features of uncertainty which are overlooked in probability theory. These include possibility theory, fuzzy
numbers, qualitative algebra, imprecise probabilities, belief functions, certainty factors, and the like. Nonmonotonic
reasoning systems attempt to capture reasoning about knowledge, or reasoning from partial knowledge; they include
default logic, defeasible logic, abductive logic, and autoepistemic logic, to name a few.

This document is a draft for review purposes only and does not constitute Agency policy.

6-4 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

Cognitive Uncertainty. This concerns uncertainty regarding what is the case. Not
knowing "what is the case" may be conceived as uncertainty over the set of all
possibilities, sometimes expressed as 'uncertainty over the set of possible worlds.'
Uncertainty over possible worlds may be represented formally as probability; that is, the
uncertainty of a given situation is represented as a number between zero and one, and the
uncertainty of either of two mutually exclusive situations is the sum of the uncertainties
of each situation.55 Two interpretations or operationalizations of the probability
formalism are current: the objective or frequentist interpretation and the subjective or
Bayesian interpretation. These interpretations are not mutually exclusive, as subjective
probabilities can and often do track relative frequencies.

Volitional Uncertainty. This concerns uncertainty regarding what to do. In the natural
language, being unsure which course of action to choose is also called "uncertainty."
Insofar as uncertainty on the best course of action can be translated into a claim about the
state of the world, volitional uncertainty can be translated into cognitive uncertainty. For
example, a regulatory body charged with setting a speed limit is obliged to make a
decision. The decision may be cautious or reckless, well or poorly motivated, wise or
foolish; but it cannot be true or false. Since the decision makes no claim about the state
of the world, it cannot be uncertain in the cognitive sense. The uncertainty cannot be
analyzed by sampling from some distribution. However, if the decision is based on the
claim that the chosen speed limit minimizes accidents while maintaining a prescribed
traffic volume, that claim may be uncertain and may be subjected to quantitative
uncertainty analysis. A discretionary decision of a regulatory body may entrain cognitive
uncertainty, but it becomes amenable for quantitative uncertainty analysis only when it is
linked to a claim about the state of the world.

Aleatoric/Epistemic Uncertainty. This terminology has become standard in the technical
uncertainty analysis literature, and it has been called Variability/Uncertainty in some
areas, particularly dealing with human populations. A variable whose uncertainty is
aleatoric for a given population takes different, uncertain, values for each member of the
population. If its uncertainty is epistemic, it takes the same uncertain value for all
members of the population. Issues involving uncertainty and variability or epistemic and
aleatory uncertainty translate into issues of dependence, when conducting a quantitative
uncertainty analysis (see Section 6.1.3.3). In its Science and Judgment report, NRC
(1994, 006424) correctly remarks that "the amount of variability is generally itself an
uncertain parameter." It is natural to ask whether a given uncertainty is aleatoric or
epistemic, whereas it is awkward to ask whether this uncertainty is uncertain or
variable—which explains the preference for the epistemic/aleatoric terminology.

55These are known collectively as Kolmogorov's probability axioms. The additivity of probability for exclusive
alternatives states, e.g., that the probability of an unseen object being red or green is the sum of the probability that it
is red and the probability that it is green. This of course assumes that "red" and "green" are clearly defined, such
that nothing can be simultaneously red and green. Many alternative representations of uncertainty contest this
additivity property.

This document is a draft for review purposes only and does not constitute Agency policy.

6-5 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

6.1.3. Key Elements of a Quantitative Uncertainty Analysis

The uncertainty propagation can be performed by some rough estimation, as for example
the delta method (Oehlert, 1992, 543742). or in rare cases it can be performed analytically, as in
simple error propagation.56 Most often, however, it will be performed using Monte Carlo
simulation. A joint distribution is assigned to the parameters of a quantitative model and then
propagated through the model by sampling repeatedly from this joint distribution, computing
model output and generating a distribution of model output. Every uncertainty analysis is
conditional on initial assumptions. A "complete" uncertainty analysis is an unattainable goal; the
best that can be done in practice is to identify and motivate the assumptions that are used. This
section is not a how-to guide, but a to-do list of key elements of any quantitative uncertainty
analysis.57

6.1.3.1.	Quantitative Model

The starting point of any quantitative uncertainty analysis is a mathematical model or
procedure for calculating quantities of interest. A structured narrative explains the choice of
quantitative models. If some values of input parameters for this calculation are not known with
certainty, then the question arises: "What is the uncertainty attending the use of this model?"

This is the question a quantitative uncertainty analysis seeks to answer.

6.1.3.2.	Marginal Distributions over Model Parameter

If the model parameters are directly measurable with sampling error, then the sampling
distribution may itself be used in the quantitative uncertainty analysis. If the model parameters
are fit to data that are sampled from a known or hypothesized distribution, then by resampling
this distribution and refitting the model, distributions over the model parameters may be
constructed. Physically-based simulation models, such as pharmacokinetic models or
environmental transport models, may be solved analytically if equilibrium reaction rates (the

56Simple measurement error is often represented by adding a normally distributed random variable with mean zero
to a "true" value. If several measurements are performed in succession, and the errors on each measurement are
assumed to be independent, then the error induced by adding the measurement results is also a normally distributed
random variable whose mean is zero and whose variance is the sum of the variances on the individual
measurements.

57These key elements of quantitative uncertainty analysis are discussed in many publications such as Saltelli et al.
(2000, 5437561. Cox (2006, 5943421. Kurowicka and Cooke (2006, 5437581. NRC (2007, 5437481 and EPA (2009,
5229271.

This document is a draft for review purposes only and does not constitute Agency policy.

6-6 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

transfer coefficients) are constant. If these rates are not constant, as when concentrations are
near saturation levels, then simulating the pharmacokinetics or transport is indicated. Structured
expert judgment has been applied for uncertainty quantification within the engineering
community since the time of the Rasmussen Report (U.S. NRC, 1975, 543729). More recently,
this approach has been "test-driven" by EPA in assessing health effects of fine particulates
(Walker et al., 1999, 198615). and its potential application has been identified in the Agency's
Guidelines for Carcinogen Risk Assessment, commonly referred to as the Cancer Guidelines
(U.S. EPA, 2005, 086237).58

6.1.3.3. Dependence between Parameter Uncertainties: Aleatoric and Epistemic (Uncertainty
and Variability)

Two uncertain quantities are independent if knowledge about one of them does not alter
our uncertainty regarding the other. The quantities are dependent if they are not independent.
The role of dependence modeling in quantitative uncertainty analysis must be addressed. To
illustrate, cigarette smoking and body fat are both thought to influence biomarkers for toxic
response to dioxin exposure, such as ethoxyresorufin-O-deethylase (EROD) activity (Pereg et al.,
2002, 199797). In an individual sampled at random from a target population, both percent body
fat and whether (and how much) he or she smokes are uncertain.59 However, these uncertainties
are not independent, inasmuch as smokers tend to have less body fat (Vanni et al., 2009,

543754).

Issues involving uncertainty and variability, or epistemic and aleatory uncertainty,
translate into issues of dependence when conducting a quantitative uncertainty analysis. For
example, a constant used to estimate the biokinetic behavior of dioxin may be uncertain. If it is
believed to be the same for every member of the population, the uncertainty is termed

58The EPA (2005, 0862371 cancer guidelines state: "In many of these scientific and engineering disciplines,
researchers have used rigorous expert elicitation methods to overcome the lack of peer-reviewed methods and
data...." These cancer guidelines are flexible enough to accommodate the use of expert elicitation to characterize
cancer risks, as a complement to the methods presented in the cancer guidelines. According to NRC (2002,

0353121. the rigorous use of expert elicitation for the analyses of risks is considered to be quality science."

59Because dioxins generally distribute to body fat/lipid, the percent body fat is often used to estimate body burden; a
default value of 25% is common (Connor and Ayhvard. 2006, 1976321. However, body fat percentage varies
widely between individuals, from a minimum essential level (e.g., 2% for men, 10% for women) to obesity (e.g.,
38% or more for men, 42% for women). Considering that current estimates suggest 30% of the U.S. population are
obese, an uncertainty analysis of dioxin risk in this population should sample individuals from their gender/body fat
distribution and correlate this with other known or suspected covariates influencing toxic response (such as diet,
smoking, natural and endogenous ligands, disease, and age).

This document is a draft for review purposes only and does not constitute Agency policy.

6-7 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

"epistemic." In a quantitative uncertainty analysis, this factor would be sampled from its
uncertainty distribution on each Monte Carlo run and used for all members of the population.
Body fat, in contrast, is aleatoric. We do not sample one value from the body fat distribution and
use this value for all members of the population on each Monte Carlo run. Instead we sample a
body fat value for each individual on each run. Because body fat is correlated with other
relevant variables (e.g., smoking, gender, age, and socioeconomic status), all of these variables
should be sampled in a manner that reflects their dependences. Kinetic constants whose
uncertainty is epistemic are completely correlated across individuals: if the value is 0.5 for one
individual, it is 0.5 for everyone. Body fat values vary from individual to individual, and they
are correlated through a host of other variables.

6.1.3.4. Model Uncertainty

All models, being idealizations, are false; on this there is no uncertainty to quantify.
However, the choice of model may constrain the ability to represent uncertainty in observable
phenomena. Thus, in a linear low-dose model, uncertainty over a cancer slope factor may be
quantified, but uncertainty regarding changes in slope at distinct low-dose regimes cannot be
captured. When the model choice imposes severe and potentially unwelcome constraints on
uncertainty quantification, this must be addressed. Distributions over model parameters may be
selected and evaluated based on their ability to reflect uncertainty distributions over observable
phenomena predicted by the models.60 In such cases, the uncertainty propagated through the
quantitative model is not strongly model-dependent. In other cases, multiple model alternatives
may be applied, whose "probability of being the true model" is known or assumed. Since
different models can always be regarded as specializations of more general models, the
distinction between parameter and model uncertainty is sometimes more apparent than real. For
example, as illustrated in the EPA Benchmark Dose Software (BMDS) (U.S. EPA, 2000,
052150s). the multistage and Weibull dose-response models both contain the model Pr(x) = y +
(1 - y) (1 - e ^l x) as a submodel, to which they collapse if other parameters are zero (multistage)
or one (Weibull). Recalling that the function 1/(1 + x) is first-order equivalent to (1 - x) for

60 Such techniques were first used on a large scale in the U.S. NRC-EU joint uncertainty analysis of consequence
models for accidents at nuclear power plants, see Goossens et al. (1996, 548727: 2001, 548737: 2001, 548738: 2001,
548731: 2001, 548732: 2001, 5487351 (Bock et al.. 1998, 5487521.

This document is a draft for review purposes only and does not constitute Agency policy.

6-8 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

small x, the same may be said for logistic models as well. In this case, these models could easily
be parameterized within one family, rendering the choice between them a choice of parameter
values. Similarly, the choice between sub-, supra-, and linear models is sometimes reduced to
parameter estimation within a more general class of model (Hoel and Portier, 1994, 198741).

In other cases, the reduction of model uncertainty to parameter uncertainty is less natural.
For example, according to the "chemoprotection model" of Greenlee et al. (2001, 0154001
dioxin's binding to the aryl hydrocarbon receptor (AhR) inhibits proliferation in tumor cells and
thus suppresses mammary tumors. Dose-dependent protection and cancer induction can both be
true, each applying to different tissues. Although mathematical models exhibiting these twin
features have been suggested (e.g., Kohn and Mel nick, 2002, 199104). these models are not yet
readily estimable from data, and the choice between them is referred to the structured narrative.

6.1.3.5.	Sampling Method

All sampling on a computer is "pseudo random." Significant issues arise in choosing a
method for sampling high-dimensional distributions with dependence. If evaluating the
quantitative model is very time consuming, various "quasi random" schemes may be applied,
including Latin hypercube sampling, importance sampling, and Hammersley sampling. These
methods involve trade-offs between economy and accuracy of the dependence modeling.

6.1.3.6.	Method for Extracting and Communicating Results

When a large quantitative uncertainty analysis has been performed, the method for
identifying important contributors and communicating this information to users is not
straightforward. Analysts have proposed many ways to quantify the uncertainty contribution of
one variable, or set of variables, on others,61 and the analyst's decision at this juncture may
strongly impact the "take-home" message from the study. An importance measure that averages

61A few examples may suffice. The standard Pearson correlation coefficient measures the linear dependence
between two variables, positive or negative. The rank or Spearman correlation coefficient measures the monotone
dependence. The correlation ratio measures the (unsigned) variance contribution of an explanatory variable on a
target variable. The regression coefficient measures the expected change in standard (not natural!) units of a target
variable, per standard unit change in an explanatory variable, and assumes this expected change is independent of
the values of the explanatory variables. Multiple correlation measures the correlation between a given variable and
its best linear predictor based on another set of variables. The partial correlation of two variables given a set of
other variables is their correlation after discounting the influence of the other variables. The correlation ratio,
multiple correlation, and the regression coefficient are not symmetric; the correlation ratio and multiple correlation
are always non-negative (Kurowicka and Cooke. 2006, 543758: Saltelli et al.. 2000, 543756).

This document is a draft for review purposes only and does not constitute Agency policy.

6-9 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

over an entire sample space may obscure the features of real interest. For example, the drivers of
cancer induction at low doses might be different from the drivers at high doses. If the drivers of
low-dose cancer induction are of interest, then importance measures that average over all doses
should not be considered.

6.2. EPA APPROACHES FOR ORAL CANCER AND NONCANCER ASSESSMENT

Different types of toxicity information have historically been used in EPA's oral cancer
and noncancer dose-response assessments, although efforts to harmonize these approaches are
ongoing. For oral exposures, noncancer endpoints are commonly assessed using the RfD
methodology to derive "an estimate (with uncertainty spanning perhaps an order of magnitude) of
a daily oral exposure to the human population (including sensitive subgroups) that is likely to be
without an appreciable risk of deleterious effects during a lifetime." In contrast, cancer
endpoints are commonly assessed using a dose-response function with the probability of excess
risk above background modeled as a linear function of dose, for doses down to zero. The RfD
method relies on a POD. The cancer dose-response method uses a POD if the linear model is
chosen. From the Cancer Guidelines, cancer endpoints can also be assessed using the RfD
methodology if the proof burden is satisfactorily met (as described in Section 5.2.3.4.1.2).

Toxicity reference values have typically been derived for human noncancer endpoints
based on a no-observed-adverse-effect level (NOAEL) or lowest-observed-adverse-effect level
(LOAEL) from animal bioassay studies. This terminology suggests a biological population
threshold beneath which no harm is anticipated. Reference values such as the oral RfD or
inhalation reference concentration are derived by applying uncertainty factors (UFs) to a POD.
Depending on the nature of available data and modeling choice, a POD can be selected from
values other than a NOAEL or LOAEL, such as an EDX (effective dose eliciting x percent
response), or a benchmark dose (BMD) or its lower confidence bound (BMDL). The BMD is
the dose that induces a benchmark response (BMR), which is often chosen to represent a 5 or
10% increase in excess risk above background. The POD is divided by one or more uncertainty
factors that represent knowledge gaps (see Section 6.4.1.2 for details on specific types of UFs).

An RfD is described as "likely to be without appreciable risk" but the probabilistic
language has not as yet been operationalized. A quantitative definition of "appreciable" has not
been articulated, and methods to compute risks above the RfD as a function of dose have not

This document is a draft for review purposes only and does not constitute Agency policy.

6-10 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

been designated for use by the EPA; thus, it is not current practice to ascertain that the risk is
indeed not appreciable. In addition, different participants in discussions over
threshold/nonthreshold models for dioxin may have different perspectives regarding how to
define "appreciable risk." Under the current POD/UF framework, dose-response functions are
not provided for calculating the actual risk at or above the RfD. Instead, to provide a "risk
indicator" for use in screening for health hazards, a hazard quotient (HQ) is computed as the
ratio of a given oral exposure to the RfD, or a margin of exposure (MOE) is estimated as the
ratio of the POD to the human exposure level.

For the cancer endpoint, an oral cancer slope factor may be derived for human health risk
assessment, typically based on tumor incidence data from an animal bioassay or on cancer
incidence or deaths from an epidemiologic study. In the EPA Cancer Guidelines, cancer is
predominantly thought to have no population biological threshold and a linear extrapolation to
zero is applied from the POD based on extra risk above background, i.e., the probability of the
endpoint decreases linearly in dose from the POD to zero or to a population background level. In
the absence of sufficient information on the cancer mode of action (MOA), the linear model is
applied as a default. The linear model also can be applied when there is sufficient MOA
evidence supporting this choice for low-dose cancer induction. Cancer endpoints could also be
evaluated using a "nonlinear" model. In this case, the proof burden clearly rests on the nonlinear
model; there must be sufficient evidence to override the health-protective default or
scientifically-based choice of a linear model, as described in the Cancer Guidelines. These
Guidelines state, "When adequate data on mode of action provide sufficient evidence to support
a nonlinear mode of action for the general population (emphasis added) and/or any
subpopulations of concern, a different approach—a reference dose/reference concentration that
assumes that nonlinearity—is used." In current terminology, the RfD methodology applies to the
cancer endpoint if there is sufficient evidence supporting a "zero slope at zero" model;
otherwise, the linear nonthreshold model applies by default. (See Section 5.2.3 for a detailed
discussion of linear vs. nonlinear extrapolations below the observed data, population vs.
individual thresholds, and how the Cancer Guidelines are applied in choosing dose-response
model forms for risk assessment.)

This document is a draft for review purposes only and does not constitute Agency policy.

6-11 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

6.3. HIGHLIGHTS OF NAS REVIEW COMMENTS ON UNCERTAINTY
QUANTIFICATION FOR THE 2003 REASSESSMENT

The NAS (2006, 198441; 2006, 543760) identified a number of uncertainty
characterization issues for the 2003 Reassessment; key sources of uncertainty for which
quantification is suggested are highlighted in Table 6-1. The discussion in this section focuses
on comments related to dose response.

There are several nuances in the NAS position relative to the need for substantial
improvement in transparency, thoroughness, and clarity in quantitative uncertainty analysis for
the 2003 Reassessment. These nuances concern whether the nonlinear model (note that the NAS
committee uses "sublinear" and "nonlinear" interchangeably) is scientifically better supported
than the linear model, and if the sublinear model is better supported, whether this is based on
data or on apodictic knowledge (knowledge without uncertainty) of the MOA. The NAS
committee does not distinguish between individual and population dose-response models;
however the criteria from the EPA Cancer Guidelines clearly apply to population models.
Assuming that the AhR-mediated MOA implies a threshold for each individual, the step to a
population "zero slope at zero" model requires the following, as identified and discussed in detail
in Section 5.2.3.:

1.	The distribution of the individual thresholds induced by the MOA, and

2.	The dose-response function for values above the thresholds.

This information can either come from data or from known information of the MOA, but
the burden of proof clearly rests on the nonlinear model. This section summarizes the NAS
committee's overall positions. Responses to specific suggestions are given in Section 6.4 and
summarized in Section 6.5. Several excerpts of specific comments from NAS (2006, 198441)
illustrate key issues.

The NAS committee favors the nonlinear model with a threshold:

This document is a draft for review purposes only and does not constitute Agency policy.

6-12 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

.. .the committee concludes that, although it is not possible to scientifically prove
the absence of linearity at low doses, the scientific evidence, based largely on
mode of action, is adequate to favor the use of a nonlinear model that would
include a threshold response over the use of the default linear assumption.

(p. 122)

The committee does not state whether the threshold applies to the population, or whether each
individual has his/her own threshold.

The NAS also comments on whether the nonlinear model should be used to compare with
the linear default:

Because the committee concludes that the data support the hypothesis that the
dose-response relationship for dioxin and cancer is sublinear, it recommends that
EPA include a nonlinear model for cancer risk estimates but also use the current
linear models for comparative purposes, (p. 16)

The committee does not suggest what the (population/individual) threshold might be, nor how it
might be supported on the basis of data. Rather, the apodictic knowledge that there is a
(population/individual) threshold places the dioxin risk assessment within the RfD framework,
using a HQ or MOE as the basis for indicating the potential risks from exposure. The committee
further asks for a quantitative characterization of the range of uncertainty:

The committee determined that the available data support the use of a nonlinear
model, which is consistent with receptor-mediated responses and a potential
threshold, with subsequent calculations and interpretation of MOEs. EPA's sole
use of the default assumption of linearity and selection of ED0i as the only POD
to quantify cancer risk does not provide an adequate quantitative characterization
of the overall range of uncertainties associated with the final estimates of cancer
risk. (p. 24)

Regarding the Cancer Guidelines' requirement of sufficient evidence to use a nonlinear
approach for cancer risk assessment, the committee indicates that quantitative evidence will not
decide the linearity/nonlinearity (nonthreshold/threshold) issue, but knowledge (without
uncertainty) of the MOA should be used:

This document is a draft for review purposes only and does not constitute Agency policy.

6-13 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

Quantitative evidence of nonlinearity below the point of departure (POD), the
ED0i62 will never be available because the POD is chosen to be at the bottom end
of the available dose-response data. ... EPA should give greater weight to
knowledge about the mode of action and its impact on the shape of the
dose-response relationship, (p. 178)

The comment continues, with the committee implicitly acknowledging that there is no
evidence arguing against linearity, but that the lack of evidence should not justify using the linear
model.

The committee considers that the absence of evidence that argues against linearity
is not sufficient justification for adopting linear extrapolation, even over a dose
range of one to two orders of magnitude or to the assumption of linearity through
zero, which would not normally be applied to receptor-mediated effects, (p. 178)

In addition, the committee recommended that EPA explore both linear and nonlinear
approaches to TCDD cancer assessment:

On the whole, the committee concluded that the empirical evidence supports a
nonlinear dose response below the EDoi, while acknowledging that the possibility
of a linear response cannot be completely ruled out. The Reassessment
emphasizes the lack of such nonlinear models, hence its adoption of the approach
of linear extrapolation below the POD level. Although this approach remains
consistent with the cancer guidelines...., EPA should acknowledge the qualitative
evidence of a nonlinear dose response in a more balanced way, continue to fill in
the quantitative data gaps, and look for opportunities to incorporate mechanistic
information as it becomes available. The committee recommends adopting both
linear and nonlinear methods of risk characterization to account for the
uncertainty of dose-response relationship shape below EDoi (p. 72).

In this document, EPA has applied its own guidance on cancer risk assessment and
adopted linearity (and an assumption of no threshold) as a health-protective default approach in
the absence of sufficient evidence of MO A involving a threshold for all tumors resulting from
TCDD exposures (volitional uncertainty). (Note that the NAS report appears to view the
absence of evidence as imposing a burden of proof on the linear model [cognitive uncertainty];
see Sections 5.2.3.4.1.2 and 6.2 regarding the burden of proof.) In addition, the NAS
committee's request to apply nonlinear methods for the cancer assessment is addressed, in

62 Eeffective dose (ED) is the dose corresponding to a X% increase (in this case a 1%) in an adverse effect such as a
concer endpoint, relative to the control response.

This document is a draft for review purposes only and does not constitute Agency policy.

6-14 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

Section 5.2.3.4.1.4 of this document. That evaluation describes the application of nonlinear
methods to TCDD data and presents two illustrative examples of RfD development for
carcinogenic effects: one based on tumorigenesis in experimental animals, and the other on
hypothesized key events in TCDD's MO As for liver and lung tumors.

The thrust of the NAS remarks regarding transparency, thoroughness and clarity in
quantitative uncertainty analysis relevant to dose-response can be summarized as follows:

1.	The uncertainty of cancer risks due to dioxin exposure should be quantified.

2.	Dioxin cancer risk should be treated either as a threshold phenomenon, thus following the
basic RfD methodology, or should be modeled using a sublinear dose-response function
below the observed data, with the linear model used for comparison.

3.	The POD should be subjected to quantitative uncertainty analysis.

A similar point of view has been indicated by others.63 Detailed suggestions regarding specific
improvements for quantitative uncertainty analysis in the 2003 Reassessment are outlined in the
next section and summarized in Section 6.5.

6.4. FEASIBILITY OF CONDUCTING A QUANTITATIVE UNCERTAINTY
ANALYSIS FOR TCDD

This section focuses on uncertainty analysis for TCDD dose response, which involves a

range of issues as highlighted in Table 6-1.

6.4.1. Feasibility of Conducting a Quantitative Uncertainty Analysis under the RfD
Methodology

This discussion applies to all noncancer endpoints of TCDD, and to cancer endpoints
insofar as they fall under the RfD methodology. An RfD is obtained through the following steps:

1.	Choose a POD, then

2.	Apply uncertainty factors (UFs) to account for knowledge shortfalls.

' "For example, from Popp et al. (2006, 1970741. "Overall, the evidence indicates that (1) TCDD causes cancer via a
receptor-mediated process; (2) this dose-response is non-linear; and (3) a threshold region exists for TCDD-induced
cancer below which adverse effects are unlikely to occur."

This document is a draft for review purposes only and does not constitute Agency policy.

6-15 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

The method of uncertainty factors harkens back to the engineering practice of safety
factors (Lehman and Fitzhugh, 1954, 003195). To illustrate, if the reference load for an
engineered structure is X, then engineers might design the structure to withstand load 3X, using a
safety factor of 3 to create a margin of safety. If the structure functions in a corrosive
environment, another factor could be multiplied to guarantee safety for that condition, and
another factor could be applied for heat, another for vibrations, and so on. The choice of values
is simply based on good engineering practice, i.e., reflecting what has worked in the past.
Although safety factors are still common in engineering, they are giving way to probabilistic
design in many applications. The reason is that compounding safety factors quickly leads to
overdesigning. Compounding safety margins for spaceflight systems may render them too heavy
to fly. As our understanding of a system increases, it becomes possible to guarantee the requisite
safety by leveraging our scientific understanding of the materials and processes. That of course
requires formulating clear probabilistic safety goals and developing the techniques to
demonstrate compliance.

The engineering community has never sought to account for uncertainty by treating
safety factors as random variables and assigning them (marginal) distributions; such an approach
would not counteract the overdesigning inherent in safety factors. Many authors, including the
recent national committee for Science and Decisions (NRC, 2009, 194810). have advocated just
such a probabilistic approach to the apparent "overdesigning" of the RfD when multiple UFs are
used in its derivation.

The NAS committee that evaluated the 2003 Reassessment does not discuss how to
perform uncertainty analysis. But their call for substantial improvement in quantitative
uncertainty analysis with TCDD falling under the RfD framework entails examining the
feasibility of quantitative uncertainty analysis within this framework. (Note that the EPA
Integrated Risk Information System (IRIS) database uses uncertainty factors without
probabilistic interpretations; some context for this is offered in Section 6.4.1.2.)

6.4.1.1. Feasibility of Conducting a Quantitative Uncertainty Analysis for the Point of
Departure

The POD plays a role in both the noncancer RfD methodology and the cancer
dose-response methodology. The POD can be selected from various options, such as a NOAEL

This document is a draft for review purposes only and does not constitute Agency policy.

6-16 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

or LOAEL, a BMDL, or an EDX. The feasibility of quantitative uncertainty analysis for each of
these three options is considered below.

By definition, the NOAEL is the highest of the tested doses in a toxicological experiment
that is judged not to have caused an adverse effect (with dose expressed as a dose rate, in
mg/kg-day). A quantitative uncertainty analysis for a NOAEL or LOAEL encounters the
following problem. In an experiment involving a small, positive response, the probability of
seeing no response can be calculated using a binomial model with the number of exposed
animals and the observed number of responses. However, in an experiment with no response,
the probability of having observed a response cannot be calculated without assuming a response
probability. Such an assumption could not be based on observed data. The probability of a
higher NOAEL or higher LOAEL can be computed, but not that of a lower NOAEL or LOAEL.
In other words, the probability that an experiment with a positive result may have yielded a null
response can be estimated, but not the probability that an experiment with a null response might
have yielded a positive response.64

In addressing uncertainty quantification for a BMDL or EDX, two questions must be
distinguished regarding the response:

1.	What is the distribution of possible doses that causes an x% increase over background?

2.	What is the distribution for possible values of increase over background caused by a
given dose?

The BMD is defined as the dose that realizes a BMR. It is an estimate from bioassay data
that requires choosing a BMR and fitting a dose-response curve. The BMR, being a choice, is
not amenable to quantitative uncertainty analysis, but the choice can be motivated in a structured
narrative. The BMDL is the lower confidence limit on the dose that realizes a BMR (e.g., 95%)
that can be found based on the uncertainty in the parameters of the dose-response relationship.
Thus, the BMDL is addressed to the first question above, and represents in this case the
95% lower confidence band of the distribution of possible doses causing an x% increase over
background. In the standard approach, the uncertainty captured by the BMDL is sampling

64The probability associated with a null response is often estimated by fitting a dose-response model to the data.

This document is a draft for review purposes only and does not constitute Agency policy.

6-17 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

uncertainty conditional on the truth of the dose-response model. Different models might fit the
data equally well yet lead to different BMDLs.

The BMDL is also influenced by the constraints imposed on the parameter fitting.
Suppose that the slope is expected to be greater than one, and that the maximum likelihood
estimate of the slope is slightly greater than one. Since the constraint is not binding, the
constrained and unconstrained model would have the same Akaike Information Criterion and
would be equivalent in this sense. However, computing the BMDL with the slope constraint can
lead to very different values than without this constraint. In the latter case, slope values less than
one contribute to the uncertainty in the dose causing the BMR (see Cooke, 2009, 543763).

The EDX can also be taken as a POD. It is similar in spirit to the BMD; however, as used
here, the term EDX applies when the dose causing an x% extra risk over background has actually
been observed, not estimated from a fitted dose-response model.65 The observations are subject
to sample fluctuations, and if the experiment on which the EDX is based were repeated, different
values might be found. It is helpful to consider a numerical example. Suppose a background
response rate of 10% is established based on many observations of nonexposed individuals. In a
given experiment, involving say 100 individuals given dose 14 individuals responded. The
percent increase x over background (extra risk) is found by solving:

14/100 = 10/100 + x x 90/100, or x = 4.4%.

We conclude that d = ED4 4. We may assume that if the experiment were repeated with 100 new
individuals sampled independently from the whole population, the response would be given by a
binomial distribution with parameters (14, 100). The number of responses might be greater or
smaller than four, there is a 16% chance of observing 10 or fewer responses. The response to
dose d would not be distinguished from the background in that case, and a higher dose would be
used for the POD.

The uncertainty analysis of EDX as the POD involves addressing the second question
above, without a quantitative dose-response model. A quantitative uncertainty analysis is
hampered, however, by the possibility that dose d would produce a response less than or equal to

65This definition of EDX is adopted to distinguish the modeled response (BMD) and the observed response (EDX),
and it is more restrictive than usages common in the literature.

This document is a draft for review purposes only and does not constitute Agency policy.

6-18 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

the background, in which case the POD is indeterminate—another experiment with a different
dose would be chosen as the POD. A true quantitative uncertainty analysis of EDX as the POD
would thus require a full bioassay experimental design, with binomial sampling of response rates
at each dose level in the assay. Absent that, quantitative uncertainty analysis is not possible.

The interplay of choice and estimation ingredients in the POD depends on the type of
POD. The main features of the above discussion are captured in Table 6-2. This table notes that
the BMDL captures the uncertainty caused by sampling fluctuations given that the dose-response
model is true. Other methods are available to compute the BMDL using (1) model-independent,
observable uncertainty; (2) nonparametric Bayesian dose-response models; or (3) Bayesian
model averaging (Cooke, 2009, 543763). Only the EDX can be subject to a quantitative
uncertainty analysis, and then only if a full bioassay data set is available.

6.4.1.2. Feasibility of Conducting a Quantitative Uncertainty Analysis with Uncertainty
Factors

Uncertainty factors are chosen based on a structured narrative characterizing knowledge
shortfalls involving the following issues:

1.	Interspecies extrapolation (UFA: from animal data to humans).

2.	Intraspecies extrapolation (UFH: to account for human interindividual variability,
considering sensitive subgroups).

3.	LOAEL to NOAEL extrapolation (UFL: to estimate the dose corresponding to no adverse
effect, from a reported LOAEL).

4.	Subchronic to chronic extrapolation (UFS: to estimate effects of chronic exposures, from
a subchronic study).

5.	Database deficiency (UFD: to extrapolate from an incomplete data set, e.g., in terms of
endpoints assessed or study design, i.e., from a poor to a sufficient or rich data context).

The standard chronic RfD can represent a sensitive human (H) response to a toxic
substance under chronic (C) exposure conditions. Suppose a BMDL POD were based on animal
(A) data from a subchronic (S) study. The database for that chemical could be rich (R), e.g.,
involving multiple (and at least sensitive) species/strains, both sexes, multiple life stages, with
multiple endpoints observed under sound study designs. Or the data could be poor (P), with
limited measurements from only a subchronic animal study (ASP) forming the basis for

This document is a draft for review purposes only and does not constitute Agency policy.

6-19 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

estimating a general reference value for humans (including sensitive subgroups) under chronic
exposure conditions. For that case, the UF method would be applied as follows:

A SP

RfD =	—		(Eq. 6-1)

UFaxUFsxUFdxUFh

where UFA, UFS, UFD, and UFH are the uncertainty factors for extrapolating from animals to
humans (UFA), subchronic to chronic exposure conditions (UFS), without adequate endpoint
coverage (UFD), and considering sensitive human subpopulations (UFH). It is possible to assign
distributions to the UFs in Eq. 6-1, and to perform a Monte Carlo analysis to produce a
quantitative uncertainty distribution over the dose or value likely to be without appreciable risk
to humans for chronic exposures. Many authors have proposed such an approach,66 and the
recent NRC (2009, 194810) report on science and decisions encourages EPA to move in this
direction.

The idea of using a Monte Carlo analysis to develop quantitative uncertainty distributions
for the RfD is simple, but the data on which the UFs are based and the assumptions that would
need to be made should be further explored. For example, it is assumed that the extrapolation
from subchronic to chronic exposure (UFS) is the same whether applied to animals or humans,
and whether applied to sufficient (rich) or deficient (poor) data contexts. Swartout et al. (1998,
093460) noted "Within the current RfD methodology, UFS does not consider differences among
species, endpoints, or severity of effects; the same factor is applied in all cases." In addition, due
to the paucity of relevant human data, the same authors suggested the use of other endpoints as
surrogates in estimating the extrapolation from animals to humans, UFa- Further, few data exist

66There has been considerable work on giving a probabilistic interpretation of the UFs, including by Abdel-Rahman
and Kadry (1995), Vermeire et al. (1999), Baird et al. (1996), Swartout et al. (1998, 093460). Slob and Pieters
(1998, 087256). Evans and Baird (1998), Calabrese and Gilbert (1993), Calabrese and Baldwin (1995), Hattis et al.
(2002, 548720). Kang et al. (2000, 548722). and Pekelis et al. (2003, 548723). These evaluations can be considered
to frame what might be called a random chemical approach. Several authors adduce properties based on log normal
distributions. Insightful studies by Kodell and Gay lor (1999;)(Gaylor and Kodell. 2000, 548724) suggest that
uncertainty factors are independent log normal variables. Combining uncertainty factors involves multiplying the
median values, and combining the "error factors" according to the formula KSxH = exp[1.6449 x V((js2 + (jH2)],
where cr/ aH2 are the variances of ln(UFs) and ln(UFH). Thus UFS x UFH is a lognormal variable with Median(UFs
x UFh) = Median(UFs) x Median(UFH), and 95th percentile given by Median(UFs x UFH) xKSxE. If Us and UH
each have an error factor or 10, then the error factor of UFS x UFH is not 100 but 25.95. Several authors suggest that
multiplying uncertainty factors might over-protect. Recent proposals from the National Research Council reflect the
random chemical concept, and they inherit its problems (NRC. 2009, 194810).

This document is a draft for review purposes only and does not constitute Agency policy.

6-20 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

in humans to accurately portray the interindividual variability represented by UFH. Much of the
data gathered to date on distributions of UFs have aggregated across other extrapolations; that is,
data from subchronic to chronic ratios are aggregated over different species and different data
contexts. Finally, it may be noted that an important issue is the data on which empirical
distributions of response ratios are based. Brand et al. (1999, 007629; 2001, 543765) studied the
sampling behavior of response ratios and raised concerns with regard to their informativeness.

Detailed analyses of the data underlying a Monte Carlo uncertainty analysis of Eq. 6-1
would afford the possibility of verifying at least some of the assumptions and numerical
estimations such an analysis must make. Even if the assumption that the same UFS is applicable
for all species, endpoints, and effect severities is thought to be biological plausible, the question
of whether a given set of chemicals reflects this assumption, and hence they are suitable for a
Monte Carlo analysis of Eq. 6-1, can only be decided by data evaluation. Data are the ultimate
arbiter of whether quantitative uncertainty analysis with uncertainty factors, as currently
envisioned, has sufficient evidentiary support.

6.4.1.3. Uncertainty Reduction Using Quantitative Data for Species Extrapolation

Expressing dose in units of exposure that are more closely related to target tissue than to
contact with administered feed (or an environmental medium) can reduce uncertainty in
extrapolations of dose, route or species. This concept underlies EPA's establishment of the
Inhalation Reference Concentration Methodology (U.S. EPA, 1994, 006488). Under this
method, species differences in tissue exposure for inhalation toxicants serve as the basis for
interspecies adjustments of dose. Likewise, the International Programme on Chemical Safety
(IPCS) has established guidance for chemical-specific adjustment factors (IPCS, 2005), which
also uses a measure of internal exposure (dose) to normalize (e.g., make equivalent) the dose
between species. Certain more recent IRIS values also reflect such an approach, with
data-derived extrapolation factors replacing default adjustments. Under such approaches, the
relationship between external exposure and target tissue exposure is determined in each species,
and the applied doses are normalized on the basis of the same level of the internal tissue
exposure. One distinction between the two approaches is that the IPCS (2005) approach is
based on the attainment of the same levels of the toxicant in the blood (the central compartment)
rather than in the actual target tissue (a consideration based in part on the fact that typically the

This document is a draft for review purposes only and does not constitute Agency policy.

6-21 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

only data available to evaluate a human toxicokinetic model will be venous blood
concentrations, rather than concentrations in a responding tissue or organ). Further, it has been
shown that species differences in internal dosimetry are more a function of species differences
in blood solubility than differences in tissue solubility—that is, once distributed to blood,
species differences in tissue exposure are less likely to be based on species differences in tissue
solubility.

The approach to development of interspecies extrapolation factors for inter- and
intraspecies extrapolation of effective dose for the oral RfD for dioxin, which is described in
Sections 3 and 4 of this document, is in agreement with both of these approaches. All tissues in
the body are exposed to dioxin via the bloodstream. Even in instances where the specific target
tissues for observed effects may be other than the tissue where the effect is observed (e.g.,
effects mediated through the endocrine system), this biologically-based approach remains valid
and reduces uncertainty in dose extrapolation. The approach to extrapolation of dosimetry—on
the basis of circulating levels of dioxin in blood—makes optimal use of human
exposure-response data, human biomonitoring data, and toxicokinetic modeling to estimate
equivalent exposures for humans and test species without requiring that the target tissue be
conclusively identified. The decision to base animal-to-human extrapolation on circulating
levels of dioxin in blood, as predicted by a well-evaluated PBPK model, reduces some potential
sources of uncertainty.

6.4.1.4. Conclusion on Feasibility of Quantitative Uncertainty Analysis with the RfD
Approach

A quantitative uncertainty analysis of the POD is not feasible for PODs based on
NOAELs or LOAELs. For the BMDL, such an analysis is not appropriate because the BMDL is
already a quantile from an uncertainty distribution of the BMD. However, this uncertainty
distribution can be obtained in different ways that capture different aspects of uncertainty.
Quantitative uncertainty analysis is feasible if the POD is based on the EDX (as defined above)
and is supported by a full set of bioassay data. A quantitative uncertainty analysis based on a
probabilistic interpretation of uncertainty factors in their present form invokes strong
assumptions. The data on which the distributions of uncertainty factors are based could be used
to check at least some of these assumptions.

This document is a draft for review purposes only and does not constitute Agency policy.

6-22 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

6.4.2. Feasibility of Conducting a Quantitative Uncertainty Analysis for TCDD under the
Dose-Response Methodology

Quantitative uncertainty analysis starts with a mathematical model and seeks to quantify
the uncertainty attending the use of this model. Dose-response relations are mathematical
models expressing the probability of response as a mathematical function of dose. For several
decades, the uncertainty attending the use of dose-response models has been an abiding concern
in many sectors, including the chemical and nuclear industries as well as the public health sector.
Given a set of animal bioassay data, quantifying dose-response uncertainty may be approached in
different ways. The differences reflect different types of uncertainty that are captured. A recent
evaluation enumerates the following possible methodologies (Bussard et al., 2009, 543770"):

Benchmark Dose Modeling (BMD): Choose the 'best' model, and assess
uncertainty assuming this model is true. Supplemental results can compare
estimates obtained with different models, and sensitivity analyses can investigate
other modeling issues.

Probabilistic Inversion with Isotonic Regression (PI-IR): Define
model-independent 'observational' uncertainty, and look for a model that captures
this uncertainty by assuming the selected model is true and providing for a
distribution over its parameters.

Non-Parametric Bayes (NPB): Choose a prior mean response (potency)
curve (potentially a "non-informative prior") and a precision parameter to express
prior uncertainty over all increasing dose-response relations, and update this prior
distribution with the bioassay data.

Bayesian Model Averaging (BMA) (as considered here): Choose an
initial set of models, and then estimate the parameters of each model with
maximum likelihood. Use classical methods to estimate parameter uncertainty,
given the truth of the model. Determine a probability weight for each model
using the Bayes Information Criterion, and use these weights to average the model
results.

The first of the above methods involves standard classical statistical methods and
captures sampling uncertainty conditional on the truth of the model used. The other methods are
"exotic" in the sense that they attempt to capture uncertainty that is not conditional on the truth
of a given model. All have been subjected to peer review and published, but they do not enjoy
the wide usage of the standard classical methods. The Bayesian models involve subjective
choices of prior distributions. Insofar as the final result is largely independent of the choice of

This document is a draft for review purposes only and does not constitute Agency policy.

6-23 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

prior, these methods conform to the current starting point of focusing on data-driven methods
and not appealing to structured expert judgment. (Structured expert judgment can also be
considered an exotic method; an explanation of this approach falls outside the scope of this
report.)

A quantitative uncertainty analysis of TCDD capturing uncertainty in extrapolating data
from animal bioassays to human reference values together with consideration of epidemiological
data from studies of workers (routine exposures) or the general public (including dietary
exposures and those reflecting discrete poisonings or accidental releases) would raise many
issues. The major issues are summarized below.

6.4.2.1. Feasibility of Quantitatively Characterizing the Uncertainties Encountered when
Determining Appropriate Types of Studies (Epidemiological, Animal, Both, and
Other)

The risk assessor must choose the data set(s) that will serve as a starting point for
dose-response modeling. With respect to TCDD, a wealth of animal bioassay data exist in the
scientific literature, across species ranging from rats, mice, guinea pigs, and hamsters to mink,
dogs and monkeys, and a variety of tissues, organs, and systems. In addition, a considerable
amount of human data is available from clinical/case reports, accidental releases, and
occupational exposures, including epidemiological data for several cohorts. As detailed in
Sections 2, 4 and 5, some of the main sources of uncertainty in the TCDD epidemiological data
include the healthy worker effect, confounding and exposure misclassification. Epidemiological
data are usually attended with large uncertainties regarding the doses actually received by
individuals. The difficulty in characterizing individual-level exposures largely stems from
having limited internal measures of TCDD exposure, as biomonitoring data may only be
available for one point in time or on a subset of the exposed population. Although there is little
direct evidence of strong confounding in the cohorts of TCDD and dioxin-like compounds, some
of the confounders that have been evaluated in a few of the epidemiological studies include
gender, body mass index, age, cigarette and alcohol consumption, and hair and eye color
(Baccarelli et al., 2005, 197053: 2006, 197036: Eskenazi et al., 2002, 197168: 2002, 197164:
Pereg et al., 2002, 199797). As discussed in Section 5 on TCDD carcinogenicity, an additional
limitation of the epidemiological evidence includes the lack of organ specificity, as many of the

This document is a draft for review purposes only and does not constitute Agency policy.

6-24 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

studies have shown associations between TCDD exposure and all-cause mortality. With
disagreement in the literature over the nature, scope, and quality of the epidemiological data for
TCDD, given the lack of precedent for a multisite carcinogen without particular sites
predominating, some have urged caution in the interpretation of the epidemiological data based
on small relative risks Popp et al. (2006, 197074).

Despite these uncertainties, the EPA Cancer Guidelines express a clear preference for
epidemiological studies over animal data. The question here is whether quantitative uncertainty
analyses based on either a collection of bioassay data or on several epidemiological studies can
be combined in some overall uncertainty assessment. Diverse human studies are sometimes
combined into a meta-analysis, and the issues arising in this regard are instructive. A primary
challenge of meta-analytical approaches is combining heterogeneous effects that may result from
studies of different populations, study designs or analytical techniques. The question of whether
uncertainty arising from combining such different studies can be taken into account in
quantitative uncertainty analysis is similar to that of accounting for uncertainty due to missing
covariates in Cox regression (see Section 6.4.2.2).

Existing standard statistical tools are insufficient to address this issue, as they quantify
uncertainty in model parameters estimated from data. However, exotic methods, such as
Bayesian methods, probabilistic inversion, or structured expert judgment may be applicable.
These methods can be applied when a quantitative model predicts other phenomena, even though
these phenomena could not be used to estimate the model. The question of whether such
methods could remain sufficiently tethered to data, or whether structured expert judgment is
unavoidable, is a subject for future research.

6.4.2.2. Uncertainty in TCDD Exposure/Dose in Epidemiological Studies

Uncertainties in epidemiological studies arise from a variety of study characteristics.
There are many types of epidemiological study designs which determine the data structure,
including intervention trials, case-control studies, cohort studies and cross-sectional studies. A
variety of mathematical models some of these can be used to analyze epidemiological data; some
of these includes Cox proportional hazard, Poisson regression, linear and logistic regression.
The model outputs are based on different measures of association such as rate ratios, risk ratios,
odds ratios, and standardized mortality ratios (SMRs, ratio of observed to expected deaths).

This document is a draft for review purposes only and does not constitute Agency policy.

6-25 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

Exposure uncertainties often concern back-casted exposures based on current serum lipid
concentrations, estimated/self reported dietary habits, fish consumption, placenta lipid
concentrations, and other measures.

Uncertainty in exposure is often dealt with by coarsely grouping a cohort into exposed
and unexposed groups. The output of such a study can be coarse grained in a similar way;
instead of computing dose-dependent risk estimates, standard mortality ratios might be used to
compare the exposed and unexposed groups. Packages computing the outputs routinely produce
confidence intervals that reflect sampling fluctuations (e.g., can indicate the potential for chance
to explain the association), assuming truth of the model. Additional uncertainty could be
factored in with exotic methods. A significant issue in epidemiological studies is the effect of
omitted covariates. Omitted covariates in Cox regression will bias the estimates of effects of
included covariates. If the omitted covariates are independent of the included covariates, the bias
is toward zero in absolute value (Bretagnolle and Huber-Carol, 1988, 543772); if the omitted
covariates are not independent, little can be inferred.

With regard to individual studies, it might be possible to identify specific opportunities
for uncertainty quantification. This is illustrated here using the study of Steenland et al. (2001,
198589) of more than 3,500 male workers exposed to TCDD-contaminated products at eight
U.S. chemical plants. Each worker was assigned an exposure score based on an estimated level
of contact with TCDD, the degree of TCDD contamination of product at each plant over time,
and the fraction of a workday in contact with the product. For 170 workers, the serum TCDD
levels were also measured. The serum levels were back-extrapolated to the last time of exposure
using a constant biological half life, and regressed on the exposure scores. This regression
model was used to predict the dose in all workers, and predicted dose was correlated with cancer
mortality. Figure 6-1 shows a scatter plot of back-casted versus predicted TCDD serum levels
for the 170 workers on which the regression was based.

Given a predicted TCDD level, the uncertainty on the back-casted TCDD value could be
inferred from such data by various techniques. A key question is whether the actual cancer
mortalities among 170 back-casted workers are randomly placed in the conditional distribution
given predicted TCDD. Imagine, in other words, that the mortalities among the 170 back-casts
are colored red in Figure 6-1. At any given level of TCDD prediction, are the red points evenly
distributed, or are they shifted to the right? In principle, the correlation between mortality and

This document is a draft for review purposes only and does not constitute Agency policy.

6-26 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

back-casted TCDD level, given the predicted level, could be estimated. This amounts to
estimating heteroscedasticity in the regression model.67 Then, for each of the 3,538 workers,
given his predicted TCDD level, we could sample a back-casted TCDD level, appropriately
correlating with mortality, and recompute the dose response analysis. Repeating this many times
we could build up a distribution for excess lifetime cancer mortality risk.

It is instructive to step through similar issues with regard to biological half life,
background and body fat. The Steenland et al. (2001, 197433s) analysis assumed a constant
TCDD biological half life (8.7 years). A distribution over this half life could plausibly be
developed from published sources. Assuming this half life is constant for all workers, but
uncertain (epistemic uncertainty), this distribution could easily supplement the previous
distribution: first sample a half life (to be applied to all workers), then estimate the regression
model for this half life, and sample back-casted TCDD levels given each worker's exposure
score, taking account of correlation with mortality. This works if the half life uncertainty is
epistemic. However, since the half life is estimated from data, it is more reasonable to suppose
that the half life varies from worker to worker (aleatoric uncertainty). Here again the correlation
with mortality must be taken into account, indeed it seems reasonable to suppose that the
256 cancer deaths involved workers with longer half lives. However, there is no way ex post of
determining the biological half life in the deceased workers.

The potential impact of uncertainty regarding background exposure and body fat is likely
similar to the uncertainty of estimating the half life of TCDD. Steenland et al. (2001, 197433)
held the background level constant at the median level (6.1 ppt, range 2.0 to 19.7) for
79 nonexposed workers from whom blood was also drawn (see also Section 6.4.2.4). The full
distribution of TCDD levels for these nonexposed workers could be used as well. Is it
reasonable to suppose that responsive workers (i.e., those exhibiting the response) have
background levels that are sampled randomly from this distribution, or might they not plausibly
come from the high end of the distribution? The analysis also assumed a constant percentage of
body fat (30%), whereas body fat percentage varies in the general population, e.g., for men this
has been reported to range from 2 to 38% or more (see Footnote in Section 6.1.3.3). The body

67 Heteroscedasticity occurs when the variance of the dependent variable in a regression analysis varies across the
data, violating the assumption of equal variance commonly used in many regression models.

This document is a draft for review purposes only and does not constitute Agency policy.

6-27 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

fat distribution in the worker population could have been ascertained, but again the question
arises, are the responsive workers sampled randomly from this distribution?

These three factors, variable half life, variable background, and variable body fat
percentage, might combine to make the effective dose level among the responsive workers
significantly higher than would appear in a study that assumes these factors to be constant.
However, such concerns cannot be addressed in a quantitative uncertainty analysis, unless cancer
mortality can be correlated with these variables. In an optimal study design, this information
could be retrieved from the data. However, in most observational epidemiological studies such
data are not available, and it might be possible to estimate these correlations in some other
defensible manner, in which case the effect of exposure uncertainty could be quantified and
propagated. Such an analysis would involve substantial effort and should not be undertaken
under assumptions that are themselves implausible. Protocols for epidemiological studies do not
currently require such uncertainty quantification. In any event, Steenland et al. (2001, 197433)
should be recognized for conscientiously identifying these key issues.

6.4.2.3. Uncertainty in Toxicity Equivalence (TEQ) Exposures in Epidemiological Studies

Toxicity equivalence factors (TEFs) are used to infer the health effects of dioxin-like
compounds based on their relative potencies compared to TCDD. These factors are not known
with certainty, and the question arises whether uncertainty in TEFs can be incorporated into a
quantitative uncertainty analysis. The process of deriving TEFs applied by the World Health
Organization (WHO, 2005, 198739) is described in Van den Berg et al. (2006, 543769).
Distributions of relative potencies (REPs) were developed from the scientific literature, with
preference for in vivo studies, as supplemented by in vitro studies. An expert panel used a
consensus process to select a TEF value for each congener, in half log steps "Thus, the TEF is a
central value with a degree of uncertainty assumed to be at least ± half a log, which is one order
of magnitude. However, it should be realized that TEF assignments are usually within the 50th to
75th percentile of the REP distribution, with a general inclination toward the 75th percentile in
order to be health protective" (Van den Berg et al., 2006, 543769) (see Figure 6-2 of this
document).

The WHO considers the uncertainty in TEFs to span one order of magnitude (presumably
log uniformly distributed). It would be tempting to use the distributions in Figure 6-2 to quantify

This document is a draft for review purposes only and does not constitute Agency policy.

6-28 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

uncertainty in the TEFs in a quantitative uncertainty analysis. However, the issue of dependence
in this case is daunting. For example, should values of 1,2,3,7,8,-pentachlorodibenzofuran and
2,3,4,7,8-pentachlorodibenzofuran be sampled independently? The choice of dependence
structure will have a large effect. As described by (Van den Berg et al., 2006, 543769). the
differences in REPs reflect differences in dosing regimens, species, endpoints, mechanisms, and
calculation methods. In a quantitative uncertainty analysis one must insure that these are not
double counted.

Reasons for significant differences in REPs for the same congener can be caused
by the use of different dosing regimens (acute vs. subchronic), different endpoints,
species, and mechanisms (e.g., tumor promotion caused by at least two different
mechanisms as for mono-ort/zo-substituted PCBs), as well as different methods
used for calculating REPs. Thus, different methodological approaches used in
different studies clearly provide uncertainties when deriving and comparing REPs.
If future study designs to derive REPs were more standardized and similar, the
variation in REPs when using the same congener, endpoint, and species might be
expected to be smaller (Van den Berg et al., 2006, 543769).

Although the TEFs themselves and the distributions underlying them are based on expert
judgment, it is possible to incorporate these into a quantitative uncertainty analysis; however, it
is not simply a matter of taking the distributions in Figure 6-2 to predict the results, with
uncertainty, of exposure to dioxin4ike compounds. The issues of dependence and double
counting must first be addressed. Inasmuch as the distributions are the result of expert judgment,
this would reasonably involve structured expert judgment as well. (Procedures for this type of
assessment have been developed and applied, and it would entail a significant level of effort.)

6.4.2.4. Uncertainty in Background Feed Exposures in Bioassays

TCDD is not produced intentionally but rather is formed as a byproduct of volcano
eruptions, forest fires, manufacturing of steel and certain chemicals (including some pesticides
and paints), pulp and paper bleaching, exhaust emissions, and incineration. It enters the food
supply primarily via aerial transport and deposition of emissions, and it bioaccumulates in animal
fat. In general, food of animal origin contributes to about 80% of the overall human exposure.
For example, Schecter et al. (1997, 198396) measured dioxins in pooled food samples collected
in 1995 from supermarkets across the United States. Reported as parts per trillion (ppt) toxicity

This document is a draft for review purposes only and does not constitute Agency policy.

6-29 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

equivalences (TEQs), fresh water fish had the highest level (1.43); followed by butter (1.07);
hotdog/bologna (0.54); ocean fish (0.47); cheese (0.40); beef (0.38); eggs (0.34); ice cream
(0.33); chicken (0.32); pork (0.32); milk (0.12); and vegetables, fruits, grains, and legumes
(0.07). More recent exposure studies indicate dietary levels have decreased over time. Values
reported for the early 2000s by Lorber et al. (2009, 543766). in ppt TEQ, are: fish (0.33); beef
(0.12); dairy, other than milk (0.079); eggs (0.06); pork (0.036); poultry (0.018); other meat
(0.058); and milk (0.012).

These results illustrate that a person's dietary intake of dioxins depends on the relative
intake of foods with high or low levels of contamination, and human background levels will vary
accordingly. The same applies to experimental animals in bioassays, although in those cases the
background intake can in principle be controlled. Some of the effects of TCDD and other AhR
agonists in enhancing the early initiation stages of cancers are considered to occur as a result of
prenatal exposures that are not included in the standard National Toxicology Program (NTP)
bioassay protocol (Brown et al., 1998, 051311; Muto et al., 2001, 548713). Further, to enhance
reproducibility and keep statistical fluctuations to a minimum, the standard NTP assays are
deliberately run on groups of animals that are relatively uniform genetically, fed uniform diets,
and have the minimum possible exposures to toxicants other than the agent(s) being tested. This
tends to reduce the potential for observing the consequences of potential interactive effects that
might occur in the diverse human population with its variety of dietary and other exposures to a
wide range of potentially interacting substances and conditions.

A critical question is the extent to which the background exposure influences the
dose-response curve, and how this background should be taken into account. One idea,
articulated in the recent NRC (2009, 194810) report on science and decisions, involves an
"interacting background."68 This can be implemented by computing a virtual dose B which,
according to the selected dose-response model, would explain a chosen fraction of the
background response. If the chosen model for dose 8 is f(8), the model can be adapted to

68"Effects of exposures that add to background processes and background endogenous and exogenous exposures can
lack a threshold if a baseline level of dysfunction occurs without the toxicant and the toxicant adds to or augments
the background process. Thus, even small doses may have a relevant biologic effect. That may be difficult to
measure because of background noise in the system but may be addressed through dose-response modeling
procedures" (NRC, 2009).

This document is a draft for review purposes only and does not constitute Agency policy.

6-30 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

account for an interacting background by writing P(S) = f(5 + B) - f(B). This can alter the
model's behavior at zero dose.

For example, if f(8) = 8n/(8n + EC so'1), the derivative d(f)/d(8) is nSn 'ECso'VfS11 + ECso")2,
which goes to zero as 8—>0, if n > 1. However, replacing 8 with (8 + B) evidently changes the
derivative at zero to nBn 'ECso'VfB'1 + ECso")- This model is not yet estimable from data, as we
have no way of choosing from the available animal data the fraction of background response to
be explained by the model when applied to humans (although judgments could be made if we
had better information about the details of the processes that are involved in causing various
human health effects). However, as a conceptual model, it serves to remind us that the manner
of accounting for background exposures can influence a model's behavior in the low-dose
region. (Note that sensitivity analyses can be done showing the consequences of assuming
different amounts of interacting background within the context of a specific nonlinear model.)

6.4.2.5.	Feasibility of Quantifying the Uncertainties Encountered When Choosing Specific
Studies and Subsets of Data (e.g., Species and Gender)

Species, strain, gender, life stage, and other characteristics of experimental animals are
selected for a given study based on previous knowledge (e.g., of the species sensitivity,
availability of strains having little genetic variation for the endpoints in question, relevance of
the MO A, and degree to which the endpoints are similar for humans). Many other decisions are
made in designing a bioassay study; will the animals be sacrificed at the termination of the study
(if not a lifetime study), or will they be allowed to live out their natural lives? What dosing
regimen should be applied? How will the animals be fed and handled? Although such questions
may engender uncertainty in the minds of the experimenters, and reviewers; such uncertainty is
not amenable for quantitative uncertainty analysis unless and until there are quantitative models,
with parameters estimable from data, that can predict the effect of these choices on the response
function.

6.4.2.6.	Feasibility of Quantifying the Uncertainties Encountered when Choosing Specific
Endpoints for Dose-Response Modeling

Standard experimental protocols guide the selection of exposure/dosing conditions for a
given bioassay, including the amount, delivery vehicle, route, timing, dosing frequency and

This document is a draft for review purposes only and does not constitute Agency policy.

6-31 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

duration, and dose spacing. The goal is to find the dose range where the experimental animals
begin to respond adversely, to help anchor the lower end of the dose-response relationship, and
to avoid multiple experiments in which all or none of the animals respond. A common
recommendation is that the dose levels be chosen such that the increments in probability of
response are roughly equal. Hence, the choice of endpoint, dose spacing, and number of animals
should be made with these factors in mind. Of particular importance is the number of animals at
each dose level in relation to the choice of endpoint and probability of response. Using more
animals at the lower dose levels increases the probability of seeing some animals respond; on the
other hand, it will give higher weight to the low-dose responses in model fitting and uncertainty
quantification. Including many low-dose groups in a study with no expected response can
produce a bias in the event of model mis-specification (see Text Box 6-1). The conclusion with
regard to the feasibility of this quantitative uncertainty analysis echoes that of the previous
paragraph: such uncertainty is not amenable for quantitative analysis unless and until there are
quantitative models, with parameters estimable from data, that predict the effect of these choices
on the response function.

6.4.2.7. Feasibility of Quantifying the Uncertainties Encountered when Choosing a Specific
Dose Metric (Trade-Off between Confidence in Estimated Dose and Relevance of
MOA)

The concept of dose is not straightforward. To review, the Cancer Guidelines provide the
following taxonomy:

•	Exposure is contact of an agent with the outer boundary of an organism.

•	Exposure concentration is the concentration of a chemical in its transport or
carrier medium at the point of contact.

•	Dose is the amount of a substance available for interaction with metabolic
processes or biologically significant receptors after crossing the outer boundary of
an organism.

•	Potential dose is the amount ingested, inhaled, or applied to the skin.

•	Applied dose is the amount of a substance presented to an absorption barrier and
available for absorption (although not necessarily having yet crossed the outer
boundary of the organism).

This document is a draft for review purposes only and does not constitute Agency policy.

6-32 DRAFT—DO NOT CITE OR QUOTE

-------
Text Box 6-1. Model Mis-Specification and Maximum Likelihood Estimation.

The maximum likelihood estimate (MLE) is widely used in statistics because of its attractive properties: //"the
true model generating the data is from the class whose parameters are being estimated, then under regularity
conditions, the expected MLE converges to the true value, and its variance converges to zero. The caveat against
what is called "mis-specification" is very important and easily overlooked. An illustration can be extracted from
the NTP (2006a) data for female rat tumor incidence of cholangiocarcinoma, representative of the data which
persuaded the NAS committee that the cancer dose response for dioxin was "sublinear."

NTP (2006a) Female Rat Tumor Incidence Data for Cholangiocarcinoma

Blood concentration (ng/kg)

2.56

5.69

9.79

16.57

29.70

Number exposed

48

46

50

49

53

Number responding

0

0

1

4

25

Relative frequency

0

0

0.02

0.08

0.47

The Hill model with MLE in this case has zero slope at zero. The default Linear Low Dose (LLD) model fits a
Hill model to doses with positive responses, but it extrapolates linearly from the lowest observed nonzero response
frequency. Both models have the same two parameters, but the parameter values of the Hill model used in the LLD
model are different from those in Hill model fit to all doses, including doses with zero response. Although the null
responses are expected on the LLD model, the Hill model has greater log likelihood since it gives higher probability
to the null responses (see below).

NTP (2006a) Female Rat Tumor Incidence Data for Cholangiocarcinoma:
Low-Dose Linear and Hill Models

Blood concentration (ng/kg)

2.56

5.69

9.79

16.57

29.70

Number exposed

48

46

50

49

53

Response probability: Linear Low Dose (LLD)

0.005

0.012

0.014

0.09

0.47

Response probability: Hill model

0.00009

0.0017

0.013

0.09

0.47

Probability of cohort null response: LLD

0.77

0.58







Probability of cohort null response: Hill

0.99

0.92







Log Likelihood

LLD

2.46





Hill

2.16





Suppose, for the sake of illustration, that the data were generated with the response probabilities from the LLD
model. The Hill model would be mis-specified in this case, as the model generating the data is not a Hill model.
Because of the small cohort size, the probability of null responses is such that the Hill model has greater likelihood
than the LLD model with probability (based on bootstrapping) about 0.43, even though the latter, by construction,
is the true model. Averaging over many simulated responses from the LLD model, the Hill model underestimates
the response probabilities for doses 2.56 and 5.69 by factors of 7.5 and 2.1 respectively. In the event of such mis-
specification, the bias in the Hill model would be aggravated by including more 50-rat experiments with doses
lower than 2.56.

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

6-33 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

•	Absorbed dose is the amount crossing a specific absorption barrier (e.g., the
exchange boundaries of skin, lung, and digestive tract) through uptake processes.

•	Internal dose is a more general term, used without respect to specific absorption
barriers or exchange boundaries. Delivered dose is the amount of the chemical
available for interaction by any particular organ or cell

Due to their greater causal proximity to the affected organs, using the absorbed dose or
internal dose would yield statistically more powerful results and enable more precise predictions
than potential dose. If it is not possible to measure these or they were not measured during the
conduct of the study (as is commonly the case), then other available dose metrics, such as
potential dose or exposure, are used. Due to toxicokinetic variability, different individuals
receiving the same exposure may not have the same absorbed dose. Hence, use of either
exposure or exposure concentration adds variability to the predicted results. The dose metric
should be selected that (1) has the most proximate possible causal relation to the production of an
adverse health endpoint, and (2) can be readily related to the units of (external) exposure that
will be the basis for assessing human exposures.

6.4.2.8. Feasibility of Quantifying the Uncertainties Encountered When Choosing Model
Type and Form

The EPA (2009, 522927)draft white paper on probabilistic methods notes: "There is no
consensus on any one well-accepted general methodology for dealing with model uncertainty,
although there are various examples of efforts to do so." Model uncertainty was introduced in
Section 6.1.3.4. Many statistical techniques are available to evaluate model adequacy or to
choose a "best" model. Although it is tempting to qualify such deliberations as "uncertainty that
a model is true," one must remember that all models, being idealizations, are false. Ultimately,
one is interested in uncertainty with regard to observable phenomena, not with regard to models.
Models are merely tools for describing the phenomena. Nonetheless, the choice of a model
constrains the ways in which uncertainty can be represented, so the question is how to deal with
these constraints. A recent study of uncertainty modeling in dose response (Cooke, 2009,
543763) addresses precisely this issue and provides technical details to frame possible options.

Before exploring exotic approaches to model uncertainty (i.e., those not yet widely used
in dose-response analyses), one feature in the standard statistical treatment of uncertainty must

This document is a draft for review purposes only and does not constitute Agency policy.

6-34 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

be appreciated. Consider a model based on experimental data, typically bioassay data, in which
a certain number of study subjects are exposed to varying doses of a test substance, and in which
the numbers of subjects exhibiting a response are tallied. Values for the parameters in the model
are chosen by the principle of maximal likelihood: those values are chosen which render the data
as likely as possible. According to standard practice, a model is chosen that best fits the data
according to one of the accepted criteria, such as reduced R2, or the Akaike Information
Criterion. There might be many incompatible models that are nearly as good.

One can ask the following: If the experiments on which the model is based were repeated,
sampling the same number of experimental subjects from the distribution posited by the model,
how much could our parameter estimates change? This is described by a joint distribution over
the model's parameters, which captures sampling uncertainty under the assumption that the
model is true. Now, all models are false, and as our sample sizes grow the lack of fit in the
model becomes increasingly apparent. At the same time, the sample fluctuations in parameter
estimates—assuming the model is true—become smaller and smaller. In very large
epidemiological studies, standard statistical methods can produce razor-thin confidence bands in
this way, which fail to capture experts' uncertainty regarding observable phenomena.69

The exotic methods sketched in the beginning of Section 6.4.2 may be viewed as attempts
to deal with this feature. Probabilistic inversion methods were deployed on a large scale in the
joint U.S. NRC-EU uncertainty analyses noted in Section 6.1. Distributions over model
parameters are intended to capture an antecedently defined uncertainty over observable
phenomena predicted by the model. This method was applied in dispersion and deposition
modeling and further environmental transport models (including uptake) for radionuclides. In
most cases, the observable uncertainty was based on structured expert judgment, but it has also
been based on binomial uncertainty in bioassay studies. A potential drawback is that it may not
prove possible to capture the observable uncertainty in this way with a classically best-fitting
model, and new models may be required.

Nonparametric Bayesian methods arose in the biomedical and reliability fields. They
start with a prior distribution over all nondecreasing dose-response functions, and update these

69See, for example, Tuomisto et al. (2008, 548715. Table 6) for a comparison of experts' uncertainty in health
effects of fine particulates with uncertainties derived from sampling uncertainty from large epidemiological studies.
Although the experts generally agree with the studies' central estimates, their uncertainty bands are often much
wider than those surrounding the published estimates.

This document is a draft for review purposes only and does not constitute Agency policy.

6-3 5 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

with observations from a bioassay. No further assumptions regarding parametric form are
introduced, but the prior distribution remains important for doses outside the range of
observations. Bayesian model averaging starts with a prior distribution over a set of candidate
models, and updates this distribution with bioassay data. The method is flexible and intuitive,
though attenuation of the effect of the prior on the posterior must be verified.

All these approaches represent attempts to capture "extramodel uncertainty," that is,
uncertainty that is not conditional on the truth of the model. This is an active research area, and
improvements in methods for capturing extramodel uncertainty in quantitative uncertainty
analysis are anticipated. A major effort with regard to TCDD dose-response would be indicated
when the strengths and weakness of the exotic methods are well understood.

6.4.2.9. Threshold MOA for Cancer

The NAS committee avers that knowledge of the AhR binding MOA implies that there is
a response threshold for TCDD cancer induction. The differences between individual and
population thresholds are not discussed, but the following two possibilities are distinguishable:

1.	The threshold is the same for each individual; since human variability in AhR binding
affinity is rather large (see Section 5.2.3.3), this entails that the threshold is not affected
by the binding affinity.

2.	The threshold varies across individuals and is related to the individual AhR binding
affinity.

These two positions are different. As shown in Section 5.2.3 it is quite possible that each
individual in a population has a threshold, whereas the population dose-response relation is
linear. Because the NAS committee does not distinguish which of these positions it holds, the
feasibility of quantitative uncertainty analysis is examined here for both.

i.	Quantitative uncertainty analysis concerns a mathematical model. In case (1), this model
would show how the existence of the AhR binding would induce a threshold,
independently of the strength of the binding. Assessing the feasibility of quantitative
uncertainty analysis must await the elaboration of such a model.

ii.	In case (2), it must be shown that the distribution of thresholds, and the dose-response
function above the threshold, are able to induce a population "zero slope at zero dose"
(ZS@Z) model. Recall, the burden of proof is on this (ZS@Z) model. Scoping the

This document is a draft for review purposes only and does not constitute Agency policy.

6-36 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

population variability with regard to AhR-mediated mechanisms in general, and dioxin
sensitivity in particular, is an active area of research. It involves phenotyping human
AhR-mediated responsiveness and relating this to polymorphisms in the human
population. Harper et al. (2002, 198124) report that a 10-fold variation in binding
affinity of AhR for TCDD in human placental tissue did not reveal any polymorphisms,
suggesting that the relation between phenotypical and genotypical variation is tenuous.
Tuomisto et al. (1999, 548717) demonstrate large variations in efficacy in two rat strains
whose binding affinity is similar (Long-Evans, Kd = 3.4, Han/Wistar, Kd = 3.9 (as also
discussed in Connor and Ay 1 ward, 2006, 197632)). and they also show that this variation
is endpoint-specific. The responses in both strains are similar for cytochrome P450
(CYP)l Al induction, but very dissimilar for thymus atrophy, serum bilirubin, and
mortality. Toide et al., (2003, 548792) suggest that common biochemical measures of
EROD activity might be mediated by CYP1B1 and CYP1A2. The differences in serum
bilirubin at doses around 10 |ig/kg are about a factor of 30. Han/Wistar rats seldom die at
this dose, while mortality of Long Evans rats is about 50%. The mechanisms are not
understood.

Although the mass action dose-response model does not have a threshold, it is possible
that certain enzymes block the receptor binding, and until these are overwhelmed, no response
occurs. The availability of such enzymes may vary from individual to individual, and may or
may not covary with the dissociation constant, Kd. Pursuing these lines of research may result in
a convincing demonstration of a population (ZS@Z) model. Such a model would express the
individual threshold in terms of parameters that could be estimated with uncertainty from the
data.

6.4.2.10. Feasibility of Quantifying the Uncertainties Encountered when Selecting the BMR

The NAS committee explicitly requested that the uncertainty attending the choice of a
BMR be quantified. Although selecting relevant alternative values for the BMR may provide
information of interest, it does not constitute a quantitative analysis of uncertainty. The
alternative values must be sampled from some uncertainty distribution. Since this concerns
volitional uncertainty, there is no underlying distribution from which to sample, unless the
choice of BMR is related to some claim about the state of the world.

However, in response to the NAS concerns, this document provides some limited
quantitative comparisons of BMR choices. BMDs, BMDLs and OSFs from the animal cancer
bioassay benchmark dose modeling assuming 1, 5, and 10% extra risk are compared in units of
blood concentrations and human equivalent doses in Tables 5-18 and 5-19, respectively. In

This document is a draft for review purposes only and does not constitute Agency policy.

6-37 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

addition, MLE and upper bound slope factor estimates based on Cheng et al. (2006, 523122) are
presented (see Tables 5-3 and 5-4). For the noncancer effects, key animal study PODs
(ng/kg-day) are shown based on different dose metrics: administered dose, first-order body
burden HED, and blood concentration (see Tables 4-3 and 4-4).

6.5. CONCLUSIONS REGARDING THE FEASIBILITY OF QUANTITATIVE
UNCERTAINTY ANALYSIS

In this section the main conclusions regarding the feasibility of quantitative uncertainty
analysis are summarized in relation to specific suggestions made by the NAS committee (see
Section 6.5.1). Following this, a suggested research agenda for moving forward in this area is
provided (see Section 6.5.2).

6.5.1. Summary of NAS Suggestions and Responses

On page 130 of their report (NAS, 2006, 198441), NAS makes specific suggestions
regarding uncertainty quantification. These are reformatted and presented in italics below.
Following each suggestion, a summary of the discussion in this section is given, with reference
to the section in which it is addressed.

EPA should have addressed quantitatively the following sources of uncertainty:

• Basis for risk quantification:

1.	bioassay data,

2.	occupational cohort data.

Response: (1) Classical statistical methods yield distributions on model parameters
which reflect sample fluctuations, assuming that the model is true. This type of
uncertainty is taken into account in the BMDL. Exotic methods can account for
uncertainty which is not conditional on the truth of a model, at least for bioassay data
(see Section 6.4.2). (2) For epidemiological data, the dose reconstruction often involves
assumptions which may support data driven uncertainty analysis, if sufficient data can
be retrieved. Examples discussed above include back-casted TCDD level, biological
half life, body fat and background (see Section 6.4.2.2). Uncertainty in the choice of
bioassay data sets or choice of occupational cohort data sets is volitional, and is not
quantified by sampling an input distribution. To be amenable for quantitative
uncertainty analysis, the choice must be linked to a statement about the state of the
world (see Section 6.1.1).

This document is a draft for review purposes only and does not constitute Agency policy.

6-3 8 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

•	Epidemiology data to use:

1.	risk estimate developed with data aggregatedfrom all suitable studies,

2.	risk estimate or estimates developed using each study individually.

•	Factors affecting extrapolation from occupational to general population cohorts,

including differences in baseline health status, age distribution, the healthy worker
survivor effect, and background exposures.

Response: (1) Quantitative uncertainty analysis based on meta-analysis data poses
challenges owing to differences in study protocols. Exotic methods might take us further,
the question is whether the restriction to data driven methods (as opposed to expert
judgment or Bayesian methods) could be maintained (see Sections 6.4.2.2 and 6.4.2.3).
(2) If the general population is characterized by distributions over known confounders
whose coefficients are estimated from the epidemiological studies, then uncertainty over
these coefficients can be extracted with the methods mentioned in Section 6.4.2.1.
Uncertainty due to missing covariates is intractable for data driven uncertainty analysis
(see Section 6.4.2.2).

•	Bioassay data to use:

1.	risk estimate developed with the single data set implying the greatest risk (that is,
single study, tumor site, gender),

2.	risk estimate developed with multiple data sets satisfying an a priori set of
selection criteria.

Response: (1) Uncertainty in choice of data sets is volitional and is not quantified by
sampling an input distribution. To be amenable for quantitative uncertainty analysis the
choice must be linked to a statement about the state of the world (see Section 6.1.1).
(2) The issue here is similar to the meta-analysis addressed in (2.a).

•	Dose-response model:

1.	linear dose response,

2.	nonlinear dose.

Response: (1) When low dose extrapolation is done using a linear model by default, the
uncertainty is volitional. To be amenable for quantitative uncertainty analysis, the choice
must be linked to a statement about the state of the world (see Section 6.1.1). The EDX as
POD for the linear extrapolation can be subjected to quantitative uncertainty analysis, if
based on sufficient bioassay data. (2) With respect to nonlinear dose response, it is
possible that human thresholds exist, and that the distribution of thresholds can be
characterized in the human population. In as much as the mechanisms for this are not yet
understood, there is no quantitative model expressing threshold as a function of
parameters which could be estimated, with uncertainty, from data. This currently limits
the application of uncertainty quantification (see Section 6.4.2.9).

This document is a draft for review purposes only and does not constitute Agency policy.

6-39 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

•	Dose metric:

1.	average daily intake,

2.	area under the blood concentration-time curve,

3.	lifetime average body burden,

4.	peak body burden,

5.	other.

Response: (1-5) The dose metric is chosen to maximize causal proximity to the endpoint,
while maintaining the link to measured exposure (see Section 6.4.2.7). There may be
uncertainty with regard to which metric is optimal. If an inappropriate metric is chosen
in a bioassay study, this would be expressed in noisier responses which would tend to
suppress the dependence of endpoint on dose. A data driven quantitative uncertainty
analysis of dose metric would require a mathematical model expressing endpoints as a
function, inter alia, of dose metric, with parameters estimated from data.

•	Dose metric—biological measure:

1.	free dioxin,

2.	bound dioxin.

Response: (1-2) The issue is whether all TCDD available for AhR binding, or only the
bound TCDD, should be used as a dose metric. Binding affinity is determined by more
factors than genetic polymorphisms and these other factors are poorly understood (see
Section 6.4.2.9). A quantitative uncertainty analysis must await the formulation of a
quantitative model expressing binding affinity in terms of parameters which can be
estimated from data.

•	POD:

1.

ED io,

2.

EDos,

3.

EDoi

Response: (1-3) Uncertainty in choosing a POD is volitional. Uncertainty in the value
of an EDX can be quantified in a data driven manner if sufficient bioassay data is at hand
(see Section 6.4.1.1).

• Value from ED distribution to use:

1.	ED,

2.	lower confidence bound value for the ED (LED),

3.	upper confidence boundfor the ED (UED).

This document is a draft for review purposes only and does not constitute Agency policy.

6-40 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

Response: (1-3) Given that uncertainty on the POD is quantified, a distribution of the
slopes of a linear low dose extrapolation is readily derived, and hence a distribution of a
risk specific dose.

• Where alternative assumptions or methodologies could not be ruled out as implausible or
unreasonable, EPA could have estimated the corresponding risks and reported the
impact of these alternatives on the risk assessment results. The potential impacts offour
sources of uncertainty are discussed below.

1.	The full range ofplausible parameter values for the dose-response functions used
to characterize the dose-response relationship for the three occupational cohort
studies selected by EPA (Becher et al., 1998, 197173; Ott and Zober, 1996,
198483; Steenland et al., 2001, 197433)).

2.	Use of other points of departure, not just the ED0i (or LED0i), to develop a CSF.

3.	Alternative dose-response functional forms as well as goodness of fit of all
models, especially at low doses.

4.	Uncertainty introduced by estimation of occupational exposures.

Response: (1) The study of Steenland et al. (2001, 197433) was selected to illustrate the
possibilities and limitations of quantitative uncertainty analysis for this type of study (see
Section 6.4.2.2). (2) The possibilities for uncertainty quantification with regard to the
POD are discussed in Section 6.4.1.1 and in the POD bullet above. (3) Goodness of fit at
any measured dose is evaluated in standard packages. There may be different models
with comparable goodness of fit at observed doses which differ strongly at doses outside
the measured range. Extra model uncertainty, that is, uncertainty which is not conditional
on the truth of any given model, is addressed by the exotic methods (see Section 6.4.2).
(4) The feasibility of quantifying uncertainty in occupational exposure is study specific.
The example of Steenland et al. (2001, 197433) was discussed in some detail (see
Section 6.4.2.2). In general, the problem is not so much quantifying the exposure
uncertainty, but in quantifying the dependence between the endpoints and the exposure
uncertainty.

6.5.2. How Forward? Beyond RfDs and Cancer Slope Factors to Development of
Predictive Human Dose-Response Functions

Uncertainty quantification is an emerging area in science. There are many examples of
highly vetted and peer-reviewed uncertainty analyses based on structured expert judgment.

Under this process, experts in effect synthesize a wide diversity of information in generating
their subjective probability distributions. Where considerable data exist for an environmental
pollutant, such as for the well-studied TCDD, it is natural to ask whether these extensive data can
be leveraged more directly in uncertainty quantification. This is an area where research could be
focused. The requisite knowledge does not yet exist, but there are promising lines of attack. It is

This document is a draft for review purposes only and does not constitute Agency policy.

6-41 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

therefore not a question of convening blue-ribbon panels to reveal the proper approach; instead
multiple approaches should be encouraged, to try out new ideas and share experiences.

An important idea that has been pioneered in Europe is to organize bench-test exercises
where different approaches are applied to a common problem. This focuses the discussion on
real issues and builds a community of capable practitioners. Such initiatives have proven much
more productive than simply supporting individual researchers to explore their ideas.

Areas for which bench-test exercises might be appropriate include:

•	Testing "exotic" methods for capturing model uncertainty;

•	Combining bioassay and epidemiological data for uncertainty quantification;

•	Assessing applicability of structured expert judgment, e.g., for low-dose extrapolation;
and

•	Conducting dependence modeling, dependence inference, and dependence elicitation
(such as with regard to TEFs).

Looking beyond compounds for which considerable data exist, there will always be a
need to evaluate new substances. The target will be a simple method that:

1.	Can yield predictions of toxicological indicators with uncertainty via a valid probabilistic
mechanism;

2.	Could evolve from approaches based on similarities (such as a random chemical model)
under which the new substance could be seen as a random sample from a reference
distribution of chemicals considered sufficiently similar, e.g., in terms of structure,
physicochemical properties, and biological activity (potency); and

3.	Is consistent with current risk assessment science and approaches, peer-reviewed and
accepted as EPA policy.

This last feature is important because advancements in risk assessment approaches should
extend logically from current methodology based on data analysis and scientific methods. For
example, the discussion surrounding uncertainty factors suggests that a probabilistically valid
inference system could substantially differ from the current system. Nonetheless, to meld with
current practice, it must initialize on the current system and allow this system to evolve in a
measured fashion. Ideally, methodological changes should be undertaken in a forum where such
issues are being addressed and not within an assessment of a single chemical.

This document is a draft for review purposes only and does not constitute Agency policy.

6-42 DRAFT—DO NOT CITE OR QUOTE

-------
1	Additional research topics relevant to dioxin that could further inform health assessments

2	include population variability of biokinetic constants, threshold mechanisms for the mass action

3	model, and low-frequency polymorphisms (e.g., less than 1%). Further data and improved

4	methodologies in these areas, combined with developments illustrated elsewhere in this report,

5	will help reduce uncertainties and strengthen our understanding of potential health implications

6	of environmental contaminants.

This document is a draft for review purposes only and does not constitute Agency policy.

6-43 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

Table 6-1. Key sources of uncertainty

Selection of endpoint and of species/strain, gender, life stage, other subject characteristics

-	critical effect

-	sensitivity (e.g., species, life stage)

	- human relevance	

Selection of key study(ies): human data and bioassays (strength, inclusion criteria)

-	epidemiological studies, clinical/case reports (exposure estimate)

-	adequacy of study design, statistical power (exposure term, histopathology)

-	human relevance of bioassays (TK, MO A, endpoint)

	- data uncertainty, confidence in data; database deficiencies	

Use of TK, dosimetry; body burden; species differences, cross-species extrapolation

-	bioavailability, dose dependence

-	half life, life stage, body fat, other compartments, age, other factors

-	body burden (peak, steady state, lifetime average)

-	physiologically-based pharmacokinetic (PBPK) modeling

	- scaling (human equivalents), adjustments (default and nondefault; with TP)	

Selection of dose metric

-	intake (averaging time)

-	background (what place on the dose-response curve)

-	free vs. receptor-bound TCDD

-	tissue-specific concentration

	- lipid-normalized level	

Selection of POD

-	selection (e.g., NOAEL/LOAEL, BMDL, ED0i, 05,10)

-	derivation method (e.g., BMD)

-	choice of model form (e.g., Hill, Weibull)

	- statistical uncertainty at/confidence in POD	

Selection of dose-response model (e.g., biologically based, multistage) and of BMR

-	biological plausibility, MOA

-	model type and form, alternative functional forms

-	range of plausible parameter values

	- goodness of fit, especially at low doses	

Selection of low-dose extrapolation approach

-	linear/nonlinear

	- threshold/nonthreshold	

Human population variability

-	subpopulations (e.g., occupational, general public, sensitive groups)

-	polymorphisms

-	life stage, other features

	- individual vs. population threshold	

Characterization of risk/effect

-	adversity of effect (vs. in normal range of variation and adaptation)

-	uncertainty factors (TK; TD; chemical-specific vs. default; justification)

-	consistency of methods for endpoints with common MOA

-	back-extrapolation from occupational data

	- MOE, RfD; beyond a point estimate for SF	

3	PBPK = physiologically-based pharmacokinetic; SF = slope factor; TD = toxicodynamic;

4	TK = toxicokinetic. (Other acronyms are as defined elsewhere within this section.)

This document is a draft for review purposes only and does not constitute Agency policy.

6-44 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table 6-2. PODs and amenability for uncertainty quantification

2

POD

Data profile

Choice

Uncertainty quantification

LOAEL

Experimental dose
level from set of
exposure-response data

Choose set of

exposure-response

measurements

No

NOAEL

Experimental dose
level from set of
exposure-response data

Choose set of

exposure-response

measurements

No

BMDL

Estimate from
bioassay data

Choose BMR, choose
dose-response relation

No, the BMDL is a quantile of
an uncertainty distribution
assuming that the
dose-response model is true

EDX

Estimate from set of
exposure-response data

Choose bioassay
experiments to estimate EDX

Yes, if full bioassay data are
available

3

This document is a draft for review purposes only and does not constitute Agency policy.

6-45 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10000.0

1000.0

Q

Q 00.0

u

0)

u

8>

CL

1.0

0.1

'

w. ••

i .<#•>.

o i	• 	

:	jsp

o	J	• • •

10.0:

i i i -*171	1—i—rrri rn	r m:

Tvnf

0.1	1,0	10.0 100.0 1000.0 10000.0 100000.0

Back-Extrapolated TCDD

Figure 6-1. Back-casted vs. predicted TCDD serum levels for a worker
subset.

Source: Steenland et al. (2001, 197433).

This document is a draft for re\>iew purposes only and does not constitute Agency policy.

6-46 DRAFT—DO NOT CITE OR QUOTE

-------
LEGEND

12378-PeCDD
123478-HxCDD
123789-HxCDD
1234678-HpCDD
OCDD
TCDF
12378-PeCDF
23478-PeCDF
123478-HxCDF
123678-HxCDF
234678-HxCDF
OCDF

h

H

10th 25th: 50th 75th 90th

who1996tef

3]—I

b

I	HE

I

i—rn—
i>+

3—I'

0.00001

0.0001	0.001	0.01

Relative Potencies (REPs)

0.1

PCB-77
q PCB-126

a

PCB-169

PCB-105
PCB-114
PCB-118
PCB-123
PCB-156
PCB-157
PCB-189

i—m—i

:—i

i	c

KB"

O

i	cn:

LEGEND

4-CTH

HT^H

10th 25th; 50th 75th 90th

who1998tef

0.0000001 0.000001 0.00001	0.0001	0.001	0.01	0.1	1

Relative Potencies (REPs)

Figure 6-2. Distribution of in vivo unweighted REP values in the 2004
database.

Source: Van den Berg et al. (2006, 543769). reprinted with permission from Haws

etal. (2006,198416)-

This document is a draft for review purposes only and does not constitute Agency policy.

6-47 DRAFT—DO NOT CITE OR QUOTE

-------
1	REFERENCES

2

3

4	Abbott BD; Birnbaum LS; Diliberto JJ (1996). Rapid Distribution of 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD)

5	to Embryonic Tissues in C57BL/6N Mice and Correlation with Palatal Uptakein Vitro. Toxicol Appl Pharmacol,

6	141:256-263.155093

7	Abraham K; Geusau A; Tosun Y; Helge H; Bauer S; Brockmoller J (2002). Severe 2,3,7,8-tetrachlorodibenzo-p-

8	dioxin (TCDD) intoxication: insights into the measurement of hepatic cytochrome P450 1A2 induction. Clin

9	Pharmacol Ther, 72: 163-174. 197034

10	Abraham K; Knoll A; Ende M; Papke O; Helge H (1996). Intake, fecal excretion, and body burden of

11	polychlorinated dibenzo-p-dioxins and dibenzofurans in breast-fed and formula-fed infants. Pediatr Res, 40: 671-

12	679.548782

13	Abraham K; Krowke R; Neubert D (1988). Pharmacokinetics and biological activity of 2,3,7,8-tetrachlorodibenzo-

14	p-dioxin. 1. Dose-dependent tissue distribution and induction of hepatic ethoxyresorufin O-deethylase in rats

15	following a single injection. Arch Toxicol, 62: 359-368. 199510

16	Ailhaud G (2006). Adipose tissue as a secretory organ: from adipogenesis to the metabolic syndrome. C R Biol, 329:

17	570-577. 549255

18	Aittomaki A; Lahelma E; Roos E; Leino-Aijas P; Martikainen P (2005). Gender differences in the association of age

19	with physical workload and functioning. Br Med J, 62: 95-100. 197139

20	Akhmedkhanov A, Revich B, Adibi JJ, Zeilert V, Masten SA, Patterson DG Jr, Needham LL, Toniolo P (2002).

21	Characterization of dioxin exposure in residents of Chapaevsk, Russia. J Expo Anal Environ Epidemiol, 12: 409-

22	417.197140

23	Akhtar FZ; Garabrant DH; Ketchum NS; Michalek JE (2004). Cancer in US Air Force veterans of the Vietnam war.

24	J Occup Environ Med, 46: 123. 197141

25	Alaluusua S; Calderara P; Gerthoux PM; Lukinmaa PL; Kovero O; Needham L; Patterson Jr DG; Tuomisto J;

26	Mocarelli P (2004). Developmental dental aberrations after the dioxin accident in Seveso. Environ Health Perspect,

27	112: 1313-1318. 197142

28	Alvarez-Pedrerol M; Ribas-Fito N; Torrent M; Carrizo D; Garcia-Esteban R; Grimalt JO; Sunyer J (2008). Thyroid

29	disruption at birth due to prenatal exposure to beta-hexachlorocyclohexane. Environ Int, 34: 737-740. 594407

30	Amin S; Moore RW; Peterson RE; Schantz SL (2000). Gestational and lactational exposure to TCDD or coplanar

31	PCBs alters adult expression of saccharin preference behavior in female rats. Neurotoxicol Teratol, 22: 675-682.

32	197169

33	Andersen ME; Birnbaum LS; Barton HA; Eklund CR (1997). Regional hepatic CYP1A1 and CYP1A2 induction

34	with 2,3,7,8-tetrachlorodibenzo-p-dioxin evaluated with a multicompartment geometric model of hepatic zonation.

35	Toxicol Appl Pharmacol, 144: 145-155. 197172

36	Andersen ME; Mills JJ; Gargas ML; Kedderis L; Birnbaum LS; Neubert D; Greenlee WF (1993). Modeling

3 7	receptor-mediated processes with dioxin: Implications for pharmacokinetics and risk assessment. Risk Anal, 13:25-

38	36. 196991

39	Anderson LM; Beebe LE; Fox SD; Issaq HJ; Kovatch RM (1991). Promotion of mouse lung tumors by

40	bioaccumulated polychlorinated aromatic hydrocarbons. Exp Lung Res, 17: 455-471. 201761

This document is a draft for review purposes only and does not constitute Agency policy.

R-1 DRAFT—DO NOT CITE OR QUOTE

-------
1	Andersson P; McGuire J; Rubio C; Gardin K; Whitelaw ML; Pettersson S; Hanberg A; Poellinger L (2002). A

2	constitutively active dioxin/aryl hydrocarbon receptor induces stomach tumors. PNAS, 99: 9990-9995. 197101

3	Ariens EJ; van Rossum JM; Koopman PC (1960). Receptor reserve and threshold phenomena. I. Theory and

4	experiments with autonomic drugs tested on isolated organs. Arch Int Pharmacodyn Ther, 127: 459-478. 594279

5	Armstrong BG (1995). Comparing standardized mortality ratios. Ann Epidemiol, 5: 60-64. 594397

6	ATSDR (1998). Toxicological profile for chlorinated dibenzo-p-dioxins (CDDs). Agency for Toxic Substances and

7	Disease Registry. Atlanta, GA.http://www.atsdr.cdc.gov/toxprofiles/tpl04.pdf. 197033

8	Aylward L; Kirman C; Cher D; Hays S (2003). Re: analysis of dioxin cancer threshold. Environ Health Perspect,

9	111: A510. 594305

10	Aylward LL; Bodner KM; Collins JJ; Hays SM (2007). Exposure reconstruction for a dioxin-exposed cohort:

11	Integration of serum sampling data and work histories., 69: 2063-2066. 197175

12	Aylward LL; Bodner KM; Collins JJ; Wilken M, McBride D; Burns CJ; Hays SM; Humphry N (2009). TCDD

13	exposure estimation for workers at a New Zealand 2,4,5-T manufacturing facility based on serum sampling data. J

14	Expo Sci Environ Epidemiol, TB A: 1-10. 197187

15	Aylward LL; Brunet RC; Carrier G; Hays SM; Cushing CA; Needham LL; Patterson DG; Gerthoux PM; Brambilla

16	P; Mocarelli P (2005). Concentration-dependent TCDD elimination kinetics in humans: Toxicokinetic modeling for

17	moderately to highly exposed adults from Seveso, Italy, and Vienna, Austria, and impact on dose estimates for the

18	NIOSH cohort. JExpo Anal Environ Epidemiol, 15: 51-65. 197114

19	Aylward LL; Brunet RC; Starr TB; Carrier G; Delzell E; Cheng H; Beall C (2005). Exposure reconstruction for the

20	TCDD-exposed NIOSH cohort using a concentration- and age-dependent model of elimination. Risk Anal, 25: 945-

21	956.197014

22	Aylward LL; Goodman JE; Charnley G; Rhomberg LR (2008). A margin-of-exposure approach to assessment of

23	noncancer risks of dioxins based on human exposure and response data. Environ Health Perspect, 116: 1344-1351 .

24	197068

25	Aylward LL; Hays SM; Karch NJ; Paustenbach DJ (1997). Relative susceptibility of animals and humans to the

26	cancer hazard posed by 2,3,7,8-tetrachlorodibenzo-p-dioxin using internal measures of dose. Environ Sci Tech, 31:

27	1252.594365

28	Baccarelli A; Giacomini SM; Corbetta C; Landi MT; Bonzini M; Consonni D; Grillo P; Patterson DG; Pesatori AC;

29	Bertazzi PA (2008). Neonatal thyroid function in Seveso 25 years after maternal exposure to dioxin. PLoS Med, 5:

30	el61. 197059

31	Baccarelli A; Hirt C; Pesatori AC; Consonni D; Patterson DG Jr; Bertazzi PA; Dolken G; Landi MT (2006). t(14; 18)

32	translocations in lymphocytes of healthy dioxin-exposed individuals from Seveso, Italy. Carcinogenesis, 27: 2001-

33	2007.197036

34	Baccarelli A; Mocarelli P; Patterson DG Jr; Bonzini M; Pesatori AC; Caporaso N; Landi MT (2002). Immunologic
3 5	effects of dioxin: new results from Seveso and comparison with other studies. Environ Health Perspect, 110:1169-

36	1173.197062

37	Baccarelli A; Pesatori AC; Consonni D; Mocarelli P; Patterson DG Jr; Caporaso NE; Bertazzi PA; Landi MT
3 8	(2005). Health status and plasma dioxin levels in chloracne cases 20 years after the Seveso, Italy accident. Br J
39	Dermatol, 152: 459-465. 197053

This document is a draft for review purposes only and does not constitute Agency policy.

R-2 DRAFT—DO NOT CITE OR QUOTE

-------
1	Baccarelli A; Pesatori AC; Masten SA; Patterson DG Jr; Needham LL; Mocarelli P; Caporaso NE; Consonni D;

2	Grassman JA; Bertazzi PA; Landi MT (2004). Aryl-hydrocarbon receptor-dependent pathway and toxic effects of

3	TCDD in humans: a population-based study in Seveso, Italy. Toxicol Lett, 149: 287-293. 197045

4	Bang KM; Kim JH (2001). Prevalence of cigarette smoking by occupation and industry in the United States. Am J

5	Ind Med, 40: 233-239. 197081

6	Banks YB; Birnbaum LS (1991). Absorption of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) after low dose dermal

7	exposure. Toxicol Appl Pharmacol, 107: 302-310. 548742

8	Banks YB; Brewster DW; Birnbaum LS (1990). Age-related changes in dermal absorption of 2,3,7,8-

9	tetrachlorodibenzo-p-dioxinand2,3,4,7,8-pentachlorodibenzofuran. Fundam Appl Toxicol, 15: 163-173. 548741

10	Baron JM; Zwadio-Klarwasser G; JugertF; Hamann W; Riibben A; MukhtarH; MerkHF (1998). Cytochrome P450

11	IB 1: A major P450 isoenzyme in human blood monocytes and macrophage subsets. Biochem Pharmacol, 56: 1105-

12	1110.548791

13	Barouki R; Coumoul X; Fernandez-Salguero PM (2007). The aryl hydrocarbon receptor, more than a xenobiotic-

14	interacting protein. FEBS J, 581: 3608-3615. 543778

15	Bastomsky CH (1977). Enhanced thyroxine metabolism and high uptake goiters in rats after a single dose of 2,3,7,8-

16	tetrachlorodibenzo-p-dioxin. Endocrinology, 101: 292-296. 548760

17	Bates MN; Buckland SJ; Garrett N; Ellis H; Needham LL; Patterson DG Jr; Turner WE; Russell DG (2004).

18	Persistent organochlorines in the serum of the non-occupationally exposed New Zealand population. Chemosphere,

19	54: 1431-1443. 197113

20	Becher H; Flesch-Janys D; Kauppinen T; Kogevinas M; Steindorf K; Manz A; Wahrendorf J (1996). Cancer

21	mortality in German male workers exposed to phenoxy herbicides and dioxins. Cancer Causes Control, 7:312-321.

22	197121

23	Becher H; Steindorf K; Flesch-Janys D (1998). Quantitative cancer risk assessment for dioxins using an

24	occupational cohort. Environ Health Perspect, 106: 663-670. 197173

25	Beebe LE; Anver MR; Riggs CW; Fornwald LW; Anderson LM (1995). Promotion of N-nitrosodimethylamine-

26	initiated mouse lung tumors following single or multiple low dose exposure to 2,3,7,8- tetrachlorodibenzo-p-dioxin.

27	Carcinogenesis, 16: 1345-1349. 548754

28	Bell DR; Clode S; Fan MQ; Fernandes A; Foster PM; Jiang T; Loizou G; MacNicoll A; Miller BG; Rose M; Tran L;

29	White S (2007). Relationships between tissue levels of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), mRNAs and

30	toxicity in the developing male Wistar(Han) rat. Toxicol Sci, 99: 591-604. 197050

31	Bell DR; Clode S; Fan MQ; Fernandes A; Foster PM; Jiang T; Loizou G; MacNicoll A; Miller BG; Rose M; Tran L;

32	White S (2007). Toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin in the developing male Wistar(Han) rat. II: Chronic

33	dosing causes developmental delay . Toxicol Sci, 99: 224-233. 197041

34	Bernert JT; Turner WE; Patterson DG; Needham LL (2007). Calculation of serum total lipid concentrations for the

35	adjustment of persistent organohalogen toxicant measurements in human samples. Chemosphere, 68: 824-831.

36	594270

37	Bertazzi A; Pesatori AC; Consonni D; Tironi A; Landi MT; Zocchetti C (1993). Cancer incidence in a population

38	accidentally exposed to 2,3,7,8-tetrachlorodibenzo-para-dioxin. Epidemiology, 4: 398-406. 192445

39	Bertazzi PA; Consonni D; Bachetti S; Rubagotti M; Andrea Baccarelli A; Zocchetti C; Pesatoril AC (2001). Health

40	effects of dioxin exposure: a 20-year mortality study. Am J Epidemiol, 153: 1031-1044. 197005

This document is a draft for review purposes only and does not constitute Agency policy.

R-3 DRAFT—DO NOT CITE OR QUOTE

-------
1	Bertazzi PA; Zocchetti C; Guercilena S; Consonni D; Tironi A; Landi MT; Pesatori AC (1997). Dioxin exposure

2	and cancer risk: A 15-year mortality study after the "Seveso accident". Epidemiology, 8: 646-652. 197097

3	Bertazzi PA; Zocchetti C; Pesatori AC; Guercilena S; Sanarico M; Radice L (1989). Ten-year mortality study of the

4	population involved in the Seveso incident in 1976. Am J Epidemiol, 129: 1187-1200. 197013

5	Birnbaum LS (1986). Distribution and excretion of 2,3,7,8-tetrachlorodibenzo-p-dioxin in congenic strains of mice

6	which differ at the Ah locus. Drug Metab Dispos, 14: 34-40. 548749

7	Blankenship A; Matsumura F (1997). 2,3,7,8-Tetrachlorodibenzo-p-dioxin-induced activation of a protein tryosine

8	kinase, pp60src, in murine hepatic cytosol using a cell-free system. Mol Pharmacol, 52: 667-675. 543751

9	Bock KW (1994). Aryl hydrocarbon or dioxin receptor: biologic and toxic responses. Rev Physiol Biochem

10	Pharmacol, 125: 1-42. 543755

11	Bock KW; Gschaidmeier H; Heel H; Lehmkoster T; Miinzel PA; Raschko F; Bock-Hennig B (1998). AH receptor-

12	controlled transcriptional regulation and function of rat and human UDP-glucuronosyltransferase isoforms. Adv

13	Enzyme Regul, 38: 207-222. 548752

14	Bodner K; Collins J; Bloemen L; Carson M (2003). Cancer risk for chemical workers exposed to 2,3,7,8-

15	tetrachlorodibenzo-p-dioxin. Occup Environ Med, 60: 672-675. 197135

16	Bond GG; McLaren EA; Brenner FE; Cook RR (1989). Incidence of chloracne among chemical workers potentially

17	exposed to chlorinated dioxins. J Occup Environ Med, 31: 771-774. 064967

18	Bond GG; Wetterstroem NH; Roush GJ; McLaren EA; Lipps TE; Cook RR (1988). Cause specific mortality among

19	employees engaged in the manufacture, formulation, or packaging of 2,4-dichlorophenoxyacetic acid and related

20	salts. Occup Environ Med, 45: 98-105. 197183

21	Boverhoff DR; BurgoonLD; Tashiro C; ChittimB; Harkema JR; Jump DB; Zacharewski TR (2005). Temporal and

22	dose-dependent hepatic gene expression patterns in mice provide new insights into TCDD-mediated hepatotoxicity.

23	Toxicol Sci, 85: 1048-1063. 594260

24	Bowman RE; Schantz SL; Gross ML; Ferguson SA (1989). Behavioral effects in monkeys exposed to 2,3,7,8-

25	TCDD transmitted maternally during gestation and for four months of nursing. Chemosphere, 18: 235-242. 543745

26	Bowman RE; Schantz SL; Weerasinghe NCA; Gross ML; Barsotti DA (1989). Chronic dietary intake of 2,3,7,8-

27	tetrachlorodibenzo-p-dioxin (TCDD) at 5 or 25 parts per trillion in the monkey: TCDD kinetics and dose-effect

28	estimate of reproductive toxicity. Chemosphere, 18: 243-252. 543744

29	Brand KP; Catalano PJ; Hammitt JK; Rhomberg L; Evans JS (2001). Limitations to empirical extrapolation studies:

30	the case of BMD ratios. Risk Anal, 21: 625-640. 543765

31	Brand KP; Rhomberg L; Evans JS (1999). Estimating noncancer uncertainty factors: are ratios NOAELs

32	informative? Risk Anal, 19: 295-308. 007629

33	Bretagnolle J; Huber-Carol C (1988). Effects of omitting covariates in Cox's model of survival data., 15: 125-138.

34	543772

3 5	Brouwer A; Morse DC; Lans MC; Schuur AG; Murk AJ; Klasson-Wehler E; Bergman A; Visser TJ (1998).

36	Interactions of persistent environmental organohalogens with the thyroid hormone system: Mechanisms and possible

37	consequences for animal and human health. Toxicol Ind Health, 14: 59-84. 201801

This document is a draft for review purposes only and does not constitute Agency policy.

R-4 DRAFT—DO NOT CITE OR QUOTE

-------
1	Brown J; Goossens LH; Kraan BCP (1997). Probabilistic accident consequence uncertainty study: food chain

2	uncertainty assessment. U.S. Nuclear Regulatory Commission; Commission of the European Communities.

3	Washington, DC; Brussels, Belgium. NUREG/CR-6523, EUR 16771, SAND97-0335. 543739

4	Brown NM; Manzolillo PA; Zhang J-X; Wang J; Lamartiniere CA (1998). Prenatal TCDD and predisposition to

5	mammary cancer in the rat. Carcinogenesis, 19: 1623-1629. 051311

6	Budinsky RA; Paustenbach D; Fontaine D; Landenberger B; Starr TB (2006). Recommended relative potency

7	factors for 2,3,4,7,8 pentachlorodibenzofuran: The impact of different dose metrics. Toxicol Sci, 91: 275-285.

8	594248

9	Buelke-Sam J; Holson JF; Nelson CJ (1982). Blood flow during pregnancy in the rat: II Dynamics of and litter

10	variability in uterine flow. Teratology, 26: 279-288. 020478

11	Buelke-Sam J; Nelson CJ; Byrd RA; Holson JF (1982). Blood flow during pregnancy in the rat: I Flow patterns to

12	maternal organs. Teratology, 26: 269-277. 020477

13	Bueno de Mesquita HB; Doornbos G; Van der Kuip DA; Kogevinas M; Winkelmann R (1993). Occupational

14	exposure to phenoxy herbicides and chlorophenols and cancer mortality in The Netherlands., 23: 289-300. 196993

15	Burleson GR; Lebrec H; Yang YG; Ibanes JD; Pennington KN; Birnbaum LS (1996). Effect of 2,3,7,8-

16	tetrachlorodibenzo-p-dioxin (TCDD) on influenza virus host resistance in mice. Fundam Appl Toxicol, 29: 40-47.

17	196998

18	Bussard D; Preuss P; White P (2009). Conclusions. In RM Cooke (Ed.),Uncertainty modeling in dose response:

19	bench testing environmental toxicity (pp. 217-224). New York, NY: John Wiley & Sons, Inc. 543770

20	Calvo RM; Jauniaux E; Gulbis B; Asuncion M; Gervy C; Contempre B; Morreale De Escobar G (2002). Fetal

21	tissues are exposed to biologically relevant free thyroxine concentrations during early phases of development. J Clin

22	Endocrinol Metab, 87: 1768-1777. 051690

23	Cantoni L; Salmona M; Rizzardini M (1981). Porphyrogenic effect of chronic treatment with 2,3,7,8-

24	tetrachlorodibenzo-p-dioxin in female rats. Dose-effect relationship following urinary excretion of porphyrins.

25	Toxicol Appl Pharmacol, 57: 156-163. 197092

26	Carrier G; Brunet RC; Brodeur J (1995). Modeling of the toxicokinetics of polychlorinated dibenzo-p-dioxins and

27	dibenzofurans in mammalians, including humans. I. Nonlinear distribution of PCDD/PCDF body burden between

28	liver and adipose tissues. Toxicol Appl Pharmacol, 131: 253-266. 197618

29	Carrier G; Brunet RC; Brodeur J (1995). Modeling of the toxicokinetics of polychlorinated dibenzo-p-dioxins and

30	dibenzofurans in mammalians, including humans: II. Kinetics of absorption and disposition of PCDDs/PCDFs .

31	Toxicol Appl Pharmacol, 131: 267-276. 543780

32	CDC (2004). The health consequences of smoking: A report of the Surgeon General. Centers for Disease Control

33	and Prevention, U.S. Department of Health and Human Services. Washington, DC. 056384

34	Cesana GC; de Vito G; Ferrario M; Sega R; Mocarelli P (1995). Trends of smoking habits in northern Italy (1986-
3 5	1990). The WHO MONICA Project in Area Brianza, Italy. MONICA Area Brianza Research Group. Eur J

36	Epidemiol, 11: 251-258. 594366

37	Checkoway H; Pearce N; Crawford-Brown DJ (1989). Research methods in occupational epidemiology. 027173

3 8	Cheng H; Aylward L; Beall C; Starr TB; Brunet RC (2006). TCDD exposure-response analysis and risk assessment.

39	Risk Anal. 26: 1059-1071. 523122

This document is a draft for review purposes only and does not constitute Agency policy.

R-5 DRAFT—DO NOT CITE OR QUOTE

-------
1	Chevrier J; Eskenazi B; Bradman A; Fenster L; Barr DB (2007). Associations between prenatal exposure to

2	polychlorinated biphenyls and neonatal thyroid-stimulating hormone levels in a Mexican-American population,

3	Salinas Valley, California. Environ Health Perspect, 115: 1490-1496. 594408

4	Chiaro CR; Morales JL; Prabhu KS; Perdew GH (2008). Leukotriene A4 metabolites are endogenous ligands for the

5	AH receptor. Biochemistry, 47: 8445-8455. 543771

6	Choi BC (1992). Definition, sources, magnitude, effect modifiers, and strategies of reduction of the healthy worker

7	effect. J Occup Med, 34: 979-988. 594250

8	Chu I; Lecavalier P; Hakansson H; Yagminas A; Valli VE; Poon P; Feeley M (2001). Mixture effects of 2,3,7,8-

9	tetrachlorodibenzo-p-dioxin and polychlorinated biphenyl congeners in rats . Chemosphere, 43: 807-814. 521829

10	Clark GC; Tritscher A; Maronpot R; Foley J; Lucier G (1991). Tumor promotion by TCDD in female rats. In

11	Banbury Report 35: biological basis for risk assessment of dioxin and related compounds (pp. 389-404). Cold

12	Spring Harbor, NY: Cold Spring Harbor Laboratory. 594378

13	Clegg LX; Li FP; Hankey BF; Chu K; Edwards BK (2002). Cancer survival among US whites and minorities: a

14	SEER (Surveillance, Epidemiology, and End Results) Program population-based study. Arch Intern Med, 162:

15	1985-1993.594267

16	Clewell HJ; Gentry PR; Covington TR; Sarangapani R; Teeguarden JG (2004). Evaluation of the potential impact of

17	age- and gender-specific pharmacokinetic differences on tissue dosimetry. Toxicol Sci, 79: 381-383. 056269

18	Cohen SM; Boobis AR; Meek ME; Preston RJ; McGregor DB (2006). 4-Aminobiphenyl and DNA reactivity: Case

19	study within the context of the 2006 IPCS Human Relevance Framework for Analysis of a cancer mode of action for

20	humans. CritRev Toxicol, 36: 803-819. 197621

21	Cole P; Trichopoulos D; Pastides H; Starr T; Mandel JS (2003). Dioxin and cancer: A critical review. Regul Toxicol

22	Pharmacol, 38: 378-388. 197626

23	Collins JJ; Bodner K; Aylward LL; Wilken M; Bodnar CM (2009). Mortality rates among trichlorophenol workers

24	with exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Am J Epidemiol, 170: 501-506. 197627

25	Connor KT; Aylward LL (2006). Human response to dioxin: Aryl hydrocarbon receptor (AhR) molecular structure,

26	function, and dose-response data for enzyme induction indicate an impaired human AhR. J Toxicol Environ Health

27	B CritRev, 9: 147-171. 197632

28	Consonni D; Pesatori AC; Zocchetti C; Sindaco R; D'Oro LC; Rubagotti M; Bertazzi PA (2008). Mortality in a

29	population exposed to dioxin after the Seveso, Italy, accident in 1976: 25 years of follow-up. Am J Epidemiol, 167:

30	847-858. 524825

31	Cooke RM (2009). Uncertainty modeling in dose response: bench testing environmental toxicity. New York, NY:

32	Wiley, John & Sons, Inc. 543763

3 3	Cooper GS; Klebanoff MA; Promislow J; Brock JW; Longnecker MP (2005). Polychlorinated biphenyls and

34	menstrual cycle characteristics. Epidemiology, 16: 191-200. 594401

3 5	Cox DR (2006). Combination of data. In Kotz S; Read CB; Balakrishnan N et al. (Ed.),Encyclopedia of statistical

36	sciences (pp. 1074-1081). Hoboken: Wiley. 594342

37	Crofton KM; Craft ES; Hedge JM; Gennings C; Simmons JE; Carchman RA; Carter WH Jr; DeVito MJ (2005).
3 8	Thyroid-hormone-disrupting chemicals: Evidence for dose-dependent additivity or synergism. Environ Health
39	Perspect, 113: 1549-1554. 197381

This document is a draft for review purposes only and does not constitute Agency policy.

R-6 DRAFT—DO NOT CITE OR QUOTE

-------
1	Croutch CR; Lebofsky M; Schramm KW; Terranova PF; Rozman KK (2005). 2,3,7,8-Tetrachlorodibenzo-p-dioxin

2	(TCDD) and 1,2,3,4,7,8-hexachlorodibenzo-p-dioxin (HxCDD) alter body weight by decreasing insulin-like growth

3	factor I (IGF-I) signaling. Toxicol Sci, 85: 560-571. 197382

4	Crump K (2002). Critical issues in benchmark calculations from continuous data. Crit Rev Toxicol, 32: 133-153.

5	035681

6	Crump Kenny S; Chiu Weihsueh A; Subramaniam Ravi P (2010). Issues in using human variability distributions to

7	estimate low-dose risk. Environ Health Perspect, 118: 387-393. 380192

8	Crump KS; Canady R; Kogevinas M (2003). Meta-analysis of dioxin cancer dose response for three occupational

9	cohorts. Environ Health Perspect, 111: 681-687. 197384

10	Crump KS; Hoel DG; Langley CH; Peto R (1976). Fundamental carcinogenic processes and their implications for

11	low dose risk assessment. Cancer Res, 36: 2973-2979. 003192

12	D'Amico M; Agozzino E; Biagino A; Simonetti A; Marinelli P (1999). Ill-defined and multiple causes on death

13	certificates-a study of misclassification in mortality statistics. Eur J Epidemiol, 15: 141-148. 197389

14	DeCaprio AP; McMartin DN; O'Keefe PW; Rej R; Silkworth JB; Kaminsky LS (1986). Subchronic oral toxicity of

15	2,3,7,8-tetrachlorodibenzo-p-dioxin in the guinea pig: Comparisons with a PCB-containing transformer fluid

16	pyrolysate. Fundam Appl Toxicol, 6: 454-463. 197403

17	DeKoning EP; Karmaus W (2000). PCB exposure in utero and via breast milk. A review. J Expo Anal Environ

18	Epidemiol, 10: 285-293. 548801

19	Delia Porta G; Dragani TA; Sozzi G (1987). Carcinogenic effects of infantile and long-term 2,3,7,8-

20	tetrachlorodibenzo-p-dioxin treatment in the mouse. Tumori, 73: 99-107. 197405

21	Denison MS; Nagy SR (2003). Activation of the aryl hydrocarbon receptor by structurally diverse exogenous and

22	endogenous chemicals. Annu Rev Pharmacol Toxicol, 43: 309-334. 197226

23	DeVito MJ; Ma X; Babish JG; Menache M; Birnbaum LS (1994). Dose-response relationships in mice following

24	subchronic exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin: CYP1A1, CYP1A2, estrogen receptor, and protein

25	tyrosine phosphorylation. Toxicol Appl Pharmacol, 124: 82-90. 197278

26	Diliberto JJ; Akubue PI; Luebke RW; Birnbaum LS (1995). Dose-response relationships of tissue distribution and

27	induction of CYP1A1 and CYP1A2 enzymatic activities following acute exposure to 2,3,7,8-tetrachlorodibenzo-p-

28	dioxin (TCDD) in mice. Toxicol Appl Pharmacol, 130: 197-208. 197309

29	Diliberto JJ; Burgin DE; Birnbaum LS (1997). Role of CYP1A2 in hepatic sequestration of dioxin: Studies using

30	CYP1A2 knock-out mice. Biochem Biophys Res Commun, 236: 431-433. 548755

31	Diliberto JJ; Burgin DE; Birnbaum LS (1999). Effects of CYP1A2 on Disposition of 2,3,7,8-Tetrachlorodibenzo-p-

32	dioxin, 2,3,4,7,8-Pentachlorodibenzofuran, and 2,2',4,4',5,5'-Hexachlorobiphenyl in CYP1A2 Knockout and Parental

33	(C57BL/6N and 129/Sv) Strains of Mice. Toxicol Appl Pharmacol, 159: 52-64. 143713

34	Diliberto JJ; DeVito MJ; Ross DG; Birnbaum LS (2001). Subchronic Exposure of [3H]- 2,3,7,8-tetrachlorodibenzo-
3 5	p-dioxin (TCDD) in female B6C3F1 mice: Relationship of steady-state levels to disposition and metabolism.

36	Toxicol Sci, 61: 241-255. 197238

37	Diliberto JJ; Jackson JA; Birnbaum LS (1996). Comparison of 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD)

3 8	Disposition Following Pulmonary, Oral, Dermal, and Parenteral Exposures to Rats. Toxicol Appl Pharmacol, 138:

39	158-168. 143712

This document is a draft for review purposes only and does not constitute Agency policy.

R-7 DRAFT—DO NOT CITE OR QUOTE

-------
1	Dolwick KM; Schmidt JV; Carver LA; Swanson HI; Bradfield CA (1993). Cloning and expression of a human Ah

2	receptor cDNA. Mol Pharmacol, 44: 911-917. 543762

3	Dragan YP; Schrenk D (2000). Animal studies addressing the carcinogenicity of TCDD (or related compounds) with

4	an emphasis on tumour promotion. Food Addit Contam, 17: 289-302. 197243

5	Dunson DB; Baird DD (2001). A flexible parametric model for combining current status and age at first diagnosis

6	data. Biometrics, 57: 396-403. 197248

7	EC (2009). Nuclear energy library: Archives. Retrieved 17-JUL-09, from http://cordis.europa.eu/lp5-

8	euratom/src/lib_docs.htm. 543738

9	Ema M; Ohe N; Suzuki M; Mimura J; Sogawa K; Ikawa S; Fujii-Kuriyama Y (1994). Dioxinbinding activities of

10	polymorphic forms of mouse and human arylhydrocarbon receptors. J Biol Chem, 269: 27337-27343. 197313

11	Emond C; Birnbaum LS; DeVito MJ (2004). Physiologically based pharmacokinetic model for developmental

12	exposures to TCDD in the rat. Toxicol Sci, 80: 115-133. 197315

13	Emond C; Birnbaum LS; DeVito MJ (2006). Use of a physiologically based pharmacokinetic model for rats to study

14	the influence of body fat mass and induction of CYP1A2 on the pharmacokinetics of TCDD. Environ Health

15	Perspect, 114: 1394-1400. 197316

16	Emond C; Michalek JE; Birnbaum LS; DeVito MJ (2005). Comparison of the use of a physiologically based

17	pharmacokinetic model and a classical pharmacokinetic model for dioxin exposure assessments. Environ Health

18	Perspect, 113: 1666-1668. 197317

19	Eskenazi B; Mocarelli P; Warner M; Chee WY; GerthouxPM; Samuels S; NeedhamLL; Patterson DG Jr (2003).

20	Maternal serum dioxin levels and birth outcomes in women of Seveso, Italy. Environ Health Perspect, 111: 947-953.

21	197158

22	Eskenazi B; Mocarelli P; Warner M; Needham L; Patterson DG Jr; Samuels S; Turner W; Gerthoux PM; Brambilla

23	P (2004). Relationship of serum TCDD concentrations and age at exposure of female residents of Seveso, Italy.

24	Environ Health Perspect, 112: 22-27. 197160

25	Eskenazi B; Mocarelli P; Warner M; Samuels S; Vercellini P; Olive D; Needham L; Patterson D; Brambilla P

26	(2000). Seveso Women's Health Study: A study of the effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin on

27	reproductive health. Chemosphere, 40: 1247-1253. 197162

28	Eskenazi B; Mocarelli P; Warner M; Samuels S; Vercellini P; Olive D; Needham LL; Patterson DG, Jr.; Brambilla

29	P; Gavoni N; Casalini S; Panazza S; Turner W; Gerthoux PM (2002). Serum dioxin concentrations and

30	endometriosis: A cohort study in Seveso, Italy. Environ Health Perspect, 110: 629-634. 197164

31	Eskenazi B; Warner M; Marks AR; Samuels S; Gerthoux PM; Vercellini P; Olive DL; Needham L; Patterson D Jr;

32	Mocarelli P (2005). Serum dioxin concentrations and age at menopause. Environ Health Perspect, 113: 858-862.

33	197166

34	Eskenazi B; Warner M; Mocarelli P; Samuels S; Needham LL; Patterson DG Jr; Lippman S; Vercellini P; Gerthoux

35	PM; Brambilla P; Olive D (2002). Serum dioxin concentrations and menstrual cycle characteristics. Am J

36	Epidemiol, 156: 383-392. 197168

37	Eskenazi B; Warner M; Samuels S; Young J; Gerthoux PM; Needham L; Patterson D; Olive D; Gavoni N;
3 8	Vercellini P; Mocarelli P (2007). Serum dioxin concentrations and risk of uterine leiomyoma in the Seveso
39	Women's Health Study. Am J Epidemiol, 166: 79-87. 197170

This document is a draft for review purposes only and does not constitute Agency policy.

R-8 DRAFT—DO NOT CITE OR QUOTE

-------
1	Fattore E; Trossvik C; Hakansson H (2000). Relative potency values derived from hepatic vitamin A reduction in

2	male and female Sprague-Dawley rats following subchronic dietary exposure to individual polychlorinated dibenzo-

3	p-dioxin and dibenzofuran congeners and a mixture therof. Toxicol Appl Pharmacol, 165: 184-194. 197446

4	Fernandez-Salguero PM; Hilbert DM; Rudikoff S; Ward JM; Gonzalez FJ (1996). Aryl-hydrocarbon receptor-

5	deficient mice are resistant to 2,3,7,8-tetrachlorodibenzo-p-dioxin-induced toxicity. Toxicol Appl Pharmacol, 140:

6	173-179. 197650

7	Ferriby LL; Knutsen JS; Harris M; Unice KM; Scott P; Nony P; Haws LC; Paustenbach D (2007). Evaluation of

8	PCDD/F and dioxin-like PCB serum concentration data from the 2001-2002 National Health and Nutrition

9	Examination Survey of the United States population. J Expo Sci Environ Epidemiol, 17: 358-371. 548789

10	Fielden MR; Brennan R; Gollub J (2007). A gene expression biomarker provides early prediction and mechanistic

11	assessment of hepatic tumor induction by nongenotoxic chemicals. Toxicol Sci, 99: 90-100. 197298

12	Fingerhut MA; Halperin WE; Marlow DA; Piacitelli LA; Honchar PA; Sweeney MH; Greife AL; Dill PA;

13	SteenlandK; Suruda AJ (1991). Cancer mortality in workers exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin. NEngl

14	J Med, 324: 212-218. 197301

15	Fingerhut MA; Halperin WE; Marlow DA; Piacitelli LA; Honchar PA; Sweeney MH; Greife AL; Dill PA;

16	Steenland K; Suruda AJ (1991). Mortality of U.S. workers employed in the production of chemicale contaminated

17	with 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). U.S. Department of Health and Human Services. Cincinnati, OH.

18	197375

19	Fisher JW; Whittaker TA; Taylor DH; Clewell HJ III; Andersen ME (1989). Physiologically based pharmacokinetic

20	modeling of the pregnant rat: a multiroute exposure model for trichloroethylene and its metabolite, trichloroacetic

21	acid. Toxicol Appl Pharmacol, 99: 395-414. 065288

22	Flesch-Janys D (1997). Analyses of exposure to polychlorinated dibenzo-p-dioxins, furans, and

23	hexachlorocyclohexane and different health outcomes in a cohort of former herbicide-producing workers in

24	Hamburg, Germany. Teratog Carcinog Mutagen, 17: 257-264. 197305

25	Flesch-Janys D; Becher H; Gurn P; Jung D; Konietzko J; Manz A; Papke O (1996). Elimination of polychlorinated

26	dibenzo-p-dioxins and dibenzofurans in occupationally exposed persons. J Toxicol Environ Health, 47: 363-378.

27	197351

28	Flesch-Janys D; Berger J; Gurn P; Manz A; Nagel S; Waltsgott H; Dwyer JH (1995). Exposure to polychlorinated

29	dioxins and furans (PCDD/F) and mortality in a cohort of workers from a herbicide-producing plant in Hamburg,

30	Federal Republic of Germany. Am J Epidemiol, 142: 1165-1175. 197261

31	Flesch-Janys D; Gurn P; Jung D; Konietzko J; Manz A; Papke O (1994). First results of an investigation of the

32	elimination of polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/F) in occupationally exposed persons.,

33	21: 93-99. 197372

34	Flesch-Janys D; Steindorf K; Gurn P; Becher H (1998). Estimation of the cumulated exposure to polychlorinated

3 5	dibenzo-p-dioxins/furans and standardized mortality ratio analysis of cancer mortality by dose in an occupationally

36	exposed cohort. Environ Health Perspect, 106: 655-662. 197339

37	Flodstrom S; Ahlborg UG (1991). Promotion of hepatocarcinogenesis in rats by PCDDs and PCDFs. In Gallo MA;
3 8	Scheuplein RJ; van der Heijden (Ed.),Banbury Report 35: biological basis for risk assessment of dioxin and related

39	compounds (pp. 405-414). Cold Spring Harbor, NY: Cold Spring Harbor Laboratory. 548728

40	Fox TR; Best LL; Goldsworthy SM; Mills JJ; Goldsworthy TL (1993). Gene expression and cell proliferation in rat

41	liver after 2,3,7,8-tetrachlorodibenzo-p-dioxin exposure. Cancer Res, 53: 2265-2271. 197344

This document is a draft for review purposes only and does not constitute Agency policy.

R-9 DRAFT—DO NOT CITE OR QUOTE

-------
1	Franc MA; Pohjanvirta R; Tuomisto J; Okey AB (2001). Persistent, low-dose 2,3,7,8-tetrachlorodibenzo-p-dioxin

2	exposure: effect on aryl hydrocarbon receptor expression in a dioxin-resistance model. Toxicol Appl Pharmacol,

3	175: 43-53. 197353

4	Franczak A; Nynca A; Valdez KE; Mizinga KM; Petroff BK (2006). Effects of acute and chronic exposure to the

5	aryl hydrocarbon receptor agonist 2,3,7,8-tetrachlorodibenzo-p-dioxin on the transition to reproductive senescence

6	in female Sprague-Dawley rats. Biol Reprod, 74: 125-130. 197354

7	Fretland AJ; Safe S; Hankinson O (2004). Lack of antagonism of 2,3,7,8-tetrachlorodibenzo-p-dioxin's (TCDDs)

8	induction of cytochrome P4501A1 (CYP1A1) by the putative selective aryl hydrocarbon receptor modulator 6-alkyl-

9	1,3,8-trichlorodibenzofuran (6-MCDF) in the mouse hepatoma cell line Hepa-lclc7. Chem Biol Interact, 150: 161-

10	170.197357

11	Fritz W; Lin TM; Safe S; Moorea RW; Peterson RE (2009). The selective aryl hydrocarbon receptor modulator 6-

12	methyl-1,3,8-trichlorodibenxofuran inhibits prostate tumor metastasis in TRMP mice. Biochem Pharmacol, 77:

13	1151-1160.594372

14	Fujii-Kuriyama Y; Ema M; Mimura J; Matsushita N; Sogawa K (1995). Polymorphic forms of the Ah receptor and

15	induction of the CYP1A1 gene. Pharmacogenetics, 5 (S): 149-153. 543727

16	Funatake CJ; Dearstyne EA; Steppan LB; Shepherd DM; Spanjaard ES; Marshak-Rothstein A; Kerkvliet NI (2004).

17	Early consequences of 2,3,7,8-tetrachlorodibenzo-p-dioxin exposure on the activation and survival of antigen-

18	specific T cells. Toxicol Sci, 82: 129-142. 197267

19	Gasiewicz TA; Henry EC; Collins LL (2008). Expression and activity of aryl hydrocarbon receptors in development

20	and cancer. Crit Rev Eukaryot Gene Expr, 18: 279-321. 473406

21	Gaylor DW; Kodell RL (2000). Percentiles of the product of uncertainty factors for establishing probabilistic risk

22	doses. Risk Anal, 20: 245-250. 548724

23	Ge NL; Elferink CJ (1998). A direct interaction between the aryl hydrocarbon receptor and retinoblastoma protein:

24	linking dioxin signaling to the cell cycle. J Biol Chem, 273: 22708-22713. 197702

25	Geusau A; Abraham K; Geissler K; Sator MO; Stingl G; Tschachler E (2001). Severe 2,3,7,8-tetrachlorodibenzo-p-

26	dioxin (TCDD) intoxication: Clinical and laboratory effects. Environ Health Perspect, 109: 865-869. 197444

27	Geusau A; Schmaldienst S; DerflerK; (2002). Severe 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) intoxication:

28	Kinetics and trials to enhance elimination in two patients. Arch Toxicol, 76: 316-325. 594259

29	Geyer H; Scheunert I; Korte F (1986). Bioconcentration potential of organic environmental chemicals in humans.

30	Regul Toxicol Pharmacol, 6: 313-347. 064899

31	Geyer HJ; Scheuntert I; Rapp K; Kettrup A; Korte F; Greim H; Rozman K (1990). Correlation between acute

32	toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and total body fat content in mammals. Toxicology, 65: 97-

33	107.197700

34	Geyer HJ; Schramm KW; Scheunert I; Schughart K; Buters J; Wurst W; Greim H; Kluge R; Steinberg CE; Kettrup

35	A; Madhukar B; Olson JR; Gallo MA (1997). Considerations on genetic and environmental factors that contribute to

36	resistance or sensitivity of mammals including humans to toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)

37	and related compounds. Ecotoxicol Environ Saf, 36: 213-230. 543768

38	Gielen JE; Nebert DW (1971). Aryl hydrocarbon hydroxylase induction in mammalian liver cell culture. I.

39	Stimulation of enzyme activity in nonhepatic cells and in hepatic cells by phenobarbital, polycyclic hydrocarbons,

40	and 2,2-bis(p-chlorophenyl)-l,l,l-trichloroethane. J Biol Chem, 246: 5189-5198. 543775

This document is a draft for review purposes only and does not constitute Agency policy.

R-10 DRAFT—DO NOT CITE OR QUOTE

-------
1	Goodman DG; Sauer RM (1992). Hepatotoxicity and carcinogenicity in female Sprague-Dawley rats treated with

2	2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD): a pathology working group reevaluation. Regul Toxicol Pharmacol,

3	15: 245-252. 197667

4	Goossens LH; Harrison JD; Harper FT; Kraan BCP; Cooke RM; Hora SC (1998). Probabilistic accident

5	consequence uncertainty assessment: uncertainty assessment for internal dosimetry. U.S. Nuclear Regulatory

6	Commission; Commission of the European Communities. Washington, DC; Brussels-Luxembourg. NUREG/CR-

7	6571, EUR 16773, SAND98-0119. 548726

8	Goossens LH; Kraan BCP; Cooke RM; Ehrhardt J; Fischer F; Hasemann I; Brown J; Jones JA; Smith JG (2001).

9	Nuclear science and technology: Probabilistic accident consequence uncertainty assessment using Cosyma:

10	Uncertainty from the food chain module. European Commission. Luxemborg. EUR 18823EN. 548737

11	Goossens LH; Kraan BCP; Cooke RM; Ehrhardt J; Fischer F; Hasemann I; Jones JA; Brown J; Khursheed A;

12	Phipps A (2001). Probabilistic accident consequence uncertainty assessment using Cosyma: Uncertainty from the

13	dose module. European Commission. Luxemborg. EUR 18825EN. 548738

14	Goossens LH; Kraan BCP; Cooke RM; Jones J; Brown J; Ehrhardt J; Fischer F; Hasemann I (2001). Overall

15	uncertainty analysis. European Commission. Luxemborg. EUR 18826EN. 548731

16	Goossens LH; Kraan BCP; Cooke RM; Jones J; Ehrhardt J (2001). Nuclear science and technology:

17	countermeasures uncertainty assessment. European Commission. Luxemborg. EUR 18821EN. 548732

18	Goossens LH; Kraan BCP; Cooke RM; Jones JA; Ehrhardt J; Fischer F; Hasemann I (2001). Uncertainty from the

19	early and late health effects module. European Commission. Luxemborg. EUR 18824EN. 548735

20	Goossens LHJ; Cooke RM; Kraan BCP (1996). Evaluation of weighting schemes for expert judgment studies. Delft

21	University of Technology. Delft, The Netherlands. 548727

22	Goossens LHJ; Kraan BCP; Cooke RM; Boardman J; Jones JA; Harper FT; Young ML; Hora SC (1997).

23	Probabilistic accident consequence uncertainty analysis: uncertainty assessment for deposited material and external

24	doses. Office for Official Publications of the European Communities. Washington, DC; Brussels-Luxembourg.

25	NUREG/CR-6526, EUR 16772, SAND97-2323. 543752

26	Goossens LJH; Kraan BCP; Cooke RM; Jones J; Brown J; Ehrhardt J; Fischer F; Hasemann I (2001). Methodology

27	and processing techniques. European Commission. Luxembourg. EUR 18827EN. 548730

28	Goossens, LH; Kraan, BCP; Cooke, RM; Jones JA; Ehrhardt J; Fischer F; Hasemann I (2001). Probabilistic accident

29	consequence uncertainty assessment using Cosyma: Uncertainty from the atmospheric dispersion and deposition

30	module. European Commission. Luxemborg. EUR 18822EN. 548734

31	Graham MJ; Lucier GW; Linko P; Maronpot RR; Goldstein JA (1988). Increases in cytochrome P-450 mediated

32	17B-estradiol 2-hydroxylase activity in rat liver microsomes after both acute administration and subchronic

33	administration of 2,3,7,8-tetrachlorodibenzo-p-dioxin in a two-stage hepatocarcinogenises model. Carcinogenesis, 9:

34	1935-1941.594375

35	Grassman JA; Needham LL; Masten SA; Patterson D; Portier CJ; Lucier GW; Walker NJ (2000). Evidence of

36	hepatic sequestration of dioxin in humans? An examination of tissue levels and CYP1A2 expression., 48: 87-90.

37	548762

38	Greenlee WF; Hushka LJ; Hushka DR (2001). Molecular basis of dioxin actions: evidence supporting

39	chemoprotection. Toxicol Pathol, 29: 6-7. 015400

This document is a draft for review purposes only and does not constitute Agency policy.

R-11 DRAFT—DO NOT CITE OR QUOTE

-------
1	Greer MA; Goodman G; Pleus RC; Greer SE (2002). Health effects assessment for environmental perchlorate

2	contamination: The dose response for inhibition of thyroidal radioiodine uptake in humans. Environ Health Perspect,

3	110:927-937.051202

4	Guess HA; Hoel DG (1977). The effect of dose on cancer latency period. J Environ Pathol Toxicol, 1: 279-286.

5	197464

6	Haarmann-Stemmann T; Bothe H; Abel J (2009). Growth factors, cytokines and their receptors as downstream

7	targets of arylhydrocarbon receptor (AhR) signaling pathways. Biochem Pharmacol, 77: 508-520. 197874

8	Haddow JE; Palomaki GE; Allan WC; Williams JR; Knight GJ; Gagnon J; O'Heir CE; Mitchell ML; Hermos RJ;

9	Waisbren SE; Faix JD; Klein RZ (1999). Maternal thyroid deficiency during pregnancy and subsequent

10	neuropsychological development of the child. N Engl J Med, 341: 549-555. 002176

11	HahnME (2002). Aryl hydrocarbon receptors: Diversity and evolution. ChemBiol Interact, 141: 131-160. 099302

12	Hahn ME; Allan LL; Sherr DH (2009). Regulation of constitutive and inducible AHR signaling: complex

13	interactions involving the AHR repressor. Biochem Pharmacol, 77: 485-497. 548725

14	Hahn MW (2009). Distinguishing Among Evolutionary Models for the Maintenance of Gene Duplicates. J Hered,

15	100:605-617.477460

16	Hakk H; Diliberto JJ; Birnbaum LS (2009). The effect of dose on 2,3,7,8-TCDD tissue distribution, metabolism and

17	elimination in CYP1A2 (-/-) knockout and C57BL/6N parental strains of mice. Toxicol Appl Pharmacol, 241: 119-

18	126.594256

19	Harper N; Connor K; Steinberg M; Safe S (1995). Immunosuppressive activity of polychlorinated biphenyl mixtures

20	and congeners: nonadditive (antagonistic) interactions. Fundam Appl Toxicol, 27: 131-139. 202317

21	Harper PA; Wong JY; Lam MS; Okey AB (2002). Polymorphisms in the human AH receptor. Chem Biol Interact,

22	141: 161-187. 198124

23	Harrad S; Wang Y; Sandaradura S; Leeds A (2003). Human dietary intake and excretion of dioxin-like compounds.

24	J Environ Monit, 5: 224-228. 197324

25	Hassoun EA; Al-Ghafri M; Abushaban A (2003). The role of antioxidant enzymes in TCDD-induced oxidative

26	stress in various brain regions of rats after subchronic exposure. Free Radic Biol Med, 35: 1028-1036. 198726

27	Hassoun EA; Li F; Abushaban A; Stohs SJ (2000). The relative abilities of TCDD and its congeners to induce

28	oxidative stress in the hepatic and brain tissues of rats after subchronic exposure. Toxicology, 145: 103-113. 197431

29	Hassoun EA; Wang H; Abushaban A; Stohs SJ (2002). Induction of oxidative stress following chronic exposure to

30	TCDD, 2,3,4,7,8-pentachlorodibenzofuran, and 2,3',4,4',5-pentachlorobiphenyl. J Toxicol Environ Health A Curr

31	Iss, 65: 825-842. 543725

32	Hassoun EA; Wilt SC; Devito MJ; Van Birgelen A; Alsharif NZ; Birnbaum LS; Stohs SJ (1998). Induction of

33	Oxidative Stress in Brain Tissues of Mice after Subchronic Exposure to 2,3,7,8-Tetrachlorodibenzo-p-dioxin., 42:

34	23-27. 136626

3 5	Hattis D; Baird S; Goble R (2002). A straw man proposal for a quantitative definition of the RfD. Drug Chem

36	Toxicol, 25: 403-436. 548720

37	Hattis D; Banati P; Goble R (1999). Distributions of individual susceptibility among humans for toxic effects~for
3 8	what fraction of which kinds of chemicals and effects does the traditional 10-fold factor provide how much

39	protection? AnnNY Acad Sci, 23: 117-142. 594299

This document is a draft for review purposes only and does not constitute Agency policy.

R-12 DRAFT—DO NOT CITE OR QUOTE

-------
1	Hattis D; Burmaster DE (1994). Assessment of variability and uncertainty distributions for practical risk analyses.

2	Risk Anal. 14: 713 - 730. 594301

3	Hattis D; Ginsberg G; Sonawane B; Smolenski S; Russ A; Kozlak M; Goble R (2003). Differences in

4	pharmacokinetics between children and adults- II. Childrens variability in drug elimination half-lives and in some

5	parameters needed for physiologically-based pharmacokinetic modeling. Risk Anal, 23: 117-142. 548773

6	Haws LC; Su SH; Harris M; Devito MJ; Walker NJ; Farland WH; Finley B; Birnbaum LS (2006). Development of a

7	refined database of mammalian relative potency estimates for dioxin-like compounds. Toxicol Sci, 89: 4-30. 198416

8	Henck JM; New MA; Kociba RJ; Rao KS (1981). 2,3,7,8-Tetrachlorodibenzo-p-dioxin: acute oral toxicity in

9	hamsters. Toxicol Appl Pharmacol, 59: 405-407. 543779

10	Henriksen GL; Ketchum NS; Michalek J; Swaby JA (1997). Serum dioxin and diabetes mellitus in veterans of

11	Operation Ranch Hand. Epidemiology, 8: 252-258. 197645

12	Hertz-Picciotto I (1995). Epidemiology and quantitative risk assessment: a bridge from science to policy. Am J

13	Public Health, 85: 484-491. 065678

14	Higgins JPT; Thompson SG; Spiegelhalter DJ (2009). Re-evaluation of random-effects meta analysis., 172: 137 -

15	159.594339

16	HochsteinMS, Jr.; Render JA; Bursian SJ; AulerichRJ (2001). Chronic toxicity of dietary 2,3,7,8-

17	tetrachlorodibenzo-p-dioxin to mink. Vet Hum Toxicol, 43: 134-139. 197544

18	Hoel DG; Portier CJ (1994). Nonlinearity of dose-response functions for carcinogenicity. Environ Health Perspect

19	Suppl, 102 (Suppl 1): 109-113. 198741

20	Hoglund M; Sehn L; Connors JM; Gascoyne RD; Siebert R; Sail T; Mitelman F; Horsman DE (2004). Identification

21	of cytogenetic subgroups and karyotypic pathways of clonal evolution in follicular lymphomas. Genes

22	Chromosomes Cancer, 39: 195-204. 199130

23	Hojo R; Stern S; Zareba G; Markowski VP; Cox C; Kost JT; Weiss B (2002). Sexually dimorphic behavioral

24	responses to prenatal dioxin exposure. Environ Health Perspect, 110: 247-254. 198785

25	Hooiveld M; Heederik DJ; Kogevinas M; Boffetta P; Needham LL; Patterson DG Jr; Bueno-de-Mesquita HB

26	(1998). Second follow-up of a Dutch cohort occupationally exposed to phenoxy herbicides, chlorophenols, and

27	contaminants. Am J Epidemiol, 147: 891-901. 197829

28	Huff JE (1992). 2,3,7,8-TCDD: A potent and complete carcinogen in experimental animals. Chemosphere, 25: 173-

29	176.548757

30	Huff JE; Salmon AG; Hooper NK; Zeise L (1991). Long-term carcinogenesis studies on 2,3,7,8-tetrachlorodibenzo-

31	p-dioxin and hexachlorodibenzo-p-dioxins . Cell Biol Toxicol, 7: 67-94. 197981

32	Hurst CH; Abbott BD; DeVito MJ; Birnbaum LS (1998). 2,3,7,8-Tetrachlorodibenzo-p-dioxin in Pregnant Long

33	Evans Rats: Disposition to Maternal and Embryo/Fetal Tissues., 45: 129-136. 134516

34	Hurst CH; DeVito MJ; Birnbaum LS (2000). Tissue disposition of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) in

35	maternal and developing long-evans rats following subchronic exposure . Toxicol Sci, 57: 275-283. 198806

36	Hurst CH; DeVito MJ; Setzer RW; Birnbaum LS (2000). Acute administration of 2,3,7,8-tetrachlorodibenzo-p-
3 7	dioxin (TCDD) in pregnant Long Evans rats: association of measured tissue concentrations with developmental
38	effects. Toxicol Sci, 53: 411-420. 199045

This document is a draft for review purposes only and does not constitute Agency policy.

R-13 DRAFT—DO NOT CITE OR QUOTE

-------
1	Hutt KJ; Shi Zhanquan; Albertini DF; Petroff BK (2008). The environmental toxicant 2,3,7,8-tetrachlorodibenzo-p-

2	dioxin disrupts morphogenesis of the rat pre-implantation embryo. BMC Developmental Biology, 8: 1-12. 198268

3	IARC (1997). IARC monographs on the evaluation of carcinogenic risks to humans. International Agency for

4	Research on Cancer. Lyon, France. 537123

5	Ikeda M; Tamura M; Yamashita J; Suzuki C; Tomita T (2005). Repeated in utero and lactational 2,3,7,8-

6	tetrachlorodibenzo-p-dioxin exposure affects male gonads in offspring, leading to sex ratio changes in F2 progeny.

7	Toxicol Appl Pharmacol, 206: 351-355. 197834

8	ILSI (1994). Physiological parameter values for PBPK models. Risk Science Institute. Washington, DC. 046436

9	Institute of Medicine (1994). Veterans and Agent Orange. Washington, DC: National Acadmies Press. 594376

10	Institute of Medicine (2006). Veterans and Agent Orange: update 2000. Washington, DC: National Academies

11	Press. 594374

12	Ishihara K; Warita K; Tanida T; Sugawara T; Kitagawa H; Hoshi N (2007). Does paternal exposure to 2,3,7,8-

13	tetrachlorodibenzo-p-dioxin (TCDD) affect the sex ratio of offspring. J Vet Med Sci, 69: 347-352. 197677

14	James WH (1995). What stabilizes the sex ratio? Ann Hum Genet, 59: 243-249. 197722

15	Jorgensen N; Andersen AG; Eustache F; Irvine DS; Suominen J; Petersen JH; Andersen AN; Auger J; Cawood EH;

16	Horte A; Jensen TK; Jouannet P; Keiding N; Vierula M; Toppari J; Skakkebaek NE (2001). Regional differences in

17	semen quality in Europe. Hum Reprod, 16: 1012-1019. 594402

18	Kang HK; Dalager NA; Needham LL; Patterson DG Jr; Lees PS; Yates K; Matanoski GM (2006). Health status of

19	Army Chemical Corps Vietnam veterans who sprayed defoliant in Vietnam. Am J Ind Med, 49: 875-884. 199133

20	Kang SH; Kodell RL; Chen JJ (2000). Incorporating model uncertainties along with data uncertainties in microbial

21	risk assessment. Regul Toxicol Pharmacol, 31: 68-72. 548722

22	Kattainen H; Tuukkanen J; Simanainen U; Tuomisto JT; Kovero O; Lukinmaa P-L; Alaluusua S; Tuomisto J;

23	Viluksela M (2001). In Utero/Lactational 2,3,7,8-Tetrachlorodibenzo-p-dioxin Exposure Impairs Molar Tooth

24	Development in Rats . Toxicol Appl Pharmacol, 174: 216-224. 198952

25	Kauppinen T; Kogevinas M; Johnson E; Becher H; Bertazzi PA; Bueno de Mesquita HB; Coggon D; Green L;

26	Littorin M; Lynge E Mathews J; Neuberger M; Osman J; Pannett B; Pearce N; Winkelmann R; Saracci R (1993).

27	Chemical exposure in manufacture of phenoxy herbicides and chlorophenols and in spraying of phenoxy herbicides.

28	Am J Ind Med, 23: 903-920. 594388

29	Keller JM; Huet-Hudson Y; Leamy LJ (2008). Effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin on molar development

30	among non-resistant inbred strains of mice: A geometric morphometric analysis. Growth Development and Aging,

31	71:3-16.198033

32	Keller JM; Huet-Hudson YM; Leamy LJ (2007). Qualitative effects of dioxin on molars vary among inbred mouse

33	strains. Arch Oral Biol, 52: 450-454. 198526

34	Keller JM; Zelditch ML; Huet YM; Leamy LJ (2008). Genetic differences in sensitivity to alterations of mandible

35	structure caused by the teratogen 2,3,7,8-tetrachlorodibenzo-p-dioxin. Toxicol Pathol, 36: 1006-1013. 198531

36	Kerger BD; Leung H-W; Scott P; Paustenbach DJ; Needham LL; Patterson DG Jr; Gerthoux PM; Mocarelli P

37	(2006). Age- and concentration-dependent elimination half-life of 2,3,7,8-tetrachlorodibenzo-p-dioxin in Seveso

38	children. Environ Health Perspect, 114: 1596-1602. 198651

This document is a draft for review purposes only and does not constitute Agency policy.

R-14 DRAFT—DO NOT CITE OR QUOTE

-------
1	Kerger BD; Leung HW; Scott PK; Paustenbach DJ (2007). Refinements on the age-dependent half-life model for

2	estimating child body burdens of polychlorodibenzodioxins and dibenzofurans. Chemosphere, 67: S272-S278.

3	548784

4	Ketchum NS; Michalek JE; Burton JE (1999). Serum dioxin and cancer in veterans of Operation Ranch Hand. Am J

5	Epidemiol, 149: 630-639. 198120

6	Kim AH; Kohn MC; Nyska A; Walker NJ (2003). Area under the curve as a dose metric for promotional responses

7	following 2,3,7,8-tetrachlorodibenzo-p-dioxin exposure. Toxicol Appl Pharmacol, 191: 12-21. 199146

8	Kitchin KT; Woods JS (1979). 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) effects on hepatic microsomal

9	cytochrome P-448-mediated enzyme activities. Toxicol Appl Pharmacol, 47: 537-546. 198750

10	Kociba RJ; Keeler PA; Park CN; Gehring PJ (1976). 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD): Results of a 13-

11	week oral toxicity study in rats. Toxicol Appl Pharmacol, 35: 553-574. 198594

12	Kociba RJ; Keyes DG; Beyer JE; Carreon RM; Wade CE; Dittenber DA; Kalnins RP; Frauson LE; Park CN;

13	Barnard SD; Hummel RA; Humiston CG (1978). Results of a two-year chronic toxicity and oncogenicity study of

14	2,3,7,8-tetrachlorodibenzo-p-dioxin in rats. Toxicol Appl Pharmacol, 46: 279-303. 001818

15	Kogevinas M; Becher H; Benn T; Bertazzi PA; Boffetta P; Bueno-de-Mesquita HB; Coggon D; Colin D; Flesch-

16	Janys D; Fingerhut M; Green L; Kauppinen T; LJttorin M; Lynge E; Mathews JD; Neuberger M; Pearce N; Saracci

17	R (1997). Cancer mortality in workers exposed to phenoxy herbicides, chlorophenols, and dioxins an expanded and

18	updated international cohort study . Am J Epidemiol, 145: 1061-1075. 198598

19	KohnMC; Lucier GW; Clark GC; Sewall C; Tritscher AM; Portier CJ (1993). A mechanistic model of effects of

20	Dioxin on gene expression in the rat liver. Toxicol Appl Pharmacol, 120: 138-154. 198601

21	Kohn MC; Melnick RL (2002). Biochemical origins of the non-monotonic receptor-mediated dose-response. Journal

22	of Molecular Endocrinology, 29: 113-123. 199104

23	Kohn MC; Sewall CH; Lucier GW; Portier CJ (1996). A mechanistic model of effects of dioxin on thyroid

24	hormones in the rat. Toxicol Appl Pharmacol, 165: 29-48. 022626

25	Kohn MC; Walker NJ; Kim AH; Portier CJ (2001). Physiological modeling of a proposed mechanism of enzyme

26	induction by TCDD. Toxicology, 162: 193-208. 198767

27	Kolluri SK; Weiss C; Koff A; Gottlicher M (). p27(Kipl) induction and inhibition of proliferation by the

28	intracellular Ah receptor in developing thymus and hepatoma cells. Genes Dev, 13: 1742-1753. 548721

29	Kopylev L; Chen C; White P (2007). Towards quantitative uncertainty assessment for cancer risks: central estimates

30	and probability distributions of risk in dose-response modeling. Regul Toxicol Pharmacol, 49: 203-207. 194860

31	Kopylev L; John Fox J; Chen C (2009). Combining risks from several tumors using Markov Chain Monte Carlo. In

32	RM Cooke (Ed.),Uncertainty Modeling in Dose Response (pp. 197-205). Hoboken, NJ: John Wiley & Sons. 198071

33	Kreuzer PE; Csanady GA; Baur C; Kessler W; Papke O; GreimH; Filser JG (1997). 2,3,7,8-Tetrachlorodibenzo-p -

34	dioxin (TCDD) and congeners in infants. A toxicokinetic model of human lifetime body burden by TCDD with

35	special emphasis on its uptake by nutrition. Arch Toxicol, 71: 383-400. 198088

36	Krishnan K; Andersen ME (1991). Interspecies scaling in pharmacokinetics. In A Rescingo; A Thakkur (Ed.),New

37	trends in pharmacokinetics (pp. 203-226). New York, NY: Plenum Press. 548799

This document is a draft for review purposes only and does not constitute Agency policy.

R-15 DRAFT—DO NOT CITE OR QUOTE

-------
1	Krowke R; Chahoud I; Baumann-Wilschke I; Neubert D (1989). Pharmacokinetics and biological activity of 2,3,7,8-

2	tetrachlorodibenzo-p-dioxin 2: pharmacokinetics in rats using a loading-dose/maintenance-dose regime with high

3	doses. Arch Toxicol, 63: 356-360. 198808

4	Kuchiiwa S; Cheng SB; Nagatomo I; Akasaki Y; UchidaM; TominagaM; Hashiguchi W; Kuchiiwa T (2002). In

5	utero and lactational exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin decreases serotonin-immunoreactive neurons

6	in raphe nuclei of male mouse offspring. Neurosci Lett, 317: 73-76. 198355

7	Kurowicka D; Cooke RM (2006). Uncertainty analysis with high dimensional dependence modelling. West Sussex,

8	England: John Wiley & Sons. 543758

9	LaKind JS; Berlin CM; Park CN; Naiman DQ; Gudka NJ (2000). Methodology for characterizing distributions of

10	incremental body burdens of 2,3,7,8-TCDD and DDE from breast milk in North American nursing infants. J Toxicol

11	Environ Health A Curr Iss, 59: 605-639. 198094

12	Lakshmanan MR; Campbell BS; Chirtel SJ; Ekarohita N; Ezekiel M (1986). Studies on the mechanism of absorption

13	and distribution of 2,3,7,8-tetrachlorodibenzo-p-dioxin in the rat. J Pharmacol Exp Ther, 239: 673-677. 548729

14	Landi MT, Consonni D, Patterson DG Jr, Needham LL, Lucier G, Brambilla P, Cazzaniga MA, Mocarelli P,

15	Pesatori AC, Bertazzi PA, Caporaso NE.. (1998). 2,3,7,8-Tetrachlorodibenzo-p-dioxin plasma levels in Seveso 20

16	years after the accident. Environ Health Perspect, 106: 273-277. 594409

17	Landi MT; Bertazzi PA; Baccarelli A; Consonni D; Masten S; Lucier G; Mocarelli P; Needham L; Caporaso N;

18	Grassman J (2003). TCDD-mediated alterations in the AhR-dependent pathway in Seveso, Italy, 20 years after the

19	accident. Carcinogenesis, 24: 673-680. 198362

20	Larsen JC (2006). Risk assessments of polychlorinated dibenzo-p-dioxins, polychloriniated dibenzofurans, and

21	dioxin-like polychlorinated biphenyls in food. Mol Nutr Food Res, 50: 885-896. 548744

22	Latchoumycandane C; Chitra C; Mathur P (2002). Induction of oxidative stress in rat epididymal sperm after

23	exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Arch Toxicol, 76: 113-118. 197839

24	Latchoumycandane C; Chitra KC; Mathur PP (2002). The effect of 2,3,7,8-tetrachlorodibenzo-p-dioxin on the

25	antioxidant system in mitochondrial and microsomal fractions of rat testis. Toxicology, 171: 127-135. 198365

26	Latchoumycandane C; Chitra KC; Mathur PP (2003). 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) induces

27	oxidative stress in the epididymis and epididymal sperm of adult rats. Arch Toxicol, 77: 280-284. 543746

28	Latchoumycandane C; Mathur PP (2002). Effects of vitamin E on reactive oxygen species-mediated 2,3,7,8-

29	tetrachlorodibenzo-p-dioxin toxicity in rat testis. J Appl Toxicol, 22: 345-351. 197498

30	Lawrence GS; Gobas FAPC (1997). A pharmacokinetic analysis of interspecies extrapolation in dioxin risk

31	assessment. Chemosphere, 35: 427-452. 199072

32	Lean MEJ; Han TS; Deurenberg P (1996). Predicting body composition by densitometry from simple

33	anthropometric measurements. Am J Clin Nutr, 63: 4-14. 548770

34	Lee DJ; Fleming LE; Arheart KL; LeBlanc WG; Caban AJ; Chung-Bridges K; Christ SL; McCollister KE; Pitman T
3 5	(2007). Smoking rate trends in U.S. occupational groups: the 1987 to 2004 National Health Interview Survey. J

36	Occup Environ Med, 49: 75-81. 594391

37	Lehman AJ; Fitzhugh OG (1954). 100-fold margin of safety., 18: 33-35. 003195

38	Leo A; Hansch C; Elkins D (1971). Partition coefficients and their uses. ChemRev, 71: 557-558. 019600

This document is a draft for review purposes only and does not constitute Agency policy.

R-16 DRAFT—DO NOT CITE OR QUOTE

-------
1	Leung H-W; Poland A; Paustenbach DJ; Murray FJ; Andersen ME (1990). Pharmacokinetics of [125I]-2-iodo-3,7,8-

2	trichlorodibenzo-p-dioxin in mice: analysis with a physiological modeling approach. Toxicol Appl Pharmacol, 103:

3	411-419. 192833

4	Leung HW; Kerger BD; Paustenbach DJ (2006). Elimination half-lives of selected polychlorinated dibenzodioxins

5	and dibenzofurans in breast-fed human infants. J Toxicol Environ Health A Curr Iss, 69: 437-443. 548779

6	Leung HW; Ku RH; Paustenbach DJ; Andersen ME (1988). A physiologically based pharmacokinetic model for

7	2,3,7,8-tetrachlorodibenzo-p-dioxin in C57BL/6J and DBA/2J mice. Toxicol Lett, 42: 15-28. 198815

8	Li B; Liu HY; Dai LJ; Lu JC; Yang ZM; Huang L (2006). The early embryo loss caused by 2,3,7,8-

9	tetrachlorodibenzo-p-dioxin may be related to the accumulation of this compound in the uterus. Reprod Toxicol, 21:

10	301-306. 199059

11	Li CY; Sung FC (1999). A review of the healthy worker effect in occupational epidemiology. Occup Med (Lond),

12	49: 225-9. 198427

13	Li X; Johnson DC; Rozman KK (1997). 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) increases release of

14	luteinizing hormone and follicle-stimulating hormone from the pituitary of immature female rats in vivo and in vitro.

15	Toxicol Appl Pharmacol, 142: 264-269. 199060

16	Limbird LE (1996). Cell surface receptors: a short course on theory and method. 594276

17	Longnecker MP; Gladen BC; Patterson DG; Rogan WJ (2000). Polychlorinated biphenyl (PCB) exposure in relation

18	to thyroid hormone levels in neonates. Epidemiology, 11: 249-254. 201463

19	LorberM; Patterson D; Huwe J; KahnH (2009). Evaluation of background exposures of Americans to dioxin-like

20	compounds in the 1990s and the 2000s . Chemosphere, 77: 640-651. 543766

21	Lorenzen A; Okey AB (1991). Detection and characterization of Ah receptor in tissue and cells from human tonsils.

22	Toxicol Appl Pharmacol, 107: 203-214. 198397

23	Lucier GW (1991). Humans are a sensitive species to some of the biochemical effects of structural analogs of

24	dioxin. Environ Toxicol Chem, 10: 727-735. 198691

25	Lucier GW; Rumbaugh RC; McCoy Z; Hass R; Harvan D; Albro P (1986). Ingestion of soil contaminated with

26	2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) alters hepatic enzyme activities in rats. Fundam Appl Toxicol, 6: 364-

27	371.198398

28	Lucier GW; Tritscher A; Goldsworthy T; Foley J; Clark G; Goldstein J; Maronpot R (1991). Ovarian hormones

29	enhance 2,3,7,8-tetrachlorodibenzo-p-dioxin-mediated increases in cell proliferation and preneoplastic foci in a two-

30	stage model for rat hepatocarcinogenesis. Cancer Res, 51: 1391-1397. 199007

31	Lutz WK (1990). Dose-response relationship and low dose extrapolation in chemical carcinogenesis.

32	Carcinogenesis, 11: 1243-1247. 000399

33	Lutz WK (1999). Dose-response relationships in chemical carcinogenesis reflect differences in individual

34	susceptibility. Hum Exp Toxicol, 18: 707-712. 594298

35	Lutz WK (2001). Susceptibility differences in chemical carcinogenesis linearize the dose-response relationship:

36	threshold doses can be defined only for individuals. DNA Repair (Amst), 482: 71-76. 053426

37	Lutz WK; Gaylor DW (2008). Letter to the editor. Dose-response relationships for cancer incidence reflect

38	susceptibility distributions. Chem Res Toxicol, 21: 971-972. 594297

This document is a draft for review purposes only and does not constitute Agency policy.

R-17 DRAFT—DO NOT CITE OR QUOTE

-------
1	Lutz WK; Gaylor DW; Conolly RB; Lutz RW (2005). Nonlinearity and thresholds in dose-response relationships for

2	carcinogenicity due to sampling variation, logarithmic dose scaling, or small differences in individual susceptibility.

3	Toxicol Appl Pharmacol, 207: S565-S569. 087763

4	Mackie D; Liu J; Loh Y-S; Thomas V (2003). No evidence of dioxin cancer threshold. Environ Health Perspect,

5	111:1145-1147.594303

6	Mally A; Chipman JK (2002). Non-genotoxic carcinogens: Early effects on gap junctions, cell proliferation and

7	apoptosis in the rat. Toxicology, 180: 233-248. 198098

8	Manchester DK; Gordon SK; Golas CL; Roberts EA; Okey AB (1987). Ah receptor in human placenta: stabilization

9	by molydate and characterization of binding of 2,3,7,8-tetrachlorodibenzo-p-dioxin, 3-methylcholanthrene, and

10	benzo(a)pyrene. Cancer Res, 47: 4861-4868. 198054

11	Manz A; Berger J; Dwyer JH; Flesch-Janys D; Nagel S; Waltsgott H (1991). Cancer mortality among workers in

12	chemical plant contaminated with dioxin. Lancet, 338: 959-964. 199061

13	Markowski VP; Zareba G; Stern S; Cox C; Weiss B (2001). Altered operant responding for motor reinforcement and

14	the determination of benchmark doses following perinatal exposure to low-level 2,3,7,8-tetrachlorodibenzo-p-

15	dioxin. Environ Health Perspect, 109: 621-627. 197442

16	Maronpot RR; Foley JF; Takahashi K; Goldsworthy T; Clark G; Tritscher A; Portier C; Lucier G (1993). Dose

17	response for TCDD promotion of hepatocarcinogenesis in rats initiated with DEN: histologic, biochemical, and cell

18	proliferation endpoints., 101: 643-642. 198386

19	Maronpot RR; Montgomery CA; Boorman GA; McConnell EE (1986). National Toxicology Program nomenclature

20	for hepatoproliferative lesions of rats. Toxicol Pathol, 14: 263-273. 013967

21	Maronpot RR; Pitot HC; Peraino C (1989). Use of rat liver altered focus models for testing chemicals that have

22	completed two-year carcinogenicity studies. Toxicol Pathol, 17: 651-652. 548778

23	Maruyama W; Yoshida K; Tanaka T; Nakanishi J (2002). Determination of tissue-blood partition coefficients for a

24	physiological model for humans, and estimation of dioxin concentration in tissues. Chemosphere, 46: 975-985.

25	198448

26	Matsumoto Y; Ide F; Kishi R; Akutagawa T; Sakai S; Nakamura M; Ishikawa T; Fujii-Kuriyama Y; Nakatsuru Y

27	(2007). Aryl hydrocarbon receptor plays a significant role in mediating airborne particulate-induced carcinogenesis

28	in mice. Environ Sci Tech, 41: 3775-3780. 548748

29	McBride DI, Collins JJ, Humphry NF, Herbison P, Bodner KM, Aylward LL, Burns CJ, Wilken M (2009).

30	Mortality in workers exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin at a trichlorophenol plant in New Zealand. J

31	Occup Med, 51: 1049-56. 198490

32	McBride DI; Burns CJ; Herbison GP; Humphry NF; Bodner K; Collins JJ (2009). Mortality in employees at a New

33	Zealand agrochemical manufacturing site. Occup Med (Lond), 59: 255-263. 197296

34	McEwen LN, Kim C, Haan M, Ghosh D, Lantz PM, Mangione CM, Safford MM, Marrero D, Thompson TJ,

3 5	Herman WH; TRIAD Study Group (2006). Diabetes reporting as a cause of death: results from the Translating

36	Research Into Action for Diabetes (TRIAD) study. Diabetes Care, 29: 247-253. 594400

37	McMichael AJ (1976). Standardized mortality ratios and the "healthy worker effect": scratching beneath the surface.

38	J Occup Environ Med, 18: 165-168. 073484

39	McMillan BJ; Bradfield CA (2007). The aryl hydrocarbon receptor sans xenobiotics: endogenous function in genetic

40	model systems. Mol Pharmacol, 72: 487-498. 543777

This document is a draft for review purposes only and does not constitute Agency policy.

R-18 DRAFT—DO NOT CITE OR QUOTE

-------
1	McNulty WP; Nielsen-Smith KA; Lay JO Jr; Lippstreu DL; Kangas NL; Lyon PA; Gross ML (1982). Persistence of

2	TCDD in monkey adipose tissue. Food Chem Toxicol, 20: 985-986. 543782

3	Michalek JE; Pavuk M (2008). Diabetes and cancer in veterans of Operation Ranch Hand after adjustment for

4	calendar period, days of spraying, and time spent in Southeast Asia. J Occup Environ Med, 50: 330-340. 199573

5	Michalek JE; Pirkle JL; Needham LL; Patterson DG Jr; Caudill SP; Tripathi RC; Mocarelli P (2002).

6	Pharmacokinetics of 2,3,7,8-tetrachlorodibenzo-p-dioxin in Seveso adults and veterans of operation Ranch Hand. J

7	Expo Anal Environ Epidemiol, 12:44-53. 199579

8	Michalek JE; Pirkle JL; Caudill SP; Tripathi RC; Patterson DG Jr; Needham LL (1996). Pharmacokinetics of TCDD

9	in veterans of Operation Ranch Hand: 10-year follow-up. J Toxicol Environ Health, 47: 209-220. 198893

10	Micka J; Milatovich A; Menon A; Grabowski GA; Puga A; Nebert DW (1997). Human Ah receptor (AHR) gene:

11	Localization to 7pl5 and suggestive correlation of polymorphism with CYP1A1 inducibility. Pharmacogenetics, 7:

12	95-101. 548797

13	Miettinen HM; Sorvari R; Alaluusua S; Murtomaa M; Tuukkanen J; Viluksela M (2006). The Effect of Perinatal

14	TCDD exposure on caries susceptibility in rats. Toxicol Sci, 91: 568-575. 198266

15	Milbrath MO; Wenger Y; Chang CW; Emond C; Garabrant D; Gillespie BW; Jolliet O (2009). Apparent half-lives

16	of dioxins, furans, and polychlorinated biphenyls as a function of age, body fat, smoking status, and breast-feeding.

17	Environ Health Perspect, 117: 417-425. 198044

18	Mocarelli P (2001). Seveso: ateaching story. Chemosphere, 43: 391-402. 197002

19	Mocarelli P; Needham LL; MarocchiA; Patterson DG Jr; BrambillaP; GerthouxPM; Meazza L; Carreri V

20	(1991). Serum concentrations of 2,3,7,8-tetrachlorodibenzo-p-dioxin and test results from selected residents of

21	Seveso, Italy . J Toxicol Environ Health A Curr Iss, 32: 357-366. 199600

22	Mocarelli P; Brambilla P; Gerthoux PM; Patterson Jr DG; Needham LL (1996). Change in sex ratio with exposure

23	to dioxin. Lancet, 348: 409. 197637

24	Mocarelli P; Gerthoux PM; Ferrari E; Patterson Jr DG; Kieszak SM; Brambilla P; Vincoli N; Signorini S;

25	Tramacere P; Carreri V; Sampson EJ; Turner WE (2000). Paternal concentrations of dioxin and sex ratio of

26	offspring. Lancet, 355: 1858-1863. 197448

27	Mocarelli P; Gerthoux PM; Patterson DG Jr; Milani S; Limonata G; Bertona M; Signorini S; Tramacere P; Colombo

28	L; Crespi C; Brambilla P; Sarto C; Carreri V; Sampson EJ; Turner WE; Needham LL (2008). Dioxin exposure, from

29	infancy through puberty, produces endocrine disruption and affects human semen quality . Environ Health Perspect,

30	116:70-77.199595

31	Monson RR (1986). Observations on the healthy worker effect. J Occup Environ Med, 28: 425-433. 001410

32	Morreale de Escobar G; Obregon MJ; Escobar del Ray F (2000). Is neuropsychological development related to

33	maternal hypothyroidism or to maternal hypothyroxinemia? J Clin Endocrinol Metab, 85: 3975-3987. 019231

34	Moser GA; McLachlan MS (2001). The influence of dietary concentration on the absorption and excretion of
3 5	persistent lipophilic organic pollutants in the human intestinal tract. Chemosphere, 45: 201 -211. 198045

36	Muller A; De La Rochebrochard E; Labbe-Decleves C; Jouannet P; Bujan L; Mieusset R; Le Lannou D; Guerin JF;

37	Benchaib M; Slama R; Spira A (2004). Selection bias in semen studies due to self-selection of volunteers. Hum

38	Reprod, 19: 2838-2844. 594403

This document is a draft for review purposes only and does not constitute Agency policy.

R-19 DRAFT—DO NOT CITE OR QUOTE

-------
1	Murdoch DJ; Krewski D (1988). Carcinogenic risk assessment with time-dependent exposure patterns. Risk Anal, 8:

2	521-530. 548718

3	Murdoch DJ; Krewski D; Wargo J (1992). Cancer risk assessment with intermittent exposure. Risk Anal, 12: 569-

4	577.548719

5	Murphy JM; Sexton DM; Barnett DN; Jones GS; Webb MJ; Collins M; Stainforth DA (2004). Quantification of

6	modeling uncertainties in a large ensemble of climate change simulations. Nature, 430: 768-772. 543741

7	Murray FJ; Smith FA; Nitschke KD; Humiston CG; Kociba RJ; Schwetz BA (1979). Three-generation reproduction

8	study of rats given 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) in the diet. Toxicol Appl Pharmacol, 50: 241-252.

9	197983

10	Muto T; Wakui S; Imano N; Nakaaki K; Hano H; Furusato M; Masaoka T (2001). In-utero and lactational exposure

11	of 3,3',4,4',5-pentachlorobiphenyl modulate dimethlbenz[a]anthracene-induced rat mammary carcinogenesis. J

12	Toxicol Pathol, 4: 213-224. 548713

13	Myers JE; ThompsonML (1998). Meta-analysis and occupational epidemiology. Occup Med (Lond), 48: 99-101.

14	594395

15	Nagel S; Berger J; Flesch-Janys D; Manz A; Ollroge I (1994). Mortality and cancer mortality in a cohort of female

16	workers of a herbicide producing plant exposed to polychlorinated dibenzo-p-dioxins and furans. Inform Biomet

17	Epidemiol Med Biol, 25: 32-38. 594369

18	NAS (2006). Health risks from dioxin and related compounds. Retrieved 09-FEB-10, from

19	http://www.nap.edu/webcast/webcast_detail.php?webcast_id=328. 543760

20	NAS (2006). Health risks from dioxin and related compounds: Evaluation of the EPA reassessment. National

21	Academy of Science. Washington, DC.http://www.nap.edu/catalog.php?record_id=11688. 198441

22	NAS (2009). Toward a unified approach to dose-response assessment: the need for an improved dose-response

23	framework. National Academics Press. Washington DC. 594307

24	NASA (2002). Probabilistic risk assessment procedures guide for NASA managers and practitioners. National

25	Aeronautics and Space Administration. Washington, DC. 543734

26	Nebert DW; Petersen DD; Fornace AJ Jr (1990). Cellular responses to oxidative stress: the [Ah] gene battery as a

27	paradigm. Environ Health Perspect, 88: 13-25. 548756

28	Nebert DW; Peterson DD; Puga A (1991). Human Ah locus polymorphism and cancer: Inducibility of CYPIA1 and

29	other genes by combustion products and dioxin. Pharmacogenetics, 1: 68-78. 543728

30	Needham LL; Barr DB; Caudill SP; Pirkle JL; Turner WE; Osterloh J; Jones RL; Sampson EJ (2005).

31	Concentrations of environmental chemicals associated with neurodevelopmental effects in the US population.

32	Neurotoxicology, 26: 531-545. 594295

33	Needham LL; Gerthoux PM; Patterson Jr DG; Brambilla P; Prikle JL; Tramacere PL; Turner WE; Beretta c;

34	Sampson EJ; Mocarelli P (1994). Half-life of 2,3,7,8-tetrachlorodibenzo-p-dioxin in serum of Seveso adults: interim

35	report. ,21: 81-85. 200030

36	Nessel CS; Amoruso MA; Umbreit TH; Meeker RJ; Gallo MA (1992). Transpulmonary uptake and bioavailability

37	of 2,3,7,8-TCDD from respirable soil particles. Chemosphere, 25: 29-32. 548743

3 8	Nilsson CB; Hakansson H (2002). The retinoid signaling system- a target in dioxin toxicity. Crit Rev Toxicol, 32:

39	211-232.548746

This document is a draft for review purposes only and does not constitute Agency policy.

R-20	DRAFT—DO NOT CITE OR QUOTE

-------
1	Nishimura N; Yonemoto J; Nishimura H; Ikushiro S; Tohyama C (2005). Disruption of thyroid hormone

2	homeostasis at weaning of Holtzman rats by lactational but not in utero exposure to 2,3,7,8-tetrachlorodibenzo-p-

3	dioxin. Toxicol Sci, 85: 607-614. 197860

4	Niskar A; Needham LL; Rubin C; Turner WE; Martin CA; Patterson DG Jr; Hasty L; Wong LY; Marcus M (2009).

5	Serum dioxin, poly chlorinated biphenyls, and endometriosis: A case-control study in Atlanta. Chemosphere, 74:

6	944-949. 548802

7	Nohara K; Fujimaki H; Tsukumo S; Ushio H; Miyabara Y; Kijima M; Tohyama C; Yonemoto J (2000). The effects

8	of perinatal exposure to low doses of 2,3,7,8-tetrachlorodibenzo-p-dioxin on immune organs in rats. Toxicology,

9	154: 123-133. 200027

10	Nohara K; Izumi H; Tamura S; Nagata R; Tohyama C (2002). Effect of low-dose 2,3,7,8-tetrachlorodibenzo-p-

11	dioxin (TCDD) on influenza A virus-induced mortality in mice. Toxicology, 170: 131-138. 199021

12	Nolan KJ; Smith FA; Hefner JG (1979). Elimination and tissue distribution of 2,3,7,8-tetrachlorodibenzo-p-dioxin

13	(TCDD) in female guinea pigs following a single oral dose. Toxicol Appl Pharmacol, 48: 162. 543785

14	NRC (1983). Risk assessment in the federal government: Managing the process. National Academy Press.

15	Washington, DC. 194806

16	NRC (1989). Improving risk communication. Washington, DC: National Academy Press. 000858

17	NRC (1991). Human exposure assessment for airborne pollutants: advances and opportunities. Washington, DC:

18	National Academies Press. 037823

19	NRC (1993). Issues in risk assessment. Committee on Risk Assessment Methodology, National Research Council.

20	Washington, DC.http://www.nap.edu/catalog.php?record_id=2078. 078637

21	NRC (1994). Science and judgment in risk assessment. National Research Council; National Academy Press.

22	Washington, DC. 006424

23	NRC (2002). Estimating the public health benefits of proposed air pollution regulations. Washington, DC: National

24	Academy of Sciences. 035312

25	NRC (2007). Scientific review of the proposed risk assessment bulletin from the Office of Management and Budget.

26	National Research Council. Washington, DC.http://www.nap.edu/catalog.php?record_id=l 1811. 543748

27	NRC (National Research Council) (2009). Science and decisions: advancing risk assessment. National Academy

28	Press. Washington, DC. 194810

29	NTP (1982). Carcinogenesis bioassay of BIS(2-chloro-l-methylethyl) ether (70%) (CAS no. 108-60-1) containing

30	2-chloro-l-methylethyl(2-chloropropyl) ether ( 30%) (CAS no. 83270-31-9) inB6C3Fl mice (gavage study).

31	National Toxicology Program. Research Triangle Park, NC and Bethesda, MD. NTP-81-55. 200870

32	NTP (1982). NTP Technical Report on carcinogenesis bioassay of 2,3,7,8-tetrachlorodibenzo-p-dioxin in Osborne-

33	Mendel rats and B6C3F1 mice (gavage study). Public Health Service, U.S. Department of Health and Human

34	Services, National Toxicology Program. Research Triangle Park, NC. 543764

35	NTP (1982). NTP Technical Report on carcinogenesis bioassay of 2,3,7,8-tetrachlorodibenzo-p-dioxin in Osborne-

36	Mendel rats and B6C3F1 mice (gavage study). Public Health Service, U.S. Department of Health and Human

37	Services; NTP TR 209. NIEHS. Research Triangle Park, NC. 594255

3 8	NTP (2006). NTP technical report on the toxicology and carcinogenesis studies of 2,3,7,8-tetrachlorodibenzo-p-

39	dioxin (TCDD) in female harlan Sprague-Dawley rats. National Toxicology Program. RTP, NC. 06-4468. 197605

This document is a draft for review purposes only and does not constitute Agency policy.

R-21	DRAFT—DO NOT CITE OR QUOTE

-------
1	NTP (2006). Toxicology and carcinogenesis studies of a mixture of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)

2	(CAS No. 1746-01-6), 2,3,4,7,8-pentachlorodibenzofuran (PeCDF) (CAS No. 57117-31-4), and 3,3',4,4',5-

3	pentachlorobiphenyl (PCB 126) (CAS No. 57465-28-8) in female Harlan Sprague-Dawley rats (gavage studies).

4	Public Health Service, U.S. Department of Health and Human Services, tional Toxicology Program. Research

5	Triangle Park, NC.http://ntp.niehs.nih.gov/index.cfm?objectid=070B7300-0E62-BF12-F4C3E3B5B645A92B.

6	543749

7	Oehlert GW (1992). A note on the delta method. Am Stat, 46: 27-29. 543742

8	Ohsako S; Miyabara Y; Nishimura N; Kurosawa S; Sakaue M; Ishimura R; Sato M; Takeda K; Aoki Y; Sone H;

9	Tohyama C; Yonemoto J (2001). Maternal exposure to a low dose of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)

10	suppressed the development of reproductive organs of male rats: Dose-dependent increase of mRNA levels of 5a-

11	reductase type 2 in contrast to descrease of androgen receptor in the pubertal ventral prostate. Toxicol Sci, 60: 132-

12	143.198497

13	Okey AB; Riddick DS; Harper PA (1994). The Ah receptor: Mediator of the toxicity of 2,3,7,8-tetrachlorodibenzo-

14	p-dioxin (TCDD) and related compounds. Toxicol Lett, 70: 1-22. 548759

15	Olson JR; Holscher MA; Neal RA (1980). Toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin in the Golden Syrian

16	hamster . Toxicol Appl Pharmacol, 55: 67-78. 197976

17	Olson JR; McGarrigle BP; Gigliotti PJ; Kumar S; McReynolds JH (1994). Hepatic uptake and metabolism of

18	2,3,7,8-tetrachlorodibenzo-p-dioxin and 2,3,7,8-tetrachlorodibenzofuran. Fundam Appl Toxicol, 22: 631-640.

19	198008

20	Ott MG; Messerer P; Zober A (1993). Assessment of past occupational exposure to 2,3,7,8-tetrachlorodibenzo-p-

21	dioxin using blood lipid analyses. Int ArchOccup Environ Health, 65: 1-8. 594322

22	Ott MG; Olson RA; Cook RR; Bond GG (1987). Cohort mortality study of chemical workers with potential

23	exposure to the higher chlorinated dioxins. J Occup Environ Med, 29: 422-429. 064994

24	Ott MG; Zober A (1996). Cause specific mortality and cancer incidence among employees exposed to 2,3,7,8-

25	TCDD after a 1953 reactor accident. Occup Environ Med, 53: 606-612. 198408

26	Ott MG; Zober A (1996). Morbidity study of extruder personnel with potential exposure to brominated dioxins and

27	furans. II. Results of clinical laboratory studies. Occup Environ Med, 53: 844-846. 198101

28	Papke O; Ball M; Lis A (1994). PCDD/PCDF in humans, a 1993-update of background data. Chemosphere, 29:

29	2355-2360. 198279

30	Pekelis M; Nicolich MJ; Gauthier JS (2003). Probabilistic framework for the estimation of the adult and child

31	toxicokinetic intraspecies uncertainty factors. Risk Anal, 23: 1239-1255. 548723

32	Percy C; Stanek E III; Gloeckler L (1981). Accuracy of cancer death certificates and its effect on cancer mortality

33	statistics. Am J Public Health, 71: 242-250. 004891

34	Pereg D; Dewailly E; Poirier GG; Ayotte P (2002). Environmental exposure to polychlorinated biphenyls and

35	placental CYP1A1 activity in Inuit women from northern Quebec. Environ Health Perspect, 110: 607-612. 199797

36	Pesatori AC; Consonni D; Bachetti S; Zocchetti C; Bonzini M; Baccarelli A; Bertazzi PA (2003). Short- and long-

37	term morbidity and mortality in the population exposed to dioxin after the "Seveso accident". Ind Health, 41: 127-

38	138. 197001

39	Pesatori AC; Zocchetti C; Guercilena S; Consonni D; Turrini D; Bertazzi PA (1998). Dioxin exposure and non-

40	malignant health effects: A mortality study. Occup Environ Med, 55: 126-131. 523076

This document is a draft for review purposes only and does not constitute Agency policy.

R-22 DRAFT—DO NOT CITE OR QUOTE

-------
1	Piacitelli LA; Sweeney MH; Fingerhut MA; Patterson DG; Turner WE; Connally LB; Wille KK; Tompkins B

2	(1992). Serum levels of PCDDS and PCDFS among workers exposed to 2,3,7,8-TCDD contaminated chemicals.

3	Chemosphere, 25: 251-254. 197275

4	Pipe NG; Smith T; Halliday D; Edmonds CJ; Williams C; Coltart TM (1979). Changes in fat, fat-free mass and body

5	water in human normal pregnancy. Br J Obstet Gynaecol, 86: 929-940. 548786

6	Pirkle JL; Wolfe WH; Patterson DG; Needham LL; Michalek JE; Miner JC; Peterson MR; Phillips DL (1989).

7	Estimates of the half-life of 2,3,7,8-tetrachlorodibenzo-p-dioxin in Vietnam Veterans of Operation Ranch Hand. J

8	Toxicol Environ Health, 27: 165-171. 197861

9	Pitot H; Goldsworthy T; Campbell H; Poland A (1980). Quantitative evaluation of the promotion by 2,3,7,8-

10	tetrachlorodibenzo-p-dioxin of hepatocarcinogenesis from diethylnitrosamine. Cancer Res, 40: 3616-3620. 197885

11	Pohjanvirta R; Tuomisto J (1994). Short-term toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin in laboratory animals:

12	Effects, mechanisms, and animal models. Pharmacol Rev, 46: 483-549. 543767

13	Pohjanvirta R; Tuomisto L; Tuomisto J (1989). The central nervous system may be involved in TCDD toxicity.

14	Toxicology, 58: 167-174. 548766

15	PoigerH; Schlatter C (1986). Pharmacokinetics of 2,3,7,8-TCDD in man. Chemosphere, 15: 1489-1494. 197336

16	Poland A; Glover E (1980). 2,3,7,8-tetrachlorodibenzo-p-dioxin: segregation of toxicity with the Ah locus. Mol

17	Pharmacol, 17: 86-94. 543761

18	Poland A; Glover E (1990). Characterization and strain distribution pattern of the murine Ah receptor specified by

19	the Ahd and Ahb-3 alleles. Mol Pharmacol, 38: 306-312. 543759

20	Poland A; Palen D; Glover E (1982). Tumour promotion by TCDD in skin of HRS/J hairless mice. Nature, 300:

21	271-273. 199756

22	Poland A; Palen D; Glover E (1994). Analysis of the four alleles of the murine aryl hydrocarbon receptor. Mol

23	Pharmacol, 46: 915-921. 198439

24	Popp JA; Crouch E; McConnell EE (2006). A Weight-of-evidence analysis of the cancer dose-response

25	characteristics of 2,3,7,8-tetrachlorodibenzodioxin (TCDD). Toxicol Sci, 89: 361-369. 197074

26	Potter CL; Moore RW; Inhorn SL; Hagen TC; Peterson RE (1986). Thyroid status and thermogenesis in rats treated

27	with 2,3,7,8-tetrachlorodibenzo-p-dioxin. Toxicol Appl Pharmacol, 84: 45-55. 548771

28	Potter CL; Sipes IG; Russell DH (1983). Hypothyroxinemia and hypothermia in rats in response to 2,3,7,8-

29	tetrachlorodibenzo-p-dioxin administration. Toxicol Appl Pharmacol, 69: 89-95. 548769

3 0	Poulin P; Theil FP (2001). Prediction of pharmicokinetics prior to in vivo studies. 1. mechanism-based prediction of

31	volume of distribution. J Pharm Sci, 91: 129-156. 594269

32	Puga A; Nebert DW; Carier F (1992). Dioxin induces expression of c-fos and c-jun proto-oncogenes and a large

33	increases in transcription factor AP-1. Toxicol Appl Pharmacol, 55: 67-78. 543784

34	Ramadoss P; Perdew GH (2004). Use of 2-azido-3-[125I]iodo-7,8-dibromodibenzo-p-dioxin as a probe to determine
3 5	the relative ligand affinity of human versus mouse aryl hydrocarbon receptor in cultured cells. Mol Pharmacol, 66:
36	129-136. 198824

This document is a draft for review purposes only and does not constitute Agency policy.

R-23 DRAFT—DO NOT CITE OR QUOTE

-------
1	Ramsey JC; Hefner JG; Karbowski RJ; Braun WH; GeMng PJ (1982). The in vivo biotransformation of 2,3,7,8-

2	tetrachlorodibenzo-p-dioxin (TCDD) in the rat. Toxicol Appl Pharmacol, 65: 180-184. 548750

3	Rao MS; Subbarao V; Prasad JD; Scarpelli DG (1988). Carcinogenicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin in

4	the Syrian golden hamster. Carcinogenesis, 6: 1677-1679. 199032

5	Reddy M; Yang R; Clewell HJ; Andersen ME (2005). Physiologically based pharmacokinetic modeling: Science

6	and applications. Hoboken, New Jersey: John Wiley & Sons. 594251

7	Revich B; Aksel E; Ushakova T; Ivanova I; Zhuchenko N; Klyuev N; Brodsky B; Sotskov Y (2001). Dioxin

8	exposure and public health in Chapaevsk, Russia. Chemosphere, 43: 951-966. 199843

9	Revich B; Sergeyev O; Zeilert V; Hauser R (2005). Chapaevsk, Russia: 40 years of dioxins exposure on the human

10	health and 10 years of Russian ?USA epidemiological studies. Presented at Almaty 2005, Almaty, Kazakhstan.

11	198777

12	Rier SE; Coe CL; Lemieux AM; Martin DC; Morris R; Lucier GW; Clark GC (2001). Increased tumor necrosis

13	factor-alpha production by peripheral blood leukocytes from TCDD-exposed rhesus monkeys. Toxicol Sci, 60: 327-

14	337.543773

15	Rier SE; Martin DC; Bowman RE; Becker JL (1995). Immunoresponsiveness in endometriosis: Implications of

16	estrogenic toxicants. Environ Health Perspect, 103: 151-156. 198566

17	Rier SE; Martin DC; Bowman RE; Dmowski WP; Becker JL (1993). Endometriosis in Rhesus Monkeys (Macaca

18	mulatta) Following Chronic Exposure to 2,3,7,8-Tetrachlorodibenzo-p-dioxin. Fundam Appl Toxicol, 21: 433-441.

19	199987

20	Rier SE; Turner WE; Martin DC; Morris R; Lucier GW; Clark GC (2001). Serum levels of TCDD and dioxin-like

21	chemicals in Rhesus monkeys chronically exposed to dioxin: Correlation of increased serum PCB levels with

22	endometriosis. Toxicol Sci, 59: 147-159. 198776

23	Roberts EA; Golas CL; Okey AB (1986). Ah receptor mediating induction of aryl hydrocarbon hydroxylase:

24	Detection in human lung by binding of 2,3,7,8-[H]tetrachlorodibenzo-p-dioxin. Cancer Res, 46: 3739-3743. 198780

25	Roberts EA; Shear NH; Okey AB; Manchester DK (1985). The Ah receptor and dioxin toxicity: From rodent to

26	human tissues . Chemosphere, 14: 661-674. 198706

27	Rohde S; Moser GA; Papke O; McLachlan MS (1999). Clearance of PCDD/Fs via the gastrointestinal tract in

28	occupationally exposed persons. Chemosphere, 38: 3397-3410. 548764

29	Roth WL; Ernst S; Weber LWD; Kerescen L; Rozman KK (1994). A pharmacodynamically responsive model of

30	2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) transfer between liver and fat at low and high doses. Toxicol Appl

31	Pharmacol, 127: 151-162. 198063

32	RothmanKJ (1986). Modern epidemiology. 046091

33	Roy T; Hammerstrom K; Schaum J (2008). Percutaneous absorption of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)

34	from soil. J Toxicol Environ Health A Currlss, 71: 1509-1515.548747

3 5	Rozman KK (2000). The role of time in toxicology or Haber's c x t product. Toxicology, 149: 35-42. 548758

36	Ryan JJ; Amirova Z; Carrier G (2002). Sex Ratios of Children of Russian Pesticide Producers Exposed to Dioxin.

37	Environ Health Perspect, 110: A699-A701. 198508

This document is a draft for review purposes only and does not constitute Agency policy.

R-24 DRAFT—DO NOT CITE OR QUOTE

-------
1	Ryan JJ; Schecter A (2000). Exposure of Russian phenoxy herbicide producers to dioxin. J Occup Environ Med, 42:

2	861-870. 594412

3	Saltelli A; Chan K; Scott EM (2000). Sensitivity analysis. England: John Wiley & Sons Ltd. 543756

4	Sandau CD; Ayotte P; Dewailly E; Duffe J; Norstrom RJ (2002). Pentachlorophenol and hydroxylated

5	poly chlorinated biphenyl metabolites in umbilical cord plasma of neonates from coastal populations in Quebec.

6	Environ Health Perspect, 110:411-417.594406

7	Santostefano MJ; Johnson KL; Whisnant NA; Richardson VM; Devito MJ; Birnbaum LS (1996). Subcellular

8	localization of TCDD differs between the liver, lungs, and kidneys after acute and subchronic exposure:

9	Species/dose comparison and possible mechanism. Fundam Appl Toxicol, 34: 365-375. 594258

10	Santostefano MJ; Wang X; Richardson VM; Ross DG; DeVito MJ; Birnbaum LF (1998). A pharmacodynamic

11	analysis of TCDD-Induced Cytochrome 450 gene expression in multiple tissues: Dose and time-dependent effects.

12	Toxicol Appl Pharmacol, 151: 294-310. 200001

13	Saracci R; Kogevinas M; Bertazzi PA; Bueno de Mesquita BH; Coggon D; Green LM; Kauppinen T; L'Abbe KA;

14	Littorin M; Lynge E; Mathews JD; Neuberger M; Osman J; Pearce N; Winkelmann R (1991). Cancer mortality in

15	workers exposed to chlorophenoxy herbicides and chlorophenols. Lancet, 338():: 1027-1032. 199190

16	Sauer RM (1990). 2,3,7,8-Tetrachlorodibenzo-p-dioxin in sprague-dawley rats. PATHCO, INC. Maryland. 198829

17	Schantz SL; Bowman RE (1989). Learning in monkeys exposed perinatally to 2,3,7,8-tetrachlorodibenzo-p-dioxin

18	(TCDD). Neurotoxics Teratol, 11: 13-19. 198104

19	Schantz SL; Laughlin NK; Van Valkenberg HC; Bowman RE (1986). Maternal care by rhesus monkeys of infant

20	monkeys exposed to either lead or 2,3,7,8-tetrachlorodibenzo-P-dioxin. Neurotoxicology, 7: 637-650. 088206

21	Schantz SL; Seo BW; Moshtaghian J; Peterson RE; Moore RW (1996). Effects of gestational and lactational

22	exposure to TCDD or coplanar PCBs on spatial learning. Neurotoxicol Teratol, 18: 305-313. 198781

23	Schecter A; Cramer P; Boggess K; Stanley J; Olson JR (1997). Levels of Dioxins, Dibenzofurans, PCB and DDE

24	congeners in pooled food samples collected in 1995 at supermarkets across the United States. Chemosphere, 34:

25	1437-1447. 198396

26	Schwartz M; Appel KE (2005). Carcinogenic risks of dioxin: mechanistic considerations. Regul Toxicol Pharmacol,

27	43: 19-34. 543737

28	Seidel SD; Winters GM; Rogers WJ; Ziccardi MH; Li V; Keser B; Denison MS (2001). Activation of the Ah

29	receptor signaling pathway by prostaglandins. J Biochem Mol Toxicol, 15: 187-196. 543776

30	Self SG; Liang KY (1987). Asymptotic properties of maximum likelihood estimators and likelihood ratio tests under

31	nonstandard conditions. J Am Stat Assoc, 82: 605-610. 594398

32	Seo BW; Li MH; Hansen LG; Moore RW; Peterson RE; Schantz SL (1995). Effects of gestational and lactational

33	exposure to coplanar polychlorinated biphenyl (PCB) congeners or 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) on

34	thyroid hormone concentrations in weanling rats. Toxicol Lett, 78: 253-262. 197869

3 5	Sewall C; Lucier G; Tritscher A; Clark G (1993). TCDD-mediated changes in hepatic epidermal growth factor

36	receptor may be a critical event in the hepatocarcinogenic action of TCDD. Carcinogenesis, 14: 1885-1893. 197889

37	Sewall CH; Flagler N; Vanden Heuvel JP; Clark GC; Tritscher AM; Maronpot RM; Lucier GW (1995). Alterations

38	in thyroid function in female Sprague-Dawley rats following chronic treatment with 2,3,7,8-tetrachlorodibenzo-p-

39	dioxin. Toxicol Appl Pharmacol, 132: 237-244. 198145

This document is a draft for review purposes only and does not constitute Agency policy.

R-25 DRAFT—DO NOT CITE OR QUOTE

-------
1	Shi Z; Valdez KE; Ting AY; Franczak A; Gum SL; Petroff BK (2007). Ovarian endocrine disruption underlies

2	premature reproductive senescence following environmentally relevant chronic exposure to the aryl hydrocarbon

3	receptor agonist 2,3,7,8-tetrachlorodibenzo-p-dioxin. Biol Reprod, 76: 198-202. 198147

4	Shu H; Teitelbaum P; Webb AS; Marple L; Brunck B; Dei Rossi D; Murray FJ; Paustenbach D (1988).

5	Bioavailability of soil-bound TCDD: Dermal bioavailability in the rat. Fundam Appl Toxicol, 2: 335-343. 548739

6	Siemiatycki J; Wacholder S; Dewar R; Cardis E; Greenwood C; Richardson L (1988). Degree of confounding bias

7	related to smoking, ethnic group, and socioeconomic status in estimates of the associations between occupation and

8	cancer. J Occup Med, 30: 617-625. 198556

9	Sikov M (1970). Radiation biology of the fetal and juvenile mammal. Science, 167: 1640-1641. 594274

10	SimanainenU; Haavisto T; Tuomisto JT; Paranko J; Toppari J; Tuomisto J; PetersonRE; VilukselaM (2004).

11	Pattern of male reproductive system effects after in utero and lactational 2,3,7,8-tetrachlorodibenzo-p-dioxin

12	(TCDD) exposure in three differentially TCDD-sensitive rat lines Pattern of male reproductive system effects after

13	in utero and lactational 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) exposure in three differentially TCDD-sensitive

14	rat lines. Toxicol Sci, 80: 101-108. 198948

15	Simanainen U; Tuomisto JT; Pohjanvirta R; Syijala P; Tuomisto J; Viluksela M (2004). Postnatal development of

16	resistance to short-term high-dose toxic effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin in TCDD-resistant and -

17	semiresistant rats. Toxicol Appl Pharmacol, 196: 11-19. 198106

18	Simanainen U; Tuomisto JT; Tuomisto J; Viluksela M (2002). Structure-Activity relationships and dose responses

19	ofPolychlorinated Dibenzo-p-dioxins for short-term effects in 2,3,7,8-Tetrachlorodibenzo-p-dioxin-Resistant and

20	sensitive rat strains. Toxicol Appl Pharmacol, 181: 38-47. 201369

21	Simanainen U; Tuomisto JT; Tuomisto J; Viluksela M (2003). Dose-response analysis of short-term effects of

22	2,3,7,8-tetrachlorodibenzo-p-dioxin in three differentially susceptible ratlines., 187: 128-136. 198582

23	Simon T; Aylward LL; Kirman CR; Rowlands JC; Budinsky RA (2009). Estimates of cancer potency of 2,3,7,8-

24	tetrachlorodibenzo(p)dioxin using linear and non-linear dose-response modeling and toxicokinetics. Toxicol Sci,

25	112:490-506.594321

26	Slezak BP; Hatch GE; DeVito MJ; Diliberto JJ; Slade R; Crissman K; Hassoun E; Birnbaum LS (2000). Oxidative

27	stress in female B6C3F1 mice following acute and subchronic exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin

28	(TCDD). Toxicol Sci, 54: 390-398. 199022

29	Slob W; Pieters MN (1998). A probabilistic approach for deriving acceptable human intake limits and human health

30	risks from toxicological studies: general framework. Risk Anal, 18: 787-798. 087256

31	Smart J; Daly A (2000). Variation in induced CYP1A1 levels: Relationship to CYP1A1, Ah receptor, and GSTM1

32	polymorphisms. Pharmacogenetics, 10: 11-24. 548794

33	Smialowicz RJ; Burgin DE; Williams WC; Diliberto JJ; Setzer RW; Birnbaum LS (2004). CYP1A2 is not required

34	for 2,3,7,8-tetrachlorodibenzo-p-dioxin-induced immunosuppression. Toxicology, 197: 15-22. 110937

3 5	Smialowicz RJ; DeVito MJ; Williams WC; Birnbaum LS (2008). Relative potency based on hepatic enzyme

36	induction predicts immunosuppressive effects of a mixture of PCDDS/PCDFS and PCBS. Toxicol Appl Pharmacol,

37	227:477-484. 198341

3 8	Smith AH; Fisher DO; Pearce N; Chapman CJ (1982). Congenital defects and miscarriages among New Zealand 2,

39	4, 5-T sprayers. Arch Environ Health, 37: 197-200. 198586

This document is a draft for review purposes only and does not constitute Agency policy.

R-26 DRAFT—DO NOT CITE OR QUOTE

-------
1	Smith AH; Lopipero P (2001). Invited commentary: how do the Seveso findings affect conclusions concerning

2	TCDD as a human carcinogen? Am J Epidemiol, 153: 1045-1047. 198585

3	Spiegelhalter D; Thomas A; Best N; Gilks W (2003). BUGS 0.5 Bayesian inference using Gibbs sampling manual,

4	version ii. MRC Biostatistics Units, Institute of Public Health, Cambridge. 594261

5	Squire RA (1980). Pathologic evaluations of selected tissues from the Dow Chemical TCDD and 2,4,5-T rat studies.

6	U.S. Environmental Protection Agency. Washington DC. 594272

7	Squire RA (1990). Pathologic evaluations of selected tissues from the Dow Chemical TCDD and 2,4,5-T rat studies.

8	Submitted to Carcinogen Assessment Group, U.S. Environmental Protection Agency. Washington, DC. 548781

9	Starr TB (2003). Significant issues raised by meta-analyses of cancer mortality and dioxin exposure. Environ Health

10	Perspect, 111: 1443-1447. 594271

11	Staskal DF; Diliberto JJ; DeVito MJ; Birnbaum LS (2005). Inhibition of human and rat CYP1A2 by TCDD and

12	dioxin-like chemicals. Toxicol Sci, 84: 225-231. 198276

13	Stayner L; Bailer AJ; Smith R; Gilbert S; Rice F; Kuempel E (1999). Sources of uncertainty in dose-response

14	modeling of epidemiological data for cancer risk assessment. Ann N Y Acad Sci, 895: 212-222. 198654

15	Stayner L; Steenland K; Dosemeci M; Hertz-Picciotto I (2003). Attenuation of exposure-response curves in

16	occupational cohort studies at high exposure levels. Scand J Work Environ Health, 29: 317-324. 054922

17	Steenland K; Calvert G; Ketchum N; Michalek J (2001). Dioxin and diabetes mellitus: an analysis of the combined

18	NIOSH and Ranch Hand data. Occup Environ Med, 58: 641-648. 198589

19	Steenland K; Deddens J (2003). Dioxin: Exposure-response analyses and risk assessment. Ind Health, 41: 175-180.

20	198587

21	Steenland K; Deddens J; Piacitelli L (2001). Risk assessment for 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) based

22	on an epidemiologic study. Am J Epidemiol, 154: 451-458. 197433

23	Steenland K; Piacitelli L; Deddens J; Fingerhut M; Chang LI (1999). Cancer, heart disease, and diabetes in workers

24	exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin. J Natl Cancer Inst, 91: 779-786. 197437

25	Stellman SD; Stellman JM (1986). Estimation of exposure to Agent Orange and other defoliants among American

26	troops in Vietnam: a methodological approach. Am J Ind Med, 9: 305-321. 594380

27	StephensonRP (1956). A modification of receptor theory. Br J Pharmacol, 11: 379-393. 594280

28	Sugita-Konishi Y; Kobayashi K; Naito H; Miura K; Suzuki Y (2003). Effect of lactational exposure to 2,3,7,8-

29	tetrachlorodibenzo-p-dioxin on the susceptibility to Listeria infection. Biosci Biotechnol Biochem, 67: 89-93.

30	198375

31	Swartout JC; Price PS; Dourson ML; Carlson-Lynch HL; Keenan RE (1998). A probabilistic framework for the

32	reference dose (probabilistic RfD). Risk Anal, 18: 271-282. 093460

33	t' Mannetje A; McLean D; Cheng S; Boffetta P; Colin D; Pearce N (2005). Mortality in New Zealand workers

34	exposed to phenoxy herbicides and dioxins. Occup Environ Med, 62: 34-40. 197593

35	Takemoto K; NakajimaM; Fujiki Y; KatohM; Gonzalez FJ; Yokoi T (2004). Role of the aryl hydrocarbon receptor

36	and Cyplbl in the antiestrogenic activity of 2,3,7,8-tetrachlorodibenzo-p-dioxin. Arch Toxicol, 78: 309-315.

37	543753

This document is a draft for review purposes only and does not constitute Agency policy.

R-27 DRAFT—DO NOT CITE OR QUOTE

-------
1	Teeguarden JG;, Dragan YP; Singh J; Vaughan J; Xu YH; Goldsworthy T; HC Pitot HC (1999). Quantitative

2	analysis of dose- and time-dependent promotion of four phenotypes of altered hepatic foci by 2,3,7,8-

3	tetrachlorodibenzo-p- dioxin in female Sprague-Dawley rats. Toxicol Sci, 51: 211-223. 198274

4	Thiess AM; Frentzel-Beyme R (1977). Mortality study of persons exposed to dioxin following an accident which

5	occurred in the BASF on 17 November 1953. Presented at Proceedings of the 5th International Conference

6	Medichem, 1977, San Francisco, CA. 594302

7	Thiess AM; Frentzel-Beyme R; Link R (1982). Mortality study of persons exposed to dioxin in a trichlorophenol-

8	process accident that occurred in the BASF AGon November 17, 1953. Am JIndMed, 3: 179-189. 064999

9	Tian Y; Ke S; Denison MS; Rabson AB; Gallo MA (1999). Ah Receptor and NF-kB Interactions, a Potential

10	Mechanism for Dioxin Toxicity. J Biol Chem, 274: 510-515. 198378

11	Toide K; Yamazaki JH; Nagashima R; Itoh K; Iwano S; Takahashi Y; Watanabe S; Kamataki T (2003). Aryl

12	hydrocarbon hydroxylase represents CYP1B1 and not CYP1A1, in human freshly isolated white cells: Trimodal

13	distribution of Japanese population according to induction of CYP1B1 mRNA by environmental dioxins. Cancer

14	Epidemiol Biomarkers Prev, 12: 219-222. 548792

15	Toth K; Somfai-Relle S; Sugar J; Bence J (1979). Carcinogenicity testing of herbicide 2,4,5-

16	trichlorophenoxyethanol containing dioxin and of pure dioxin in Swiss mice. Nature, 278: 548-549. 197109

17	Tritscher AM; Mahler J; Portier CJ; Lucier GW; Walker NJ (2000). Induction of lung lesions in female rats

18	following chronic exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Toxicol Pathol, 28: 761-769. 197265

19	Tuomisto JT; Viluksela M; Pohjanvirta R; Tuomisto J (1999). The AH receptor and a novel gene determine acute

20	toxic responses to TCDD: segregation of the resistant alleles to different rat lines. Toxicol Appl Pharmacol, 155: 71-

21	81.548717

22	Tuomisto JT; Wilson AM; Evans JS; Tainio M (2008). Uncertainty in mortality response to airborne fine particulate

23	matter: combining European air pollution experts. Reliab Eng Syst Saf, 93: 732-744. 548715

24	U.S. DOE (1992). DOE standard: Hazard categorization, and accident analysis techniques for compliance with DOE

25	Order 5480.23, nuclear safety analysis reports. U.S. Department of Energy. Washington, DC. DOE-STD-1027-92.

26	http://www.hss.energy.gov/nuclearsafety/ns/techstds/standard/stdl027/sl027cnl.pdf. 543733

27	U.S. EPA (1994). Methods for derivation of inhalation reference concentrations and application of inhalation

28	dosimetry. Environmental Criteria and Assessment Office, Office of Health and Environmental Assessment, Office

29	of Research and Development, U.S. Environmental Protection Agency. Research Triangle Park, NC. EPA/600/8-

30	90/066F. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=71993. 006488

31	U.S. EPA (1996). Columbus waste-to-energy municipal incinerator Dioxin soil sampling project. U.S. EPA

32	REGION 5. Chicago, IL. 905R96018.

33	http://nepis.epa.gov/Exe/ZyNET.exe/2000PCXX.TXT?ZyActionD=ZyDocument&Client=EPA&Index=1995+Thru

34	+1999&Docs=&Query=columbus+waste-to-

35	energy+municipal+incinerator&Time=&EndTime=&SearchMethod=3&TocRestrict=n&Toc=&TocEntry=&QField

36	=pubnumber%5E%22905R96018%22&QFieldYear=&QFieldMonth=&QFieldDay=&UseQField=pubnumber&Int

37	QFieldOp=l&ExtQFieldOp=l&XmlQuery=&File=D%3A%5Czyfiles%5CIndex%20Data%5C95thru99%5CTxt%5

38	C00000017%5C2000PCXX.txt&User=ANONYMOUS&Password=anonymous&SortMethod=h%7C-

39	&MaximumDocuments=10&FuzzyDegree=0&ImageQuality=r75g8/r75g8/xl50yl50gl6/i425&Display=p%7Cf&D

40	efSeekPage=x&SearchBack=ZyActionL&Back=ZyActionS&BackDesc=Results%20page&MaximumPages=l&Zy

41	Entry=l&SeekPage=x. 198087

42	U.S. EPA (1996). Proposed guidelines for carcinogen risk assessment. Risk Assessment Forum. U.S. Environmental

43	Protection Agency. Washington, D.C.. 594399

This document is a draft for review purposes only and does not constitute Agency policy.

R-28 DRAFT—DO NOT CITE OR QUOTE

-------
1	U.S. EPA (1998). Guidelines for neurotoxicity risk assessment. Federal Register 63(93):26926-26954. National

2	Center for Environmental Assessment; Office of Research and Development; U.S. Environmental Protection

3	Agency. Washington, DC. EPA/630/R-95/001Fa.

4	http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=4555 .030021

5	U.S. EPA (2000). Benchmark dose technical guidance document [external review draft]. Risk Assessment Forum,

6	U.S. Environmental Protection Agency. Washington, DC. EPA/630/R-00/001.

7	http://www.epa.gov/raf/publications/benchmark-dose-doc-draft.htm. 052150

8	U.S. EPA (2003). Exposure and human health reassessment of 2,3,7,8 tetrachlorodibenzo-p dioxin (TCDD) and

9	related compounds [NAS review draft]. U.S. Environmental Protection Agency, National Center for Environmental

10	Assessment. Washington, DC. EPA/600/P 00/001. http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/. 537122

11	U.S. EPA (2005). Guidelines for carcinogen risk assessment, Final Report. Risk Assessment Forum, U.S.

12	Environmental Protection Agency. Washington, DC. EPA/630/P-03/001F.

13	http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=l 16283. 086237

14	U.S. EPA (2006). Air quality criteria for lead, in 2 Volumes. Office of Health and Environmental Assessment,

15	Environmental Criteria and Assessment Office, Office of Research and Development, U.S. Environmental

16	Protection Agency. Research Triangle Park, NC. EPA-600/R-5/144aF-bF. 090110

17	U.S. EPA (2006). Air quality criteria for ozone and related photochemical oxidants. EPA. DC. 088089

18	U.S. EPA (2006). Provisional Assessment of Recent Studies on Health Effects of Particulate Matter Exposure. U.S.

19	Environmental Protection Agency. Research Triangle Park, NC. 157071

20	U.S. EPA (2008). 2,3,7,8 Tetrachlorodibenzo-p dioxin (TCDD) dose response studies: preliminary literature search

21	results and request for additional studies. U.S. Environmental Protection Agency. Washington, DC. EPA/600/R-

22	08/119.519261

23	U.S. EPA (2008). Framework for application of the toxicity equivalence methodology for polychlorinated dioxins,

24	furans, and biphenyls in ecological risk assessment. U.S. Environmental Protection Agency. Washington, DC.

25	EPA/100/R 08/004. http://www.epa.gov/raf/tefframework/index.htm. 543774

26	U.S. EPA (2009). Integrated risk information system (IRIS). Retrieved 24-JUN-09, from

27	http://cfpub.epa.gov/ncea/iris/index.cfm. 192196

28	U.S. EPA (2009). Summary of U.S. EPA dioxin workshop: February 18-20, 2009. U.S. Environmental Protection

29	Agency. National Center for Environmental Assessment. Cincinnati, OH. EPA/600/R-09/027. 543757

30	U.S. EPA (2009). Using probabilistic methods to enhance the role of risk analysis in decision-making with case

31	study examples. U.S. Environmental Protection Agency. Washington, DC. Washington, DC. EPA/100/R-09/001.

32	522927

33	U.S. NRC (1975). Reactor safety study-an assessment of accident risks in U.S. commercial nuclear power plants.

34	U.S. Nuclear Regulatory Commission. Rockville, MD. NUREG-75/014 (WASH-1400).
3 5	http://www.nrc.gov/reading-rm/doc-collections/nuregs/staff/sr75-014/. 543729

36	U.S. NRC (1981). Fault tree handbook. U.S. Nuclear Regulatory Commission. Washington, DC. NUREG-0492.

37	http://www.nrc.gov/reading-rm/doc-collections/nuregs/staff/sr0492/. 543730

38	U.S. NRC (1983). A guide to the performance of probabilistic risk assessments for nuclear power plants. U.S.

39	Nuclear Regulatory Commission. Washington, DC. NUREG/CR-2300. http://www.nrc.gov/reading-rm/doc-

40	collections/nuregs/contract/cr2300/. 543732

This document is a draft for review purposes only and does not constitute Agency policy.

R-29 DRAFT—DO NOT CITE OR QUOTE

-------
1	U.S. NRC (1991). Severe accident risks: an assessment for five U.S. nuclear power plants. U.S. Nuclear Regulatory

2	Commission. Washington, DC. NUREG-1150. http://www.nrc.gov/reading-rm/doc-collections/nuregs/staff/srll50/.

3	543736

4	Umemura T; Kai S; Hasgawa R; Sai K; Kurokawa Y; Williams GM (1999). Pentachlorophenol (PCP) produces liver

5	oxidative stress and promotes but does not initiate hepatocarcinogenesis inB6C3Fl mice. Carcinogenesis, 20: 1115-

6	1120.198001

7	Van Birgelen AP; Smit EA; Kampen IM; Groeneveld CN; Fase KM; Van der Kolk J; Poiger H; Van den Berg M;

8	Koeman JH; Brouwer A (1995). Subchronic effects of 2,3,7,8-TCDD or PCBs on thyroid hormone metabolism: use

9	in risk assessment. Eur J Pharmacol, 293: 77-85. 197096

10	Van den Berg M; Birnbaum L; Bosveld AT; Brunstrom B; Cook P; Feeley M; Giesy JP; Hanberg A; Hasegawa R;

11	Kennedy SW; Kubiak T; Larsen JC; van Leeuwen FX; Liem AK; Nolt C; Peterson RE; Poellinger L; Safe S;

12	Schrenk D; Tillitt D; Tysklind M; Younes M; Waern F; Zacharewski T (1998). Toxic equivalency factors (TEFs) for

13	PCBs, PCDDs, PCDFs for humans and wildlife. Environ Health Perspect, 106: 775-792. 198345

14	Vanden Heuvel JP; Clark GC; Kohn MC; Tritscher AM; Greenlee WF; Lucier GW; Bell DA (1994). Dioxin-

15	responsive genes: examination of dose-response relationships using quantitative reverse transcriptase-polymerase

16	chain reaction. Cancer Res, 54: 62-68. 197551

17	Vanden Heuvel JP; Clark GC; Tritscher A; Lucier GW (1994). Accumulation of polychlorinated dibenzo-p-dioxins

18	and dibenzofurans in liver of control laboratory rats. Fundam Appl Toxicol, 23: 465-469. 594318

19	Vanni H; Kazeros A; Wang R; Harvey BG; Ferris B; De Bishnu P; Carolan BJ; Hiibner RH; O'Connor TP; Crystal

20	RG (2009). Cigarette smoking induces overexpression of a fat-depleting gene AZGP1 in the human airway

21	epithelium. Chest, 135: 1197-1208. 543754

22	van Birgelen AP; van den Berg M (2000). Toxicokinetics. Food Addit Contam, 17: 267-273. 523248

23	Van Birgelen AP; Van der Kolk J; Fase KM; Bol I; Poiger H; Brouwer A; Van den Berg M (1995). Subchronic

24	dose-response study of 2,3,7,8-tetrachlorodibenzo-p-dioxin in female Sprague-Dawley rats. Toxicol Appl

25	Pharmacol, 132: 1-13. 198052

26	Van Den Hove MF; Beckers C; Devlieger H; De Zegher F; De Nayer P (1999). Hormone synthesis and storage in

27	the thyroid of human preterm and term newborns: effect of thyroxine treatment. Biochimie, 81: 563-570. 016478

28	Van den Berg M; Birnbaum LS; Denison M; De Vito M; Farland W; Feeley M; Fiedler H; Hakansson H; Hanberg

29	A; Haws L; Rose M; Safe S; Schrenk D; Tohyama C; Tritscher A; Tuomisto J; Tysklind M; Walker N; Peterson RE
3 0	(2006). The 2005 World Health Organization reevaluation of human and mammalian toxic equivalency factors for

31	dioxins and dioxin-like compounds. Toxicol Sci, 93: 223-241. 543769

32	Van den Berg M; de Vroom E; Olie K; Hutzinger O (1986). Bioavailability of PCDDs and PCDFs of fly ash after

33	semi-chronic oral ingestion by guinea pig and Syrian golden hamster. Chemosphere, 15: 519-533. 543781

34	Van der Molen GW; Kooijman BA; Wittsiepe J; Schrey P; Flesch-Janys D; Slob W (2000). Estimation of dioxin and
3 5	furan elimination rates with a pharmacokinetic model. J Expo Anal Environ Epidemiol, 10: 579-585. 548777

36	Van der Molen GW; Kooijman SALM; Michalek JE; Slob W (1998). The estimation of elimination rates of

37	persistent compounds: A re-analysis of 2,3,7,8-tetrachlorodibenzo-p-dioxin levels in Vietnam veterans.

38	Chemosphere, 37: 1833-1844. 548765

39	Van der Molen, G; Kooijman A; Slob W (1996). A generic toxicokinetic model for persistent lipophilic compounds

40	in humans: An application to TCDD. Fundam Appl Toxicol, 31: 83-94. 548768

This document is a draft for review purposes only and does not constitute Agency policy.

R-30 DRAFT—DO NOT CITE OR QUOTE

-------
1	Viluksela M; Bager Y; Tuomisto JT; Scheu G; Unkila M; Pohjanvirta R; Flodstrom S; Kosma VM; Maki-

2	Paakkanen J; Vartiainen T; Klimm C; Schramm KW; Warngard L; Tuomisto J (2000). Liver tumor-promoting

3	activity of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) in TCDD-sensitive and TCDD-resistant rat strains. Cancer

4	Res, 60: 6911-6920. 198968

5	Vos JG, Moore JA, Zinkl JG (1973). Effect of 2,3,7,8-tetrachlorodibenzo-p-dioxin on the immune system of

6	laboratory animals. Environ Health Perspect, 5: 149-162. 198367

7	Walker NJ; Portier CJ; Lax SF; Crofts FG; Li Y; Lucier GW; Sutter TR (1999). Characterization of the dose-

8	response of CYP1B1, CYP1A1, and CYP1A2 in the liver of female Sprague-Dawley rats following chronic

9	exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Toxicol Appl Pharmacol, 154: 279-286. 198615

10	Walker NJ; Tritscher AM; Sills RC; Lucier GW; Portier CJ (2000). Hepatocarcinogenesis in female Sprague-

11	Dawley rats following discontinuous treatment with 2,3,7,8-tetrachlorodibenzo-p-dioxin. Toxicol Sci, 54: 330-337.

12	198733

13	Wang SL; Su PH; Jong SB; Guo YL; Chou WL; Papke O (2005). In utero exposure to dioxins and polychlorinated

14	biphenyls and its relations to thyroid function and growth hormone in newborns. Environ Health Perspect, 113:

15	1645-1650. 198734

16	Wang X; Santostefano MJ; DeVito MJ; Birnbaum LS (2000). Extrapolation of a PBPK model for dioxins across

17	dosage regimen, gender, strain, and species. Toxicol Sci, 56: 49-60. 198738

18	Wang X; Santostefano MJ; Evans MV; Richardson VM; Diliberto JJ; Birnbaum LS (1997). Determination of

19	parameters responsible for pharmacokinetic behavior of TCDD in female Sprague-Dawley rats. Toxicol Appl

20	Pharmacol, 147: 151-168. 104657

21	Ware JH; Spengler JD; Neas LM; Samet JM; Wagner GR; Coultas D; Ozkaynak H; Schwab M (1993). Respiratory

22	and irritant health effects of ambient volatile organic compounds: the Kanawha County health study. Am J

23	Epidemiol, 137: 1287-1301. 004687

24	Warner M; Eskenazi B; Mocarelli P; Gerthoux PM; Samuels S; Needham L; Patterson D; Brambilla P (2002).

25	Serum dioxin concentrations and breast cancer risk in the seveso women's health study. Environ Health Perspect,

26	110:625-628.197489

27	Warner M; Eskenazi B; Olive DL; Samuels S; Quick-Miles S; Vercellini P; Gerthoux PM; Needham L; Patterson

28	DG Jr; Mocarelli P (2007). Serum dioxin concentrations and quality of ovarian function in women of seveso.

29	Environ Health Perspect, 115: 336-340. 197486

30	Warner M; Samuels S; Mocarelli P; Gerthoux PM; Needham L; Patterson DG Jr; Eskenazi B (2004). Serum dioxin

31	concentrations and age at menarche. Environ Health Perspect, 112: 1289-1292. 197490

32	Weber R; Schmitz H-J; Schrenk D; Hagenmaier H (1997). Metabolic degradation, inducing potency, and

33	metabolites of fluorinated and chlorinated-fluorinated dibenzodioxins and dibenzofurans. Chemosphere, 34: 29-40.

34	548753

3 5	Wendling JM; Orth RG; Poiger H (1990). Determination of [3H]-2,3,7,8-tetrachlorodibenzo-p-dioxin in human

36	feces to ascertain its relative metabolism in man. Anal Chem, 62: 796-800. 548751

37	White KL Jr; Lysy HH; McCay JA; Anderson AC (1986). Modulation of serum complement levels following

38	exposure to polychlorinated dibenzo-p-dioxins. Toxicol Appl Pharmacol, 84: 209-219. 197531

39	White RH; Cote I; Zeise L; Fox M; Dominici F; Burke TA; White PD; Hattis DB; Samet JM (2009). State-of-the-

40	Science Workshop Report: Issues and Approaches in Low-Dose~Response Extrapolation for Environmental Health

41	Risk Assessment. Environ Health Perspect, 117: 283-287. 622764

This document is a draft for review purposes only and does not constitute Agency policy.

R-31 DRAFT—DO NOT CITE OR QUOTE

-------
1	WHO (1978). International Classification of Diseases: Ninth Revision. Geneva, Switzerland: World Health

2	Organization. 594329

3	WHO (1988). Assessment of the health risk of dioxins: re evaluation of the tolerable daily intake (TDI). WHO

4	European Centre for Environmental Health and International Programme on Chemical Safety. Geneva, Switzerland.

5	594278

6	WHO (2005). Chemical-specific adjustment factors for interspecies differences and human variability: guidance

7	document for use of data in dose/concentration-response assessment. World Health Organization. Geneva,

8	Switzerland. Harmonization Project Document No. 2. 198739

9	Whysner J; Williams GM (1996). 2,3,7,8-Tetrachlorodibenzo-p-dioxin mechanistic data and risk assessment: gene

10	regulation, cytotoxicity, enhanced cell proliferation, and tumor promotion. Pharmacol Ther, 71: 193-223. 197556

11	Wittsiepe J; Erlenkamper B; Welge P; Hack A; Wilhelm M (2007). Bioavailability of PCDD/F from contaminated

12	soil in young Goettingen minipigs. Chemosphere, 67: S355-S364. 548736

13	Wong TK; DominBA; Bent PE; BlantonTE; Anderson MW; Philpot RM (1986). Correlation of placental

14	microsomal activities with protein detected by antibodies to rabbit cytochrome P-450 isozyme 6 in preparations

15	from humans exposed to polychlorinated biphenyls, quaterphenyls, and dibenzofurans. Cancer Res, 46: 999-1004.

16	548795

17	Woods CG; Burns AM; Bradford BU; Ross PK; Kosyk O; Swenberg JA; Cunningham ML; Rusyn I (2007). WY-

18	14,643-induced cell proliferation and oxidative stress in mouse liver are independent of NADPH oxidase. Toxicol

19	Sci, 98: 366-374. 543735

20	Wyde ME; Cambre T; Lebetkin M; Eldridge SR; Walker NJ (2002). Promotion of altered hepatic foci by 2,3,7,8-

21	Tetrachlorodibenzo-p-dioxin and 17B-estradiol in male Sprague-Dawley rats. Toxicol Sci, 68: 295-303. 197009

22	Wyde ME; Eldridge SR; Lucier GW; Walker NJ (2001). Regulation of 2,3,7,8-tetrachlorodibenzo-p-dioxin-induced

23	tumor promotion by 17 beta-estradiol in female Sprague-Dawley rats. Toxicol Appl Pharmacol, 173: 7-17. 198575

24	Yang JZ; Agarwal SK; Foster WG (2000). Subchronic exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin modulates

25	the pathophysiology of endometriosis in the cynomolgus monkey. Toxicol Sci, 56: 374-381. 198590

26	Youakim S (2006). Risk of cancer among firefighters: A quantitative review of selected malignancies. Arch Environ

27	Occup Health, 61: 223-231. 197295

28	Zack JA; Gaffey WR (1983). A mortality study of workers employed at the Monsanto Company plant in Nitro, West

29	Virginia. Environ Sci Res, 26: 575-591. 548783

30	Zack JA; Suskind RR (1980). The mortality experience of workers exposed to tetrachlorodibenzodioxin in a

31	trichlorophenol process accident. J Occup Environ Med, 22: 11-14. 065005

32	Zareba G; Hojo R; Zareba KM; Watanabe C; Markowski VP; Baggs RB; Weiss B (2002). Sexually dimorphic

33	alterations of brain cortical dominance in rats prenatally exposed to TCDD. J Appl Toxicol, 22: 129-137. 197567

34	Zeise L; Wilson R; Crouch EAC (1987). Dose-response relationships for carcinogens: a review. Environ Health
3 5	Perspect, 73: 259-308. 060867

36	Zober A; Messerer P; Huber P (1990). Thirty-four-year mortality follow-up of BASF employees exposed to 2,3,7,8-

37	TCDD after the 1953 accident. Int Arch Occup Environ Health, 62: 139-157. 197604

This document is a draft for review purposes only and does not constitute Agency policy.

R-32 DRAFT—DO NOT CITE OR QUOTE

-------
1	Zober A; Ott MG; Messerer P (1994). Morbidity follow up study of BASF employees exposed to 2,3,7, 8-

2	tetrachlorodibenzo-p-dioxin (TCDD) after a 1953 chemical reactor incident. Occup Environ Med, 51: 479-486.

3	197572

4	Zober A; Papke O (1993). Concentrations of PCDDs and PCDFs in human tissue 36 years after accidental dioxin

5	exposure. Chemosphere, 27: 413-418. 197602

6	Zober A; Schilling D; Ott MG; Schauwecker P; Riemann JF; Messerer P (1998). Helicobacter pylori infection:

7	prevalence and clinical relevance in a large company. J Occup Environ Med, 40: 586-594. 594300

8	Altekruse, SF; Kosary, CL; Krapcho, M; et al., eds. (2010) SEER Cancer Statistics Review, 1975-2007. National

9	Cancer Institute. Bethesda, MD, based on November 2009 SEER data submission, posted to the SEER web site,

10	2010. Available online at http://seer.cancer.gov/csr/1975_2007/.

11	Auso, E; Lavado-Autric, R; Cuevas, E; et al. (2004) A moderate and transient deficiency of maternal thyroid

12	function at the beginning of fetal neocorticogenesis alters neuronal migration. Endocrinology 145:4037-4047.

13	Baird, SJS; Cohen, JT; Graham, JD, et al. (1996) Noncancer risk assessment: a probabilistic alternative to current

14	practice. Human Ecol Risk Assess 2:79-102.

15	Calabrese, EJ; Gilbert, CE. (1993) Lack of total independence of uncertainty factors (Ufs): Implications for the size

16	of the total uncertainty factor. Reg Toxicol Pharmacol 17:44-51.

17	Calabrese, EJ; Baldwin, LA. (1995) A toxicological basis to derive generic interspecies uncertainty factors for

18	application in human and ecological risk assessment. Human Ecol Risk Assess l(5):555-564.

19	Calvo, RM; Jauniaux, E; Gulbis, B; et al. (2002) Fetal tissues are exposed to biologically relevant free thyroxine

20	concentrations during early phases of development. J Clin Endocrinol Metab 87(4): 1768-1777.

21	Chan, S; Franklyn, JA; Kilby, MD. (2005) Maternal thyroid hormones and fetal brain development. Curr Opinion

22	Endocrinol Diab 12:23-30.

23	Chu, I; Valli, VE; Rousseaux, CG. (2007) Combined effects of 2,3,7,8-tetrachlorodibenzo-pdioxin and

24	polychlorinated biphenyl congeners in rats. Toxicol Environ Chem 89(l):71-87.

25	Cook, RR. (1981) Dioxin, chloracne, and soft tissue sarcoma. Lancet 1:618-619.

26	Crump, KS; Chiu, WA; Subramanian, RP. (2010) Issues in using human variability distributions to estimate low-

27	dose risk. Environ Health Perspect 118(3):387—393.

28	Delange, F; Bourdoux, P; Ermans, AM. (1985) Transient disorders of thyroid function and regulation in preterm

29	infants. In: Delange, F; Fisher, D; Malvaux, P; eds. Pediatric Thyroidology. Basel, S. Karger. pp 369-393.

30	Delia Porta, G; Dragani, TA; Sozzi, G. (1987) Carcinogenic effects of infantile and long-term

31	2,3,7,8-tetrachlorodibenzo-p-dioxin treatment in the mouse. Tumori73: 99-107.

32	Denison, MS; Nagy, SR. (2003) Activation of the aryl hydrocarbon receptor by structurally diverse exogenous and

33	endogenous chemicals. Annu Rev Pharmacol Toxicol 43:309-334.

34	Evans, JS; Baird, SJS. (1998) Accounting for missing data in noncancer risk assessment. Human Ecological Risk

35	Assess 4:291-317.

36	Geusau, A; Abraham, K; Geissler, K; et al. (2001) Severe 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) intoxication:

37	clinical and laboratory effects. Environ Health Perspect 109(8):865-869.

This document is a draft for review purposes only and does not constitute Agency policy.

R-33 DRAFT—DO NOT CITE OR QUOTE

-------
1	Glinoer, D; Delange, F. (2000) The potential repercussions of maternal, fetal, and neonatal hypothyroxinemia on the

2	progeny. Thyroid 10(10):871-887.

3	Haavisto, TE; Myllymaki, SA; Adamsson, NA; et al. (2006) The effects of maternal exposure to

4	2,3,7,8-tetrachlorodibenzo-p-dioxin on testicular steroidogenesis in infantile male rats. Int J Androl 29:313-322.

5	Hahn, ME. (2002) Aryl hydrocarbon receptors diversity and evolution. Chem-Biol Interact 141:131-160.

6	Horner MJ; Ries LAG; Krapcho M; et al.; eds. (2009) SEER Cancer Statistics Review, 1975-2006. Betheda, MD:

7	National Cancer Institute. Available online at http://seer.cancer.gov/csr/1975_2006/, based on November 2008

8	SEER data submission, posted to the SEER web site, 2009.

9	Hutt, KJ; Shi, Z; Albertini, DF; et al. (2008) The environmental toxicant 2,3,7,8-tetrachlorodibenzo-p-dioxin

10	disrupts morphogenesis of the rat pre-implantation embryo. BMC Developmental Biology 8:1-12.

11	IOM (Institute of Medicine). (1994) Veterans, and Agent Orange: health effects of herbicides used in Vietnam.

12	Washington, DC: National Academy Press.

13	IPCS (International Programme on Chemical Safety). (2005) Chemical-specific adjustment factors for interspecies

14	differences and human variability: guidance document for use of data in dose/concentration-response assessment.

15	harmonization project Document No. 2. World Health Organization, Geneva, Switzerland.

16	Kahn, PC; Gochfeld, M; Nygren, M; et al. (1988) Dioxins and dibenzofurans in blood and adipose tissue of Agent

17	Orange-exposed Vietnam veterans and matched controls. JAMA 259:1661-1667.

18	Kang, HK; Dalager, NA; Needham, LL; et al. (2006) Health status of Army Chemical Corps Vietnam veterans who

19	sprayed defoliant in Vietnam. Amer J Indust Med 49:875-884.

20	Kodell, RL; Gaylor, DW. (1999) Combining uncertainty factors in deriving human exposure levels of

21	noncarcinogenic toxicants. Annals New York Academy of Sciences 895:188-195.

22	Krishnan, K; Andersen, M. (2007) Physiologically based pharmacokinetic modelling in toxicology. In Principles

23	and methods of toxicology (A.W.Hayes, Ed.), 5th ed., pp. 231-291. CRC Press, New York.

24	Landi, MT; Bertazzi, PA; Baccarelli, A; et al. (2003) TCDD-mediated alterations in the AhR-dependent pathway in

25	Seveso, Italy, 20 years after the accident. Carcinogenesis 24:673-680.

26	Lavado-Autric, R; Auso, E; Garcia-Velasco, JV; et al. (2003) Early maternal hypothyroxinemia alters histogenesis

27	and cerebral cortex cytoarchitecture of the progeny. J Clin Invest 111: 1073-1082.

28	Lutz, WK; Gaylor, DW; Conolly, RB, et al. (2005) Nonlinearity and thresholds in dose-response relationships for

29	carcinogenicity due to sampling variation, logarithmic dose scaling, or small differences in individual susceptibility.

30	Toxicol Appl Pharmacol 207(Suppl. 2):565-569.

31	Morreale de Escobar, G; Obregon, MJ; et al. (2000) Is neuropsychological development related to maternal

32	hypothyroidism or to maternal hypothyroxinemia? J Clin Endocrinol Metab 85(11):3975-3987.

33	NAS (National Academy of Sciences), ed. (2005) Health Implications of Perchlorate Ingestion. Washington DC:

34	National Research Council of the National Academies.

35	NAS (National Academy of Sciences). (2007) Toxicity testing in the 21st century. A vision and a strategy. Report

36	of the Committee on Toxicity Testing and Assessment of Environmental Agents. National Research Council of The

37	National Academies. Washington, DC: The National Academies Press. Available online at

38	www.nap.edu/catalog/11970.html.

This document is a draft for review purposes only and does not constitute Agency policy.

R-34 DRAFT—DO NOT CITE OR QUOTE

-------
1	Navarro, C; Chirlaque, MD; Tormo, MJ; et al. (2006) Validity of self reported diagnoses of cancer in a major

2	Spanish prospective cohort study. J Epidemiol Comm Health 60: 593-599.

3	Needham, LL; Gerthoux, PM; Patterson, DG, Jr; et al. (1997) Serum dioxin levels in Seveso, Italy, population in

4	1976. Teratog Carcinog Mutagen 17:225-240.

5	NRC (National Research Council). (2005) Health risks from exposure to low levels of ionizing radiation: BEIR VII.

6	Washington, DC: National Academy Press (as cited by White et al., 2008).

7	NTP (National Toxicology Program). (2006a) NTP technical report on the toxicology and carcinogenesis studies of

8	2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) (CAS No. 1746-01-6) in female Harlan Sprague-Dawley rats (Gavage

9	Studies). Natl Toxicol ProgramTech Rep 521. Public Health Service, National Institute of Health, U.S. Department

10	of Health and Human Services, Research Triangle Park, NC.

11	Okura, Y; Urban, LH; Mahoney, DW; et al. (2004) Agreement between self-report questionnaires and medical

12	record data was substantial for diabetes, hypertension, myocardial infarction and stroke but not for heart failure.

13	J Clinic Epidemiol 57: 1096-110

14	Patterson, D; Hampton, L; Lapeza, CR, Jr; et al. (1987) High resolution gas chromatographic/high resolution mass

15	spectrometer analysis of human serum on a whole-weight and lipid basis for 2,3,7,8-tetrachlorodibenzo-p-dioxin.

16	Anal Chem 59:2000-2005.

17	Patterson, DG, Jr.; Wong, L-Y; Turner, WE; et al. (2009) Levels in the U. S. population of those persistent organic

18	pollutants (2003-2004) included in the Stockholm Convention or in other Long-Range Tran boundary Air Pollution

19	Agreements. Environ Sci Technol 43(4): 1211-1218.

20	Pesonen, SA; Haavisto, TE; Viluksela, M; et al. (2006) Effects of in utero and lactational exposure to

21	2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) on rat follicular steroidogenesis. Reprod Toxicol 22:521-528.

22	Pitotetal. 1991 pg 5-36. Not listed in reference section and not in HERO. Maybe Pitot and Dragan (available in

23	HERO)?

24	Poiger, M; Schlatter, C. (1986) Pharmacokinetics of 2,3,7,8-TCDD in man. Chemosphere 15:9-12.

25	Puga, A; Barnes, SJ; Dalton, TP; et al. (2000) Aromatic hydrocarbon receptor interaction with the retinoblastoma

26	protein potentiates repression of E2F-dependent transcription and cell cycle arrest. J Biol Chem 275:2943-2950.

27	Rigon, F; Bianchin, L; Bernasconi, S; et al. (2010) Update on age at menarche in Italy: toward the leveling off of

28	the secular trend. J Adolesc Health 46(3):238-244.

29	Rovet, JF. (2002) Congenital hypothyroidism: an analysis of persisting deficits and associated factors. Child

30	Neuropsychol 8(3): 150-62.

31	Roy land, J; Parker, J; Gilbert, ME. (2008) A genomic microarray analysis of hippocampus and neocortex following

32	modest reductions thyroid hormone during development. J Neuroendocrinol 12:1319-13

33	Safe, SH. (1986) Comparative toxicology and mechanism of action of polychlorinated dibenzo-p-dioxins and

34	dibenzofurans. Annu Rev Pharmacol Toxicol 26:371-379.

3 5	Savin, S; Cvejic, D; Nedic, Oet al. (2003) Thyroid hormone synthesis and storage in the thyroid gland of human

36	neonates. J. Pediatr Endocrinol Metab 16:521-528.Schantz, SL; Bowman, RE. (1989) Learning in monkeys

37	exposed perinatally to 2,3,7,8-tetrachloridibenzo-p-dioxin (TCDD). Neurotoxicol Teratol 11:13-19.

3 8	Sharlin, DS; Tighe, D; et al. (2008) The balance between oligodendrocyte and astrocyte production in major white

39	matter tracts is linearly related to serum total thyroxine. Endocrinology 149(5):2527-2536.

This document is a draft for review purposes only and does not constitute Agency policy.

R-3 5 DRAFT—DO NOT CITE OR QUOTE

-------
1	Sharlin, DS; Gilbert, ME; Taylor, M; et al. (2010). The nature of the compensatory response to low thyroid hormone

2	in the developing brain. J Neuroendocrinol. 22(3): 153-165.

3	Slama, R; Eutache, F; Ducot, B; et al. (2002) Time to pregnancy and semen parameters: a cross-sectional study

4	among fertile couples from four European cities. Human Repro 17:503-515.

5	Subramaniam, RP; White, P; Cogliano, VJ. (2006) Comparison of cancer slope factors using different statistical

6	approaches. Risk Anal. 26(3):825-830.

7	Swan, SH; Brazil, C; Drobnis, EZ; et al. (2003) Geographic differences in semen quality of fertile U.S. males.

8	Environ Health Perspect 111(4):414-20.

9	U.S. EPA (Environmental Protection Agency). (1994) Methods for derivation of inhalation reference concentrations

10	and application of inhalation dosimetry. October. Office of Health and Environmental Assessment, Environmental

11	Criteria and Assessment Office, Washington, DC. EPA/600/8-90/066F.

12	U.S. EPA (Environmental Protection Agency). (2001) Evaluation of the carcinogenic potential of lindane. Final

13	Report. Cancer assessment document. Cancer Assessment Review Committee, Health Effects Division, Office of

14	Pesticide Programs, Washington, DC. Available online at

15	http://www.lindane.com/pdf/EPA_Cancer_Assessment_of_Lindane2001 .pdf.

16	Vermeire, T; Stevenson, H; Pieters, MN; et al. (1999) Assessment factors for human health risk assessment: a

17	discussion paper. CritRev Toxiocol 29(5):439-490.

18	Walker, NJ; Miller, BD; Kohn, MC; et al. (1998). Differences in kinetics of induction and reversibility of

19	TCDD-induced changes in cell proliferation and CYP1 Al expression in female Sprague-Dawley rat liver.

20	Carcinogenesis 19:1427-1435.

21	Walker, NJ; Tritscher, AM; Sills, RC; et al. (2000) Hepatocarcinogenesis in female Sprague-Dawley rats following

22	discontinuous treatment with 2,3,7,8-tetrachlorodibenzo-p-dioxin Toxicol Sci 54:330-337.

23	Ware, J. (1993) Appendix C: Script for personal interview SF-36 administration. In: Ware, JE, Jr; Snow, KK;

24	Kosinski, M; et al., eds. SF-36 health survey manuals and interpretation guide. Boston, MA: Nimrod Press.

25	Warner, M; Eskenazi, B. (2005) TCDD and puberty: Warner and Eskenazi Respond. Environ Health Perspect

26	113:A18-A18.

27	White, RH; Cote, I; Zeise, L; et al. (2009) State-of-the-science workshop report: issues and approaches in low-dose-

28	response extrapolation for environmental health risk assessment. Environ Health Perspect 117(2):283-287.

29	Whitlock, JP. (1999) Induction of cytochrome P4501A1. Annu Rev Pharmacol Toxicol 39:103-125.

30	WHO (World Health Organization). (1994) Indicators for assessing iodine deficiency disorders and their control

31	through salt iodization. Geneva: World Health Organization. WHO/NUT/94.6 WHO/NUT/94.6.

32	WHO (World Health Organization). (2007) Assessment of iodine deficiency disorders and monitoring their

33	elimination. Geneva: WHO Press.

34	Wijchman, JG; DeWolf, B; Graaff, R; et al. (2001) Variation in semen parameters derived from computer aided
3 5	semen analysis within donors and between donors. J Androl 22(5):773-780.

36	Wyrobek, AJ; Gordon, LA; Watchmaker, G; et al. (1982) Human sperm morphology testing: description of a

37	reliable method and its statistical power. In: Bredges, BA; Butterworth, BE; Weinstein, IB; eds. Banbury Report

38	Indicators of Genotoxic Exposure. Cold Spring Harbor, NY: Cold Spring Laboratory, pp. 527-54

This document is a draft for review purposes only and does not constitute Agency policy.

R-36 DRAFT—DO NOT CITE OR QUOTE

-------
1	Zoeller, RT; Rovet, J. (2004). Timing of thyroid hormone action in the developing brain: clinical observations and

2	experimental findings. J Neuroendocrinol 16(10):809-818.

3	Zeise, L; Wilson, R; Crouch, E.A. (1987) Dose-response relationships for carcinogens: a review. Environ Health

4	Perspect 73:259-306.

5

This document is a draft for review purposes only and does not constitute Agency policy.

R-37 DRAFT—DO NOT CITE OR QUOTE

-------
DRAFT

DO NOT CITE OR QUOTE

May 2010
External Review Draft

APPENDIX A
Dioxin Workshop Report

NOTICE

THIS DOCUMENT IS AN EXTERNAL REVIEW DRAFT. It has not been formally released
by the U.S. Environmental Protection Agency and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its technical accuracy and policy
implications.

National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH

-------
EPA/600/R-09/027
May 2009

Summary of U.S. EPA
Dioxin Workshop

February 18-20, 2009

Cincinnati, Ohio

National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268

-------
DISCLAIMER

This document summarizes the discussions presented at the Dioxin Workshop in
February 2009, in Cincinnati, OH, as documented by the Session Co-Chairs. This document is
not all inclusive or binding. Conclusions and recommendations to the U.S. EPA may not
represent full consensus. The views expressed in this document are those of the Dioxin
Workshop Panelists and do not necessarily reflect the views and policies of the U.S.
Environmental Protection Agency. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.

Preferred Citation:

U.S. Environmental Protection Agency (U.S. EPA). (2009) Summary of U.S. EPA Dioxin Workshop:

February 18-20, 2009. U.S. Environmental Protection Agency, National Center for Environmental Assessment,
Cincinnati, OH. EPA/600/R-09/027.

This document is a draft for review purposes only and does not constitute Agency policy.

A-ii DRAFT—DO NOT CITE OR QUOTE

-------
TABLE OF CONTENTS

DIOXIN WORKSHOP TEAM	A-iv

ACKNOWLEDGMENTS	A-iv

INTRODUCTION	A-l

REFERENCES	A-2

SCIENTIFIC WORKSHOP TO INFORM THE TECHNICAL WORK PLAN FOR U.S.

EPA'S RESPONSE TO NAS COMMENTS ON THE HEALTH EFFECTS OF
DIOXIN PRESENTED IN U.S. EPA'S DIOXIN REASSESSMENT	A-3

SESSION 1: QUANTITATIVE DOSE-RESPONSE MODELING ISSUES	A-3

SESSION 2: IMMUNOTOXICITY	A-6

SESSION 3 A: DOSE-RESPONSE FOR NEUROTOXICITY AND

NONREPRODUCTIVE ENDOCRINE EFFECTS	A-8

SESSION 3B: DOSE-RESPONSE FOR CARDIOVASCULAR TOXICITY

AND HEPATOTOXICITY	A-l 1

SESSION 4A: DOSE-RESPONSE FOR CANCER	A-13

SESSION 4B: DOSE-RESPONSE FOR

REPRODUCTIVE/DEVELOPMENTAL TOXICITY	A-16

SESSION 5: QUANTITATIVE UNCERTAINTY ANALYSIS OF DOSE-

RESPONSE	A-20

APPENDIX A: 2009 U.S. EPA DIOXIN WORKSHOP AGENDA	A-24

APPENDIX B: 2009 U.S. EPA DIOXIN WORKSHOP QUESTIONS TO GUIDE

PANEL DISCUSSIONS	A-31

APPENDIX C: 2009 U.S. EPA DIOXIN WORKSHOP DRAFT SELECTION
CRITERIA TO IDENTIFY KEY IN VIVO MAMMALIAN STUDIES THAT
INFORM DOSE-RESPONSE MODELING FOR

2,3,7,8-TETRACHLORODIBENZO-p-DIOXIN (TCDD)	A-34

This document is a draft for review purposes only and does not constitute Agency policy.

A-iii DRAFT—DO NOT CITE OR QUOTE

-------
DIOXIN WORKSHOP TEAM

The Dioxin Workshop Team, under the leadership of Peter W. Preuss, Director, NCEA,
comprised the following members:

National Center for Environmental Assessment, Office of Research and Development,
U.S. Environmental Protection Agency, Cincinnati, OH 45268
Belinda S. Hawkins
Janet Hess-Wilson
Glenn Rice
Jeff Swartout
Linda K. Teuschler
Bette Zwayer

Argonne National Laboratory, Argonne, IL 60439
Maryka H. Bhattacharyya
Andrew Davidson
Mary E. Finster
Margaret M. MacDonell
David P. Peterson

ACKNOWLEDGMENTS

The Track Group, Alexandria, VA 22312
Kara Hennigan
Alan Minton
Brandy Quinn

ECFlex, Inc., Fairborn, OH 45324
Dan Heing
Heidi Glick
Amy Prues
Lana Wood

IntelliTech Systems, Inc., Fairborn, OH 45324
Cris Broyles
Luella Kessler
Stacey Lewis
Linda Tackett

This document is a draft for review purposes only and does not constitute Agency policy.

A-iv DRAFT—DO NOT CITE OR QUOTE

-------
INTRODUCTION

This document provides a summary of the Scientific Workshop to Inform EPA's
Response to National Academy of Science Comments on the Health Effects of Dioxin in EPA's
2003 Dioxin Reassessment. The U.S. Environmental Protection Agency (U.S. EPA) and
Argonne National Laboratories (ANL), through an inter-Agency agreement with the U.S.
Department of Energy, convened this scientific workshop ("Dioxin Workshop") on February
18-20, 2009, in Cincinnati, Ohio. The goals of the Dioxin Workshop were to identify and
address issues related to the dose-response assessment of 2,3,7,8-tetrachlorodibenzo-p-dioxin
(TCDD). This report summarizes the discussions and conclusions from this workshop.
Previously, at the request of the U.S. EPA, the National Academy of Sciences (NAS) prepared a
report, Health Risks from Dioxin and Related Compounds: Evaluation of the EPA Reassessment
(NAS, 2006), which made a number of recommendations to improve the U.S. EPA's risk
assessment for TCDD (U.S. EPA, 2003). The 3-day Dioxin Workshop was convened
specifically to ensure that the U.S. EPA's response to the NAS recommendations focuses on the
key issues and reflects the most meaningful science.

The Dioxin Workshop included seven scientific sessions:

(1)	Session 1:

(2)	Session 2:

(3)	Session 3A

(4)	Session 3B

(5)	Session 4A

(6)	Session 4B

(7)	Session 5:

Quantitative Dose-Response Modeling Issues
Immunotoxicity

Dose-Response for Neurotoxicity and Nonreproductive Endocrine Effects
Dose-Response for Cardiovascular Toxicity and Hepatotoxicity
Dose-Response for Cancer

Dose-Response for Reproductive/Developmental Toxicity
Quantitative Uncertainty Analysis of Dose-Response

During each session, the U.S. EPA asked a panel of expert scientists to:

•	identify and discuss the technical challenges involved in addressing the key NAS
comments on the TCDD dose-response assessment in the U.S. EPA Reassessment
(U.S. EPA, 2003);

•	discuss approaches for addressing the key NAS comments; and

•	identify important published, independently peer-reviewed literature, particularly studies
describing epidemiologic and in vivo mammalian bioassays, which are expected to be
most useful for informing the U.S. EPA's response.

The sessions were followed by open comment periods during which members of the
audience were invited to address the Panels. At the conclusion of the open comment periods, the
Panel Co-Chairs were asked to summarize and present the results of the panel discussions. The
summaries could include minority opinions stated by panelists. The main points derived from
the session summaries were used to prepare this document. Additionally, this document includes
a list of the session panelists and their affiliations and three appendices. Appendix A presents
the Dioxin Workshop Agenda. Appendix B identifies the charge questions presented to the
Panel. Appendix C describes draft study selection criteria proposed by the Dioxin Workshop
Team for consideration by the workshop panelists.

This document is a draft for review purposes only and does not constitute Agency policy.

A-1 DRAFT—DO NOT CITE OR QUOTE

-------
REFERENCES

NAS (National Academy of Sciences). 2006. Health Risks from Dioxin and Related
Compounds: Evaluation of the EPA Reassessment. National Academies Press, Washington, DC
(July). Available at http://www.nap.edu/catalog.php7record id=l 1688.

U.S. EPA (U.S. Environmental Protection Agency). 2003. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds. NAS
review draft, Volumes 1-3 (EPA/600/P-00/001Cb, Volume 1). U.S. Environmental Protection
Agency, National Center for Environmental Assessment, Washington, DC (December).
Available at http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/.

This document is a draft for review purposes only and does not constitute Agency policy.

A-2 DRAFT—DO NOT CITE OR QUOTE

-------
SCIENTIFIC WORKSHOP TO INFORM I II I TECHNICAL WORK PLAN FOR U.S.
EPA'S RESPONSE TO NAS COMMENTS ON THE HEALTH EFFECTS OF DIOXIN
PRESENTED IN U.S. EPA'S DIOXIN REASSESSMENT

Dioxin Workshop Co-Chairs: Peter W. Preuss and Glenn Rice

The Dioxin Workshop session summaries were prepared by the session panel Co-Chairs
with input from the panelists, as requested by the U.S. EPA prior to the workshop. The Co-
Chairs subsequently presented these summaries to all of the workshop participants during
designated periods at the workshop. In these summaries, the U.S. EPA asked that the Co-Chairs
summarize the key issues from the panel discussions. Because the sessions were not designed to
achieve consensus among the panelists, the summaries do not necessarily represent consensus
opinions; rather, they reflect the essence of the panel discussions. Some of the specific points
may represent the views of multiple panelists, while others only the views of a single panelist.
Prior to the summarizations, there were opportunities for public comments on the discussion
topics. Some Co-Chairs met with their sessions' panelists after their sessions ended to develop
these summaries, while others developed reports based on their personal notes. Because Session
5 was the last session of the workshop—with little time provided to develop the summary—the
Co-Chairs circulated a draft for comment by the Session 5 panelists after the workshop, prior to
finalizing the session summary. The U.S. EPA collected the session summaries and then
prepared this document. A draft of this document was distributed to all of the session Co-Chairs
to provide them with a final opportunity to comment and make revisions. Finally, it should be
noted that U.S. EPA was not prescriptive to the session Co-Chairs with respect to the format of
the presentation materials and provided no specific instructions, resulting in unique formats
among the session summaries.

SESSION 1: QUANTITATIVE DOSE-RESPONSE MODELING ISSUES

This session discussed the general dose-response modeling issues related to TCDD.

Many of these issues were highlighted by NAS (2006). There was a general introductory
presentation on TCDD kinetics, including information and uncertainties pertaining to the
conversion of administered doses in animals to human body burden (BB) and additivity to
background issues. This presentation was followed by a Panel discussion on the state of the
science regarding dioxin dose-response modeling issues.

Session 1 Panelists (Session Co-Chairs are identified by asterisk)

•	Bruce Allen, Bruce Allen Consulting

•	Lesa Aylward, Summit Toxicology

•	Roger Cooke, Resources for the Future

•	Kenny Crump, Louisiana Tech University

•	MikeDeVito, U.S. EPA

•	Dale Hattis, Clark University

•	Rick Hertzberg, Biomath Consulting

•	Rob McDowell, U.S. Department of Agriculture

•	Jim Olson, State University of New York, University at Buffalo

This document is a draft for review purposes only and does not constitute Agency policy.

A-3 DRAFT—DO NOT CITE OR QUOTE

-------
•	*Lorenz Rhomberg, Gradient

•	Woody Setzer, U.S. EPA

•	*Jeff Swartout, U.S. EPA

Please note that the use of the term "concluded" or "recommended" in this summary does not mean that a consensus
was reached. Session Summaries were written from the material prepared by the non-EPA/ANL Co-Chair and
represent a synopsis of the panel discussions.

Key Study Selection Criteria

The Panel discussed the advantages and disadvantages of using key study criteria
(Appendix C). They concluded that a priori criteria foster transparency and consistency, and
could deflect a posteriori criticism. However, the Panel also acknowledged that having a priori
criteria could introduce the potential for excluding useful data. Although the key study criteria
provided by the U.S. EPA listed studies using TCDD only as a criterion, the Panel posed the
possibility of using closely related dioxin-like compounds (DLCs) as surrogates for TCDD. The
criterion for use of data from mammalian studies only was one criterion that received generalized
support due to the lack of extrapolation protocols for nonmammalian species. The Panel also
discussed the specific exposure-duration criterion and asked if there should be a preference for
longer-term rather than acute studies. The Panel made three suggestions to modify U.S. EPA's
key study selection criteria:

(1)	Define more relevant exposure-level (i.e., dose) cut points using tissue concentrations.

(2)	Reword statistical criteria to include do-it-yourself analysis.

(3)	Reword the response criteria to clarify "outside of normal range."

Dose Metrics

The Panel discussed the relative merits of various measures of dose for modeling TCDD
dose response. One general conclusion was that tissue concentration (TC) is the preferred
metric, especially lipid-adjusted TC, because this measure more closely approximates exposures
close to the target tissue when compared to administered doses. However, the Panel
acknowledged that these data are often unavailable. They further noted that BB, which is
defined as the concentration of TCDD in the body (ng/kg body weight) (U.S. EPA, 2003), might
be useful as a surrogate for TC provided the two measures were proportional.

The Panel suggested that a linear approach to BB estimation, which was utilized by
U.S. EPA (2003), is too simplistic because this approach does not take into account toxicokinetic
issues related to TCDD—e.g., sequestration in the liver and fat, age-dependent elimination, and
changing elimination rates over time. The Panel recommended the use of kinetic/mechanistic
modeling to the extent possible to quantify tissue-based metrics.

The Panel raised the issue of whether the preferred dose metric would be different for
different endpoints and exposure durations. This led to the Panel's comment that the peak
exposure might be a more important metric than average BB for variable exposure scenarios.
Given this discussion about different exposure durations being relevant to a specific endpoint,
the Panel suggested that the U.S. EPA also consider peak measures in dose-response modeling.

This document is a draft for review purposes only and does not constitute Agency policy.

A-4 DRAFT—DO NOT CITE OR QUOTE

-------
The last point raised in this part of the discussion centered on the possibility of dose
errors in experimental studies. The Panel highlighted the need for the U.S. EPA to consider dose
error (i.e., uncertainty in the x-axis of the dose-response curve) when using dose surrogates.

Dose-Response Modeling of Mammalian Bioassays

The Panel considered several issues related to dose-response modeling of mammalian
bioassay data for TCDD: supralinearity and incomplete response data ("anchoring"), defining the
benchmark response (BMR) level with respect to establishing the point of departure (POD), and
the use of threshold modeling—as further explained below.

The Panel discussed the specific issues of supralinearity and anchoring raised by the
U.S. EPA with respect to modeling noncancer endpoints. The panel recognized that, for many of
the most sensitive endpoints, the response at the lowest dose is high (e.g., quantal responses
above 25% and continuous endpoints differ substantially from the mean, often implying 100%
incidence in the treated animals). This lack of response anchoring at the low end of the dose-
response curve (near the BMR) results in the higher responses determining the shape of the
curve.

The Panel asked whether new tools might be needed or whether the current tools could be
applied differently. In the context of developing new tools, the Panel emphasized the need for
collaboration between biologists and mathematicians. When discussing application, the Panel
suggested that the problem with supralinearity might be overcome by simply dropping the
requirement for using the lower bound on the Benchmark Dose. In addition, the Panel posed
several more approaches for further consideration in dose-response modeling by the U.S. EPA:

(1)	Combine similar data sets to fill in data gaps.

(2)	Use mechanistic approaches to model the data gaps.

(3)	Dichotomize continuous data.

Finally, the Panel acknowledged that, in certain situations, there simply may not be enough
information to provide meaningful answers.

The Panel discussed the BMR level for establishing a POD in the context of deriving a
Reference Dose (RfD). The Panel generally agreed that, while the effective dose level (ED0i)
used in the 2003 Reassessment may be useful for comparative analysis across endpoints, the
EDoi estimates developed for all endpoints considered in the Reassessment were not appropriate
for deriving an RfD because they were not based on the effect's adversity. The panel noted that
EDoi also is much lower than typical EPA BMR levels. The Panel recommended that the U.S.
EPA work to define endpoint-specific BMRs based on the consideration of adversity. Given that
the same uncertainty factor framework is applied to all PODs, the Panel emphasized the need for
consistency in BMRs; numerical consistency is needed for quantal BMRs and consistency in the
choice of biological relevance should be applied for continuous BMRs.

The Panel generally discouraged threshold modeling by stating that thresholds are very
difficult to pin down and suggested that the lower bound may always be zero.

This document is a draft for review purposes only and does not constitute Agency policy.

A-5 DRAFT—DO NOT CITE OR QUOTE

-------
Dose-Response Modeling of Epidemiological Studies

The Panel noted that many studies have been published with measured concentrations of
TCDD that could be used for dose reconstruction. In this discussion, the Panel acknowledged
that use of these data would entail dealing with toxicity equivalence (TEQ) issues and
pharmacokinetic (PK) modeling. Pertaining to the use of these data for quantitative risk
assessment by the U.S. EPA, the Panel posed the question, "At what point does indirect or
confounded human data supersede controlled animal bioassay data?", or alternatively, "How
much human data uncertainty can we tolerate?" The Panel suggested, at the least, that the
epidemiologic data could be used to "ground-truth" the animal bioassay modeling results.

Supporting Information

The Panel acknowledged that Ah receptor (AhR) binding affinities are not necessarily
tied to endpoint sensitivity, but they reiterated the need to consider mechanistic modeling to aid
in developing appropriate dose metrics or filling in data gaps in the existing dose-response data.

References

NAS (National Academy of Sciences). 2006. Health Risks from Dioxin and Related
Compounds: Evaluation of the EPA Reassessment. National Academies Press, Washington, DC
(July). Available at http://www.nap.edu/catalog.php?record id=l 1688.

U.S. EPA (U.S. Environmental Protection Agency). 2003. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds. NAS
Review Draft (EPA/600/P-00/001Cb). U.S. Environmental Protection Agency, National Center
for Environmental Assessment, Washington, DC. Available at
http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/.

SESSION 2: IMMUNOTOXICITY

The U.S. EPA plans to consider development of a quantitative dose-response assessment
for the immunologic effects associated with TCDD exposure. Such an assessment would be
based on information in U.S. EPA (2003), NAS (2006) and key studies identified in this
workshop. The purpose of this session was to identify and discuss key issues pertaining to dose-
response assessment for dioxin-induced immunologic effects.

Session 2 Panelists (Session Co-Chairs are identified by asterisk)

•	Roger Cooke, Resources for the Future

•	Rob Goble, Clark University

•	*Belinda Hawkins, U.S. EPA

•	Nancy Kerkvliet, Oregon State University

•	Manolis Kogevinas, Centre for Research in Environmental Epidemiology

•	Robert Luebke, U.S. EPA

•	Paolo Mocarelli, University of Milan

•	* Allen Silverstone, State University of New York, Upstate Medical University

This document is a draft for review purposes only and does not constitute Agency policy.

A-6 DRAFT—DO NOT CITE OR QUOTE

-------
•	Courtney Sulentic, Wright State University

•	Nigel Walker, National Institute of Environmental Health Sciences

Please note that the use of the term "concluded" or "recommended" in this summary does not mean that a consensus
was reached. Session Summaries were written from the material prepared by the non-EPA/ANL Co-Chair and
represent a synopsis of the panel discussions.

Key Study Selection Criteria

The Panel first addressed the Key Study Selection Criteria proposed by the U.S. EPA
(Appendix C). The Panel raised the issue that the key study criteria do not apply to most studies
designed to investigate immunotoxicity, including those used to calculate ED0iS (U.S. EPA,
2003). The Panel observed that most dioxin immunotoxicity studies are relatively high dose
(>200 ng/kg-d) acute studies and/or use parenteral rather than oral administration.

The Panel discussed several studies often considered important for assessing the
immunotoxic effects of TCDD exposure. The Oughton et al. (1995) mouse bioassay was
discussed and, although the study does meet the proposed criteria, it could not be considered a
key study; specifically, the Panel contended that since there were no functional alterations
observed or measured in this bioassay, the changes in cellular phenotypes are only "suggestive"
of immune alterations and cannot be regarded as having immunopathologic significance.

The Panel discussed two additional studies for further consideration by the U.S. EPA:

•	Baccarelli et al. (2002). The Panel discussed this as a potentially key human
epidemiological study that should be reviewed and considered further by the U.S. EPA.
It measured the level of IgG, demonstrating a significant decline relative to dioxin body
burdens.

•	Smialowicz et al. (2008). The Panel noted that this study identified the antibody response
to sheep red blood cells (SRBCs) as the critical effect, labeling this protocol as a
functional assay. The Panel stated that if modeled, the U.S. EPA could calculate the
BMR for this endpoint as 1 standard deviation from the control mean.

References

Baccarelli, A., P. Mocarelli, D.G. Patterson et al. 2002. Immunologic effects of dioxin: New
results from Seveso and comparison with other studies. Environ. Health Perspect.
110(12): 1169-1173.

NAS (National Academy of Sciences). 2006. Health Risks from Dioxin and Related
Compounds: Evaluation of the EPA Reassessment. National Academies Press, Washington, DC
(July). Available at http://www.nap.edu/catalog.php7record id=l 1688.

Oughton, J.A., C.B. Pereira, G.K. Dekrey, J.M. Collier, A.A. Frank and N.I. Kerkvliet. 1995.
Phenotypic analysis of spleen, thymus, and peripheral blood cells in aged C57BI/6 mice
following long-term exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Toxicol. Sci. 25(l):60-69.

This document is a draft for review purposes only and does not constitute Agency policy.

A-7 DRAFT—DO NOT CITE OR QUOTE

-------
Smialowicz, R.J., M.J. DeVito, W.C. Williams and L.S. Birnbaum. 2008. Relative potency
based on hepatic enzyme induction predicts immunosuppressive effects of a mixture of
PCDDS/PCDFS and PCBS. Toxicol. Appl. Pharmacol. 227(3):477-484.

U.S. EPA (U.S. Environmental Protection Agency). 2003. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds. NAS
Review Draft (EPA/600/P-00/001Cb). U.S. Environmental Protection Agency, National Center
for Environmental Assessment, Washington, DC. Available at
http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/.

SESSION 3A: DOSE-RESPONSE FOR NEUROTOXICITY AND NONREPRODUCTIVE
ENDOCRINE EFFECTS

The U.S. EPA plans to consider development of a quantitative dose-response assessment
for neurological and/or nonreproductive endocrine effects associated with TCDD exposure.

Such an assessment would be based on information in U.S. EPA (2003), NAS (2006) and key
studies identified in this workshop. The purpose of this session was to identify and discuss key
issues pertaining to dose-response assessment for dioxin-induced neurological and/or
nonreproductive endocrine effects.

Session 3A Panelists (Session Co-Chairs are identified by asterisk)

•	*Maryka Bhattacharyya, Argonne National Laboratory

•	Mike DeVito, U.S. EPA

•	Mary Gilbert, U.S. EPA

•	Rob Goble, Clark University

•	Nancy Kerkvliet, Oregon State University

•	Fumio Matsumura, University of California-Davis

•	Paolo Mocarelli, University of Milan

•	Chris Portier, National Institute of Environmental Health Sciences

•	Lorenz Rhomberg, Gradient

•	Allen Silverstone, State University of New York, Upstate Medical University

•	Marie Sweeney, National Institute of Occupational Safety and Health

•	*Bernie Weiss, University of Rochester

Please note that the use of the term "concluded" or "recommended" in this summary does not mean that a consensus
was reached. Session Summaries were written from the material prepared by the non-EPA/ANL Co-Chair and
represent a synopsis of the panel discussions.

What Are the Key Questions Regarding These Endpoints?

The Panel used the following question to initiate discussion: "Are there identifiable
indices of neurotoxicity and nonreproductive endocrine effects in animal studies and human
populations? " Under this discussion topic, the Panel discussed three endpoints: neurotoxicity
(with focus on developmental exposures), thyroid dysfunction (e.g., thyroid hormone deficits),
and diabetes. The Panel also addressed the relevance of windows of vulnerability to each

This document is a draft for review purposes only and does not constitute Agency policy.

A-8 DRAFT—DO NOT CITE OR QUOTE

-------
endpoint. The Panel acknowledged that, in some cases, the window of exposure may precede the
window of expression of toxicity.

Epidemiological Study Selection

Developmental Neurotoxicity

The Panel recognized that an unusual feature for this endpoint is that there are sufficient
human data for dose-response modeling (e.g., Dutch children [Huisman et al., 1995; Patandin et
al., 1999] and U.S. children [Jacobson and Jacobson, 1996]) and there is an internal dose metric
(serum concentrations). Additionally, the Panel discussed recent studies that address this
endpoint in humans (from Japan [reference not provided] and Holland [e.g., Koopman-Esseboom
et al., 1996; Vreugdenhil et al., 2002]). For continued investigation into this endpoint, the Panel
raised two issues to the U.S. EPA:

•	Conduct an evaluation of whether a modeled effect can be attributed to TCDD and not
some other persistent organic pollutant (POP), although the Panel recognized that it is
unlikely U.S. EPA will be able to distinguish among these exposures because other POPs
are intrinsic confounders in the Dutch study.

•	Allow animal data to inform the dose-response modeling of epidemiological data.

Thyroid Dysfunction

The Panel identified the availability of human data for this endpoint (e.g., Calvert et al.,
1999; Koopman-Esseboom et al., 1994). Much of the thyroid dysfunction literature has been
published since the 2003 Reassessment (e.g., Wang et al., 2005; Baccarelli et al., 2008). The
Panel also noted the availability of an internal dose metric (serum concentrations). Additionally,
the Panel discussed the mechanistic studies in animals that link TCDD to thyroid dysfunction.
For continued investigation into this endpoint, the Panel raised three issues for the U.S. EPA to
consider:

•	Consider the newly available human data since the Reassessment.

•	Investigate and clarify of the role of TCDD-induced thyroid dysfunction in
developmental neurotoxicity.

•	Evaluate and determine whether an effect can be attributed to TCDD or other
contaminants.

Diabetes

The Panel discussed that data suggest that diabetes incidence in those under 55 years old
may be associated with exposure to PCBs. They acknowledged that whether this is a dioxin-like
compound (DLC) mediated effect or whether other POPs are responsible is still undetermined.
The Panel also acknowledged that no animal model exists for the investigation of xenobiotic-
induced diabetes, and that separating the injury dose level from the current body burdens would
depend on good pharmacokinetics in humans. For continued investigation into this endpoint, the
Panel listed two issues for the U.S. EPA to consider:

•	Results from the Anniston study and the Great Lakes Fishermen study (references not
provided) should be examined for dose metrics (both studies examine human PCB
exposures).

This document is a draft for review purposes only and does not constitute Agency policy.

A-9 DRAFT—DO NOT CITE OR QUOTE

-------
• Changes of adipose tissue status need to be considered, given that dieting can cause
release of lipid-soluble contaminants.

References

Baccarelli, A., S.M. Giacomini, C. Corbetta et al. 2008. Neonatal thyroid function in Seveso 25
years after maternal exposure to dDioxin. PLoS Med. 5(7):el61.
doi: 10.1371/journal.pmed. 0050161.

Calvert, G.M., M.H. Sweeney, J. Deddens and D.K. Wall. 1999. Evaluation of diabetes
mellitus, serum glucose, and thyroid function among United States workers exposed to
2,3,7,8-tetrachlorodibenzo-p-dioxin. Occ. Env. Med. 56:270-276.

Huisman, M., C. Koopman-Esseboom, V. Fidler et al. 1995. Perinatal exposure to
polychlorinated biphenyls and dioxins and its effect on neonatal neurological development.

Early Hum. Devel. 41(2): 111-127.

Jacobson, J.L. and S.W. Jacobson. 1996. Intellectual impairment in children exposed to
polychlorinated biphenyls in utero. N. Engl. J. Med. 335:783-789.

Koopman-Esseboom, C., N. Weisglas-Kuperus, M.A.J. deRidder, C.G. Van derPaauw,
L.G.M.Th. Tuinstra and P.J.J. Sauer. 1996. Effects of polychlorinated biphenyl/dioxin exposure
and feeding type on infants' mental and psychomotor development. J. Pediatr. 97(5):700-706.

Koopman-Esseboom, C., D.-C. Morse, N. Weisglas-Kuperus et al. 1994. Effects of dioxins and
polychlorinated biphenyls on thyroid hormone status of pregnant women and their infants.
Pediatr. Res. 36:468-473.

Patandin, S., C.I. Lanting, P.G.H. Mulder, E.R. Boersma, P.J.J. Sauer and N. Weisglas-Kuperus.
1999. Effects of environmental exposure to polychlorinated biphenyls and dioxins on cognitive
abilities in Dutch children at 42 months of age. J. Pediatr. 134:33-41.

NAS (National Academy of Sciences). 2006. Health Risks from Dioxin and Related
Compounds: Evaluation of the EPA Reassessment. National Academies Press, Washington, DC
(July). Available at http://www.nap.edu/catalog.php7record id=l 1688.

U.S. EPA (U.S. Environmental Protection Agency). 2003. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds. NAS
Review Draft (EPA/600/P-00/001Cb). U.S. Environmental Protection Agency, National Center
for Environmental Assessment, Washington, DC. Available at
http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/.

Vreugdenhil, H.J., C.I. Lanting, P.G. Mulder, E.R. Boersma and N. Weisglas-Kuperus. 2002.
Effects of prenatal PCB and dioxin background exposure on cognitive and motor abilities in
Dutch children at school age. J. Pediatr. 140:48-56.

This document is a draft for review purposes only and does not constitute Agency policy.

A-10 DRAFT—DO NOT CITE OR QUOTE

-------
Wang S.L., P.H. Su, S.B. Jong, Y.L. Guo, W.L. Chou and O. Papke. 2005. In utero exposure to
dioxins and polychlorinated biphenyls and its relations to thyroid function and growth hormone
in newborns. Environ. Health Perspect. 113:1645-1650.

SESSION 3B: DOSE-RESPONSE FOR CARDIOVASCULAR TOXICITY AND
HEPATOTOXICITY

The U.S. EPA plans to consider development of a quantitative dose-response assessment
for cardiovascular and/or hepatic effects associated with TCDD exposure. Such an assessment
would be based on information in U.S. EPA (2003), NAS (2006) and key studies identified in
this workshop. The purpose of this session was to identify and discuss key issues pertaining to
dose-response assessment for dioxin-induced cardiovascular and/or hepatic effects.

Session 3B Panelists (Session Co-Chairs are identified by asterisk)

•	Bob Budinksy, Dow Chemical

•	Manolis Kogevinas, Centre for Research in Environmental Epidemiology

•	Rob McDowell, U.S. Department of Agriculture

•	Jim Olson, State University of New York, University at Buffalo

•	Marian Pavuk, Agency for Toxic Substances and Disease Registry

•	*Jeff Swartout, U.S. EPA

•	*Mary Walker, University of New Mexico

•	Nigel Walker, National Institute of Environmental Health Sciences

Please note that the use of the term "concluded" or "recommended" in this summary does not mean that a consensus
was reached. Session Summaries were written from the material prepared by the non-EPA/ANL Co-chair and
represents a synopsis of the panel discussions.

Key Study Selection Criteria

The Panel initially focused on the draft key study selection criteria offered by the
U.S. EPA (Appendix C). The panel recommended that for cardiovascular effects, which are not
usually observed in rodents, the use of knockout mouse models (ApoE KO and LDLR KO) be
moved to the "primary" column because only these studies establish the cardiovascular toxicity
model in mice.

The panel also was concerned that the gavage procedure can increase mouse blood
pressure. Consequently, the panel recommended that gavage studies not be used for the blood
pressure endpoint (i.e., only dietary dosing studies should be considered).

Human Health Endpoints

In relation to the hepatic endpoint, the Panel acknowledged the large body of dose
response information on hepatic effects in rodents and that enzyme (mostly CYP1 Al) induction
was a sensitive effect. However, the Panel cited the lack of linkage of CYP1A1 to downstream
events, which complicates the toxicological interpretation of this endpoint, and concluded that

This document is a draft for review purposes only and does not constitute Agency policy.

A-l 1 DRAFT—DO NOT CITE OR QUOTE

-------
the more important liver effects in rodents are probably on the "road to cancer." The Panel noted
that hepatic effects were not seen in the epidemiological studies, but acknowledged that these
studies were not designed to detect them.

In relation to the cardiovascular endpoint, the Panel identified hypertension and ischemic
heart disease (IHD) as two key endpoints from the epidemiological studies. The Panel
recommended that the U.S. EPA perform a meta-analysis of these data. The Panel also
commented that recent animal studies support the observations linking TCDD exposure to IHD
and hypertension. In particular, the National Toxicology Program (NTP) study shows
inflammatory and structural effects on resistant vascular arterioles (NTP, 2006). Additional
evidence from the study suggests that the vascular effects may be CYP1A1-dependent. The
Panel suggested that the NTP study data might be used as a surrogate for dose-response
modeling of hypertension and that such an approach would be supported by data on the role of
AhR in vascular function and remodeling.

POD Issues

The Panel was not supportive of 1% of maximal response (ED0i), which was utilized in
the 2003 Reassessment. The Panel concluded that the POD should depend on the specific
endpoint and recommended the following to the U.S. EPA:

•	For continuous measures, base the BMR on difference from control. Consider the
adversity level—at what point does the endpoint become adverse?

•	For incidence data, set the BMR to a fixed-risk level.

Supporting Information

The Panel posed several suggestions to the U.S. EPA for reducing uncertainty and
improving the knowledge base for TCDD toxicity.

•	Use in vitro data to define uncertainties, such as the relative sensitivity between rodents
and humans and around the definition of a POD.

•	Consider studies on dioxin-like compounds (DLCs).

•	Use PK modeling to define the dose metric for hepatic effects.

•	Use body burden or serum concentrations for cardiovascular endpoints.

Finally, the Panel recommended that U.S. EPA finish the reassessment quickly and establish a
definitive plan to review and incorporate new data as they become available.

References

NAS (National Academy of Sciences). 2006. Health Risks from Dioxin and Related
Compounds: Evaluation of the EPA Reassessment. National Academies Press, Washington, DC
(July). Available at http://www.nap.edu/catalog.php7record id=l 1688.

This document is a draft for review purposes only and does not constitute Agency policy.

A-12 DRAFT—DO NOT CITE OR QUOTE

-------
NTP (National Toxicology Program). 2006. Toxicology and Carcinogenesis Studies of
2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) (CAS No. 1746-01-6) in Female Harlan Sprague-
Dawley Rats (Gavage Studies). U.S. Department of Health and Human Services. NTP TR 521.
Research Triangle Park, NC (April).

U.S. EPA (U.S. Environmental Protection Agency). 2003. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds. NAS
Review Draft (EPA/600/P-00/001Cb). U.S. Environmental Protection Agency, National Center
for Environmental Assessment, Washington, DC. Available at
http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/.

SESSION 4A: DOSE-RESPONSE FOR CANCER

The U.S. EPA plans to consider development of a quantitative dose-response assessment
for cancer associated with TCDD exposure. Such an assessment would be based on information
in U.S. EPA (2003), NAS (2006) and key studies identified in this workshop. The purpose of
this session was to identify and discuss key issues pertaining to dose-response assessment for
dioxin-induced cancer.

Session 4A Panelists (Session Co-Chairs are identified by asterisk)

•	Lesa Aylward, Summit Toxicology

•	Kenny Crump, Louisiana Tech University

•	Dale Hattis, Clark University

•	* Janet Hess-Wilson, U.S. EPA

•	Karen Hogan, U.S. EPA

•	Manolis Kogevinas, Centre for Research in Environmental Epidemiology

•	Marian Pavuk, Agency for Toxic Substances and Disease Registry

•	Chris Portier, National Institute of Environmental Health Sciences

•	Lorenz Rhomberg, Gradient

•	Jay Silkworth, General Electric

•	*Nigel Walker, National Institute of Environmental Health Sciences

Please note that the use of the term "concluded" or "recommended" in this summary does not mean that a consensus
was reached. Session Summaries were written from the material prepared by the non-EPA/ANL Co-chair and
represent a synopsis of the panel discussions.

Key Study Selection

The Panel discussed both human and rodent studies. In reviewing the epidemiological
data, the Panel agreed the EPA should focus on four cohort studies (Dutch cohort, NIOSH
cohort, BASF accident cohort, and Hamburg cohort) and pointed out that there are numerous
updates and reevaluations of data now in the literature and others will be published soon. The
Panel stated that it is appropriate for the U.S. EPA to consider the increase in total cancers for
modeling human cancer data, however, Non-Hodgkin's lymphoma, and lung tumors are the main
TCDD-related cancer types seen in humans exposed to TCDD. The Panel suggested the U.S.
EPA focus the quantitative dose-response modeling on the human data.

This document is a draft for review purposes only and does not constitute Agency policy.

A-13 DRAFT—DO NOT CITE OR QUOTE

-------
In reviewing the rat data, the Panel identified four new NTP rodent cancer bioassays with
liver and lungs as the main target organs. However, they suggested that dose-response modeling
efforts should model "all cancers" from these NTP data sets as well and use tumor incidence—
not individual rats as measures.

Key Study Selection Criteria

The Panel discussed whether data for TCDD only should be used or if PCB126 could be
used to develop a dose-response curve. From this discussion, the Panel reached a general
agreement that limiting the dose-response modeling and cancer assessment to TCDD only would
be the best approach.

Regarding the oral dosing regimens, the Panel discussed the differences in results from
different bioassays. They concluded that there were insufficient data to pick between oral feed
(Kociba et al., 1978) and oral gavage (NTP, 2006) studies, but stated "If all aspects of studies
were equal, an oral feed study is preferred." However, given that current data sets are not equal,
they agreed that U.S. EPA should consider both feed and gavage studies.

The Panel put forth the recommendation that studies that include initiation-promotion
model data and TgAC transgenic model data from oral exposure studies should be excluded from
the primary category in the key study selection criteria (Appendix C lists the draft study selection
criteria distributed prior to the meeting). Studies from both classifications should be moved to
the second tier.

The Panel was also unsupportive of the "response magnitude outside the range of normal
variability" criterion, as they did not believe it was applicable to a cancer endpoint.

Critical Endpoints to Consider

The Panel recognized that the MOA for TCDD includes cell growth/differentiation
dysregulation, that different endpoints (tumor types) across species may be expected, and that
there are differences in tumor sites across species. The Panel further acknowledged that there is
insufficient information to determine if rodent tumor types observed are relevant to humans.
Thus, the Panel suggests the following:

• U.S. EPA should consider all the observed cancer endpoints in its evaluation.

Nonlinear (aka threshold) Versus Linear Dose-Response Modeling

The Panel agreed that NTP bioassays appear to demonstrate nonlinear dose response, but
they expressed concern about using animal data to infer slope and dose response for humans.
The Panel pointed out that there are differences in slopes across different bioassays, and
specifically, that some appear linear while others appear nonlinear. Given the observation of
both nonlinear vs. linear, the Panel concluded that neither could be ruled out for extrapolation
below the POD simply based on the available data. One panelist noted that U.S. EPA Cancer
Guidelines (U.S. EPA, 2005) state that only if one can demonstrate that the MOA has a threshold
dose-response shape, and can exclude all other potential linear MO As, can one use a nonlinear
model. Lastly, the Panel noted that there are data and rationales to support use of both linear and

This document is a draft for review purposes only and does not constitute Agency policy.

A-14 DRAFT—DO NOT CITE OR QUOTE

-------
nonlinear response below POD. From this discussion, the Panel raised one possibility to the U.S.
EPA:

•	Both linear and nonlinear model functions should be considered in the dose-response
analysis.

Dose Metrics

In considering human data, the Panel expressed a preference for lipid-adjusted serum
levels over body burden (BB), and they expressed concerns over the assumptions used in the
back calculation of the BB in the epidemiologic cohorts. In considering the rat data, the Panel
supported the use of BB—especially lipid-adjusted BB. The Panel, however, did express
concern over the sequestering of TCDD in liver and then the use of liver levels in BB
calculations.

Supporting Information—Biologically-Based Dose-Response (BBDR) Models and MOA

The Panel discussed BBDR. Though once considered an attractive proposition, BBDR
models may mask uncertainty within the models, necessitating them to be used with greater
caution. The Panel suggested two issues for the U.S. EPA to consider:

•	If there is a published model, use it if it is valid—do not generate a new model.

•	Focus on the actual experimental data to drive the analysis.

References

Kociba, R.J., D.G. Keyes, J.E. Beyer et al. 1978. Results of a two-year chronic toxicity and
oncogenicity study of 2,3,7,8-tetrachlorodibenzo-p-dioxin in rats. Toxicol. Appl. Pharmacol.
46:279-303.

NAS (National Academy of Sciences). 2006. Health Risks from Dioxin and Related
Compounds: Evaluation of the EPA Reassessment. National Academies Press, Washington, DC
(July). Available at http://www.nap.edu/catalog.php7record id=l 1688.

NTP (National Toxicology Program). 2006. Toxicology and Carcinogenesis Studies of
2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) (CAS No. 1746-01-6) in Female Harlan Sprague-
Dawley Rats (Gavage Studies). U.S. Department of Health and Human Services. NTP TR 521.
Research Triangle Park, NC (April).

U.S. EPA (U.S. Environmental Protection Agency). 2003. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds. NAS
Review Draft (EPA/600/P-00/001Cb). U.S. Environmental Protection Agency, National Center
for Environmental Assessment, Washington, DC. Available at
http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/.

U.S. EPA (U.S. Environmental Protection Agency). 2005. Guidelines for Carcinogen Risk
Assessment. U.S. Environmental Protection Agency Risk Assessment Forum.
EPA/630/P-03/001F.

This document is a draft for review purposes only and does not constitute Agency policy.

A-15 DRAFT—DO NOT CITE OR QUOTE

-------
SESSION 4B: DOSE-RESPONSE FOR REPRODUCTIVE/DEVELOPMENTAL
TOXICITY

The U.S. EPA plans to consider development of a quantitative dose-response assessment
for reproductive and developmental effects associated with TCDD exposure. Such an
assessment would be based on information in U.S. EPA (2003), NAS (2006) and key studies
identified in this workshop. The purpose of this session was to identify and discuss key issues
pertaining to dose-response assessment for dioxin-induced reproductive and developmental
effects.

Session 4B Panelists (Session Co-Chairs are identified by asterisk)

•	Barbara Abbott, U.S. EPA

•	Bruce Allen, Bruce Allen Consulting

•	Roger Cooke, Resources for the Future

•	George Daston, Procter & Gamble

•	MikeDeVito, U.S. EPA

•	Rob Goble, Clark University

•	*Fumio Matsumura, University of California-Davis

•	Paolo Mocarelli, University of Milan

•	Brian Petroff, University of Kansas

•	*Glenn Rice, U.S. EPA

•	Marie Sweeney, National Institute of Occupational Safety and Health

•	Mary Walker, University of New Mexico

•	Bernie Weiss, University of Rochester

Please note that the use of the term "concluded" or "recommended" in this summary does not mean that a consensus
was reached. Session Summaries were written from the material prepared by the non-EPA/ANL Co-Chair and
represent a synopsis of the panel discussions.

A Major Question Posed During this Workshop Session was "Are Human Embryos and
Infants Less Sensitive to Dioxin Exposures Than Some Experimental Animals?"

The Panel recognized that animal data show a wide range of species sensitivity to dioxin
for a given developmental or reproductive endpoint. Presently, there are data for some endpoints
that show that human sensitivity is comparable to experimental animals (e.g., semen quality),
and for other endpoints the data demonstrate that humans are insensitive compared to other
species (e.g., cleft palate). Lastly, the Panel recognized that there are some endpoints for which
relative human sensitivity remains uncertain.

Key Study Selection

The Panel reviewed the charge questions (Appendix B), discussed them, and listed two
issues for the U.S. EPA to consider:

•	Concerning key study determination, use a stepwise approach that is dependent upon the
information available and needed to address the question.

This document is a draft for review purposes only and does not constitute Agency policy.

A-16 DRAFT—DO NOT CITE OR QUOTE

-------
•	Concerning the key studies informing the POD and the POD endpoint choice, use the
POD to depart from what is certain and use a high-confidence study that has found
effects at a low enough level at which other effects are protected.

The Panel also developed Table 1, based on the information presented in this session. Table 1
identifies specific reproductive and developmental effects of concern, listing whether an effect
has been observed in test animals and epidemiologic cohorts. It also identifies the EDi0
estimated by the U.S. EPA (2003) for health effects observed in rodent bioassays. If the U.S.
EPA did not report an EDi0 for an effect, the table identifies a study where the effect was
reported and the lowest study dose where the effect was observed. Table 1 also identifies the
epidemiologic cohort where the specific reproductive and developmental effects were observed.

Epidemiological Study Utility

The Panel reviewed the charge questions (Appendix B), discussed them, and made two
suggestions to the U.S. EPA:

•	Concerning the ability of epidemiological studies to inform critical effects, start with
concordance across species (including humans) for the spectrum of effects.

•	Concerning the ability of epidemiological studies to inform dose-response modeling, start
with the epidemiology and then go to animal data if the dose response has not been well
characterized for an endpoint of interest and compare to animal data as a reality check.

Animal Model Utility

The Panel reviewed and discussed the charge questions (Appendix B). Table 1, which
identifies the effects that occur in animals and also have relevance to humans, summarizes much
of this discussion. Regarding the influence of mode of action (MO A) on animal model choice,
the Panel concluded that by evaluating concordance among health effects reported in
epidemiologic and animal bioassay data, the U.S. EPA could identify a set of plausible
reproductive and developmental effects to consider. Actual animal and human MOA
information is helpful in that it creates comfort with the animal models and in defining the
boundaries of possible effects.

This document is a draft for review purposes only and does not constitute Agency policy.

A-17 DRAFT—DO NOT CITE OR QUOTE

-------




TABLE 1



Reproductive/Developmental Effects of Concern for Human Health

Endpoint

Rodent
(ED 10 ng/kg-d)

Human

Notes

Sperm Count/Motility

Yes (6.2-28;
66-200)

Yes

EDio bases Mabley et al. (1992a,b) caudal
sperm count and daily sperm production
range from 6.2-28; Gray et al. (1997)
epididymal sperm count and total testis sperm
counts range from 66-200.

Sex Ratio

No

Yes, Seveso



Delayed Puberty Males

Yes (94)

Yu-cheng

EDio basis rat male puberty delay Gray et al.
(1997). Need to qualify epidemiology data
because of cohort PCDD/PCDFs exposures.

Delayed Puberty in Females

Yes

No in Seveso

Gray and Ostby (2002) report delayed
puberty in female offspring of pregnant rats
receiving a single dose of 1 |ig TCDD/kg on
GD 15.

Cleft Palate

Yes(6300-6400)

No

EDio basis Birnbaum et al. (1989).

Premature Senescence

Yes

No, Seveso

Franczak et al. (2006) report that rats
prematurely entered reproductive senescence,
after receiving cumulative TCDD doses as
low as 1.7 jig TCDD/kg. They considered
first occurrence of prolonged interestrous
interval (>6 d) as evidence of onset of
reproductive senescence.

Hormones E2

Yes

Yes, Males—
Seveso

Li et al. (1995) report serum estradiol-17f^
(E2) concentrations induced by equine
Chorionic Gonadotropin injection were
significantly elevated in female rats orally
administered 10 (ig/kg TCDD onPND 22.
While E2 decreased dramatically in control
animals during the preovulatory LH surge, it
did not in TCDD-treated rats.

Low Birth Weight

Yes (190)

Suggestive
effect in Seveso
in first 8 years
after exposure

EDio basis Gray et al. (1997).

Reproductive Cycling
(prolongation)

Yes

Yes, Seveso
Prepubertal
exposure

Franczak et al. (2006) report loss of normal
cyclicity in female rats at 8 months of age
following a cumulative dose of 1.7 jig
TCDD/kg.

This document is a draft for review purposes only and does not constitute Agency policy.

A-18 DRAFT—DO NOT CITE OR QUOTE

-------
Supporting Information

The Panel reviewed the charge questions (Appendix B), discussed them, and made two
suggestions to the U.S. EPA:

•	Concerning deviation from default approaches for noncancer endpoints, there needs to be
a careful assessment of the POD and the application of uncertainty factors in light of
PK/pharmacodynamics (PD), population characteristics and variability, and MOA
information.

•	Concerning the MOA's ability to clarify endpoint and the incorporation of a cascade of
cellular event into dose-response for noncancer endpoint, any study that helps inform the
dose response should be considered—including studies not specific to dioxins.
Complicated mechanistic models need not be developed. Standard dose-response models
can be applied. One can look at the cascade of events in a stepwise, simple way.

References

Birnbaum, L.S., M.W. Harris, L.M. Stocking et al. 1989. Retinoic acid and 2,3,7,8-
tetrachlorodibenzo-p-dioxin selectively enhance teratogenesis in C57BL/6N mice. Toxicol.

Appl. Pharmacol. 98:487-500.

Franczak, A., A. Nynca, K.E. Valdez, K.M. Mizinga and B.K. Petroff 2006. Effects of acute
and chronic exposure to the aryl hydrocarbon receptor agonist 2,3,7,8-tetrachlorodibenzo-
p-dioxin on the transition to reproductive senescence in female Sprague-Dawley rats. Biol.
Reprod. 74:125-130.

Gray, L.E. and J.S. Ostby. 2002. In utero 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) alters
reproductive morphology and function in female rat offspring. Toxicol. Appl. Pharmacol.
133(2):285-294.

Gray, L.E., J.S. Ostby and W.R. Kelce. 1997. A dose-response analysis of the reproductive
effects of a single gestational dose of 2,3,7,8-tetrachlorodibenzo-p-dioxin in male Long Evans
Hooded rat offspring. Toxicol. Appl. Pharmacol. 146:11-20.

Li, X., D.C. Johnson and K.K. Rozman. 1995. Reproductive effects of 2,3,7,8-
tetrachlorodibenzo-p-dioxin (TCDD) in female rats: ovulation, hormonal regulation, and possible
mechanism(s). Toxicol. Appl. Pharmacol. 133:321-327.

Mably, T.A., D.L. Bjerke, R.W. Moore et al. 1992a. In utero and lactational exposure of male
rats to 2,3,7,8-tetrachlorodibenzo-p-dioxin. 3. Effects on spermatogenesis and reproductive
capability. Toxicol. Appl. Pharmacol. 114:118-126.

Mably, T.A., R.W. Moore, R.W. Goy et al. 1992b. In utero and lactational exposure of male
rats to 2,3,7,8-tetrachlorodibenzo-p-dioxin. 2. Effects on sexual behavior and the regulation of
luteinizing hormone secretion in adulthood. Toxicol. Appl. Pharmacol. 114:108-117.

This document is a draft for review purposes only and does not constitute Agency policy.

A-19 DRAFT—DO NOT CITE OR QUOTE

-------
NAS (National Academy of Sciences). 2006. Health Risks from Dioxin and Related
Compounds: Evaluation of the EPA Reassessment. National Academies Press, Washington, DC
(July). Available at http://www.nap.edu/catalog.php7record id=l 1688.

U.S. EPA (U.S. Environmental Protection Agency). 2003. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds. NAS
Review Draft (EPA/600/P-00/001Cb). U.S. Environmental Protection Agency, National Center
for Environmental Assessment, Washington, DC. Available at
http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/.

SESSION 5: QUANTITATIVE UNCERTAINTY ANALYSIS OF DOSE-RESPONSE

This session addressed the uncertainty analysis to be considered for the dose-response
assessments. The session opened with a presentation on current estimates of dioxin exposure
levels. Then it focused on the factors to include in the scope of an uncertainty analysis including
dioxin kinetics.

Session 5 Panelists (Session Co-Chairs are identified by asterisk)

•	Bruce Allen, Bruce Allen Consulting

•	Lesa Aylward, Summit Toxicology

•	Roger Cooke, Resources for the Future

•	Kenny Crump, Louisiana Tech University

•	MikeDeVito, U.S. EPA

•	Dale Hattis, Clark University

•	*Rick Hertzberg, Biomath Consulting

•	Nancy Kerkvliet, Oregon State University

•	Leonid Kopylev, U.S. EPA

•	Rob McDowell, U.S. Department of Agriculture

•	Lorenz Rhomberg, Gradient

•	Woody Setzer, U.S. EPA

•	Marie Sweeney, National Institute of Occupational Safety and Health

•	*Linda Teuschler, U.S. EPA

Please note that the use of the term "concluded" or "recommended" in this summary does not mean that a consensus
was reached. Session Summaries were written from the material prepared by the non-EPA/ANL Co-Chair and
represent a synopsis of the panel discussions.

The Panel summarized the NAS comments regarding uncertainty. Areas for improvement
include:

•	Ensure "transparency, thoroughness, and clarity in quantitative uncertainty analysis."

•	Describe and define (quantitatively to the extent possible) the variability and uncertainty
for key assumptions used for each key endpoint-specific risk assessment, including
choices of data set, point of departure, dose-response model, and dose metric.

•	Incorporate probabilistic models to represent the range of plausible values.

This document is a draft for review purposes only and does not constitute Agency policy.

A-20 DRAFT—DO NOT CITE OR QUOTE

-------
•	Assess goodness-of-fit of dose-response models.

•	Provide upper and lower bounds on central tendency estimates for all statistical estimates.

•	When quantification is not possible, clearly state it, and explain what would be required
to achieve quantification.

Identification of Important Uncertainties

The Panel reviewed the charge questions (Appendix B), discussed them, and listed eight
issues for consideration by the U.S. EPA:

•	Concerning species and strain differences in the U.S. EPA's Response to NAS, current
U.S. EPA procedures do not take this into account when selecting one data set for risk
assessment. Issues include "Where are humans in the distribution of potencies that can
be generated? How likely is it that human response is similar to the selected data? Can
we infer inter-individual variability from these differences?"

•	Concerning the use of animal data for cross species extrapolation to humans (PK and PD
uncertainties), issues to consider include differences in distribution and responses
following bolus doses from those of subchronic and chronic protocols; uncertainty in
liver doses due to sequestration; differences in receptor binding affinity among
congeners; and age factors (e.g., assumption of a lifetime constant daily dose for a cancer
extrapolation).

•	Concerning the description of AhR response, biochemical changes occur at lower doses
than toxicological changes. There should be an effort to identify the biochemical changes
that would mark Ah receptor binding to inform the BMR, and, thus, prevent toxicity.

•	Concerning model uncertainty, the mathematical model choice depends on endpoint.
There should be an effort towards determining what is the most sensitive endpoint(s) for
humans and conducting animal studies to model that endpoint(s).

•	Concerning exposure and dose response in human studies, ensure enough similarity to
current human exposure profiles (mixture composition) so that a dose-response
assessment can be done. Incorporate new epidemiological studies. Evaluate
concordance with animal data and consistency across studies. Panel-acknowledged
uncertainties include exposure estimates from person to person, shape of human dose-
response curve, healthy worker effect, and age dependence.

•	Concerning POD determination, uncertainty factors are inherently mathematically
inconsistent and that should be conveyed in the discussion of uncertainties when
interpreting the POD.

•	Concerning dose metric, tissue concentration is preferred. It should be evaluated against
a background of variability in AhR-binding expression. There is uncertainty in what
level of binding should be considered, in different cell types, tissues, life stage
(development). The relationship between dose metric and causation of adverse effects
should be examined.

This document is a draft for review purposes only and does not constitute Agency policy.

A-21 DRAFT—DO NOT CITE OR QUOTE

-------
Low-Dose Extrapolation

The Panel reviewed the charge questions and discussed them (Appendix B). The Panel
concluded that curve-fitting uncertainty (for a given dataset, dose metric, and model) can be
characterized and is useful, but, by itself, it is an incomplete characterization of uncertainty. The
Panel acknowledged the difficulty of fully characterizing uncertainty, especially quantitatively.
Some panelists argued that the problem is insurmountable and that no meaningful uncertainty
analysis is likely to be performable. Other panelists contended that, the difficulties
notwithstanding, "good-faith" efforts to do something practical and forthright to characterize
uncertainty in low-dose extrapolation would be useful and important. The Panel clarified "good
faith" as meaning a characterization that is useful and not misleading to decision makers and is
inclusive of approaches that have meaningful support in the scientific community as a whole.
Being in "good faith" is more important than being complete (i.e., addressing every uncertain
element), especially since completeness is not a realistic goal. From this discussion, the Panel
listed four issues for consideration by the U.S. EPA:

•	Review alternative data sets, dose metrics, and models to see where consequential
uncertainties and impacts on low-dose implications arise.

•	Consider the impacts of choices among plausible alternative data sets, dose metrics,
models, and other more qualitative choices—issues include how much difference the
choices make and also how much relative credence should be put to each alternative as a
way of gauging and describing the landscape of imperfect knowledge

regarding possibilities for the true dose-response.

¦	Hard to do quantitatively, since the factors are not readily expressed as statistical
distributions, but can describe the rationale for believing/doubting each alternative in
terms of available supporting evidence, contrary evidence, and needed assumptions.

¦	Expert judgment methods may be helpful in characterizing the relative weights of
scientific credibility among alternatives. The expert judgment process, when
conducted systematically, can be thought of as adding data to the assessment of
credibility of alternatives, rather than as just an opinion poll.

¦	Information on plausibility of alternative low-dose extrapolation approaches can
come from external considerations of mode of action, and not just from statistical
success at fitting particular (high-dose) data sets.

•	Characterizing uncertainty through a variety of approaches could be tried, and their
relative merits and shortcomings discussed, as a way forward.

•	Consider the sources of potential error, particularly in epidemiological data (e.g., TEF
uncertainty and variation in congener mixtures) and if possible quantify their impact on
the dose-response assessment.

Considerations for Conducting Uncertainty Analysis

Overall, the Panel was split on whether U.S. EPA should do quantitative uncertainty
analyses. The Panel noted that if done on only some of the uncertainties, then results would be
misleading and could be misused. Ultimately, the Panel listed seven issues for consideration by
the U.S. EPA:

This document is a draft for review purposes only and does not constitute Agency policy.

A-22 DRAFT—DO NOT CITE OR QUOTE

-------
•	The Panel recapped what some consider as being the first integrated risk assessment, with
structured expert judgment and uncertainty analysis, i.e., the Rasmussen Report
(WASH-1400; U.S. Nuclear Regulatory Commission, 1975). In their discussion of the
report, the Panel noted that in addition to standard event tree/fault tree modeling, this
report also tackled difficult model uncertainty issues involved in accident progression,
dispersion of released pollutants in the atmosphere, environmental transport, exposure,
health, and economic impacts. And though the Panel also recognized that this method
was no longer state-of-the-art, the Panel contended that it represents a good example of a
structured approach and methodology that could be built upon.

•	The Panel also discussed TEQs used in epidemiological studies, based on intake, and
recognized that the key uncertainty in what was measured was not just intake but also
involved PK/PD issues. The Panel acknowledged that the TEQ system is regularly used
on a concentration basis, but they expressed concern that the qualification becomes lost.
TEQs ignore pharmacokinetics and the common practice of rounding to orders of
magnitude introduces more error.

•	Structure the risk assessment along MOA steps—identify key biochemical measures
(-5-10) common across toxic endpoints and identify the degree of meaningful change in
effect or effect variance. Make a table with all options for data set, model, etc.; make
best estimates/choices and determine which of these choices matter the most to the
answer.

•	Use expert panels—expert judgment can be collected scientifically (procedures are
published). But there are known biases; central tendency estimates work much better
than extremes.

•	Use supporting studies to fill in critical data gaps—Info filling methods do exist (e.g., PK
modeling). Put short-term studies into the "supporting info" category (unless, of course,
the risk assessment is for acute exposures, such as chemical spills).

•	Be creative in the analysis of uncertainty. Intermediate steps between AhR binding and
the end processes can be hypothesized based on data, experiences, and analogies related
to other chemicals.

•	The 2003 Reassessment presented potency estimates on wide variety of
endpoints/models; needed to be more transparent in that discussion. Statistical graphics
can be used to convey uncertainties.

Reference

U.S. Nuclear Regulatory Commission. 1975. Reactor Safety Study: An Assessment of Accident
Risks in U.S. Commercial Nuclear Power Plants. WASH-1400 (NUREG-75-014). Washington,
DC.

This document is a draft for review purposes only and does not constitute Agency policy.

A-23 DRAFT—DO NOT CITE OR QUOTE

-------
APPENDIX A: 2009 U.S. EPA DIOXIN WORKSHOP AGENDA

SCIENTIFIC WORKSHOP
TO INFORM THE TECHNICAL WORK PLAN FOR U.S. EPA'S RESPONSE TO
NAS COMMENTS ON THE HEALTH EFFECTS OF DIOXIN
PRESENTED IN U.S. EPA'S DIOXIN REASSESSMENT

Cincinnati, OH

Date: February 18-20, 2009

BACKGROUND/WORKSHOP OBJECTIVE

At the request of the U.S. Environmental Protection Agency (U.S. EPA), the National
Academy of Sciences (NAS) prepared a report, Health Risks from Dioxin and Related
Compounds: Evaluation of the EPA Reassessment (NAS, 2006), that made a number of
recommendations to improve the U.S. EPA's risk assessment for 2,3,7,8-tetrachlorodibenzo-
/;-dioxin (TCDD). In response, the U.S. EPA will prepare a technical report that addresses key
comments on the dose-response assessment for TCDD. The U.S. EPA intends to develop its
response through a transparent process that provides multiple opportunities for input.

To assist in this effort, a Workshop will be held to inform the U.S. EPA's evaluation of
the NAS recommendations. The Workshop will be open to the public. At the Workshop, the
U.S. EPA will solicit input from expert scientists and the public.

The goal of the Workshop is to ensure that the U.S. EPA's response to the NAS
comments focuses on the key issues and reflects the most meaningful science. The three main
objectives of the Workshop are to (1) identify and discuss the technical challenges involved in
addressing the NAS key comments on the TCDD dose-response assessment in the U.S. EPA
Reassessment (U.S. EPA, 2003), (2) discuss approaches for addressing these comments, and
(3) identify key published, independently peer-reviewed literature, particularly studies describing
epidemiologic and in vivo mammalian bioassays, which are expected to be most useful for
informing the U.S. EPA response.

Workshop participants will be encouraged to think broadly about the body of scientific
information that can be used to inform the U.S. EPA's response and to participate in open
dialogue regarding ways in which the science can best be used to address the key dose-response
issues. This Workshop is similar to scientific workshops being conducted under the new review
process for the National Ambient Air Quality Standards (NAAQS)1 that assess health-related
information for criteria pollutants.

1 Please see http://www.epa.gov/ttn/naaas/ for more information on the new NAAQS review process.

This document is a draft for review purposes only and does not constitute Agency policy.

A-24 DRAFT—DO NOT CITE OR QUOTE

-------
The Workshop discussions are expected to build upon two prior publications:

1.	Exposure and Human Health Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin
(TCDD) and Related Compounds (U.S. EPA, 2003). This external review draft
provides a comprehensive reassessment of dioxin exposure and human health effects.
This "dioxin reassessment" was submitted in October 2004 to the National Academy
of Sciences (NAS) for review.

2.	Health Risks from Dioxin and Related Compounds: Evaluation of the EPA
Reassessment (NAS, 2006).

Workshop participants are encouraged to review both of these documents and other
relevant materials (e.g., the National Toxicology Program report on TCDD [NTP, 2006]) before
the meeting because they provide important insights into the key questions and challenges.

There are a number of open comment periods that are intended to facilitate a broad discussion of
the issues.

Scientists with significant expertise and experience relevant to the health effects of
TCDD or dioxin-like compounds and associated topics will be asked to serve on "expert panels"
for discussions throughout the Workshop. Workshop panelists will include a wide range of
experts representing many scientific areas needed to assess TCDD dose-response (e.g.,
epidemiology, human and animal toxicology, nuclear receptor biology, dose-response modeling,
risk assessment, and uncertainty analysis). The Workshop panelists will be asked to highlight
significant and emerging research and to make recommendations to the U.S. EPA regarding the
design and scope of the technical response to NAS comments on the dose-response analysis for
TCDD—including, but not limited to, recommendations for evaluating associated uncertainty.
Open comment periods will follow each panel discussion session. Public participation will be
encouraged by way of these designated open comment periods and, also, by participation in the
scientific poster session planned for the second evening (February 19).

U.S. EPA will use the input received during this Workshop as the foundation for its
development of a technical work plan for responding to the NAS comments on the TCDD dose-
response analysis. The work plan will outline the schedule, process, and approaches for
evaluating the relevant scientific information and addressing the key issues. The work plan also
will identify the key literature to be utilized in U.S. EPA's response.

As a follow-on activity to this Workshop, a panel is being established under the Federal
Advisory Committee Act (FACA) to guide and review the U.S. EPA's response to NAS
comments. The FACA panel will be asked to conduct a consultation with the Agency on the
draft technical work plan. At the same time, the public will also have the opportunity to provide
comments to the FACA panel on the work plan. The final technical work plan will guide the
development of the technical report that will constitute the U.S. EPA's response to NAS
comments. During the development of this response, the U.S. EPA will seek advice from the
FACA panel and the public several times. Finally, the FACA panel will be asked to review the
technical report in a public forum.

The preliminary Agenda presented on the following pages may be revised prior to the
Workshop following review by the session Co-Chairs; the dates and general timing of the

This document is a draft for review purposes only and does not constitute Agency policy.

A-25 DRAFT—DO NOT CITE OR QUOTE

-------
sessions, however, will not change. A final Agenda and a set of charge questions, intended to
provide general direction for the Workshop discussions, will be posted on the Workshop Internet
site (http://cfpub.epa.gov/ncea/cfrn/recordisplav.cfm?deid=199923) prior to the meeting.

A poster session will be held on the evening of the second day (February 19). The
purpose of this poster session is to provide a forum for scientists to present recent studies
relevant to TCDD dose-response assessment and to encourage open discussion about these
presentations.

REFERENCES

NAS (National Academy of Sciences). 2006. Health Risks from Dioxin and Related
Compounds: Evaluation of the EPA Reassessment. National Academies Press, Washington, DC
(July). Available at http://www.nap.edu/catalog.php7record id=l 1688.

NTP (National Toxicology Program). 2006. Toxicology and Carcinogenesis Studies of
2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) (CAS No. 1746-01-6) in Female Harlan Sprague-
Dawley Rats (Gavage Studies). U.S. Department of Health and Human Services. NTP TR 521.
Research Triangle Park, NC (April).

U.S. EPA (U.S. Environmental Protection Agency). 2003. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds, NAS
review draft, Volumes 1-3 (EPA/600/P-00/001Cb, Volume 1). U.S. Environmental Protection
Agency, National Center for Environmental Assessment, Washington, DC (December).
Available at http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/.

This document is a draft for review purposes only and does not constitute Agency policy.

A-26 DRAFT—DO NOT CITE OR QUOTE

-------
WORKSHOP AGENDA

Day 1

8:00-9:00	Registration

9:00-9:30	Welcome/Purpose of Meeting/Document Development Process

9:30-9:45	Panel Comments/Questions on Charge

9:45-2:45	Session 1: Quantitative Dose-Response Modeling Issues

(Hall of Mirrors)

9:45-10:10 Background/Introductory Remarks

10:10-10:35 TCDD Kinetics: Converting Administered Doses in Animals to
Human Body Burdens

Presenter: Michael Devito

10:35-11:30	Panel Discussion

11:30-1:00	Lunch

1:00-2:00	Panel Discussion cont.

2:00-2:45	Open Comment Period

2:45-3:05	Break

3:05-5:15	Session 2: Immunotoxicitv (Hall of Mirrors)

3:05-3:15	Background/Introductory Remarks

3:15-4:45	Panel Discussion

4:45-5:15	Open Comment Period

This document is a draft for review purposes only and does not constitute Agency policy.

A-27 DRAFT—DO NOT CITE OR QUOTE

-------
Day 2

8:00-8:30	Report-Outs for Sessions 1 and 2 (Hall of Mirrors)

8:00-8:15 Report-Out for 1: Quantitative Dose-Response Modeling Issues
8:15-8:30 Report-Out for 2: Immunotoxicity

8:30-11:30 Sessions 3A and 3B (concurrent sessions)

8:30-11:30	Session 3A: Dose-Response for Neurotoxicity and

Nonreproductive Endocrine Effects (Hall of Mirrors)

8:30-8:45 Background/Introductory Remarks

8:45-11:00 Panel Discussion
11:00-11:30 Open Comment Period

8:30-11:30	Session 3B: Dose-Response for Cardiovascular Toxicity and

Hepatotoxicity (Rookwood Room)

8:30-8:45 Background/Introductory Remarks

8:45-11:00 Panel Discussion

11:00-11:30 Open Comment Period

11:30-1:00 Lunch

1:00-2:00	Report-Outs for Sessions 3A and 3B (Hall of Mirrors)

The structure of the session report-outs will include the following:

Summary of session presentation including minority opinion

¦	Public comments

¦	Discussion

1:00-l: 15 Report-Out for 3A: Dose-Response for Neurotoxicity and
Nonreproductive Endocrine Effects

1:15-1:30 Open Comment Period

This document is a draft for review purposes only and does not constitute Agency policy.

A-28 DRAFT—DO NOT CITE OR QUOTE

-------
1:30—1:45 Report-Out for 3B: Dose-Response for Cardiovascular Toxicity and
Hepatotoxicity

1:45-2:00 Open Comment Period

2:00-5:15	Sessions 4A and 4B (concurrent sessions)

2:00-5:15	Session 4A: Dose-Response for Cancer (Hall of Mirrors)

2:00-2:15	Background/Introductory Remarks

2:15-4:45	Panel Discussion

4:45-5:15	Open Comment Period

2:00-5:15	Session 4B: Dose-Response for

Reproductive/Developmental Toxicity (Rookwood Room)

2:00-2:15	Background/Introductory Remarks

2:15-4:45	Panel Discussion

4:45-5:15	Open Comment Period

6:45-8:15	Poster Session (Rosewood Room)

Day 3

8:30-9:30	Report-Outs for Sessions 4A and 4B (Hall of Mirrors)

8:30-8:45 Report-Out for 4A: Dose-Response for Cancer
8:45-9:00 Open Comment Period

9:00-9:15 Report-Out for 4B: Dose-Response for Reproductive/Developmental
Toxicity

9:15-9:30 Open Comment Period

This document is a draft for review purposes only and does not constitute Agency policy.

A-29 DRAFT—DO NOT CITE OR QUOTE

-------
9:30-3:30	Session 5: Quantitative Uncertainty Analysis of Dose-

Response (Hall of Mirrors)

9:30-9:40 Background/Introductory Remarks

9:40-10:10 Evidence of a Decline in Background Dioxin Exposures in Americans
Between the 1990s and 2000s

Presenter: Matt Lorber

10:10-10:30	Break

10:30-11:30	Panel Discussion

11:30-1:00	Lunch

1:00-2:15	Panel Discussion cont.

2:15-2:30	Break

2:30-3:00	Open Comment Period

3:00-3:15	Report-Out for 5: Quantitative Uncertainty Analysis of Dose-
Response

3:15-3:30 Closing Remarks
3:30	Adjourn

This document is a draft for review purposes only and does not constitute Agency policy.

A-30 DRAFT—DO NOT CITE OR QUOTE

-------
APPENDIX B: 2009 U.S. EPA DIOXIN WORKSHOP
QUESTIONS TO GUIDE PANEL DISCUSSIONS

SESSION 1
Dose Metric

Considering all of the endpoints or target tissues, and species that U.S. Environmental Protection
Agency (U.S. EPA)'s dose-response modeling might evaluate, what are the best measures of
dose (e.g., ingested, tissue concentrations, body burden, receptor occupancy, other surrogate) and
why?

Developing Dose-Response Models from Mammalian Bioassays

How best can the point of departure (POD) be determined when the response range is
incompletely characterized (i.e., high response at the lowest dose or low response at the highest
dose; observed in several key 2,3,7,8-Tetrachlorodibenzo-p-Dioxin [TCDD] studies)?

If considered to be biologically plausible, how can a threshold be incorporated into a dose-
response function (e.g., for TCDD cancer data)?

How can nonmonotonic responses be incorporated into the dose-response function?

Developing Dose-Response Models from Epidemiological Studies

How can the epidemiological data be utilized best to inform the TCDD exposure-response
modeling? Which epidemiological studies are most relevant?

Supporting Information

For those toxicological endpoints that are Ah receptor-mediated, how would the receptor kinetics
influence the shape of the dose-response curve? How would downstream cellular events affect
the shape of the dose-response curve? How can this cascade of cellular events be incorporated
into a quantitative model of dose-response?

This document is a draft for review purposes only and does not constitute Agency policy.

A-31 DRAFT—DO NOT CITE OR QUOTE

-------
SESSIONS 2, 3A, 3B, 4A, AND 4B

Key Study Selection

For this endpoint, what refinements should be made to the draft criteria for selection of key
studies?

What are the specific effects of concern for human health for this endpoint?

Based on the draft criteria for the selection of key studies, what are the key studies informing the
shape of the dose-response curve above the POD and the choice of the POD for this endpoint?

Epidemiological Study Utility

How and to what extent do the epidemiological data inform the choice of critical effect?

How can the epidemiological data inform the quantitative dose-response modeling?

Animal Model Utility

Are there types of effects observed in animal models that are more relevant to humans than
others? To what extent does information on mode of action (MOA) influence the choice of
animal model (species, strain, sex)?

Supporting Information

Are there studies that establish a sufficient justification for departure from the default procedures
that address the shape of the dose-response curve below the POD under the cancer guidelines?

Are there studies that establish a sufficient justification for departing from U.S. EPA's default
approaches for noncancer endpoints?

To what extent can MO A information clarify the identification of endpoints of concern and dose-
response metric for this endpoint? How can the cascade of cellular events for this endpoint be
incorporated into a quantitative model of dose response?

This document is a draft for review purposes only and does not constitute Agency policy.

A-32 DRAFT—DO NOT CITE OR QUOTE

-------
SESSION 5

For cancer and noncancer TCDD dose-response assessments, U.S. EPA is interested in
developing a quantitative uncertainty analysis addressing both parameter and model uncertainty,
if feasible. Uncertainties will include, among others, choice of endpoint; underlying study
uncertainties; choice of dose metric; interspecies extrapolations such as kinetic uncertainties; and
choice of dose-response model, including threshold models. The U.S. EPA is currently
examining techniques and tools for uncertainty analysis—including Bayesian and frequentist
approaches.

Identification of Important Uncertainties

What are the major uncertainties pertaining to modeling the animal data?

Consider the dose metric (species or tissue specificity), vehicle of administration,
exposure frequency, exposure duration, and POD determination (e.g., benchmark
response selection or no-observed-adverse-effect level/lowest-observed-adverse-effect
level identification).

What are the major uncertainties pertaining to dose-response modeling below the POD?

Consider how receptor kinetics and downstream cellular event information might be used
to bound the uncertainties associated with dose-response modeling below the POD.

What are the major uncertainties in cross-species extrapolation (e.g., half-lives, tissue
distribution, and toxicodynamics)?

Consider the primary species dosed with TCDD: mice, hamsters, rats, guinea pigs, and
monkeys.

What are the major uncertainties pertaining to intrahuman variability?

Consider what data sets would be useful to represent sensitive subpopulations.

What are other significant sources of uncertainty for the cancer and noncancer assessments?
Considerations for Conducting Uncertainty Analysis

What data sets could be used to quantify uncertainties in cancer and noncancer TCDD dose-
response assessments?

Consider dioxin-like compound dose-response data.

Consider MOA information.

What are the appropriate techniques for the TCDD dose-response uncertainty analysis, and what
are their respective strengths and weaknesses of these approaches as applied to TCDD?

This document is a draft for review purposes only and does not constitute Agency policy.

A-33 DRAFT—DO NOT CITE OR QUOTE

-------
APPENDIX C: 2009 U.S. EPA DIOXIN WORKSHOP DRAFT SELECTION CRITERIA TO IDENTIFY KEY IN
VIVO MAMMALIAN STUDIES THAT INFORM DOSE-RESPONSE MODELING FOR
2,3,7,8-TETRACHLORODIBENZO-/>-DIOXIN (TCDD)a

Study Feature

Selection Rationale

Primaryb

Secondaryc

Currently Excluded

Chemical, purity,
matrix/medium

TCDD-only doses included, purity specified,
matrix in which TCDD is administered is identified

TCDD purity or matrix not clearly identified

Studies of dioxin-like compounds
(DLCs) or mixtures

Peer review

Independently peer-reviewed, publicly available

Supplementary materials accompanying
peer-reviewed publication

Not formally peer-reviewed; literature
not publicly available

Study design,
execution, and
reporting

Clearly documented and consistent with standard
toxicological principles, testing protocols,
and practice (i.e., endpoint-appropriate,
particularly for negative findings)

Testing protocol provides incomplete
coverage of relevant endpoint-specific
measures, particularly for negative findings

Studies not meeting standard
principles and practices

Study subject:
species, strain, and
sensitivity for given
endpoint; litter; life
stage; gender

Mammalian species

Strain and gender identified

Animal age at beginning of treatment identified

Litter confounders (within/between) accounted for

Mammalian species, in vivo, but only
studying an artificially sensitive subject
(e.g., knockout mouse)

Non-mammalian or not in vivo

Exposure route

Oral

Parenteral (e.g., intravenous, intramuscular,
intraperitoneal, subcutaneous)

Inhalation, dermal, ocular

Dose level

Lowest dose <200 ng/kg-d for noncancer
endpoints and <1 pg/kg-d for cancer

Lowest dose >200 ng/kg-d for noncancer
endpoints, or >1.0 [jg/kg-d for cancer



Exposure frequency,
duration, and timing

Dosing regimen characterized and explained



Characterization/explanation
missing or cannot be determined

Controls

Appropriate and well characterized

Effect reported, but with no negative control



Response

Effect relevant to human health
Magnitude outside range of normal variability

Precursor effects, or adaptive responses
potentially relevant to human health

Lethality

Statistical evaluation

Clearly described and appropriate to the endpoint
and study design (e.g., per error variance,
magnitude of effect)

Limited statistical context



a NAS (2006) commented that the selection of data sets for quantitative dose-response modeling needed to be more transparent. These draft criteria are
offered for consideration at the kickoff workshop. These criteria would be used to identify candidate studies of non-human mammals that would be used to
define the point-of-departure (POD). These criteria are not designed for hazard identification or weight-of-evidence determinations. Studies addressing data
other than direct TCDD dose-response in mammals (including toxicokinetic data on absorption, distribution, metabolism, or elimination; information on
physiologically-based pharmacokinetic [PBPK] modeling, and mode of action data) will be evaluated separately.
b Presents preliminary draft criteria for evaluating a study being considered for estimating a POD in a TCDD dose-response model.

c Presents preliminary draft criteria that could qualify a study as primary with support from other lines of evidence (e.g., PBPK modeling), when no study for an
endpoint meets the "primary" criteria.

-------
[This page intentionally left blank.]

-------
DRAFT

DO NOT CITE OR QUOTE

May 2010
External Review Draft

APPENDIX B

Evaluation of Cancer and Noncancer
Epidemiological Studies for Inclusion in
TCDD Dose-Response Assessment

NOTICE

THIS DOCUMENT IS AN EXTERNAL REVIEW DRAFT. It has not been formally released
by the U.S. Environmental Protection Agency and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its technical accuracy and policy
implications.

National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH

-------
CONTENTS—Appendix B: Evaluation of Cancer and Noncancer Epidemiological Studies
for Inclusion in TCDD Dose-Response Assessment

LIST OF TABLES	B-iii

APPENDIX B. EVALUATION OF CANCER AND NONCANCER
EPIDEMIOLOGICAL STUDIES FOR INCLUSION IN

TCDD DOSE-RESPONSE ASSESSMENT	B-l

B.l. EVALUATION 01 CANCER STUDIES	B-l

B.l.l. NIOSH Cohort Studies	B-l

B.l.2. BASF Cohort Studies	B-8

B.l.3. The Hamburg Cohort	B-ll

B.l.4. The Seveso Cohort Studies	B-16

B.l.5. The Chapaevsk Study	B-22

B. 1.6. The Air Force Health ("Ranch Hands") Study	B-23

B. 1.7. Other Studies of Potential Relevance to Dose-Response Modeling	B-26

B.2. EVALUATION OF NONCANCER STUDIES	B-31

B.2.1. NIOSH Cohort	B-3 1

B.2.2. BASF Cohort	B-33

B.2.3. Hamburg Cohort	B-35

B.2.4. The Seveso Women's Health Study	B-36

B.2.5. Other Seveso Noncancer Studies	B-45

B.2.6. Chapaevsk Study	B-55

B.2.7. Air Force Health ("Ranch Hands") Study	B-56

B.2.8. Other Noncancer Studies of Dioxin	B-57

B.3. REFERENCES	B-61

This document is a draft for review purposes only and does not constitute Agency policy.

B-ii DRAFT—DO NOT CITE OR QUOTE

-------
LIST OF TABLES

B-l. Fingerhut et al., 1991—All cancer sites, site-specific analysis	B-l

B-2. Steenland et al., 1999—All cancer sites combined, site-specific analysis	B-2

B-3. Steenland et al., 2001—All cancer sites combined	B-4

B-4. Cheng et al., 2006—All cancer sites combined	B-5

B-5. Collins et al., 2009—All cancer sites combined, site-specific analysis	B-6

B-6. Zober et al., 1990—All cancer sites combined, site-specific analysis	B-8

B-7. Ott and Zober, 1996—All cancer sites combined	B-9

B-8. Manz et al., 1991—All cancer sites combined, site-specific analyses	B-l 1

B-9.	Flesch-Janys et al., 1995; Flesch-Janys et al., 1996 erratum—All cancer sites

combined	B-12

B-10. Flesch-Janys et al., 1998—All cancer sites combined, site-specific analysis	B-13

B-ll. Becher et al., 1998—All cancer sites combined	B-15

B-12. Bertazzi et al., 2001—All cancer sites combined, site-specific analyses	B-16

B-13. Pesatori et al., 2003—All cancer sites combined, site-specific analyses	B-17

B-14. Consonni et al., 2008—All cancer sites combined, site-specific analyses	B-18

B-15. Baccarelli et al., 2006—Site-specific analysis	B-20

B-16. Warner et al., 2002—Breast cancer incidence	B-21

B-17. Revich et al., 2001—All cancer sites combined, and site-specific analyses	B-22

B-18. Akhtar et al., 2004—All cancer sites combined and site-specific analyses	B-23

B-l9. Michalek and Pavuk, 2008—All cancer sites combined	B-25

B-20. 't Mannetje et al., 2005—All cancer sites combined, site specific analyses	B-26

B-21. McBride et al., 2009b—All cancer sites combined, site-specific analysis	B-27

B-22. McBride et al., 2009a—All cancer sites combined, site-specific analysis	B-28

B-23. Hooiveld et al., 1998—All cancer sites combined, site-specific analysis	B-29

B-24. Steenland et al., 1999—Mortality (noncancer)	B-31

B-25. Collins et al., 2009—Mortality (noncancer)	B-32

This document is a draft for review purposes only and does not constitute Agency policy.

B-iii DRAFT—DO NOT CITE OR QUOTE

-------
LIST OF TABLES (continued)

B-26.	Ott and Zober, 1996—Mortality (noncancer)	B-33

B-27.	Flesch-Janys et al., 1995; Flesch-Janys et al., 1996 erratum—Mortality

(noncancer)	B-35

B-28.	Eskenazi et al., 2002a—Menstrual cycle characteristics	B-36

B-29.	Eskenazi et al., 2002b—Endometriosis	B-38

B-30.	Eskenazi et al., 2003—Birth outcomes	B-39

B-31.	Warner et al., 2004—Age at menarche	B-40

B-32.	Eskenazi et al., 2005—Age at menopause	B-41

B-33.	Warner et al., 2007—Ovarian function	B-43

B-34.	Eskenazi et al., 2007—Uterine leiomyoma	B-44

B-35.	Mocarelli et al., 2008—Semen quality	B-45

B-36.	Mocarelli et al., 2000—Sex ratio	B-46

B-37.	Baccarelli et al., 2008—Neonatal thyroid function	B-48

B-38.	Alaluusua et al., 2004—Oral hygiene	B-49

B-39.	Bertazzi et al., 2001—Mortality (noncancer)	B-50

B-40.	Consonni et al., 2008—Mortality (noncancer)	B-52

B-41.	Baccarelli et al., 2005—Chloracne	B-53

B-42.	Baccarelli et al, 2002 and 2004—Immunological effects	B-54

B-43.	Revich et al., 2001—Mortality (noncancer) and reproductive health	B-55

B-44.	Michalek and Pavuk, 2008—Diabetes	B-56

B-45.	McBride et al., 2009a—Mortality (noncancer)	B-57

B-46.	McBride et al., 2009b—Mortality (noncancer)	B-59

B-47.	Ryan et al., 2002—Sex ratio	B-60

This document is a draft for review purposes only and does not constitute Agency policy.

B-iv DRAFT—DO NOT CITE OR QUOTE

-------
1	APPENDIX B. EVALUATION OF CANCER AND NONCANCER

2	EPIDEMIOLOGICAL STUDIES FOR INCLUSION IN TCDD

3	DOSE-RESPONSE ASSESSMENT

4

5

6	B.l. EVALUATION OF CANCER STUDIES

7	B.1.1. NIOSH Cohort Studies

8

9	Table B-l. Fingerhut et al., 1991—All cancer sites, site-specific analysis

10

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. The data sources to ascertain vital status and cause of death
information were the Social Security death files, the National Death Index, and the Internal
Revenue Service. Vital status could be determined for 98% of the cohort.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. While the authors provide compelling arguments that suggest
risks are not unduly biased by lack of cigarette smoking data, they acknowledge potential
biases that could exist for other occupational exposure (e.g., asbestos) for which data were
lacking.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. There was not a statistically significant linear trend of increasing
mortality with increased duration of exposure.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration not satisfied. This study used duration of exposure, at an individual level, as a
surrogate measure of TCDD. Duration of exposure determined by number of years workers
were involved in processes involving TCDD contamination. Exposure was determined by
reviewing, at each plant, operating conditions, job duties, records of TCDD levels in industrial
hygiene samples, intermediate reactants, products, and wastes. Exposure assessment was
limited and the uncertainty related to exposure measures not fully addressed.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. This is the largest of the occupational cohorts that has been exposed
to TCDD. The cohort consisted of 5,172 workers and a total of 265 cancer deaths. Site-
specific mortality analyses, including soft tissue sarcoma (n = 4), was limited by small
numbers.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

This document is a draft for review purposes only and does not constitute Agency policy.

B-1 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. New England Journal of Medicine, 1991;324:212-218. Authors address
the possibility of bias from lack of control for potential confounders such as smoking and other
occupational exposures. They address limitations of using death certificates for identifying
certain causes of deaths, and limitations of using duration of employment as an exposure
metric.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. Since this study used duration of exposure as the exposure metric,
dose-response relationships cannot be quantified.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported
dose-is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. Models incorporated period of latency, and a surrogate measure of
cumulative TCDD exposure was modeled. The follow-up interval was sufficiently long
(1942-1987).





Conclusion

Overall, quantitative exposure data are lacking on an individual-level basis. Further
dose-response analysis should consider updated data for this cohort that includes serum-based
measures of TCDD, in addition to an extension of the follow-up period. Given these
limitations, this study is not further evaluated for TCDD dose-response assessment.

Table B-2. Steenland et al., 1999—All cancer sites combined, site-specific
analysis

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. The study evaluated mortality from all cancer sites (combined). As
described in the paper, the sources of vital status and cause of death information were received
from the Social Security death files, the National Death Index, and the Internal Revenue
Service. Vital status was known for 99.4% of the cohort members, cause of death information
is available for 98% of the decedents.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Occupational exposure to asbestos and 4-aminobiphenyl contributed
to some excess cancer, but no evidence of confounding for the relationship between TCDD
and all cancer mortality was detected following removal of workers who died of bladder
cancer. No information is available for cigarette smoking, although dose-response patterns
were stronger for nonsmoking related cancers. This finding suggests that smoking is not
responsible for excess cancer risk that was observed in the cohort.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

This document is a draft for review purposes only and does not constitute Agency policy.

B-2 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration satisfied. When a 15-year lag interval was incorporated into the exposure
metric a statistically significant dose-response pattern was observed for all cancer sites
combined with both a continuous measure of TCDD (p = 0.05) as well as one that was
log-transformed (p < 0.001).

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. The study conducted detailed sensitivity analyses and evaluated
different assumptions regarding latency, log-transformed TCDD exposures, and half-life
values for TCDD.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. This is the largest of the occupational cohorts with exposures to
TCDD. The cohort consisted of 5,132 male workers and a total of 377 cancer deaths. This
permits characterization of risk for all cancer sites (combined).





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Journal of the National Cancer Institute, 1999; 91(9):779-786. The authors
discussed the potential for bias from smoking, and other occupational exposures for which
data for both were lacking at an individual basis.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Exposure scores assigned on an individual level using a job-exposure
matrix. The job-exposure matrix was based on estimated factor of contact with TCDD in each
job, level of TCCD contamination of materials at each plant over time, and proportion of day
worker could be in contact with materials. These factors were multiplied together to derive a
daily exposure score, which was accumulated over the working history of each worker to
obtain a cumulative measure of TCDD.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. The follow-up of the cohort extended from 1942 until the end of 1993.
Greater than 25 years of follow-up have accrued in cohort allowing for latency to be
examined. Different assumptions on the half-life of TCDD were evaluated and produced
similar results. Latency intervals were incorporated, with strongest associations noted with an
interval of 15 years.





Conclusion

This study meets the criteria and considerations noted above but has been superseded and
updated by Steenland et al. (2001). Therefore, this study was not considered for further
dose-response analyses.

1

2

3

4

5

This document is a draft for review purposes only and does not constitute Agency policy.

B-3 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table B-3. Steenland et al., 2001—All cancer sites combined

2

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. The study evaluated mortality from all cancer sites (combined). As
described by Steenland et al., (1999) the sources of vital status and cause of death information
were received from the Social Security death files, the National Death Index, and the Internal
Revenue Service. Vital status was known for 99.4% of the cohort members, cause of death
information is available for 98% of the decedents.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Occupational exposure to asbestos and 4-aminobiphenyl contributed
to some excess cancer, but no evidence of confounding for the relationship between TCDD
and all cancer mortality was detected following removal of workers who died of bladder
cancer. No information is available for cigarette smoking, although dose-response patterns
were similar between smoking and nonsmoking related cancers.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Increased risk estimates were observed in the higher cumulative
exposure categories. The dose-response curve was not linear at higher doses.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied.

Exposure metrics considered included cumulative TCDD, loglOTCDD, average exposure, and
a cubic spline model was also evaluated. Exposure response relationships were also evaluated
using TEQs. Exposure scores were assigned on an individual level using a job-exposure
matrix. The job-exposure matrix was based on estimated factor of contact with TCDD in each
job, level of TCCD contamination of materials at each plant over time, and proportion of day
worker could be in contact with materials. Serum levels were measured in 199 workers at one
of 8 plants in 1998. Different estimate of the half-life of TCDD were used, and similar results
were produced. The paper presented a range in risk estimates thereby conveying the range of
uncertainties in risk estimates derived using different measures of exposure.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. This is the largest of the occupational cohorts with exposures to
TCDD. The cohort consisted of 3,538 male workers and a total of 256 cancer deaths.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied Am JEpidem, 2001, 154(5)451-458. However, additional details to assess
uncertainties associated with characterizing serum data in a subset of workers to remainder of
cohort are lacking.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

This document is a draft for review purposes only and does not constitute Agency policy.

B-4 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. The metrics considered included cumulative TCDD, loglOTCDD, average
exposure, and a cubic spline model was also evaluated. Exposure response relationships were
also evaluated using TEQs. Serum lipid TCDD measurements from 170 workers whose
TCDD levels were greater than 10 ppt (the upper ranges of a background level) were used
along with JEM information, work histories, and a pharmacokinetic elimination model to
estimate dose rates per unit exposure score. In this regression model, the estimated TCDD
level at the time of last exposure was modeled as a function of exposure scores. The
coefficient relating serum levels and exposure scores was then used to estimate serum TCDD
levels over time from occupational exposure (minus the background level) for all
3,538 workers. Time-specific serum levels were then integrated over time to derive a
cumulative serum lipid concentration due to occupational exposure for each worker.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. Greater than 25 years of follow-up have accrued in cohort allowing for
latency to be examined. Different assumptions on the half-life of TCDD were evaluated
producing similar results.





Conclusion

Overall, criteria have been satisfied. This study was modeled in the 2003 Reassessment and is
considered for further dose-response evaluations herein.

Table B-4. Cheng et al., 2006—All cancer sites combined

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. The study evaluated cancer mortality. The vital status and the
information regarding the cause of death were extracted from the Social Security death files,
the National Death Index, and the Internal Revenue Service (Steenland et al., 1999). Vital
status was known for 99.4% of the cohort members, while cause of death information is
available for 98% of the decedents.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. This is the same data set used in the Steenland et al., (2001) paper.
Occupational exposure to asbestos and 4-aminobiphenyl contributed to some excess cancer,
but no evidence of confounding for the relationship between TCDD and all cancer mortality
was detected following removal of workers who died of bladder cancer. No information is
available for cigarette smoking, although dose-response patterns were similar between
smoking and nonsmoking related cancers.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Slope coefficients are available for all cancers combined under a
varying set of assumptions. Little evidence of an association was found when lag interval was
not taken into account. Associations strengthened with incorporation of a 10 to 15 year lag
interval. Dose-response was nonlinear at higher exposures, suggesting a nonlinear
relationship or increased exposure misclassification at higher levels.

This document is a draft for review purposes only and does not constitute Agency policy.

B-5 DRAFT—DO NOT CITE OR QUOTE

-------
4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Compared to the 1st order models, the concentration, and age
dependent model (CADM) provided a better fit for the serum sampling data. CADM model
exposure estimates are higher than those based on an age only, constant 8.7-year half-life
model. As discussed by Aylward et al. (2005b), model exposure estimates are influenced not
only by choice of elimination model, but also by choices in regression procedure (e.g., log
transformation, use of intercept, and incorporation of background dose term). Other
limitations or uncertainties in exposure assessment include the following

•	Job-exposure matrix based on limited sampling data, and subjective judgment on contact
times and factors

•	Inability to take into account interindividual variability in TCDD elimination kinetics

•	Dose-rate regressions are based on a small sample of the cohort with serum measures;
therefore, regression results may not be representative of remainder of the cohort.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. Largest cohort of TCDD exposed workers. The risk estimates are
based on a total of 256 cancer deaths.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Risk Analysis, 2006; 4:1,059-1,071. Additional details to assess
uncertainties associated with characterizing serum data can be found in Aylward et al.
(2005b); Risk Anal. 25(4):945-956.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Cumulative serum lipid concentrations were estimated for each worker. No
other dioxin-like compounds were assessed in this analysis.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. Concentration and age-dependence of TCDD elimination and
two compartments (hepatic and adipose tissue) were taken into account when estimating
TCDD exposures. Nearly 50 years of follow-up were available permitting an evaluation of
latency.





Conclusion

This study met the main criteria and considerations. The study is considered for further
dose-response analyses.

Table B-5. Collins et al., 2009—All cancer sites combined, site-specific
analysis

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

This document is a draft for review purposes only and does not constitute Agency policy.

B-6 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration satisfied. Vital status complete for all but two workers.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. No information collected on smoking status, but no excess in lung
cancer or nonmalignant respiratory diseases noted. Analyses took into account potential for
exposure to pentachlorophenol.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. No dose-response pattern was observed with all cancer sites
combined, however, a dose-response pattern was observed with soft tissue sarcoma. The study
found no association between TCDD and death from most types of cancer.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. The authors used these serum from 280 former TCP workers to
estimate historical exposure levels of TCDD, furans, and polychlorinated biphenyls for all
1,615 workers. Exposure assessment included detailed work history, industrial hygiene
monitoring, and the presence of chloracne cases among groups of workers. This data was
integrated into a 1-compartment, first-order pharmacokinetic to determine the average TCDD
dose associated with jobs in each group, after accounting for the presence of background
exposures estimated from the residual serum TCDD concentration in the sampled individuals.
The authors did not evaluate departures from linearity, or examine skewness at higher
exposures. Exposure levels were not provided.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. Largest study of workers employed in one center, and a total of
177 deaths from cancer were observed. Limited precision in the relative risk estimate was
noted for soft tissue sarcoma and TCDD exposures.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Published in Am J Epidemiol, 2009, 170(4):501-506. The authors discuss
limitations of using death certificates for identifying deaths from soft tissue sarcoma for which
a positive association was noted, assumptions in exposure characterization, and effects of
cigarette smoking.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. This study has the largest number of serum samples obtained from a specific
plant.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. Although specific analyses of latency were not reported, this cohort had a
sufficient Icnalh of follow-up for cnnccr mortality outcomes

This document is a draft for review purposes only and does not constitute Agency policy.

B-7 DRAFT—DO NOT CITE OR QUOTE

-------
The authors found a statistically significant dose-response trend for soft tissue sarcoma
mortality and TCDD exposures. The all-tumor results are not amenable to dose-response
analysis because they found no effect. Therefore, this study is considered for quantitative
dose-response analysis for the soft tissue sarcoma mortality results, only.

1

2	B.1.2. BASF Cohort Studies

3

4	Table B-6. Zober et al., 1990—All cancer sites combined, site-specific

5	analysis

6

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. A large component of the cohort (94 out of 247 workers) was
assembled by actively seeking out workers who were alive in 1986 through the "Dioxin
Investigation Programme." As a result, it is likely a number of deaths were missed due to the
recruitment of survivors. This underascertainment is supported by much lower all cancer
SMR one component of the cohort (SMR = 0.48, 95% CI: 0.13-1.23) relative to the general
population.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. See above discussion of underascertainment in mortality for some of
the cohort members. Although it is likely that other coexposures occurred (e.g., among
firefighters), confounding could only occur if these coexposures were associated with both the
endpoint and exposure (TCDD) being considered.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. Workers were not categorized on the basis of their exposure, but
rather their mortality experience compared to control cohort and the general population. The
design of the study does not allow for dose-response to be examined.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration not satisfied. Although years since first exposure was examined, exposure
assessment was based on working in various occupational cohorts. Since there was no
quantitative assignment of TCDD exposures, the associated uncertainties could not be
evaluated.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration not satisfied. There were only 23 cancer deaths in the entire cohort. As such,
this study lacked adequate statistical power to detect cancer mortality differences that were
moderate in magnitude.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

This document is a draft for review purposes only and does not constitute Agency policy.

B-8 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. Int Arch Occup Envir Health, 1990,62:139-157. The authors address
issues related to the healthy worker effect, multiple comparisons, smoking, and small size of
the cohort.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. Risks were derived by comparing mortality rates of the three cohort
subsets relative to a control cohort and the general population by time since first exposure
categories. Workers were not assigned exposures. There were no quantitative estimates of
TCDD exposure.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria not satisfied. While the study was able to indirectly look at variations in risk estimates
related to latency by using time since exposure, there were no quantitative estimates of TCDD
exposure.





Conclusion

This study is not suitable for dose-response analysis, as it failed the inclusion criteria. Most
notably, the lack of exposure data does not permit the use of these data for a dose-response
analysis.

Table B-7. Ott and Zober, 1996—All cancer sites combined

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Mortality ascertainment appeared to be fairly complete. The
ascertainment of cancer incidence is more difficult to judge as geographical area not covered
by a cancer registry.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Information was collected on smoking status, body mass index, and
other occupational exposures, however a large portion of the cohort was firefighters who may
have been exposed to other occupational carcinogens. However, the recruitment of survivors
may results in under-ascertainment of mortality.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Increased cancer incidence was observed in the highest TCDD
cumulative exposure category. Risks were most pronounced when a period of 20 years since
first exposure was incorporated into the model.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

This document is a draft for review purposes only and does not constitute Agency policy.

B-9 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration satisfied. Cumulative measure of TCDD expressed was derived from serum
measures. Exposure was also estimated by chloracne status of the cohort members. The
authors have not addressed the potential implication of deriving TCDD exposure estimates for
the whole cohort using sera data that were available for only about half of the cohort.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. For all cancer sites combined, there were 31 deaths. It is the smallest
of the occupational cohorts, but the deaths can be grouped into quartiles to allow for
evaluation of dose-response relationships.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Occupational and Environmental Medicine, 1996,53:606-612. A large
component of the cohort (94 out of 247 workers) was assembled by actively seeking out
workers who were alive in 1986 through the "Dioxin Investigation Programme." As a result,
it is likely a number of deaths were missed due to the recruitment of survivors. This
underascertainment is supported by much lower all cancer SMR one component of the cohort
(SMR = 0.48, 95% CI: 0.13-1.23) relative to the general population (Zober et al., 1990).

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum samples, taken in 1989, were available for 138 surviving workers out
of 254 and allowed for cumulative TCDD levels to be estimated using regression techniques in
the remainder of the cohort.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. Exposure assignment took into the affect that body mass index had on
TCDD half-lives. TCDD levels estimates through back-extrapolation of serum levels based on
half-life estimates obtained from previous studies. Latency was considered with stronger
association observed in external comparisons incorporating a latency of 20 years. The
follow-up of the cohort was lengthy (>50 years).





Conclusion

Given a part of the cohort was based solely on survivors in the in the mid-1980s, the SMR
statistic derived from this study underestimates excess mortality relative to the general
population. The cohort also includes some firefighters who are recognized to be exposed to
other carcinogenic agents—these exposures may be confounding the associations that were
reported. However, exposure to TCDD was quantified and the effective dose and oral
exposure estimable. Overall, criteria have been satisfied. This study was modeled in the 2003
Reassessment and is considered for further dose-response evaluations herein.

1

2

3

4

5

6

This document is a draft for review purposes only and does not constitute Agency policy.

B-10 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Deaths were identified through medical records of the cohort
members. A review of death certificates of the identified cancer deaths found a high degree of
concordance (51/54). One of the 136 noncancer death certificates examined indicated an
"occult" neoplasm.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Smoking data were similar between exposed and nonexposed cohort
based on independent samples. Occupational exposure for which individual data are lacking
unlikely to explain dose-response with TCDD.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Dose-response patterns across three levels of exposure observed
among those who started work before 1954, and among those who worked for 20 years or
longer. Dose-response patterns not evident across whole cohort, among those with less than
20 years of employment, or among those who started after 1954.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Categorical exposures were based on TCDD concentrations in
precursor materials, products, waste, and soil from the plant grounds, measured after the plant
closed in 1984. Exposure uncertainty examined using a separate group of 48 workers who
provided adipose tissue samples. Other surrogate measures of exposure were considered in
this study, including duration of exposure and year of first employment.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. For all cancer sites combined, there were 65 cancer deaths for the
comparison to the comparison cohort of gas workers. The study is underpowered to look at
site-specific cancers.





1 Ciilciia

Siud\ is published in I lie peei'-iv\ icuoil scientific lilci'.iliiiv and lias an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Lance,1 1991, 338:959-964. The authors discussed potential for
misclassification using death certificates, healthy worker effect and their related use of a
comparison cohort of gas supply workers, other occupational exposures present at the plant,
potential impact and the lack of smoking data.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. Exposure consisted of a large DLC component that was not quantified.
Given crude TCDD exposure categorization data, no quantitative exposure metric was derived.

This document is a draft for review purposes only and does not constitute Agency policy.

B-11 DRAFT—DO NOT CITE OR QUOTE

B.1.3. The Hamburg Cohort

Table B-8. Manz et al., 1991—All cancer sites combined, site-specific
analyses

-------
3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. Exposure metrics were constructed that took into account duration of
exposure, and periods when exposure was highest. However, exposure estimates did not
consider lagged exposure.





Conclusion

This study is not amenable to further TCDD dose-response analysis and is not considered
further here because it consisted of a large DLC component that was quantified and no
quantitative exposure metric was derived. The dose-response patterns of risks observed across
the three exposure groups provide compelling support for an association between TCDD and
cancer mortality, particularly, given the associations observed when analyses restricted to
those who were hired when TCDD exposures were known to be much higher, and among
those who worked for at least 20 years. Subsequent studies improved the exposure assessment
through the use of serum measures.

Table B-9. Flesch-Janys et al., 1995; Flesch-Janys et al., 1996 erratum—All
cancer sites combined

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Medical records used to identify deaths over the period 1952-1992.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Similarity in smoking rates between control cohort and the exposed
workers was similar based on independent surveys. Occupational exposures to benzene, and
dimethyl sulfate were unlikely to bias dose-response pattern observed as these exposures
occurred in production departments with low-medium levels of exposure.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Dose-response relationship observed across 6 exposure categories,
with the cohort of gas supply workers used as the referent.

4. Consideration

Consideration satisfied. Exposure assessment methodology is clear and adequately
characterizes individual-level exposures. The limitations and uncertainties in the exposure
assessment are considered.

Response

The exposure measure was an integrated TCDD concentration over time estimate that
back calculated TCDD exposures to the end of the employment. Categorical and continuous
TCDD exposures were examined in relation to the health outcome. These efforts improve the
exposure assessment of earlier studies.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. For all cancer sites combined, there were 124 deaths in the exposed
cohort, and 283 in the cohort of gas supply workers. No site-specific cancers were examined
in this paper.

This document is a draft for review purposes only and does not constitute Agency policy.

B-12 DRAFT—DO NOT CITE OR QUOTE

-------




1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Am J Epidemiol, 1995, 1442:1165-1175. The authors discuss the potential
role of other occupational exposures (i.e., dimethyl sulfate, solvents, and benzene), smoking,
and suitability of the comparison cohort of gas supply workers.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum and adipose tissues were used to estimate TCDD exposure in
190 workers. A one-compartment first-order kinetic model was used to estimate exposure at
end of exposure for these workers. Regression methods were then used to estimates TCDD
exposures for all workers.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria not satisfied. Exposure was based on half-life estimates from individuals with
repeated serum measures. Other dioxin-like compounds were considered with the TOTTEQ
exposure metric. No consideration, however, was given to latency or lagged exposures.





Conclusion

The exposure data used within this study are well-suited to a dose-response analysis given the
associations observed, the characterization of exposure using serum, and quality of
ascertainment of cancer outcomes. However, subsequent methods have been applied to the
cohort to derive different exposures to TCDD using area under the curve approaches, which
updates the analysis herein. Therefore, subsequent studies (i.e., Becher et al., 1998) will
supersede this evaluation.

Table B-10. Flesch-Janys et al., 1998—All cancer sites combined, site-
specific analysis

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Mortality follow-up was extended until the end of 1992, an increase
in 3 years from previous analyses of the cohort.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Exposure was well characterized using sera data. While serum
samples provided only from a subsample of surviving workers, these levels were consistent
with expected levels in different production departments. The authors examined other
potential occupational coexposures (e.g., p-hexachlorocyclohexane) and indirectly examined
the potential effect of smoking on the associations that were detected.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

This document is a draft for review purposes only and does not constitute Agency policy.

B-13 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration satisfied. A dose-response relationship across quartiles of TCDD was observed
with cancer mortality based on the SMR statistic (SMRs = 1.24, 1.34, 1.34, 1.73), and a linear
test for trend was statistically significant (p = 0.01).

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. The exposure measure was an integrated TCDD concentration over
time estimate that back-calculated TCDD exposures to the end of the employment.

Categorical and continuous TCDD exposures were examined in relation to the health outcome.
These efforts improve the exposure assessment of earlier studies.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. For all cancer sites combined, there were 124 cancer deaths.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Environ Health Perspect, 1998, 106(2):655-662. The authors address
uncertainties in the estimation of exposure, describe the potential for confounding from
(3-2,4,5-T, hexachlorocyclohexane, and cigarette smoking. In fact, they showed that blood
levels of TCDD were not associated with smoking in a subsample suggesting little bias from
lack of smoking data.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum samples, taken from 190 workers were used to derive TCDD levels
for the entire cohort. Methods used to estimate exposure took into account elimination of
TCDD during employment periods when exposure took place, and the methods of the area
under the curve was used as it takes into account variations in concentration over time, and
reflects cumulative exposure.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. Exposure estimated based on half-lives observed in individuals with
repeated samples. Area under the curve approach was used which is an improvement from
past characterizations of exposure in this cohort.





Conclusion

The study provides data suitable for dose-response modeling. Derivation of exposure was
done using current understanding of elimination of TCDD. Estimates of risks were derived
from external comparisons to the general population that are unlikely to be biased by healthy
worker effect, but risks generated using internal cohort comparisons would be preferable.
Becher et al., (1998) assessed this same data taking cancer latency into account, therefore
Flesch-Janys et al., (1998) will not be further considered for dose-response modeling.

1

2

3

4

5

This document is a draft for review purposes only and does not constitute Agency policy.

B-14 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table B-ll. Becher et al., 1998—All cancer sites combined

2

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Medical records used to identify deaths over the period 1952-1992.
The follow-up interval was lengthy.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Risks adjusted for exposures to TEQ, (3-hexachlorbenzene, and
employment characteristics. Smoking was shown to be similar to the comparison cohort of
gas workers.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. A variety of exposure measures for both TCDD and TEQs found
positive associations with cancer mortality.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. The exposure measure was an integrated TCDD concentration over
time estimate that back-calculated TCDD exposures to the end of the employment.

Categorical and continuous TCDD exposures were examined in relation to the health outcome.
Different models explored the shape of the dose-response curve. These efforts improve the
exposure assessment of earlier studies.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. For all cancer sites combined, there were 124 cancer deaths.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Environ Health Perspect, 1998, 106(2):663-670. The authors discuss
uncertainties associated with their use of exposure metrics, inability to evaluate effects for
PCDD/Fs other than dioxin due to high correlations with (3-HCH, and inability to characterize
risks associated with exposures in children.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. The authors derived a measure of cumulative dose as a time-dependent
variable ("area under curve") using serum measures available in a sample of 275 workers.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. TCDD levels estimates through back-extrapolation of serum levels based on
half-life estimates obtained from previous studies. Latency was considered, and a variety of
exposure metrics including nonlinear relationships were evaluated.





This document is a draft for review purposes only and does not constitute Agency policy.

B-15 DRAFT—DO NOT CITE OR QUOTE

-------
In this paper, a variety of exposure metrics were found to be positively associated with cancer
mortality. The additional lifetime risk of cancer corresponded to a daily intake of lpg ranged
between .01 and 0.001. This study was modeled in the 2003 Reassessment and is considered
for further dose-response evaluations herein.

1	B.1.4. The Seveso Cohort Studies

2

3	Table B-12. Bertazzi et al., 2001—All cancer sites combined, site-specific

4	analyses

5

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Mortality appears to be well captured from the vital statistics
registries in the region (99% complete). Vital status was ascertained using similar methods for
both the exposed and reference populations. Both cancer and noncancer mortality outcomes
were evaluated. Ideally, would have evaluated incident rather than decedent outcomes for
cancer.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Individual-level data on potential confounders (i.e., age, calendar
period, and gender) were adjusted for. Information from other independent surveys suggests
similarity between smoking behaviors across the regions. Comparison of cancer mortality
rates before the time of the accident between the regions also revealed no differences.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied (for all cancers combined). No statistically significant excesses noted
in Zone A, or Zone B relative to reference area. Evidence of an exposure-response
relationship was detected for lymphatic and hematopoietic tissues by number of years since
first exposure.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration not satisfied. Subjects were assigned to one of the zones (A, B, R, or reference)
based on official residence on the day of the accident or at entry into the area. Exposure
misclassification is likely and lack of individual-level data precludes an examination of this
source of error.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. In total, 27, and 222, cancer deaths were found among residents of
Zones A, and B, respectively. This allowed examined of gender-specific effects.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

This document is a draft for review purposes only and does not constitute Agency policy.

B-16 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. Am J Epidemiol, 2001 Jun 1; 153(11): 1031—1044. Authors discuss
completeness of mortality ascertainment, diagnostic accuracy of death certificates particularly
with respect to diabetes, limited available of blood dioxin measures that did not permit
estimation of TCDD dose on an individual-level basis.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. Individual-level exposure data are unavailable. Exposure based on place
of residence at time of the explosion. Soil sampling performed indicated considerable
variability in TCDD levels within each region. In addition, place of residency at time of
explosion does not ensure individuals were at their home around the time of the accident.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria not satisfied. An ecological measure of exposure (region of residency at time of
accident) was used to categorize individuals according to their possible exposure. Latencies
were considered. While such an approach has value for identifying wherever excesses
occurred among highly exposed populations, it is not precise enough to conduct a quantitative
dose-response analysis.





Conclusion

The lack of individual-level exposure data precludes quantitative dose-response modeling
using these data.

Table B-13. Pesatori et al., 2003—All cancer sites combined, site-specific
analyses

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Mortality was ascertained from 1977-1996, and, as reported in other
related manuscripts, appears to be well captured from the vital statistics registries in the region
(99% complete). Cancer incidence data was available from 1977-1991.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Individual-level data on potential confounders (i.e., age, calendar
period, and gender) were adjusted for.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. Although risk of all cancer mortality was not associated with zone
of residence, increased risk of cancer incidence was observed in Zone A. Among men, excess
lymphatic and hematopoietic cancer incidence was observed in Zone A (primarily to
non-Hodgkin's lymphoma). Soft tissues sarcoma cancer incidence was also associated with
residence in Zone R among males, but not the more highly exposed zones (A and B). Among
females living in Zones A and B, higher rates were observed for multiple myeloma (RR = 4.9,
95% CI = 1.5-16.1), cancer of the vagina (RR = 5.5, 95% CI = 1.3-23.8), and cancer of the
biliary tract (RR = 3.0, 95% CI = 1.1-8.2).

This document is a draft for review purposes only and does not constitute Agency policy.

B-17 DRAFT—DO NOT CITE OR QUOTE

-------
4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration not satisfied. Subjects were assigned to one of the zones (A, B, R, or reference)
based on official residence on the day of the accident or at entry into the area. Exposure
misclassification is likely and lack of individual-level data precludes an examination of this
source of error.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied for some endpoints, although several of the cancer specific mortality
results among women were based on very small number of deaths (i.e., <5).





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. OccupEnvMed, 1998;55:126-131. Authors discuss limitations such as
residency-based exposure assignment, absence of smoking, differential and death certification
in exposed versus nonexposed areas.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. Individual-level exposure data are unavailable. Exposure based on place
of residence at time of the explosion. Soil sampling performed indicated considerable
variability in TCDD levels within each region. In addition, place of residency at time of
explosion does not ensure individuals were at their home around the time of the accident.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria not satisfied. An ecological measure of exposure (region of residency at time of
accident) was used to categorize individuals according to their possible exposure. Latencies
were considered. While such an approach has value for identifying wherever excesses
occurred among highly exposed populations, it is not precise enough to conduct a quantitative
dose-response analysis.





Conclusion

No dose-response patterns evident in the study, and the study lacked quantifiable measures of
TCDD at an individual-level basis. The data are not well suited for dose-response analysis.

Table B-14. Consonni et al., 2008—All cancer sites combined, site-specific
analyses

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Mortality appears to be well captured from the vital statistics
registries in the region (99% complete). Both cancer and noncancer mortality evaluated,
although diagnostic accuracy of death certificates is likely low. Ideally, would have evaluated
incident rather than decedent outcomes for cancer.

This document is a draft for review purposes only and does not constitute Agency policy.

B-18 DRAFT—DO NOT CITE OR QUOTE

-------
2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Individual-level data on potential confounders (i.e., age, calendar
period, and gender) were adjusted for. Comparison of cancer mortality rates before the time of
the accident between the regions also revealed no differences. Information from other
independent surveys suggests similarity between smoking behaviors across the regions.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied for some outcomes. For all cancer sites combined, no evidence of
dose-response was observed relative to general population across Zones A, B and R. Only
statistically significant excess found in Zone A was for chronic rheumatic disease but based on
only three deaths. Higher cancer excesses were found in Zone A after a latency period was
incorporated; however, no dose-response relationship observed with this latency period.
Evidence of an exposure-response relationship was detected for lymphatic and hematopoietic
tissues by zone of residence.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration not satisfied. Subjects were assigned to one of the zones (A, B, R, or reference)
based on official residence on the day of the accident or at entry into the area. Exposure
misclassification is likely and lack of individual-level data precludes an examination of this
source of error.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. In total, 42, 244, and 1,848 cancer deaths were found among residents
of Zones A, B, and R respectively.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Am J Epidemiol, 2008, 167:847-858. Authors discuss potential for
selection bias, limitation of residential based measure of exposure, similarities of mortality
ascertainment in exposed and referent populations, and multiple testing.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. Individual-level exposure data are unavailable. Exposure based on place
of residence at time of the explosion. Soil sampling performed indicated considerable
variability in TCDD levels within each region. In addition, place of residency at time of
explosion does not ensure individuals were at their home around the time of the accident.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria not satisfied. An ecological measure of exposure (region of residency at time of
accident) was used to categorize individuals according to their possible exposure. Latencies
were considered. While such an approach has value for identifying wherever excesses
occurred among highly exposed populations, it is not precise enough to conduct a quantitative
dose-response analysis.

This document is a draft for review purposes only and does not constitute Agency policy.

B-19 DRAFT—DO NOT CITE OR QUOTE

-------




Conclusion

The lack of individual-level exposure data precludes quantitative dose-response modeling
using these data.

Table B-15. Baccarelli et al., 2006—Site-specific analysis

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Polymerase chain reaction (PCR) methods were used to describe
outcome measures. The prevalence of t(14; 18) was estimated as those individuals having a
t(14; 18) positive blood sample divided by the t(14; 18) frequency (number of copies per
million lymphocytes).

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Questionnaire data were used to collect information on cigarette
smoking. Other potential confounders (age, smoking status, and duration of smoking). In
addition, both exposure and outcome were objectively and accurately measured.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration was not satisfied. Associations were detected between the frequency of t(14;
18) and plasma TCDD levels as well as zone of residence at the time of the explosion. No
association was detected for these exposure measures and prevalence of t(14; 18). A dose-
response trend was detected for TCDD and the mean number of t(14; 18)
translocations/106 lymphocytes, however the relevance of t(14; 18) in lymphocytes to
non-Hodgkin's lymphoma is uncertain.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. The authors highlight that exposure metrics represent both past and
current body burdens. They employ several different exposure metrics of TCDD: place of
residence (Zone A, B, R or reference), categorical serum measures, a linear term, log (base 10)
transformed TCDD, and individuals with chloracne diagnosed after the accident.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. Analyses are made using 72 highly exposed, and 72 low exposed
individuals.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Carcinogenesis, 2006, 27(10):2001-2007. The authors discuss the
limitation of using t(14; 18) translocations as an outcome measure, and the uncertain role it
plays in the development of non-Hodgkin's lymphoma.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

This document is a draft for review purposes only and does not constitute Agency policy.

B-20 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. A total of 144 subjects were included in the study. This included
72 subjects who had low exposures, and 72 who had high exposures based on serum
concentrations.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. A variety of measures were employed including current TCDD levels, as
well as surrogates of exposure at the time of the accident.





Conclusion

While an association was observed with the frequency of t(14; 18) translocation, it is uncertain
whether this translates into an increased risk of non-Hodgkin's lymphoma. Given the
speculative nature of this endpoint and lack of demonstrated adverse effect, dose-response
analyses for this outcome were not conducted.

Table B-16. Warner et al., 2002—Breast cancer incidence

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Diagnoses of incident breast cancer were based on interview and
information from medical records appears thorough. Of the 15 cases of breast cancer, 13 were
confirmed by pathology and the remaining 2 by surgery report only. Three cases of breast
cancer were excluded which represents a large proportion of the total cases identified. This
would reduce sample size and could result in bias if the exclusion was association with TCDD
exposure.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Information was collected on an extensive series of risk factors by
using an interviewer administered questionnaire. Participation rates for the survey were fairly
good (80%).

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Limited evidence (not statistically significant) of a dose-response
when TCDD was analyzed as a categorical variable; only one breast cancer case was in the
referent exposure category. In the analysis of TCDD as a continuous measure (logi0TCDD),
the hazard ratio associated with a 10-fold increase in TCDD serum levels was 2.1
(95% CI: 1.0-4.6).

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Different exposure metrics were considered in these analyses
(categorical, continuous, measures on a log-scale). Exposure data are of high quality as they
are based on serum samples taken among women near the time of the accident. As such,
exposure assignment is not dependent on as many assumption as used in occupational cohorts
were back-extrapolation for many years had to be performed.

This document is a draft for review purposes only and does not constitute Agency policy.

B-21 DRAFT—DO NOT CITE OR QUOTE

-------
5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration somewhat satisfied. Inadequate follow-up for cancer limited the number of
cases available. Sample size also limited the conclusions draw from the categorical analysis
based on very few cases for some exposure categories.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Paper published in Environ Health Perspect, 2002 Jul, 110(7):625-628. A
major limitation of the study is the small number of incident cases of breast cancer (n = 15),
important strengths of the study include characterization of TCDD using serum collected near
the time of the accident.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum was used to estimate TCDD levels in 981 of 1271 eligible women
who had lived in either of the two contaminated sites in 1976. Data represent an objective
measure of TCDD near the time of the exposure. Data obtained near the time of exposure
which minimized the potential for exposure misclassification.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. Exposure characterized using serum measures obtained close to the time of
the accident.





Conclusion

While characterization of exposure and availability of other risk factor data at an
individual-level basis are important strengths of this study, small sample size (n= 15 cases)
based on inadequate follow-up is a key limitation. Quantitative dose-response analyses were
conducted using this study, but continued follow-up of the study population or consideration of
all cancer outcomes would be valuable.

1

2

3

4

5

6

7

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration cannot be evaluated. Insufficient details are provided in the paper to gauge the
completeness and coverage of the cancer registry and mortality data. Health outcomes were
studied on the basis of information in the official medical statistics.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. Given that this is an ecological study, bias may be present.

This document is a draft for review purposes only and does not constitute Agency policy.

B-22 DRAFT—DO NOT CITE OR QUOTE

B.1.5. The Chapaevsk Study

Table B-17. Revich et al., 2001—All cancer sites combined, and site-specific
analyses

-------
3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration cannot be evaluated. Dose-response was not evaluated as exposure was based on
residency in the region vs. no residency.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration not satisfied. No individual-level exposure estimates were used.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. A total of 476 cancer deaths were observed among males, and 376
cancer deaths observed among females. The precision of the SMRs is demonstrated with fairly
narrow confidence intervals for many causes of death.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria not satisfied. Published in Chemosphere, 2001, 43(4-7):951-966. Authors do not
address the completeness of the mortality follow-up, and whether there are differences in death
registrations between regions. The authors do acknowledge, however, that new investigations
being undertaken would characterize exposure using serum-based measures.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. It is a cross-sectional study that compares mortality rates between
regions. No individual-level exposure data available.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of exposure
are consistent with the current biological understanding of dose. The reported dose is
consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria not satisfied. No individual-level exposure estimates were used in the study.





Conclusion

These cancer data are cross-sectional in nature and not appropriate for a dose-response analysis.

B.1.6. The Air Force Health ("Ranch Hands") Study

Table B-18. Akhtar et al., 2004—All cancer sites combined and site-specific
analyses

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Cancer incidence and mortality based on information from repeated
medical examinations, medical records and death certificate.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

This document is a draft for review purposes only and does not constitute Agency policy.

B-23 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration not satisfied. The risk estimates were adjusted for a number of factors
measured on an individual level including smoking. However, analyses are unable to
distinguish between exposure to TCDD and 2,4-D as both were used in equal parts in the
formulation of Agent Orange.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. There is evidence of a dose-response for all cancers and for some
site-specific cancers (i.e., malignant melanoma, and prostate cancer).

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. High quality exposure data for most veterans was collected, so
extrapolation to other members of the cohort was not required. The serum dioxin
measurements also correlated well with reported skin exposure to herbicide in Vietnam, but
collection of the samples 25 years later required back-extrapolation.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. In total, 117 incidence cancers identified in the Ranch Hands cohort.
For those sites with a dose-response association, malignant melanoma and prostate cancer,
there were 16 and 34 incident cases, respectively.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Published in J Occup Environ Med, 2004, 46(2): 123-136. Authors
highlight that this is only cancer incidence study in US veterans, and the lengthy interval of
follow-up (35-40 years)—both important strengths of the study. They addressed potential
bias from healthy-worker effect, and uncertainties surrounding the estimation of TCDD
exposure (extrapolation 30 years after exposure), as well as exposure to other chemical
exposures. Study uses incident outcomes for cancer.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Individual exposure estimates are based on measurements of dioxin serum
lipid concentrations. They were available for 1,009 Ranch Hands and 1,429 in the
comparison cohort.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. TCDD exposures at the end of duty were estimated by back-extrapolating
1987 serum values.





Conclusion

The major limitation of the study is the inability to isolate effects of TCDD from other
chemicals used in the formulation of the herbicides. This limitation precludes dose-response
modeling of the TCDD and cancer outcomes data.

1

2

3

This document is a draft for review purposes only and does not constitute Agency policy.

B-24 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table B-19. Michalek and Pavuk, 2008—All cancer sites combined

2

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Cancer incidence was ascertained through the use of medical records.
Death certificate were used to identify some malignancies. Little data is provided on the
number of individuals lost to follow-up, however the same mechanisms of case ascertainment
were applied to both the comparison and Ranch Hand cohorts.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. Information collected from repeated physical examinations
allowed for the adjustment of risk factors such as smoking. Agent Orange was a 50% mixture
of 2,4-D and TCDD; therefore, potential for confounding by other coexposures is likely.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied for some comparisons. Statistically significant associations were
noted with cancer incidence and TCDD when analyses were restricted to workers who served
at most two years in Southeast Asia and those who sprayed more than 30 days before 1967.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Initial TCDD dose were estimated at the end of the tour of duty for
the Ranch Hands. Individual-level serum dioxin measurements correlated well with
correlated with days of spraying and calendar period of service, but collection of the samples
roughly 20 years later required back-extrapolation.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. A total of 347 incident cases of cancer were used in the analyses.
For stratified analyses, statistical power is more limited. For example, only 67 incident
cancer in the subset of workers who spent less than 2 years in Southeast Asia, and sprayed for
at least 30 days before 1967.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied J Occup Environ Med 2008; 50:330-340. The authors discuss issues related
to exposure misclassification error, and suggest approaches for improving characterization of
days of spraying. Congener specific data were unavailable, thereby not allowing for congener
specific risks or adjustments to be made.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. TCDD data was available for 986 veterans in the Ranch Hand cohort, and
1,597 members of the comparison cohort.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

This document is a draft for review purposes only and does not constitute Agency policy.

B-25 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. TCDD exposures at the end of duty were estimated by back-extrapolating
1987 serum values.





Conclusion

Ranch Hand veterans were exposed to other contaminants in the herbicides that were mixed,
thereby making it difficult to determine independent effects of TCDD on cancer. In
particular, 2,4-D has been shown to be associated with some cancers, notable cancer of the
prostate. This limitation precludes dose-response modeling of TCDD and cancer using data
from this cohort.

1

2

3

4

5

6

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. National records for death registrations through the New
Zealand Health Information Service (NZHIS). Subjects not registered as having died during
the study period were confirmed to be actually alive and resident in New Zealand using the
New Zealand Electoral Roll, drivers' license, and social security records.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. Seventeen percent of workers were lost to follow up but it is
unclear if bias resulted. The dichotomous exposure measure was based on exposure to
TCDD, chlorinated dioxins and phenoxy herbicides, so confounding is a possibility by these
coexposures.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Dose-response evidence for duration of employment and elevated
mortality noted only in synthesis workers.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Exposure measures were limited to duration of employment and exposed/unexposed.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. For all cancer sites combined, there were 43 cancer deaths among the
production workers, and 35 such deaths among the sprayers. Site-specific cancer analyses are
limited by small sample sizes.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria not satisfied Occup Env Med, 2005; 62:34-40. A high percentage of the cohort was
lost to follow-up (17%). The authors fail to mention this important limitation in this paper.

This document is a draft for review purposes only and does not constitute Agency policy.

B-26 DRAFT—DO NOT CITE OR QUOTE

B.1.7. Other Studies of Potential Relevance to Dose-Response Modeling

Table B-20. 't Mannetje et al., 2005—All cancer sites combined, site specific
analyses

-------
2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. This study used duration of exposure, at an individual level, as a
surrogate measure of TCDD.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria not satisfied. Exposure was defined according to duration, and not concentrations of
TCDD. Latency intervals were not evaluated.





Conclusion

Overall, quantitative exposure data are lacking for TCDD and limited dose-response
relationships were observed across duration of exposure categories. Furthermore,
confounding by coexposures is a possibility. Taken together, these data are not suitable for
inclusion in a dose-response analysis

Table B-21. McBride et al., 2009b—All cancer sites combined, site-specific
analysis

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. The New Zealand Health Information Service Mortality Collection
and the Registrar-General's Index to Deaths. Additional searches were based on the last
known address from the work record; the electoral roll and the habitation index; the telephone
book; the internet; and Terranet property information database. An additional search was
carried out through the Births, Deaths, and Marriages office of the New Zealand Department
of Internal Affairs. Lastly, automated personnel and pension records were also used to locate
past New Plymouth workers and identify some deaths.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. Considerable amount of workers were lost to follow up (22%), but
it is unclear if bias resulted. The dichotomous exposure measure was based on exposure to
TCDD, chlorinated dioxins and phenoxy herbicides, so confounding is a possibility by these
coexposures.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. There was no examination of dose-response effects.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Dichotomous exposure (exposed/unexposed) and duration of
employment were examined from job exposure classification assessed via occupational history
records industrial hygienists/factory personnel knowledge and questionnaires. Authors discuss
limitations in the assignment of exposure among cohort members.

This document is a draft for review purposes only and does not constitute Agency policy.

B-27 DRAFT—DO NOT CITE OR QUOTE

-------
5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration not satisfied. A low number of deaths (n = 16) may have limited ability to
detect effects small in magnitude and exposure-response relationships.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Published in Occup Medicine, 2009; 59(4):255-263. The authors highlight
cohort lost to follow-up, the limited size of the cohort, differences in cohort definitions
between sprayers and producers, and the potential for other exposures during employment at
the plant.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. TCDD exposures were not quantified.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria not satisfied. Effective dose could not be estimated given the lack of individual-level
exposure data.





Conclusion

The study lacks the quantification of exposures at an individual level, precluding dose-
response analysis. This study is not considered further in the dose-response modeling analysis.

Table B-22. McBride et al., 2009a—All cancer sites combined, site-specific
analysis

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. The New Zealand Health Information Service Mortality Collection
and the Registrar-General's Index to Deaths were used to identify deaths. Additional searches
were based on the last known address from the work record; the electoral roll and the
habitation index; the telephone book; the internet; and several other public databases in New
Zealand. An additional search was carried out through the Births, Deaths, and Marriages
office of the New Zealand Department of Internal Affairs. Lastly, automated personnel and
pension records were also used to locate past New Plymouth workers and identify some
deaths.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Workers lost to follow-up were an unlikely source of bias especially
for internal analyses. Confounding by other coexposures (e.g., 2,4,6-TCP) unlikely to have
resulted in bias, due to presumed poor correlation with TCDD.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

This document is a draft for review purposes only and does not constitute Agency policy.

B-28 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration not satisfied. The linear test for trend for TCDD exposure was not statistically
significant for all cancer sites (combined), as well as lung cancer mortality. Dose-response
relationships were not apparent across quartiles of TCDD exposure for all cancer sites
combined, digestive cancers, lung cancer, soft tissue sarcomas or non-Hodgkin's Lymphoma.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Cumulative exposure to TCDD as a time-dependent metric was
estimated for each worker from serum samples, but the authors did not examine a continuous
measure of TCDD exposure (lagged orunlagged).

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Published in J Occup Environ Med 51:1049-1056. This paper discussed the
22% of the cohort lost to follow-up, differences in cohort definitions between sprayers and
producers, and the potential for other exposures during employment at the plant.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum measures available for 346 workers were used to derive TCDD
exposures for the entire cohort using the area under the curve approach.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria satisfied. Effective dose could be estimated from serum-derived cumulative exposure
estimates.





Conclusion

Given that no dose-response associations were found, the data are not suited to dose-response
analysis.

Table B-23. Hooiveld et al., 1998—All cancer sites combined, site-specific
analysis

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Outcomes were mortality. Few deaths expected to be missed since
only 5% of the cohort was lost to follow-up or had emigrated.

This document is a draft for review purposes only and does not constitute Agency policy.

B-29 DRAFT—DO NOT CITE OR QUOTE

-------
2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. Although dioxin-like compounds (PCDDs, PCDFs, and PCBs)
were measured in the serum samples, these were not incorporated into the analysis. Therefore,
confounding cannot be ruled out as an explanation of the reported association.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. A dose-response pattern was observed for internal cohort comparison
for all cancer mortality, with RRs of 5.0 and 5.6 for the medium and high exposure,
respectively. Dose-response patterns evident for lung cancer as well.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Detailed occupational histories to assign dichotomous exposures
(exposed/unexposed) based on maximum exposure levels. Although serum data also collected
for TCDD and other coexposures (PCDDs, PCDFs, and PCBs), study only presents data for
TCDD exposure. TCDD exposures at time of maximum exposure were extrapolated from
measured serum.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration not satisfied for internal cohort comparisons in either men or women. Among
men, only 7 cancer deaths were observed among those in the unexposed part of the cohort, and
51 among exposed workers. For external cohort comparisons, a total of 20 deaths were
observed.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Am J Epidemiol, 1998, 147:891-901. The authors address potential
limitations of estimating TCDD exposure from a subsample of surviving workers, lack of
smoking data, the healthy worker effect, and relevance of other occupational exposures.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum samples were obtained from 94 of 144 subjects who were asked to
participate in serum measurement study. Of these, a further 44 excluded due to absence due to
holiday or work (n = 22), and nonexposed workers excluded because matching exposed
worker not participating (n = 20). TCDD levels were extrapolated to the time of maximum
exposure.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined.

Response

Criteria not satisfied. Exposures assigned based on levels at maximum exposure. Assignment
of exposure based on nonrepresentative sample of 50 survivors among the occupational cohort.





Conclusion

The small number of identified cancer deaths, limitations in terms of the exposure assignment
(based on nonrepresentative sample, and maximum exposure level) and concern over potential
confounding by coexposures preclude using these data for a dose-response analysis.

This document is a draft for review purposes only and does not constitute Agency policy.

B-30 DRAFT—DO NOT CITE OR QUOTE

-------
1	B.2. EVALUATION OF NONCANCER STUDIES

2	B.2.1. NIOSH Cohort

3

4	Table B-24. Steenland et al., 1999—Mortality (noncancer)

5

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. The study evaluated mortality from all cancer sites (combined). As
described in the paper, the sources of vital status and cause of death information were received
from the Social Security death files, the National Death Index, and the Internal Revenue
Service. Vital status was known for 99.4% of the cohort members, cause of death information
is available for 98% of the decedents.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. External comparisons for all-cause and cardiovascular mortality
do not appear to be affected by the "healthy worker effect" as similar patterns were observed
with internal cohort comparisons. Nonetheless, internal cohort comparisons are unable to
adjust for many of the individual-level risk factors for cardiovascular disease.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. A dose-response relationship was observed with ischemic heart
disease (linear test for trendp = 0.05), and with TCDD on a log-transformed scale the p-valuc
was <0.001.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. The study conducted detailed sensitivity analyses and evaluated
different assumptions regarding latency, log-transformed TCDD exposures, and half-life
values for TCDD. Associations were stronger for log-transformed values, and latency
intervals of 15 years.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. This is the largest of the occupational cohorts with exposures to
TCDD. The cohort consisted of 5,132 male workers and a total of 456 deaths from ischemic
heart disease. This permits characterization of risk for all cancer sites (combined).





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied Journal of the National Cancer Institute, 1999, 91(9):779-786. The authors
discussed the potential for bias from smoking, and other occupational exposures for which
data for both were lacking at an individual basis.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

This document is a draft for review purposes only and does not constitute Agency policy.

B-31 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria not satisfied. Exposure scores assigned at an individual level based on job-exposure
matrix (JEM). The JEM was based on estimated factor of contact with TCDD in each job,
level of TCCD contamination of materials at each plant over time, and proportion of day
worker could be in contact with materials. These factors were multiplied together to derive a
daily exposure score, which was accumulated over the working history of each worker to
obtain a cumulative measure of TCDD.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. The follow-up of the cohort extended from 1942 until the end of 1993.
Greater than 25 years of follow-up have accrued in cohort allowing for latency to be
examined. Different assumptions on the half-life of TCDD were evaluated and produced
similar results. Latency intervals were incorporated, with strongest associations noted no lag.
Suggests mechanisms occur at the same time as exposure. However, noncancer mortality is
ik»I a \ iable endpoiiil U» consider for (mllier dose-response anaUsis

Conclusion

TCDD exposures were quantified in this study, and a dose-response relationship was observed
with ischemic heart disease mortality. The sample size was sufficient, and the follow-up
interval was lengthy. However, no individual-level data were available for cardiovascular
conditions, and the inability to adjust for these exposures introduces considerable uncertainty
into the risk estimates. Furthermore, noncancer mortality is not considered a viable endpoint
for dose-response analysis.

Table B-25. Collins et al., 2009—Mortality (noncancer)

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Vital status complete for all but two workers.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. No information collected on smoking status, but no excess in lung
cancer or nonmalignant respiratory diseases noted. Analyses took into account potential for
exposure to pentachlorophenol. External cohort comparisons should be interpreted cautiously
due to healthy worker effect, but internal cohort comparisons should not be influence by this
bias.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. No statistically significant mortality excess for any noncancer
mortality outcome evaluated. This included ischemic heart disease, stroke, nonmalignant
respiratory disease, ulcers, cirrhosis, and external causes of death (accidents). Modeling of
continuous measure of TCDD was not related to diabetes, ischemic heart disease, or
nonmalignant respiratory mortality.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

This document is a draft for review purposes only and does not constitute Agency policy.

B-32 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration satisfied. The authors used these serum from 280 former TCP workers to
estimate historical exposure levels of TCDD, furans, and polychlorinated biphenyls for all
1,615 workers. Exposure assessment included detailed work history, industrial hygiene
monitoring, and the presence of chloracne cases among groups of workers. This data was
integrated into a 1-compartment, first-order pharmacokinetic to determine the average TCDD
dose associated with jobs in each group, after accounting for the presence of background
exposures estimated from the residual serum TCDD concentration in the sampled individuals.
The authors did not evaluate departures from linearity, or examine skewness at higher
exposures. No presentation of exposure levels was provided.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. A total of 662 deaths were observed. Of these, 218 were from
ischemic heart disease, and 16 from diabetes (two outcomes for which associations have been
noted elsewhere).





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Published in Am J Epidemiol, 2009, 170(4):501-506. The authors discuss
potential for exposure misclassification, large size of the cohort, lengthy follow-up interval,
and large number of workers who provided serum from which TCDD exposures were
estimated.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. This study has the greatest number of serum samples obtained from a
specific plant.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. Noncancer mortality is not a viable endpoint to consider for further
dose-response analysis.





Conclusions

No dose-response associations were noted for noncancer mortality outcomes. The data are,
therefore, not suited for dose-response modeling.

B.2.2. BASF Cohort

Table B-26. Ott and Zober, 1996—Mortality (noncancer)

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Mortality ascertainment appeared to be fairly complete.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

This document is a draft for review purposes only and does not constitute Agency policy.

B-33 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration satisfied. Information was collected on smoking status, body mass index, and
other occupational exposures, however a large portion of the cohort was firefighters who may
have been exposed to other occupational carcinogens. However, the recruitment of survivors
may results in under-ascertainment of mortality.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. For external cohort comparisons across the three TCDD exposure
categories, there was no dose-response pattern observed for any of the noncancer causes of
death. Cox regression risk estimates for all cause or circulatory disease mortality when TCDD
was modeled as a continuous variable were not statistically significant.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Cumulative measure of TCDD expressed was derived from serum
measures. Exposure was also estimated by chloracne status of the cohort members. The
authors have not addressed the potential implication of deriving TCDD exposure estimates for
the whole cohort using sera data that were available for only about half of the cohort.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. For all causes of death, there were 92 deaths, while 37 circulatory
deaths. Many of the cause-specific death had less than 5 deaths in the upper exposure
category.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Occup Environ Med, 1996, 53:606-612. A large component of the cohort
was assembled by actively seeking out workers who were alive in the mid 1980s. As a result,
it is likely a number of deaths were missed. This is supported by much lower SMRs in this
component of the cohort published in earlier studies of the cohort. This underascertainment of
mortality results in biased SMR statistics (underestimated). The authors do highlight the value
of the serum based measures to estimate TCDD exposure

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum samples, taken in 1989, were available for 138 surviving workers out
of 254 and allowed for cumulative TCDD levels to be estimated using regression techniques in
the remainder of the cohort.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. Exposure assignment took into the affect that body mass index had on
TCDD half-lives. TCDD levels estimates through back-extrapolation of serum levels based on
half-life estimates obtained from previous studies. Latency was considered with stronger
association observed in external comparisons incorporating a latency of 20 years. The follow-
up of the cohort was lengthy (>50 years). However, noncancer mortality is not a viable
endpoint to consider for further dose-response analysis.





This document is a draft for review purposes only and does not constitute Agency policy.

B-34 DRAFT—DO NOT CITE OR QUOTE

-------
No associations noted with any noncancer deaths. External comparisons should be treated
cautiously especially for cardiovascular mortality which is recognized to often be biased by
the healthy-worker effect. In the absence of any outcome with an association with TCDD
exposure, dose-response analyses of these data were not undertaken.

1	B.2.3. Hamburg Cohort

2

3	Table B-27. Flesch-Janys et al., 1995; Flesch-Janys et al., 1996 erratum—

4	Mortality (noncancer)

5

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Medical records used to identify deaths over the period 1952-1992.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Similarity in smoking rates between control cohort and the exposed
workers was similar based on independent surveys. Occupational exposures to benzene, and
dimethyl sulfate were unlikely to bias dose-response pattern observed as these exposures
occurred in production departments with low to medium levels of TCDD exposure.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Dose-response relationship observed for all-cause mortality,
cardiovascular mortality, and ischemic heart disease mortality across 6 exposure categories,
with the cohort of gas supply workers used as the referent. The linear tests for trend for these
three outcomes were all statistically significant (p < 0.05).

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. The exposure measures was an integrated TCDD concentration over
time estimate that back-calculated TCDD exposures to the end of the employment.

Categorical and continuous TCDD exposures were examined in relation to the health outcome.
These efforts improve the exposure assessment of earlier studies.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. For all causes of death combined, there were 414 deaths in the
exposed cohort, and 943 in the cohort of gas supply workers. A total of 157 and 76 deaths
from cardiovascular disease, and ischemic heart disease were noted. The corresponding
number in the cohort of gas supply workers was 459, and 205, respectively.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Am J Epidemiol, 1995, 1442:1165-1175. The authors discuss the potential
role of other occupational exposures (i.e., dimethyl sulfate, solvents, benzene), smoking, and
suitability of the comparison cohort of gas supply workers.

This document is a draft for review purposes only and does not constitute Agency policy.

B-35 DRAFT—DO NOT CITE OR QUOTE

-------
2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum and adipose tissues were used to estimate TCDD exposure in
190 workers. A one-compartment first-order kinetic model was used to estimate exposure at
end of exposure for these workers. Regression methods were then used to estimates TCDD
exposures for all workers.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. Exposure based on half-life estimates from individuals with repeated
serum measures. Other dioxin-like compounds were considered with the TOTTEQ exposure
metric. Noncancer mortality, however, is not a viable endpoint to consider for further dose-
response analysis.





Conclusion

Although, the exposure data used within this study are well-suited to a dose-response analysis
for all-cause and cardiovascular mortality given the associations observed, use of noncancer
mortality endpoint is not amenable for further dose-response analysis.

1	B.2.4. The Seveso Women's Health Study

2

3	Table B-28. Eskenazi et al., 2002a—Menstrual cycle characteristics

4

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Information was also obtained from medical records for all obstetric
and gynecologic conditions. Information on menstrual cycles was obtained from
questionnaires. Women were asked about length of cycles, regularity, how many days flow
lasted, and heaviness of menstrual flow (scanty, moderate, or heavy). Measurement error is
likely for the subjective nature of self-reported menstrual parameters but specificity and
sensitivity is difficult to ascertain due to lack of validation data for these measures.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Detailed risk factor information was collected from questionnaire,
allowing for the potential confounding influence of many risk factors to be controlled for. The
length of cycle study findings may have been affected by the presence of a few outliers.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. A positive dose-response relationship was found with TCDD among
women who were premenarcheal at time of the explosion and longer menstrual cycle. Increased
TCDD resulted in a reduced odds of scanty menstrual flow. No association was noted with
these two outcomes among postmenarcheal women. A decreased risk of irregular cycles was
observed with higher TCDD levels.

This document is a draft for review purposes only and does not constitute Agency policy.

B-36 DRAFT—DO NOT CITE OR QUOTE

-------
4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Criteria satisfied. Serum concentrations of TCDD offer improved exposure assessment,
although delineating the critical exposure window is challenging given the nature of the very
high initial exposure.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure adequate
statistical power.

Response

Consideration satisfied. Cohort was large enough as analyses were conducted on 301 women.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion of
the strengths and limitations.

Response

Criteria satisfied. Am J Epidemiol, 2002; 156(4) 383-392. Limitations included an inability to
assess affects on menstrual cycle at time body burdens were the highest (at time of the
accident). Also, TCDD was estimated for 1976, not concurrent with their cycles in the previous
year, and a large number of women were excluded due to intrauterine device or oral
contraceptive use. Strengths included population-based nature of study, with characterization
of exposure using serum, and levels of other polychlorinated dibenzo-p-dioxins and
dibenzofurans were at background levels. Findings for length of menstrual cycle may be
unduly influenced by the presence of some outliers.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. The study population was based on 301 women as those who were over the
age of 44 were excluded, as well as women with surgical of natural menopause, women with
Turner's syndrome, those who had been pregnant or breastfed in the past year, and those who
had used an intrauterine device or oral contraceptives. For 272 women, TCDD levels were
based on serum data provided in 1976; TCDD levels were back-extrapolated to 1976 levels for
the other 29 women.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of exposure
are consistent with the current biological understanding of dose. The reported dose is consistent
with a toxicologically relevant dose. Latency and appropriate window(s) of exposure
examined. Response had to be a nonfatal endpoint.

Response

Criteria satisfied. Ideally, TCDD exposures would be concurrent with reporting of cycle
characteristics. Herein, TCDD exposures were based on levels in 1976; however, given the
long half-life of TCDD and the same follow-up interval for all women, TCDD exposures in
1976 should correlate well with levels near the time of interview. Further, the critical window
of exposure can be estimated for the women that were premenarcheal at the time of the accident
(13 years).





Conclusion

This study meets all of the criteria and considerations for further dose-response analysis. The
determination of the relevant time interval over which TCDD dose should be considered is
uncertain.

1

2

3

4

5

6

This document is a draft for review purposes only and does not constitute Agency policy.

B-37 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table B-29. Eskenazi et al., 2002b—Endometriosis

2

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration not satisfied. Results of a pilot study showed that ultrasounds had excellent
specificity and sensitivity for ovarian endometriosis.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design or
statistical analysis.

Response

Consideration not satisfied. More than half of the women were classified as 'uncertain' with
respect to endometriosis disease status.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of an
exposure-response relationship.

Response

Consideration not satisfied. While an increased risk of endometriosis was observed across the
3 TCDD categories, these risks were not statistically significant relative to the lowest exposure
category. The test for trend based on a continuous measure (logi0TCDD) was also not statistically
significant.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Criteria satisfied. Serum concentrations of TCDD offer improved exposure assessment, although
delineating the critical exposure window is challenging given the nature of the very high initial
exposure.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure adequate
statistical power.

Response

Consideration not satisfied. Only a total of 19 cases of endometriosis were identified.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion of
the strengths and limitations.

Response

Criteria satisfied. Environ Health Perspect 2002; 110(7)629-634. Author's highlight that this is
the first study to examine the relationship between TCDD and endometriosis, and the availability
of sera data to estimate TCDD levels. Limitations included the small number of women with
endometriosis, and inability to confirm disease status using laparoscopy. Finally, young women
may have been underrepresented due to cultural difficulties in examining women who had never
been sexually active.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response relationships
can be assessed.

Response

Criteria satisfied. Eligible study subjects were women between 1 month and 40 years of age at
time of accident. These analyses excluded virgins, those with Turner's syndrome, and women
who refused the examination of ultrasound. Serum data were available for the 601 participants on
which the analyses are based. Of these, 559 had serum measures taken in 1976/77, 25 between
1978 and 1981, and 17 women in 1996.

This document is a draft for review purposes only and does not constitute Agency policy.

B-38 DRAFT—DO NOT CITE OR QUOTE

-------
3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of exposure
are consistent with the current biological understanding of dose. The reported dose is consistent
with a toxicologically relevant dose. Latency and appropriate window(s) of exposure examined.
Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. TCDD exposure was estimated at the time of "conception attempt" using
serum measures, with extrapolation from 1976 levels using half-life assumptions. It is difficult to
identify the relevant time interval over which TCDD dose should be considered for dose-response
analysis. The critical window of exposure is unknown.





Conclusion

The lack of a statistically significant association coupled with a large number of women for which
endometriosis disease status was "uncertain", precludes the use of these data to conduct dose-
response analysis.

1

2

3	Table B-30. Eskenazi et al., 2003—Birth outcomes

4

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration not satisfied. Outcomes were identified through self-reported questionnaires.
Women were found to over-report birth weight, and have a tendency to underreport birth
defects in children. As a large number of women in Seveso underwent voluntary abortion in
the first year after the explosion, an awareness bias may have contributed to differential
reporting of pregnancy histories.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. See above.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. There was no association between spontaneous abortions and
logioTCDD, or with births small for gestational age. An inverse association with birth weight
was noted in first eight years following the accident as were the number of births small for
gestational age; however, none achieved statistical significance atp< 0.05.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Criteria satisfied. Serum concentrations of TCDD offer improved exposure assessment,
although delineating the critical exposure window is challenging given the nature of the very
high initial exposure.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. For spontaneous abortions there were 769 pregnancies. Fetal growth
and gestational age analysis was carried out on 608 singleton births that occurred post-
explosion.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

This document is a draft for review purposes only and does not constitute Agency policy.

B-39 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. Environ Health Perspect, 2003, lll(7):947-953. The authors highlight
potential limitation of reliance on self-reported data to ascertain pregnancy outcomes. They
also address the relevance of paternal exposures to TCDD on the developing fetus—such
exposure data were not considered in this study.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. A total of 745 women in the SWHS had reported getting pregnant, of these
510 women were pregnant after the explosion (888 pregnancies). Analyses of spontaneous
abortions based on 476 women (excludes those with voluntary abortion, ectopic pregnancy, or
molar pregnancy). TCDD measured for 413 women in 1976/77, 12 women between 1978 and
1981, and 1996 for 19 women.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. TCDD exposures were extrapolated to 1976 values. However, it is
difficult to identify the relevant time interval over which TCDD dose should be considered for
dose-response analysis.





Conclusion

The findings of the study are somewhat limited due to the reliance on self-reported information
for pregnancy outcomes, and lack of paternal exposures. The findings were not statistically
significant. Considered together, quantitative dose-response analyses for this study population
were not undertaken.

Table B-31. Warner et al., 2004—Age at menarche

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. In this study age at menarche was based on retrospective recall 5 to
19 years before the interview. Previous work suggests moderate to high correlations between
actual and recalled menarche, misclassification of outcome would bias risk estimates towards
the null (assuming nondifferential misclassification).

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Data collected from self-reported questionnaires allow for the
potential confounding influence of many risk factors to be taken into account. Some
misclassification of outcome may bias risk estimates towards the null.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. There was no association between TCDD levels and the age at
menarche with either the continuous or categorical measures of TCDD.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

This document is a draft for review purposes only and does not constitute Agency policy.

B-40 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. Serum concentrations of TCDD offer improved exposure assessment,
although delineating the critical exposure window is challenging given the nature of the very
high initial exposure.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. Cohort was large enough as analyses were performed using
282 women who wore piemennichenl nl Ihe lime of Ihe explosion

1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Environ Health Perspect, 2004, 112:1289-1292. Authors discuss use of
pooled serum from residents of the unexposed zone, and that those in lowest exposure group
had high exposures relative with contemporary levels for the area. Strengths of study include
use of serum to estimate TCDD exposure.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. The SWHS included women between 1 month and 40 years of age at time of
accident who attempted to get pregnant after the explosion (n = 463). This study is restricted
to those who were premenarcheal at the time of the explosion (n = 282). Serum was collected
for these women, primarily in 1976-1977 (n = 257), between 1978 and 1981 for 23, and in
1996-1997 for the 2 remaining women.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. TCDD exposures in 1976 were estimated by extrapolation serum levels
obtained after this date using the Filser model. Both categorical and continuous measures of
exposure were modeled. In utero measures of exposure are likely most relevant exposure
based on findings from animal studies.





Conclusion

No association between TCDD levels and age at menarche was found. There may be some
misclassification of age at menarche based on self-report, and biologically, the most relevant
dose as suggested by animal studies occurs in utero. Additionally, it is difficult to identify the
relevant time interval over which TCDD dose should be considered for dose-response analysis.
For these reasons, these data are not suited to a dose-response analysis.

Table B-32. Eskenazi et al., 2005—Age at menopause

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Outcome measures were obtained based on self-reported data
collected from questionnaires. Studies have shown that self-reports of age at menopause are
reported with accuracy and reliability, and among women with surgical menopause, the self-
reported age correlated well with that on the medical records.

This document is a draft for review purposes only and does not constitute Agency policy.

B-41 DRAFT—DO NOT CITE OR QUOTE

-------
2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Data obtained from the questionnaire allow for the potential
confounding influence of several potential confounders to be controlled for.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. Although risks of earlier menopause increased in the first
four quintiles, with a statistically significant trend, no increased risk was noted in the highest
exposure category (hazard ratio = 1.0 relative to lowest exposure group). Study authors
suggest this is due to the "inverted U" dose response often seen with hormonally active
compounds. Additionally, no statistically significant association was noted with logi0TCDD
for the individual quintiles.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Criteria satisfied. Serum concentrations of TCDD offer improved exposure assessment,
although delineating the critical exposure window is challenging given the nature of the very
high initial exposure.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. The study included 616 women. Of these, 260 were premenopausal,
169 classified as natural menopause, 83 as surgical menopause, 24 as impending menopause,
33 as premenopausal, and 58 in an "othef' category.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Environ Health Perspect, 113:858-862 (2005). Authors highlight this is
first study to look at relationship between dioxin and age at menopause. Other limitations of
the study include lowest exposure group (< 20.4 ppt) includes exposures level that are far
higher than background, and age at menopause was based on retrospective recall. Strength of
study is ability to characterize TCDD using serum measures.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. The Seveso Women's Health Study collected serum sample which allowed
TCDD exposures to be characterized. Those women (n = 616) who had not reached natural
menopause at the time of the accident were included in the study. Serum measures collected
in 1976/77 were available for 564 women, for 28 women, sera was collected between 1978
and 1981, while for 24 women, sera was collected in 1996/97.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. TCDD levels were estimated at the time of the explosion using available
information on TCDD half-life. However, it is difficult to identify the relevant time interval
over which TCDD dose should be considered for dose-response analysis. The critical window
of exposure can be estimated but is large and highly uncertain.





This document is a draft for review purposes only and does not constitute Agency policy.

B-42 DRAFT—DO NOT CITE OR QUOTE

-------
Conclusion

The findings do not provide strong support for a dose-response relationship. As such, they are
not well suited to a quantitative dose-response analysis.

Table B-33. Warner et al., 2007—Ovarian function

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Ovarian cyst analysis based on women who underwent ultrasound
(n = 310). Ovarian follicle analysis based on self-report on menstrual cycle and done in
women in preovulatory cycle (n = 96) at time of ultrasound. Hormonal analysis based on
women in last 14 days of cycle (n = 129).

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Data collected from self-reported questionnaires allow for the
potential confounding influence of many risk factors to be taken into account. Some
misclassification of outcome based on self-reports of menstrual cycle may bias risk estimates
towards the null.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. There was no association between serum TCDD levels and the
number or size of ovarian follicles. TCDD was also not associated wit the odds of ovulation.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Criteria satisfied. Serum concentrations of TCDD offer improved exposure assessment,
although delineating the critical exposure window is challenging given the nature of the very
high initial exposure.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. Cohort was large enough as analyses were performed using
129 women for ovulation outcome, and hormone analyses based on 87 women in luteal, and
55 in midluteal phases.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Environ Health Perspect, 2007,115:336-340. An important limitation cited
by the authors was that women may not have been exposed at critical period (prenatally).
Phases of the cycle may also have been misclassified as this was based on self-reported data.
Strength, first study to have examined ovarian function and TCDD exposures.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. The SWHS included women between 1 month and 40 years of age at time of
accident who were between 20-40 years of age and not using oral contraceptives at follow-up
(n = 363).Of these, serum was collected for 330 women between 1976 and 1977, between
1978 and 1982 for 25 women, and between 1996 and 1997 for 8 women.

This document is a draft for review purposes only and does not constitute Agency policy.

B-43 DRAFT—DO NOT CITE OR QUOTE

-------
3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. The women may not have been exposed at critical period (prenatally).





Conclusion

No association between TCDD levels and ovarian function was found. There may be some
misclassification of period of the cycle based on self-report, and biologically, the most relevant
dose as suggested by animal studies occurs in utero. For these reasons, these data are not
suited to a dose-response analysis.

Table B-34. Eskenazi et al., 2007—Uterine leiomyoma

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Outcomes were determined using two definitions: current fibroids, or
past diagnosis of fibroids. For past diagnosis of fibroids, self-reported data and medical
records were used to determine whether women were previously diagnosed with fibroids, these
were confirmed with medical records. A total of 25 women indicated they had never been
diagnosed with fibroids. Medical records indicate a past diagnosis for these women, and they
were classified as such. For current fibroids, this was determined at the time of the interview
for 634 women using transvaginal ultrasound examinations.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. In the SWHS questionnaires were administered to the participants and
detailed data for reproductive characteristics, smoking, body mass index, and alcohol use were
collected so risks could readily be adjusted for these covariates.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied, but inversely. An inverse dose-response pattern with the percentage of
women diagnosed (current and past history—combined) with fibroids across 3 categories of
exposure. Namely, the percentages of women with fibroids in the <20, 20.1-75.0, and
>75.0 ppt categories were 41.1%, 26.8%, and 20.0%, respectively.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. A variety of different exposure metrics were considered including
linear, categorical, splines, and logi0TCDD.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. A total of 251 women were found to have fibroids, and there were 62,
110, and 79 women with fibroids diagnosed in the 3 TCDD exposure categories.





1. Criicria

Study is published in the peer-reviewed scientific lilciaiuic and has an appropriate discussion
of the strengths and limitations.

This document is a draft for review purposes only and does not constitute Agency policy.

B-44 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. Am J Epidemiol, 2007, 166:79-87. In this study, the authors found an
inverse association between TCDD and uterine leiomyoma risk. The authors highlighted
strengths of the study that included the longitudinal design, serum measures taken at an
individual-level basis and most taken within 2 years of the accident, ability to include
outcomes among those who did not take an ultrasound by using an adapted statistical
approach. An important limitation that was the differences in risk by the stage of development
could not be assessed as all women were exposed postnatally, and only 4 cases were observed
among those who were premenarcheal at the time of exposure.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Final sample consisted of 956 women in the Seveso Women's Health Study
without a history of fibroids. For 872 of these women, serum was collected in 1976 and 1977.
For 56 women, TCDD was measured in women between 1978 and 1981, and for 28 women
the serum was collected in 1996.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. TCDD exposures were back extrapolated to expected levels in 1976 (at
the time of the accident). However, it is difficult to identify the relevant time interval over
which TCDD dose should be considered for dose-response analysis. The critical window of
exposure is unknown.





Conclusion

The data suggest an inverse (protective) effect between fibroids and exposure to TCDD. As
such, these data are not suited to further dose-response analyses.

B.2.5. Other Seveso Noncancer Studies

Table B-35. Mocarelli et al., 2008—Semen quality

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Serum levels of TCDD were measured on an individual basis for men
in exposed areas; pooled samples from men in uncontaminated areas were measured to assess
background TCDD exposure levels.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. While compliance rates may have introduced some possible bias, this
does not seem likely as different effects noted between the 22-31 and 32-39 year old age
groups. Information collected for other risks factors, which have been used as adjustment
factors in the models.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

This document is a draft for review purposes only and does not constitute Agency policy.

B-45 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration satisfied. Figure 3 suggests dose-response relationship among those aged 1-9 at
the time of the accident for sperm concentration and motility.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Serum concentrations of TCDD offer improved exposure assessment,
although delineating the critical exposure window is challenging.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. Analyses are based on 135 males exposed to TCDD.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Environmental Health Perspective s, 2008, 116(l):70-77. The authors
describe strengths associated with characterization of exposure (using serum samples), and
representativeness of study population. Limitation of study includes low compliance (but high
for semen sample studies), namely, 60% among a group of healthy men. The compliance rate
was higher among exposed group (69%).

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Involved males, <16 years old at time of accident.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria satisfied. TCDD exposures were based on serum samples. Serum samples were
drawn (in 1997/1998) from participants whose 1976 samples were above 15 ppt. Pooled
samples obtained in 1997/98 were used to describe background TCDD levels in
uncontaminated areas. The associated between TCDD exposure and semen quality was found
statistically significant for the boys with 1 and 9 years of age at the time of the accident. This
provides a critical window of exposure to estimate TCDD concentration.





Conclusion

Health outcomes are exposures are well characterized using serum data. However, the men
exposed between the ages of 1 and 9 to elevated TCDD levels had reduced semen quality
22 years later. It is difficult to discern whether this effect is a consequence of the initial high
exposure between 1 and 9 years of age or a function of the cumulative exposure for this entire
exposure window beginning at the early age. Nonetheless, quantitative dose-response analyses
for this outcome were conducted.

Table B-36. Mocarelli et al., 2000—Sex ratio

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Birth records examined for those who lived in parents who lived in
the area and who provided serum samples.

This document is a draft for review purposes only and does not constitute Agency policy.

B-46 DRAFT—DO NOT CITE OR QUOTE

-------
2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Paternal TCDD exposures were associated with an increased
probability of female births (p = 0.008).

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Serum samples were used to estimate maternal and paternal TCDD
levels. No discussion of exposure levels in reference population.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. Statistically significant findings achieved.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria not satisfied. The Lancet, 2000, 355:1858-1863. There is no discussion on the
strengths and limitations of this study.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum levels of TCDD were obtained from parents using samples provided
in 1976/77. Serum measures available for 296 mothers and 239 fathers.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. Serum based measures of TCDD were obtained shortly after the
accident. TCDD levels were also extrapolated to the time of conception. However, it is
difficult to identify the relevant time interval over which TCDD dose should be considered for
dose-response analysis. The critical window of exposure is unknown.





Conclusion

The data from this study demonstrate a positive dose-response relationship with paternal
TCDD levels at the time of the accident and increased likelihood for female births. However,
It is difficult to identify the relevant time interval over which TCDD dose should be
considered; specifically, it is difficult to discern whether this effect is a consequence of the
initial high exposure during childhood or a function of the cumulative exposure for this entire
exposure window beginning at the early age. Using the initial exposures in a dose-response
model would yield LOAELs that are too high to be relevant to factor into the RfD calculation.
Dose-response analysis for this outcome is, therefore, was not conducted.

1

2

3

4

5

This document is a draft for review purposes only and does not constitute Agency policy.

B-47 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table B-37. Baccarelli et al., 2008—Neonatal thyroid function

2

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Measures of b-TSH are taken using a standardized protocol 72 hours
after birth. These b-TSH measures are taken on all newborns born in the region of Lombardy of
which Seveso if a part of.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design or
statistical analysis.

Response

Consideration satisfied for component of the study based on plasma dioxin measures. For the
comparisons involving place of residence at the time of the accident, exposure misclassification is
likely given variability in soil TCDD exposure levels within these areas.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of an
exposure-response relationship.

Response

Consideration satisfied. Mean neonatal b-TSH was 0.98|iU/ml [0.90-1.08] in the reference area,
1,35|iU/ml [1.22-1.49] in zone B, and 1,66|iU/ml [1.19-2.31] in zone A (p < 0.001). The plotted
frequency distributions have similar shapes, but have shifted to the right for areas of higher
exposures. Neonatal b-TSH was correlated with current maternal plasma TCDD ((3-0.47,
p < 0.001) in the 51 newborns for which individual maternal serum TCDD values were available.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. TEQs were measured among the 38 women for which serum samples
were available and were defined for a mixture of dioxin-like compounds. Maternal mean total
TEQs (PCDDs, PCDFs, coplanar PCBs, and noncoplanar PCBs) was 41.8 ppt. Two measures of
exposure included place of residence at time of accident and plasma samples obtained from
mothers at the time of delivery. Similarities in positive dose-response relationships give stronger
weight to the findings.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure adequate
statistical power.

Response

Consideration satisfied for exposure metric that was based on 'place of residence'. For plasma
based estimate of maternal TCDD there were only 51 mother-child pairs. Only seven children in
total were found to have b-TSH levels in excess of 5 uU/ml; this implies limited statistical power
involving this health outcome.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion of
the strengths and limitations.

Response

Criteria satisfied. PLOS Medicine 2008; 5(7)1133-1142. The authors discuss the strength of the
study related to characterization of exposure using serum sampling, and ability to adjust for
factors related to b-TSH or TCDD levels (gender, birth weight, birth order, maternal age, hospital
and type of delivery). They also highlight that a limitation of study was that the influence of
mother-child dioxin transfer through colostrum could not be assessed because no information on
breastfeeding before b-TSH measurement was available.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response relationships
can be assessed.

This document is a draft for review purposes only and does not constitute Agency policy.

B-48 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. In the population-based study, eligible women who resided in zones A and B at
the time of the accident (n = 1,772) were matched to nonexposed women. In the study based on
plasma dioxin measurements, participants were the 51 children born to 38 women from zones A,
B, R, or a reference zone for which plasma dioxin measurements were available.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of exposure
are consistent with the current biological understanding of dose. The reported dose is consistent
with a toxicologically relevant dose. Latency and appropriate window(s) of exposure examined.
Response has to be a nonfatal endpoint.

Response

Criteria satisfied. Maternal TCDD levels were estimated at the time of delivery based on plasma
samples, and the critical window of exposure can be defined as the 9 month gestation period.





Conclusion

The data provide an opportunity for quantitative dose-response analyses.

1

2

3

4

Table B-38. Alaluusua et al., 2004—Oral hygiene

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Ascertainment of dental health was done blind to place of residence,
used standard protocol for caries developed by the WHO, and the clinical examination
supplemented by radiographic examination.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Additional risk factor information was collected on questionnaires.
These factors were considered as adjustment factors. Findings potentially susceptible to
participation biases.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Increased prevalence of developmental enamel effects found with
increased TCDD serum measures. Namely, prevalence in unexposed region was 26%,
whereas in the low, middle, and high TCCD groups the prevalence was 10, 40, and 60%,
respectively.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. TCDD exposure level based on serum lipids. No discussion of
exposure levels in reference population.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Criteria satisfied. Despite small numbers, statistically significant findings were achieved.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

This document is a draft for review purposes only and does not constitute Agency policy.

B-49 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. Environmental Health Perspectives, 2004, 112(13)1313-1318. Authors
mention two important strength of the study: characterization of TCDD exposure using serum
collected shortly after the time of the accident, and the fact that developmental defects are
permanent in nature. Therefore, they represent a health outcome can evaluated years later.
Little discussion was made of the impact of differential compliance rates between the exposed
(74%) and nonexposed (58%) groups. Authors mention two important strength of the study:
characterization of TCDD exposure using serum collected shortly after the time of the
accident, and the fact that developmental defects are permanent in nature. Therefore, they
represent a health outcome can evaluated years later. Little discussion was made of the impact
of differential compliance rates between the exposed (74%) and nonexposed (58%) groups.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum levels of TCDD could be estimated for children in exposed areas.
No serum levels were available for reference group of children, and assumption of zero
exposure was made. This seems reasonable.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria satisfied. It is difficult to discern whether this effect is a consequence of the initial
high exposure during childhood or a function of the cumulative exposure of the entire
exposure window beginning at early age. However, assumptions can be made regarding the
critical window of exposure and the relevant dose can be calculated.





Conclusion

The considerations for conducting a dose-response analysis have been satisfied with the study
population of only those subjects who lived in the ABR zone at the time of the accident;
exposure data are unavailable for those in the referent area. While is difficult to identify the
relevant time interval over which TCDD dose should be considered, quantitative
dose-response analysis for this outcome was conducted.

Table B-39. Bertazzi et al., 2001—Mortality (noncancer)

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied for some causes of death, but not others. Mortality appears to be well
captured from the vital statistics registries in the region (99% complete). Some health
outcomes (e.g., diabetes) are subject to misclassification using death certificate data.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Although individual-level data for individual risk factors are not
available, the potential for confounding is likely minimal. For e.g., independent surveys
suggests similarity between smoking behaviors across the regions. Exposure misclassification
based on place of residency likely to bias risk estimates towards the null.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

This document is a draft for review purposes only and does not constitute Agency policy.

B-50 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration not satisfied for most causes of death. An exception was the dose-response
relationship was observed for chronic obstructive pulmonary disease across Zones A, and B.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration not satisfied. Exposure classification was based on the address of the
residence on the date of the accident or when the person first entered the area. Although
TCDD blood levels were also measured, these were not examined with respect to health
outcomes. The lack of individual-level data also precluded an examination of these
uncertainties.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. A total of 494 noncancer deaths were found among residents of
Zones A, and B, respectively. This allowed examined of gender-specific effects.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Am J Epidemiol, 2001, 153:1031-1044. Authors discuss lack of
individual-level exposure data and other risk factors (e.g., smoking), difficulties in
extrapolating to background levels, diagnostic accuracy of using death certificates. Strengths
included similarities between exposed and comparison population for several risk factors,
completeness of follow-up, and consistent methods to identify mortality outcomes in the
exposed and comparison populations.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. Individual-level exposure data are unavailable. Exposure based on place
of residence at time of the explosion. Soil sampling performed indicated considerable
variability in TCDD levels within each region. In addition, place of residency at time of
explosion does not ensure individuals were at their home around the time of the accident.

3. Critieria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. An ecological measure of exposure (region of residency at time of
accident) was used to categorize individuals according to their possible exposure. Latencies
were considered. While such an approach has value for identifying whether excesses occurred
among highly exposed populations, it is not precise enough to conduct a quantitative dose-
response analysis. Furthermore, noncancer mortality is not a viable endpoint to consider for
further dose-response analysis.





Conclusion

Study is not suitable for dose-response analysis due to mortality as endpoint and lack of
individual-level exposure data.

1

2

3

4

5

This document is a draft for review purposes only and does not constitute Agency policy.

B-51 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table B-40. Consonni et al., 2008—Mortality (noncancer)

2

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied for some causes of death, but not others. Mortality appears to be well
captured from the vital statistics registries in the region (99% complete). Some health
outcomes (e.g., diabetes) are subject to misclassification using death certificate data.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Although individual-level data for individual risk factors are not
available, the potential for confounding is likely minimal. For e.g., information from other
independent surveys suggests similarity between smoking behaviors across the regions.
Exposure misclassification based on place of residency is likely to bias risk estimates towards
the null.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. Statistically significant association noted in most highly exposed
area for chronic rheumatic disease and chronic obstructive pulmonary disease. Dose-response
pattern noted across Zones A, B and R for circulatory disease mortality 5-9 years after the
accident.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration not satisfied. Lack of individual-level data precludes an examination of these
uncertainties.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied for some causes of death but not others. For example, only
three deaths from diabetes occurred among residents of Zone A. The limitation related to
statistical power is exacerbated for stratified analyses carried out by number of years since the
accident.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Am J Epidemiol, 2008, 167:847-858. Authors discuss potential for
selection bias, limitation of residential based measure of exposure, similarities of mortality
ascertainment in exposed and referent populations, and multiple testing.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. Individual-level exposure data are unavailable. Exposure based on place
of residence at time of the explosion. Soil sampling performed indicated considerable
variability in TCDD levels within each region. In addition, place of residency at time of
explosion does not ensure individuals were at their home around the time of the accident.

This document is a draft for review purposes only and does not constitute Agency policy.

B-52 DRAFT—DO NOT CITE OR QUOTE

-------
3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. An ecological measure of exposure (region of residency at time of
accident) was used to categorize individuals according to their possible exposure. Latencies
were considered. While such an approach has value for identifying whether excesses occurred
among highly exposed populations, it is not precise enough to conduct a quantitative
dose-response analysis. Furthermore, noncancer mortality is not a viable endpoint to consider
for further dose-response analysis.





Conclusion

Study is not suitable further dose-response evaluation due to noncancer moialils ciidpoml

Table B-41. Baccarelli et al., 2005—Chloracne

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Chloracne cases identified using standardized criteria.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Plasma TCDD was associated with an increased risk of chloracne.
The odds ratios increased in a dose-response pattern across zone of residence.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Authors discussed implications of differential elimination rates by age
and body growth.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. A total of 101 chloracne cases were identified, and 211 controls were
selected. Statistically significant findings were observed in several comparisons.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. British Journal of Dermatology, 2005, 152,459-465. The authors detail the
limited statistical power they had available in the study. They also highlight a strength of the
study that included uniqueness of age and sex distribution of chloracne cases, characterization
of TCDD that could be done using sera samples, and availability of both clinical and
epidemiological data.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

This document is a draft for review purposes only and does not constitute Agency policy.

B-53 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria satisfied. TCDD was estimated in both chloracne cases and control using serum
measures.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria satisfied. Serum based measures of TCDD were obtained shortly after the accident.
Chloracne is thought to be caused by the initial high exposure.





Conclusion

Exposure to TCDD at sufficiently high levels is recognized to cause chloracne. This study
provides limited relevance to dose-response modeling of TCDD as exposure levels typically
observed in the general population are much lower.

Table B-42. Baccarelli et al, 2002 and 2004—Immunological effects

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Common methods were used to describe blood levels of plasma
immunoglobulins (IgA, IgG, and IgM) and complement components (C3 and C4).

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Both exposure and outcome were objectively and accurately measured.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. Plasma IgG levels were inversely related with TCDD.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Both categorical (quintiles) and continuous measures of TCDD were
examined in the dose-response analysis.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. Analyses are made using 72 highly exposed, and 72 low exposed
individuals.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Toxicology letters, 2004, 149:287-293 and Environ Health Perspect, 2002,
110(12): 1169-1173. The authors highlight that few studies have looked at immunological
effects of TCDD in humans, that the current study was able to exclude those with concurrent
medical conditions, and the ability to characterize exposure using serum measures. Limitations
addressed were the uncertainty about the clinical relevance of the dose-response pattern found,
and the relatively small size of the study population.

This document is a draft for review purposes only and does not constitute Agency policy.

B-54 DRAFT—DO NOT CITE OR QUOTE

-------
2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. A total of 120 subjects were included in the study. This included
62 randomly selected from the high exposed zone, and 58 selected from the reference area.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of exposure
are consistent with the current biological understanding of dose. The reported dose is
consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. Dose-response relationships were examined using current TCDD levels.
However, it is difficult to identify the relevant time interval over which TCDD dose should be
considered for dose-response analysis.





Conclusion

An inverse dose-response association between IgG and TCDD was observed, however, because
the relationship can not be described in terms of clinical relevance with respect to a specific
health outcome, it is our view that these data are not suited to dose-response modeling.

1

2

3

4

5

6

7

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration cannot be evaluated. Insufficient details are provided in the paper to gauge the
completeness and coverage of the cancer registry and mortality data. Health outcomes were
studied on the basis of information in the official medical statistics

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. It is an ecological study.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration cannot be evaluated. Dose-response was not evaluated as exposure was based on
residency in the region vs. no residency.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration not satisfied. No individual-level exposure estimates were used.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. Population-based data over several years were used to make ecological
comparisons.





This document is a draft for review purposes only and does not constitute Agency policy.

B-55 DRAFT—DO NOT CITE OR QUOTE

B.2.6. Chapaevsk Study

Table B-43. Revich et al., 2001—Mortality (noncancer) and reproductive
health

-------
1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Published in Chemosphere, 2001, 43(4-7):951-966.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. It is a cross-sectional study that compares mortality rates between
regions. No individual-level exposure data available.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of exposure
are consistent with the current biological understanding of dose. The reported dose is
consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. No exposure estimates were used in the study.





Conclusion

These cancer data are cross-sectional in nature and not appropriate for a dose-response analysis.

B.2.7. Air Force Health ("Ranch Hands") Study

Table B-44. Michalek and Pavuk, 2008—Diabetes

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. Prevalent diabetes identified from medical records from repeated
medical check-ups. Preferred method of ascertaining outcome relative to use of death
certificates.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. Adjustment was made for a number of risk factors related to
diabetes (e.g., BMI, family history, smoking). However, Agent Orange was a 50% mixture of
2,4-D and TCDD; therefore, potential for confounding by other coexposures is likely.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration satisfied. The RR for an increase in 10 units was 1.29 (p < 0.001), and the risks
across the background, low and high exposure categories, relative to the unexposed were 0.86,
1.45, and 1.68.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Initial TCDD dose were estimated at the end of the tour of duty for
the Ranch Hands. Individual-level serum dioxin measurements correlated well with correlated
with days of spraying and calendar period of service, but collection of the samples roughly
20 years later required back-extrapolation.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

This document is a draft for review purposes only and does not constitute Agency policy.

B-56 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Consideration satisfied. There were a total of 439 cases of diabetes identified.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. J Occup Environ Medicine, 2008, 50:330-340. The authors address
strengths and limitations related to the accuracy of the one-compartment pharmacokinetic
model, impact of the covariate time spent in Southeast Asia, and potential exposure
misclassification on days sprayed.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. TCDD estimates were derived using serum samples. However, Ranch
Hand veterans were exposed to other compounds in the herbicides, such as 2,4-D.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria satisfied. TCDD levels at the end of service were estimated. Extrapolation was done
using a half-life of 7.6 years. Exposures were grouped into comparison, background, low and
high. This allows for a shape of the dose-response curve to be evaluated. A continuous
measure of TCDD was also examined (logi0TCDD).





Conclusion

Ranch Hand veterans were exposed to other contaminants in the herbicides that were mixed,
thereby making it difficult to determine independent effects of TCDD on diabetes. In our
view, this limitation precludes dose-response modeling of TCDD and diabetes using data from
this cohort.

1

2

3

4

5

6

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. The New Zealand Health Information Service Mortality Collection
and the Registrar-General's Index to Deaths were used to identify deaths. Additional searches
were based on the last known address from the work record; the electoral roll and the
habitation index; the telephone book; the internet; and Terranet property information database.
An additional search was carried out through the Births, Deaths, and Marriages office of the
New Zealand Department of Internal Affairs. Lastly, automated personnel and pension
records were also used to locate past New Plymouth workers and identify some deaths.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration satisfied. Workers lost to follow-up were an unlikely source of bias especially
for internal analyses. Confounding by other coexposures (e.g., 2,4,6-TCP) unlikely to have
resulted in bias, due to presumed poor correlation with TCDD.

This document is a draft for review purposes only and does not constitute Agency policy.

B-57 DRAFT—DO NOT CITE OR QUOTE

B.2.8. Other Noncancer Studies of Dioxin

Table B-45. McBride et al., 2009a—Mortality (noncancer)

-------
3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. There was no cause of death among those considered for which a
dose-response trend was observed across four exposure categories of TCDD.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Dichotomous exposure (exposed/unexposed) and duration of
employment were examined from job exposure classification assessed via occupational history
records industrial hygienists/factory personnel knowledge and questionnaires.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration not satisfied.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Published in J Occup Environ Med, 2009, 51:1049-1056. The other studies
in the cohort highlight the 22% of the cohort lost to follow-up, the limited size of the cohort
tissue sarcomas, differences in cohort definitions between sprayers and producers, and the
potential for other exposures during employment at the plant.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria satisfied. Serum measures available for 346 workers were used to derive TCDD
exposures for the entire cohort using the area under the curve approach.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. Dichotomous exposure assessment did not allow individual estimates of
dose to be developed. However, noncancer mortality is not a viable endpoint to consider for
further dose-response analysis.





Conclusion

A considerable portion of the cohort was lost to follow-up, and no dose-response associations
noted. As a result, the data are not suited to dose-response analysis.

1

2

This document is a draft for review purposes only and does not constitute Agency policy.

B-58 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table B-46. McBride et al., 2009b—Mortality (noncancer)

2

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration satisfied. The New Zealand Health Information Service Mortality Collection
and the Registrar-General's Index to Deaths were used to identify deaths. Additional searches
were based on the last known address from the work record; the electoral roll and the
habitation index; the telephone book; the internet; and Terranet property information database.
An additional search was carried out through the Births, Deaths, and Marriages office of the
New Zealand Department of Internal Affairs. Lastly, automated personnel and pension
records were also used to locate past New Plymouth workers and identify some deaths.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. Considerable amount of workers were lost to follow up (22%),
but it is unclear if bias resulted. The dichotomous exposure measure was based on exposure
to TCDD, chlorinated dioxins and phenoxy herbicides, so confounding is a possibility by
these coexposures.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. Because no individual exposure estimates were available for
these analyses, dose-response could not be evaluated.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration satisfied. Consideration satisfied. Dichotomous exposure
(exposed/unexposed) and duration of employment were examined from job exposure
classification assessed via occupational history records industrial hygienists/factory personnel
knowledge and questionnaires. Authors discuss limitations in the assignment of exposure
among cohort members.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

Response

Criteria satisfied. Published in Occup Medicine, 2009, 59(4):255-263. The authors highlight
cohort lost to follow-up, the limited size of the cohort, differences in cohort definitions
between sprayers and producers, and the potential for other exposures during employment at
the plant.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. Exposures were not quantified. The dichotomous exposure measure
was based on exposure to TCDD, chlorinated dioxins and phenoxy herbicides.

This document is a draft for review purposes only and does not constitute Agency policy.

B-59 DRAFT—DO NOT CITE OR QUOTE

-------
3. Critiera

The effective dose and oral exposure can be reasonably estimated and the measures of
exposure are consistent with the current biological understanding of dose. The reported dose
is consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Effective dose could not be estimated given the lack of individual-level exposure data.
Noncancer mortality is not a viable endpoint to consider for further dose-response analysis.





Conclusion

The study lacks the quantification of exposures at an individual level, and a considerable
portion of the cohort was lost to follow-up. As a result, the data are not suited to
dose-response analysis.

Table B-47. Ryan et al., 2002—Sex ratio

1. Consideration

Methods used to ascertain health outcomes identified were unbiased, highly sensitive, and
specific.

Response

Consideration not satisfied. Company records were used to identify births, the date of birth,
and the sex of the child. No information was provided on the expected completeness of
identifying births in this manner. Moreover, the study was expanded to include workers who
heard about the study in a public forum. Therefore, the study could be influenced by
participation bias.

2. Consideration

Risk estimates are not susceptible to biases from confounding exposures or from study design
or statistical analysis.

Response

Consideration not satisfied. See above.

3. Consideration

Study demonstrates an association between TCDD and adverse health effect with evidence of
an exposure-response relationship.

Response

Consideration not satisfied. The study compared birth ratios among men and women employed
at the plant to the general population. No categories of exposure were examined.

4. Consideration

Exposure assessment methodology is clear and adequately characterizes individual-level
exposures. The limitations and uncertainties in the exposure assessment are considered.

Response

Consideration not satisfied. This is not relevant as no analyses were done in relation to
exposure levels.

5. Consideration

Study size and follow-up are large enough to yield precise estimates of risk and ensure
adequate statistical power.

Response

Consideration satisfied. For the categories of exposure used (yes/no), and the stratified
analyses by sex and subcohort, the study allows for the birth ratios to be estimated with
sufficient precision.





1. Criteria

Study is published in the peer-reviewed scientific literature and has an appropriate discussion
of the strengths and limitations.

This document is a draft for review purposes only and does not constitute Agency policy.

B-60 DRAFT—DO NOT CITE OR QUOTE

-------
Response

Criteria not satisfied. Published in Environ Health Perspect, 2002, 110(11):A699-A701. The
authors discussed the limitations of using serum collected many years after they stopped
working to estimate TCDD exposures when the preferred metric would be TCDD levels at the
time of conception. They did not address issues about the representativeness of the study
participants to the entire cohort of workers, nor did they address the limitation of not being able
to conduct dose-response analyses using individual-level TCDD data.

2. Criteria

Exposure must be primarily TCDD and is properly quantified so that dose-response
relationships can be assessed.

Response

Criteria not satisfied. While serum measures were available for 84 of the 198 participants of
the study, birth ratios were compared between the cohort of 2,4,5-T and 2,4,5-trichlorphgenol
workers relative to the city of Ufa. There was no attempt to derive birth ratios in relation to
exposure levels. The serum data were only used to demonstrate that these workers, on average,
had TCDD levels 30 times higher than Ufa residents.

3. Criteria

The effective dose and oral exposure can be reasonably estimated and the measures of exposure
are consistent with the current biological understanding of dose. The reported dose is
consistent with a toxicologically relevant dose. Latency and appropriate window(s) of
exposure examined. Response has to be a nonfatal endpoint.

Response

Criteria not satisfied. TCDD exposures were based on serum measures taken in some cases
many years after children were born; no attempt was made to back-extrapolate to the time of
conception.





Conclusion

The data are not suitable for dose-response modeling. Risk estimates have not been derived in
relation to TCDD exposure levels. There exist uncertainties about the representativeness of the
participants in relation to the cohort as a whole, and insufficient details are provided to evaluate
the extent in which all births were identified. While these data should not be used for
quantitative dose-response modeling, the much lower M/F birth ratio among exposed fathers is
consistent with the finding by Mocarelli et al, and lends support to those findings.

1

2

3	B.3. REFERENCES

4	Akhtar, FZ; Garabrant, DH; Ketchum, NS; et al. (2004) Cancer in US Air Force veterans of the Vietnam War. J

5	Occup Environ Med 46(2): 123-136.

6	Alaluusua, S; Calderara, P; Gerthoux, PM; et al. (2004) Developmental dental aberrations after the dioxin accident

7	inSeveso. Environ Health Perspect 112(13): 1313-1318.

Aylward, LL; Brunet, RC; Starr, TB; et al. (2005a) Exposure reconstruction for the TCDD-exposed NIOSH cohort
using a concentration- and age-dependent model of elimination. Risk Anal 25(4):945-956.

Aylward, LL; Brunet, RC; Carrier, G; et al. (2005b) Concentration-dependent TCDD elimination kinetics in
humans: toxicokinetic modeling for moderately to highly exposed adults from Seveso, Italy, and Vienna, Austria,
and impact on dose estimates for the NIOSH cohort. J Expo Anal Environ Epidemiol 15(1):51-65.

8	Baccarelli, A; Mocarelli, P; Patterson, DG, Jr.; et al. (2002) Immunologic effects of dioxin: new results from Seveso

9	and comparison with other studies. Environ Health Perspect 110(12): 1169—1173.

10	Baccarelli, A; Pesatori, AC; Masten, SA; et al. (2004) Aryl-hydrocarbon receptor-dependent pathway and toxic

11	effects of TCDD in humans: a population-based study in Seveso, Italy. Toxicol Lett 149(l-3):287-293.

This document is a draft for review purposes only and does not constitute Agency policy.

B-61 DRAFT—DO NOT CITE OR QUOTE

-------
1	Baccarelli, A; Pesatori, AC; Consonni, D; et al. (2005) Health status and plasma dioxin levels in chloracne cases

2	20 years after the Seveso, Italy accident. Br J Dermatol 152(3):459-465.

3	Baccarelli, A; Hirt, C; Pesatori, AC; et al. (2006) t(14; 18) translocations in lymphocytes of healthy dioxin-exposed

4	individuals from Seveso, Italy. Carcinogenesis 27(10):2001-2007.

5	Baccarelli, A; Giacomini, SM; Corbetta, C; et al. (2008) Neonatal thyroid function in Seveso 25 years after maternal

6	exposure to dioxin. PLoS Med 5(7): 1133-1142.

7	Becher, H; Steindorf, K; Flesch-Janys, D. (1998) Quantitative cancer risk assessment for dioxins using an

8	occupational cohort. Environ Health Perspect 106(Suppl 2):663-670.

9	Bertazzi, PA; Consonni, D; Bachetti, S; et al. (2001) Health effects of dioxin exposure: a 20-year mortality study.

10	Am J Epidemiol 153(11): 1031-1044.

11	Cheng, H; Aylward, L; Beall, C; et al. (2006) TCDD exposure-response analysis and risk assessment. Risk Anal

12	26:1059-1071.

13	Collins, JJ; Bodner, K; Aylward, LL; et al. (2009) Mortality rates among trichlorophenol workers with exposure to

14	2,3,7,8-tetrachlorodibenzo-p-dioxin. Am J Epidemiol 170(4):501-506.

15	Consonni, D; Pesatori, AC; Zocchetti, C; et al. (2008) Mortality in a population exposed to dioxin after the Seveso,

16	Italy, accident in 1976: 25 years of follow-up. Am J Epidemiol 167(7):847-858.

17	Eskenazi, B; Warner, M; Mocarelli, P; et al. (2002a) Serum dioxin concentrations and menstrual cycle

18	characteristics. Am J Epidemiol 156(4):383-392.

19	Eskenazi, B; Mocarelli, P; Warner, M; et al. (2002b) Serum dioxin concentrations and endometriosis: a cohort study

20	in Seveso, Italy. Environ Health Perspect 110(7):629-634.

21	Eskenazi, B; Mocarelli, P; Warner, M; et al. (2003) Maternal serum dioxin levels and birth outcomes in women of

22	Seveso, Italy. Environ Health Perspect 111(7), 947-953.

23	Eskenazi, B; Warner, M; Marks, AR; et al. (2005) Serum dioxin concentrations and age at menopause. Environ

24	Health Perspect 113(7):858-862.

25	Eskenazi, B; Warner, M; Samuels, S; et al. (2007) Serum dioxin concentrations and risk of uterine leiomyoma in the

26	Seveso Women's Health Study. Am J Epidemiol 166(l):79-87.

27	Fingerhut, MA; Halperin, WE; Marlow, DA; et al. (1991) Cancer mortality in workers exposed to

28	2,3,7,8-tetrachlorodibenzo-p-dioxin. N Engl J Med 324(4):212-218.

29	Flesch-Janys, D; Berger, J; Gurn, P; et al. (1995) Exposure to polychlorinated dioxins and furans (PCDD/F) and

3 0	mortality in a cohort of workers from a herbicide-producing plant in Hamburg, Federal Republic of Germany. Am J

31	Epidemiol 142(11): 1165-1175.

32	Flesch-Janys, D; Becher, H; Gurn, P; et al. (1996) Elimination of polychlorinated dibenzo-p-dioxins and

33	dibenzofurans in occupationally exposed persons. J Tox Environ Health 47(4):363-378.

34	Flesch-Janys, D; Steindorf, K; Gurn, P; et al. (1998) Estimation of the cumulated exposure to polychlorinated

3 5	dibenzo-p-dioxins/furans and standardized mortality ratio analysis of cancer mortality by dose in an occupationally

36	exposed cohort. Environ Health Perspect 106(Suppl 2):655-662.

37	Hooiveld, M; Heederik, DJ; Kogevinas, M; et al. (1998) Second follow-up of a Dutch cohort occupationally

38	exposed to phenoxy herbicides, chlorophenols, and contaminants. Am J Epidemiol 147(9):891-901.

This document is a draft for review purposes only and does not constitute Agency policy.

B-62 DRAFT—DO NOT CITE OR QUOTE

-------
1	Manz, A; Berger, J; Dwyer, JH; et al. (1991) Cancer mortality among workers in chemical plant contaminated with

2	dioxin. Lancet 338(8773):959-964.

3	McBride, DI; Collins, JJ; Humphry, NF; et al. (2009a) Mortality in workers exposed to

4	2,3,7,8-tetrachlorodibenzo-p-dioxin at a trichlorophenol plant in New Zealand. J Occup Environ Med

5	51(9):1049-1056.

6	McBride, DI; Burns, CJ; Herbison, GP; et al. (2009b) Mortality in employees at a New Zealand agrochemical

7	manufacturing site. Occup Med (Oxford, England) 59(4):255-263.

8	Michalek, JE; Pavuk, M. (2008) Diabetes and cancer in veterans of Operation Ranch Hand after adjustment for

9	calendar period, days of spraying, and time spent in Southeast Asia. J Occup Environ Med 50(3):330-340.

10	Mocarelli, P; Gerthoux, PM; Ferrari, E; et al. (2000) Paternal concentrations of dioxin and sex ratio of offspring.

11	Lancet 355(9218): 1858-1863.

12	Mocarelli, P; Gerthoux, PM; Patterson, DG, Jr.; et al. (2008) Dioxin exposure, from infancy through puberty,

13	produces endocrine disruption and affects human semen quality. Environ Health Perspect 116(l):70-77.

14	Ott, MG; Zober, A. (1996) Cause specific mortality and cancer incidence among employees exposed to

15	2,3,7,8-TCDD after a 1953 reactor accident. Occup Environ Med 53(9):606-612.

16	Pesatori, AC; Consonni, D; Bachetti, S; et al. (2003) Short- and long-term morbidity and mortality in the population

17	exposed to dioxin after the "Seveso accident". Ind Health 41(3): 127—138.

18	Revich, B; Aksel, E; Ushakova, T; et al. (2001) Dioxin exposure and public health in Chapaevsk, Russia.

19	Chemosphere 43(4-7):951-966.

20	Ryan, JJ; Amirova, Z; Carrier, G. (2002) Sex ratios of children of Russian pesticide producers exposed to dioxin.

21	Environ Health Perspect, 110(11):A699-701.

22	Steenland, K; Piacitelli, L; Deddens, J; et al. (1999) Cancer, heart disease, and diabetes in workers exposed to

23	2,3,7,8-tetrachlorodibenzo-p-dioxin. J Natl Cancer I 91(9):779-786.

24	Steenland, K; Deddens, J; Piacitelli, L. (2001) Risk assessment for 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)

25	based on an epidemiologic study. Am J Epidemiol 154(5):451-458.

26	't Mannetje, A; McLean, D; Cheng, S; et al. (2005) Mortality in New Zealand workers exposed to phenoxy

27	herbicides and dioxins. Occup Environ Med 62(l):34-40.

28	Warner, M; Eskenazi, B; Mocarelli, P; et al. (2002) Serum dioxin concentrations and breast cancer risk in the

29	Seveso Women's Health Study. Environ Health Perspect 110(7):625-628.

30	Warner, M; Samuels, S; Mocarelli, P; et al. (2004) Serum dioxin concentrations and age at menarche. Environ

31	Health Perspect 112(13): 1289-1292.

32	Warner, M; Eskenazi, B; Olive, DL; et al. (2007) Serum dioxin concentrations and quality of ovarian function in

33	women of Seveso. Environ Health Perspect 115(3):336-340.

34	Zober, A; Messerer, P; Huber, P. (1990) Thirty-four-year mortality follow-up of BASF employees exposed to

35	2,3,7,8-TCDD after the 1953 accident. Int Arch Occup Environ Health 62(2): 139-157.

This document is a draft for review purposes only and does not constitute Agency policy.

B-63 DRAFT—DO NOT CITE OR QUOTE

-------
[This page intentionally left blank.]

-------
DRAFT

DO NOT CITE OR QUOTE

May 2010
External Review Draft

APPENDIX C
Kinetic Modeling

NOTICE

THIS DOCUMENT IS AN EXTERNAL REVIEW DRAFT. It has not been formally released
by the U.S. Environmental Protection Agency and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its technical accuracy and policy
implications.

National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH

-------
CONTENTS—APPENDIX C: Kinetic Modeling

APPENDIX C. kinetic Modeling	C-l

C. 1. LITERATURE SEARCH STRATEGY AND RESULTS—IDENTIFYING
RECENT PUBLICATIONS FOR UPDATING TCDD TOXICOKINETIC

MODEL INPUT PARAMETERS	C-l

C.l.l. Data Bases Searched	C-l

C.1.2. Literature Search Strategy and Approach	C-2

C.1.2.1. Chemical Search Terms—DIALOG Search	C-2

C.1.2.2. Primary Search Terms (Species)—DIALOG Search	C-2

C.1.2.3. Secondary Search Terms (Toxicology)—DIALOG Search	C-3

C.1.3. Citation Screening Procedures and Results	C-3

C. 1.4. References Selected for More Detailed Review for Updating the PBPK

Models	C-6

C.1.5. Brief Descriptions of DIALOG Bibliographic Data Bases Searched	C-8

C.2. TOXICOKINETIC MODELING CODE	C-l 1

C.2.1. Human Standard Model	C-ll

C.2.1.1. Model Code	C-ll

C.2.1.2. Input File	C-l9

C.2.2. Human Gestational Model	C-20

C.2.2.1. Model Code	C-20

C.2.2.2. Input File	C-3 1

C.2.3. Rat Standard Model	C-32

C.2.3.1. Model Code	C-32

C.2.3.2. Input Files	C-40

C.2.4. Rat Gestational Model	C-55

C.2.4.1. Model Code	C-55

C.2.4.2. Input Files	C-65

C.2.5. Mouse Standard Model	C-73

C.2.5.1. Model Code	C-73

C.2.5.2. Input Files	C-81

C.2.6. Mouse Gestational Model	C-85

C.2.6.1. Model Code	C-85

C.2.6.2. Input Files	C-96

C.3. TOXICOKINETIC MODELING RESULTS FOR KEY ANIMAL BIOASSAY

STUDIES	C-98

C.3.1. Nongestational Studies	C-98

C.3.1.1. Cantoni etal. (1981)	C-98

C.3.1.2. Chu et al. (2007)	C-l00

C.3.1.3. Crofton etal. (2005)	C-l02

C.3.1.4. Delia Porta et al. (2001) (female)	C-105

C.3.1.5. Delia Porta etal. (2001) (male)	C-l 07

C.3.1.6. Fattore et al. (2000)	C-l08

C.3.1.7. Franc et al. (2001) Sprague Dawley Rats	C-110

C.3.1.8. Franc etal. (2001) Long-Evans Rats	C-l 11

C.3.1.9. Franc etal. (2001) Hans Wi star Rats	C-113

This document is a draft for review purposes only and does not constitute Agency policy.

C-ii DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

C.3.1.10. Hassoun et al. (2000)	C-115

C.3.1.11. Hutt et al. (2008)	C-117

C.3.1.12. Kitchin and Woods (1979)	C-118

C.3.1.13. Kociba et al. (1976)	C-121

C.3.1.14. Kociba etal. (1978) Female	C-123

C.3.1.15. Kociba etal. (1978) Male	C-125

C.3.1.16. Latchoumycandane and Mathur (2002)	C-127

C.3.1.17. Li etal. (1997)	C-128

C.3.1.18. Murray et al. (1979) Adult Portion	C-132

C.3.1.19. NTP (1982)—Female Rats, Chronic	C-133

C.3.1.20. NTP (1982)—Male Rats, Chronic	C-135

C.3.1.21. NTP (1982)—Female Mice, Chronic	C-137

C.3.1.22. NTP (1982)—Male Mice, Chronic	C-138

C.3.1.23. NTP (2006) 14 Weeks	C-140

C.3.1.24. NTP (2006) 31 Weeks	C-142

C.3.1.25. NTP (2006) 53 Weeks	C-144

C.3.1.26. NTP (2006) 2 Years	C-146

C.3.1.27. Sewall et al. (1995)	C-148

C.3.1.28. Shi etal. (2007) Adult Portion	C-150

C.3.1.29. Smialowicz et al. (2008)	C-151

C.3.1.30. Toth et al., 1 Year (1979)	C-153

C.3.1.31. Van Birgelen etal. (1995)	C-155

C.3.1.32. Yanden IIeuvel etal. (1994)	C-157

C.3.1.33. White etal. (1986)	C-160

C.3.2. Gestational Studies	C-163

C.3.2.1. Bell et al. (2007)	C-163

C.3.2.2. Haavisto et al. (2006)	C-164

C.3.2.3. Hojo et al. (2002)	C-166

C.3.2.4. Ikeda et al. (2005)	C-167

C.3.2.5. Kattainen et al. (2001)	C-168

C.3.2.6. Keller et al. (2007)	C-170

C.3.2.7. Li et al. (2006) 3-Day	C-171

C.3.2.8. Markowski et al. (2001)	C-173

C.3.2.9. Mietinnen et al. (2006)	C-174

C.3.2.10. Nohara et al. (2000)	C-176

C.3.2.11. Ohsako et al. (2001)	C-177

C.3.2.12. Schantzetal. (1996) and Amin et al. (2000)	C-179

C.3.2.13. Seo etal. (1995)	C-180

C.4. RESPONSE SURFACE TABLES	C-183

C.4.1. Nongestational Lifetime	C-184

C.4.2. Nongestational 5-Year Average	C-192

C.4.3. Gestational	C-198

C.5. REFERENCES	C-204

This document is a draft for review purposes only and does not constitute Agency policy.

C-iii DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

APPENDIX C. KINETIC MODELING

C.l. LITERATURE SEARCH STRATEGY AND RESULTS—IDENTIFYING RECENT
PUBLICATIONS FOR UPDATING TCDD TOXICOKINETIC MODEL INPUT
PARAMETERS

The purpose of this literature search was to identify recent publications that address the
input parameters for the physiologically based pharmacokinetic (PBPK) models Aylward and
colleagues (described in articles published in 2005 and 2009) and Emond and colleagues
(described in articles published in 2004, 2005, and 2006). This literature search was part of the
U.S. Environmental Protection Agency (EPA)'s preparation of a response to the National
Academy of Sciences' review (Health Risks from Dioxin and Related Compounds: Evaluation of
the EPA Reassessment, NAS, 2006]) of EPA Exposure and Human Health Reassessment of
2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds (U.S. EPA, 2003), herein
called the "2003 Reassessment." English-only references from 2003 to May 2009 were searched
using bibliographic data bases relevant to health effects and toxicology of
2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). The search focused on toxicokinetic data that
could be used to update the dynamic disposition of 2,3,7,8-TCDD in mice, rats, guinea pigs,
monkeys, and humans.

In the primary search, EPA identified 775 distinct citations based on the literature search
criteria described below. EPA also performed an independent supplemental search to avoid
missing key studies. EPA identified 28 papers for further analysis that appeared on first review
to report data to update the input parameters of the Aylward and Emond PBPK models;
considerations for selection are described in Section C.1.3.

C.l.l. Data Bases Searched

EPA used the following DIALOG bibliographic data bases in the primary search. Brief
descriptions of the DIALOG data bases searched are provided in Section C.l.5.

1.	File 6: NTIS

2.	File 41: Pollution Abstracts

3.	File 55: Biosis

4.	File 153: IPA Toxicology

This document is a draft for review purposes only and does not constitute Agency policy.

C-l DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

5.	File 155: MedLine

6.	File 156: ToxFile

7.	File 157: Biosis Toxicology

8.	File 159: CancerLit

9.	File 336: RTECS

The PUB MED data base was used for the supplemental search.

C.1.2. Literature Search Strategy and Approach

The primary search used a tiered key-word approach, as documented below. The
principal search term was the Chemical Abstract Service Registry Number (CASRN) or specific
chemical name, 2,3,7,8-tetrachlorodibenzo-p-dioxin or 2,3,7,8-TCDD. The next tier of search
terms was species, and finally toxicokinetic keywords, as listed below. The period of the search
was 2003 through May 2009, and articles were limited to English language.

The supplemental PUB MED search was limited to the most recent five years (2004 to
present) and used four combinations of key words:

•	TCDD + pharmacokinetic + humans,

•	TCDD + toxicokinetic + humans,

•	TCDD + pharmacokinetic + animals, and

•	TCDD + toxicokinetic + animals.

C.l.2.1. Chemical Search Terms—DIALOG Search

•	CASRN: 1746-01-6

•	2,3,7,8 -tetrachl orodib enzo-p-di oxin

•	2,3,7,8-TCDD

C. 1.2.2. Primary Search Terms (Species)—DIALOG Search

•	Guinea pig(s)

•	Human(s)

•	Monkey(s)

•	Mouse

This document is a draft for review purposes only and does not constitute Agency policy.

C-2 DRAFT—DO NOT CITE OR QUOTE

-------
1	• Mice

2	• Rodent(s)

3	• Rat(s)

4

5	C.l.2.3. Secondary Search Terms (Toxicologyj—DIALOG Search

6	* = truncated

7	lw = terms are within 1 word of each other and in the order specified (see search term 32)

8

1.

Absor*

16. Elimin*

32. Mechanism (lw)

2.

ADME

17. Excret*

action

3.

Aryl hydrocarbon

18. Epidemiolog*

33. Metabo*



receptor

19. Feces

34. Oral*

4.

AhR

20. Feed*

35. P450

5.

Bioavail*

21. First order kinetics

36. Partition coefficient

6.

Biliar*

22. Food*

37. PBPK

7.

Biotransform*

23. Gastro*

38. Pharmacodynamic*

8.

Cytochrome

24. Gavage*

39. Pharmacokinetic*

9.	CYP*

10.	CYP1A1

11.	CYP1A2

12.	Diet, dietary, diets

13.	Disposit*

14.	Distrib*

25.	Half-life

26.	Induct*

27.	Ingest*

28.	In silico

29.	Kinetic*

30.	Liver

40.	Physiologically
based

41.	pharmacokinetic

42.	Protein bind*

43.	Toxicokinetic*

44.	Urin*

15.

Drink*

31. Lymph*



1

2	ADME = absorption, distribution, metabolism, elimination; AhR = aryl hydrocarbon receptor; CYP = cytochrome

3	P450.

4

5

6	C.1.3. Citation Screening Procedures and Results

7	Initial DIALOG searches resulted in a very large number of citation hits. Therefore,

8	some title and key word restrictions were applied iteratively to screen out less relevant citations

9	(e.g., requiring some search terms in title, requiring 2,3,7,8-TCDD rather than just TCDD).

10	Then, using reference management software, pooled information obtained from the various

11	DIALOG data bases was screened to remove duplicates. Citations then were numbered

This document is a draft for review purposes only and does not constitute Agency policy.

C-3	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

sequentially (as a unique identifier). Information retrieved included the following (when
available): author(s), publication year, title, source document name, volume, and page numbers.

The DIALOG search and duplicate removal procedure produced 775 unique citations. In
the next step, all 775 citations were screened for potential applicability to updating parameters in
the Aylward and Emond PBPK models. Of these 775 citations, 26 were selected for more
detailed review to determine their potential applicability, and full publications were retrieved.
Two citations were added from the supplemental search, giving a total of 28 articles identified
for further review.

Bibliographic information for the 28 articles selected for full review is provided in the
reference list at the end of this section. Table C-l summarizes the model input parameters
potentially addressed by the selected articles.

During 2003 to May 2009, the authors of the two kinetic models under consideration
published several articles. For the Emond model, which was first published in 2004 (Emond
et al., 2004), two subsequent papers have been published (Emond et al., 2005, 2006). The
Aylward model, which originated from the 1995 papers by Carrier et al. (1995a, b), was later
updated by the same group (Aylward et al., 2005a, b). The major change implemented in the last
two papers was the description of a desorption process in the digestive tract. The transfer rate
described is slow, but for a low body burden of TCDD, this process remains significant. This
concept was reported in 2002 by Moser and McLachlan (2002). The major modifications
expected to update the Emond model are (1) consideration of the desorption process in the
gastrointestinal tract and (2) rearrangement of the elimination constant, which will have a
negligible impact on the simulation. These changes are motivated by plausible observations
reported in the literature.

Because of the body burden found in humans and the importance of selecting an
appropriate dose metric in human risk assessment, the physiological model is an important tool
for assessing the kinetics following exposure to TCDD (Kim et al., 2003). Based on the
literature identified in this search, the major contributions that should be reviewed with respect to
the Aylward and Emond kinetic models are not modes of action or pharmacokinetic mechanisms,
but rather information for verifying or improving the accuracy of some model parameters.

Pharmacokinetics typically refers to four distinct steps including absorption, distribution,

metabolism, and excretion. Physiologically-based models consider each step. In the model each

This document is a draft for review purposes only and does not constitute Agency policy.

C-4 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

step is parameterized to reflect better predictions of the real observations. Occasionally,
reviewing these models is essential to determine if any key processes or parameters might be
described with better accuracy. This perspective underlies the review of the literature described
here. The review indicates TCDD disposition has become recognized as relatively significant
since the publication of the Emond and Aylward models. The literature that provides
information related to improving these models, however, is limited. For the benefit of this
exercise, EPA selected the literature that would likely contribute significantly to model response,
or to clarify or confirm different key issues driving the model results. Regarding the two TCDD
models, the two major issues that should be evaluated with respect to the recent literature
identified are the elimination profile and the induction of CYP1A2.

Reviewing the elimination variation in different species and testing variable elimination
with a data set appears to be appropriate. The literature reports that various factors might
influence elimination rate. Recent publications report the influence of diverse predictors such
age, body fat, or smoking habit on the elimination half-life (Milbrath et al., 2009; Kerger et al.,
2006, 2007). Determining whether using the Milbrath et al. information would help account for
intraspecies variability in elimination rate in the Emond and Aylward kinetic models would be
useful. In 2006, Emond et al. reviewed the influence of body fat mass and CYP1A2 induction on
the pharmacokinetics of TCDD. These two factors appear to contribute significantly to
elimination and their influences seem to be driven by TCDD body burden. Mullerova and
Kopecky (2007) discussed the influence of adipose tissue and the "yoyo" effects on various
diseases that might be influenced by persistent organic pollutant distribution. One group
explored the importance of variable elimination and compared these predictions to first-order
elimination using the Aylward and Emond models and supported these approaches for risk
assessment (Heinzl et al., 2007). Two groups of authors considered a one-compartment model to
derive the elimination half-life (Aylward et al., 2009; Nadal et al., 2008). Comparing the
half-life they obtained using this approach for a range of body burden to the variable elimination
half-life would be interesting.

The second important mechanism driving the distribution and elimination of TCDD is the
induction of CYP1A2, identified as the major ligand protein in liver (Diliberto et al., 1997). For
that process, authors suggested different aspects that should be investigated, including the
importance of the dose metrics in the target tissue and the inducible level of CYP1A2 (Wilkes

This document is a draft for review purposes only and does not constitute Agency policy.

C-5 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

et al., 2008; Staskal et al., 2005). Other papers address the intraspecies variability of lethal
potency in mature species versus the developing fetus (Kransler et al., 2007; Korkalainen et al.,
2004). Still others point out pronounced differences among species (namely, guinea pigs,
hamsters, mice, and rats) (Bohonowych and Denison, 2007), as observed in studies of long-term
effects of low TCDD dose in liver and in studies comparing hepatic accumulation and clearance
of TCDD (Korenaga et al., 2007; Boverhof et al., 2005). The interspecies variation of the
binding affinity constant of AhR also has been reported (Connor and Aylward, 2006; Nohara
et al., 2006).

The articles identified in this literature review should be adequate to update the Aylward
and Emond models, which need to be evaluated according to the same structure of compartments
described in the literature by the two model authors.

C.1.4. References Selected for More Detailed Review for Updating the PBPK Models

Aylward, LL; Brunet, RC; Carrier, G; et al. (2004). Concentration-dependent TCDD elimination
kinetics in humans: toxicokinetic modeling for moderately to highly exposed adults from Seveso,
Italy, and Vienna, Austria, and impact on dose estimates for the NIOSH cohort. J Expo Anal
Environ Epidemiol 15(1):51—65.

Aylward, LL; Brunet, RC; Starr, TB; et al. (2005). Exposure reconstruction for the TCDD-
exposed NIOSH cohort using a concentration- and age-dependent model of elimination. Risk
Anal 25(4):945-956.

Aylward, LL; Bodner, KM; Collins, JJ; et al. (2009). TCDD exposure estimation for workers at
a New Zealand 2,4,5-T manufacturing facility based on serum sampling data. J Expo Sci
Environ Epidemiol, doi: 10.1038/jes.2009.31.

Bohonowych, JE; Denison, MS. (2007). Persistent binding of ligands to the aryl hydrocarbon
receptor. Toxicol Sci 98(1):99-109.

Boverhof, DR; Burgoon, LD; Tashiro, C; et al. (2005). Temporal and dose-dependent hepatic
gene expression patterns in mice provide new insights into TCDD-mediated hepatotoxicity.
Toxicol Sci 85(2): 1048-1063.

Connor, KT; Aylward, LL. (2006). Human response to dioxin: aryl hydrocarbon receptor (AhR)
molecular structure, function, and dose-response data for enzyme induction indicate an impaired
human AhR. J Toxicol Environ Health B 9(2): 147-171.

Heinzl, H; Mittlback, M; Edler, L. (2007). On the translation of uncertainty from toxicokinetic
to toxicodynamic models - the TCDD example. Chemosphere 67(9):S365-S374.

This document is a draft for review purposes only and does not constitute Agency policy.

C-6 DRAFT—DO NOT CITE OR QUOTE

-------
Irigaray, P; Mejean, L; Laurent, F. (2005). Behaviour of dioxin in pig adipocytes. Food Chem
Toxicol 43(3):457-460.

Kerger, BD; Leung, HW; Scott, P; et al. (2006). Age- and concentration-dependent elimination
half-life of 2,3,7,8-tetrachlorodibenzo-p-dioxin in Seveso children. Environ Health Perspect
114(10): 1596—1602.

Kerger, BD; Leung, HW; Scott, PK; et al. (2007). Refinements on the age-dependent half-life
model for estimating child body burdens of polychlorodibenzodioxins and dibenzofurans.
Chemosphere 67(9):S272-S278.

Kim, AH; Kohn, MC; Nyska, A; et al. (2003). Area under the curve as a dose metric for
promotional responses following 2,3,7,8-tetrachlorodibenzo-p-dioxin exposure. Toxicol Appl
Pharmacol 191(1): 12-21.

Korenaga, T; Fukusato, T; Ohta, M; et al. (2007). Long-term effects of subcutaneously injected
2,3,7,8-tetrachlorodibenzo-p-dioxin on the liver of rhesus monkeys. Chemosphere
67(9):S399-S404.

Korkalainen, M; Tuomisto, J; Pohjanvirta, R. (2004). Primary structure and inducibility by
2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) of aryl hydrocarbon receptor repressor in a TCDD-
sensitive and a TCDD-resistant rat strain. Biochem Biophys Res Communications
315(1): 123-131.

Kransler, KM; McGarrigle, BP; Olson, JR. (2007). Comparative developmental toxicity of
2,3,7,8-tetrachlorodibenzo-p-dioxin in the hamster, rat and guinea pig. Toxicology
229(3):214-225.

Maruyama, W; Yoshida, K; Tanaka, T; et al. (2002). Determination of tissue-blood partition
coefficients for a physiological model for humans, and estimation of dioxin concentration in
tissues. Chemosphere 46(7):975-985.

Maruyama, W; Yoshida, K; Tanaka, T; et al. (2003). Simulation of dioxin accumulation in
human tissues and analysis of reproductive risk. Chemosphere 53(4):301 -313.

Maruyama, W; Aoki, Y. (2006). Estimated cancer risk of dioxins to humans using a bioassay
and physiologically based pharmacokinetic model. Toxicol Appl Pharmacol 214(2): 188-198.

Milbrath, MO; Wenger, Y; Chang, C-W; et al. (2009). Apparent Half-Lives of Dioxins, Furans,
and Polychlorinated Biphenyls as a Function of Age, Body Fat, Smoking Status, and Breast-
Feeding. Environ Health Perspect 117(3):417-425.

Moser, GA; McLachlan, MS. (2002). Modeling digestive tract absorption and desorption of
lipophilic organic contaminants in humans. Environ Sci Technol 36(15):3318-25.

Mullerova, D; Kopecky, J. (2007). White adipose tissue: storage and effector site for
environmental pollutants. Physiol Res 56(4):375-381.

This document is a draft for review purposes only and does not constitute Agency policy.

C-7 DRAFT—DO NOT CITE OR QUOTE

-------
Nadal, M; Perello, G; Schuhmacher, M; et al. (2008). Concentrations of PCDD/PCDFs in
plasma of subjects living in the vicinity of a hazardous waste incinerator: Follow-up and
modeling validation. Chemosphere 73(6):901-906.

Nohara, K; Ao, K; Miyamoto, Y; et al. (2006). Comparison of the 2,3,7,8-tetrachlorodibenzo-
p-dioxin (TCDD)-induced CYP1A1 gene expression profile in lymphocytes from mice, rats, and
humans: Most potent induction in humans. Toxicology 225(2-3):204-213.

Olsman, H; Engwall, M; Kammann, U; et al. (2007). Relative differences in aryl hydrocarbon
receptor-mediated response for 18 polybrominated and mixed halogenated dibenzo-p-dioxins
and -furans in cell lines from four different species. Environ Toxicol Chem 26(11):2448-2454.

Saghir, SA; Lebofsky, M; Pinson, DM; et al. (2005). Validation of Haber's Rule (doseX
time=constant) in rats and mice for monochloroacetic acid and 2,3,7,8-tetrachlorodibenzo-
p-dioxin under conditions of kinetic steady state. Toxicology 215(l-2):48-56.

Schecter, A; Pavuk, M; Popke, O; et al. (2003). Dioxin, dibenzofuran, and coplanar PCB Levels
in Laotian blood and milk from Agent Orange-sprayed and nonsprayed areas, 2001. J Toxicol
Environ Health A 66(21):2067-2075.

Staskal, DF; Diliberto, JJ; Devito, MJ; et al. (2005). Inhibition of human and rat CYP1A2 by
TCDD and dioxin-like chemicals. Toxicol Sci 84(2):225-231.

Toyoshiba, H; Walker, NJ; Bailer, AJ; et al. (2004). Evaluation of toxic equivalency factors for
induction of cytochromes P450 CYP1A1 and CYP1A2 enzyme activity by dioxin-like
compounds. Toxicol Appl Pharmacol 194(2): 156-168.

Wilkes, JG; Hass, BS; Buzatu, DA; et al. (2008). Modeling and assaying dioxin-like biological
effects for both dioxin-like and certain non-dioxin-like compounds. Toxicol Sci
102(1): 187—195.

1	C.1.5. Brief Descriptions of DIALOG Bibliographic Data Bases Searched

2	The National Technical Information Service (NTIS) database comprises summaries of

3	U.S. government-sponsored research, development, and engineering, plus analyses prepared by

4	federal agencies, their contractors, or grantees. It is the means through which unclassified,

5	publicly available, unlimited distribution reports are made available for sale from 240 agencies.

6	Additionally, some state and local government agencies contribute summaries of their reports to

7	the database. NTIS also provides access to the results of government-sponsored research and

8	development from countries outside the United States. Organizations that currently contribute to

9	the NTIS database include but are not limited to the following: the Japan Ministry of

10	International Trade and Industry (MITI); laboratories administered by the United Kingdom

This document is a draft for review purposes only and does not constitute Agency policy.

C-8 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

Department of Industry; the German Federal Ministry of Research and Technology (BMFT); and
the French National Center for Scientific Research (CNRS).

Pollution Abstracts provides access to environmental information that combines
information on scientific research and government policies in a single resource. Topics of
growing concern are extensively covered from the standpoints of atmosphere, emissions,
mathematical models, effects on people and animals, and environmental action in response to
global pollution issues. This database also contains material from conference proceedings and
hard-to-find summarized documents along with information from primary journals in the field of
pollution.

BIOSIS Previews® contains citations from Biological Abstracts® (BA) and Biological
Abstracts/Reports, Reviews, and Meetings® (BA/RRM) (formerly BioResearch Index®), the
major publications of BIOSIS®. These publications constitute the major English-language
service providing comprehensive worldwide coverage of research in the biological and
biomedical sciences. Biological Abstracts includes approximately 350,000 accounts of original
research yearly from nearly 5,000 primary journal and monograph titles. BA/RRM includes an
additional 200,000+ citations a year from meeting abstracts, reviews, books, book chapters,
notes, letters, and selected reports.

IPA Toxicology provides focused toxicology information on all phases of the
development and use of drugs and on professional pharmaceutical practice. The scope of the
database ranges from the clinical and practical to the theoretical aspects of toxicology literature.
A unique feature of abstracts reporting clinical studies is the inclusion of the study design,
number of patients, dosage, dosage forms, and dosage schedule.

Medical Literature, Analysis, and Retrieval System Online (MEDLINE®), produced by
the U.S. National Library of Medicine (NLM), is NLM's premier bibliographic database. It
contains more than 15 million references to journal articles in life sciences with a concentration
on biomedicine. The broad coverage of the database includes basic biomedical research and the
clinical sciences since 1950, including nursing, dentistry, veterinary medicine, pharmacy, allied
health, and pre-clinical sciences. MEDLINE® also covers life sciences that are vital to
biomedical practitioners, researchers, and educators, including some aspects of biology,
environmental science, marine biology, and plant and animal science, as well as biophysics and
chemistry. MEDLINE® is indexed using NLM's controlled vocabulary, Medical Subject

This document is a draft for review purposes only and does not constitute Agency policy.

C-9 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

Headings (MeSH®). Approximately 400,000 records are added per year, of which more than
76 percent are in English. MEDLINE® contains AIDSLINE, HealthSTAR, Toxline, In Process
(formerly known as Pre-MEDLINE®), In Data Review, and POPLINE.

ToxFile covers the toxicological, pharmacological, biochemical, and physiological
effects of drugs and other chemicals. Adverse drug reactions, chemically induced diseases,
carcinogenesis, mutagenesis, teratogenesis, environmental pollution, waste disposal, radiation,
and food contamination are typical areas of coverage. The databases Environmental Mutagen
Information Center (EMIC), Developmental and Reproductive Toxicology (DART), and Toxic
Substances Control Act Test Submissions (TSCATS) are included in ToxFile. It is not clearly
stated whether the Chemical Carcinogenesis Research Information System (CCRIS), Hazardous
Substances Data Bank (HSDB), or Genetic Toxicology Data Bank (GENE-TOX) are included in
ToxFile. Consequently, a separate, on-line search was conducted to ensure that these databases
were searched.

BIOSIS® Toxicology contains citations from BA and BA/RRM (formerly BioResearch
Index®), the major publications of BIOSIS®, that focus on toxicology and related topics.
Records are drawn from journal articles, conference papers, monographs and book chapters,
notes, letters, and reports, as well as original research. U.S. patent records are also included.

CANCERLIT® is produced by the International Cancer Research DataBank Branch
(ICRDB) of the U.S. National Cancer Institute. The database consists of bibliographic records
referencing cancer research publications dating from 1963 to 2002. Most records contain
abstracts, and all records contain citation information and additional descriptive fields such as
document type and language. Beginning with the June 1983 CANCERLIT update, records from
the MEDLINE® database dealing with cancer topics have been added to CANCERLIT.

The Registry of Toxic Effects of Chemical Substances (RTECS®) is a comprehensive
database of basic toxicity information for over 150,000 chemical substances including
prescription and non-prescription drugs, food additives, pesticides, fungicides, herbicides,
solvents, diluents, chemical wastes, reaction products of chemical waste, and substances used in
both industrial and household situations. Reports of the toxic effects of each compound are
cited. In addition to toxic effects and general toxicology reviews, data on skin and/or eye
irritation, mutation, reproductive consequences and tumorigenicity are provided. Federal
standards and regulations, National Institute for Occupational Safety and Health (NIOSH)

This document is a draft for review purposes only and does not constitute Agency policy.

C-10 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

recommended exposure limits and information on the activities of EPA, NIOSH, National
Toxicology Program (NTP), and Occupational Safety and Health Administration (OSHA)
regarding the substance are also included. The toxic effects are linked to literature citations from
both published and unpublished governmental reports, and published articles from the scientific
literature. The database corresponds to the print version of the RTECS®, formerly known as the
Toxic Substances List, which was started in 1971. Originally prepared by the NIOSH, the
RTECS® database is now produced and distributed by Symyx Technologies, Inc.

C.2. TOXICOKINETIC MODELING CODE (EMOND ET AL., 2005)

C.2.1. Human Standard Model
C.2.1.1. Model Code

PROGRAM: 'Three Compartment PBPK Model for TCDD in Human: Standard Model
(Non-Gestation)'

!HUM_NON_GEST_ICF_F083109.csl

INITIAL !INITIALIZATION OF PARAMETERS
!SIMULATION PARAMETERS ====

CONSTANT EXP_TIME_ON
(HOUR)

CONSTANT EXP_TIME_0FF
(HOUR)

CONSTANT DAY_CYCLE
(HOUR)

CONSTANT BCK_TIME_ON
EXPOSURE BEGINS (HOUR)
CONSTANT BCK_TIME_OFF
EXPOSURE ENDS (HOUR)

0.	! TIME AT WHICH EXPOSURE BEGINS
6.132e5 ! TIME AT WHICH EXPOSURE ENDS

24.0	! NUMBER OF HOURS BETWEEN DOSES

6.132e5	! TIME AT WHICH BACKGROUND

6.132e5	! TIME AT WHICH BACKGROUND

!EXPOSURE DOSES
CONSTANT MSTOTBCKGR
(NG/KG)

CONSTANT MSTOT
CONSTANT DOSEIV
CONSTANT MW
MSTOT_NM = MSTOT/MW
MSTOT_NMBCKGR = MSTOTBCKGR/MW
DOSEIV_NM = DOSEIV/MW
NMOL/KG

0.0

! ORAL BACKGROUND EXPOSURE DOSE

1.0E-7	! ORAL EXPOSURE DOSE (NG/KG)

0.0	! INJECTED DOSE (NG/KG)

322.0	! MOLECULAR WEIGHT (G/MOL)

! CONVERTS THE DOSE TO NMOL/KG
!CONVERTS THE BACKGROUND DOSE TO NMOL/KG

! CONVERTS THE INJECTED DOSE TO

!INITIAL GUESS OF THE FREE CONCENTRATION IN THE LIGAND (COMPARTMENT
INDICATED BELOW) ====

CONSTANT CFLLI0	=0.0	! LIVER (NMOL/L)

This document is a draft for review purposes only and does not constitute Agency policy.

C-11 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!BINDING CAPACITY (AhR) FOR NON LINEAR BINDING (COMPARTMENT INDICATED
BELOW) ===

CONSTANT LIBMAX	=0.35	! LIVER (NMOL/L)

! PROTEIN AFFINITY CONSTANTS (1A2 OR AhR, COMPARTMENT INDICATED BELOW)

CONSTANT KDLI	=	0.1	! LIVER (AhR) (NMOL/L) WANG

ET AL.. 1997

CONSTANT KDLI2	=	4 0.0	! LIVER (1A2) (NMOL/L) EMOND ET

AL. 2004

!EXCRETION AND ABSORPTION CONSTANTS
CONSTANT KST	=	0.01	! GASTRIC RATE CONSTANT (HR-

1), EMOND ET AL., 2005

CONSTANT KABS	=	0.06	! INTESTINAL ABSORPTION CONSTANT

(HR-1), EMOND ET AL. 2005

!ELIMINATION CONSTANTS
CONSTANT CLURI	=	4.17D-8	! URINARY CLEARANCE (L/HR), EMOND

ET AL., 2005

CONSTANT KELV	=	l.le-3	! INTERSPECIES VARIABLE

ELIMINATION CONSTANT (1/HOUR)

!CONSTANT TO DIVIDE THE ABSORPTION INTO LYMPHATIC AND PORTAL FRACTIONS
CONSTANT A	=0.7	! LYMPHATIC FRACTION,

WANG ET AL. (1997)

!PARTITION COEFFICIENTS
CONSTANT PF	=	1.0e2

WANG ET AL. 1997

CONSTANT PRE	=	1.5

WANG ET AL. 1997

CONSTANT PLI	=	6.0

AL. 1997

! ADIPOSE TISSUE/BLOOD,
! REST OF THE BODY/BLOOD,
! LIVER/BLOOD, WANG ET

!PARAMETERS FOR INDUCTION OF CYP1A2
CONSTANT PAS_INDUC
= NO)

CONSTANT CYP1A2_10UTZ =

OF 1A2 (NMOL/L)

CONSTANT CYP1A2_1A1
(NMOL/L)

CONSTANT CYP1A2_1EC5 0 =

(NMOL/L)

CONSTANT CYP1A2_1A2
(NMOL/L)

CONSTANT CYP1A2_1K0UT =

(H-l)

CONSTANT CYP1A2_1TAU
CONSTANT CYP1A2_1EMAX =

(UNITLESS)

CONSTANT HILL
BINDING EFFECT CONSTANT (UNITLESS)

! DIFFUSIONAL PERMEABILITY FRACTION
CONSTANT PAFF	=	0.12

1.0

1. 6e3

1. 6e3

1. 3e2

1. 6e3

0.1

0.25
9 . 3e3

0.6

! INCLUDE INDUCTION? (1 = YES, 0

! DEGRADATION CONCENTRATION CONSTANT

! BASAL CONCENTRATION OF 1A1

! DISSOCIATION CONSTANT TCDD-CYP1A2

! BASAL CONCENTRATION OF 1A2

! FIRST ORDER RATE OF DEGRADATION

! HOLDING TIME (H)

! MAXIMUM INDUCTION OVER BASAL EFFECT

!HILL CONSTANT; COOPERATIVELY LIGAND
! ADIPOSE (UNITLESS)

This document is a draft for review purposes only and does not constitute Agency policy.

C-12 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CONSTANT PAREF	=	0.03

CONSTANT PALIF	=	0.35

!TISSUE BLOOD FLOW EXPRESSED AS
CONSTANT QFF	=	0.05

(UNITLESS), KRISHNAN 2008
CONSTANT QLIF	=	0.2 6

! REST OF BODY (UNITLESS)
! LIVER (UNITLESS)

A FRACTION OF CARDIAC OUTPUT =========

! ADIPOSE TISSUE BLOOD FLOW FRACTION

! LIVER (UNITLESS), KRISHNAN 2008

!COMPARTMENT TISSUE BLOOD EXPRESSED AS A FRACTION OF THE TOTAL
COMPARTMENT VOLUME =========

CONSTANT WFB0	=	0.050 ! ADIPOSE TISSUE, WANG ET AL. 1997

CONSTANT WREB0	=	0.030 ! REST OF THE BODY, WANG ET AL. 1997

CONSTANT WLIB0	=	0.266 ! LIVER, WANG ET AL. 1997

!EXPOSURE SCENARIO FOR UNIQUE OR
!NUMBER OF EXPOSURES PER WEEK
CONSTANT WEEK_LACK	=	0.0

(WEEK)

CONSTANT WEEK_PERIOD =	168.0

(HOURS)

CONSTANT WEEK_FINISH =	168.0

!NUMBER OF EXPOSURES PER MONTH
CONSTANT MONTH_LACK =	0.0

(MONTH)

REPETITIVE WEEKLY OR MONTHLY EXPOSURE
! DELAY BEFORE EXPOSURE ENDS
! NUMBER OF HOURS IN THE WEEK
! TIME EXPOSURE ENDS (HOURS)
! DELAY BEFORE EXPOSURE BEGINS

!SET FOR BACKGROUND EXPOSURE===========

!TIME CONSTANT FOR BACKGROUND EXPOSURE===========

CONSTANT Day_LACK_BG	=0.0	! DELAY BEFORE EXPOSURE BEGINS

(HOUR)

CONSTANT Day_PERIOD_BG = 24.0	! LENGTH OF EXPOSURE (HOUR)

!TIME CONSTANT FOR WEEKLY EXPOSURE
CONSTANT WEEK_LACK_BG	= 0.0	! DELAY BEFORE BACKGROUND EXPOSURE

BEGINS (WEEK)

CONSTANT WEEK_PERIOD_BG = 168.0	! NUMBER OF HOURS IN THE WEEK

(HOURS)

CONSTANT WEEK FINISH BG = 168.0	! TIME EXPOSURE ENDS (HOURS)

! CONSTANT USED IN CARDIAC OUTPUT EQUATION
CONSTANT QCC	= 15.36	! (L/KG-H), EMOND ET AL.

2004

! COMPARTMENT LIPID EXPRESSED AS THE FRACTION OF TOTAL LIPID
!Data from Emonds Thesis 2001

CONSTANT F_TOTLIP
(UNITLESS)

CONSTANT B_TOTLIP
CONSTANT RE_TOTLIP
(UNITLESS)

CONSTANT LI_TOTLIP
CONSTANT MEANLIPID

0.8000

0.0057
0.0190

0.0670
974 . 0

! ADIPOSE TISSUE

! BLOOD (UNITLESS)
! REST OF THE BODY

! LIVER (UNITLESS)

END ! END OF THE INITIAL SECTION

DYNAMIC ! DYNAMIC SIMULATION SECTION

This document is a draft for review purposes only and does not constitute Agency policy.

C-13 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

ALGORITHM IALG
CINTERVAL CINT
MAXTERVAL MAXT
MINTERVAL MINT
VARIABLE T
CONSTANT TIMELIMIT
CONSTANT	Y0

SIMULATION
CONSTANT GROWON
GROWTH? (1 = YES, 0 = NO)
CINTXY = CINT
PFUNC = CINT

2	! GEAR METHOD

10.0	! COMMUNICATION INTERVAL

1.0e+10	IMAXIMUM INTERVAL CALCULATION

1.0E-10	!MINIMUM INTERVAL CALCULATION
0.0

1.752e5	!SIMULATION LIMIT TIME (HOUR)

0.0	! AGE (YEARS) AT BEGINNING OF

1.0	! INCLUDE BODY WEIGHT AND HEIGHT

DAY=T/24.0

WEEK =T/168.0

MONTH =T/730 . 0

YEAR=Y 0+T/8760.0

GYR =Y0 + growon*T/8 7 60.0

! TIME IN DAYS
! TIME IN WEEKS
! TIME IN MONTHS
! TIME IN YEARS
TIME FOR USE IN GROWTH EQUATION (YEARS)

DERIVATIVE ! PORTION OF CODE THAT SOLVES DIFFERENTIAL EQUATIONS

! CHRONIC OR SUBCHRONIC EXPOSURE SCENARIO
! NUMBER OF EXPOSURES PER DAY
DAY_LACK	= EXP_TIME_ON

DAY_PERIOD = DAY_CYCLE
DAY_FINISH = CINTXY
MONTH_PERIOD = TIMELIMIT
MONTH FINISH = EXP TIME OFF

DELAY BEFORE EXPOSURE BEGINS (HOURS)
EXPOSURE PERIOD (HOURS)

LENGTH OF EXPOSURE (HOURS)

EXPOSURE PERIOD (MONTHS)

LENGTH OF EXPOSURE (MONTHS)

! NUMBER OF EXPOSURES PER DAY AND MONTH
DAY_FINISH_BG = CINTXY

MONTH_LACK_BG = BCK_TIME_ON !DELAY BEFORE BACKGROUD EXPOSURE BEGINS
(MONTHS)

MONTH_PERIOD_BG = TIMELIMIT	! BACKGROUND EXPOSURE PERIOD (MONTHS)

MONTH_FINISH_BG = BCK_TIME_OFF ! LENGTH OF BACKGROUND EXPOSURE (MONTHS)

B = 1.0-A ! FRACTION OF DIOXIN ABSORBED IN THE PORTAL FRACTION OF THE LIVER

!HUMAN BODY WEIGHT GROWTH EQUATION========

! POLYNOMIAL REGRESSION EXPRESSION WRITTEN
!APRIL 10 2008, OPTIMIZED WITH DATA OF PELEKIS ET AL. 2001
! POLYNOMIAL REGRESSION EXPRESSION WRITTEN WITH
!HUH AND BOLCH 2 0 03 FOR BMI CALCULATION

! BODY WEIGHT CALCULATION

WTO = (0.0006*GYR**3 - 0.0912*GYR**2 + 4.32*GYR + 3.652)

! BODY MASS INDEX CALCULATION

BH = -2D-5*GYR**4+4.2D-3*GYR**3.0-0.315*GYR**2.0+9.7465*GYR+72.098

!HEIGHT EQUATION FORMULATED FOR USE FROM 0 TO 7 0 YEARS

BHM= (BH/10 0.0)	!HUMAN HEIGHT IN METERS (BHM)

HBMI= WTO/(BHM**2.0) ! HUMAN BODY MASS INDEX (BMI)

This document is a draft for review purposes only and does not constitute Agency policy.

C-14 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

! ADIPOSE TISSUE FRACTION

WT0GR= WT0*1.0e3 ! BODY WEIGHT IN GRAMS

WF0= -6.3 6D-2 0*WT0GR**4.0 +1.12D-14*WT0GR**3.0 -5.8D-10*WT0GR**2.0 +1.2D-
5*WT0GR+5.91D-2

! LIVER,VOLUME,

! APPROACH BASED ON LUECKE (2007)

WLI0= (3.59D-2 -(4.76D-7*WT0GR)+(8.50D-12*WT0GR**2.0)-(5.45D-
17*WT0GR**3.0))

WRE0 = (0.91 -(WLIB0*WLI0+WFB0*WF0+WLI0+WF0))/(1.0+WREB0)

!REST OF THE BODY FRACTION; UPDATED FOR

EPA ASSESSMENT

QREF = 1.0-(QFF+QLIF)	!REST OF BODY BLOOD FLOW

QTTQF = QFF+QREF+QLIF	! SUM MUST EQUAL 1

!COMPARTMENT VOLUME (L
WF = WF0 * WTO
WRE = WRE0 * WTO
WLI = WLI0 * WTO
WB=0.07 5*WT0

OR KG) =========

! ADIPOSE

! REST OF THE BODY
! LIVER
! BLOOD

!COMPARTMENT TISSUE BLOOD (L OR KG) =========

WFB = WFB0 * WF	! ADIPOSE

WREB = WREB0 * WRE	! REST OF THE BODY

WLIB = WLIB0 * WLI	! LIVER

!CARDIAC OUTPUT FOR THE GIVEN BODY WEIGHT
QC= QCC*(WTO**0.75)	! [L BLOOD/HOUR]

QF = QFF*QC
[L/HR]

QLI = QLIF*QC
QRE = QREF*QC

! ADIPOSE TISSUE BLOOD FLOW RATE

! LIVER TISSUE BLOOD FLOW RATE [L/HR]
!REST OF THE BODY BLOOD FLOW RATE [L/HR]

QTTQ = QF+QRE+QLI

! TOTAL FLOW RATE [L/HR]

!PERMEABILITY ORGAN FLOW [L/HR] :
PAF = PAFF*QF
PARE = PAREF*QRE
PALI = PALIF*QLI

ADIPOSE

REST OF THE BODY
LIVER TISSUE

! ABSORPTION SECTION
! INTRAVENOUS
IV	= DOSEIV_NM * WTO

MSTTBCKGR = MSTOT_NMBCKGR *WT0
MSTT	= MSTOT NM * WTO

!AMOUNT IN NMOL
!AMOUNT IN (NMOL)
!AMOUNT IN NMOL

!REPETITIVE ORAL BACKGROUND EXPOSURE SCENARIOS
DAY_EX P O S U RE_B G = PULSE(DAY_LACK_BG,DAY_PERIOD_BG,DAY_FINISH_BG)
WEEK_EXPOSURE_BG = PULSE(WEEK_LACK_BG,WEEK_PERIOD_BG,WEEK_FINISH_BG)
MONTH_EXPOSURE_BG = PULSE(MONTH_LACK_BG,MONTH_PERIOD_BG,MONTH_FINISH_BG)

MSTTCH_BG = (DAY_EXPOSURE_BG*WEEK_EXPOSURE_BG*MONTH_EXPOSURE_BG)*MSTTBCKGR
MSTTFR_BG = MSTTBCKGR/CINT

CYCLE BG =DAY EXPOSURE BG*WEEK EXPOSURE BG*MONTH EXPOSURE BG

This document is a draft for review purposes only and does not constitute Agency policy.

C-15 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

! CONDITIONAL ORAL EXPOSURE (BACKGROUND EXPOSURE)
IF (MSTTCH_BG.EQ.MSTTBCKGR) THEN
ABSMSTT_GB= MSTTFR_BG

ELSE

ABSMSTT_GB =0.0
END IF

!REPETITIVE ORAL MAIN EXPOSURE SCENARIO
DAY_EXPOSURE = PULSE(DAY_LACK,DAY_PERIOD,DAY_FINISH)

WEEK_EXPOSURE = PULSE(WEEK_LACK,WEEK_PERIOD,WEEK_FINISH)

MONTH_EXPOSURE = PULSE(MONTH_LACK,MONTH_PERIOD,MONTH_FINISH)

MSTTCH = (DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE)*MSTT
CYCLE = DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE
MSTTFR=MSTT/CINT

!CONDITIONAL ORAL EXPOSURE
IF (MSTTCH.EQ.MSTT) THEN

ABSMSTT= MSTTFR
ELSE

ABSMSTT = 0.

END IF

CYCLETOT=INTEG(CYCLE, 0.0)

! MASS Balance CHANGE IN THE LUMEN
RMSTT= -(KST+KABS)*MST+ABSMSTT +ABSMSTT_GB ! RATE OF CHANGE (NMOL/H)
MST = INTEG(RMSTT,0.)	!AMOUNT REMAINING IN GI TRACT

(NMOL)

! ABSORPTION IN LYMPH CIRCULATION
LYRMLUM = KABS*MST*A
LYMLUM = INTEG(LYRMLUM,0.0)

! ABSORPTION IN PORTAL CIRCULATION
LIRMLUM = KABS*MST*B

LIMLUM = INTEG(LIRMLUM,0.0)

! PERCENT OF DOSE REMAINING IN THE GI TRACT
PRCT_remain_GIT = 100.0*MST/(MSTT+1E-30)

!IV ABSORTPION SCENARIO 	

IVR= IV/PFUNC ! RATE FOR IV INFUSION IN BLOOD
EXPIV= IVR * (1.0-STEP(PFUNC))

IVDOSE = integ(EXPIV,0.0)

!SYSTEMIC BLOOD COMPARTMENT
! MODIFICATION OCT 8 2009
CB=(QF*CFB+QRE*CREB+QLI*CLIB+EXPIV+LYRMLUM)/(QC+CLURI) !

CA = CB	!CONCENTRATION (NMOL/L)

!CB=(QF*CFB+QRE*CREB+QLI*CLIB+EXPIV+LYRMLUM-RAURI)/QC !

! CA = CB	! CONCENTRATION (NMOL/L)

This document is a draft for review purposes only and does not constitute Agency policy.

C-16 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!URINARY EXCRETION BY KIDNEY
! MODIFICATION OCT 8 2009
RAURI = CLURI *CB

AURI = INTEG(RAURI,0.0)

! CONCENTRATION UNIT
PRCT_B = 100.0*CB/(MSTT+1E-30)	! PERCENT OF DOSE

CBSNGKGLIADJ = CB*MW/(0.55*B TOTLIP) !serum concentration in lipid adjust
(PG/G LIPID=PPT)	~

CBPPT = CBSNGKGLIADJ
CBNGKG = CB*MW

CBpptRH = CB*MW*10000/(0.55*MEANLIPID) !SERUM CONCENTRATION IN LIPID ADJUST
(PG/G LIPID=PPT)

AUC CBSNGKGLIADJ=INTEG(CBSNGKGLIADJ,0.0)

!ADIPOSE TISSUE COMPARTMENT
RAFB= QF*(CA-CFB)-PAF*(CFB-CF/PF)
AFB = INTEG(RAFB,0.0)
CFB = AFB/WFB

!TISSUE SUBCOMPARTMENT
RAF = PAF*(CFB-CF/PF)
AF = INTEG(RAF,0.0)

CF = AF/WF

!(NMOL/HR)

!(NMOL)
!(NMOL/KG)

!(NMOL/HR)

!(NMOL)
!(NMOL/KG)

!POST SIMULATION UNIT CONVERSION
CFTOTAL = (AF + AFB)/(WF + WFB) ! TOTAL CONCENTRATION NMOL/ML
PRCT_F = 100.0*CFTOTAL/(MSTT+1E-30)

CFNGKG =CFTOTAL*MW

!REST OF THE BODY COMPARTMENT========

RAREB= QRE*(CA-CREB)-PARE*(CREB-CRE/PRE) !(NMOL/HR)
AREB = INTEG(RAREB,0.0)	!(NMOL)

CREB = AREB/WREB	!(NMOL/KG)

!TISSUE SUBCOMPARTMENT
RARE = PARE*(CREB-CRE/PRE)	!(NMOL/HR)

ARE = INTEG(RARE,0.0)	!(NMOL)

CRE = ARE/WRE	!(NMOL/KG)

!POST SIMULATION UNIT CONVERSION
CRETOTAL = (ARE + AREB)/(WRE + WREB) ! TOTAL CONCENTRATION IN NMOL/ML
PRCT RE = 100.0*CRETOTAL/(MSTT+1E-30) ! PERCENT OF DOSE

!LIVER COMPARTMENT
!TISSUE BLOOD SUBCOMPARTMENT
RALIB = QLI*(CA-CLIB)-PALI*(CLIB-CFLLIR)+LIRMLUM	!(NMOL/HR)

ALIB = INTEG(RALIB,0.0)	!(NMOL)

CLIB = ALIB/WLIB

!TISSUE SUBCOMPARTMENT
RALI = PALI*(CLIB-CFLLIR)-REXCLI	!(NMOL/HR)

ALI = INTEG(RALI,0.0)	!(NMOL)

CLI = ALI/WLI	!(NMOL/KG)

This document is a draft for review purposes only and does not constitute Agency policy.

C-17 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!FREE TCDD IN LIVER
! MODIFICATION OCTOBER 8 2009
CFLLI= IMPLC(CLI-(CFLLIR*PLI+(LIBMAX*CFLLIR/(KDLI+CFLLIR)) &

+ ( (CYP1A2_103*CFLLIR/(KDLI2+CFLLIR)*PAS_INDUC) ) )-CFLLI, CFLLI0) !
CONCENTRATION OF FREE TCDD IN LIVER
CFLLIR=DIM(CFLLI,0.0)

!MODIFIED FROM:

!PARAMETER (LIVER_1RMN = 1.0E-30)

! CFLLI= IMPLC(CLI-(CFLLIR*PLI+(LIBMAX*CFLLIR/(KDLI+CFLLIR &
+LIVER_1RMN))+((CYP1A2_103*CFLLIR/(KDLI2+CFLLIR &

!	+LIVER_1RMN)*PAS_INDUC)))-CFLLI,CFLLI0)

! CFLLIR=DIM(CFLLI,0.0)

CBNDLI= LIBMAX*CFLLIR/(KDLI+CFLLIR) !CONC OF TCDD BOUDN TO AhR
!CBNDLI= LIBMAX*CFLLIR/(KDLI+CFLLIR+LIVER 1RMN) !CONC BIND

!POST SIMULATION UNIT CONVERSION
CLITOTAL = (ALI + ALIB)/(WLI + WLIB)

PRCT_LI = 100.0*CLITOTAL/(MSTT+1.0E-30)
rec_occ_AHR= 100.0*CFLLIR/(KDLI+CFLLIR+1.0)
OCCUPANCY

PROT_occ_lA2= 100.0*CFLLIR/(KDLI2+CFLLIR)
OCCUPANCY

CLINGKG= CLITOTAL*MW
CBNDLINGKG = CBNDLI*MW

! TOTAL CONCENTRATION IN NMOL/ML

! PERCENT BOUND TO AhR

! PERCENT BOUND TO 1A2

[NG TCDD/KG]

!FRACTION INCREASE OF INDUCTION OF CYP1A2
fold_ind=CYPlA2_10UT/CYPlA2_lA2

VARIATIONOFAC =(CYP1A2_10UT-CYP1A2_1A2)/CYP1A2_1A2

!VARIABLE ELIMINATION BASED ON THE CYP1A2
KBILE_LI_T = Kelv*VARIATIONOFAC!

REXCLI = KBILE_LI_T*CFLLIR*WLI ! DOSE-DEPENDENT RATE OF BILLIARY EXCRETION
OF DIOXIN

EXCLI = INTEG(REXCLI,0.0) !TOTAL AMOUNT OF DIOXIN EXCRETED

!CHEMICAL IN CYP450 (1A2) COMPARTMENT
!PARAMETER FOR INDUCTION OF CYP1A2

CYP1A2_1KINP = CYP1A2_1K0UT*CYP1A2_10UTZ ! BASAL RATE OF CYP1A2 PRODUCTION
SET EQUAL TO BASAL RATE OF DEGRDATION AT STEADY STATE

! MODIFICATION OCTOBER 8 2009
CYP1A2_10UT =INTEG(CYP1A2_1KINP * (1.0 + CYP1A2_1EMAX *(CBNDLI+1.0e-30)**HILL
&

/(CYP1A2_1EC50**HILL + (CBNDLI+1.0e-30)**HILL)) &
- CYP1A2_1K0UT*CYP1A2_10UT, CYP1A2_10UTZ) ! LEVELS OF CYP1A2
! MODEIFIED FROM:

!PARAMETER (CYP1A2_1RMN = le-30)

!CYP1A2_10UT =INTEG(CYP1A2_1KINP * (1 + CYP1A2_1EMAX *(CBNDLI &

!	+CYP1A2 1RMN)**HILL/(CYP1A2 1EC50 + (CBNDLI + CYP1A2 1RMN)**HILL) &

This document is a draft for review purposes only and does not constitute Agency policy.

C-18 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

!	+CYP1A2_1RMN) - CYP1A2_1K0UT*CYP1A2_1&

!	OUT, CYP1A2_10UTZ)

! EQUATIONS INCORPORATING DELAY OF CYP1A2 PRODUCTION (NOT USED IN
SIMULATIONS)

CYP1A2_1R02 = (CYP1A2_10UT - CYP1A2_102)/ CYP1A2_1TAU
CYP1A2_102 =INTEG(CYP1A2_1R02, CYP1A2_1A1)

CYP1A2_1R03 = (CYP1A2_102 - CYP1A2_103)/ CYP1A2_1TAU
CYP1A2_103 =INTEG(CYP1A2_1R03, CYP1A2_1A2)

!CHECK MASS BALANCE
BDOSE= LYMLUM+LIMLUM+IVDOSE

BMASSE = EX C LI+AURI+AFB+AF+AREB+ARE+ALIB+ALI
BDIFF = BDOSE-BMASSE

! BODY BURDEN IN TERMS OF CONCENTRATION (NG/KG)

BBNGKG = (AFB+AF+AREB+ARE+ALIB+ALI)*MW/WT0	!

!COMMAND END OF THE SIMULATION
TERMT (T.GE. TIMELIMIT, 'Time limit has been reached.')

END ! END OF THE DERIVATIVE SECTION
END ! END OF THE DYTNAMIC SECTION
END ! END OF THE PROGRAM

C.2.1.2. Input File

% base file name = "TESTJULY2009.m"

%clear ©variable
output 0clear

prepare 0clear year T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG
%output 0all

% PARAMETERS FOR SIMULATION
CINT = 1 %0.5

% TIME AT WHICH EXPOSURE BEGINS (HOUR)

%324120	% HOUR/YEAR !TIME AT WHICH EXPOSURE

EXP_TIME_ON = 0.
EX P_TIME_0 FF = 613200
ENDS (HOUR)

DAY_CYCLE =24
B C K_TIME_ON = 613200
BEGINS (HOUR)
BCK_TIME_OFF = 613200
ENDS (HOUR)

TIMELIMIT = 613200
(HOUR)

MSTOTBCKGR = 0.

5 NUMBER OF HOURS BETWEEN DOSES (HOUR)

5324120	% TIME AT WHICH BACKGROUND EXPOSURE

5324120	% TIME AT WHICH BACKGROUND EXPOSURE

5324120	%324120	% SIMULATION TIME LIMIT

5 ORAL BACKGROUND EXPOSURE DOSE (UG/KG)

% oral dose oral dose oral dose
MSTOT	= 9.97339283634997E-07

DOSEIV	= 0	%NG/KG

% oral dose oral dose oral dose

ORAL DAILY EXPOSURE DOSE (NG/KG)

MEANLIPID = 730
PAS INDUC= 1

INDUCTION INCLUDED? (1=YES, 0=NO)

This document is a draft for review purposes only and does not constitute Agency policy.

C-19 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

C.2.2. Human Gestational Model

C.2.2.1. Model Code

PROGRAM: 'Three Compartment PBPK Model for TCDD in Human (Gestation)'

! Parameters were change may 16, 2002
! Come from {8MAI_CHR_PRE-EXP_GD}

! Come from {12 Mouse GDJfile

!{{IMPORTANT-IMPORTANT-IMPORTANT-IMPORTANT}}

! REDUCTION OF MOTHER AND FETUS COMPARTMENT
! 2M_R_TCDD_JULY2 0 02 ////(JULY 18,20 02)////

!TCDD_RED_4Species_2003_4	////(APR 8 ,2003)////

!TCDD_RED_4Species_2003_9	////(APR 17 ,2003)////

!TCDD_RED_4Species_2003_12	////(APR 17 ,2003)////

!APRIL 18 2003

!TCDD_4C_4SP_2 0 03	////(APR 18 ,2003)////

! was ''Gest 4 species l.csl'' but update July 2009

!GEST HUM 0 45Y 4 ICF afterKKfix v3 humangestational.csl
!HUM_GESTATIONAL_ICF_F083109.csl
!HUM_GESTATIONAL_ICF_F10 07 0 9.csl

!Legend/Legend/Legend/Legend/Legend/Legend/Legend/Legend/

!Legend for this PBPK model

IMating: control the tenure of exchange between fetus and
IMother and also control imitated tissue growth
!Control: WTFE, WPLA0, QPLAF
!(for rat, mouse, human, and monkey)

!Control transfer from mother to fetus and fetus to mother by TRANSTIME ON
!SWITCH_trans = 0 NO TRANSFER	~

!SWITCH_trans = 1 TRANSFER OCCURS
! These switches are also controlled by mating parameters

INITIAL !

!SIMULATION PARAMETERS
CONSTANT PARA_ZERO
CONSTANT EXP_TIME_ON
CONSTANT EXP_TIME_0FF
CONSTANT DAY_CYCLE
CONSTANT BCK_TIME_ON
BEGINS (HOURS)

CONSTANT BCK_TIME_OFF
(HOURS)

CONSTANT TRANSTIME_ON

AT 9 WEEKS OR 1512 HOURS OF GESTATION

! INTRAVENOUS SEQUENCY
CONSTANT IV_LACK	=0.0

CONSTANT IV_PERIOD	=0.0

!PREGNANCY PARAMETER

= le-30

0.0	!TIME AT WHICH EXPOSURE BEGINS (HOURS)

530.0	!TIME AT WHICH EXPOSURE ENDS (HOURS)

2 4.0	!NUMBER OF HOURS BETWEEN DOSES (HOURS)

0.0	!TIME AT WHICH BACKGROUND EXPOSURE

0.0	!TIME AT WHICH BACKGROUND EXPOSURE ENDS

0.0	!CONTROL TRANSFER FROM MOTHER TO FETUS

This document is a draft for review purposes only and does not constitute Agency policy.

C-20 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CONSTANT MATTING
CONSTANT PFETUS
CONSTANT CLPLA_FET
(L/HR)

=0.0	!BEGINNING OF MATING (HOUR)

=4.0	!PARTITION COEFFICIENT

= 1.0e-3 !CLEARANCE TRANSFER FOR MOTHER TO FETUS

! CONSTANT EXPOSURE CONTROL

!ACUTE, SUBCHRONIC, CHRONIC EXPOSURE =====

!OR BACKGROUND EXPOSURE (IN THIS CASE 3 TIMES A DAY)===

CONSTANT MSTOTBCKGR	=0.0	! ORAL BACKGROUND EXPOSURE DOSE (NG/KG)

CONSTANT MSTOT	=0.0	! ORAL EXPOSURE DOSE (NG/KG)

!ORAL ABSORPTION

! MSTT= MSTOT/10 0 0 *WT0 *1/322*1000 !AMOUNT IN NMOL
MSTOT NM = MSTOT/MW	!CONVERTS THE DOSE TO NMOL/KG

!INTRAVENOUS ABSORPTION

CONSTANT DOSEIV =0.0	! INJECTED DOSE (NG/KG)

DOSEIV_NM = DOSEIV/MW	! CONVERTS THE INJECTED DOSE TO NMOL/KG

CONSTANT DOSEIVLATE =0.0	!INJECTED DOSE LATE (UG/KG)

DOSEIVNMlate = DOSEIVLATE/MW	!AMOUNT IN NMOL/G

!INITIAL GUESS OF THE FREE CONCENTRATION IN THE LIGAND (COMPARTMENT
INDICATED BELOW)====

CONSTANT CFLLI0	= 0.0	!LIVER (NMOL/L)

CONSTANT CFLPLA0	= 0.0	!PLACENTA (NMOL/L)

!BINDING CAPACITY	(AhR) FOR NON LINEAR BINDING (COMPARTMENT INDICATED
BELOW) (NMOL/L) ===

CONSTANT LIBMAX	=0.35 ! LIVER (NMOL/L)

CONSTANT PLABMAX	= 0.2 !TEMPORARY PARAMETER

!PROTEIN AFFINITY CONSTANTS	(1A2 OR AhR, COMPARTMENT INDICATED BELOW)
(NMOL/ML)===

CONSTANT KDLI = 0.1	!LIVER (AhR) (NMOL/L), WANG ET AL. 1997

CONSTANT KDLI2 = 40.0	!LIVER (1A2) (NMOL/L), EMOND ET AL.
2004

CONSTANT KDPLA =0.1	!ASSUME IDENTICAL TO KDLI (AhR)

!EXCRETION AND ABSORPTION CONSTANT
CONSTANT KST	=0.01	! GASTRIC RATE CONSTANT (HR-1), EMOND ET

AL. 2005

CONSTANT KABS	= 0.06	! INTESTINAL ABSORPTION CONSTANT (HR-1),

EMOND ET AL. (2005)

!INTERSPECIES ELIMINATION CONSTANT
!TEST ELIMINATION VARIABLE, EMOND ET AL. 2005
CONSTANT KELV	=	l.le-3 !4.OD-3	! INTERSPECIES VARIABLE

ELIMINATION CONSTANT (1/HOUR)

! ELIMINATION CONSTANTS
CONSTANT CLURI	= 4.17e-8 ! URINARY CLEARANCE (L/HR), EMOND ET AL.

2005

! CONSTANT TO DIVIDE THE ABSORPTION INTO LYMPHATIC AND PORTAL FRACTIONS
CONSTANT A	=0.7	! LYMPHATIC FRACTION, WANG ET AL. 1997

This document is a draft for review purposes only and does not constitute Agency policy.

C-21 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!PARTITION COEFFICIENTS

CONSTANT PF	= 1.0e2

CONSTANT PRE	=1.5
1997

CONSTANT PLI	= 6.0

CONSTANT PPLA	=1.5
WANG ET AL. 1997

! ADIPOSE TISSUE/BLOOD, WANG ET AL. 1997
! REST OF THE BODY/BLOOD, WANG ET AL.

! LIVER/BLOOD, WANG ET AL. 19 97
! TEMPORARY PARAMETER NOT CONFIGURED,

!PARAMETER FOR INDUCTION OF CYP

CONSTANT PAS_INDUC
CONSTANT CYP1A2_10UTZ
1A2 (NMOL/L)

CONSTANT CYP1A2_1A1
CONSTANT CYP1A2_1EC5 0
(NMOL/L)

CONSTANT CYP1A2_1A2
CONSTANT CYP1A2_1K0UT
CONSTANT CYP1A2_1TAU
CONSTANT CYP1A2_1EMAX
(UNITLESS)

CONSTANT HILL
BINDING EFFECT CONSTANT

= 1.0
= 1.6e3

= 1.6e3
= 1.3e2

= 1.6e3
= 0.1
= 0.25
= 9.3e3

= 0.6
(UNITLESS)

1A2, WANG ET AL. 1997

! INCLUDE INDUCTION? (1 = YES, 0 = NO)
! DEGRADATION CONCENTRATION CONSTANT OF

! BASAL CONCENTRATION OF 1A1 (NMOL/L)
! DISSOCIATION CONSTANT TCDD-CYP1A2

!BASAL CONCENTRATION OF 1A2 (NMOL/ML)
! FIRST ORDER RATE OF DEGRADATION (H-l)
!HOLDING TIME (H)

! MAXIMUM INDUCTION OVER BASAL EFFECT

!HILL CONSTANT; COOPERATIVELY LIGAND

!DIFFUSIONAL PERMEABILITY FRACTION, WANG ET AL (1997)

CONSTANT PAFF
CONSTANT PAREF
CONSTANT PALIF
CONSTANT PAPLAF

= 0. 12
= 0. 03
= 0.35
= 0.3

ADIPOSE (UNITLESS)

REST OF THE BODY (UNITLESS)
LIVER (UNITLESS)

OPTIMIZED PARAMETER

!TISSUE BLOOD FLOW EXPRESSED AS A FRACTION OF CARDIAC OUTPUT, KRISHNAN 2007
CONSTANT QFF	=0.05	! ADIPOSE TISSUE BLOOD FLOW FRACTION

(UNITLESS), KRISHNAN 2008

CONSTANT QLIF	= 0.26	! LIVER (UNITLESS), KRISHNAN 2008

!===FRACTION OF TISSUE BLOOD WEIGHT Wang et al

(1997)

CONSTANT WFB0
CONSTANT WREB0
CONSTANT WLIB0
CONSTANT WPLAB0

0.050	!ADIPOSE TISSUE, WANG ET AL. 1997

0.030	!REST OF THE BODY, WANG ET AL. 1997

0.266	!LIVER, WANG ET AL. 1997

0.500	!ASSUME HIGHLY VASCULARIZED

! EXPOSURE SCENARIO FOR UNIQUE OR REPETITIVE WEEKLY OR MONTHLY EXPOSURE
! NUMBER OF EXPOSURES PER WEEK

CONSTANT WEEK_LACK	=0.0	!DELAY BEFORE EXPOSURE ENDS (WEEK)

CONSTANT WEEK_PERIOD	= 168.0	! NUMBER OF HOURS IN THE WEEK (HOURS)

CONSTANT WEEK FINISH	= 168.0	! TIME EXPOSURE ENDS (HOURS)

! NUMBER OF EXPOSURES PER MONTH
CONSTANT MONTH LACK	=0.0

!DELAY BEFORE EXPOSURE BEGINS (MONTHS)

CONSTANT FOR BACKGROUND EXPOSURE=

CONSTANT Day_LACK_BG	=0.0

CONSTANT Day_PERIOD_BG =24.0

! DELAY BEFORE EXPOSURE BEGINS (HOURS)
!LENGTH OF EXPOSURE (HOURS)

! NUMBER OF EXPOSURES PER WEEK
CONSTANT WEEK_LACK_BG =0.0
(WEEK)

!DELAY BEFORE BACKGROUD EXPOSURE BEGINS

This document is a draft for review purposes only and does not constitute Agency policy.

C-22 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CONSTANT WEEK_PERIOD_BG = 168.0
CONSTANT WEEK FINISH BG = 168.0

! NUMBER OF HOURS IN THE WEEK (HOURS)
!TIME EXPOSURE ENDS (HOURS)

! CONSTANT USED IN CARDIAC OUTPUT EQUATION

CONSTANT QCC	= 15.36 ![L/KG-H], EMOND ET AL. 2004

! COMPARTMENT LIPID EXPRESSED AS THE FRACTION OF TOTAL LIPID
!Data from Emonds Thesis 2001

CONSTANT F_TOTLIP
CONSTANT B_TOTLIP
CONSTANT RE_TOTLIP
CONSTANT LI_TOTLIP
CONSTANT PLA_TOTLIP
CONSTANT FETUS TOTLIP

0.8000	! ADIPOSE TISSUE (UNITLESS)

0.0057	! BLOOD (UNITLESS)

0.0190	! REST OF THE BODY (UNITLESS)

0.0670	! LIVER (UNITLESS)

0.019	! PLACENTA (UNITLESS)

0.019	! FETUS (UNITLESS)

CONSTANT MEANLIPID

974

END ! END OF THE INITIAL SECTION

DYNAMIC ! DYNAMIC SIMULATION SECTION

ALGORITHM

CINTERVAL

MAXTERVAL

MINTERVAL

VARIABLE

CONSTANT

CONSTANT

SIMULATION

CONSTANT

IALG
CINT
MAXT
MINT
T

TIMELIMIT
Y0

GROWON

GROWTH? (1=YES, 0=NO)

CINTXY = CINT
PFUNC = CINT

2
0.1
1.0e+10
1.0E-10
0.0
100
0.0

1.0

GEAR METHOD

COMMUNICATION INTERVAL
MAXIMUM CALCULATION INTERVAL
MINIMUM CALCULATION INTERVAL

!SIMULATION LIMIT TIME (HOUR)
! AGE (YEARS) AT BEGINNING OF

! INCLUDE BODY WEIGHT AND HEIGHT

!TIME TRANSFORMATION
DAY= T/24.0
WEEK =T/168.0
YEAR=Y0+T/8760 . 0
GYR =Y0 + growon*T/8 7 60.0
EQUATION

! TIME IN YEARS
! TIME FOR USE IN GROWTH

DERIVATIVE ! PORTION OF CODE THAT SOLVES DIFFERENTIAL EQUATIONS

!====== CHRONIC OR SUBCHRONIC EXPOSURE SCENARIO

! NUMBER OF EXPOSURES PER DAY

DAY_LACK
DAY_PERIOD
DAY_FINISH
MONTH_PERIOD
MONTH FINISH

EXP_TIME_ON
DAY_CYCLE
CINTXY
TIMELIMIT
EXP TIME OFF

DELAY BEFORE EXPOSURE BEGINS
EXPOSURE PERIOD (HOURS)
LENGTH OF EXPOSURE (HOURS)
EXPOSURE PERIOD (MONTHS)
LENGTH OF EXPOSURE (MONTHS)

(HOURS)

! NUMBER OF EXPOSURES PER DAY AND MONTH

This document is a draft for review purposes only and does not constitute Agency policy.

C-23 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

DAY_FINISH_BG	= CINTXY

MONTH_LACK_BG	= BCK_TIME_ON !DELAY BEFORE BACKGROUND EXPOSURE BEGINS
(MONTHS)

MONTH_PERIOD_BG	= TIMELIMIT	!BACKGROUND EXPOSURE PERIOD (MONTHS)

MONTH FINISH BG	= BCK TIME OFF !LENGTH OF BACKGROUND EXPOSURE (MONTHS)

! INTRAVENOUS LATE
IV_FINISH = CINTXY

B = 1-A ! FRACTION OF DIOXIN ABSORBED IN THE PORTAL FRACTION OF THE LIVER

! MOTHER BODY WEIGHT GROWTH EQUATION
! MODIFICATION TO ADAPT THIS MODEL AT HUMAN MODEL

! BECAUSE LINEAR DESCRIPTION IS NOT GOOD ENOUGH FOR MOTHER GROWTH
! MOTHER BODY WEIGHT GROWTH
! HUMAN BODY WEIGHT (0 TO 4 5 YEARS)

! POLYNOMIAL REGRESSION EXPRESSION WRITTEN

!APRIL 10 2008, OPTIMIZED WITH DATA OF PELEKIS ET AL. 2001
! POLYNOMIAL REGRESSION EXPRESSION WRITTEN WITH
!HUH AND BOLCH 2 0 03 FOR BMI CALCULATION

! BODY WEIGHT CALCULATION. UNIT IN KG FOR GESTATIONAL PORTION
WTO = (0.0006*GYR**3 - 0.0912*GYR**2 + 4.32*GYR + 3.652)
!BODY MASS INDEX CALCULATION

BH = -2D-5*GYR**4+4.2D-3*GYR**3.0-0.315*GYR**2.0+9.7465*GYR+72.098
!HEIGHT EQUATION FORMULATED FOR USE FROM 0 TO 7 0 YEARS
BHM= (BH/100.0)!HUMAN HEIGHT IN METER (BHM)

HBMI= WTO/(BHM**2.0) ! HUMAN BODY MASS INDEX (BMI)

!MODIFICATION IN KG

RTESTGEST= T-MATTING ! STARTING TIME FOR FETAL GROWTH
TESTGEST=DIM(RTESTGEST,0.0)

! GROWTH OF FETAL TISSUE

GESTATTION_FE=((4d-15*TESTGEST**4 -3d-ll*TESTGEST**3 +ld-7*TESTGEST**2 -8d-
5*TESTGEST +0.0608))

WTFER= DIM(GESTATTION_FE,0.0) ! FETAL COMPARTMENT WEIGHT
WTFE= WTFER

!///////////////////////////////////////////////////////////////////////
! FAT GROWTH EXPRESSION LINEAR DURING PREGNANCY
! FROM O'FLAHERTY_19 92
!///////////////////////////////////////////////////////////////////////

WT0GR= WT0*1.0e3 ! MOTHER BODY WEIGHT IN G

WF0 =(-6.36D-20*WT0GR**4.0 +1.12D-14*WT0GR**3.0 &

-5.8D-10*WT0GR**2.0+1.2D-5*WT0GR+5.91D-2) ! MOTHER FAT COMPARTMENT

GROWTH

!///////////////////////////////////////////////////////////////////////
! WPLA PLACENTA GROWTH EXPRESSION, SINGLE EXPONENTIAL WITH OFFSET
! FROM O'FLAHERTY_19 92 ! FOR EACH PUP
!///////////////////////////////////////////////////////////////////////

This document is a draft for review purposes only and does not constitute Agency policy.

C-24 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!SAME EQUATION THEN THE FORST MODEL. BODY WEIGHT KEPT IN G
!A CORRECTION FOR THE BODY WEIGHT (WTO(KG)*1000 = WTOGR)

WPLA0N_HUMAN= (850*exp(-9.434*(exp(-5.23d-4*(TESTGEST)))))
WPLA0R = WPLA0N_HUMAN/WTOGR

WPLA0W = DIM(WPLA0R,0.0) ! PLACENTA WEIGHT
WPLA0=WPLA0W

!///////////////////////////////////////////////////////////////////////
! QPLA PLACENTA GROWTH EXPRESSION, DOUBLE EXPONENTIAL WITH OFFSET
! FROM O'FLAHERTY_19 92
!///////////////////////////////////////////////////////////////////////

QPLAF_HUMAN = SWITCH_trans*( (ld-10*TESTGEST**3.0 -5D-7*TESTGEST**2.0
+0.0 017*TESTGEST+1.1937)/QC)

GEST_QPLAF=DIM(QPLAF_HUMAN,0.0) ! PLACENTA BLOOD FLOW RATE
QPLAF =GEST_QPLAF

! LIVER,VOLUME (HUMAN 0 TO 7 0 YEARS)

! APPROACH BASED ON LUECKE (2007)

WLI0= (3.5 9D-2 -(4.7 6D-7*WT0GR)+(8.5 0D-12*WT0GR**2.0)-(5.4 5D-17*WT0GR**3.0))
! LIVER VOLUME IN GROWING HUMAN

! VARIABILITY OF REST OF THE BODY DEPENDS ON OTHER ORGAN
WRE0 = (0.91-(WLIB0*WLI0+WFB0*WF0+ WPLAB0*WPLA0 + WLI0 + WF0 +
WPLA0))/(1+WREB0)

QREF = 1-(QFF+QLIF+QPLAF)	!REST BODY BLOOD FLOW (ML/HR)

QTTQF = QFF+QREF+QLIF+QPLAF	! SUM MUST EQUAL 1

! COMPARTMENT TISSUE BLOOD VOLUME (L)
WF = WF0 * WTO
WRE = WRE0 * WTO
WLI = WLI0 * WTO
WPLA= WPLA0* WTO

ADIPOSE TISSUE
REST OF THE BODY
LIVER
PLACENTA

! COMPARTMENT TISSUE VOLUME (L)
WFB = WFB0 * WF
WREB = WREB0 * WRE
WLIB = WLIB0 * WLI
WPLAB = WPLAB0* WPLA

ADIPOSE TISSUE
REST OF THE BODY
LIVER
PLACANTA

! TOTAL VOLUME OF COMPARTMENT (L)
WFT = WF
WRET = WRE
WLIT = WLI
WPLAT= WPLAB

TOTAL ADIPOSE TISSUE
TOTAL REST OF THE BODY
TOTAL LIVER TISSUE
TOTAL PLACENTA TISSUE

! CONSTANT USED IN CARDIAC OUTPUT EQUATION

! UNIT CHANGED ON JULY 14 2009 (L/HR)
QC= QCC*(WTO)**0.75

QF = QFF*QC
QLI = QLIF*QC
QRE = QREF*QC
QPLA = QPLAF*QC

ADIPOSE TISSUE BLOOD FLOW RATE (L/HR)
LIVER TISSUE BLOOD FLOW RATE (L/HR)
REST OF THE BODY BLOOD FLOW RATE (L/HR)
PLACENTA TISSUE BLOOD FLOW RATE (L/HR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-25 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

QTTQ = QF+QRE+QLI+QPLA	!TOTAL FLOW RATE (L/HR)

! ========= DIFFUSIONAL PERMEABILITY

PAF = PAFF*QF
PARE = PAREF*QRE
(L/HR)

PALI = PALIF*QLI
PAPLA = PAPLAF*QPLA

FACTORS FRACTION ORGAN FLOW =========

! ADIPOSE TISSUE BLOOD FLOW RATE (L/HR)
! REST OF THE BODY BLOOD FLOW RATE

! LIVER TISSUE BLOOD FLOW RATE (L/HR)
! PLACENTA TISSUE BLOOD FLOW RATE (L/HR)

ABSORPTION SECTION
ORAL

INTRAPERITONEAL

SUBCUTANEOUS

INTRAVENOUS

!BACKGROUND EXPOSURE

!EXPOSURE FOR STEADY STATE CONSIDERATION
!REPETITIVE EXPOSURE SCENARIO

MSTOT_NMBCKGR = MSTOTBCKGR/322	!AMOUNT IN NMOL/G

MSTTBCKGR =MSTOT_NMBCKGR *WT0

DAY_EX P O S U RE_B G = PULSE(DAY_LACK_BG,DAY_PERIOD_BG,DAY_FINISH_BG)
WEEK_EXPOSURE_BG = PULSE(WEEK_LACK_BG,WEEK_PERIOD_BG,WEEK_FINISH_BG)
MONTH_EXPOSURE_BG = PULSE(MONTH_LACK_BG,MONTH_PERIOD_BG,MONTH_FINISH_BG)

MSTTCH_BG = (DAY_EXPOSURE_BG*WEEK_EXPOSURE_BG*MONTH_EXPOSURE_BG)*MSTTBCKGR
MSTTFR_BG = MSTTBCKGR/CINT

CYCLE_BG =DAY_EXPOSURE_BG*WEEK_EXPOSURE_BG*MONTH_EXPOSURE_BG
! CONDITIONAL ORAL EXPOSURE (BACKGROUND EXPOSURE)

IF (MSTTCH_BG.EQ.MSTTBCKGR) THEN
ABSMSTT_GB= MSTTFR_BG

ELSE

ABSMSTT_GB =0.0
END IF

CYCLETOTBG=INTEG(CYCLE_BG,0.0)

!MULTIROUTE EXPOSURE
!REPETITIVE EXPOSURE SCENARIO

MSTT= MSTOT_NM * WTO	!AMOUNT IN NMOL

DAY_EXPOSURE = PULSE(DAY_LACK,DAY_PERIOD,DAY_FINISH)
WEEK_EXPOSURE = PULSE(WEEK_LACK,WEEK_PERIOD,WEEK_FINISH)
MONTH_EXPOSURE = PULSE(MONTH_LACK,MONTH_PERIOD,MONTH_FINISH)

MSTTCH = (DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE)*MSTT

MSTTFR = MSTT/CINT

This document is a draft for review purposes only and does not constitute Agency policy.

C-26 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CYCLE = DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE

SUMEXPEVENT= INTEG (CYCLE,0.0) !NUMBER OF CYCLES GENERATED DURING SIMULATION

! CONDITIONAL ORAL EXPOSURE
IF (MSTTCH.EQ.MSTT) THEN

ABSMSTT= MSTTFR
ELSE

ABSMSTT =0.0
END IF

CYCLETOT=INTEG(CYCLE, 0.0)

! MASS CHANGE IN THE LUMEN

RMSTT= -(KST+KABS)*MST +ABSMSTT +ABSMSTT_GB ! RATE OF CHANGE (NMOL/H)

MST = INTEG(RMSTT,0.0)	!AMOUNT REMAINING IN DUODENUM

(NMOL)

! ABSORPTION IN LYMPH CIRCULATION
LYRMLUM = KABS*MST*A
LYMLUM = INTEG(LYRMLUM,0.0)

! ABSORPTION IN PORTAL CIRCULATION
LIRMLUM = KABS*MST*B
LIMLUM = INTEG(LIRMLUM,0.0)

!IV ABSORPTION SCENARIO	

IV= DOSEIV_NM * WTO !AMOUNT IN NMOL
IVR= IV/PFUNC ! RATE FOR IV INFUSION IN BLOOD
EXPIV= IVR * (1-STEP(PFUNC))

IVDOSE = integ(EXPIV,0.0)

!IV LATE IN THE CYCLE
!MODIFICATION JANUARY 13 2004
IV_RlateR = DOSEIVNMlate*WT0

IV_EXPOSURE=PULSE(IV_LACK,IV_PERIOD,IV_FINISH)

IV_lateT = IV_EXPOSURE *IV_RlateR
IV_late = IV_lateT/CINT

SUMEXPEVENTIV= integ(IV_EXPOSURE,0.0) !NUMBER OF CYCLE GENERATE DURING
SIMULATION

!SYSTEMIC BLOOD COMPARTMENT
! MODIFICATION OCT 8 2009
CB=(QF*CFB+QRE*CREB+QLI*CLIB+EXPIV+LYRMLUM+QPLA*CPLAB+IV_late)/(QC+CLURI) !
CA = CB	! CONCENTRATION (NMOL/L)

!CB=(QF*CFB+QRE*CREB+QLI*CLIB+EXPIV+LYRMLUM+QPLA*CPLAB+IV_late-RAURI)/QC
!(NMOL/L)

!URINARY EXCRETION BY KIDNEY
! MODIFICATION OCT 8 2009
RAURI = CLURI *CB

This document is a draft for review purposes only and does not constitute Agency policy.

C-27 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

AURI = INTEG(RAURI,0.0)

!RAURI = CLURI * CRE
!AURI = INTEG(RAURI,0.0)

!UNIT CONVERSION POST SIMULATION
CONSTANT MW=322 IMOLECULAR WEIGHT (NG/NMOL)

CONSTANT SERBLO =0.55
CONSTANT UNITCORR = 1.0e3

CBSNGKGLIADJ = CB*MW/(0.55*B_TOTLIP) !NG SERUM LIPID ADJUSTED/KG
AUCBS_NGKGLIADJ=integ(CBSNGKGLIADJ,0.)

CBNGKG= CB*MW !NG/KG

PRCT_B = 100.0*CB/(MSTT+1E-30)	!PERCENT OF ORAL DOSE IN BLOOD

PRCT_BIV = 100.0*CB/(IV_RlateR+lE-30) ! PERCENT OF IV DOSE IN BLOOD

!ADIPOSE COMPARMTENT
!TISSUE BLOOD SUBCOMPARTMENT
RAFB= QF*(CA-CFB)-PAF*(CFB-CF/PF)

AFB = INTEG(RAFB,0.0)

CFB = AFB/WFB

!TISSUE SUBCOMPARTMENT
RAF = PAF*(CFB-CF/PF)

AF = INTEG(RAF,0.0)

CF = AF/WF

!UNIT CONVERSION POST SIMULATION
CFTOTAL= (AF + AFB)/(WF + WFB) ! TOTAL CONCENTRATION IN NMOL/ML
PRCT_F = 100.0*CFTOTAL/(MSTT+1E-30) !PERCENT OF ORAL DOSE IN FAT
PRCT_FIV = 100.0*CFTOTAL/(IV_RlateR+lE-30) !PERCENT OF IV DOSE IN FAT
CFNGKG=CFTOTAL*MW ! FAT CONCENTRATION IN NG/KG
AUCF_NGKGH=integ(CFNGKG,0.)

!(NMOL/H)

!(NMOL)
!(NMOL/L)

!(NMOL/H)
!(NMOL)
!(NMOL/L)

!REST OF THE BODY COMPARTMENT
!TISSUE BLOOD SUBCOMPARTMENT
RAREB= QRE *(CA-CREB)-PARE*(CREB-CRE/PRE)
AREB = INTEG(RAREB,0.0)

CREB = AREB/WREB

!TISSUE SUBCOMPARTMENT
RARE = PARE*(CREB - CRE/PRE)
ARE = INTEG(RARE,0.0)

CRE = ARE/WRE
ARETOT = ARE +AREB

!(NMOL/H)
!(NMOL)

(NMOL/L)

!(NMOL/H)
!(NMOL)

(NMOL/L)

!POST SIMULATION UNIT CONVERSION
CRETOTAL= (ARE + AREB)/(WRE + WREB)	! TOTAL CONCENTRATION (NMOL/L)

PRCT_RE = 100.0*CRETOTAL/(MSTT+1E-30) ! PERCENT OF ORAL DOSE IN REST OF BODY
PRCT_REIV = 100.0*CRETOTAL/(IV_RlateR+lE-30) ![ PERCENT OF IV DOSE IN REST
OF BODY

CRENGKG=CRETOTAL*MW	! REST OF THE BODY

CONCENTRATION (NG/KG)

!LIVER COMPARTMENT
!TISSUE BLOOD SUBCOMPARTMENT

This document is a draft for review purposes only and does not constitute Agency policy.

C-28 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

RALIB = QLI*(CA-CLIB)-PALI*(CLIB-CFLLIR)+LIRMLUM ! (NMOL/HR)

ALIB = INTEG(RALIB,0.0)	!(NMOL)

CLIB = ALIB/WLIB	!(NMOL/L)

!TISSUE SUBCOMPARMTENT

RALI = PALI*(CLIB - CFLLIR)-REXCLI	! (NMOL/HR)

ALI = INTEG(RALI,0.0)	!(NMOL)

CLI = ALI/WLI	!(NMOL/L)

!FREE TCDD CONCENTRATION IN LIVER
! MODIFICATION OCTOBER 8 2009
CFLLI= IMPLC(CLI-(CFLLIR*PLI+(LIBMAX*CFLLIR/(KDLI+CFLLIR)) &

+((CYP1A2_103*CFLLIR/(KDLI2+CFLLIR)*PAS_INDUC)))-CFLLI,CFLLI0)
CFLLIR=DIM(CFLLI,0.0) ! FREE TCDD CONCENTRATION IN LIVER
!MODIFIED FROM:

!PARAMETER (LIVER_1RMN = 1.0E-30)

! CFLLI= IMPLC(CLI-(CFLLIR*PLI+(LIBMAX*CFLLIR/(KDLI+CFLLIR &
!+LIVER_1RMN))+((CYP1A2_103*CFLLIR/(KDLI2 + CFLLIR &
!+LIVER_1RMN)*PAS_INDUC)))-CFLLI,CFLLI0)

!CFLLIR=DIM(CFLLI,0.0)

! MODIFICATION OCTOBER 8 2009

CBNDLI= LIBMAX*CFLLIR/(KDLI+CFLLIR) !BOUND CONCENTRATION (NMOL/L)

!POST SIMULATION UNIT CONVERSION
CLITOTAL= (ALI + ALIB)/(WLI + WLIB) ! TOTAL CONCENTRATION (NMOL/L)

PRCT_LI = 100.0*CLITOTAL/(MSTT+1E-30) ! PERCENT OF ORAL DOSE IN LIVER
PRCT_LIIV = 100.0*CLITOTAL/(IV_RlateR+lE-30) ! PERCENT OF IV DOSE IN LIVER
Rec_occ= CFLLIR/(KDLI+CFLLIR)

CLINGKG=CLITOTAL*MW ! LIVER CONCENTRATION IN NG/KG

AUCLI_NGKGH=integ(CLINGKG, 0.0)

CBNDLINGKG = CBNDLI*MW ! BOUND CONCENTRATION IN NG/KG
AUCBNDLI_NGKGH =INTEG(CBNDLINGKG,0.0)

!FRACTION INCREASE OF INDUCTION OF CYP1A2
fold_ind=CYPlA2_10UT/CYPlA2_lA2

VARIATIONOFAC =(CYP1A2_10UT-CYP1A2_1A2)/CYP1A2_1A2

!VARIABLE ELIMINATION BASED ON THE CYP1A2
! MODIFICATION OCTOBER 8 2009

KBILE_LI_T = Kelv*VARIATIONOFAC! ! DOSE-DEPENDENT EXCRETION RATE CONSTANT

REXCLI = KBILE_LI_T*CFLLIR*WLI ! DOSE-DEPENDENT BILLIARY EXCRETION RATE
EXCLI = INTEG(REXCLI,0.0)

!KBILE LI T =((CYP1A2 10UT-CYP1A2 1A2)/CYP1A2 lA2)*Kelv !

!CHEMICAL IN CYP450 (1A2) COMPARTMENT

CYP1A2_1KINP = CYP1A2_1K0UT* CYP1A2_10UTZ ! BASAL PRODCUTION RATE OF CYP1A2
SET EQUAL TO BASAL DEGREDATION RATE

! MODIFICATION OCTOBER 8 2009
CYP1A2_10UT =INTEG(CYP1A2_1KINP * (1.0 + CYP1A2_1EMAX *(CBNDLI+1.0e-30)**HILL
&

/(CYP1A2_1EC50**HILL + (CBNDLI+1.Oe-30)**HILL)) &

This document is a draft for review purposes only and does not constitute Agency policy.

C-29 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

- CYP1A2_1K0UT*CYP1A2_10UT, CYP1A2_10UTZ)

!MODIFIED FROM:

!PARAMETER (CYP1A2_1RMN = 1E-30)

!CYP1A2_10UT =INTEG(CYP1A2_1KINP * (1 + CYP1A2_1EMAX *(CBND&
!LI +CYP1A2_1RMN)**HILL/(CYP1A2_1EC50 + (CBNDLI + CYP1A2_1&
!RMN)**HILL) +CYP1A2_1RMN) - CYP1A2_1K0UT*CYP1A2_1&
!OUT, CYP1A2_10UTZ)

! EQUATIONS INCORPORATING DELAY OF CYP1A2 PRODUCTION (NOT USED IN
SIMULATIONS)

CYP1A2_1R02 = (CYP1A2_10UT - CYP1A2_102)/ CYP1A2_1TAU
CYP1A2_102 =INTEG(CYP1A2_1R02, CYP1A2_1A1)

CYP1A2_1R03 = (CYP1A2_102 - CYP1A2_103)/ CYP1A2_1TAU
CYP1A2_103 =INTEG(CYP1A2_1R03, CYP1A2_1A2)

!PLACENTA COMPARTMENT
!TISSUE BLOOD SUBCOMPARTMENT
RAPLAB= QPLA*(CA - CPLAB)-PAPLA*(CPLAB -CFLPLAR)

APLAB = INTEG(RAPLAB,0.0)

CPLAB = APLAB/(WPLAB+1E-30)

!TISSUE SUBCOMPARTMENT
RAPLA = PAPLA*(CPLAB-CFLPLAR)-RAMPF + RAFPM
APLA = INTEG(RAPLA,0.0)

CPLA = APLA/(WPLA+le-30)

! NEW EQUATION AUGUST 2 8 2 009
PARAMETER (PARA_ZERO = 1.0E-30)

CFLPLA= IMPLC(CPLA-(CFLPLAR*PPLA +(PLABMAX*CFLPLAR/(KDPLA&

+CFLPLAR+PARA_ZERO)))-CFLPLA,CFLPLA0)

CFLPLAR=DIM(CFLPLA,0.0)

!POST SIMULATION UNIT CONVERSION
CPLATOTAL = ((APLAB+APLA)/(WPLAB+WPLA))

PRCT_PLA = (CPLATOTAL/(MSTT+1E-30))*100
PRCT_PLAIV = (CPLATOTAL/(IV_RlateR+lE-30))*100

!FETUS COMPARTMENT
RAFETUS= RAMPF-RAFPM

AFETUS=INTEG(RAFETUS,0.0)

CFETUS=AFETUS/(WTFE+1.0e-30)

CFETOTAL= CFETUS
CFETUS_v = CFETUS/PFETUS

!POST SIMULATION UNIT CONVERSION
CFETUSNGKG = CFETUS*MW	!(NG/KG)

PRCT_FE = 100.0*CFETOTAL/(MSTT+1E-30)

PRCT_FEIV = 100.0*CFETOTAL/(IV_RlateR+lE-30)

!TRANSFER OF DIOXIN FROM PLACENTA TO FETUS
!FETAL EXPOSURE ONLY DURING EXPOSURE

IF (T.LT.TRANSTIME_ON) THEN

SWITCH_trans =0.0
ELSE

SWITCH trans = 1

! NMOL/HR)

! (NMOL)
! (NMOL/ML)

! (NMOL/HR)

! (NMOL)
! (NMOL/ML)

This document is a draft for review purposes only and does not constitute Agency policy.

C-30 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

END IF

!TRANSFER OF DIOXIN FROM PLACENTA TO FETUS
! MODIFICATION 26 SEPTEMBER 2003

RAMPF = (CLPLA_FET*CPLA)*SWITCH_trans
AMPF=INTEG(RAMPF,0.0)

!TRANSFER OF DIOXIN FROM FETUS TO PLACENTA
RAFPM = (CLPLA_FET*CFETUS_v)*SWITCH_trans!

AFPM = INTEG(RAFPM,0.0)

!CHECK MASS BALANCE 	

BDOSE= IVDOSE +LYMLUM+LIMLUM

BMASSE = EXCLI+AURI+AFB+AF+AREB+ARE+ALIB+ALI+APLA+APLAB+AFETUS !
BDIFF = BDOSE-BMASSE

!BODY BURDEN (NMOL)

BODY_BURDEN = AFB +AF+ARE B +ARE +ALIB+ALI+AP LA+AP LAB

!BODY BURDEN CONCENTRATION (NG/KG)

BBNGKG =(AFB+AF+AREB+ARE+ALIB+ALI+AP LA+AP LAB)*MW/WT0

! END SIMULATION COMMAND

TERMT (T.GE. TimeLimit, 'Time limit has been reached.')

END ! END OF THE DERIVATIVE SECTION
END ! END OF THE DYNAMIC SECTION
END ! END OF THE PROGRAM

C.2.2.2. Input File

output 0clear
prepare 0clear T

year CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

CINT = 1 %168 %100
%EXPOSURE SCENARIO

EXP_TIME_ON
EX P_T I ME_0 F F
DAY_CYCLE
B C K_TIME_ON
(HOUR)

BCK_TIME_OFF
IV_LACK
IV PERIOD

0

401190
24

401190

401190
401190
401190

%GESTATION CONTROL
MATTING	= 393120

TIMELIMIT	= 399840

TRANSTIME_ON	= 394632

GESTATION

%EXPOSURE DOSE
MSTOT	= 9.9733

MSTOTBCKGR	=0. %

DOSEIV	=0. %

%INTEGRATION TIME

% TIME AT WHICH EXPOSURE BEGINS (HOUR)

%TIME AT WHICH EXPOSURE ENDS (HOUR)

^NUMBER OF HOURS BETWEEN DOSES (HOUR)

%TIME AT WHICH BACKGROUND EXPOSURE BEGINS

%TIME AT WHICH BACKGROUND EXPOSURE ENDS (HOUR)

5 BEGINNING OF MATING (HOUR) AT 4 5 YEARS OLD
i;SIMULATION TIME LIMIT (HOUR)

5 TRANSFER FROM MOTHER TO FETUS AT 1512 HOURS

9283634997E-07	% NG OF TCDD PER KG OF BW

0.1	% ORAL BACKGROUND EXPOSURE DOSE (NG/KG)

10

This document is a draft for review purposes only and does not constitute Agency policy.

C-31 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

DOSEIVLATE

= 0. %10

% TRANFER MOTHER TO FETUS CLEARANCE
CLPLA_FET	= 0.001 % MOTHER TO FETUS TRANFER CLEARANCE (L/HR)

C.2.3. Rat Standard Model
C.2.3.1. Model Code

PROGRAM: 'Three Compartment PBPK Model in Rat: Standard Model (Non-Gestation)'

!Rat Dioxin 3C June09 2clean icf afterKKfix v3 ratnongest.csl
!RAT~NON_GEST_ICF_F0 8 310 9.CSL ~	~ ~

!RAT_NON_GEST_ICF_F10 0 60 9.CSL

INITIAL ! INITIALIZATION OF PARAMETERS

!SIMULATION PARAMETERS
CONSTANT PARA_ZERO	=	Id-3 0

CONSTANT EXP_TIME_ON	=	0.0

(HOURS)

CONSTANT EXP_TIME_0FF =	900.0

(HOURS)

CONSTANT DAY_CYCLE	=	9 0 0.0

DOSES (HOURS)

CONSTANT BCK_TIME_ON	=	0.0

EXPOSURE BEGINS (HOURS)

CONSTANT BCK_TIME_OFF =	0.0

EXPOSURE ENDS (HOURS)

! TIME AT WHICH EXPOSURE BEGINS
! TIME AT WHICH EXPOSURE ENDS
! NUMBER OF HOURS BETWEEN
! TIME AT WHICH BACKGROUND
! TIME AT WHICH BACKGROUND

CONSTANT MW=322 IMOLECULAR WEIGHT (NG/NMOL)
CONSTANT SERBLO =0.55
CONSTANT UNITCORR = 1000

! EXPOSURE DOSES
CONSTANT MSTOTBCKGR
(UG/KG)

CONSTANT MSTOT
CONSTANT MSTOTsc
(UG/KG)

CONSTANT DOSEIV

0.0	!ORAL BACKGROUND EXPOSURE DOSE

10	!ORAL EXPOSURE DOSE (UG/KG)

0.0	!SUBCUTANEOUS EXPOSURE DOSE

0.0	! INJECTED DOSE (UG/KG)

!ORAL DOSE
MSTOT_NM
MSTOT NMBCKGR

MSTOT/MW

!AMOUNT IN NMOL/G

MSTOTBCKGR/MW !AMOUNT IN NMOL/G

!INTRAVENOUS DOSE
DOSEIV NM

DOSEIV/MW

!AMOUNT IN NMOL/G

!INITIAL GUESS OF THE FREE CONCENTRATION IN THE LIGAND (COMPARTMENT
INDICATED BELOW)====

CONSTANT CFLLI0	=	0.0	!LIVER (NMOL/ML)

This document is a draft for review purposes only and does not constitute Agency policy.

C-32 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

!BINDING CAPACITY (AhR) FOR NON LINEAR BINDING (COMPARTMENT INDICATED
BELOW) (NMOL/ML) ===

CONSTANT LIBMAX	= 3.5e-4	! LIVER (NMOL/ML), WANG ET AL.

1997

! PROTEIN AFFINITY CONSTANTS (1A2
(NMOL/ML)===

CONSTANT KDLI	= 1.0e-4

ET AL. 1997

CONSTANT KDLI2	= 4.Oe-2

ET AL. 2004

OR AhR, COMPARTMENT INDICATED BELOW)
! LIVER (AhR) (NMOL/ML), WANG
!LIVER (1A2) (NMOL/ML), EMOND

!EXCRETION AND ABSORPTION CONSTANT [RAT]

CONSTANT KST	= 0.36	! GASTRIC RATE CONSTANT (HR-1),

WANG ET AL. (1997)

CONSTANT KABS	= 0.4 8	!INTESTINAL ABSORPTION CONSTANT

(HR-1), WANG ET AL. 1997

!URINARY ELIMINATION CLEARANCE (ML/HR)

CONSTANT CLURI	=	0.01	!URINARY CLEARANCE (ML/HR),

EMOND ET AL. 2004

!INTERSPECIES VARIABLE ELIMINATION
CONSTANT KELV	=	0.15	! INTERSPECIES VARIABLE

ELIMINATION CONSTANT (1/HOUR) (OPTIMIZED), EMOND ET AL. 2004

! CONSTANT TO DIVIDE THE ABSORPTION INTO LYMPHATIC AND PORTAL FRACTIONS

CONSTANT A	=0.7

AL. 1997

!PARTITION COEFFICIENTS
CONSTANT PF	=	100

AL. 1997

CONSTANT PRE	=	1.5

ET AL. 1997

CONSTANT PLI	=	6.0

1997

! LYMPHATIC FRACTION, WANG ET

! ADIPOSE TISSUE/BLOOD, WANG ET
! REST OF THE BODY/BLOOD, WANG
! LIVER/BLOOD, WANG ET AL.

!PARAMETER FOR INDUCTION OF CYP 1A2 [MOUSE] ===
CONSTANT PAS_INDUC	=	1.0

0 = NO)

CONSTANT CYP1A2_10UTZ =	1.6

CONSTANT OF 1A2 (NMOL/ML), WANG ET AL. 1997

CONSTANT CYP1A2_1A1	=	1.6

(NMOL/ML), WANG ET AL. 1997

CONSTANT CYP1A2_1EC5 0 =	0.13

CYP1A2 (NMOL/ML) , WANG ET AL.	19 97

CONSTANT CYP1A2_1A2	=	1.6

(NMOL/ML) Wang et al (1997)

CONSTANT CYP1A2_1K0UT =	0.1

DEGRADATION (H-l), WANG ET AL.	1997

CONSTANT CYP1A2_1TAU
1997

CONSTANT CYP1A2 1EMAX

0.25

600

!	INCLUDE INDUCTION? (1 = YES,

!	DEGRADATION CONCENTRATION

!	BASAL CONCENTRATION OF 1A1

!	DISSOCIATION CONSTANT TCDD-

!	BASAL CONCENTRATION OF 1A2

!	FIRST ORDER RATE OF

!	HOLDING TIME (H), WANG ET AL.

!	MAXIMUM INDUCTION OVER BASAL

EFFECT (UNITLESS), WANG ET AL. 1997

This document is a draft for review purposes only and does not constitute Agency policy.

C-33 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CONSTANT HILL	=	0.6	!HILL CONSTANT; COOPERATIVELY LIGAND

BINDING EFFECT CONSTANT (UNITLESS)

!TISSUE BLOOD FLOW EXPRESSED AS A FRACTION OF CARDIAC OUTPUT
CONSTANT QFF = 0.069	! ADIPOSE TISSUE BLOOD FLOW

FRACTION (UNITLESS), WANG ET AL. 1997

CONSTANT QLIF = 0.183	! LIVER (UNITLESS), WANG ET AL.

1997

!DIFFUSIONAL PERMEABILITY FRACTION
CONSTANT PAFF	= 0.0910	! ADIPOSE (UNITLESS), WANG ET

AL. 1997

CONSTANT PAREF	= 0.0298	! REST OF THE BODY (UNITLESS),

WANG ET AL. 1997

CONSTANT PALIF	= 0.35	! LIVER (UNITLESS), WANG ET AL.

1997

!FRACTION OF TISSUE VOLUME (UNITLESS)

CONSTANT WLI0	= 0.0360	! LIVER, WANG ET AL. 1997

CONSTANT WF0	= 0.069	! BLOOD, WANG ET AL. 1997

!COMPARTMENT TISSUE BLOOD EXPRESSED AS A FRACTION OF THE TOTAL
COMPARTMENT VOLUME =========

CONSTANT WFB0	= 0.050	! ADIPOSE TISSUE, WANG ET

1997

CONSTANT WREB0	= 0.030	! REST OF THE BODY, WANG

1997

CONSTANT WLIB0	= 0.266	! LIVER , WANG ET AL. 199

!EXPOSURE SCENARIO FOR UNIQUE OR
! NUMBER OF EXPOSURES PER WEEK
CONSTANT WEEK_LACK	=0.0

(WEEK)

CONSTANT WEEK_PERIOD	= 168.0

(HOURS)

CONSTANT WEEK FINISH	= 168.0

AL.
ET AL.
7

REPETITIVE WEEKLY OR MONTHLY EXPOSURE
! DELAY BEFORE EXPOSURE ENDS
! NUMBER OF HOURS IN THE WEEK
! TIME EXPOSURE ENDS (HOURS)

!NUMBER OF EXPOSURES PER MONTH
CONSTANT MONTH_LACK	=0.0	! DELAY BEFORE EXPOSURE BEGINS

(MONTH)

!SET FOR BACKGROUND EXPOSURE=====:
!CONSTANT FOR BACKGROUND EXPOSURE^
CONSTANT Day_LACK_BG	=0.0

(HOURS)

CONSTANT Day_PERIOD_BG =24.0

!NUMBER OF EXPOSURES PER WEEK
CONSTANT WEEK_LACK_BG =0.0
EXPOSURE (WEEK)

CONSTANT WEEK_PERIOD_BG = 168.0
(HOURS)

CONSTANT WEEK FINISH BG = 168.0

! DELAY BEFORE EXPOSURE BEGINS
! LENGTH OF EXPOSURE (HOURS)

! DELAY BEFORE BACKGROUND

!NUMBER OF HOURS IN THE WEEK
! TIME EXPOSURE ENDS (HOURS)

!GROWTH CONSTANT FOR RAT

'CONSTANT FOR MOTHER BODY WEIGHT GROWTH

This document is a draft for review purposes only and does not constitute Agency policy.

C-34 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CONSTANT BW TO

= 250.0

!CHANGED FOR SIMULATION

! CONSTANT USED IN CARDIAC OUTPUT EQUATION

CONSTANT QCCAR =311.4
AL.

!CONSTANT (ML/MIN/KG), WANG ET

! COMPARTMENT LIPID EXPRESSED AS THE FRACTION OF TOTAL LIPID

CONSTANT F_TOTLIP
CONSTANT B_TOTLIP
CONSTANT RE_TOTLIP
CONSTANT LI TOTLIP

= 0.855
= 0.0033
= 0.019
= 0. 06

END	!END OF THE INITIAL SECTION

DYNAMIC !DYNAMIC SIMULATION SECTION

!ADIPOSE TISSUE (UNITLESS)
!BLOOD (UNITLESS)

!REST OF THE BODY (UNITLESS)
!LIVER (UNITLESS)

ALGORITHM
CINTERVAL
MAXTERVAL
MINTERVAL
VARIABLE
CONSTANT
(HOURS)

CINTXY = CINT
PFUNC = CINT

IALG
CINT
MAXT
MINT
T

TIMELIMIT

2
0.1
1.0e+10
1.0E-10
0.0
900. 0

GEAR METHOD

COMMUNICATION INTERVAL
MAXIMUM CALCULATION INTERVAL
MINIMUM CALCULATION INTERVAL

!SIMULATION TIME LIMIT

!TIME CONVERSION
DAY=T/24.0
WEEK =T/168.0
MONTH =T/730 . 0
YEAR=T/8 7 60.0

TIME IN DAYS
TIME IN WEEKS
TIME IN MONTHS
TIME IN YEARS

DERIVATIVE ! PORTION OF CODE THAT SOLVES DIFFERENTIAL EQUATIONS

!CHRONIC OR SUBCHRONIC EXPOSURE SCENARIO

!NUMBER OF EXPOSURES PER DAY
DAY_LACK	= EXP_TIME_ON

(HOURS)

DAY_PERIOD = DAY_CYCLE
DAY_FINISH = CINTXY
MONTH_PERIOD = TIMELIMIT
MONTH FINISH = EXP TIME OFF

! DELAY BEFORE EXPOSURE BEGINS

EXPOSURE PERIOD (HOURS)
LENGTH OF EXPOSURE (HOURS)
EXPOSURE PERIOD (MONTHS)
LENGTH OF EXPOSURE (MONTHS)

!NUMBER OF EXPOSURES PER DAY AND MONTH

DAY_FINISH_BG = CINTXY
MONTH_LACK_BG = B C K_TIME_ON
EXPOSURE BEGINS (MONTHS)
MONTH_PERIOD_BG = TIMELIMIT
(MONTHS)

MONTH_FINISH_BG = BCK_TIME_OFF
(MONTHS)

!	LENGTH OF EXPOSURE (HOURS)

!	DELAY BEFORE BACKGROUND

!	BACKGROUND EXPOSURE PERIOD

!	LENGTH OF BACKGROUND EXPOSURE

B = 1-A	! FRACTION OF DIOXIN ABSORBED IN

THE PORTAL FRACTION OF THE LIVER

This document is a draft for review purposes only and does not constitute Agency policy.

C-3 5 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

! BODY WEIGHT GROWTH EQUATION========

PARAMETER (BW_RMN = 1.0E-30)

WT0= (BW TO *(1.0+(0.41*T)/(1402.5+T+BW RMN)))

!VARIABILITY OF REST OF THE BODY DEPEND OTHERS ORGAN
WRE0 = (0.91 - (WLIB0*WLI0 + WFB0*WF0 + WLI0 + WF0))/(1.0+WREB0) !REST OF
THE BODY FRACTION; UPDATED FOR EPA ASSESSMENT
QREF = 1.0-(QFF+QLIF)	!REST OF BODY BLOOD FLOW

QTTQF = QFF+QREF+QLIF	! SUM MUST EQUAL 1

!COMPARTMENT VOLUME (G)
WF = WF0 * WTO
WRE = WRE0 * WTO
WLI = WLI0 * WTO

ADIPOSE

REST OF THE BODY
LIVER

!COMPARTMENT TISSUE BLOOD VOLUME
WFB = WFB0 * WF
WREB = WREB0 * WRE
WLIB = WLIB0 * WLI

(G)

ADIPOSE

REST OF THE BODY
LIVER

!CARDIAC OUTPUT FOR THE GIVEN BODY WEIGHT
QC= QCCAR* 60.0*(WTO/UNITCORR)**0.75

! COMPARTMENT BLOOD FLOW (ML/HR)
QF = QFF*QC
QLI = QLIF*QC
QRE = QREF*QC

RATE
QTTQ

= QF+QRE+QLI

ADIPOSE TISSUE BLOOD FLOW RATE
LIVER TISSUE BLOOD FLOW RATE
REST OF THE BODY BLOOD FLOW

! TOTAL FLOW RATE

!PERMEABILITY ORGAN FLOW (ML/HR)
PAF = PAFF*QF
PARE = PAREF*QRE
PALI = PALIF*QLI

ADIPOSE

REST OF THE BODY
LIVER TISSUE

!CONDITIONAL ORAL EXPOSURE (BACKGROUND EXPOSURE)

!EXPOSURE + !REPETITIVE EXPOSURE SCENARIO
IV= DOSEIV_NM * WTO !AMOUNT IN NMOL
MSTT= MSTOT_NM * WTO !AMOUNT IN NMOL
MSTTBCKGR =MSTOT_NMBCKGR *WT0

!REPETITIVE ORAL BACKGROUND EXPOSURE SCENARIOS
DAY_EX P O S U RE_B G = PULSE(DAY_LACK_BG,DAY_PERIOD_BG,DAY_FINISH_BG)
WEEK_EXPOSURE_BG = PULSE(WEEK_LACK_BG,WEEK_PERIOD_BG,WEEK_FINISH_BG)
MONTH_EXPOSURE_BG = PULSE(MONTH_LACK_BG,MONTH_PERIOD_BG,MONTH_FINISH_BG)

MSTTCH_BG = (DAY_EXPOSURE_BG*WEEK_EXPOSURE_BG*MONTH_EXPOSURE_BG)*MSTTBCKGR
MSTTFR_BG = MSTTBCKGR/CINT

CYCLE BG =DAY EXPOSURE BG*WEEK EXPOSURE BG*MONTH EXPOSURE BG

IF (MSTTCH_BG.EQ.MSTTBCKGR) THEN
ABSMSTT_GB= MSTTFR_BG

ELSE

ABSMSTT GB = 0.0

This document is a draft for review purposes only and does not constitute Agency policy.

C-36 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

END IF

!REPETITIVE ORAL MAIN EXPOSURE SCENARIO
DAY_EXPOSURE = PULSE(DAY_LACK,DAY_PERIOD,DAY_FINISH)

WEEK_EXPOSURE = PULSE(WEEK_LACK,WEEK_PERIOD,WEEK_FINISH)

MONTH_EXPOSURE = PULSE(MONTH_LACK,MONTH_PERIOD,MONTH_FINISH)

MSTTCH = (DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE)*MSTT
CYCLE = DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE
MSTTFR = MSTT/CINT

SUMEXPEVENT= integ (CYCLE,0.0) INUMBER OF CYCLE GENERATE DURING SIMULATION

!CONDITIONAL ORAL EXPOSURE
IF (MSTTCH.EQ.MSTT) THEN

ABSMSTT= MSTTFR
ELSE

ABSMSTT =0.0
END IF

CYCLETOT=INTEG(CYCLE, 0.0)

!MASS CHANGE IN THE LUMEN
RMSTT = -(KST+KABS)*MST+ABSMSTT +ABSMSTT_GB ! RATE OF CHANGE (NMOL/H)
MST = INTEG(RMSTT,0.0) !AMOUNT OF STAY IN DUODENUM (NMOL)

!ABSORPTION IN LYMPH CIRCULATION
LYRMLUM = KABS*MST*A

LYMLUM = INTEG(LYRMLUM,0.0)

!ABSORPTION IN PORTAL CIRCULATION
LIRMLUM = KABS*MST*B

LIMLUM = INTEG(LIRMLUM,0.0)

!PERCENT OF DOSE REMAINING IN THE GI TRACT
PRCT_remain_GIT = (MST/(MSTT+PARA_ZERO))*100.0

!ABSORPTION of Dioxin by IV route	

IVR= IV/PFUNC ! RATE FOR IV INFUSION IN BLOOD
EXPIV= IVR * (1.0-STEP(PFUNC))

IVDOSE = integ(EXPIV,0.0)

!SYSTEMIC BLOOD COMPARTMENT
! MODIFICATION ON OCTOBER 6, 2009
CB=(QF*CFB+QRE*CREB+QLI*CLIB+EXPIV+LYRMLUM)/(QC+CLURI) !

CA = CB

!URINARY EXCRETION BY KIDNEY
! MODIFICATION ON OCTOBER 6, 2009
RAURI = CLURI *CB

AURI = INTEG(RAURI,0.0)

!CONVERSION EQUATION POST SIMULATION
PRCT_B = (CB/(MSTT+PARA_ZERO))*100.0

This document is a draft for review purposes only and does not constitute Agency policy.

C-37 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CBNGKG = CB*MW*UNITCORR ![NG/KG]

CBSNGKGLIADJ= (CB*MW*UNITCORR*(1.0/B_TOTLIP)*(1.0/SERBLO))![NG of TCDD
Serum/Kg OF LIPIP]

!ADIPOSE TISSUE COMPARTMENT
! TISSUE BLOOD SUBCOMPARTMENT

RAFB = QF*(CA-CFB)-PAF*(CFB-CF/PF)	!(NMOL/HR)

AFB = INTEG(RAFB,0.0)	!(NMOL)

CFB = AFB/WFB	!(NMOL/ML)
!TISSUE SUBCOMPARTMENT

RAF = PAF*(CFB-CF/PF)	!(NMOL/HR)

AF = INTEG(RAF,0.0)	!(NMOL)

CF = AF/WF	!(NMOL/ML)

!CONVERSION EQUATION POST SIMULATION
CFTOTAL = (AF + AFB)/(WF + WFB)	!TOTAL CONCENTRATION IN NMOL/ML

PRCT_F = (CFTOTAL/(MSTT+PARA_ZERO))*100.0 ! PRCENT OF DOSE IN FAT
CFNGKG = CFTOTAL*MW*UNITCORR	! CONCENTRATION [NG/KG]

!REST OF THE BODY COMPARTMENT
! TISSUE BLOOD SUBCOMPARTMENT

RAREB= QRE*(CA-CREB)-PARE*(CREB-CRE/PRE)	!(NMOL/HR)

AREB = INTEG(RAREB,0.0)	!(NMOL)

CREB = AREB/WREB	!(NMOL/ML)
! TISSUE COMPARTMENT

RARE = PARE*(CREB - CRE/PRE)	!(NMOL/HR)

ARE = INTEG(RARE,0.0)	!(NMOL)
CRE = ARE/WRE !(NMOL/ML)

!CONVERSION EQUATION POST SIMULATION

CRETOTAL= (ARE + AREB)/(WRE + WREB)	! TOTAL CONCENTRATION IN

NMOL/ML

PRCT_RE = (CRETOTAL/(MSTT+PARA_ZERO))*100.0
CTREPGG= CRETOTAL*MW*UNITCORR !(PG/ML)

AUC_REPPG = integ(CTREPGG,0.0)

!LIVER COMPARTMENT
! TISSUE BLOOD COMPARTMENT
RALIB = QLI*(CA-CLIB)-PALI*(CLIB-CFLLIR)+LIRMLUM
ALIB = INTeg(RALIB,0.0)

CLIB = ALIB/WLIB
! TISSUE COMPARTMENT
RALI = PALI*(CLIB-CFLLIR)-REXCLI
ALI = integ(RALI,0.0)

CLI = ALI/WLI

!(NMOL/HR)
!(NMOL)

!(NMOL/HR)

!(NMOL)
!(NMOL/ML)

PARAMETER (LIVER_1RMN = 1.0E-30)

CFLLI= IMPLC(CLI-(CFLLIR*PLI+(LIBMAX*CFLLIR/(KDLI+CFLLIR &
+LIVER_1RMN))+((CYP1A2_103*CFLLIR/(KDLI2+CFLLIR &

+LIVER_1RMN)*PAS_INDUC)))-CFLLIR,CFLLI0) ! FREE TCDD CONCENTRATION IN LIVER
CFLLIR=DIM(CFLLI,0.0)

CBNDLI= LIBMAX*CFLLIR/(KDLI+CFLLIR+LIVER_1RMN) !BOUND CONCENTRATION
This document is a draft for review purposes only and does not constitute Agency policy.

C-3 8 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!CONVERSION EQUATION POST SIMULATION
CLITOTAL= (ALI + ALIB)/(WLI + WLIB)

NMOL/ML

PRCT_LI = (CLITOTAL/(MSTT+PARA_ZERO))*100.0
rec_occ_AHR= (CFLLIR/(KDLI+CFLLIR+1) )*100 . 0
OCCUPANCY

PROT_occ_lA2= (CFLLIR/(KDLI2+CFLLIR))*100.0
OCCUPANCY

CLINGKG =(CLITOTAL*MW*UNITCORR)

CBNDLINGKG = CBNDLI*MW*UNITCORR
AUCLI_NGKGH=INTEG(CLINGKG, 0.0)

CLINGG=CLITOTAL*MW

!VARIABLE ELIMINATION HALF-LIFE BASED ON THE CONCENTRATION OF CYP1A2
KBILE_LI_T =((CYP1A2_10UT-CYP1A2_1A2)/CYP1A2_1A2)*Kelv ! INDUCED BILIARY
EXCRETION RATE CONSTANT

REXCLI= (KBILE_LI_T*CFLLIR*WLI) ! DOSE-DEPENDENT BILIARY EXCRETION RATE
EXCLI = INTEG(REXCLI,0.0)

!CHEMICAL IN CYP450 (1A2) COMPARTMENT
!===PARAMETER FOR INDUCTION OF CYP1A2

CYP1A2_1KINP = CYP1A2_1K0UT* CYP1A2_10UTZ ! BASAL RATE OF CYP1A2 PRODUCTION
SET EQUAL TO BASAL RATE OF DEGREDATION

! TOTAL CONCENTRATION IN

! PERCENT OF AhR
! PERCENT OF 1A2

! MODIFICATION ON OCTOBER 6, 2009
CYP1A2_10UT =INTEG(CYP1A2_1KINP * (1.0 + CYP1A2_1EMAX *(CBNDLI+1.0e-
30)**HILL &

/(CYP1A2_1EC50**HILL + (CBNDLI+1.Oe-30)**HILL)) &-
- CYP1A2_1K0UT*CYP1A2_10UT, CYP1A2_10UTZ)

! EQUATIONS INCORPORATING DELAY OF CYP1A2 PRODUCTION (NOT USED IN
SIMULATIONS)

CYP1A2_1R02 = (CYP1A2_10UT - CYP1A2_102)/ CYP1A2_1TAU

CYP1A2_102 =INTEG(CYP1A2_1R02, CYP1A2_1A1)

CYP1A2_1R03 = (CYP1A2_102 - CYP1A2_103)/ CYP1A2_1TAU
CYP1A2_103 =INTEG(CYP1A2_1R03, CYP1A2_1A2)

! 	CHECK MASS BALANCE	

BDOSE= LYMLUM+LIMLUM+IVDOSE

BMASSE = EXCLI+AURI+AFB+AF+AREB+ARE+ALIB+ALI
BDIFF = BDOSE-BMASSE

!	BODY BURDEN	

BBNGKG =(((AFB+AF+AREB+ARE+ALIB+ALI)*MW)/(WTO/UNITCORR)) !
! 	END OF THE SIMULATION COMMAND	

TERMT (T.GE. TimeLimit, 'Time limit has been reached.')

END ! END OF THE DERIVATIVE SECTION

END ! END OF THE DYNAMIC SIMULATION SECTION

END ! END OF THE PROGRAM.

This document is a draft for review purposes only and does not constitute Agency policy.

C-39 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

C.2.3.2. Input Files

C.2.3.2.1. Cantoni et al. (1981).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

%Cantoni et al. 1981

%protocol: oral exposure 1 dose/week for 45 weeks; female CD-COBS rats

%Rat Dioxin 3C June09 2clean.csl

%RAT_NON_GEST_ICF_F083109.CSL (now 09-11-09)

%dose levels: 0.01, 0.1, 1 ug/kg 1 dose/week for 45 weeks

%dose levels: 10, 100, 1000 ng/kg 1 dose/week for 45 weeks

%dose levels equivalent to: 1.43, 14.3 143 ng/kg 7 days/weeks for 45 weeks

MAXT
CINT

EXP_TIME_ON

EX P_T I ME_0 F F

DAY_CYCLE

B C K_TIME_ON

BCK_TIME_OFF

TIMELIMIT

BW_T0

(g)

o. 01

0.1

0.

7560
168
0.

0.

7560
125

%delay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

¦;DELAY BEFORE BACGROUND EXPOSURE (HOUR)
iTIME OF BACKGROUND EXPOSURE STOP (HOUR)
¦;SIMULATION LIMIT TIME (HOUR)

i Body weight at the beginning of the simulation

iEXPOSURE DOSE SCENARIOS	(UG/KG)

%MSTOT =0.01	% exposure dose ug/kg

%MSTOT =0.1	% exposure dose ug/kg

MSTOT =1	% exposure dose ug/kg

C.2.3.2.2. Chu et al (2007).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% Chu et al. 2 0
%protocol: ora
%dose levels: 0
% dose levels =
MAXT
CINT

EXP_TIME_ON
after start of
EX P_T I ME_0 F F
every two weeks
DAY_CYCLE
B C K_TIME_ON
BCK_TIME_OFF
TIMELIMIT
BW_T0

simulation (g) ;

07

1 exposure daily for 28 days

.0025, 0.025, 0.250, 1.0 ug/kg every day for 28 days
2.5, 25, 250, 1000 ng/kg every day for 28 days
= 0. 01
= 0.1

= 0.	%delay before begin exposure (HOUR) 5 weeks

experiment (age = 12 weeks)

= 672.	%TIME EXPOSURE STOP (HOUR); 30 doses, 1

24 .

0.

0.

672 .
200.

once every two weeks

%DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
%TIME OF BACKGROUND EXPOSURE STOP (HOUR)
i;SIMULATION LIMIT TIME (HOUR)

% Body weight at the beginning of the

corresponds to 12 week old female

iEXPOSURE DOSE SCENARIOS (UG/KG)

This document is a draft for review purposes only and does not constitute Agency policy.

C-40 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

%MSTOT	= 0.0025

%MSTOT	= 0.025

%MSTOT	= 0.250

MSTOT	=1.0

% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)

C.2.3.2.3. Crofton et al. (2005).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% Crofton et al. 2005

%protocol: oral exposure daily for 4 days

%dose levels: 0.0001, 0.003, 0.01, 0.03, 0.1, 0.3, 1, 3, and 10 ug/kg every
day for four days

%dose levels: 0.1, 3, 10, 30, 100, 300, 1000, 3000, and 10000 ng/kg every day
for four days

MAXT	=	0.01

CINT	=	0.1

EXP_TIME_ON	=	0.

after start of experiment (age

EX P_TIME_0 FF =	96.

every two weeks

DAY_CYCLE	=	24.

B C K_T I ME_ON	=	0.

BCK_TIME_OFF =	0.

TIMELIMIT	=	96.

BW TO	=	250

%delay before begin exposure (HOUR) 5 weeks
12 weeks)

%TIME EXPOSURE STOP (HOUR); 30 doses, 1

% once every two weeks

%DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
%TIME OF BACKGROUND EXPOSURE STOP (HOUR)
%SIMULATION LIMIT TIME (HOUR)

% Body weight at the beginning of the

simulation (g); corresponds to 12 week old female

iEXPOSURE DOSE SCENARIOS (UG/KG)

MSTOT	= 0.0001

%MSTOT	= 0.003

%MSTOT	=0.01

%MSTOT	=0.03

%MSTOT	=0.1

%MSTOT	=0.3

%MSTOT	=1. %

%MSTOT	=3. %

MSTOT	=10. %

% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)

>	ORAL EXPOSURE DOSE (UG/KG)

>	ORAL EXPOSURE DOSE (UG/KG)
ORAL EXPOSURE DOSE (UG/KG)
ORAL EXPOSURE DOSE (UG/KG)
ORAL EXPOSURE DOSE (UG/KG)

C.2.3.2.4. Fattore et al. (2000).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% Fattore et al. 2000

%built and check in August 7 2009

%protocol: oral exposure in diet for 13 weeks; SD rats

%dose levels: 0.02, 0.1, 0.2, 2 ug/kg 7 days/week for 13 weeks

%dose levels equivalent to: 20, 100, 200, 2000 ng/kg 7 days/week for 13 weeks

MAXT = 0.01

This document is a draft for review purposes only and does not constitute Agency policy.

C-41 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

CINT =0.1

EXP_TIME_ON

EX P_T I ME_0 F F

DAY_CYCLE

B C K_TIME_ON

BCK_TIME_OFF

TIMELIMIT

BW_T0

(G)

0.

2184
24
0.

0.

2184
150

%TIME AT WHICH EXPOSURE BEGINS (HOUR)
iTIME AT WHICH EXPOSURE ENDS (HOUR)

iTIME AT WHICH BACKGROUND EXPOSURE BEGINS (HOUR)
iTIME AT WHICH BACKGROUND EXPOSURE ENDS (HOUR)
¦;SIMULATION TIME LIMIT (HOUR)

i BODY WEIGHT AT THE BEGINNING OF THE SIMULATION

iEXPOSURE DOSE	SCENARIOS (UG/KG)

%MSTOT	=0.02

%MSTOT	=0.1

%MSTOT	= 0.2

MSTOT	= 2

% EXPOSURE DOSE IN UG/KG
% EXPOSURE DOSE IN UG/KG
% EXPOSURE DOSE IN UG/KG
EXPOSURE DOSE IN UG/KG

C.2.3.2.5. Franc et al. (2001). Sprague Dawley rats

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

%	Franc et al. 2001

%	Non-gestational rat model

%	dose levels: 0.140, 0.420, and 1.400 ug/kg every 2 weeks for 22 weeks

%	dose levels: 140, 420, and 1400 ng/kg every 2 weeks for 22 weeks

%	dose levels equivalent to 10, 30, and 100 ng/kg/day

MAXT

II

o
o



CINT

II

o



EXP TIME ON

= 0.

%delay before begin exposure (HOUR)

EXP TIME OFF

= 3696.

%TIME EXPOSURE STOP (HOUR)

DAY CYCLE

= 336.



BCK TIME ON

= 0.

%DELAY BEFORE BACGROUND EXPOSURE (HOUR)

BCK TIME OFF

= 0.

%TIME OF BACKGROUND EXPOSURE STOP (HOUR)

TIMELIMIT

= 3696.

%SIMULATION LIMIT TIME (HOUR)

BW TO

= 200.

% Body weight at the beginning of the

simulation (g) ;

corresponds

to approximate weight of females 10 weeks old

iEXPOSURE DOSE	SCENARIOS	(UG/KG)

%MSTOT	=0.14	% ORAL EXPOSURE DOSE (UG/KG)

%MSTOT	=0.42	% ORAL EXPOSURE DOSE (UG/KG)

MSTOT	=1.4	% ORAL EXPOSURE DOSE (UG/KG)

C.2.3.2.6. Franc et al. (2001). Long-Evans rats

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

%	Franc et al. 2001

%	Non-gestational rat model

%	dose levels: 0.140, 0.420, and 1.400 ug/kg every 2 weeks for 22 weeks

%	dose levels: 140, 420, and 1400 ng/kg every 2 weeks for 22 weeks

%	dose levels equivalent to 10, 30, and 100 ng/kg/day

This document is a draft for review purposes only and does not constitute Agency policy.

C-42 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

MAXT

II

o
o

I—1



CINT

II

o

I—1



EXP TIME ON

= 0.

%delay before begin exposure (HOUR)

EXP TIME OFF

= 3696.

%TIME EXPOSURE STOP (HOUR)

DAY CYCLE

= 336.



BCK TIME ON

= 0.

%DELAY BEFORE BACGROUND EXPOSURE (HOUR)

BCK TIME OFF

= 0.

%TIME OF BACKGROUND EXPOSURE STOP (HOUR)

TIMELIMIT

= 3696.

%SIMULATION LIMIT TIME (HOUR)

BW TO

= 190.

% Body weight at the beginning of the

simulation (g) ;

corresponds

to approximate weight of females 10 weeks old

iEXPOSURE DOSE	SCENARIOS	(UG/KG)

%MSTOT	=0.14	% ORAL EXPOSURE DOSE (UG/KG)

%MSTOT	=0.42	% ORAL EXPOSURE DOSE (UG/KG)

MSTOT	=1.4	% ORAL EXP

C.2.3.2.7. Franc et al. (2001). Hans Wistar rats

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

%	Franc et al. 2001

%	Non-gestational rat model

%	dose levels: 0.140, 0.420, and 1.400 ug/kg every 2 weeks for 22 weeks

%	dose levels: 140, 420, and 1400 ng/kg every 2 weeks for 22 weeks

%	dose levels equivalent to 10, 30, and 100 ng/kg/day

MAXT

II

o
o

I—1



CINT

II

o

I—1



EXP TIME ON

= 0.

%delay before begin exposure (HOUR)

EXP TIME OFF

= 3696.

%TIME EXPOSURE STOP (HOUR)

DAY CYCLE

= 336.



BCK TIME ON

= 0.

%DELAY BEFORE BACGROUND EXPOSURE (HOUR)

BCK TIME OFF

= 0.

%TIME OF BACKGROUND EXPOSURE STOP (HOUR)

TIMELIMIT

= 3696.

%SIMULATION LIMIT TIME (HOUR)

BW TO

= 205.

% Body weight at the beginning of the

simulation (g) ;

corresponds

to approximate weight of females 10 weeks old

iEXPOSURE DOSE	SCENARIOS	(UG/KG)

%MSTOT	=0.14	% ORAL EXPOSURE DOSE (UG/KG)

%MSTOT	=0.42	% ORAL EXPOSURE DOSE (UG/KG)

MSTOT	=1.4	% ORAL EXP

C.2.3.2.8. Hassoun et al. (2000).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG
% Hassoun et al. 2000

%protocol: oral exposure for 13 weeks; SD rats

%dose levels: 0.003, 0.010, 0.022, 0.046 0.1 ug/kg 5 days/weeks for 13 weeks
%dose levels equivalent to: 3, 10, 22, 46 100 ng/kg 5 days/weeks for 13 weeks
%dose levels equivalent to: 2.14, 7.14, 15.7, 32.9 71.4 ng/kg 7 days/weeks
for 13 weeks

This document is a draft for review purposes only and does not constitute Agency policy.

C-43 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

MAXT	= 0.01

CINT	=0.1

EXP_TIME_ON	= 0.

EX P_TIME_0 FF	= 2184.

DAY_CYCLE	= 24.

WEEK_PERIOD	= 168.

WEEK_FINISH	= 119.

B C K_T I ME_ON	= 0.

BCK_TIME_OFF	= 0.

TIMELIMIT	= 2184.

BW_T0	=215.
simulation (g)

idelay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

iDELAY BEFORE BACKGROUND EXPOSURE (HOUR)
iTIME OF BACKGROUND EXPOSURE STOP (HOUR)
iSIMULATION LIMIT TIME (HOUR)

% Body weight at the beginning of the

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT	= 0.003	% exposure dose ug/kg

%MSTOT	= 0.010	% exposure dose ug/kg

%MSTOT	= 0.022	% exposure dose ug/kg

%MSTOT	= 0.046	% exposure dose ug/kg

MSTOT	=0.1	% exposure dose ug/kg

C.2.3.2.9. Hutt et al. (2008).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

%	Hutt et al. 2008

%	Non-gestational rat model

%	dose levels: 0.050 ug/kg every week for 13 weeks

%	dose levels: 50 ng/kg every week for 13 weeks

%	dose levels equivalent to 7.14 ng/kg/day

MAXT
CINT

EXP_TIME_ON
EX P_T I ME_0 F F
DAY_CYCLE
B C K_TIME_ON
BCK_TIME_OFF
TIMELIMIT
BW TO

0. 01

0.1

0.

2184 .
168 .
0.

0.

2184 .
4 . 5

idelay before begin exposure (HOUR)

%TIME EXPOSURE STOP (HOUR)

%DELAY BEFORE BACGROUND EXPOSURE (HOUR)
%TIME OF BACKGROUND EXPOSURE STOP (HOUR)

%SIMULATION LIMIT TIME (HOUR)
i Body weight at the beginning of the

simulation (g); corresponds to approximate weight of females 10 weeks old

iEXPOSURE DOSE SCENARIOS (UG/KG)

MSTOT

= 0. 05

ORAL EXPOSURE DOSE (UG/KG)

C.2.3.2.10. Kitchin and Woods (1979)

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% Kitchen and Woods 1979
%protocol: single oral gavage

This document is a draft for review purposes only and does not constitute Agency policy.

C-44 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

idose levels: 0.0006, 0.002, 0.004, 0.020, 0.060, 0.200, 0.600, 2.000,

5.000, 20.000 ug/kg single
% dose levels = 0.6, 2, 4,
oral gavage

MAXT	=	0.001

CINT	=	0.1

EXP_TIME_ON	=	0.

EX P_TIME_0 FF	=	24.

DAY_CYCLE	=	24.

B C K_T I ME_ON	=	0.

BCK_TIME_OFF	=	0.

TIMELIMIT	=	24.

BW_T0	=	225.
simulation (g)

oral gavage

20, 60, 200, 600, 2000, 5000, 20000 ng/kg single

idelay before begin exposure (HOUR)

%TIME EXPOSURE STOP (HOUR)

% daily

%DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
%TIME OF BACKGROUND EXPOSURE STOP (HOUR)
%SIMULATION LIMIT TIME (HOUR)

% Body weight at the beginning of the

iEXPOSURE DOSE
%MSTOT
%MSTOT
%MSTOT
%MSTOT
%MSTOT
%MSTOT
%MSTOT
%MSTOT
%MSTOT
MSTOT

SCENARIOS (UG/KG)
= 0.0006
= 0.002
= 0.004
= 0.020
= 0.060
= 0.200
= 0.600
= 2.000
= 5.000
= 20.000

i ORAL EXPOSURE DOSE (UG/KG)
i ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
ORAL EXPOSURE DOSE (UG/KG)
i ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)

C.2.3.2.11. Kociba et al. (1976) (13 weeks).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% Kociba et al. 1976.

%built and check in August 7 2009

%protocol: 5 days/week exposure for 13 weeks; SD rats
%Rat Dioxin 3C June09 2clean.csl
%RAT~NON_GEST_ICF_F083109.CSL (now 09-11-09)

%dose levels: 0.001, 0.01, 0.1, 1 ug/kg 5 days/weeks for 13 weeks

%dose levels: 1, 10, 100, 1000 ng/kg 5 days/weeks for 13 weeks

%dose levels equivalent to: 0.714, 7.14, 71.4, 714 ng/kg/d (adj) 7 days/weeks

for 13 weeks

MAXT
CINT

EXP_TIME_ON

EX P_T I ME_0 F F

WEEK_PERIOD

WEEK_FINISH

DAY_CYCLE

B C K_TIME_ON

BCK_TIME_OFF

TIMELIMIT

BW_T0

simulation (g)

0. 001

0.1

0.

2184
168
119
24
0.

0.

2184
180

idelay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

iDELAY BEFORE BACGROUND EXPOSURE (HOUR)
iTIME OF BACKGROUND EXPOSURE STOP (HOUR)
iSIMULATION LIMIT TIME (HOUR)
i Body weight at the begeniong of the

This document is a draft for review purposes only and does not constitute Agency policy.

C-45 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

•;EXPOSURE	DOSE SCENARIOS (UG/KG)

iMSTOT	= 0.001

%MSTOT	=0.01

%MSTOT	=0.1

MSTOT	= 1

C.2.3.2.12. Kociba et al. (1978) (female) (104 weeks).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% Kociba et al, 1978.

%built and check in August 7 2009

%protocol: daily dietary exposure for 104 weeks; SD rats
%dose levels: 0.001, 0.01, 0.1 ug/kg 7 days/week for 104 weeks
%dose levels: 1, 10, 100 ng/kg 7 days/week for 104 weeks

MAXT	=	0.01

CINT	=	0.1

EXP_TIME_ON	=	0.

EX P_T I ME_0 F F	=	17472

DAY_CYCLE	=	2 4

B C K_T I ME_ON	=	0.
(HOUR)

BCK_TIME_OFF	=	0.
(HOUR)

TIMELIMIT	=	17472

BW_T0	=	180
SIMULATION (G)

iTIME AT WHICH EXPOSURE BEGINS (HOUR)
iTIME AT WHICH EXPOSURE ENDS (HOUR)

iTIME AT WHICH BACKGROUND EXPOSURE BEGINS

iTIME AT WHICH BACKGROUND EXPOSURE ENDS

iSIMULATION TIME LIMIT (HOUR)
i BODY WEIGHT AT THE BEGINNING OF THE

%EXPOSURE DOSE SCENARIOS (UG/KG)
%MSTOT	= 0.001

%MSTOT	=0.01

MSTOT	=0.1

% EXPOSURE DOSE IN UG/KG
% EXPOSURE DOSE IN UG/KG
% EXPOSURE DOSE IN UG/KG

C.2.3.2.13. Kociba et al (1978) (male) (104 weeks).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% Kociba et al, 1978.

%built and check in August 7 2009

%protocol: daily dietary exposure for 104 weeks; SD rats
%dose levels: 0.001, 0.01, 0.1 ug/kg 7 days/week for 104 weeks
%dose levels: 1, 10, 100 ng/kg 7 days/week for 104 weeks

MAXT

II

o
o

I—1







CINT

II

o

I—1







EXP TIME ON

= 0.

%TIME

AT

WHICH

EXP TIME OFF

= 17472

%TIME

AT

WHICH

DAY CYCLE

= 24







BCK TIME ON

= 0.

%TIME

AT

WHICH

(HOUR)









(HOUR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-46 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

BCK_TIME_OFF = 0.
(HOUR)

TIMELIMIT	= 17472

BW_T0	= 250

SIMULATION (G)

iTIME AT WHICH BACKGROUND EXPOSURE ENDS

iSIMULATION TIME LIMIT (HOUR)
i BODY WEIGHT AT THE BEGINNING OF THE

•;EXPOSURE	DOSE SCENARIOS (UG/KG)

%MSTOT	= 0.001

%MSTOT	=0.01

MSTOT	=0.1

% EXPOSURE DOSE IN UG/KG
% EXPOSURE DOSE IN UG/KG
EXPOSURE DOSE IN UG/KG

C.2.3.2.14. Latchoumycandane and Mathur (2002).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% Latchoumycandane and Mathur 2002.

%built and check in August 7 2009

%protocol: 1 time per day for 45 days oral gavage
%Rat Dioxin 3C June09 2clean.csl
%RAT~NON_GEST_ICF_F083109.CSL (now 09-11-09)

%dose levels: 0.001, 0.01, 0.1 ug/kg daily for 45 days
%dose levels: 1, 10, 100 ng/kg daily for 45 days

MAXT
CINT

EXP_TIME_ON

EX P_T I ME_0 F F

DAY_CYCLE

B C K_TIME_ON

BCK_TIME_OFF

TIMELIMIT

BW_T0

simulation (g)

0. 01

0.1

0.

1080
24
0.

0.

1080
200

delay before begin exposure (HOUR)

TIME EXPOSURE STOP (HOUR)

DELAY BEFORE BACGROUND EXPOSURE (HOUR)
TIME OF BACKGROUND EXPOSURE STOP (HOUR)
SIMULATION LIMIT TIME (HOUR)

Body weight at the beginning of the

iEXPOSURE DOSE	SCENARIOS (UG/KG)

%MSTOT	= 0.001

%MSTOT	=0.01

MSTOT	=0.1

% exposure dose ug/kg
i; exposure dose ug/kg
exposure dose ug/kg

C.2.3.2.15. Lietal. (1997).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% Li et al 1997
% created 1/10/10
% Non-gestational rat model

% dose levels: 3, 10, 30, 100, 300, 1000, 3000, 10000, 30000 nkd one dose via
gavage, sacrificed 24 hrs later

MAXT	= 0.1

CINT	= 0.1

This document is a draft for review purposes only and does not constitute Agency policy.

C-47 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

EXP TIME ON

= 0.

%delay before begin exposure (HOUR)



EXP TIME OFF

= 24.

%TIME EXPOSURE STOP (HOUR)



DAY CYCLE

= 24.





BCK TIME ON

= 0.

%DELAY BEFORE BACKGROUND EXPOSURE

(HOUR)

BCK TIME OFF

= 0.

%TIME OF BACKGROUND EXPOSURE STOP

(HOUR)

TIMELIMIT

= 24.

%SIMULATION LIMIT TIME (HOUR)



BW TO

= 56.5

% Body weight at the beginning of

the

simulation (g)







iEXPOSURE DOSE
MSTOT

SCENARIOS (UG/KG)

= 0.003	% ORAL EXPOSURE DOSE (UG/KG)

%MSTOT

= 0.

01

O

o

ORAL

EXPOSURE

DOSE

(UG/KG)

%MSTOT

= 0.

03

o
o

ORAL

EXPOSURE

DOSE

(UG/KG)

%MSTOT

= 0.

1

o
o

ORAL

EXPOSURE

DOSE

(UG/KG)

%MSTOT

= 0.

3

o
o

ORAL

EXPOSURE

DOSE

(UG/KG)

%MSTOT

= 1.



o
o

ORAL

EXPOSURE

DOSE

(UG/KG)

%MSTOT

= 3.



o
o

ORAL

EXPOSURE

DOSE

(UG/KG)

%MSTOT

= 10



o
o

ORAL

EXPOSURE

DOSE

(UG/KG)

%MSTOT

= 30



o
o

ORAL

EXPOSURE

DOSE

(UG/KG)

C.2.3.2.16. Murray et al. (1979).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% Murray et al 1979

%built and check in August 7 2009

%protocol: dietary exposure for 3 generations (assume 120 day exposure for
each)

%dose levels: 0.001 0.01, 0.1 ug/kg/d
%dose levels: 1, 10, 100 ng/kg/d

MAXT	=	0.01

CINT	=	0.1

EXP_TIME_ON	=	0.

EX P_TIME_0 FF	=	2880

CORRESPONDS TO	12 0	DAYS

DAY_CYCLE	=	24.

B C K_T I ME_ON	=	0.
(HOUR)

BCK_TIME_OFF	=	0.
(HOUR)

TIMELIMIT	=	2880

BW_T0	=	4.5
SIMULATION (G)

%TIME AT WHICH EXPOSURE BEGINS (HOUR)
iTIME AT WHICH EXPOSURE ENDS (HOUR);

OF EXPOSURE

%TIME AT WHICH BACKGROUND EXPOSURE BEGINS

%TIME AT WHICH BACKGROUND EXPOSURE ENDS

i;SIMULATION TIME LIMIT (HOUR)

% BODY WEIGHT AT THE BEGINNING OF THE

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT
%MSTOT
MSTOT

= 0.001
= 0. 01
= 0.1

% ORAL EXPOSURE DOSE IN UG/KG
i ORAL EXPOSURE DOSE IN UG/KG
ORAL EXPOSURE DOSE IN UG/KG

This document is a draft for review purposes only and does not constitute Agency policy.

C-48 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

C.2.3.2.17. NTP (1982) (female) (chronic).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

%NTP 1982

%built and check in August 7 2009

%protocol: twice weekly gavage for 104 weeks + 3 week observation period
%Rat Dioxin 3C June09 2clean.csl
%RAT~NON_GEST_ICF_F083109.CSL (now 09-11-09)

%dose levels: 0.005, 0.025, 0.25 ug/kg biweekly for 104 weeks + 3 week
observation period

%dose levels: 5, 25, 250 ng/kg biweekly for 104 weeks + 3 week observation
period

%dose levels equivalent to: 1.43, 7.14, 71.4 ng/kg/day (adj)

MAXT
CINT

EXP_TIME_ON

EX P_T I ME_0 F F

DAY_CYCLE

B C K_TIME_ON

BCK_TIME_OFF

TIMELIMIT

BW_T0

simulation (g)

0. 01

0.1

0.

17472
84
0.

0.

17472
250

idelay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

¦;DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
iTIME OF BACKGROUND EXPOSURE STOP (HOUR)
¦;SIMULATION LIMIT TIME (HOUR)
i Body weight at the beginning of the

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT	= 0.005	% exposure dose ug/kg

%MSTOT	= 0.025

MSTOT	= 0.25

C.2.3.2.18. NTP (1982) (male) (chronic).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

%NTP 1982

%built and check in august 7 2009

%protocol: twice weekly gavage for 104 weeks + 3 week observation period
%Rat Dioxin 3C June09 2clean.csl
%RAT~NON_GEST_ICF_F083109.CSL (now 09-11-09)

%dose levels: 0.005, 0.025, 0.25 ug/kg biweekly for 104 weeks + 3 week
observation period

%dose levels: 5, 25, 250 ng/kg biweekly for 104 weeks + 3 week observation
period

%dose levels equivalent to: 1.43, 7.14, 71.4 ng/kg/day (adj)

MAXT
CINT

EXP_TIME_ON
EX P_T I ME_0 F F
DAY CYCLE

= 0. 01
= 0.1
= 0.
= 17472
= 84

idelay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-49 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

B C K_TIME_ON
BCK_TIME_OFF
TIMELIMIT
BW_T0

simulation (g)

= 0.
= 0.
= 17472
= 350

¦;DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
iTIME OF BACKGROUND EXPOSURE STOP (HOUR)
¦;SIMULATION LIMIT TIME (HOUR)
i Body weight at the beginning of the

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT
%MSTOT
MSTOT

= 0.005
= 0.025
0.25

exposure dose ug/kg

C.2.3.2.19. NTP (2006) 14 weeks.

output @clear
prepare @clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG
% NTP 2006

%built and check in August 7 2009
%protocol: oral exposure for 14 weeks; SD rats
%Rat_Dioxin_3C June09_2clean.csl
%RAT_NON_GEST_ICF_F08310 9.CSL (now 09-11-09)

%dose levels: 0.003, 0.010, 0.022, 0.046 0.1 ug/kg 5 days/weeks for 14 weeks
%dose levels equivalent to: 3, 10, 22, 46 100 ng/kg 5 days/weeks for 14 weeks
%dose levels equivalent to: 2.14, 7.14, 15.7, 32.9 71.4 ng/kg 7 days/weeks for

14 weeks







MAXT

= 0.01





CINT

= 0.1





EXP TIME ON

= 0.

%delay before begin exposure (HOUR)

EXP TIME OFF

= 2352

%TIME EXPOSURE STOP (HOUR)



DAY CYCLE

= 24





WEEK PERIOD

= 168





WEEK FINISH

= 119





BCK TIME ON

= 0.

%DELAY BEFORE BACKGROUND EXPOSURE

(HOUR)

BCK TIME OFF

= 0.

%TIME OF BACKGROUND EXPOSURE STOP

(HOUR)

TIMELIMIT

= 2352

%SIMULATION LIMIT TIME (HOUR)



BW TO

= 215

% Body weight at the beginning of

the simulation

(g)







%EXPOSURE DOSE

SCENARIOS (UG/KG)



%MSTOT

= 0.003

% exposure dose ug/kg



%MSTOT

= 0.010

% exposure dose ug/kg



%MSTOT

= 0.022

% exposure dose ug/kg



%MSTOT

= 0.046

% exposure dose ug/kg



MSTOT

II

o

I-1

% exposure dose ug/kg



C.2.3.2.20. NTP (2006) 31 weeks.





output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% NTP 2006

%built and check in August 7 2009
%protocol: oral exposure for 31 weeks; SD rats
%Rat Dioxin 3C June09 2clean.csl
%RAT_NON_GEST_ICF_F083109.CSL (now 09-11-09)

%dose levels: 0.003, 0.010, 0.022, 0.046 0.1 ug/kg 5 days/weeks for 31 weeks
This document is a draft for review purposes only and does not constitute Agency policy.

C-50 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

%dose levels equivalent to: 3, 10, 22, 46 100 ng/kg 5 days/weeks for 31 weeks
%dose levels equivalent to: 2.14, 7.14, 15.7, 32.9 71.4 ng/kg 7 days/weeks
for 31 weeks

MAXT	= 0.01

CINT	=0.1

EXP_TIME_ON	= 0.

EX P_TIME_0 FF	= 52 0 8

DAY_CYCLE	= 2 4

WEEK_PERIOD	= 168

WEEK_FINISH	= 119

B C K_T I ME_ON	= 0.

BCK_TIME_OFF	= 0.

TIMELIMIT	= 5208

BW_T0	=215
simulation (g)

%delay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

%DELAY BEFORE BACKGROUND EXPOSURE	(HOUR)

%TIME OF BACKGROUND EXPOSURE STOP	(HOUR)
i;SIMULATION LIMIT TIME (HOUR)

% Body weight at the beginning of	the

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT	= 0.003	% exposure dose ug/kg

%MSTOT	= 0.010	% exposure dose ug/kg

%MSTOT	= 0.022	% exposure dose ug/kg

%MSTOT	= 0.046	% exposure dose ug/kg

MSTOT	=0.1	% exposure dose ug/kg

C.2.3.2.21. NTP (2006) 53 weeks.

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% NTP 2006

%built and check in August 7 2009
%protocol: oral exposure for 53 weeks; SD rats
%Rat Dioxin 3C June09 2clean.csl
%RAT~NON_GEST_ICF_F083109.CSL (now 09-11-09)

%dose levels: 0.003, 0.010, 0.022, 0.046 0.1 ug/kg 5 days/weeks for 53 weeks
%dose levels equivalent to: 3, 10, 22, 46 100 ng/kg 5 days/weeks for 53 weeks
%dose levels equivalent to: 2.14, 7.14, 15.7, 32.9 71.4 ng/kg 7 days/weeks
for 53 weeks

MAXT	= 0.01

CINT	=0.1

EXP_TIME_ON	= 0.

EX P_T I ME_0 F F	= 8904

DAY_CYCLE	= 2 4

WEEK_PERIOD	= 168

WEEK_FINISH	= 119

B C K_T I ME_ON	= 0.

BCK_TIME_OFF	= 0.

TIMELIMIT	= 8904

BW_T0	=215
simulation (g)

%delay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

%DELAY BEFORE BACKGROUND EXPOSURE	(HOUR)

%TIME OF BACKGROUND EXPOSURE STOP	(HOUR)
i;SIMULATION LIMIT TIME (HOUR)

% Body weight at the beginning of	the

%EXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT	= 0.003	% exposure dose ug/kg

%MSTOT	= 0.010	% exposure dose ug/kg

This document is a draft for review purposes only and does not constitute Agency policy.

C-51 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

%MSTOT
%MSTOT
MSTOT

= 0. 022
= 0. 046
= 0.1

% exposure dose ug/kg

% exposure dose ug/kg
% exposure dose ug/kg

C.2.3.2.22. NTP (2006) 2 year.

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% NTP 2006

%built and check in August 7 2009

%protocol: oral exposure for 105 weeks; SD rats

%dose levels: 0.003, 0.010, 0.022, 0.046, 0.1 ug/kg 5 days/week for 105
weeks

%dose levels equivalent to: 3, 10, 22, 46, 100 ng/kg 5 days/week for 105
weeks

%dose levels equivalent to: 2.14, 7.14, 15.7, 32.9, 71.4 ng/kg 7 days/week
for 105 weeks

MAXT	=	0.01

CINT	=0.1

EXP_TIME_ON	=	0.

EX P_TIME_0 FF	=	17 64 0

DAY_CYCLE	=	2 4

WEEK_PERIOD	=	168

WEEK_FINISH	=	119

B C K_T I ME_ON	=	0.
(HOUR)

BCK_TIME_OFF	=	0.

TIMELIMIT	=	17640

BW_T0	=215
SIMULATION (G)

iTIME AT WHICH EXPOSURE BEGINS (HOUR)
iTIME AT WHICH EXPOSURE ENDS (HOUR)

iTIME AT WHICH BACKGROUND EXPOSURE BEGINS

iTIME AT WHICH BACKGROUND EXPOSURE ENDS (HOUR)
iSIMULATION TIME LIMIT (HOUR)
i BODY WEIGHT AT THE BEGINNING OF THE

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT
%MSTOT
%MSTOT
%MSTOT
MSTOT

= 0.003
= 0.010
= 0.022
= 0. 046
0.1

EXPOSURE DOSE IN UG/KG

% EXPOSURE DOSE IN UG/KG
% EXPOSURE DOSE IN UG/KG

% EXPOSURE DOSE IN UG/KG
% EXPOSURE DOSE IN UG/KG

C.2.3.2.23. Sewall et al. (1995).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% Sewall et al. 1995

%Rat Dioxin 3C June09 2clean.csl

%RAT~NON_GEST_ICF_F083109.CSL (now 09-11-09)

%protocol: gavage every 2 weeks for 30 weeks

%dose levels: 0.049, 0.1498, 0.49, and 1.75 ug/kg every 2 weeks

%dose levels: 3.5, 10.7, 35, and 125 ng/kg/d or 49, 149.8, 490, and 1750

ng/kg every 2 weeks

MAXT	= 0.01

CINT	= 0.1

This document is a draft for review purposes only and does not constitute Agency policy.

C-52 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

EXP_TIME_ON	= 0.

after start of experiment

EX P_T I ME_0 F F

every two weeks

DAY_CYCLE

B C K_TIME_ON

BCK_TIME_OFF

TIMELIMIT

BW TO

=	5040

=	336.

=	0.

=	0.

=	5040

=	250

%delay before begin exposure (HOUR) 5 weeks
(age = 12 weeks)

%TIME EXPOSURE STOP (HOUR); 30 doses, 1

% once every two weeks
%DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
%TIME OF BACKGROUND EXPOSURE STOP (HOUR)
%SIMULATION LIMIT TIME (HOUR)

% Body weight at the beginning of the

simulation (g); corresponds to 12 week old female

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT
%MSTOT
%MSTOT
MSTOT

= 0.049
= 0. 1498
= 0.49
= 1.75

% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
i ORAL EXPOSURE DOSE (UG/KG)
ORAL EXPOSURE DOSE (UG/KG)

C.2.3.2.24. Shi et al. (2007), adult portion.

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% Shi et al 2007

%built and check in August 7 2009
%protocol: gavage once per week for 322 days

%dose levels: 0.001, 0.005, 0.05 and 0.2 ug TCDD:kg body weight by gavage
once per week

%dose levels: 1, 5, 50 and 200 ng/kg ng TCDD:kg body weight by gavage once
per week

% dose equivalent adjusted 0.143, 0.714, 7.14 and 28.6 ng/kg/d

MAXT
CINT

EXP_TIME_ON
EX P_T I ME_0 F F
CORRESPONDS TO
DAY_CYCLE
B C K_TIME_ON
BEGINS (HOUR)
BCK_TIME_OFF
(HOUR)
TIMELIMIT
BW_T0

SIMULATION (G)

= 0.0001
= 0.1
= 504 .
= 7728
DAYS
168 .

0.

322

= 0.

% TIME AT WHICH EXPOSURE BEGINS (HOUR)
iTIME AT WHICH EXPOSURE ENDS (HOUR);

OF EXPOSURE

TIME AT WHICH BACKGROUND EXPOSURE

TIME AT WHICH BACKGROUND EXPOSURE ENDS

7728
4 . 5

i;SIMULATION TIME
% BODY WEIGHT AT

LIMIT (HOUR)
THE BEGINNING OF THE

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT
%MSTOT
%MSTOT
MSTOT

= 0.001	% ORAL EXPOSURE DOSE IN UG/KG

= 0.005	% ORAL EXPOSURE DOSE IN UG/KG

=0.05	% ORAL EXPOSURE DOSE IN UG/KG

=0.2	% ORAL EXPOSURE DOSE IN UG/KG

C.2.3.2.25. Van Birgelen et al (1995).

output 0clear
prepare 0clear

This document is a draft for review purposes only and does not constitute Agency policy.

C-53 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% Van Birgelen et al. (1995)

%protocol: daily dietary exp
%dose levels: 0.0135, 0.0264,
weeks

% dose levels = 13.5, 26.4, 46.9, 320, 1024 ng/kg every day for 13 weeks
MAXT	= 0.01

CINT	= 0.1

EXP_TIME_ON	= 0.

EX P_TIME_0 FF = 2184.

DAY_CYCLE	= 24.

B C K_T I ME_ON	= 0.

BCK_TIME_OFF = 0.

TIMELIMIT	= 2184.

BW_T0	= 150.

simulation (g)

osure for 13 weeks
0.0469, 0.320, 1.024 ug/kg every day for 13

idelay before begin exposure (HOUR)

%TIME EXPOSURE STOP (HOUR)

% once every two weeks

%DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
%TIME OF BACKGROUND EXPOSURE STOP (HOUR)
%SIMULATION LIMIT TIME (HOUR)

% Body weight at the beginning of the

iEXPOSURE
%MSTOT
%MSTOT
%MSTOT
%MSTOT
MSTOT

DOSE SCENARIOS (UG/KG)
= 0.0135
= 0.0264
= 0. 0469
= 0.320
= 1. 024

ORAL EXPOSURE DOSE (UG/KG)
i ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
i ORAL EXPOSURE DOSE (UG/KG)
ORAL EXPOSURE DOSE (UG/KG)

C.2.3.2.26. Vanden Heuvel et al. (1994).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% Vanden Heuvel et al. 1994.

%built and check in August 7 2009

%protocol: single gavage

%Rat Dioxin 3C June09 2clean.csl

%RAT_NON_GEST_ICF_F083109.CSL (now 09-11-09)

%dose levels:0.00005, 0.0001, 0.001, 0.010, 0.1, 1, 10 ug/kg/d

%dose levels equivalent to: 0.05, 0.1, 1, 10, 100, 1000, 10000 ng/kg/d

MAXT
CINT

EXP_TIME_ON

EX P_T I ME_0 F F

DAY_CYCLE

B C K_TIME_ON

BCK_TIME_OFF

TIMELIMIT

BW_T0

simulation (g)

0. 001

0.1

0.

24

24

0.

0.

24

250

%delay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

%DELAY BEFORE BACKGROUND EXPOSURE	(HOUR)

%TIME OF BACKGROUND EXPOSURE STOP	(HOUR)
%SIMULATION LIMIT TIME (HOUR)

% Body weight at the beginning of	the

iEXPOSURE DOSE SCENARIOS (UG/KG)

iMSTOT	= 0.00005	% exposure dose ug/kg

iMSTOT	= 0.0001	%	exposure dose ug/kg

iMSTOT	= 0.001	%	exposure dose ug/kg

iMSTOT	=0.01	%	exposure dose ug/kg

This document is a draft for review purposes only and does not constitute Agency policy.

C-54 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

%MSTOT
%MSTOT
MSTOT

= 0.1
= 1
= 10

% exposure dose ug/kg
% exposure dose ug/kg
% exposure dose ug/kg

C.2.4. Rat Gestational Model
C.2.4.1. Model Code

PROGRAM: 'Three Compartment PBPK Model for TCDD in Rat (Gestation)'

! Parameters were change May 16, 2002
! Come from {8MAI_CHR_PRE-EXP_GD}

! Come from {12 Mouse GDJfile

!{{IMPORTANT-IMPORTANT-IMPORTANT-IMPORTANT}}

! REDUCTION OF MOTHER AND FETUS COMPARTMENT
! 2M_R_TCDD_JULY2 0 02 ////(JULY 18,20 02)////
!TCDD_RED_4Species_2003_4	////(APR 8 ,2003)////

!TCDD_RED_4Species_2003_9	////(APR 17 ,2003)////

!TCDD_RED_4Species_2003_12	////(APR 17 ,2003)////

!APRIL 18 2003

!TCDD_4C_4SP_2 0 03	////(APR 18 ,2003)////

! was ''Gest 4 species l.csl'' but update July 2009

!DevTCDD4Species ICF afterKKfix v3 ratgest.csl
!RAT_GESTATIONAL_ICF_F083109.csl
!RAT_GESTATIONAL_ICF_F10 0 60 9.csl

!Legend/Legend/Legend/Legend/Legend/Legend/Legend/Legend/

!Legend for this PBPK model

IMating: control the tenure of exchange between fetus and
IMother and also control imitated tissue growth
!Control: WTFE, WFO, WPLA0, QPLAF,WTO
!(for rat, mouse, human, and monkey)

!Control transfer from mother to fetus or fetus to mother by TRANSTIME ON
!SWITCH_trans = 0 NO TRANSFER	~

!SWITCH_trans = 1 TRANSFER OCCURS
!Gest off = 1
!Gest on= 0 . 0
! These switches are also controlled by mating parameters

INITIAL !

!SIMULATION PARAMETERS ====
CONSTANT PARA_ZERO	= 1E-30

CONSTANT EXP_TIME_ON	=0.0

CONSTANT EXP_TIME_0FF	= 530

CONSTANT DAY_CYCLE	= 2 4.0

CONSTANT BCK_TIME_ON	=0.0
BEGINS (HOURS)

CONSTANT BCK_TIME_OFF	=0.0
(HOURS)

CONSTANT TRANSTIME_ON	= 144.0
AT GESTATIONAL DAY 6

TIME AT WHICH EXPOSURE BEGINS (HOURS)
TIME AT WHICH EXPOSURE ENDS (HOURS)
NUMBER OF HOURS BETWEEN DOSES (HOURS)
TIME AT WHICH BACKGROUND EXPOSURE

! TIME AT WHICH BACKGROUND EXPOSURE ENDS

!CONTROL TRANSFER FROM MOTHER TO FETUS

This document is a draft for review purposes only and does not constitute Agency policy.

C-55 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!UNIT CONVERSION
CONSTANT MW=322 ! MOLECULAR WEIGHT (NG/NMOL)
CONSTANT SERBLO =0.55
CONSTANT UNITCORR = 1000

!INTRAVENOUS SEQUENCE
constant IV_LACK	=0.0

constant IV PERIOD	=0.0

!PREGNANCY PARAMETER ====

CONSTANT MATTING	=0.0	!BEGINNING OF MATING (HOUR)

CONSTANT N_FETUS	= 10.0	!NUMBER OF FETUS PRESENT

!CONSTANT EXPOSURE CONTROL ===========

!ACUTE, SUBCHRONIC, CHRONIC EXPOSURE =====

!OR BACKGROUND EXPOSURE (IN THIS CASE 3 TIMES A DAY)===

CONSTANT MSTOTBCKGR	=0.0	! ORAL BACKGROUND EXPOSURE DOSE (UG/KG)

CONSTANT MSTOT	=0.0	! ORAL EXPOSURE DOSE (UG/KG)

!ORAL ABSORPTION
MSTOT NM = MSTOT/MW

! CONVERTS THE DOSE TO NMOL/G

!INTRAVENOUS ABSORPTION
CONSTANT DOSEIV	=0.0

DOSEIV_NM = DOSEIV/MW
CONSTANT DOSEIVLATE =0.0

DOSEIVNMlate = DOSEIVLATE/MW

! INJECTED DOSE (UG/KG)

! CONVERTS THE INJECTED DOSE TO NMOL/G

! INJECTED DOSE LATE (UG/KG)

!AMOUNT IN NMOL/G

!INITIAL GUESS OF	THE FREE CONCENTRATION IN THE LIGAND (COMPARTMENT
INDICATED BELOW)====

CONSTANT CFLLI0	=0.0 !LIVER (NMOL/ML)

CONSTANT CFLPLA0	=0.0 !PLACENTA (NMOL/ML)

!BINDING CAPACITY	(AhR) FOR NON LINEAR BINDING (COMPARTMENT INDICATED
BELOW) (NMOL/ML) ===

CONSTANT LIBMAX	= 3.5E-4 ! LIVER (NMOL/ML), WANG ET AL. 1997

CONSTANT PLABMAX	= 2.OE-4 !TEMPORARY PARAMETER

! PROTEIN AFFINITY	CONSTANTS (1A2 OR AhR, COMPARTMENT INDICATED BELOW)
(NMOL/ML)===

CONSTANT KDLI	= 1.0E-4 !LIVER (AhR) (NMOL/ML), WANG ET AL. 1997

CONSTANT KDLI2	= 4.OE-2 !LIVER (1A2) (NMOL/ML), EMOND ET AL. 2004

CONSTANT KDPLA	= 1.0E-4 !TEMPORARY PARAMETER; ASSUME IDENTICAL TO
KDLI (AhR)

!EXCRETION AND ABSORPTION CONSTANT
CONSTANT KST	= 0.36	! GASTRIC RATE CONSTANT (HR-1), WANG ET

AL. 1997

CONSTANT KABS	= 0.48	!INTESTINAL ABSORPTION CONSTANT (HR-1) ),

WANG ET AL. 1997

! ELIMINATION CONSTANTS
CONSTANT CLURI	=0.01	! URINARY CLEARANCE (ML/HR), EMOND ET

AL. 2004

This document is a draft for review purposes only and does not constitute Agency policy.

C-56 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

! INTERSPECIES
CONSTANT kelv
CONSTANT (1/HOUR)

ELIMINATION VARIABLE

=0.15	! INTERSPECIES VARIABLE ELIMINATION

! CONSTANT TO DIVIDE THE ABSORPTION INTO LYMPHATIC AND PORTAL FRACTIONS
CONSTANT A	=0.7	! LYMPHATIC FRACTION, WANG ET AL. 1997

!PARTITION COEFFICIENTS

CONSTANT PF	= 100

CONSTANT PRE	=1.5
1997

CONSTANT PLI	=6.0

CONSTANT PPLA	=1.5
WANG ET AL. 1997

! ADIPOSE TISSUE/BLOOD, WANG ET AL. 1997
! REST OF THE BODY/BLOOD, WANG ET AL.

! LIVER/BLOOD, WANG ET AL. 19 97
! TEMPORARY PARAMETER NOT CONFIGURED,

!PARAMETER FOR INDUCTION OF CYP

CONSTANT PAS_INDUC	=1.0

CONSTANT CYP1A2_10UTZ	=1.6
1A2 (NMOL/ML)

CONSTANT CYP1A2_1A1	=1.6

CONSTANT CYP1A2_1EC5 0	=0.13
(NMOL/ML)

CONSTANT CYP1A2_1A2	=1.6

CONSTANT CYP1A2_1K0UT	=0.1

CONSTANT CYP1A2_1TAU	= 0.25

CONSTANT CYP1A2_1EMAX	= 600
(UNITLESS)

CONSTANT HILL	=0.6

BINDING EFFECT CONSTANT	(UNITLESS)

1A2, WANG ET AL. 1997

! INCLUDE INDUCTION? (1 = YES, 0 = NO)
! DEGRADATION CONCENTRATION CONSTANT OF

! BASAL CONCENTRATION OF 1A1 (NMOL/ML)
! DISSOCIATION CONSTANT TCDD-CYP1A2

BASAL CONCENTRATION OF 1A2 (NMOL/ML)

FIRST ORDER RATE OF DEGRADATION (H-l)
HOLDING TIME (H)

MAXIMUM INDUCTION OVER BASAL EFFECT

!HILL CONSTANT; COOPERATIVELY LIGAND

!DIFFUSIONAL PERMEABILITY FRACTION

CONSTANT PAFF
CONSTANT PAREF
AL. 1997
CONSTANT PALIF
CONSTANT PAPLAF

0.0910	!ADIPOSE (UNITLESS), WANG ET AL. 1997

0.0298	!REST OF THE BODY (UNITLESS), WANG ET

0.3500	!LIVER (UNITLESS), WANG ET AL. 1997

0.3	!TEMPORARY PARAMETER NOT CONFIGURED

!FRACTION OF TISSUE WEIGHT =========

CONSTANT WLI0	= 0.0360 !LIVER, WANG ET AL. 1997

!TISSUE BLOOD FLOW EXPRESSED AS A FRACTION OF CARDIAC OUTPUT
CONSTANT QFF	= 0.069 ! ADIPOSE TISSUE BLOOD FLOW FRACTION

(UNITLESS), WANG ET AL. 1997

CONSTANT QLIF	= 0.183 !LIVER (UNITLESS), WANG ET AL. 1997

!COMPARTMENT TISSUE BLOOD EXPRESSED AS A FRACTION OF THE TOTAL COMPARTMENT

VOLUME

CONSTANT WFB0
CONSTANT WREB0
CONSTANT WLIB0
CONSTANT WPLAB0

0.050	!ADIPOSE TISSUE, WANG ET AL. 1997

0.030	!REST OF THE BODY, WANG ET AL. 1997

0.266	!LIVER, WANG ET AL. 1997

0.500	!TEMPORARY PARAMETER NOT CONFIGURED

!EXPOSURE SCENARIO FOR UNIQUE OR REPETITIVE WEEKLY OR MONTHLY EXPOSURE
!NUMBER OF EXPOSURES PER WEEK
CONSTANT WEEK_LACK	=0.0	!DELAY BEFORE EXPOSURE ENDS (WEEK)

This document is a draft for review purposes only and does not constitute Agency policy.

C-57 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CONSTANT WEEK_PERIOD = 168	! NUMBER OF HOURS IN THE WEEK (HOURS)

CONSTANT WEEK_FINISH = 168	! TIME EXPOSURE ENDS (HOURS)

!NUMBER OF EXPOSURES PER MONTH

CONSTANT MONTH_LACK =0.0	!DELAY BEFORE EXPOSURE BEGINS (MONTHS)

!CONSTANT FOR BACKGROUND EXPOSURE===========

CONSTANT Day_LACK_BG =0.0	!DELAY BEFORE EXPOSURE BEGINS (HOURS)

CONSTANT Day_PERIOD_BG =24	!LENGTH OF EXPOSURE (HOURS)

!NUMBER OF EXPOSURES PER WEEK
CONSTANT WEEK_LACK_BG	=0.0	!DELAY BEFORE BACKGROUD EXPOSURE BEGINS

(WEEKS)

CONSTANT WEEK_PERIOD_BG = 168	!NUMBER OF HOURS IN THE WEEK (HOURS)

CONSTANT WEEK FINISH BG = 168	!TIME EXPOSURE ENDS (HOURS)

!INITIAL BODY WEIGHT
CONSTANT BW_T0	= 250

CONSTANT RAT10_RATF_MOUSEF =1.0
GESTATIONAL DAY 22

! WANG ET AL. 1997

!RATIO OF FETUS MOUSE/RAT AT

! COMPARTMENT LIPID EXPRESSED AS THE FRACTION OF TOTAL LIPID, POULIN ET AL
2000

CONSTANT F_TOTLIP	= 0.855

CONSTANT B_TOTLIP	= 0.0023

CONSTANT RE_TOTLIP	= 0.019
(UNITLESS)

CONSTANT LI_TOTLIP	= 0.060

CONSTANT PLA_TOTLIP	= 0.019

CONSTANT FETUS TOTLIP	= 0.019

! ADIPOSE TISSUE (UNITLESS)

! BLOOD (UNITLESS)

! REST OF THE BODY

! LIVER (UNITLESS)

END

! END OF THE INITIAL SECTION

DYNAMIC ! DYNAMIC SIMULATION SECTION

ALGORITHM IALG	=	2

CINTERVAL CINT	=	0.1

MAXTERVAL MAXT	=	1.0e+10

MINTERVAL MINT	=	1.0E-10

VARIABLE T	=	0.0

CONSTANT TIMELIMIT	=	100
CINTXY = CINT

GEAR METHOD

COMMUNICATION INTERVAL
MAXIMUM CALCULATION INTERVAL
MINIMUM CALCULATION INTERVAL

!SIMULATION LIMIT TIME (HOURS)

PFUNC

= CINT

!TIME CONVERSION
DAY	= T/24

WEEK	= T/168

MONTH	= T/730

YEAR	= T/87 60

TIME IN DAYS
TIME IN WEEKS
TIME IN MONTHS
TIME IN YEARS

DERIVATIVE ! PORTION OF CODE THAT SOLVES DIFFERENTIAL EQUATIONS

!CHRONIC OR SUBCHRONIC EXPOSURE SCENARIO =======

!NUMBER OF EXPOSURES PER DAY
DAY_LACK	= EXP_TIME_ON ! DELAY BEFORE EXPOSURE BEGINS (HOURS)

DAY_PERIOD	= DAY_CYCLE	! EXPOSURE PERIOD (HOURS)

DAY FINISH	= CINTXY	! LENGTH OF EXPOSURE (HOURS)

This document is a draft for review purposes only and does not constitute Agency policy.

C-58 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

MONTH_PERIOD
MONTH FINISH

TIMELIMIT
EXP TIME OFF

! EXPOSURE PERIOD (MONTHS)
! LENGTH OF EXPOSURE (MONTHS)

!NUMBER OF EXPOSURES PER DAY AND MONTH

DAY_FINISH_BG
MONTH_LACK_BG
(MONTHS)

MONTH PERIOD BG

TIMELIMIT

CINTXY
BCK TIME ON

!DELAY BEFORE BACKGROUD EXPOSURE BEGINS

!BACKGROUND EXPOSURE (MONTHS)

MONTH_FINISH_BG = BCK_TIME_OFF !LENGTH OF BACKGROUND EXPOSURE (MONTHS)

!INTRAVENOUS LATE
IV_FINISH = CINTXY

B = 1-A ! FRACTION OF DIOXIN ABSORBED IN THE PORTAL FRACTION OF THE LIVER

!FETUS,VOLUME,FETUS,VOLUME,FETUS,VOLUME,FETUS,VOLUME,FETUS,VOLUME,FETUS,VOLUM
E

! FROM OFLAHERTY_19 92

RTESTGEST= T-MATTING
TESTGEST=DIM(RTESTGEST,0.0)

WTFER_RODENT= (2.3d-3*EXP(1.49d-2*(TESTGEST))+1.3d-2)*Gest_on
WTFER = (WTFER_RODENT*RATIO_RATF_MOUSEF*N_FETUS)

WTFE = DIM(WTFER,0.0)

FAT,VOLUME,FAT,VOLUME,FAT,VOLUME,FAT,VOLUME,FAT,VOLUME,FAT,VOLUME,FAT,VOLUME
! FAT GROWTH EXPRESSION LINEAR DURING PREGNANCY
! FROM O'FLAHERTY_19 92

WF0= (((9.66d-5*(TESTGEST))*gest_on)+0.069)

! PLACENTA,VOLUME, PLACENTA,VOLUME, PLACENTA,VOLUME, PLACENTA,VOLUME
! WPLA PLACENTA GROWTH EXPRESSION, SINGLE EXPONENTIAL WITH OFFSET
! FROM O'FLAHERTY_19 92 ! FOR EACH PUP

WPLA0N_RODENT = (0.6/(1+(5d+3*EXP(-0.0225*(TESTGEST)))))*N_FETUS
WPLA0R = (WPLA0N_RODENT/WT0)*Gest_on
WPLA0 = DIM(WPLA0R,0.0)

! PLACENTA,FLOW RATE, PLACENTA,FLOW RATE, PLACENTA,FLOW RATE, PLACENTA,FLOW
RATE

! QPLA PLACENTA GROWTH EXPRESSION, DOUBLE EXPONENTIAL WITH OFFSET
! FROM O'FLAHERTY_19 92

QPLARF = (1.67d-7 *exp(9.6d-3*(TESTGEST)) &

+1.6d-3*exp(7.9d-3*(TESTGEST))+0.0)*Gest_on*SWITCH_trans
QPLAF=DIM(QPLARF,0.0)	!FRACTION OF FLOW RATE IN PLACENTA

! GESTATION CONTROL
IF (T.LT.MATTING) THEN
Gest off = 1.0
Gest on= 0 . 0

ELSE

Gest off = 0.0

This document is a draft for review purposes only and does not constitute Agency policy.

C-59 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

Gest on =1.0
END IF

MOTHER BODY WEIGHT GROWTH EQUATION========

MODIFICATION TO ADAPT THIS MODEL AT HUMAN MODEL

BECAUSE LINEAR DESCRIPTION IS NOT GOOD ENOUGH FOR MOTHER GROWTH
MOTHER BODY WEIGHT GROWTH

PARAMETER (BW_RMN = 1.0E-30)

WT0= BW_T0 *(1+(0.41*T)/(1402.5+T+BW_RMN))

! VARIABILITY OF REST OF THE BODY DEPENDS ON OTHER ORGANS
WRE0 = (0.91 - (WLIB0*WLI0 + WFB0*WF0 +WPLAB0*WPLA0 + WLI0 + WF0 +
WPLA0))/(1+WREB0) ! REST OF THE BODY FRACTION; UPDATED FOR EPA ASSESSMENT
QREF = 1-(QFF+QLIF+QPLAF)	!REST OF BODY BLOOD FLOW RATE (ML/HR)

QTTQF = QFF+QREF+QLIF+QPLAF	! SUM MUST EQUAL 1

! COMPARTMENT VOLUME (ML OR G)
WF = WF0 * WTO
WRE = WRE0 * WTO
WLI = WLI0 * WTO
WPLA= WPLA0* WTO

ADIPOSE TISSUE
REST OF THE BODY
LIVER
PLACENTA

! COMPARTMENT TISSUE BLOOD	(ML OR G)

WFB = WFB0 * WF	!

WREB = WREB0 * WRE	!

WLIB = WLIB0 * WLI	!

WPLAB = WPLAB0* WPLA	!

ADIPOSE TISSUE
REST OF THE BODY
LIVER
PLACANTA

! CARDIAC OUTPUT FOR THE GIVEN BODY WEIGHT (ML/H) =========

!QC= QCCAR* 60*(WTO/1000.0)**0.75
CONSTANT QCC=18684.0	! EQUIVALENT TO 311.4 * 60

QC= QCC*(WTO/UNITCORR)**0.75

!COMPARTMENT BLOOD FLOW RATE (ML/HR)

QF = QFF*QC	!ADIPOSE TISSUE BLOOD FLOW RATE

QLI = QLIF*QC	!LIVER TISSUE BLOOD FLOW RATE

QRE = QREF*QC	!REST OF THE BODY BLOOD FLOW RATE

QPLA = QPLAF*QC	!PLACENTA TISSUE BLOOD FLOW RATE

QTTQ = QF+QRE+QLI+QPLA !TOTAL FLOW RATE

!PERMEABILITY ORGAN FLOW (ML/HR)
PAF = PAFF*QF
PARE = PAREF*QRE
PALI = PALIF*QLI
PAPLA = PAPLAF*QPLA

ADIPOSE TISSUE
REST OF THE BODY
LIVER TISSUE
PLACENTA

ABSORPTION SECTION
ORAL

INTRAPERITONEAL
INTRAVENOUS

!REPETITIVE ORAL BACKGROUND EXPOSURE SCENARIO

This document is a draft for review purposes only and does not constitute Agency policy.

C-60 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

MSTOT_NMBCKGR = MSTOTBCKGR/MW	! CONVERTS THE BACKGROUND DOSE TO NMOL/G

MSTTBCKGR =MSTOT_NMBCKGR *WT0

DAY_EX P O S U RE_B G = PULSE(DAY_LACK_BG,DAY_PERIOD_BG,DAY_FINISH_BG)
WEEK_EXPOSURE_BG = PULSE(WEEK_LACK_BG,WEEK_PERIOD_BG,WEEK_FINISH_BG)
MONTH_EXPOSURE_BG = PULSE(MONTH_LACK_BG,MONTH_PERIOD_BG,MONTH_FINISH_BG)

MSTTCH_BG = (DAY_EXPOSURE_BG*WEEK_EXPOSURE_BG*MONTH_EXPOSURE_BG)*MSTTBCKGR
MSTTFR_BG = MSTTBCKGR/CINT

CYCLE_BG =DAY_EXPOSURE_BG*WEEK_EXPOSURE_BG*MONTH_EXPOSURE_BG

! CONDITIONAL ORAL EXPOSURE (BACKGROUND EXPOSURE)

IF (MSTTCH_BG.EQ.MSTTBCKGR) THEN
ABSMSTT_GB= MSTTFR_BG

ELSE

ABSMSTT_GB =0.0
END IF

CYCLETOTBG=INTEG(CYCLE_BG,0.0)

!REPETITIVE ORAL EXPOSURE SCENARIO

MSTT= MSTOT_NM * WTO	!AMOUNT IN NMOL

DAY_EXPOSURE = PULSE(DAY_LACK,DAY_PERIOD,DAY_FINISH)

WEEK_EXPOSURE = PULSE(WEEK_LACK,WEEK_PERIOD,WEEK_FINISH)

MONTH_EXPOSURE = PULSE(MONTH_LACK,MONTH_PERIOD,MONTH_FINISH)

MSTTCH = (DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE)*MSTT
MSTTFR = MSTT/CINT

CYCLE = DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE

SUMEXPEVENT= INTEG (CYCLE,0.0) !NUMBER OF CYCLE GENERATE DURING SIMULATION

! CONDITIONAL ORAL EXPOSURE
IF (MSTTCH.EQ.MSTT) THEN

ABSMSTT= MSTTFR
ELSE

ABSMSTT =0.0
END IF

CYCLETOT=INTEG(CYCLE, 0.0)

! MASS CHANGE IN THE LUMEN
RMSTT= -(KST+KABS)*MST +ABSMSTT +ABSMSTT_GB ! RATE OF CHANGE (NMOL/H)

MST = INTEG(RMSTT,0.0)	!AMOUNT REMAINING IN DUODENUM

(NMOL)

! ABSORPTION IN LYMPH CIRCULATION
LYRMLUM = KABS*MST*A
LYMLUM = INTEG(LYRMLUM,0.0)

! ABSORPTION IN PORTAL CIRCULATION

This document is a draft for review purposes only and does not constitute Agency policy.

C-61 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

LIRMLUM = KABS*MST*B
LIMLUM = INTEG(LIRMLUM,0.0)

! 	IV EXPOSURE 	

IV= DOSEIV_NM * WTO !AMOUNT IN NMOL
IVR= IV/PFUNC ! RATE FOR IV INFUSION IN BLOOD
EXPIV= IVR * (1.0-STEP(PFUNC))

IVDOSE = integ(EXPIV,0.0)

!	IV LATE IN THE CYCLE

! MODIFICATION ON January 13 2004
IV_RlateR = DOSEIVNMlate*WT0

IV_EXPOSURE=PULSE(IV_LACK,IV_PERIOD,IV_FINISH)

IV_lateT = IV_EXPOSURE *IV_RlateR
IV_late = IV_lateT/CINT

SUMEXPEVENTIV= integ (IV_EXPOSURE,0.0) !NUMBER OF CYCLE GENERATE DURING
SIMULATION

!SYSTEMIC CONCENTRATION OF TCDD

! MODIFICATION ON OCTOBER 6, 2009
CB= (QF*CFB+QRE*CREB+QLI*CLIB+EXPIV+LYRMLUM+QPLA*CPLAB+IV_late)/(QC+CLURI) !
CA = CB ! CONCENTRATION (NMOL/ML)

!URINARY EXCRETION BY KIDNEY
! MODIFICATION ON OCTOBER 6, 2009
RAURI = CLURI *CB

AURI = INTEG(RAURI,0.0)

!UNIT CONVERSION POST SIMULATION
CBSNGKGLIADJ=(CB*MW*UNITCORR*(1.0/B_TOTLIP)*(1.0/SERBLO))![NG of TCDD
Serum/Kg OF LIPIP]

AUCBS_NGKGLIADJ=integ(CBSNGKGLIADJ,0.0)

PRCT_B = (CB/(MSTT+1E-30))*100.0 !PERCENT OF ORAL DOSE IN BLOOD
PRCT_BIV = (CB/(IV_RlateR+lE-3 0))*100.0 ! PERCENT OF IV DOSE IN BLOOD
CBNGKG= CB*MW*UNITCORR

!ADIPOSE COMPARTMENT
! TISSUE BLOOD COMPARTMENT
RAFB= QF*(CA-CFB)-PAF*(CFB-CF/PF)
AFB = INTEG(RAFB,0.0)
CFB = AFB/WFB
! TISSUE COMPARTMENT
RAF = PAF*(CFB-CF/PF)
AF = INTEG(RAF,0.0)

CF = AF/WF

!(NMOL/H)
!(NMOL)

(NMOL/ML)

!(NMOL/H)
!(NMOL)
!(NM/ML)

This document is a draft for review purposes only and does not constitute Agency policy.

C-62 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!UNIT CONVERSION POST SIMULATION
CFTOTAL= (AF + AFB)/(WF + WFB) ! TOTAL CONCENTRATION IN NMOL/ML
CFTFREE = CFB + CF !TOTAL FREE CONCENTRATION IN FAT (NM/ML)

PRCT_F = (CFTOTAL/(MSTT+1E-30))*100.0 ! PERCENT OF ORAL DOSE IN FAT
PRCT_FIV = (CFTOTAL/(IV_RlateR+lE-30))*100.0 ! PERCENT OF IV DOSE IN FAT
CFNGKG=CFTOTAL*MW*UNITCORR ! FAT CONCENTRATION NG/KG
AUCF_NGKGH=integ(CFNGKG, 0.0)

!REST OF THE BODY COMPARTMENT
RAREB= QRE *(CA-CREB)-PARE*(CREB-CRE/PRE) !(NMOL/H)

AREB = INTEG(RAREB,0.0)	!(NMOL)

CREB = AREB/WREB	!(NMOL/H)
!TISSUE COMPARTMENT

RARE = PARE*(CREB - CRE/PRE)	!(NMOL/H)

ARE = INTEG(RARE,0.0)	!(NMOL)

CRE = ARE/WRE	!(NMOL/ML)

!UNIT CONVERSION POST SIMULATION
CRETOTAL= (ARE + AREB)/(WRE + WREB)	! TOTAL CONCENTRATION IN

NMOL/ML

PRCT_RE = (CRETOTAL/(MSTT+1E-30))*100.0 ! PERCENT OF ORAL DOSE IN REST OF
THE BODY

PRCT_REIV = (CRETOTAL/(IV_RlateR+lE-30))*100.0 !PERCENT OF IV DOSE IN
REST OF THE BODY

CRENGKG=CRETOTAL*MW*UNITCORR ! REST OF THE BODY CONCENTRATION IN NG/KG

!LIVER COMPARTMENT
!TISSUE BLOOD COMPARTMENT
RALIB = QLI*(CA-CLIB)-PALI*(CLIB-CFLLIR)+LIRMLUM !

ALIB = INTEG(RALIB,0.0)	!(NMOL)

CLIB = ALIB/WLIB	!(NMOL/ML)

!TISSUE COMPARTMENT
RALI = PALI*(CLIB - CFLLIR)-REXCLI	! (NMOL/HR)

ALI = INTEG(RALI,0.0)	!(NMOL)

CLI = ALI/WLI	!(NMOL/ML)

!FREE TCDD CONCENTRATION IN LIVER COMPARTMENT
PARAMETER (LIVER_1RMN = 1.0E-30)

CFLLI= IMPLC(CLI-(CFLLIR*PLI+(LIBMAX*CFLLIR/(KDLI+CFLLIR &
+LIVER_1RMN))+((CYP1A2_103*CFLLIR/(KDLI2 + CFLLIR &
+LIVER_1RMN)*PAS_INDUC)))-CFLLI,CFLLI0)

CFLLIR=DIM(CFLLI,0.0) ! FREE CONCENTRATION IN LIVER

CBNDLI= LIBMAX*CFLLIR/(KDLI+CFLLIR+LIVER_1RMN) !BOUND CONCENTRATION

!VARIABLE ELIMINATION BASED ON THE CYP1A2
KBILE_LI_T =((CYP1A2_10UT-CYP1A2_1A2)/CYP1A2_1A2)*Kelv ! INDUCED BILIARY
EXCRETION RATE CONSTANT IN LIVER

REXCLI = KBILE_LI_T*CFLLIR*WLI ! DOSE-DEPENDENT BILIARY EXCRETION RATE
EXCLI = INTEG(REXCLI,0.0)

!UNIT CONVERSION POST SIMULATION

CLITOTAL= (ALI + ALIB)/(WLI + WLIB) ! TOTAL CONCENTRATION IN NMOL/ML

PRCT_LI = (CLITOTAL/(MSTT+1E-30))*100

PRCT_LIIV = (CLITOTAL/(IV_RlateR+lE-30))*100.0

This document is a draft for review purposes only and does not constitute Agency policy.

C-63 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

Rec_occ= CFLLIR/(KDLI+CFLLIR)

CLINGKG=CLITOTAL*MW*UNITCORR ! LIVER CONCENTRATION NG/KG

AUCLI_NGKGH=INTEG(CLINGKG, 0.0)

CBNDLINGKG = CBNDLI*MW*UNITCORR

AUCBNDLI NGKGH =INTEG(CBNDLINGKG,0.0)

!CHEMICAL IN CYP450 (1A2) COMPARTMENT
CYP1A2 1KINP = CYP1A2 1KOUT* CYP1A2 10UTZ

! MODIFICATION ON OCTOBER 6, 2009
CYP1A2_10UT =INTEG(CYP1A2_1KINP * (1.0 + CYP1A2_1EMAX *(CBNDLI+1.0e-30)**HILL
&

/(CYP1A2_1EC50**HILL + (CBNDLI+1.0e-30)**HILL)) &
- CYP1A2_1K0UT*CYP1A2_10UT, CYP1A2_10UTZ)

! EQUATIONS INCORPORATING DELAY OF CYP1A2 PRODUCTION (NOT USED IN
SIMULATIONS)

CYP1A2_1R02 = (CYP1A2_10UT - CYP1A2_102)/ CYP1A2_1TAU
CYP1A2_102 =INTEG(CYP1A2_1R02, CYP1A2_1A1)

CYP1A2_1R03 = (CYP1A2_102 - CYP1A2_103)/ CYP1A2_1TAU
CYP1A2_103 =INTEG(CYP1A2_1R03, CYP1A2_1A2)

! TRANSFER OF DIOXIN FROM PLACENTA TO FETUS
! FETAL EXPOSURE ONLY DURING EXPOSURE

IF (T.LT.TRANSTIME_ON) THEN

SWITCH_trans =0.0
ELSE

SWITCH_trans =1.0
END IF

!TRANSFER OF DIOXIN FROM PLACENTA TO FETUS
! MODIFICATION 26 SEPTEMBER 2003

CONSTANT PFETUS= 4.0 !

CONSTANT CLPLA_FET =0.17 !

RAMPF = (CLPLA_FET*CPLA) *SWITCH_trans
AMPF=INTEG(RAMPF,0.0)

!TRANSFER OF DIOXIN FROM FETUS TO PLACENTA
RAFPM = (CLPLA_FET*CFETUS_v)* SWITCH_trans !

AFPM = INTEG(RAFPM,0.0)

! TCDD IN PLACENTA (MOTHER) COMPARTMENT
RAPLAB= QPLAMCA - CPLAB) - PAPLA* (CPLAB -CFLPLAR)

APLAB = INTEG(RAPLAB,0.0)

CPLAB = APLAB/(WPLAB+1E-30)

RAPLA = PAPLA*(CPLAB-CFLPLAR)-RAMPF + RAFPM
APLA = INTEG(RAPLA,0.0)

CPLA = APLA/(WPLA+le-30)

! NMOL/H)
! (NMOL)
! (NMOL/ML)
! (NMOL/H)

! (NMOL)
! (NMOL/ML)

This document is a draft for review purposes only and does not constitute Agency policy.

C-64 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

PARAMETER (PARA_ZERO = 1.0E-30)

CFLPLA= IMPLC(CPLA-(CFLPLAR*PPLA +(PLABMAX*CFLPLAR/(KDPLA&

+CFLPLAR+PARA_ZERO)))-CFLPLA,CFLPLAO)

CFLPLAR=DIM(CFLPLA, 0.0)

!UNIT CONVERSION POST SIMULATION
CPLATOTAL= (APLA + APLAB)/((WPLA + WPLAB)+le-30)! TOTAL CONCENTRATION IN
NMOL/ML

PRCT_PLA = (CPLATOTAL/(MSTT+1E-30))*100

PRCT PLAIV = (CPLATOTAL/(IV RlateR+lE-30) )* 100

!FETUS COMPARTMENT
RAFETUS= RAMPF-RAFPM

AFETUS=INTEG(RAFETUS,0.0)

CFETUS=AFETUS/(WTFE+1E-30)

CFETOTAL= CFETUS
CFETUS_v = CFETUS/PFETUS

! UNIT CONVERSION POST SIMULATION
CFETUSNGKG = CFETUS*MW*UNITCORR	!(NG/KG)

AU C_FEN GKGH = INTEG(CFETUSNGKG,0.0)

PRCT_FE = (CFETOTAL/(MSTT+1E-30))*100
PRCT FEIV = (CFETOTAL/(IV RlateR+lE-30) )* 100

! 	CONTROL MASS BALANCE	

BDOSE= IVDOSE +LYMLUM+LIMLUM

BMASSE = EXCLI+AURI+AFB+AF+AREB+ARE+ALIB+ALI+APLA+APLAB+AFETUS
BDIFF = BDOSE-BMASSE

!BODY BURDEN (NG)

BODY_BURDEN = AFB +AF+ARE B +ARE +ALIB+ALI+AP LA+AP LAB !

BBFETUSNG	= AFETUS*MW*UNITCORR ! UNIT (NG)

! BODY BURDEN IN TERMS OF CONCENTRATION (NG/KG)

BBNGKG =( ( (AFB+AF+AREB+ARE+ALIB+ALI+AP LA+AP LAB)/WTO)*MW*UNITCORR) !
AUC BBNGKGH=INTEG(BBNGKG,0.0)

! 	COMMAND OF THE END OF SIMULATION	

TERMT (T.GE. TimeLimit, 'Time limit has been reached.')
END ! END OF THE DERIVATIVE SECTION
END ! END OF THE DYNAMIC SECTION
END ! END OF THE PROGRAM

C.2.4.2. Input Files
C.2.4.2.1. Bell et al. (2007).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC CBNGKGH

This document is a draft for review purposes only and does not constitute Agency policy.

C-65 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

%output 0nciout=l T BBFETUSNG %AJS turned off 9/21/09
%Bell et al. 2007 (rat species)

%protocol: daily dietary dose for 12 weeks followed by a two-week mating
time and 21-day gestation period
%DevTCDD4Species.csl

%RAT_GESTATIONAL_ICF_F083109.csl (now 09-11-09)

%dose levels: 0.0024, 0.008, 0.046 ug/kg/d with 0.00003 ug/kg/d background
%dose levels: 2.4, 8, 46 ng/kg/d with 0.03 ng/kg/day background

i;EXPOSURES SCENARIOS

MAXT
CINT

EXP_TIME_ON
EXP TIME OFF

0. 01

0.1

0

2856

exposure + 2 weeks for mating + 21

DAY_CYCLE

B C K_TIME_ON

BCK_TIME_OFF

IV_LACK

IV_PERIOD

TIMELIMIT

BW_T0

MATTING

TRANSTIME_ON

N FETUS

24
0.

2856.

505.

505.

2856

85

2352
2496
10

iEXPOSURE DOSE SCENARIOS (UG/KG)
MSTOT	= 0.00243

delay before begin exposure (HOUR)

TIME EXPOSURE STOP (HOUR) 12 weeks
days gestation with exposure

DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
% TIME OF BACKGROUND EXPOSURE STOP (HOUR)

SIMULATION LIMIT TIME (HOUR)

BEGINNING MATING (HOUR)

SHOULD BE MATING TIME + 6 DAYS(144 HOURS)

iMSTOT

iMSTOT = 0.0461

% ORAL EXPOSURE DOSE (UG/KG)
= 0.008	% ORAL EXPOSURE DOSE (UG/KG)

% ORAL EXPOSURE DOSE (UG/KG)

C.2.4.2.2. Haavisto et al. (2006).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC CBNGKGH

%Haavisto et al. 2006

%protocol: single dose on GD 13

%dose levels: 0.04, 0.2, and 1.0 ug/kg on GD 13

%dose levels: 40, 200, and 1,000 ng/kg on GD 13

MAXT = 0.001
CINT =0.1

%EXPOSURES SCENARIOS
EXP_TIME_ON	=312

EX P_TIME_0 FF	= 335

DAY CYCLE	=24

TIME AT WHICH EXPOSURE BEGINS (HOUR)
TIME AT WHICH EXPOSURE ENDS (HOUR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-66 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

B C K_TIME_ON
BEGINS (HOUR)
BCK_TIME_OFF
ENDS (HOUR)
IV_LACK
IV_PERIOD
TIMELIMIT
BW_T0
MATTING
TRANSTIME_ON
HOURS)

N FETUS

= 0.

= 0.

= 505
= 505
= 336
= 190
= 0.
= 144.

= 10

iEXPOSURE DOSE	SCENARIOS (UG/KG)

%MSTOT	=0.04

%MSTOT	= 0.2

MSTOT	=1.0

TIME AT WHICH BACKGROUND EXPOSURE
TIME AT WHICH BACKGROUND EXPOSURE

SIMULATION LIMIT TIME (HOUR)

BEGINNING MATTING (HOUR)

SHOULD BE MATTING TIME + 6 DAYS(144

i ORAL EXPOSURE DOSE (UG/KG)
i ORAL EXPOSURE DOSE (UG/KG)
ORAL EXPOSURE DOSE (UG/KG)

C.2.4.2.3. Hojo et al (2002).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC_CBNGKGH
%Hojo et al. 2002

%protocol: single oral dose at GD8
%DevTCDD4Species . csl

%RAT_GESTATIONAL_ICF_F083109.csl (now 09-11-09)
%RAT_GESTATIONAL_ICF_F092009.csl (now 09-21-09
%dose levels: 0.02 0.06, 0.18 ug/kg at GD8
%dose levels: 20, 60, 180 ng/kg at GD8

% author provided the body weight for each group at the beginning og
gestation (g)

%20 ng/kg BW = 271g
%60 ng/kg BW = 275g
%180 ng/kg BW = 262g

%EXPOSURES SCENARIOS
MAXT= 0.001
CINT =0.1

EXP_TIME_ON	=192

EX P_TIME_0 FF	= 216

DAY_CYCLE	=24

B C K_T I ME_ON	= 0.

BCK_TIME_OFF	= 0.

IV_LACK	= 5 05

IV_PERIOD	= 505

TIMELIMIT	= 216

% BW_T0	= 190

MATTING	= 0.

TRANSTIME_ON	= 144.
HOURS)

N FETUS	=10

delay before begin exposure (HOUR)

TIME EXPOSURE STOP (HOUR)

DELAY BEFORE BACGROUND EXPOSURE (HOUR)
TIME OF BACKGROUND EXPOSURE STOP (HOUR)

SIMULATION LIMIT TIME (HOUR)

BEGINNING MATTING (HOUR)

SHOULD BE MATTING TIME + 6 DAYS(144

This document is a draft for review purposes only and does not constitute Agency policy.

C-67 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

%EXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT
% BW_T 0

%MSTOT
% BW_T 0

MSTOT
BW TO

= 0. 02
= 275

= 0. 06
= 262

= 0. 18
= 278

% ORAL EXPOSURE DOSE (UG/KG)
% 20 ng/kg BW = 271g

% ORAL EXPOSURE DOSE (UG/KG)
%60 ng/kg BW = 275g

5 ORAL EXPOSURE DOSE (UG/KG)
i;180 ng/kg BW = 262g

C.2.4.2.4. Ikeda et al. (2005).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH

%Ikeda et al. 2005 (rat species)

%protocol: loading dose of 400 ng/kg followed by weekly maintenance doses of
80 ng/kg for 6 weeks,

%dose levels: 0.4 ug/kg/day followed by weekly 0.08 ug/kg/day
%dose levels: 400 ng/kg/day followed by weekly 80 ng/kg/day

%EXPOSURES SCENARIOS
MAXT	=. 1

CINT	=0.1 ?

EXP_TIME_ON	= 0

EX P_T I ME_0 F F	= 1008

MATING (2 WEEKS) + MATING

DAY_CYCLE
B C K_TIME_ON
(HOUR)

BCK_TIME_OFF
(HOUR)

IV_LACK
IV_PERIOD
TIMELIMIT
BW_T0
MATTING
TRANSTIME_ON
N FETUS

= 168
= 0.

= 167.

= 505.
= 505.
= 1008
= 250
= 504
= 648
= 10

% TIME AT WHICH EXPOSURE
% TIME AT WHICH EXPOSURE
(1 WEEK) + GESTATION (3 WEEKS)
% WEEKLY CYCLE
% TIME AT WHICH BACKGROUND

BEGINS (HOUR)

ENDS (HOUR); PRE-

EXPOSURE BEGINS

TIME AT WHICH BACKGROUND EXPOSURE ENDS

b SIMULATION TIME LIMIT (HOUR)

BEGINNING OF MATING (HOUR)

SHOULD BE MATING TIME + 6 DAYS (144 HOURS)

iEXPOSURE DOSE SCENARIOS (UG/KG)
MSTOT	=0.08

MSTOTBCKGR	=0.32

5 ORAL EXPOSURE DOSE IN UG/KG
BACKGROUND EXPOSURE IN UG/KG

C.2.4.2.5. Kattainen et al. (2001).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC CBNGKGH

This document is a draft for review purposes only and does not constitute Agency policy.

C-68 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

%Kattainen et al. 2001
%protocol: single gavage at GD15
%DevTCDD4Species . csl

%RAT_GESTATIONAL_ICF_F083109.csl (now 09-11-09)
%dose levels: 0.03 0.1, 0.3, 1 ug/kg at GD15
%dose levels: 30, 100 300, 1000 ng/kg at GD15

MAXT=0.001
CINT =0.1

%EXPOSURES SCENARIOS

EXP_TIME_ON	= 336

EX P_TIME_0 FF	= 360

DAY_CYCLE	=24

B C K_T I ME_ON	= 0.
(HOUR)

BCK_TIME_OFF	= 0.
(HOUR)

IV_LACK	= 5 05

IV_PERIOD	= 505

TIMELIMIT	= 360

BW_T0	= 190

MATTING	= 0.

TRANSTIME_ON	= 144.
HOURS)

N FETUS	=10

delay before begin exposure (HOUR)
TIME EXPOSURE STOP (HOUR)

DELAY BEFORE BACKGROUND EXPOSURE

TIME OF BACKGROUND EXPOSURE STOP

SIMULATION LIMIT TIME (HOUR)

BEGINNING MATTING (HOUR)

SHOULD BE MATTING TIME + 6 DAYS(144

•;EXPOSURE	DOSE SCENARIOS (UG/KG)

%MSTOT	=0.03

%MSTOT	=0.1

%MSTOT	=0.3

MSTOT	= 1

ORAL EXPOSURE DOSE	(UG/KG)

ORAL EXPOSURE DOSE	(UG/KG)

ORAL EXPOSURE DOSE	(UG/KG)

ORAL EXPOSURE DOSE	(UG/KG

C.2.4.2.6. Markowski et al. (2001).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC CBNGKGH

%Markowski et al. 2001
%protocol: single gavage at GD18
%DevTCDD4Species . csl

%RAT_GESTATIONAL_ICF_F083109 . csl (now 09-11-09)
%dose levels: 0.02 0.06, 0.18 ug/kg at GD18
%dose levels: 20, 60, 180 ng/kg at GD18

•;EXPOSURES SCENARIOS
MAXT=0 . 0001
CINT =0.1

EXP_TIME_ON	= 4 08

EX P_T I ME_0 F F	=4 32

DAY_CYCLE	=24

B C K_T I ME_ON	= 0.

BCK TIME OFF	= 0.

delay before begin exposure (HOUR)

TIME EXPOSURE STOP (HOUR)

DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
TIME OF BACKGROUND EXPOSURE STOP (HOUR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-69 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

IV_LACK	= 5 05

IV_PERIOD	= 505

TIMELIMIT	= 432

BW_T0	= 190

MATTING	= 0.
TRANSTIME_ON = 144.

N FETUS	=10

% SIMULATION LIMIT

% BEGINNING MATING
% SHOULD BE MATING

TIME (HOUR)

(HOUR)

TIME + 6 DAYS(144 HOURS)

iEXPOSURE DOSE SCENARIOS

%MSTOT
%MSTOT
MSTOT

= 0. 02
= 0. 06
= 0. 18

(UG/KG)

% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)

C.2.4.2.7. Miettinen et al. (2006).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC CBNGKGH

%Miettinen et al. 2006

%protocol: single oral dose at GD15

%DevTCDD4Species . csl

%RAT_GESTATIONAL_ICF_F083109 . csl (now 09-11-09)
%dose levels: 0.03 0.1, 0.3, 1 ug/kg at GD15
%dose levels: 30, 100, 300, 1000 ng/kg at GD15

MAXT=0.01

CINT =0.1	%

%EXPOSURES SCENARIOS

EXP TIME ON

=

336

O
O

delay before begin exposure (HOUR)

EXP TIME OFF

=

360

O
O

TIME EXPOSURE STOP (HOUR)

DAY CYCLE

=

24





BCK TIME ON

=

0.

O
O

DELAY BEFORE BACKGROUND EXPOSURE (HOUR)

BCK TIME OFF

=

0.

O
O

TIME OF BACKGROUND EXPOSURE STOP (HOUR)

IV LACK

=

505





IV PERIOD

=

505





TIMELIMIT

=

360

O
O

SIMULATION LIMIT TIME (HOUR)

BW TO

=

180





MATTING

=

0.

O
O

BEGINNING MATING (HOUR)

TRANSTIME ON

=

144 .

O
O

SHOULD BE MATING TIME + 6 DAYS(144 HOURS)

N FETUS

=

10





iEXPOSURE DOSE SCENARIOS	(UG/KG)

%MSTOT =0.03	% ORAL EXPOSURE DOSE (UG/KG)

%MSTOT =0.1	% ORAL EXPOSURE DOSE (UG/KG)

%MSTOT =0.3	% ORAL EXPOSURE DOSE (UG/KG)

MSTOT =1	% ORAL EXPOSURE DOSE (UG/KG)

C.2.4.2.8. Nohara et al. (2000).

%clear variable
output 0clear

This document is a draft for review purposes only and does not constitute Agency policy.

C-70 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC_CBNGKGH

%Nohara et al. 2000

%protocol: single gavage at GD15

%DevTCDD4Species.csl

%RAT_GESTATIONAL_ICF_F083109.csl (now 09-11-09)

%dose levels: 0.0125, 0.050, 0.2, or 0.8 ug TCDD:kg body weight by gavage on
GD15 .

%dose levels: 12.5, 50, 200, or 800 ng TCDD:kg body weight by gavage on GD15.
MAXT=0.01

CINT =0.1	%

%EXPOSURES

SCENARIOS





EXP TIME ON

=

336

O
O

delay before begin exposure (HOUR)

EXP TIME OFF

=

360

O
O

TIME EXPOSURE STOP (HOUR)

DAY CYCLE

=

24





BCK TIME ON

=

0.

O
O

DELAY BEFORE BACKGROUND EXPOSURE (HOUR)

BCK TIME OFF

=

0.

O
O

TIME OF BACKGROUND EXPOSURE STOP (HOUR)

IV LACK

=

505





IV PERIOD

=

505





TIMELIMIT

=

360



% SIMULATION LIMIT TIME (HOUR)

BW TO

=

180





MATTING

=

0.



% BEGINNING MATTING (HOUR)

TRANSTIME ON

=

144 .

% SHOULD BE MATTING TIME + 6 DAYS(144 HOURS)

N FETUS

=

10





iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT
%MSTOT
%MSTOT
MSTOT

= 0.0125	% ORAL EXPOSURE DOSE (UG/KG)

= 0.050 % ORAL EXPOSURE DOSE (UG/KG)
=0.2	% ORAL EXPOSURE DOSE (UG/KG)

=0.8	% ORAL EXPOSURE DOSE (UG/KG)

C.2.4.2.9. Ohsako et al. (2001).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC CBNGKGH

iOhsako et al. 2001

^protocol: single oral dose at GD15
•;DevTCDD4Species . csl
oRAT_GESTATIONAL_ICF_F083109.csl
oRAT_GESTATIONAL_ICF_F0 92 0 0 9.csl
idose levels: 0.0125, 0.05, 0.2,

(now 09-11-09)
(now 09-21-09)
0.8 ug/kg at GD15

idose levels: 12.5, 50, 200, 800 ng/kg at GD15

%EXPOSURES SCENARIOS
MAXT=0.01

CINT =0.1	%

EXP TIME ON	= 360	% delay before begin exposure (HOUR)

EXP TIME OFF = 384	% TIME EXPOSURE STOP (HOUR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-71 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

DAY_CYCLE	=24

B C K_T I ME_ON	= 0.

BCK_TIME_OFF	= 0.

IV_LACK	= 5 05

IV_PERIOD	= 505

TIMELIMIT	= 384

BW_T0	= 200

MATTING	= 0.

TRANSTIME_ON	= 144.
HOURS)

N FETUS	=10

DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
TIME OF BACKGROUND EXPOSURE STOP (HOUR)

SIMULATION LIMIT TIME (HOUR)

BEGINNING MATTING (HOUR)

SHOULD BE MATTING TIME + 6 DAYS(144

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT
%MSTOT
%MSTOT
MSTOT

0.0125	% ORAL EXPOSURE DOSE (UG/KG)

0.05	% ORAL EXPOSURE DOSE (UG/KG)

0.20	% ORAL EXPOSURE DOSE (UG/KG)

=0.80	% ORAL EXPOSURE DOSE (UG/KG)

C.2.4.2.10. Schantzetal. (1996) andAmin et al. (2000).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC_CBNGKGH

%Amin et al. 2000 (rat species) and Schantz et al. 1996
%protocol: daily doses on GDs 10 to 16
%DevTCDD4Species.csl

%RAT_GESTATIONAL_ICF_F083109.csl (now 09-11-09)

%dose levels: 25 and 100 ng/kg/day
%dose levels: 0.025 and 0.100 ug/kg/day

i;EXPOSURES SCENARIOS

MAXT
CINT

EXP_TIME_ON
EX P_T I ME_0 F F
to 16
DAY_CYCLE
B C K_TIME_ON
BCK_TIME_OFF
IV_LACK
IV_PERIOD
TIMELIMIT
BW_T0
MATTING
TRANSTIME_ON
N FETUS

0. 001
0.1 =?
240.
384 .

24

1000.

1000.

505.

505.

384 .

250.

0

144 .

10

% TIME AT WHICH EXPOSURE BEGINS (HOUR)
i TIME AT WHICH EXPOSURE ENDS (HOUR) GD 10

weekly cycle

% DELAY BEFORE BACKGROUND EXPOSURE (HOUR)
% TIME OF BACKGROUND EXPOSURE STOP (HOUR)

% SIMULATION LIMIT TIME (HOUR)

BEGINNING MATTING (HOUR)

% SHOULD BE MATTING TIME + 6 DAYS(144 HOURS)

ORAL EXPOSURE DOSE (UG/KG)

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT	= .025

MSTOT	= .100

MSTOTBCKGR	=0 % Background Exposure (UG/KG)

This document is a draft for review purposes only and does not constitute Agency policy.

C-72 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

C.2.4.2.11. Seoetal. (1995).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC CBNGKGH

%Seo et al. 1995

%protocol: daily doses on GDs 10-16
%DevTCDD4Species.csl

%RAT_GESTATIONAL_ICF_F083109.csl (now 09-11-09)
%dose levels: 0.025 and 0.1 ug/kg on GDs 10-16
%dose levels: 25 and 100 ng/kg on GDs 10-16

MAXT = 0.01
CINT =0.1

%EXPOSURES SCENARIOS

EXP_TIME_ON	= 24 0

EX P_T I ME_0 F F	= 384

DAY_CYCLE	=24

B C K_T I ME_ON	= 0.
BEGINS (HOUR)

BCK_TIME_OFF	= 0.
ENDS (HOUR)

IV_LACK	= 5 05

IV_PERIOD	= 505

TIMELIMIT	= 384

BW_T0	= 190

MATTING	= 0.

TRANSTIME_ON	= 144.
HOURS)

N FETUS	=10

TIME AT WHICH EXPOSURE BEGINS (HOUR)
TIME AT WHICH EXPOSURE ENDS (HOUR)

TIME AT WHICH BACKGROUND EXPOSURE

TIME AT WHICH BACKGROUND EXPOSURE

SIMULATION LIMIT TIME (HOUR)

BEGINNING MATTING (HOUR)

SHOULD BE MATTING TIME + 6 DAYS(144

•;EXPOSURE DOSE SCENARIOS (UG/KG)
%MSTOT	= 0.025

MSTOT	=0.1

% ORAL EXPOSURE DOSE (UG/KG)
ORAL EXPOSURE DOSE (UG/KG)

C.2.5. Mouse Standard Model
C.2.5.1. Model Code

PROGRAM: 'Three Compartment PBPK Model for TCDD in Mice: Standard Model (Non-
Gestation)'

IMice Dioxin 3C June09 1 icf afterKKfix v3 mousenongest.csl
!MICE~NON_GESTAT_ICF_F0831097csl	~ ~

!MICE_NON_GESTAT_ICF_F0 93 0 0 9.csl
!MICE_NON_GESTAT_ICF_Fl0 0 609.csl

INITIAL ! INITIALIZATION OF PARAMETERS
!SIMULATION PARAMETERS ====

This document is a draft for review purposes only and does not constitute Agency policy.

C-73 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CONSTANT PARA_ZERO	= 1D-30

CONSTANT EXP_TIME_ON	=	0.0	! TIME AT WHICH EXPOSURE BEGINS

(HOURS)

CONSTANT EXP_TIME_0FF = 2832	! TIME AT WHICH EXPOSURE ENDS

(HOURS)

CONSTANT DAY_CYCLE	=24	! NUMBER OF HOURS BETWEEN DOSES

(HOURS)

CONSTANT BCK_TIME_ON	=	0.0	! TIME AT WHICH BACKGROUND EXPOSURE

BEGINS (HOURS)

CONSTANT BCK_TIME_OFF =	0.0	! TIME AT WHICH BACKGROUND EXPOSURE

ENDS (HOURS)

CONSTANT MW=322 ! MOLECULAR WEIGHT (NG/NMOL)

CONSTANT SERBLO =0.55
CONSTANT UNITCORR = 1000

!CONSTANT EXPOSURE CONTROL ===========

!ACUTE, SUBCHRONIC, CHRONIC	EXPOSURE =====

!OR BACKGROUND EXPOSURE (IN	THIS CASE 3 TIMES A DAY)===

CONSTANT MSTOTBCKGR	=	0.0	!ORAL BACKGROUND EXPOSURE DOSE

(UG/KG)

CONSTANT MSTOT	=	0.15	!ORAL EXPOSURE DOSE (UG/KG)

CONSTANT MSTOTsc	=	0.0	! SUBCUTANEOUS EXPOSURE DOSE

(UG/KG)

!ORAL ABSORPTION
MSTOT_NM	= MSTOT/MW !AMOUNT IN NMOL/G

! INTRAVENOUS ABSORPTION
CONSTANT DOSEIV =0.0	!INJECTED DOSE (UG/KG)

DOSEIV_NM = DOSEIV/MW ! CONVERTS THE INJECTED DOSE TO NMOL/G

!INITIAL GUESS OF THE FREE CONCENTRATION IN THE LIGAND (COMPARTMENT
INDICATED BELOW)====

CONSTANT CFLLI0	=	0.0	!LIVER (NMOL/ML)

!BINDING CAPACITY (AhR) FOR NON LINEAR BINDING (COMPARTMENT INDICATED
BELOW) (NMOL/ML)

CONSTANT LIBMAX	=	3.5e-4	! LIVER (NMOL/ML), WANG ET AL.

1997

! PROTEIN AFFINITY CONSTANTS (1A2 OR AhR, COMPARTMENT INDICATED BELOW)
(NMOL/ML)===

CONSTANT KDLI	=	1.0e-4	!LIVER (AhR)(NMOL/ML), WANG ET AL.

1997

CONSTANT KDLI2	=	2.Oe-2	!LIVER (1A2 ) (NMOL/ML) , EMOND ET AL.

2004

!===EXCRETION AND ABSORPTION CONSTANT (OPTIMIZED)

CONSTANT KST	= 0.3 ! GASTRIC RATE CONSTANT (HR-1),

CONSTANT KABS	= 0.48 !INTESTINAL ABSORPTION CONSTANT (HR-1) ),

WANG ET AL. 1997

! ELIMINATION CONSTANTS

CONSTANT CLURI	=	0.09 ! URINARY CLEARANCE (ML/HR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-74 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

! ==test elimination variable

constant kelv	=	0.4	! INTERSPECIES VARIABLE ELIMINATION

CONSTANT (1/HOUR)

! CONSTANT TO DIVIDE THE ABSORPTION INTO LYMPHATIC AND PORTAL FRACTIONS
CONSTANT A	=0.7	! LYMPHATIC FRACTION, WANG ET AL.

1997

!PARTITION COEFFICIENTS OPTIMIZED
CONSTANT PF	= 400

CONSTANT PRE	= 3

AL. 2 0 0 0

CONSTANT PLI	= 6

! ADIPOSE TISSUE/BLOOD
! REST OF THE BODY/BLOOD, WANG ET

! LIVER/BLOOD, WANG ET AL. 19 97

!===PARAMETER FOR INDUCTION OF CYP 1A2

CONSTANT PAS_INDUC=
CONSTANT CYP1A2_10UTZ
(NMOL/ML)

CONSTANT CYP1A2_1A1 =
CYP1A2_1EC5 0
CYP1A2_1A2 =
CYP1A2_1K0UT
CYP1A2_1TAU
CYP1A2 1EMAX

5

13
5
1
5

CONSTANT
CONSTANT
CONSTANT

CONSTANT	_

CONSTANT CYP1A2_1EMAX = 600
(UNITLESS)

CONSTANT HILL	=0.6

EFFECT CONSTANT (UNITLESS)

!DIFFUSIONAL PERMEABILITY
CONSTANT PAFF	=0.12

CONSTANT PAREF =0.03
CONSTANT PALIF =0.35

! INCLUDE INDUCTION? (1 = YES, 0 = NO)
! DEGRADATION CONCENTRATION CONSTANT OF 1A2

! BASAL CONCENTRATION OF 1A1 (NMOL/ML)

! DISSOCIATION CONSTANT TCDD-CYP1A2 (NMOL/ML)
! BASAL CONCENTRATION OF 1A2 (NMOL/ML)

! FIRST ORDER RATE OF DEGRADATION (H-l)

HOLDING TIME (H)

! MAXIMUM INDUCTION OVER BASAL EFFECT

!HILL CONSTANT; COOPERATIVELY LIGAND BINDING

FRACTION

ADIPOSE (UNITLESS), WANG ET AL.
REST OF THE BODY (UNITLESS)
LIVER (UNITLESS)

2000

!COMPARTMENT TISSUE BLOOD VOLUME =========

CONSTANT WLI0	= 0.0549 ! LIVER, ILSI 1994

CONSTANT WF0	= 0.069 ! ADIPOSE

!TISSUE BLOOD FLOW EXPRESSED AS A FRACTION OF CARDIAC OUTPUT
CONSTANT QFF = 0.070	! ADIPOSE TISSUE BLOOD FLOW FRACTION

(UNITLESS), LEUNG ET AL. 1990

CONSTANT QLIF = 0.161	! LIVER (UNITLESS) ILSI ET AL. 1994

!COMPARTMENT TISSUE BLOOD EXPRESSED AS A FRACTION OF THE TOTAL
COMPARTMENT VOLUME

CONSTANT WFB0 = 0.050
CONSTANT WREB0 = 0.030
CONSTANT WLIB0 = 0.266

ADIPOSE TISSUE, WANG ET AL. 1997
REST OF THE BODY, WANG ET AL. 1997
LIVER, WANG ET AL. 19 97

! EXPOSURE SCENARIO FOR UNIQUE OR REPETITIVE WEEKLY OR MONTHLY EXPOSURE
! NUMBER OF EXPOSURES PER WEEK
CONSTANT WEEK_LACK =0.0	! DELAY BEFORE EXPOSURE ENDS (WEEK)

CONSTANT WEEK_PERIOD = 168	! NUMBER OF HOURS IN THE WEEK (HOURS)

CONSTANT WEEK_FINISH = 120 ! TIME EXPOSURE ENDS (HOURS)

! NUMBER OF EXPOSURES PER MONTH
CONSTANT MONTH LACK =0.0 ! DELAY BEFORE EXPOSURE (MONTH)

This document is a draft for review purposes only and does not constitute Agency policy.

C-75 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!SET FOR BACKGROUND EXPOSURE===========

!CONSTANT FOR BACKGROUND EXPOSURE===========

CONSTANT Day_LACK_BG =0.0	! DELAY BEFORE EXPOSURE BEGINS (HOURS)

CONSTANT Day_PERIOD_BG =24 ! LENGTH OF EXPOSURE (HOURS)

! NUMBER OF EXPOSURES PER WEEK

CONSTANT WEEK_LACK_BG =0.0
CONSTANT WEEK_PERIOD_BG = 168
CONSTANT WEEK FINISH BG = 168

DELAY BEFORE BACKGROUD EXPOSURE (WEEK)
NUMBER OF HOURS IN THE WEEK (HOURS)

TIME EXPOSURE ENDS (HOURS)

!GROWTH CONSTANT FOR RAT AND MOUSE
!CONSTANT FOR MOTHER BODY WEIGHT GROWTH ======

CONSTANT BW_T0 = 20	!CHANGED FOR SIMULATION

!CONSTANT USED IN CARDIAC OUTPUT EQUATION, HADDAD 2001
CONSTANT QCCAR =275	!CONSTANT (ML/MIN/KG)

! COMPARTMENT LIPID EXPRESSED AS THE FRACTION OF TOTAL LIPID

CONSTANT F_TOTLIP = 0.855
CONSTANT B_TOTLIP = 0.0033
CONSTANT RE_TOTLIP = 0.019
CONSTANT LI TOTLIP =0.06

!ADIPOSE TISSUE (UNITLESS)
!BLOOD (UNITLESS)

!REST OF THE BODY (UNITLESS)
!LIVER (UNITLESS)

END ! END OF THE INITIAL SECTION

DYNAMIC ! DYNAMIC SIMULATION SECTION

ALGORITHM
CINTERVAL
MAXTERVAL
MINTERVAL
VARIABLE
CONSTANT
(HOURS)

CINTXY = CINT
PFUNC = CINT

IALG
CINT
MAXT
MINT
T

TIMELIMIT

2	!GEAR METHOD

1.0	!COMMUNICATION INTERVAL

1.0e+10	IMAXIMUM CALCULATION INTERVAL

1.0E-10	!MINIMUM CALCULATION INTERVAL

0.0	!HOUR

2904.0	!SIMULATION TIME LIMIT

!TIME CONVERSION

DAY
WEEK
MONTH
YEAR

= T/24.0
= T/168.0
= T/730.0
= T/87 60.0

TIME IN DAYS
TIME IN WEEKS
TIME IN MONTHS
TIME IN YEARS

!NMAX =MAX(T,CTFNGKG)
nmax =max(T,CFNGKG)

DERIVATIVE ! PORTION OF CODE THAT SOLVES DIFFERENTIAL EQUATIONS

!CHRONIC OR SUBCHRONIC EXPOSURE SCENARIO
!NUMBER OF EXPOSURES PER DAY

DAY_LACK	= EXP_TIME_ON

DAY_PERIOD = DAY_CYCLE
DAY_FINISH = CINTXY
MONTH_PERIOD = TIMELIMIT
MONTH FINISH = EXP TIME OFF

DELAY BEFORE EXPOSURE BEGINS
EXPOSURE PERIOD (HOURS)
LENGTH OF EXPOSURE (HOURS)
EXPOSURE PERIOD (MONTHS)
LENGTH OF EXPOSURE (MONTHS)

(HOURS)

This document is a draft for review purposes only and does not constitute Agency policy.

C-76 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!NUMBER OF EXPOSURES PER DAY AND MONTH
DAY_FINISH_BG = CINTXY

MONTH_LACK_BG = BCK_TIME_ON ! DELAY BEFORE BACKGROUD EXPOSURE BEGINS
(MONTHS)

MONTH_PERIOD_BG = TIMELIMIT	! BACKGROUND EXPOSURE PERIOD (MONTHS)

MONTH_FINISH_BG = BCK_TIME_OFF ! LENGTH OF BACKGROUND EXPOSURE (MONTHS)

! FRACTION OF DIOXIN ABSORBED IN THE PORTAL FRACTION OF THE LIVER
B = 1.0-A

!GROWTH UP EQUATION (G)

PARAMETER (BW_RMN = 1.0E-30)

WT0= (BW_T0 *(1.0+(0.41*T)/(1402.5+T+BW_RMN) ) )

! VARIABILITY OF REST OF THE BODY DEPENDS ON OTHER ORGANS
!REST OF THE BODY FRACTION; UPDATED FOR EPA ASSESSMENT
WRE0 = (0.91 - (WLIB0*WLI0 + WFB0*WF0 + WLI0 + WF0))/(1+WREB0)

! REST OF THE BODY BLOOD	FLOW FRACTION

QREF = 1.0-(QFF+QLIF)	!REST OF BODY BLOOD FLOW (ML/HR)

!SUMMATION OF BLOOD FLOW	FRACTION (SHOULD BE EQUAL TO 1)

QTTQF = QFF+QREF+QLIF	! SUM MUST EQUAL 1

!COMPARTMENT VOLUME (G)
WF = WF0 * WTO
WRE = WRE0 * WTO
WLI = WLI0 * WTO

ADIPOSE

REST OF THE BODY
LIVER

!COMPARTMENT TISSUE BLOOD	(G)

WFB = WFB0 * WF	! ADIPOSE

WREB = WREB0 * WRE	! REST OF THE BODY

WLIB = WLIB0 * WLI	! LIVER

!CARDIAC OUTPUT FOR THE GIVEN BODY WEIGHT
QC= QCCAR* 60*(WTO/1000.0)**0.75

QF = QFF*QC	! ADIPOSE TISSUE BLOOD FLOW RATE (ML/HR)

QLI = QLIF*QC	! LIVER TISSUE BLOOD FLOW RATE (ML/HR)

QRE = QREF*QC	! REST OF THE BODY BLOOD FLOW RATE (ML/HR)

QTTQ = QF+QRE+QLI !TOTAL FLOW RATE (ML/HR)

!PERMEABILITY ORGAN FLOW (ML/HR) =======

PAF = PAFF*QF	! ADIPOSE TISSUE

PARE = PAREF*QRE	! REST OF THE BODY

PALI = PALIF*QLI	! LIVER TISSUE

!ABSORPTION SECTION
! ORAL

!BACKGROUND EXPOSURE

!EXPOSURE FOR STEADY STATE CONSIDERATION
!REPETITIVE EXPOSURE SCENARIO

MSTOT_NMBCKGR = MSTOTBCKGR/322 !AMOUNT IN NMOL/G

This document is a draft for review purposes only and does not constitute Agency policy.

C-77 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

MSTTBCKGR =MSTOT_NMBCKGR *WT0

!REPETITIVE ORAL BACKGROUND EXPOSURE SCENARIOS
DAY_EX P 0 S U RE_B G = PULSE(DAY_LACK_BG,DAY_PERIOD_BG,DAY_FINISH_BG)
WEEK_EXPOSURE_BG = PULSE(WEEK_LACK_BG,WEEK_PERIOD_BG,WEEK_FINISH_BG)
MONTH_EXPOSURE_BG = PULSE(MONTH_LACK_BG,MONTH_PERIOD_BG,MONTH_FINISH_BG)

MSTTCH_BG = (DAY_EXPOSURE_BG*WEEK_EXPOSURE_BG*MONTH_EXPOSURE_BG)*MSTTBCKGR
MSTTFR_BG = MSTTBCKGR/CINT

totalBG= integ (MSTTCH_BG,0.0)

CYCLE BG =DAY EXPOSURE BG*WEEK EXPOSURE BG*MONTH EXPOSURE BG

!CONDITIONAL ORAL EXPOSURE (BACKGROUND EXPOSURE)
IF (MSTTCH_BG.EQ.MSTTBCKGR) THEN
ABSMSTT_GB= MSTTFR_BG

ELSE

ABSMSTT_GB =0.0
END IF

!EXPOSURE + !REPETITIVE EXPOSURE SCENARIO
IV= DOSEIV_NM * WTO !AMOUNT IN NMOL
MSTT= MSTOT_NM * WTO !AMOUNT IN NMOL

DAY_EXPOSURE = PULSE(DAY_LACK,DAY_PERIOD,DAY_FINISH)

WEEK_EXPOSURE = PULSE(WEEK_LACK,WEEK_PERIOD,WEEK_FINISH)
MONTH_EXPOSURE = PULSE(MONTH_LACK,MONTH_PERIOD,MONTH_FINISH)

MSTTCH = (DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE)*MSTT
CYCLE = DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE

SUMEXPEVENT= integ (CYCLE,0.0)*cint INUMBER OF CYCLE GENERATE DURING
SIMULATION

MSTTFR = MSTT/CINT

! CONDITIONAL ORAL EXPOSURE
IF (MSTTCH.EQ.MSTT) THEN

ABSMSTT= MSTTFR
ELSE

ABSMSTT =0.0
END IF

CYCLETOT=INTEG(CYCLE, 0.0)

!MASS CHANGE IN THE LUMEN
RMSTT= -(KST+KABS)*MST+ABSMSTT +ABSMSTT_GB ! RATE OF CHANGE (NMOL/H)
MST = INTEG(RMSTT,0.0) !AMOUNT OF STAY IN DUODENUM (NMOL)

!ABSORPTION IN LYMPH CIRCULATION
LYRMLUM = KABS*MST*A
LYMLUM = INTEG(LYRMLUM,0.0)

!ABSORPTION IN PORTAL CIRCULATION

This document is a draft for review purposes only and does not constitute Agency policy.

C-78 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

LIRMLUM = KABS*MST*B

LIMLUM = INTEG(LIRMLUM,0.0)

!PERCENT OF DOSE REMAINING IN THE GI TRACT
PRCT_remain_GIT = (MST/(MSTT+1E-30))*100

RFECES = KST*MST + REXCLI

FECES = INTEG(RFECES,0.0)
prctFECES = (FECES/(BDOSE_TOTAL+lE-30))*100

!ABSORPTION OF DIOXIN BY IV ROUTE	

IVR= IV/PFUNC ! RATE FOR IV INFUSION IN BLOOD
EXPIV= IVR * (1.0-STEP(PFUNC))

IVDOSE = integ(EXPIV,0.0)

!SYSTEMIC BLOOD CONCENTRATION (NMOL/ML)
! MODIFICATION ON OCTOBER 6, 2009
CB=(QF*CFB+QRE*CREB+QLI*CLIB+EXPIV+LYRMLUM)/(QC+CLURI) !
CA = CB

!URINARY EXCRETION BY KIDNEY
! MODIFICATION ON OCTOBER 6, 2009
RAURI = CLURI *CB

AURI = INTEG(RAURI,0.0)

prctAURI = (AURI/(BDOSE_TOTAL+lE-30))*100

!UNIT CONVERSION POST SIMULATION
PRCT_B = (CB/(MSTT+1E-30))*100 ! PERCENT OF DOSE/G TISSUE
CBNGKG=CB*MW*UNITCORR

CBSNGKGLIADJ= (CB*MW*UNITCORR*(1.0/B_TOTLIP)*(1.0/SERBLO))![NG of TCDD
Serum/Kg OF LIPIP]

CBPMOL_KG= CB*UNITCORR*UNITCORR
CBNGG = CB*MW

!ADIPOSE TISSUE COMPARTMENT
! TISSUE BLOOD SUBCOMPARTMENT
RAFB = QF*(CA-CFB)-PAF*(CFB-CF/PF)
AFB = INTEG(RAFB,0.0)
CFB = AFB/WFB

!TISSUE SUBCOMPARTMENT
RAF = PAF*(CFB-CF/PF)
AF = INTEG(RAF, 0.0)

CF = AF/WF

!CONCENTRATION IN PMOL/KG

(NMOL/HR)
!(NMOL)
(NMOL/ML)

(NMOL/HR)
!(NMOL)
(NMOL/ML)

!POST SIMULATION UNIT CONVERSION
CFTOTAL = (AF + AFB)/(WF + WFB) ! TOTAL CONCENTRATION IN FAT(NM/ML)
PRCT_F = (CFTOTAL/(MSTT+1E-30))*100 ! PERCENT OF DOSE IN FAT
CFNGKG = CFTOTAL*MW*UNITCORR
CFUGG=(CFTOTAL*MW)/UNITCORR

CFPMOL_KG= CFTOTAL*UNITCORR*UNITCORR	!CONCENTRATION IN PMOL/KG

CFNGG = CFTOTAL*MW

!REST OF THE BODY COMPARTMENT
!TISSUE BLOOD SUBCOMPARTMENT

This document is a draft for review purposes only and does not constitute Agency policy.

C-79 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

RAREB= QRE*(CA-CREB)-PARE*(CREB-CRE/PRE)	!(NMOL/HR)

AREB = INTEG(RAREB,0.0)	!(NMOL)

CREB = AREB/WREB	!(NMOL/ML)
!TISSUE SUBCOMPARTMENT

RARE = PARE*(CREB - CRE/PRE)	!(NMOL/HR)

ARE = INTEG(RARE,0.0)	!(NMOL)

CRE = ARE/WRE	!(NMOL/ML)

!POST SIMULATION UNIT CONVERSION
CRETOTAL= (ARE + AREB)/(WRE + WREB)	! CONCENTRATION AT STEADY

STATE

PRCT RE = (CRETOTAL/(MSTT+1E-30))*100

!LIVER COMPARTMENT
!TISSUE BLOOD SUBCOMPARTMENT
RALIB = QLI*(CA-CLIB)-PALI*(CLIB-CFLLIR)+LIRMLUM
ALIB = INTeg(RALIB,0.0)

CLIB = ALIB/WLIB

!TISSUE SUBCOMPARTMENT
RALI = PALI*(CLIB-CFLLIR)-REXCLI
ALI = integ(RALI,0.0)

CLI = ALI/WLI

!(NMOL/HR)
!(NMOL)

!(NMOL/HR)

!(NMOL)
!(NMOL/ML)

!FREE TCCD CONCENTRATION IN LIVER (NMOL/ML)

PARAMETER (LIVER_1RMN = 1.0E-30)

CFLLI= IMPLC(CLI-(CFLLIR*PLI+(LIBMAX*CFLLIR/(KDLI+CFLLI &
+LIVER_1RMN))+((CYP1A2_103*CFLLIR/(KDLI2+CFLLIR &
+LIVER_1RMN)*PAS_INDUC)))-CFLLI,CFLLI0)

CFLLIR=DIM(CFLLI,0.0) ! FREE CONCENTRATION IN LIVER

CBNDLI= LIBMAX*CFLLIR/(KDLI+CFLLIR+LIVER_1RMN) !BOUND CONCENTRATION

!POST SIMULATION UNIT CONVERSION
CLITOTAL= (ALI + ALIB)/(WLI + WLIB)!

PRCT_LI = (CLITOTAL/(MSTT+1E-30))*100 ! PERCENT OF DOSE IN LIVER
rec_occ_AHR= (CFLLIR/(KDLI+CFLLIR+1E-30))*100.0 ! PERCENT OF AhR OCCUPANCY
PROT_occ_lA2= (CFLLIR/(KDLI2+CFLLIR))*100.0 ! PERCENT OF 1A2 OCCUPANCY
CLINGKG =(CLITOTAL*MW*UNITCORR)

CBNDLINGKG = CBNDLI*MW*UNITCORR
CLIUGG=(CLITOTAL*MW)/UNITCORR

CLIPMOL_KG= CLITOTAL*UNITCORR*UNITCORR	!CONCENTRATION IN PMOL/KG

CLINGG = CLITOTAL*MW

!Fraction increase of induction of CYP1A2
fold_ind=(CYP1A2_10UT/CYP1A2_1A2)

VARIATIONOfAC =(CYP1A2_10UT-CYP1A2_1A2)/CYP1A2_1A2

!VARIABLE ELIMINATION BASED ON THE CYP1A2
KBILE_LI_T =((CYP1A2_10UT-CYP1A2_1A2)/CYP1A2_1A2)*Kelv !INDUCED BILIARY
EXCRETION RATE CONSTANT

REXCLI= (KBILE_LI_T*CFLLIR*WLI) !DOSE-DEPENDENT EXCRETION RATE
EXCLI = INTEG(REXCLI,0.0)

!CHEMICAL IN CYP450 (1A2) COMPARTMENT

This document is a draft for review purposes only and does not constitute Agency policy.

C-80 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

!EQUATION FOR INDUCTION OF CYP1A2
CYP1A2_1KINP = CYP1A2_1K0UT* CYP1A2_10UTZ

! MODIFICATION ON OCTOBER 6, 2009
CYP1A2_10UT =INTEG(CYP1A2_1KINP * (1.0 + CYP1A2_1EMAX *(CBNDLI+1.0e-30)**HILL
&

/(CYP1A2_1EC50**HILL + (CBNDLI+1.0e-30)**HILL)) &
- CYP1A2_1K0UT*CYP1A2_10UT, CYP1A2_10UTZ)

! EQUATIONS INCORPORATING DELAY OF CYP1A2 PRODUCTION (NOT USED IN
SIMULATIONS)

CYP1A2_1R02 = (CYP1A2_10UT - CYP1A2_102)/ CYP1A2_1TAU

CYP1A2_102 =INTEG(CYP1A2_1R02, CYP1A2_1A1)

CYP1A2_1R03 = (CYP1A2_102 - CYP1A2_103)/ CYP1A2_1TAU
CYP1A2_103 =INTEG(CYP1A2_1R03, CYP1A2_1A2)

! MASS BALANCE CONTROL
BDOSE= LYMLUM+LIMLUM+IVDOSE

BMASSE = EX C LI+AURI+AFB+AF+AREB+ARE+ALIB+ALI
BDIFF = BDOSE-BMASSE

! AMOUNT TOTAL PRESENT IN THE GI TRACT
BDOSE_TOTAL =LYMLUM+LIMLUM+FECES

!BODY BURDEN IN NG
Body burden =(AFB+AF+AREB+ARE+ALIB+ALI)*MW

!BODY BURDEN CONCENTRATION (NG/KG)

BBNGKG =(((AFB+AF+AREB+ARE+ALIB+ALI)*MW)/(WTO/UNITCORR)) !

!COMMAND FOR END OF SIMULATION
TERMT (T.GE. TimeLimit, 'Time limit has been reached.')

END ! END OF THE DERIVATIVE SECTION
END ! END OF THE DYNAMIC SECTION
END ! END OF PROGRAM

C.2.5.2. Input Files

C.2.5.2.1. Delia Porta (1987) (female)

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% Delia Porta 1987 for female mice.

%dose levels: 2.5 and 5 ug/kg/week for 52 weeks
%dose levels: 2500 and 5000 ng/kg/week for 52 weeks
%dose levels equivalent to: 357 and 714 ng/kg/d

MAXT = 0.01
CINT =0.1
EXP_TIME_ON
EX P_T I ME_0 F F
DAY_CYCLE
BCK TIME ON

0.

8736

168

0.

idelay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

¦;DELAY BEFORE BACGROUND EXPOSURE (HOUR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-81 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

BCK_TIME_OFF	= 0.	%TIME OF BACKGROUND EXPOSURE STOP (HOUR)

TIMELIMIT	= 8736	%SIMULATION LIMIT TIME (HOUR)

BW TO	= 20	% Body weight at the beginning of the simulation

(g); corresponds to 6 weeks of age and taken from Figure 3

%EXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT	=2.5	% exposure dose ug/kg

MSTOT	=5.0	% exposure dose ug/kg

C.2.5.2.2. Delia Porta (1987) (male)

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% Delia Porta 1987 for male mice.

%dose levels: 2.5 and 5 ug/kg/week for 52 weeks
%dose levels: 2500 and 5000 ng/kg/week for 52 weeks
%dose levels equivalent to: 357 and 714 ng/kg/d

MAXT = 0.01
CINT =0.1
EXP_TIME_ON
EX P_T I ME_0 F F
DAY_CYCLE
B C K_TIME_ON
BCK_TIME_OFF
TIMELIMIT
BW TO

0.

8736
168
0.

0.

8736
26

idelay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

¦;DELAY BEFORE BACGROUND EXPOSURE (HOUR)
iTIME OF BACKGROUND EXPOSURE STOP (HOUR)
¦;SIMULATION LIMIT TIME (HOUR)

i Body weight at the beginning of the simulation

(g); corresponds to 6 weeks of age and taken from Figure 3

%EXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT	=2.5	% exposure dose ug/kg

MSTOT	=5.0	% exposure dose ug/kg

C.2.5.2.3. NTP (1982) (female) (chronic)

%RAT2.m

%clear variable
output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG
% output 0nciout=168 T SUMEXPEVENT

% NTP 1982.

%built and check in September 20, 2009
%protocol: twice weekly gavage for 104 weeks
%Rat Dioxin 3C June09 2clean 2.csl
%MICE_NON_GESTAT_ICF_F083109.csl
%MICE_NON_GESTAT_ICF_F092009.csl (now 09-20-09)

%dose levels: 0.02, 0.1, 1 ug/kg/biweekly, ug/kg for 104 weeks
%dose levels: 20, 100, 1000 ng/kg/biweekly,ng/kg for 104 weeks
%dose levels equivalent to: 5.71, 28.57, 285.1 ng/kg/d

This document is a draft for review purposes only and does not constitute Agency policy.

C-82 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

MAXT = 0.01
CINT =0.1
EXP_TIME_ON
EX P_T I ME_0 F F
DAY_CYCLE
B C K_TIME_ON
BCK_TIME_OFF
TIMELIMIT
BW_T0

(g)

o.

17472
84
0.

0.

17472
23

idelay before begin exposure (HOUR)

%TIME EXPOSURE STOP (HOUR)

¦;DELAY BEFORE BACGROUND EXPOSURE (HOUR)
iTIME OF BACKGROUND EXPOSURE STOP (HOUR)
%SIMULATION LIMIT TIME (HOUR)
i Body weight at the beginning of the simulation

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT	=0.02	% exposure dose ug/kg

%MSTOT	=0.1	% exposure dose ug/kg

MSTOT	=1.0	% exposure dose ug/kg

C.2.5.2.4. NTP (1982) (male) (chronic).

%RAT2.m

%clear variable
output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG
% output 0nciout=168 T SUMEXPEVENT

% NTP 1982.

%built and check in September 20, 2009
%protocol: twice weekly gavage for 104 weeks
%Rat Dioxin 3C June09 2clean 2.csl
%MICE_NON_GESTAT_ICF_F083109.csl
%MICE_NON_GESTAT_ICF_F092009.csl (now 09-20-09)

%dose levels: 0.005, 0.025, 0.25 ug/kg/biweekly, ug/kg for 104 weeks
%dose levels: 5, 25, 250 ng/kg/biweekly,ng/kg for 104 weeks
%dose levels equivalent to: 1.4, 7.1, 71 ng/kg/d

MAXT = 0.01





CINT =0.1





EXP TIME ON

= 0.

%delay before begin exposure (HOUR)

EXP TIME OFF

= 17472

%TIME EXPOSURE STOP (HOUR)

DAY CYCLE

= 84



BCK TIME ON

= 0.

%DELAY BEFORE BACGROUND EXPOSURE (HOUR)

BCK TIME OFF

= 0.

%TIME OF BACKGROUND EXPOSURE STOP (HOUR)

TIMELIMIT

= 17472

%SIMULATION LIMIT TIME (HOUR)

BW TO

= 25

% Body weight at the beginning of the simulation

iEXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT	= 0.005	% exposure dose ug/kg

%MSTOT	= 0.025	% exposure dose ug/kg

MSTOT	=0.25	% exposure dose ug/kg

C.2.5.2.5. Smialowicz et al. (2008).

output 0clear

This document is a draft for review purposes only and does not constitute Agency policy.

C-83 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% Smialowicz et al. 2008.

%built and check in August 7 2009

%protocol: oral gavage 5 days/week for 13 weeks
%Mice Dioxin 3C June09 l.csl

%MICE~NON_GESTAT_ICF_F083109.csl (now 09-11-09)

%dose levels: 0, 0.0015, 0.015, 0.15, 0.45 ug/kg

%dose levels: 0, 1.5, 15, 150, 450 nkd (0, 1.07, 10.7, 107, 321 nkd adj)

MAXT	= 0.01

CINT	=0.1

TIMELIMIT	= 2184

EXP_TIME_ON	= 0.

EX P_TIME_0 FF	= 2184

DAY_CYCLE	= 2 4

WEEK_PERIOD	= 168

WEEK_FINISH	= 119

B C K_TIME_ON	= 0.

BCK_TIME_OFF	= 0.

BW_T0	=28

(g)

iSIMULATION LIMIT TIME (HOUR)
idelay before begin exposure (HOUR)
iTIME EXPOSURE STOP (HOUR)

¦;DELAY BEFORE BACGROUND EXPOSURE (HOUR)

iTIME OF BACKGROUND EXPOSURE STOP (HOUR)

i Body weight at the beginning of the simulation

iEXPOSURE DOSE SCENARIOS
%MSTOT	= 0.0015

%MSTOT = 0.015
%MSTOT = 0.150
MSTOT = 0.450

(UG/KG)

% exposure dose (ug/kg)
% exposure dose (ug/kg)
% exposure dose (ug/kg)
% exposure dose (ug/kg)

C.2.5.2.6. Toth et al. (1979) (1 year).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG CBNGKG

% Toth et al. 1979
%built and check in August 7 2009
%protocol: weekly gavage for 1 year
%Mice Dioxin 3C June09 l.csl

%MICE_NON_GESTAT_ICF_F083109.csl (now 09-11-09)

%dose levels: 7, 700, 7000 ng/kg 1/week for 52 weeks (1 year)
%dose levels: 0.007, 0.7, 7 ug/kg 1/week for 52 weeks (1 year)
%dose equivalent: 1, 100, 1000 ng/kg/day

MAXT	= 0.01

CINT	=0.1

TIMELIMIT = 8760
EXP_TIME_ON = 0.
EX P_TIME_0 FF = 8760
DAY_CYCLE = 168
WEEK_PERIOD = 8760
WEEK_FINISH = 8760
B C K_TIME_ON = 0.
BCK TIME OFF = 0.

idelay before begin exposure (HOUR)
i2208 %TIME EXPOSURE STOP (HOUR)

¦;DELAY BEFORE BACGROUND EXPOSURE (HOUR)
iTIME OF BACKGROUND EXPOSURE STOP (HOUR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-84 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

BW_T0

(g)

= 27

% Body weight at the beginning of the simulation

%EXPOSURE DOSE SCENARIOS (UG/KG)

%MSTOT = 0.007	% exposure dose (ug/kg)

%MSTOT =0.7	% exposure dose (ug/kg)

MSTOT =7	% exposure dose (ug/kg)

C.2.5.2.7. White et al (1986).

output 0clear
prepare 0clear

prepare T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CBNDLINGKG

% White et al 1986

%built and check in August 7 2009

%protocol: oral exposure single dose

%dose levels: 0.714, 3.57, 7.14, 35.71, 71.43, 142.86 ng /kg/d ug/kg 1/day
for 14 consecutive days

%dose have been modified following Jeff email on Friday August 21 2009
%dose levels: 10, 50, 100, 500, 1000, 2000 ng /kg/d ug/kg 1/day for 14
consecutive days

%dose levels: 0.010, 0.050, 0.100, 0.500, 1.0, 2.0 ug /kg/d ug/kg 1/day for
14 consecutive days

MAXT

=

o
o

I—1

CINT

=

0.1

TIMELIMIT

=

336

EXP TIME ON

=

0.

EXP TIME OFF

=

336

DAY CYCLE

=

24

WEEK PERIOD

=

336

WEEK FINISH

=

336

BCK TIME ON

=

0.

BCK TIME OFF

=

0.

BW TO

=

23

iTIME AT WHICH EXPOSURE BEGINS (HOUR)
iTIME AT WHICH EXPOSURE ENDS (HOUR)

iTIME AT WHICH BACKGROUND EXPOSURE BEGINS (HOUR)
iTIME AT WHICH BACKGROUND EXPOSURE ENDS (HOUR)
i BODY WEIGHT AT THE BEGINNING OF THE SIMULATION

(G)

iEXPOSURE
%MSTOT
%MSTOT
%MSTOT
%MSTOT
%MSTOT
MSTOT

DOSE SCENARIOS
= 0.010
= 0.050
= 0.100
= 0.500
= 1
= 2	%

(UG/KG)

% EXPOSURE DOSE IN UG/KG
i EXPOSURE DOSE IN UG/KG
i EXPOSURE DOSE IN UG/KG
i EXPOSURE DOSE IN UG/KG
i EXPOSURE DOSE IN UG/KG
EXPOSURE DOSE IN UG/KG

C.2.6. Mouse Gestational Model
C.2.6.1. Model Code

PROGRAM: 'Three Compartment PBPK Model for TCDD in Mice (Gestation)'

! Parameters were change may 16, 2002
! Come from {8MAI_CHR_PRE-EXP_GD}

This document is a draft for review purposes only and does not constitute Agency policy.

C-85 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

! Come from {12 Mouse GDJfile

!{{IMPORTANT-IMPORTANT-IMPORTANT-IMPORTANT}}

! REDUCTION OF MOTHER AND FETUS COMPARTMENT
! 2M_R_TCDD_JULY2 0 02 ////(JULY 18,20 02)////
!TCDD_RED_4Species_2003_4	////(APR 8 ,2003)////

!TCDD_RED_4Species_2003_9	////(APR 17 ,2003)////

!TCDD_RED_4Species_2003_12	////(APR 17 ,2003)////

!APRIL 18 2003

!TCDD_4C_4SP_2 0 03	////(APR 18 ,2003)////

! was ''Gest 4 species l.csl'' but update July 2009

!DevTCDD4Species ICF afterKKfix v3 ratgest.csl
!MICE_GESTATIONAL_ICF_F092309.csl ~
!MICE_GESTATIONAL_ICF_Fl00 609. csl

!Legend/Legend/Legend/Legend/Legend/Legend/Legend/Legend/

!Legend for this PBPK model

IMating: control the tenure of exchange between fetus and
IMother and also control imitated tissue growth
!Ctrl: WTFE, WFO, WPLA0, QPLAF,WTO
!(for rat, mouse, human, and monkey)

!Control transfer from mother to fetus and fetus to mother by TRANSTIME ON
!SWITCH_trans = 0 NO TRANSFER	~

!SWITCH_trans = 1 TRANSFER OCCURS
!Gest off = 1
!Gest on= 0.

! These switches are also controlled by mating parameters
INITIAL !

!SIMULATION PARAMETERS ====
CONSTANT PARA_ZERO	= 1E-30

CONSTANT EXP_TIME_ON	= 288.

CONSTANT EXP_TIME_0FF	= 504

CONSTANT DAY_CYCLE	= 504.

CONSTANT BCK_TIME_ON	=0.0
BEGINS (HOURS)

CONSTANT BCK_TIME_OFF	=0.0
(HOURS)

CONSTANT TRANSTIME_ON	= 144
AT GESTATIONAL DAY 6

! TIME AT WHICH EXPOSURE BEGINS (HOURS)
! TIME AT WHICH EXPOSURE ENDS (HOURS)
! NUMBER OF HOURS BETWEEN DOSES (HOURS)
! TIME AT WHICH BACKGROUND EXPOSURE

! TIME AT WHICH BACKGROUND EXPOSURE ENDS

!CONTROL TRANSFER FROM MOTHER TO FETUS

!UNIT CONVERSION
CONSTANT MW=322 ! MOLECULAR WEIGHT (NG/NMOL)
CONSTANT SERBLO =0.55
CONSTANT UNITCORR = 1000

!INTRAVENOUS SEQUENCY
constant IV_LACK	=0.0

constant IV_PERIOD	=0.0

!PREGNANCY PARAMETER ====

CONSTANT MATTING	=0.0	!BEGINNING OF MATING (HOUR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-86 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CONSTANT N_FETUS	=10	!NUMBER OF FETUS PRESENT

!CONSTANT EXPOSURE CONTROL ===========

!ACUTE, SUBCHRONIC, CHRONIC EXPOSURE =====

!OR BACKGROUND EXPOSURE (IN THIS CASE 3 TIMES A DAY)===

CONSTANT MSTOTBCKGR	=0.0	! ORAL BACKGROUND EXPOSURE DOSE (UG/KG)

CONSTANT MSTOT	=0.0	! ORAL EXPOSURE DOSE (UG/KG)

!ORAL ABSORPTION
MSTOT NM = MSTOT/MW

!CONVERTS THE DOSE TO NMOL/G

! INTRAVENOUS ABSORPTION
CONSTANT DOSEIV	=0.0

DOSEIV_NM = DOSEIV/MW
CONSTANT DOSEIVLATE =0.0

DOSEIVNMlate = DOSEIVLATE/MW

! INJECTED DOSE (UG/KG)

! CONVERTS THE INJECTED DOSE TO NMOL/G

! INJECTED DOSE LATE (UG/KG)

!AMOUNT IN NMOL/G

!INITIAL GUESS OF THE FREE CONCENTRATION IN THE LIGAND (COMPARTMENT
INDICATED BELOW)====

CONSTANT CFLLI0	=0.0 !LIVER (NMOL/ML)

CONSTANT CFLPLA0	=0.0 !PLACENTA (NMOL/ML)

!BINDING CAPACITY	(AhR) FOR NON LINEAR BINDING (COMPARTMENT INDICATED
BELOW) (NMOL/ML) ===

CONSTANT LIBMAX	= 3.5E-4 ! LIVER (NMOL/ML), WANG ET AL. 1997

CONSTANT PLABMAX	= 2.OE-4 !TEMPORARY PARAMETER

! PROTEIN AFFINITY CONSTANTS (1A2 OR AhR, COMPARTMENT INDICATED BELOW)
(NMOL/ML)===

CONSTANT KDLI	= 1.0E-4 !LIVER (AhR) (NMOL/ML), WANG ET AL. 1997

CONSTANT KDLI2	= 4.OE-2 !LIVER (1A2) (NMOL/ML), EMOND ET AL. 2004

CONSTANT KDPLA	= 1.0E-4 !TEMPORARY PARAMETER (AhR)

!EXCRETION AND ABSORPTION	CONSTANT

CONSTANT KST	=	0.3 ! GASTRIC RATE CONSTANT (HR-1)

CONSTANT KABS	=	0.48 !INTESTINAL ABSORPTION CONSTANT (HR-1) ),

WANG ET AL. 1997

! ELIMINATION CONSTANTS

CONSTANT CLURI	=	0.09 ! URINARY CLEARANCE (ML/HR)

!TEST ELIMINATION VARIABLE
constant kelv	=	0.4	! INTERSPECIES VARIABLE ELIMINATION

CONSTANT (1/HOUR)

! CONSTANT TO DIVIDE THE ABSORPTION INTO LYMPHATIC AND PORTAL FRACTIONS
CONSTANT A	=0.7	! LYMPHATIC FRACTION, WANG ET AL. 1997

!PARTITION COEFFICIENTS
CONSTANT PF	= 400

CONSTANT PRE	= 3

CONSTANT PLI	= 6

CONSTANT PPLA	= 3

! ADIPOSE TISSUE/BLOOD

! REST OF THE BODY/BLOOD, WANG ET AL. 2000
! LIVER/BLOOD, WANG ET AL. 19 97
! TEMPORARY PARAMETER NOT CONFIGURED

!PARAMETER FOR INDUCTION OF CYP 1A2, WANG ET AL. 1997 OR OPTIMIZED
CONSTANT PAS INDUC	= 1	! INCLUDE INDUCTION? (1 = YES, 0 = NO)

This document is a draft for review purposes only and does not constitute Agency policy.

C-87 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CONSTANT CYP1A2_10UTZ = 1.6
1A2 (NMOL/ML) (OPTIMIZED)
CONSTANT CYP1A2_1A1
WANG ET AL . (2 000)

CONSTANT CYP1A2_1EC5 0
(NMOL/ML)

CONSTANT CYP1A2 1A2

= 1.5
= 0. 13

= 1.5

(NMOL/ML),WANG ET AL.	(2000)

CONSTANT CYP1A2_1K0UT	=0.1

CONSTANT CYP1A2_1TAU	=1.5
. (2000)

CONSTANT CYP1A2_1EMAX	= 600
(UNITLESS)

CONSTANT HILL	=0.6
BINDING EFFECT CONSTANT (UNITLESS)

! DEGRADATION CONCENTRATION CONSTANT OF

! BASAL CONCENTRATION OF 1A1 (NMOL/ML),

! DISSOCIATION CONSTANT TCDD-CYP1A2

!BASAL CONCENTRATION OF 1A2

! FIRST ORDER RATE OF DEGRADATION (H-l)
!HOLDING TIME (H) (OPTIMIZED), WANG ET AL

! MAXIMUM INDUCTION OVER BASAL EFFECT

!HILL CONSTANT; COOPERATIVELY LIGAND

!DIFFUSIONAL PERMEABILITY FRACTION, WANG ET AL. 1997
CONSTANT PAFF	=0.12 !ADIPOSE (UNITLESS) OPTIMIZED, WANG ET AL.

2000

CONSTANT PAREF	=0.03 !REST OF THE BODY (UNITLESS)

CONSTANT PALIF	=0.35 !LIVER (UNITLESS)

CONSTANT PAPLAF	=0.03 !TEMPORARY PARAMETER NOT CONFIGURED

!FRACTION OF TISSUE WEIGHT =========

CONSTANT WLI0	= 0.0549 !LIVER ILSI (1994)

!TISSUE BLOOD FLOW EXPRESSED AS A FRACTION OF CARDIAC OUTPUT CONSTANT QFF
= 0.070 ! ADIPOSE TISSUE BLOOD FLOW FRACTION (UNITLESS), LEUNG ET AL. 1990
CONSTANT QLIF	= 0.161 !LIVER (UNITLESS), ILSI 1994

!COMPARTMENT TISSUE BLOOD EXPRESSED AS A FRACTION OF THE TOTAL COMPARTMENT
VOLUME

CONSTANT WFB0	= 0.050 IADIPOSE TISSUE, WANG ET AL. 1997

CONSTANT WREB0	= 0.030 !REST OF THE BODY, WANG ET AL. 1997

CONSTANT WLIB0	= 0.266 !LIVER, WANG ET AL. 1997

CONSTANT WPLAB0	= 0.500 !TEMPORARY PARAMETER NOT CONFIGURED

!EXPOSURE SCENARIO FOR UNIQUE OR REPETITIVE WEEKLY OR MONTHLY EXPOSURE

!NUMBER OF EXPOSURES	PER WEEK

CONSTANT WEEK_LACK	=0.0

CONSTANT WEEK_PERIOD	= 168

CONSTANT WEEK FINISH	= 168

!DELAY BEFORE EXPOSURE ENDS (WEEK)
! NUMBER OF HOURS IN THE WEEK (HOURS)
! TIME EXPOSURE ENDS (HOURS)

!NUMBER OF EXPOSURES PER MONTH
CONSTANT MONTH LACK	=0.0

!DELAY BEFORE EXPOSURE BEGINS (MONTH)

!CONSTANT FOR BACKGROUND EXPOSURE=

CONSTANT Day_LACK_BG	=0.0

CONSTANT Day_PERIOD_BG =24

! DELAY BEFORE EXPOSURE BEGINS (HOUR)
!LENGTH OF EXPOSURE (HOUR)

!NUMBER OF EXPOSURES PER WEEK

CONSTANT WEEK_LACK_BG
CONSTANT WEEK_PERIOD_BG
CONSTANT WEEK FINISH BG

0.0
168
168

!DELAY BEFORE BACKGROUD EXPOSURE (WEEK)
! NUMBER OF HOURS IN THE WEEK (HOURS)
!TIME EXPOSURE ENDS (HOURS)

!INITIAL BODY WEIGHT

This document is a draft for review purposes only and does not constitute Agency policy.

C-88 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

CONSTANT BW_T0	=30

CONSTANT RAT10_RATF_MOUSEF = 0.2
GESTATIONAL DAY 22

! WANG ET AL. 1997

!RATIO OF FETUS MOUSE/RAT AT

! FOR RAT (1) AND FOR MOUSE (0.2)

!COMPARTMENT LIPID EXPRESSED AS THE FRACTION OF TOTAL LIPID, POULIN ET AL.
2000

CONSTANT F_TOTLIP
CONSTANT B_TOTLIP
CONSTANT RE_TOTLIP
(UNITLESS)

CONSTANT LI_TOTLIP
CONSTANT PLA_TOTLIP
CONSTANT FETUS TOTLIP

855

0033

019

060
019

= 0.019

! ADIPOSE TISSUE (UNITLESS)

! BLOOD (UNITLESS)

! REST OF THE BODY

! LIVER (UNITLESS)
! PLACENTA (UNITLESS)
! FETUS (UNITLESS)

END

! END OF THE INITIAL SECTION

DYNAMIC ! DYNAMIC SIMULATION SECTION

ALGORITHM

CINTERVAL

MAXTERVAL

MINTERVAL

VARIABLE

CONSTANT

IALG
CINT
MAXT
MINT
T

TIMELIMIT

CINTXY = CINT

2
0.1
1.0e+10
1.0E-10
0.0
313

GEAR METHOD

COMMUNICATION INTERVAL
MAXIMUM CALCULATION INTERVAL
MINIMUM CALCULATION INTERVAL

!SIMULATION LIMIT TIME (HOUR)

PFUNC

= CINT

!TIME CONVERSION

DAY
WEEK
MONTH
YEAR

= T/24
= T/168
= T/730
= T/87 60

TIME IN DAYS
TIME IN WEEKS
TIME IN MONTHS
TIME IN YEARS

DERIVATIVE ! PORTION OF CODE THAT SOLVES DIFFERENTIAL EQUATIONS

!CHRONIC OR SUBCHRONIC EXPOSURE SCENARIO
!NUMBER OF EXPOSURES PER DAY

DAY_LACK
DAY_PERIOD
DAY_FINISH
MONTH_PERIOD
MONTH FINISH

= EXP_TIME_ON
= DAY_CYCLE
= CINTXY
= TIMELIMIT
= EXP TIME OFF

DELAY BEFORE EXPOSURE BEGINS (HOURS)
EXPOSURE PERIOD (HOURS)

LENGTH OF EXPOSURE (HOURS)

EXPOSURE PERIOD (MONTHS)

LENGTH OF EXPOSURE (MONTHS)

!NUMBER OF EXPOSURES PER DAY AND MONTH

DAY_FINISH_BG
MONTH_LACK_BG
(MONTHS)

MONTH_PERIOD_BG
MONTH FINISH BG

= CINTXY
= B C K_TIME_ON

= TIMELIMIT
= BCK TIME OFF

!DELAY BEFORE BACKGROUD EXPOSURE BEGINS

!BACKGROUND EXPOSURE PERIOD (MONTHS)
!LENGTH OF BACKGROUND EXPOSURE (MONTHS)

!INTRAVENOUS LATE
IV_FINISH = CINTXY

B = 1-A ! FRACTION OF DIOXIN ABSORBED IN THE PORTAL FRACTION OF THE LIVER

This document is a draft for review purposes only and does not constitute Agency policy.

C-89 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

!FETUS,VOLUME,FETUS,VOLUME,FETUS,VOLUME,FETUS,VOLUME,FETUS,VOLUME,FETUS,VOLUM
E

! FROM OFLAHERTY_19 92

RTESTGEST= T-MATTING
TESTGEST=DIM(RTESTGEST,0.0)

WTFER_RODENT= (2.3d-3*EXP(1.49d-2*(TESTGEST))+1.3d-2)*Gest_on
WTFER = (WTFER_RODENT*RATIO_RATF_MOUSEF*N_FETUS)

WTFE = DIM(WTFER,0.0)

!

FAT,VOLUME,FAT,VOLUME,FAT,VOLUME,FAT,VOLUME,FAT,VOLUME,FAT,VOLUME,FAT,VOLUME
! FAT GROWTH EXPRESSION LINEAR DURING PREGNANCY
! FROM O'FLAHERTY_19 92

WF0= (((9.66d-5*(TESTGEST))*gest_on)+0.069)

! PLACENTA,VOLUME, PLACENTA,VOLUME, PLACENTA,VOLUME, PLACENTA,VOLUME
! WPLA PLACENTA GROWTH EXPRESSION, SINGLE EXPONENTIAL WITH OFFSET
! FROM O'FLAHERTY_19 92 ! FOR EACH PUP

WPLA0N_RODENT = (0.6/(1+(5d+3*EXP(-0.0225*(TESTGEST)))))*N_FETUS
WPLA0R = (WPLA0N_RODENT/WT0)*Gest_on
WPLA0 = DIM(WPLA0R,0.0)

! PLACENTA,FLOW RATE, PLACENTA,FLOW RATE, PLACENTA,FLOW RATE, PLACENTA,FLOW
RATE

! QPLA PLACENTA GROWTH EXPRESSION, DOUBLE EXPONENTIAL WITH OFFSET
! FROM O'FLAHERTY_19 92

QPLARF = (1.67d-7 *exp(9.6d-3*(TESTGEST)) &

+1.6d-3*exp(7.9d-3*(TESTGEST))+0.0)*Gest_on*SWITCH_trans
QPLAF=DIM(QPLARF,0.0)	!FRACTION OF FLOW RATE IN PLACENTA

! GESTATION CONTROL
IF (T.LT.MATTING) THEN
Gest off = 1
Gest on= 0 . 0
ELSE ~

Gest off = 0.0
Gest on =1
END IF ~

! MOTHER BODY WEIGHT GROWTH EQUATION========

! MODIFICATION TO ADAPT THIS MODEL AT HUMAN MODEL

! BECAUSE LINEAR DESCRIPTION IS NOT GOOD ENOUGH FOR MOTHER GROWTH
! MOTHER BODY WEIGHT GROWTH

PARAMETER (BW_RMN = 1.0E-30)

WT0= BW_T0 *(1.0+(0.41*T)/(1402.5+T+BW_RMN))

! VARIABILITY OF REST OF THE BODY DEPENDS ON OTHER ORGANS
WRE0 = (0.91 - (WLIB0*WLI0 + WFB0*WF0 +WPLAB0*WPLA0 + WLI0 + WF0 +
WPLA0))/(1.0+WREB0) ! REST OF THE BODY FRACTION; UPDATED FOR EPA ASSESSMENT

This document is a draft for review purposes only and does not constitute Agency policy.

C-90 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

QREF = 1.0-(QFF+QLIF+QPLAF)
(ML/HR)

QTTQF = QFF+QREF+QLIF+QPLAF

!REST OF BODY BLOOD FLOW RATE
! SUM MUST EQUAL 1

! COMPARTMENT VOLUME (ML OR G)
WF = WF0 * WTO
WRE = WRE0 * WTO
WLI = WLI0 * WTO
WPLA= WPLAO* WTO

ADIPOSE TISSUE
REST OF THE BODY
LIVER
PLACENTA

! COMPARTMENT TISSUE BLOOD	(ML OR G)

WFB = WFBO * WF	!

WREB = WREBO * WRE	!

WLIB = WLIBO * WLI	!

WPLAB = WPLABO* WPLA	!

ADIPOSE TISSUE
REST OF THE BODY
LIVER
PLACANTA

! CARDIAC OUTPUT FOR THE GIVEN BODY WEIGHT
!QC= QCCAR* 60*(WTO/1000.0)**0.75
CONSTANT QCC=16500	! EQUIVALENT TO 275 * 60

QC= QCC*(WTO/UNITCORR)**0.75

!COMPARTMENT BLOOD FLOW RATE (ML/HR)

QF = QFF*QC

QLI = QLIF*QC

QRE = QREF*QC

QPLA = QPLAF*QC

QTTQ = QF+QRE+QLI+QPLA

!ADIPOSE TISSUE BLOOD FLOW RATE
!LIVER TISSUE BLOOD FLOW RATE
!REST OF THE BODY BLOOD FLOW RATE
!PLACENTA TISSUE BLOOD FLOW RATE
!TOTAL FLOW RATE

!PERMEABILITY ORGAN FLOW (ML/HR)
PAF = PAFF*QF
PARE = PAREF*QRE
PALI = PALIF*QLI
PAPLA = PAPLAF*QPLA

ADIPOSE TISSUE
REST OF THE BODY
LIVER TISSUE
PLACENTA

ABSORPTION SECTION
ORAL,

INTRAPERITONEAL,
INTRAVENOUS

!REPETITIVE ORAL BACKGROUND EXPOSURE SCENARIO

MSTOT_NMBCKGR = MSTOTBCKGR/322	!AMOUNT IN NMOL/G

MSTTBCKGR =MSTOT_NMBCKGR *WT0

DAY_EX P O S U RE_B G = PULSE(DAY_LACK_BG,DAY_PERIOD_BG,DAY_FINISH_BG)
WEEK_EXPOSURE_BG = PULSE(WEEK_LACK_BG,WEEK_PERIOD_BG,WEEK_FINISH_BG)
MONTH_EXPOSURE_BG = PULSE(MONTH_LACK_BG,MONTH_PERIOD_BG,MONTH_FINISH_BG)

MSTTCH_BG = (DAY_EXPOSURE_BG*WEEK_EXPOSURE_BG*MONTH_EXPOSURE_BG)*MSTTBCKGR
MSTTFR_BG = MSTTBCKGR/CINT

CYCLE_BG =DAY_EXPOSURE_BG*WEEK_EXPOSURE_BG*MONTH_EXPOSURE_BG

! CONDITIONAL ORAL EXPOSURE (BACKGROUND EXPOSURE)

This document is a draft for review purposes only and does not constitute Agency policy.

C-91 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

IF (MSTTCH_BG.EQ.MSTTBCKGR) THEN
ABSMSTT_GB= MSTTFR_BG

ELSE

ABSMSTT_GB =0.0
END IF

CYCLETOTBG=INTEG(CYCLE_BG,0.0)

!REPETITIVE ORAL EXPOSURE SCENARIO

MSTT= MSTOT_NM * WTO	!AMOUNT IN NMOL

DAY_EXPOSURE = PULSE(DAY_LACK,DAY_PERIOD,DAY_FINISH)

WEEK_EXPOSURE = PULSE(WEEK_LACK,WEEK_PERIOD,WEEK_FINISH)
MONTH_EXPOSURE = PULSE(MONTH_LACK,MONTH_PERIOD,MONTH_FINISH)

MSTTCH = (DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE)*MSTT
MSTTFR = MSTT/CINT

CYCLE = DAY_EXPOSURE*WEEK_EXPOSURE*MONTH_EXPOSURE

SUMEXPEVENT= INTEG (CYCLE,0.0)/cint !NUMBER OF CYCLES GENERATED DURING
SIMULATION

! CONDITIONAL ORAL EXPOSURE
IF (MSTTCH.EQ.MSTT) THEN

ABSMSTT= MSTTFR
ELSE

ABSMSTT =0.0
END IF

CYCLETOT=INTEG(CYCLE, 0.0)

! MASS CHANGE IN THE LUMEN
RMSTT= -(KST+KABS)*MST +ABSMSTT +ABSMSTT_GB ! RATE OF CHANGE (NMOL/H)

MST = INTEG(RMSTT,0.0)	!AMOUNT REMAINING IN DUODENUM

(NMOL)

! ABSORPTION IN LYMPH CIRCULATION
LYRMLUM = KABS*MST*A
LYMLUM = INTEG(LYRMLUM,0.0)

! ABSORPTION IN PORTAL CIRCULATION
LIRMLUM = KABS*MST*B
LIMLUM = INTEG(LIRMLUM,0.0)

	IV EXPOSURE 	

IV= DOSEIV_NM * WTO !AMOUNT IN NMOL
IVR= IV/PFUNC ! RATE FOR IV INFUSION IN BLOOD
EXPIV= IVR * (1.0-STEP(PFUNC))

IVDOSE = integ(EXPIV,0.0)

!	IV late in the cycle

This document is a draft for review purposes only and does not constitute Agency policy.

C-92 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

! MODIFICATION ON January 13 2004
IV_RlateR = DOSEIVNMlate*WTO

IV_EXPOSURE=PULSE(IV_LACK,IV_PERIOD,IV_FINISH)

IV_lateT = IV_EXPOSURE *IV_RlateR
IV_late = IV_lateT/CINT

SUMEXPEVENTIV= integ (IV_EXPOSURE,0.0) !NUMBER OF CYCLE GENERATE DURING
SIMULATION

!SYSTEMIC CONCENTRATION OF TCDD
! MODIFICATION ON OCTOBER 6, 2009
CB=(QF*CFB+QRE*CREB+QLI*CLIB+EXPIV+LYRMLUM+QPLA*CPLAB+IV_late)/(QC+CLURI) !
CA = CB ! CONCENTRATION (NMOL/ML)

!URINARY EXCRETION BY KIDNEY
!MODIFICATION ON OCTOBER 6, 2009
RAURI = CLURI *CB

AURI = INTEG(RAURI,0.0)

!UNIT CONVERSION POST SIMULATION
CBSNGKGLIADJ=(CB*MW*UNITCORR*(l/B_TOTLIP)*(1/SERBLO))![NG of TCDD Serum/Kg
OF LIPIP]

AUCBS_NGKGLIADJ=integ(CBSNGKGLIADJ,0.0)

PRCT_B = (CB/(MSTT+1E-30))*100 ! PERCENT OF ORAL DOSE IN BLOOD
PRCT_BIV = (CB/(IV_RlateR+lE-30))*100 ! PERCENT OF IV DOSE IN BLOOD
CBNGKG= CB*MW*UNITCORR
CBNGG = CB*MW

!ADIPOSE COMPARTMENT
! TISSUE BLOOD COMPARTMENT
RAFB= QF*(CA-CFB)-PAF*(CFB-CF/PF)
AFB = INTEG(RAFB,0.0)
CFB = AFB/WFB
! TISSUE COMPARTMENT
RAF = PAF*(CFB-CF/PF)
AF = INTEG(RAF,0.0)

CF = AF/WF

!(NMOL/H)
!(NMOL)
!(NMOL/ML)

!(NMOL/H)
!(NMOL)

(NMOL/ML)

!UNIT CONVERSION POST SIMULATION
CFTOTAL= (AF + AFB)/(WF + WFB) ! TOTAL CONCENTRATION IN NMOL/ML
CFTFREE = CFB + CF !TOTAL FREE CONCENTRATION IN FAT (NM/ML)

PRCT_F = (CFTOTAL/(MSTT+1E-30))*100 ! PERCENT OF ORAL DOSE IN FAT
PRCT_FIV = (CFTOTAL/(IV_RlateR+lE-30))*100 ! PERCENT OF IV DOSE IN FAT
CFNGKG=CFTOTAL*MW*UNITCORR ! FAT CONCENTRATION IN NG/KG

AUCF_NGKGH=integ(CFNGKG, 0.0)

CFNGG = CFTOTAL*MW

!REST OF THE BODY COMPARTMENT
RAREB= QRE *(CA-CREB)-PARE*(CREB-CRE/PRE)
AREB = INTEG(RAREB,0.0)

CREB = AREB/WREB
!TISSUE COMPARTMENT
RARE = PARE*(CREB - CRE/PRE)
ARE = INTEG(RARE,0.0)

(NMOL/H)
!(NMOL)
!(NMOL/H)

(NMOL/H)
(NMOL)

This document is a draft for review purposes only and does not constitute Agency policy.

C-93 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

CRE = ARE/WRE

!(NMOL/ML)

!UNIT CONVERSION POST SIMULATION
CRETOTAL= (ARE + AREB)/(WRE + WREB)	! TOTAL CONCENTRATION IN

NMOL/ML

PRCT_RE = (CRETOTAL/(MSTT+1E-30))*100 ! PERCENT OF ORAL DOSE IN REST OF
BODY

PRCT_REIV = (CRETOTAL/(IV_RlateR+lE-30))*100 ![ PERCENT OF IV DOSE IN
REST OF THE BODY ]

CRENGKG=CRETOTAL*MW*UNITCORR ! REST OF THE BODY CONCENTRATION IN NG/KG

!LIVER COMPARTMENT
!TISSUE BLOOD COMPARTMENT
RALIB = QLI*(CA-CLIB)-PALI*(CLIB-CFLLIR)+LIRMLUM !

ALIB = INTEG(RALIB,0.0)	!(NMOL)

CLIB = ALIB/WLIB	!(NMOL/ML)

!TISSUE COMPARTMENT
RALI = PALI*(CLIB - CFLLIR)-REXCLI	! (NMOL/HR)

ALI = INTEG(RALI,0.0)	!(NMOL)

CLI = ALI/WLI	!(NMOL/ML)

!FREE TCDD IN LIVER COMPARTMENT
PARAMETER (LIVER_1RMN = 1.0E-30)

CFLLI= IMPLC(CLI-(CFLLIR*PLI+(LIBMAX*CFLLIR/(KDLI+CFLLIR &
+LIVER_1RMN))+((CYP1A2_103*CFLLIR/(KDLI2 + CFLLIR &
+LIVER_1RMN)*PAS_INDUC)))-CFLLI,CFLLI0)

CFLLIR=DIM(CFLLI,0.0) ! FREE CONCENTRATION IN LIVER

CBNDLI= LIBMAX*CFLLIR/(KDLI+CFLLIR+LIVER_1RMN) !BOUND CONCENTRATION

!VARIABLE ELIMINATION BASED ON THE CYP1A2
KBILE_LI_T =((CYP1A2_10UT-CYP1A2_1A2)/CYP1A2_1A2)*Kelv ! INDUCED BILIARY
EXCRETION RATE CONSTANT

REXCLI = KBILE_LI_T*CFLLIR*WLI ! DOSE-DEPENDENT EXCRETION RATE
EXCLI = INTEG(REXCLI,0.0)

!UNIT CONVERSION POST SIMULATION

CLITOTAL= (ALI + ALIB)/(WLI + WLIB) ! TOTAL CONCENTRATION IN NMOL/ML
PRCT_LI = (CLITOTAL/(MSTT+1E-30))*100 ! PERCENT ORAL DOSE IN LIVER
PRCT_LIIV = (CLITOTAL/(IV_RlateR+lE-30))*100 ! PERCENT IV DOSE IN LIVER
Rec_occ= CFLLIR/(KDLI+CFLLIR)

CLINGKG=CLITOTAL*MW*UNITCORR ! LIVER CONCENTRATION IN NG/KG

AUCLI_NGKGH=INTEG(CLINGKG, 0.0)

CBNDLINGKG = CBNDLI*MW*UNITCORR

AUCBNDLI_NGKGH =INTEG(CBNDLINGKG,0.0)

CLINGG = CLITOTAL*MW

!CHEMICAL IN CYP450 (1A2) COMPARTMENT
CYP1A2_1KINP = CYP1A2_1K0UT* CYP1A2_10UTZ ! BASAL RATE OF CYP1A2 PRODUCTION
SET EQUAL TO BASAL RATE OF DEGREDATION

! MODIFICATION ON OCTOBER 6, 2009
CYP1A2_10UT =INTEG(CYP1A2_1KINP * (1.0 + CYP1A2_1EMAX *(CBNDLI+1.0e-30)**HILL
&

/(CYP1A2_1EC50**HILL + (CBNDLI+1.0e-30)**HILL)) &

This document is a draft for review purposes only and does not constitute Agency policy.

C-94 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

- CYP1A2_1K0UT*CYP1A2_10UT, CYP1A2_10UTZ)

! EQUATIONS INCORPORATING DELAY OF CYP1A2 PRODUCTION (NOT USED IN
SIMULATIONS)

CYP1A2_1R02 = (CYP1A2_10UT - CYP1A2_102)/ CYP1A2_1TAU
CYP1A2_102 =INTEG(CYP1A2_1R02, CYP1A2_1A1)

CYP1A2_1R03 = (CYP1A2_102 - CYP1A2_103)/ CYP1A2_1TAU
CYP1A2_103 =INTEG(CYP1A2_1R03, CYP1A2_1A2)

! TRANSFER OF DIOXIN FROM PLACENTA TO FETUS
! FETAL EXPOSURE ONLY DURING EXPOSURE

IF (T.LT.TRANSTIME_ON) THEN

SWITCH_trans =0.0
ELSE

SWITCH_trans = 1
END IF

!TRANSFER OF DIOXIN FROM PLACENTA TO FETUS
! MODIFICATION 26 SEPTEMBER 2003

CONSTANT PFETUS= 4 !

CONSTANT CLPLA_FET =0.17 !

RAMPF = (CLPLA_FET*CPLA) *SWITCH_trans
AMPF=INTEG(RAMPF,0.0)

!TRANSFER OF DIOXIN FROM FETUS TO PLACENTA
RAFPM = (CLPLA_FET*CFETUS_v)* SWITCH_trans !

AFPM = INTEG(RAFPM,0.0)

! TCDD IN PLACENTA MOTHER COMPARTMENT

RAPLAB= QPLA*(CA - CPLAB) - PAPLA* (CPLAB -CFLPLAR) ! NMOL/H)

APLAB = INTEG(RAPLAB,0.0)	! (NMOL)

CPLAB = APLAB/(WPLAB+1E-30)	! (NMOL/ML)

RAPLA = PAPLA*(CPLAB-CFLPLAR)-RAMPF + RAFPM	! (NMOL/H)

APLA = INTEG(RAPLA,0.0)	! (NMOL)

CPLA = APLA/(WPLA+le-30)	! (NMOL/ML)

PARAMETER (PARA_ZERO = 1.0E-30)

CFLPLA= IMPLC(CPLA-(CFLPLAR*PPLA +(PLABMAX*CFLPLAR/(KDPLA&

+CFLPLAR+PARA_ZERO)))-CFLPLA,CFLPLA0)

CFLPLAR=DIM(CFLPLA,0.0)

!UNIT CONVERSION POST SIMULATION
CPLATOTAL= (APLA + APLAB)/((WPLA + WPLAB)+le-30)! TOTAL CONCENTRATION IN
NMOL/ML

PRCT_PLA = (CPLATOTAL/(MSTT+1E-30))*100
PRCT_PLAIV = (CPLATOTAL/(IV_RlateR+lE-30))*100
CPLANGG = CPLATOTAL*MW

!FETUS COMPARTMENT
RAFETUS= RAMPF-RAFPM
AFETUS=INTEG(RAFETUS,0.0)

This document is a draft for review purposes only and does not constitute Agency policy.

C-95 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

CFETUS=AFETUS/(WTFE+1E-30)

CFETOTAL= CFETUS
CFETUS_v = CFETUS/PFETUS

! UNIT CONVERSION POST SIMULATION
CFETUSNGKG = CFETUS*MW*UNITCORR	!(NG/KG)

AU C_FEN GKGH = INTEG(CFETUSNGKG,0.0)

PRCT_FE = (CFETOTAL/(MSTT+1E-30))*100
PRCT_FEIV = (CFETOTAL/(IV_RlateR+lE-30))*100
CFETUSNGG = CFETOTAL*MW

! 	CONTROL MASS BALANCE	

BDOSE= IVDOSE +LYMLUM+LIMLUM

BMASSE = EXCLI+AURI+AFB+AF+AREB+ARE+ALIB+ALI+APLA+APLAB+AFETUS
BDIFF = BDOSE-BMASSE

!BODY BURDEN (NG)

BODY_BURDEN = AFB +AF+ARE B +ARE +ALIB+ALI+AP LA+AP LAB !

BBFETUSNG	= AFETUS*MW*UNITCORR ! NG

! BODY BURDEN IN TERMS OF CONCENTRATION (NG/KG)

BBNGKG =( ( (AFB+AF+AREB+ARE+ALIB+ALI+AP LA+AP LAB)/WTO)*MW*UNITCORR) !
AUC BBNGKGH=INTEG(BBNGKG,0.0)

! 	COMMAND OF THE END OF SIMULATION	

TERMT (T.GE. TimeLimit, 'Time limit has been reached.')

END ! END OF THE DERIVATIVE SECTION
END ! END OF THE DYNAMIC SECTION
END ! END OF THE PROGRAM

C.2.6.2. Input Files
C.2.6.2.1. Keller et al. (2007).

%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC CBNGKGH

% output 0nciout=lO T SUMEXPEVENT wtO
%Keller et al. 2007

%protocol: single oral dose at GD13
%DevTCDD4Species.csl
%MICE_GESTATIONAL_ICF_F092309.csl
%dose levels: 0.01, 0.100 1 ug/kg at GD13
%dose levels: 10, 100 1000 ng/kg at GD13

•;EXPOSURES SCENARIOS
MAXT=0 . 01
CINT =0.1

EXP_TIME_ON	= 312.

EX P_TIME_0 FF	= 336

DAY_CYCLE	=24

BCK TIME ON	=0.

delay before begin exposure (HOUR)

TIME EXPOSURE STOP (HOUR)

DELAY BEFORE BACGROUND EXPOSURE (HOUR)

This document is a draft for review purposes only and does not constitute Agency policy.

C-96 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

BCK_TIME_OFF	= 0.

IV_LACK	= 5 05

IV_PERIOD	= 505

TIMELIMIT	= 336

BW_T0	= 24

MATTING	= 0.

TRANSTIME_ON	= 144.
HOURS)

N FETUS	=10

%EXPOSURE DOSE SCENARIOS (UG/KG)

TIME OF BACKGROUND EXPOSURE STOP (HOUR)

SIMULATION LIMIT TIME (HOUR)

BEGINNING MATTING (HOUR)

SHOULD BE MATTING TIME + 6 DAYS(144

%MSTOT
%MSTOT
MSTOT

= 0. 01
= 0.1
1

% ORAL EXPOSURE DOSE (UG/KG)
% ORAL EXPOSURE DOSE (UG/KG)
ORAL EXPOSURE DOSE (UG/KG)

C.2.6.2.2. Li et al. (2006).

%TO BE USED AFTER THE
%clear variable
output 0clear

prepare 0clear T CLINGKG CFNGKG CBSNGKGLIADJ BBNGKG CFETUSNGKG AUCLI_NGKGH
AUCF_NGKGH AUCBS_NGKGLIADJ AUC_BBNGKGH AUC_FENGKGH CBNDLINGKG AUCBNDLI_NGKGH
CBNGKG AUC_CBNGKGH
% output 0nciout=lO T SUMEXPEVENT
%Li et al.2006

%protocol: daily oral dose from GDI to GD3
%DevTCDD4Species.csl
%MICE_GESTATIONAL_ICF_F092309.csl

%dose levels: 0.002, 0.050, 0.10 ug/kg/day at GDI to GD3
%dose levels: 2, 50, 100 ng/kg/day from GDI to GD3

•;EXPOSURES SCENARIOS
MAXT=0 . 01
CINT =0.1
EXP_TIME_ON
EXP TIME OFF

= 0.
= 72

% delay before begin exposure (HOUR)
_ _	% TIME EXPOSURE STOP (HOUR) 2 HOURS LESS THAN

GD3 put 70 to be sure 3 doses will be administrate

% BECAUSE i STARTED TIME 0 FOR GDI

DAY_CYCLE
B C K_TIME_ON
BCK_TIME_OFF
IV_LACK
IV_PERIOD
TIMELIMIT
days
BW_T0
MATTING
TRANSTIME_ON
HOURS)

N FETUS

= 24
= 0.
= 0.
= 505
= 505
= 72.

= 27
= 0.
= 144.

= 10

DELAY BEFORE BACGROUND EXPOSURE (HOUR)
TIME OF BACKGROUND EXPOSURE STOP (HOUR)

SIMULATION LIMIT TIME (HOUR) Run for 3

BEGINNING MATTING (HOUR)

SHOULD BE MATTING TIME + 6 DAYS(144

iEXPOSURE DOSE SCENARIOS (UG/KG)

iMSTOT
iMSTOT

= 0.002
0. 05

% ORAL EXPOSURE DOSE (UG/KG)
ORAL EXPOSURE DOSE (UG/KG)

This document is a draft for review purposes only and does not constitute Agency policy.

C-97 DRAFT—DO NOT CITE OR QUOTE

-------
1	MSTOT	=0.10	% ORAL EXPOSURE DOSE (UG/KG)

2

3	C.3. TOXICOKINETIC MODELING RESULTS FOR KEY ANIMAL BIOASSAY

4	STUDIES

5	The simulated TCDD serum-adjusted lipid concentrations reported in this appendix for

6	the rodent bioassays were converted to TCDD concentrations in rodent whole blood. Initially,

7	EPA multiplied the serum-adjusted lipid concentrations by 0.0033, the ratio of lipid content to

8	total serum volume, then by 0.55, the value of the hematocrit. This product yields the TCDD

9	concentration in whole rodent blood as predicted by the PBPK model. EPA assumed that the

10	same whole blood TCDD concentration would result in the same effects in humans and rodents.

11	This conversion accomplishes the following:

12	1. Allows the human equivalent dose (HED) to be based on equivalent blood concentration

13	(that represents serum plus erythrocyte TCDD), which is proportional to tissue exposure;

14	2. Avoids criticism that the total blood concentration is normalized to serum lipid alone in

15	an unbalanced way (thus EPA does not contradict Centers for Disease Control and

16	Prevention (CDC) data or methods);

17	3. Factors out any impact of the lipid content used in the PBPK model; and

18	4. TCDD concentration in whole blood is encouraged for use in the assessments by the NAS

19	(NAS, 2006, p. 43); see additional information in Section 3.3.

20

21	C.3.1. Nongestational Studies

22	C.3.1.1. Cantoni et al. (1981)

Type:

Rat

Dose:

10, 100, 1000 ng/kg/week

Strain:

CD-COBS rats

Route:

Oral gavage exposure

Body weight:

BW set to 125g

Regime:

1 dose/week for 45 weeks

Sex:

Female

Simulation

7,560 hours





time:

(45 weeks)

23

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.43

Emond

1.85

^ 70 7.^02 hours')

1 82

CADM

-

-

-

14.29

Emond

8.84

26.6 (@ 7,392 hours)

7.97

This document is a draft for review purposes only and does not constitute Agency policy.

C-98 DRAFT—DO NOT CITE OR QUOTE

-------


CADM

-

-

-

142.86

Emond

50.0

227 (@ 7,392 hours)

41.9

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.43

Emond

247

328 (@ 7,398 hours)

242

CADM

374

431

431

14.29

Emond

2,176

2,860 (@ 7,231 hours)

1,928

CADM

3,884

4,330

4,330

142.86

Emond

20,500

26,978 (@ 7,399 hours)

17,255

CADM

39,067

43,329

43,329

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.43

Emond

175

200 (@ 7,431 hours)

181

CADM

250

280

244

14.29

Emond

837

937 (@ 7,427 hours)

807

CADM

1,209

1,352

1,167

142.86

Emond

4,741

5,374 (@ 7,424 hours)

4,349

CADM

10,050

11,224

9,734

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.43

Emond

26.1

31.7 (@ 7,398 hours)

26.3

CADM

32.0

35.0

35.0

14.29

Emond

170

210 (@ 7,230 hours)

156

CADM

225

243

243

142.86

Emond

1,337

1,695 (@ 7,398 hours)

1,151

CADM

2,106

2,266

2,266

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-99 DRAFT—DO NOT CITE OR QUOTE

-------
BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.43

Emond

6.04

7.76 (@. 7,396 hours)

6.01

CADM

-

-

-

14.29

Emond

23.7

29.1 (@. 7,228 hours)

22.2

CADM

-

-

-

142.86

Emond

66.8

80.0 ( a 1 hours)

63.4

CADM

-

-

-

1

2

3	C.3.1.2. Chuetal. (2007)

Type:

Rat

Dose:

2.5, 25, 250, and 1,000 ng/kg-day

Strain:

Sprague-Dawley

Route:

Oral exposure

Body weight:

200 g

Regime:

1 dose per day for 28 days

Sex:

Female

Simulation time:

672 hours

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.5

Emond

1.26

2.35 (@. 648 hours)

1.88

CADM

-

-

-

25

Emond

7.66

15.3 (@, 648 hours)

10.4

CADM

-

-

-

250

Emond

48.8

113 (@, 648 hours)

63.7

CADM

-

-

-

1,000

Emond

169

418 (@, 648 hours)

222

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.5

Emond

148

268 ( a 652 hours)

255

CADM

-

-

-

25

Emond

1,777

2,953 (@. 653 hours)

2,806

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-100 DRAFT—DO NOT CITE OR QUOTE

-------
250

Emond

19,232

30,262 (@. 653 hours)

28,668

CADM

-

-

-

1,000

Emond

77,819

120,400 (@. 653 hours)

113,890

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.5

Emond

108

180 ( a 668 hours)

180

CADM

-

-

-

25

Emond

660

1,020 (@. 659 hours)

1,015

CADM

-

-

-

250

Emond

4,210

6,433 (@. 655 hours)

6,354

CADM

-

-

-

1,000

Emond

14,576

22,610 (@. 655 hours)

22,280

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.5

Emond

16.1

27.5 (@. 652 hours)

26.9

CADM

-

-

-

25

Emond

138

222 ( a 652 hours)

214

CADM

-

-

-

250

Emond

1,239

1,935 (@. 652 hours)

1,842

CADM

-

-

-

1,000

Emond

4,801

7,444 (@. 652 hours)

7,067

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.5

Emond

4.15

6.51 (@, 652 hours)

6.21

CADM

-

-

-

25

Emond

20.5

28.5 (@. 652 hours)

27.4

CADM

-

-

-

250

Emond

63.3

76.0 (@. 652 hours)

74.7

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-101 DRAFT—DO NOT CITE OR QUOTE

-------
1,000

Emond

90.2

99.0 (@. 653 hours)

98.3

CADM

-

-

-

1 C.3.1.3. Crofton et ah (2005)

Type:

Rats

Dose:

0, 0.1, 3, 10, 30, 100, 300, 1000, 3000,
and 10,000 ng/kg-day

Strain:

Long Evans

Route:

Oral exposure

Body weight:

4 weeks old
BW set to 190 g

Regime:

One dose per day for four days

Sex:

Female

Simulation time:

96 hours

2	The CADM model was not ran because the dosing duration is lower than the resolution of the model (1 week)

3

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.1

Emond

0.0202

0.041 (@. 72 hours)

0.0244

CADM

-

-

-

3

Emond

0.488

1.10 ( a 72 hours)

0.582

CADM

-

-

-

10

Emond

1.38

3.40 ( a 72 hours)

1.62

CADM

-

-

-

30

Emond

3.46

9.44 72 hours)

3.93

CADM

-

-

-

100

Emond

9.26

29.0 72 hours)

10.2

CADM

-

-

-

300

Emond

23.1

81.8 ( a 72 hours)

24.5

CADM

-

-

-

1000

Emond

65.7

260 ((ai 72 hours)

68.2

CADM

-

-

-

3000

Emond

181

764 ( a 72 hours)

187

CADM

-

-

-

10,000

Emond

583

2,527 (@. 72 hours)

607

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-102 DRAFT—DO NOT CITE OR QUOTE

-------
LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.1

Emond

0.919

1.55 ((ci\ 75 hours)

1.18

CADM

-

-

-

3

Emond

37.4

62.6 ( a 76 hours)

53.3

CADM

-

-

-

10

Emond

145

242 11 hours)

214

CADM

-

-

-

30

Emond

494

818 ((ai 78 hours)

742

CADM

-

-

-

100

Emond

1,839

3,025 78 hours)

2,793

CADM

-

-

-

300

Emond

5,925

9,692 78 hours)

9,028

CADM

-

-

-

1000

Emond

20,717

33,738 79 hours)

31,564

CADM

-

-

-

3000

Emond

63,511

103,140 79 hours)

96,545

CADM

-

-

-

10,000

Emond

212,890

344,910 79 hours)

321,960

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.1

Emond

1.00

1.93 ( a 96 hours)

1.93

CADM

-

-

-

3

Emond

24.6

45.9 ((ai 96 hours)

45.9

CADM

-

-

-

10

Emond

70.3

129 ((ci\ 96 hours)

129

CADM

-

-

-

30

Emond

177

317 ( a 96 hours)

317

CADM

-

-

-

100

Emond

480

838 ( a 96 hours)

838

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-103 DRAFT—DO NOT CITE OR QUOTE

-------
300

Emond

1,206

2,065 (@ 96 hours)

2,065

CADM

-

-

-

1000

Emond

3,452

5,836 (@ 96 hours)

5,836

CADM

-

-

-

3000

Emond

9,522

16,050 (@ 96 hours)

16,050

CADM

-

-

-

10,000

Emond

30,657

51,918 (@ 96 hours)

51,918

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.1

Emond

0.138

0.224 (@ 79 hours)

0.223

CADM

-

-

-

3

Emond

4.04

6.56 ((ai 78 hours)

6.44

CADM

-

-

-

10

Emond

13.3

21.5 (@ 78 hours)

21.0

CADM

-

-

-

30

Emond

39.3

63.5 ((ai 78 hours)

61.5

CADM

-

-

-

100

Emond

129

208 ( a 78 hours)

200

CADM

-

-

-

300

Emond

384

618 ( a 11 hours)

590

CADM

-

-

-

1000

Emond

1,270

2,041 (@ 77 hours)

1,942

CADM

-

-

-

3000

Emond

3,793

6,094 (@ 77 hours)

5,784

CADM

-

-

-

10,000

Emond

12,595

20,226 (@ 77 hours)

19,154

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.1

Emond

0

0.115 (@ 75 hours)

0

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-104 DRAFT—DO NOT CITE OR QUOTE

-------
3

Emond

2

2.47 (@. 76 hours)

2

CADM

-

-

-

10

Emond

4

6.42 ( a 76 hours)

5

CADM

-

-

-

30

Emond

10

14.1 ( a 76 hours)

12

CADM

-

-

-

100

Emond

22

29.9 (@. 76 hours)

27

CADM

-

-

-

300

Emond

41

51.9 ( a 77 hours)

49

CADM

-

-

-

1000

Emond

68

80.2 (@, 1 hours)

77

CADM

-

-

-

3000

Emond

90

98.6 ((ci\ 1 hours)

96

CADM

-

-

-

10,000

Emond

104

108 ( a 1 hours)

107

CADM

-

-

-

1

2

3	C.3.1.4. Delia Porta et a I. (2001) (female)

Type:

Mouse

Dose:

2,500 and 5,000 ng/kg-week (equivalent
to 357 and 714 ng/kg-day)

Strain:

B6C3

Route:

Gavage

Body weight:

6 weeks old (BW
20g)

Regime:

Once a week for 52 weeks

Sex:

Female

Simulation time:

8,736 hours

4	The C ADM model was not ran because the study duration is longer than the allowed model duration

5

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

357

Emond

67.0

741 (@. 8,568 hours)

46.8

CADM

-

-

-

714

Emond

37.6

374 (@. 8,568 hours)

27.2

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-105 DRAFT—DO NOT CITE OR QUOTE

-------
LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

357

Emond

50,269

70,070 (@. 8,577 hours)

37,389

CADM

-

-

-

714

Emond

25,422

35,352 (@. 8,577 hours)

19,105

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

357

Emond

25,235

28,559 (@. 8,589 hours)

22,498

CADM

-

-

-

714

Emond

14,162

15,914 (@. 8,590 hours)

12,810

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

357

Emond

5,473

7,247 (@. 8,574 hours)

4,335

CADM

-

-

-

714

Emond

2,878

3,774 (@. 8,574 hours)

2,318

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

357

Emond

71.5

99.1 ( a 2 hours)

65.4

CADM

-

-

-

714

Emond

56.4

88.6 (@, 2 hours)

50.4

CADM

-

-

-

1

2

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-106 DRAFT—DO NOT CITE OR QUOTE

-------
1 C.3.1.5. Delia Porta et ah (2001) (male)

Type:

Mouse

Dose:

2,500 and 5,000 ng/kg-week (equivalent to
357 and 714 ng/kg-day)

Strain:

B6C3

Route:

Gavage

Body weight:

6 weeks old (BW 26g)

Regime:

Once a week for 52 weeks

Sex:

Male

Simulation time:

8,736 hours

2	The C ADM model was not ran because the study duration is longer than the allowed model duration

3

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

357

Emond

67.8

787 (@. 8,568 hours)

47.0

CADM

-

-

-

714

Emond

38.0

398 (@. 8,568 hours)

27.3

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

357

Emond

50,397

70,052 (@. 8,577 hours)

37,483

CADM

-

-

-

714

Emond

25,493

35,347 (@. 8,577 hours)

19,155

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

357

Emond

25,516

28,851 (@. 8,589 hours)

22,861

CADM

-

-

-

714

Emond

14,306

16,061 (@. 8,590 hours)

12,999

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

357

Emond

5,504

7,282 (@. 8,574 hours)

4,368

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-107 DRAFT—DO NOT CITE OR QUOTE

-------
714

Emond

2,894

3,791 (@. 8,574 hours)

2,335

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

357

Emond

71.6

99.2 (@. 2 hours)

65.4

CADM

-

-

-

714

Emond

56.4

88.6 (@, 2 hours)

50.4

CADM

-

-

-

1

2

3	C.3.1.6. Fattore et al. (2000)

Type:

Rat

Dose:

20, 200, 2,000 ng/kg-day

Strain:

Sprague Dawley

Route:

Oral in the diet

Body weight:

7 weeks old (BW
150g)

Regime:

Every day for 13 weeks

Sex:

Female and male

Simulation time:

2,184 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

20

Emond

9.59

15.0 (@ 2,160 hours)

11.1

CADM

-

-

-

200

Emond

57.6

102 (@. 2,160 hours)

63.9

CADM

-

-

-

2,000

Emond

476

903 ( a 2,160 hours)

522

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-108 DRAFT—DO NOT CITE OR QUOTE

-------
LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

20

Emond

2,448

3,228 (@. 2,164 hours)

3,078

CADM

4,471

5,639

5,639

200

Emond

24,136

30,245 (@. 2,164 hours)

28,709

CADM

45,337

56,499

56,499

2,000

Emond

234,170

288,020 (@. 2,164 hours)

272,590

CADM

454,031

565,103

565,103

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

20

Emond

890

1,113 (@ 2,166 hours)

1,101

CADM

1,545

1,796

1,756

200

Emond

5,355

6,542 (@. 2,165 hours)

6,430

CADM

13,351

15,604

15,292

2,000

Emond

44,176

54,246 (@. 2,165 hours)

53,140

CADM

131,259

153,534

150,516

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

20

Emond

187

242 (@. 2,164 hours)

233

CADM

261

324

324

200

Emond

1,556

1,940 (@. 2,164 hours)

1,850

CADM

2,496

3,084

3,084

2,000

Emond

14,432

17,797 (@ 2,164 hours)

16,891

CADM

24,836

30,674

30,674

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

20

Emond

24.9

29.8 (@. 2,164 hours)

28.8

CADM

-

-

-

200

Emond

69.4

76.0 (@. 2,164 hours)

74.7

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-109 DRAFT—DO NOT CITE OR QUOTE

-------
2,000

Emond

104

106 (@ 2,164 hours)

106

CADM

-

-

-

1

2

3	C.3.1.7. Franc et ah (2001) Sprague Dawley Rats

Type:

Rats

Dose:

140, 420, and 1400 ng/kg every two weeks
(equivalent to 10, 30, and 100 ng/kg-day)

Strain:

Sprague Dawley,

Route:

Oral gavage

Body weight:

200 g (10 weeks old)

Regime:

Once every two weeks for 22 weeks

Sex:

Female

Simulation
time:

3,696 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

6.59

34.6 (@ 3,360 hours)

5.52

CADM

-

-

-

30

Emond

14.5

98.1 (@ 3,360 hours)

11.3

CADM

-

-

-

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

100

Emond

36.4

315 (@ 3,360 hours)

26.4

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

1,447

2,458 (@ 3,368 hours)

1,150

CADM

2,616

3,620

2,174

30

Emond

4,228

7,161 (@ 3,368 hours)

3,120

CADM

7,936

10,899

6,510

100

Emond

13,821

23,417 (@ 3,368 hours)

9,658

CADM

26,564

36,361

21,703

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l 10 DRAFT—DO NOT CITE OR QUOTE

-------
FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

619

787 (@. 3,417 hours)

560

CADM

966

1,230

759

30

Emond

1,362

1,741 (@3,415 hours)

1,161

CADM

2,448

3,203

1,849

100

Emond

3,430

4,464 (@. 3,412 hours)

2,755

CADM

7,573

10,052

5,606

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

119

177 (@. 3,366 hours)

99.5

CADM

159

212

133

30

Emond

308

472 (@. 3,366 hours)

240

CADM

450

603

367

100

Emond

921

1,445 (@. 3,366 hours)

671

CADM

1,462

1,969

1,181

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

18.6

32.9 ((ci\ 1 hours)

16.4

CADM

-

-

-

30

Emond

33.7

59.2 ((ci\ 1 hours)

29.0

CADM

-

-

-

100

Emond

57.5

86.9 ((ci\ 1 hours)

50.4

CADM

-

-

-

1

2

3	C.3.1.8. Franc et ah (2001) Long-Evans Rats

Type:

Rats

Dose:

140, 420, and 1400 ng/kg every two weeks
(equivalent to 10, 30, and 100 ng/kg-day)

Strain:

Long-Evans

Route:

Oral gavage

Body weight:

190 g (10 weeks old)

Regime:

Once every two weeks for 22 weeks

Sex:

Female

Simulation
time:

3,696 hours

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l 11 DRAFT—DO NOT CITE OR QUOTE

-------
WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

6.58

34.2 (@. 3,360 hours)

5.52

CADM

-

-

-

30

Emond

14.5

97.0 (@. 3,360 hours)

11.3

CADM

-

-

-

100

Emond

36.4

312 (@ 3,360 hours)

26.4

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

1,447

2,458 (@. 3,368 hours)

1,150

CADM

2,616

3,620

2,174

30

Emond

4,228

7,161 (@ 3,368 hours)

3,121

CADM

7,936

10,899

6,510

100

Emond

13,821

23,421 (@. 3,368 hours)

9,659

CADM

26,564

36,361

21,703

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

619

788 (@. 3,417 hours)

560

CADM

966

1,230

759

30

Emond

1,362

1,742 (@. 3,414 hours)

1,160

CADM

2,448

3,203

1,849

100

Emond

3,429

4,466 (@. 3,412 hours)

2,752

CADM

7,573

10,052

5,606

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

119

177 (@. 3,366 hours)

99.5

CADM

159

212

133

30

Emond

308

472 (@. 3,366 hours)

240

CADM

450

603

367

100

Emond

921

1,445 (@. 3,366 hours)

671

CADM

1,462

1,969

1,181

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l 12 DRAFT—DO NOT CITE OR QUOTE

-------
BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

18.6

32.9 {(cf 1 hours)

16.4

CADM

-

-

-

30

Emond

33.7

59.2 {(cf 1 hours)

29.0

CADM

-

-

-

100

Emond

57.5

86.9 (@, 1 hours)

50.4

CADM

-

-

-

1

2

3	C.3.1.9. Franc et al. (2001) Hans Wistar Rats

Type:

Rats

Dose:

140, 420, and 1400 ng/kg every two weeks
(equivalent to 10, 30, and 100 ng/kg-day)

Strain:

Hans Wistar

Route:

Oral gavage

Body weight:

205 g (10 weeks old)

Regime:

Once every two weeks for 22 weeks

Sex:

Female

Simulation
time:

3,696 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

6.59

34.7 (@. 3,360 hours)

5.52

CADM

-

-

-

30

Emond

14.5

98.7 (@. 3,360 hours)

11.3

CADM

-

-

-

100

Emond

36.4

317 (@ 3,360 hours)

26.4

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

1,447

2,458 (@. 3,368 hours)

1,150

CADM

2,616

3,620

2,174

30

Emond

4,228

7,160 (@ 3,368 hours)

3,120

CADM

7,936

10,899

6,510

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l 13 DRAFT—DO NOT CITE OR QUOTE

-------
100

Emond

13,821

23,416 (@ 3,368 hours)

9,658

CADM

26,564

36,361

21,703

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

619

787 (@3,418 hours)

560

CADM

966

1,230

759

30

Emond

1,363

1,741 (@3,415 hours)

1,162

CADM

2,448

3,203

1,849

100

Emond

3,431

4,463 (@3,412 hours)

2,757

CADM

7,573

10,052

5,606

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

119

177 (@ 3,366 hours)

99.5

CADM

159

212

133

30

Emond

308

472 (@ 3,366 hours)

240

CADM

450

603

367

100

Emond

921

1,446 (@ 3,366 hours)

671

CADM

1,462

1,969

1,181

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

10

Emond

186

^2 9 (f(f> 1 hours')

164

CADM

-



-

30

Emond

33.7

5l> 2 u/ 1 hum's)



CADM

-



-

100

Emond

57.5

86.9 (@ 1 hours)

50.4

CADM

-



-

This document is a draft for review purposes only and does not constitute Agency policy.

C-114 DRAFT—DO NOT CITE OR QUOTE

-------
1 C.3.1.10. Hassoun et ah (2000)

Type:

Rat

Dose:

0, 3, 10, 22, 46, 100 ng/kg/day (2.14, 7.14,
15.7, 32.9, and 71.4 ng/kg/day adjusted doses)

Strain:

Sprague Dawley

Route:

Oral gavage

Body weight:

8 weeks old
(BW=215g)

Regime:

5 days/week for 13 weeks

Sex:

Female

Simulation
time:

2184 hours

2

WHOLE BLOOD CONCENTRATLONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

1.94

3.12 (@ 2,112 hours)

1,303.17

CADM

-

-

-

7.14

Emond

4.6136

7.71 (@ 2,112 hours)

2,901.26

CADM

-

-

-

15.7

Emond

8.147

14.2 (@ 2,112 hours)

4,947.3

CADM

-

-

-

32.9

Emond

14.009

25.8 (@2,112 hours)

8,277

CADM

-

-

-

71.4

Emond

25.34

49.7 (@2,112 hours)

14,637

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

266.8

399 (@2,116 hours)

349

CADM

-

-

-

7.14

Emond

888

1,259 (@2,117 hours)

1,079

CADM

-

-

-

15.7

Emond

1,948.499

2,689 (@2,117 hours)

2,278.182

CADM

-

-

-

32.9

Emond

4,055.031

5,484 (@2,117 hours)

4,607.265

CADM

-

-

-

71.4

Emond

8,774.97

11,692 (@2,117 hours)

9,754.31



CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l 15 DRAFT—DO NOT CITE OR QUOTE

-------
FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

179.2

243 ( a 2,126 hours)

234.9

CADM

-

-

-

7.14

Emond

427

553 (@ 2,124 hours)

528

CADM

-

-

-

15.7

Emond

755

958 (@ 2,123 hours)

908

CADM

-

-

-

32.9

Emond

1,299

1,627 (@ 2,122 hours)

1,529

CADM

-

-

-

71.4

Emond

2,349.892

2,928 (@ 2,121 hours)

2,727.240

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

27.425

38.9 (@ 2,116 hours)

35.720

CADM

-

-

-

7.14

Emond

76.87

105 (@2,116 hours)

93.67

CADM

-

-

-

15.7

Emond

153.1

205 (@2,116 hours)

180.2

CADM

-

-

-

32.9

Emond

295

390 (@2,116 hours)

339

CADM

-

-

-

71.4

Emond

600

785 (@2,116 hours)

674

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

6

8.48 (@2,116 hours)

8

CADM

-

-

-

7.14

Emond

13.7242

17.5 (@2,116 hours)

15.7348

CADM

-

-

-

15.7

Emond

21.9703

27.1 (@2,116 hours)

24.4047

CADM

-

-

-

32.9

Emond

32.817

39.2 (@2,116 hours)

35.608

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l 16 DRAFT—DO NOT CITE OR QUOTE

-------
71.4

Emond

47.54

55.0 (@ 2,116 hours)

50.63

CADM

-

-

-

1

2

3	C.3.1.11. Hutt et al. (2008)

Type:

Rat

Dose:

50 ng/kg-week

Strain:

Sprague-Dawley

Route:

Oral gavage

Body weight:

4.5 g

Regime:

1/week for 13 weeks

Sex:

Female

Simulation
time:

2,184 hours (weekly exposure)

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

7.14

Emond

4.49

8.86 (@. 2,016 hours)

4.71

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

7.14

Emond

867.4

1,363 (@ 2,021 hours)

928.1

CADM

1,678

2,007

2,007

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

7.14

Emond

423.6

555 (@. 2,040 hours)

459.9

CADM

730

787.1

769

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

7.14

Emond

76

108 (@. 2,022 hours)

81

CADM

108

126

126

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l 17 DRAFT—DO NOT CITE OR QUOTE

-------
BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

7.14

Emond

14

19.4 (@. 2,020 hours)

14

CADM

-

-

-

1

2

3	C.3.1.12. Kitchin and Woods (1979)

Type:

Rats

Dose:

0, 0.6, 2, 4, 20, 60, 200, 600, 2000, 5000,
20,000 ng/kg/day

Strain:

Sprague-Dawley

Route:

Oral exposure

Body weight:

200 to 250 g (BW set
to 225 g)

Regime:

Single dose

Sex:

Female

Simulation
time:

24 hours*

4	* 1 week is the minimum that can be simulated with the C ADM model, so the CADM model was not used.

5

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)

Model

Metric

Time-weighted Ave

Max

Terminal

0.6

Emond

0.0645

0.126 ((ai 0 hours)

0.0441

CADM

-

-

-

2

Emond

0.202

0.421 ( a 0 hours)

0.137

CADM

-

-

-

4

Emond

0.384

0.841 ((ai 0 hours)

0.258

CADM

-

-

-

20

Emond

1.61

4.21 ( a 0 hours)

1.04

CADM

-

-

-

60

Emond

4.15

12.6 (@, 0 hours)

2.55

CADM

-

-

-

200

Emond

11.6

42.1 ((ci\ 0 hours)

6.61

CADM

-

-

-

600

Emond

30.3

126 ( a 0 hours)

15.8

CADM

-

-

-

2000

Emond

90.9

422 ( a 0 hours)

42.8

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l 18 DRAFT—DO NOT CITE OR QUOTE

-------
5000

Emond

218

1,056 ( a 0 hours)

96.9

CADM

-

-

-

20000

Emond

863

4,233 ( a 0 hours)

365

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)

Model

Metric

Time-weighted Ave

Max

Terminal

0.6

Emond

2.95

3.81 ( a 4 hours)

2.31

CADM

-

-

-

2

Emond

10.5

12.9 {(cf 4 hours)

8.69

CADM

-

-

-

4

Emond

22.2

26.3 ( a 4 hours)

18.9

CADM

-

-

-

20

Emond

128

143 ( a 6 hours)

118

CADM

-

-

-

60

Emond

420

463 ( a 8 hours)

406

CADM

-

-

-

200

Emond

1,523

1,666 ( a 9 hours)

1,526

CADM

-

-

-

600

Emond

4,821

5,258 (@. 10 hours)

4,932

CADM

-

-

-

2000

Emond

16,603

18,080 (@. 11 hours)

17,226

CADM

-

-

-

5000

Emond

41,971

45,674 (@. 11 hours)

43,803

CADM

-

-

-

20000

Emond

167,820

182,580 (@. 11 hours)

175,890

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)

Model

Metric

Time-weighted Ave

Max

Terminal

0.6

Emond

1.60

2.47 (@. 24 hours)

2.47

CADM

-

-

-

2

Emond

5.07

7.71 (@. 24 hours)

7.71

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l 19 DRAFT—DO NOT CITE OR QUOTE

-------
4

Emond

9.68

14.6 (@, 24 hours)

14.6



CADM

-

-

-

20

Emond

41.7

60.7 ( a 24 hours)

60.7



CADM

-

-

-

60

Emond

110

155 ((ci\ 24 hours)

155



CADM

-

-

-

200

Emond

317

427 (@. 24 hours)

427



CADM

-

-

-

600

Emond

851

1,102 (@ 24 hours)

1,102



CADM

-

-

-

2000

Emond

2,620

3,276 ( a 24 hours)

3,276



CADM

-

-

-

5000

Emond

6,361

7,816 (@. 24 hours)

7,816



CADM

-

-

-

20000

Emond

25,401

30,827 (@. 24 hours)

30,827



CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)

Model

Metric

Time-weighted Ave

Max

Terminal

0.6

Emond

0.322

0.341 {(cf 9 hours)

0.338

CADM

-

-

-

2

Emond

1.07

1.14 {(cf 8 hours)

1.12

CADM

-

-

-

4

Emond

2.14

2.27 (@, 8 hours)

2.23

CADM

-

-

-

20

Emond

10.6

11.3 (@. 8 hours)

11.0

CADM

-

-

-

60

Emond

31.7

33.8 (@, 7 hours)

32.8

CADM

-

-

-

200

Emond

105

112 ( a 1 hours)

108

CADM

-

-

-

600

Emond

315

337 ( a 1 hours)

324

CADM

-

-

-

2000

Emond

1,049

1,123 (@ 7 hours)

1,074

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-120 DRAFT—DO NOT CITE OR QUOTE

-------
5000

Emond

2,621

2,806 ( a 1 hours)

2,680

CADM

-

-

-

20000

Emond

10,468

11,215 (@,7 hours)

10,693

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)

Model

Metric

Time-weighted Ave

Max

Terminal

0.6

Emond

0.216

0.309 ((ai 3 hours)

0.159

CADM

-

-

-

2

Emond

0.668

0.975 ((ci\ 3 hours)

0.494

CADM

-

-

-

4

Emond

1.25

1.86 (@, 3 hours)

0.927

CADM

-

-

-

20

Emond

4.87

7.67 ((ci\ 2 hours)

3.66

CADM

-

-

-

60

Emond

11.2

18.3 ((ci\ 2 hours)

8.55

CADM

-

-

-

200

Emond

25.1

40.8 (@, 1 hours)

19.7

CADM

-

-

-

600

Emond

45.8

68.2 (@, 1 hours)

37.6

CADM

-

-

-

2000

Emond

73.3

93.1 ((ci\ 1 hours)

64.7

CADM

-

-

-

5000

Emond

90.9

104 ( a 1 hours)

84.7

CADM

-

-

-

20000

Emond

106

110 ( a 1 hours)

104

CADM

-

-

-

1

2

3	C.3.1.13. Kocibaetal (1976)

Type:

Rats

Dose:

1, 10, 100, 1000 ng/kg-day

Strain:

Sprague-Dawley
(Spartan)

Route:

Diet exposure

Body weight:

170-190 g(bw=180g)

Regime:

5 days/week for 13 weeks

Sex:

Female

Simulation
time:

2,184 hours
(13wk exposed)

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-121 DRAFT—DO NOT CITE OR QUOTE

-------
WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.714

Emond

0.859

1.38 (@ 2,112 hours)

1.13

CADM

-

-

-

7.143

Emond

4.61

7.62 (@ 2,112 hours)

5.27

CADM

-

-

-

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

71.43

Emond

25.3

48.8 (@ 2,112 hours)

26.6

CADM

-

-

-

714.3

Emond

181

403 (@2,112 hours)

184

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.714

Emond

88.3

140 (@2,116 hours)

126

CADM

89.0

192

12.1

7.143

Emond

888

1,259 (@2,117 hours)

1,079

CADM

970

2,007

29.0

71.43

Emond

8,776

11,693 (@2,117 hours)

9,756

CADM

9,841

20,170

88.0

714.3

Emond

86,329

112,580 (@2,117 hours)

92,835

CADM

98,617

201,814

455

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.714

Emond

79.4

114 (@2,129 hours)

111

CADM

120

190

43.0

7.143

Emond

427

553 (@2,124 hours)

528

CADM

456

787

67.0

71.43

Emond

2,348

2,925 (@2,121 hours)

2,720

CADM

3,036

5,748

117

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-122 DRAFT—DO NOT CITE OR QUOTE

-------
714.3

Emond

16,815

21,126 (@ 2,120 hours)

19,233

CADM

28,382

55,013

274

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.714

Emond

10.8

16.1 (@ 2,116 hours)

15.1

CADM

11.5

20.0

3.75

7.143

Emond

76.9

105 (@ 2,116 hours)

93.6

CADM

65.3

126

6.22

71.43

Emond

600

785 (@2,116 hours)

673

CADM

553

1,113

12.0

714.3

Emond

5,366

6,960 (@2,116 hours)

5,842

CADM

5,401

10,967

37.0

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.714

Emond

2.89

4.17 (@2,116 hours)

3.81

CADM

-

-

-

7.143

Emond

13.7

17.5 (@2,116 hours)

15.7

CADM

-

-

-

71.43

Emond

47.5

55.0 (@2,116 hours)

50.6

CADM

-

-

-

714.3

Emond

93.4

98.2 (@2,117 hours)

95.7

CADM

-

-

-

1

2

3	C.3.1.14. Kociba et al. (1978) Female

Type:

Rats

Dose:

0, 1, 10, 100 ng/kg-day

Strain:

Sprague-Dawley
(Spartan)

Route:

Diet exposure

Body weight:

170-190 g (bw=180)

Regime:

7 days/week for 104 weeks

Sex:

Female

Simulation time:

17,472 hours

4

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-123 DRAFT—DO NOT CITE OR QUOTE

-------
WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

1.55

1.92 (@. 17,448 hours)

1.69

CADM

-

-

-

10

Emond

7.15

9.25 (@. 17,448 hours)

7.16

CADM

-

-

-

100

Emond

38.6

57.5 (@. 17,448 hours)

37.1

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

192

226 (@. 17,452 hours)

218

CADM

292

333

333

10

Emond

1,618

1,742 (@. 17,452 hours)

1,665

CADM

2,981

3,342

3,342

100

Emond

14,892

15,673 (@. 17,452 hours)

14,907

CADM

29,917

33,432

33,432

FA T CONCENTRA TLONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

147

165 (@. 17,457 hours)

164

CADM

196

229

181

10

Emond

680

713 (@. 17,454 hours)

706

CADM

861

1,015

789

100

Emond

3,663

3,788 (@. 17,454 hours)

3,731

CADM

6,756

7,939

6,203

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

21.2

24.3 (@. 17,452 hours)

23.8

CADM

26.0

27.0

27.0

10

Emond

131

140 (@. 17,452 hours)

136

CADM

169

176

176

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-124 DRAFT—DO NOT CITE OR QUOTE

-------
100

Emond

989

1,039 (@. 17,452 hours)

994

CADM

1,546

1,601

1,601

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

5.11

5.77 (@. 17,452 hours)

5.59

CADM

-

-

-

10

Emond

20.0

21.1 (@. 17,452 hours)

20.4

CADM

-

-

-

100

Emond

59.9

61.5 (@. 17,452 hours)

60.1

CADM

-

-

-

1

2

3	C.3.1.15. Kocibaetal. (1978) Male

Type:

Rats

Dose:

0, 1, 10, 100 ng/kg-day

Strain:

Sprague-Dawley
(Spartan)

Route:

Diet exposure

Body weight:

Body weight
approximated to be
250 g

Regime:

7 days/week for 104 weeks

Sex:

Male

Simulation time:

17,472 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

1.56

1.96 (@. 17,448 hours)

1.70

CADM

-

-

-

10

Emond

7.16

9.35 (@. 17,448 hours)

7.11

CADM

-

-

-

100

Emond

38.7

59.3 (@. 17,448 hours)

37.1

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

194

229 (@. 17,452 hours)

221

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-125 DRAFT—DO NOT CITE OR QUOTE

-------
10

Emond

1,616

1,723 (@. 17,452 hours)

1,649

CADM

-

-

-

100

Emond

14,898

15,671 (@. 17,452 hours)

14,912

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

148

167 (@. 17,456 hours)

166

CADM

-

-

-

10

Emond

680

709 (@. 17,454 hours)

703

CADM

-

-

-

100

Emond

3,677

3,803 (@. 17,453 hours)

3,747

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

21.4

24.6 (@. 17,452 hours)

24.1

CADM

-

-

-

10

Emond

131

139 (@. 17,452 hours)

134

CADM

-

-

-

100

Emond

991

1,041 (@. 17,452 hours)

995

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

5.15

5.83 (@. 17,452 hours)

5.64

CADM

-

-

-

10

Emond

20.0

21.0 (@. 17,452 hours)

20.3

CADM

-

-

-

100

Emond

60.0

61.5 (@. 17,452 hours)

60.1

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-126 DRAFT—DO NOT CITE OR QUOTE

-------
1 C.3.1.16. Latchoumycandane and Mathur (2002)

Type:

Rat

Dose:

0, 1, 10, lOOng/kg-day

Strain:

Wistar

Route:

Oral gavage

Body weight:

45 days old
(BW set to 200g)

Regime:

1/day for 45 days

Sex:

Male

Simulation
time:

1,080 hours (daily exposure)

2

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

0.785

1.37 (@. 1,056 hours)

1.18

CADM

-

-

-

10

Emond

4.65

8.18 (@. 1,056 hours)

6.18

CADM

-

-

-

100

Emond

27.3

53.9 (@. 1,056 hours)

33.8

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

78.5

138 (@. 1,060 hours)

133

CADM

116

217

217

10

Emond

902

1,423 (@. 1,060 hours)

1,358

CADM

1,669

2,550

2,550

100

Emond

9,579

14,015 (@. 1,061 hours)

13,306

CADM

17,681

25,915

25,915

FA T CONCENTRA TLONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

69.8

113 (@. 1,072 hours)

113

CADM

150

220

220

10

Emond

416

608 (@. 1,065 hours)

604

CADM

744

1,009

1,009

100

Emond

2,448

3,425 (@. 1,062 hours)

3,380

CADM

5,719

7,866

7,866

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-127 DRAFT—DO NOT CITE OR QUOTE

-------
BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

9.56

15.9 (@. 1,060 hours)

15.6

CADM

14.0

22.2

22.2

10

Emond

76.7

117 (@. 1,060 hours)

113

CADM

106

157

157

100

Emond

646

933 (@. 1,060 hours)

891

CADM

988

1,439

1,439

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

2.64

4.12 (@. 1,060 hours)

3.96

CADM

-

-

-

10

Emond

13.7

18.8 (@. 1,060 hours)

18.1

CADM

-

-

-

100

Emond

48.6

59.0 (@. 1,060 hours)

57.5

CADM

-

-

-

1

2

3	C.3.1.17. Li et al (1997)

Type:

Rats

Dose:

0, 3, 10, 30, 100, 300, 1000, 3000,
10000, 30000 ng/kg/day

Strain:

Sprague-Dawley

Route:

Gastric intubation

Body weight:

22 day old, 55 to 58 g
(BW set to 56.5 g)

Regime:

One dose for one day

Sex:

Female

Simulation time:

24 hours

4	The CADM model was not ran because the dosing duration is lower than the resolution of the model (1 week)

5

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)

Model

Metric

Time-weighted Ave

Max

Terminal

3

Emond

0.266

0.470 ((ci\ 1 hours)

0.180

CADM

-

-

-

10

Emond

0.799

1.57 ((ci\ 1 hours)

0.535

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-128 DRAFT—DO NOT CITE OR QUOTE

-------
30

Emond

2.10

4.68 ( a 1 hours)

1.37

CADM

-

-

-

100

Emond

5.87

15.6 ((ci\ 1 hours)

3.68

CADM

-

-

-

300

Emond

15.0

46.8 (@, 0 hours)

8.83

CADM

-

-

-

1,000

Emond

43.3

156 ( a 0 hours)

23.4

CADM

-

-

-

3,000

Emond

120

469 ( a 0 hours)

59.9

CADM

-

-

-

10,000

Emond

386

1,570 ((ci\ 0 hours)

182

CADM

-

-

-

30,000

Emond

1,172

4,762 ( a 0 hours)

535

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)

Model

Metric

Time-weighted Ave

Max

Terminal

3

Emond

14.7

18.6 (@, 4 hours)

11.9

CADM

-

-

-

10

Emond

55.0

65.2 {(cf 5 hours)

47.6

CADM

-

-

-

30

Emond

185

210 (@, 6 hours)

170

CADM

-

-

-

100

Emond

690

768 ( a 1 hours)

666

CADM

-

-

-

300

Emond

2,248

2,473 (@. 8 hours)

2,240

CADM

-

-

-

1,000

Emond

7,938

8,671 ( a 9 hours)

8,094

CADM

-

-

-

3,000

Emond

24,474

26,639 (@. 9 hours)

25,267

CADM

-

-

-

10,000

Emond

82,349

89,464 ((ai 9 hours)

85,597

CADM

-

-

-

30,000

Emond

245,610

265,670 (@. 10 hours)

255,390

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-129 DRAFT—DO NOT CITE OR QUOTE

-------
FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)

Model

Metric

Time-weighted Ave

Max

Terminal

3

Emond

8.75

12.7 (@. 24 hours)

12.7

CADM

-

-

-

10

Emond

26.6

38.0 (@. 24 hours)

38.0

CADM

-

-

-

30

Emond

70.8

98.9 ((ai 24 hours)

98.9

CADM

-

-

-

100

Emond

202

273 ( a 24 hours)

273

CADM

-

-

-

300

Emond

530

689 ( a 24 hours)

689

CADM

-

-

-

1,000

Emond

1,573

1,958 (@. 24 hours)

1,958

CADM

-

-

-

3,000

Emond

4,433

5,358 (@. 24 hours)

5,358

CADM

-

-

-

10,000

Emond

14,428

17,119 (@ 24 hours)

17,119

CADM

-

-

-

30,000

Emond

44,361

51,948 (@ 22 hours)

51,898

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)

Model

Metric

Time-weighted Ave

Max

Terminal

3

Emond

1.60

1.70 {(cf 8 hours)

1.68

CADM

-

-

-

10

Emond

5.33

5.66 (@, 8 hours)

5.56

CADM

-

-

-

30

Emond

15.9

16.9 (@, 8 hours)

16.5

CADM

-

-

-

100

Emond

52.8

56.2 {(cf 1 hours)

54.5

CADM

-

-

-

300

Emond

158

169 ( a 1 hours)

163

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-130 DRAFT—DO NOT CITE OR QUOTE

-------
1,000

Emond

525

561 (a 1 hours)

539

CADM

-

-

-

3,000

Emond

1,574

1,684 ( a 1 hours)

1,611

CADM

-

-

-

10,000

Emond

5,240

5,610 ((ci\ 1 hours)

5,360

CADM

-

-

-

30,000

Emond

15,758

16,815 (@. 7 hours)

16,041

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)

Model

Metric

Time-weighted Ave

Max

Terminal

3

Emond

0.89

1.37 ((ci\ 3 hours)

0.64

CADM

-

-

-

10

Emond

2.58

4.10 ((ci\ 2 hours)

1.88

CADM

-

-

-

30

Emond

6.37

10.5 ((ci\ 2 hours)

4.71

CADM

-

-

-

100

Emond

15.54

25.9 ((ci\ 2 hours)

11.77

CADM

-

-

-

300

Emond

31.25

50.1 ((ci\ 1 hours)

24.57

CADM

-

-

-

1,000

Emond

56.75

79.8 ((ci\ 1 hours)

47.62

CADM

-

-

-

3,000

Emond

81.28

98.4 ((ci\ 1 hours)

73.32

CADM

-

-

-

10,000

Emond

99.77

108 ( a 1 hours)

95.68

CADM

-

-

-

30,000

Emond

107.69

111 (a 1 hours)

106.24

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-131 DRAFT—DO NOT CITE OR QUOTE

-------
1 C.3.1.18. Murray et ah (1979) Adult Portion

Type:

Rat

Dose:

1,10, and 100 ng/kg-day

Strain:

Sprague Dawley

Route:

Diet oral dose

Body weight:

BW set to 4.5 g

Regime:

Once per day for 120 days

Sex:

Female

Simulation time:

2880 hours

2

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

1.12

1.51 (@ 2,856 hours)

1.42

CADM

-

-

-

10

Emond

5.88

7.59 (@. 2,856 hours)

6.75

CADM

-

-

-

100

Emond

32.7

44.3 (@. 2,856 hours)

36.0

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

128

180 (@. 2,859 hours)

173

CADM

-

-

-

10

Emond

1,273

1,618 (@. 2,860 hours)

1,540

CADM

-

-

-

100

Emond

12,601

15,281 (@. 2,860 hours)

14,460

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

106

139 (@. 2,865 hours)

138

CADM

-

-

-

10

Emond

556

665 (@. 2,864 hours)

657

CADM

-

-

-

100

Emond

3,095

3,604 (@. 2,862 hours)

3,534

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-132 DRAFT—DO NOT CITE OR QUOTE

-------
BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

14.8

20.0 ((ai 2,860 hours)

19.6

CADM

-

-

-

10

Emond

105

130 (@. 2,860 hours)

126

CADM

-

-

-

100

Emond

837

1,003 (@. 2,860 hours)

957

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

3.77

4.95 (@. 2,859 hours)

4.77

CADM

-

-

-

10

Emond

17.1

20.3 (@. 2,859 hours)

19.5

CADM

-

-

-

100

Emond

55.3

60.9 (@. 2,860 hours)

59.4

CADM

-

-

-

1

2

3	C.3.1.19. NTP (1982)—Female Rats, Chronic

Type:

Rat

Dose:

10, 50 and 500 ng/kg/wk,
two doses per week

Strain:

Osborne-Mendel

Route:

Oral exposure

Body weight

6 weeks old
(BW set to 25 Og)

Regime:

Biweekly

(Simulation has been perform using female BW

Sex:

Female

Simulation time

17,472 hours (104 weeks of exposure)

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

1.96

3.11 (@. 17,220 hours)

1.94

CADM

-

-

-

7.1

Emond

5.69

11.0 (@ 17,388 hours)

5.40

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-133 DRAFT—DO NOT CITE OR QUOTE

-------
71

Emond

29.8

82.2 (@. 17,388 hours)

26.9

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

265

308 (@. 17,226 hours)

265

CADM

15,318

20,170

7,102

7.1

Emond

1,175

1,338 (@. 17,394 hours)

1,117

CADM

30,700

40,353

14,200

71

Emond

10,734

12,182 (@. 17,395 hours)

9,882

CADM

30,700

40,353

14,200

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

186

200 (@. 17,328 hours)

193

CADM

4,655

5,748

2,107

7.1

Emond

541

569 (@. 17,409 hours)

544

CADM

9,064

11,224

3,964

71

Emond

2,826

2,973 (@. 17,404 hours)

2,769

CADM

17,879

22,172

7,671

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

27.9

31.1 (@. 17,225 hours)

28.4

CADM

855

1,113

403

7.1

Emond

99.4

110 (@. 17,393 hours)

96.7

CADM

1,695

2,208

787

71

Emond

729

814 (@. 17,393 hours)

683

CADM

3,375

4,395

1,556

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

6.37

7.26 (@. 17,224 hours)

6.38

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-134 DRAFT—DO NOT CITE OR QUOTE

-------
7.1

Emond

16.6

18.5 (@. 17,392 hours)

16.1

CADM

-

-

-

71

Emond

52.7

56.4 (@. 17,393 hours)

50.9

CADM

-

-

-

1

2

3	C.3.1.20. NTP (1982)—Male Rats, Chronic

Type:

Rat

Dose:

10, 50 and 500 ng/kg/wk,
two doses per week

Strain:

Osborne-Mendel

Route:

Oral exposure

Body weight

6 weeks old
(BW set to 350g)

Regime:

Biweekly

(Simulation has been perform using female BW

Sex:

Male

Simulation time

17,472 hours (104 weeks of exposure)

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

1.96

3.18 (@. 17,388 hours)

1.93

CADM

-

-

-

7.1

Emond

5.70

11.4 (@. 17,388 hours)

5.39

CADM

-

-

-

71

Emond

29.9

87.0 (@. 17,388 hours)

26.9

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

265

306 (@. 17,394 hours)

263

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

7.1

Emond

1,174

1,334 (@. 17,394 hours)

1,114

CADM

-

-

-

71

Emond

10,736

12,170 (@. 17,395 hours)

9,881

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-135 DRAFT—DO NOT CITE OR QUOTE

-------
FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

186

199 (@. 17,412 hours)

193

CADM

-

-

-

7.1

Emond

541

569 (@. 17,409 hours)

544

CADM

-

-

-

71

Emond

2,836

2,983 (@. 17,404 hours)

2,784

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

27.8

30.9 (@. 17,393 hours)

28.2

CADM

-

-

-

7.1

Emond

99.5

110 (@. 17,393 hours)

96.6

CADM

-

-

-

71

Emond

730

816 (@. 17,393 hours)

684

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

6.36

7.22 (@. 17,392 hours)

6.35

CADM

-

-

-

7.1

Emond

16.6

18.4 (@. 17,392 hours)

16.0

CADM

-

-

-

71

Emond

52.7

56.3 (@. 17,393 hours)

50.9

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-136 DRAFT—DO NOT CITE OR QUOTE

-------
1 C.3.1.21. NTP (1982)—Female Mice, Chronic

Type:

Mice

Dose:

40, 200 and 2000 ng/kg/wk,
two doses during the week

Strain:

B6C3F1

Route:

Oral exposure

Body weight

6 weeks old
(BW set to 23g)

Regime:

Biweekly

(Simulation has been perform using female BW)

Sex:

Female

Simulation time

17,472 hours (104 weeks of exposure)

2	* The mice chronic exposure could not be simulated with the CADM model because this model simulates for only

3	123 days.

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

5.7

Emond

1.95

4.86 (@. 16,800 hours)

1.82

CADM

-

-

-

28.6

Emond

5.84

19.8 (@. 17,388 hours)

5.17

CADM

-

-

-

286

Emond

32.1

171 (@. 16,884 hours)

26.0

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

5.7

Emond

490

582 (@. 16,807 hours)

463

CADM

-

-

-

28.6

Emond

2,236

2,629 (@. 17,395 hours)

2,025

CADM

-

-

-

286

Emond

20,841

24,353 (@. 17,396 hours)

18,182

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

5.7

Emond

737

785 (@. 17,408 hours)

757

CADM

-

-

-

28.6

Emond

2,213

2,337 (@. 17,404 hours)

2,216

CADM

-

-

-

286

Emond

12,138

12,861 (@. 17,400 hours)

11,775

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-137 DRAFT—DO NOT CITE OR QUOTE

-------
BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

5.7

Emond

91.9

103 (@. 17,393 hours)

91.2

CADM

-

-

-

28.6

Emond

329

370 (@. 17,393 hours)

313

CADM

-

-

-

286

Emond

2,400

2,740 (@. 17,393 hours)

2,176

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

5.7

Emond

6.18

7.29 (@. 16,805 hours)

5.93

CADM

-

-

-

28.6

Emond

16.3

18.9 (@. 17,393 hours)

15.3

CADM

-

-

-

286

Emond

52.3

67.8 (@, 2 hours)

49.3

CADM

-

-

-

1

2

3	C.3.1.22. NTP (1982)—Male Mice, Chronic

Type:

Mice

Dose:

10, 50 and 500ng/kg/wk,
two doses during the week

Strain:

B6C3F1

Route:

Oral exposure

Body weight

6 weeks old
(BW set to 25g)

Regime:

Biweekly

Sex:

Male

Simulation time

17,472 hours (104 weeks of exposure)

4	* The mice chronic exposure could not be simulated with the CADM model because this model simulates for only

5	123 days.

6

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

0.767

1.53 (@. 17,304 hours)

0.749

CADM

-

-

-

7.1

Emond

2.27

5.99 (@. 17,052 hours)

2.11

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-138 DRAFT—DO NOT CITE OR QUOTE

-------
71

Emond

11.2

46.7 (@. 17,388 hours)

9.59

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

138

165 (@. 17,310 hours)

136

CADM

-

-

-

7.1

Emond

606

722 (@. 17,059 hours)

571

CADM

-

-

-

71

Emond

5,409

6,328 (@. 17,395 hours)

4,805

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

290

314 (@. 17,411 hours)

306

CADM

-

-

-

7.1

Emond

860

918 (@. 17,155 hours)

883

CADM

-

-

-

71

Emond

4,257

4,490 (@. 17,402 hours)

4,204

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

32.3

36.2 (@. 17,309 hours)

33.3

CADM

-

-

-

7.1

Emond

110

123 (@. 17,057 hours)

108

CADM

-

-

-

71

Emond

710

802 (@. 17,393 hours)

660

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.4

Emond

2.56

3.03 (@. 17,309 hours)

2.53

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-139 DRAFT—DO NOT CITE OR QUOTE

-------
7.1

Emond

7.12

8.40 (@. 17,057 hours)

6.82

CADM

-

-

-

71

Emond

27.1

32.4 ((ci\ 2 hours)

25.3

CADM

-

-

-

1

2

3	C.3.1.23. NTP (2006) 14 Weeks

Type:

Rat

Dose:

0, 3, 10, 22, 46, 100 ng/kg-day

Strain:

Sprague Dawley

Route:

Oral gavage

Body
weight:

8 weeks old
(BW=215g)

Regime:

5 days/weeks for 14 weeks

Sex:

Female and male

Simulation time:

2,352 hours (14 weeks)

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

1.98

3.15 (a 2.280 hours)

2.39

CADM

-

-

-

7.14

Emond

4.69

7.75 (@. 2,280 hours)

5.30

CADM

-

-

-

15.7

Emond

8.27

14.3 (@. 2,280 hours)

9.02

CADM

-

-

-

32.9

Emond

14.2

25.9 (@. 2,280 hours)

15.1

CADM

-

-

-

71.4

Emond

25.7

49.8 (@. 2,280 hours)

26.6

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

275

404 (@. 2,284 hours)

354



CADM

-

-

-

7.14

Emond

909

1,270 (@. 2,285 hours)

1,089



CADM

-

-

-

15.7

Emond

1,988

2,703 (@. 2,285 hours)

2,291



CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-140 DRAFT—DO NOT CITE OR QUOTE

-------
32.9

Emond

4,129

5,508 (@. 2,285 hours)

4,628



CADM

-

-

-

71.4

Emond

8,921

11,734 (@ 2,285 hours)

9,792



CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

184

246 (@. 2,294 hours)

237

CADM

-

-

-

7.14

Emond

436

557 (@. 2,292 hours)

532

CADM

-

-

-

15.7

Emond

768

962 (@. 2,291 hours)

912

CADM

-

-

-

32.9

Emond

1,319

1,633 (@. 2,289 hours)

1,535

CADM

-

-

-

71.4

Emond

2,385

2,938 (@. 2,289 hours)

2,736

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

28.2

39.4 (@. 2,284 hours)

36.1

CADM

-

-

-

7.14

Emond

78.5

106 (@. 2,284 hours)

94.4

CADM

-

-

-

15.7

Emond

156

206 ( a 2,284 hours)

181

CADM

-

-

-

32.9

Emond

300

391 ( a 2,284 hours)

340

CADM

-

-

-

71.4

Emond

610

788 (@. 2,284 hours)

676

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

6.41

8.55 (@. 2,284 hours)

7.74

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-141 DRAFT—DO NOT CITE OR QUOTE

-------
7.14

Emond

13.9

17.6 (@. 2,284 hours)

15.8

CADM

-

-

-

15.7

Emond

22.2

27.2 (@. 2,284 hours)

24.5

CADM

-

-

-

32.9

Emond

33.2

39.3 (@. 2,284 hours)

35.7

CADM

-

-

-

71.4

Emond

47.9

55.1 (@. 2,284 hours)

50.7

CADM

-

-

-

1

2

3	C.3.1.24. NTP (2006) 31 Weeks

Type:

Rat

Dose:

0, 3, 10, 22, 46, 100 ng/kg-day

Strain:

Sprague Dawley

Route:

Oral gavage

Body
weight:

8 weeks old
(BW=215g)

Regime:

5 days/weeks for 31 weeks

Sex:

Female and male

Simulation time:

5,208 hours (31 weeks)

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

2.33

3.25 (@. 3,960 hours)

2.48

CADM

-

-

-

7.14

Emond

5.32

7.89 (@. 3,960 hours)

5.40

CADM

-

-

-

15.7

Emond

9.21

14.5 (@. 3,960 hours)

9.15

CADM

-

-

-

32.9

Emond

15.7

26.2 (@ 5,136 hours)

15.3

CADM

-

-

-

71.4

Emond

28.1

50.4 (@ 5,136 hours)

27.0

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

341

425 (@. 5,140 hours)

373

CADM

-

-

-

7.14

Emond

1,075

1,308 (@. 3,965 hours)

1,117

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-142 DRAFT—DO NOT CITE OR QUOTE

-------
15.7

Emond

2,296

2,756 (@ 3,965 hours)

2,336

CADM

-

-

-

32.9

Emond

4,696

5,597 (@ 5,141 hours)

4,712

CADM

-

-

-

71.4

Emond

10,033

11,905 (@ 5,141 hours)

9,953



CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

220

256 (@ 5,149 hours)

246

CADM

-

-

-

7.14

Emond

501

570 (@ 4,139 hours)

542

CADM

-

-

-

15.7

Emond

868

978 (@ 4,138 hours)

926

CADM

-

-

-

32.9

Emond

1,476

1,657 (@ 5,145 hours)

1,558

CADM

-

-

-

71.4

Emond

2,652

2,978 (@ 5,144 hours)

2,775

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

34.2

41.2 (@5,140 hours)

37.8

CADM

-

-

-

7.14

Emond

91.6

108 (@ 3,964 hours)

96.6

CADM

-

-

-

15.7

Emond

178

209 (@ 3,964 hours)

184

CADM

-

-

-

32.9

Emond

339

398 (@ 5,140 hours)

346

CADM

-

-

-

71.4

Emond

682

799 (@ 5,140 hours)

687

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

7.48

8.83 (@ 5,140 hours)

8.01

CADM

-

-

-

7.14

Emond

15.6

17.9 (@ 3,964 hours)

16.1

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-143 DRAFT—DO NOT CITE OR QUOTE

-------


CADM

-

-

-

15.7

Emond

24.3

27.4 (@. 3,964 hours)

24.8

CADM

-

-

-

32.9

Emond

35.7

39.6 (@. 5,140 hours)

36.0

CADM

-

-

-

71.4

Emond

50.9

55.4 (@, 5,140 hours)

51.1

CADM

-

-

-

1

2

3	C.3.1.25. NTP (2006) 53 Weeks

Type:

Rat

Dose:

0,3, 10, 22, 46, lOOng/kg-day

Strain:

Sprague Dawley

Route:

Oral gavage

Body
weight:

8 weeks old
(BW=215g)

Regime:

5 days/weeks for 105 weeks

Sex:

Female and male

Simulation time:

8,904 hours (53 weeks)

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

2.46

3.25 (@ 6,312 hours)

2.48

CADM

-

-

-

7.14

Emond

5.53

7.89 (@. 3,960 hours)

5.41

CADM

-

-

-

15.7

Emond

9.54

14.5 (@. 8,832 hours)

9.17

CADM

-

-

-

32.9

Emond

16.2

26.3 (@. 8,832 hours)

15.3

CADM

-

-

-

71.4

Emond

29.0

50.6 (@. 8,832 hours)

27.1

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

366

426 (@ 6,316 hours)

373

CADM

-

-

-

7.14

Emond

1,134

1,308 (@. 3,965 hours)

1,121

CADM

-

-

-

15.7

Emond

2,406

2,759 (@. 8,837 hours)

2,345

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-144 DRAFT—DO NOT CITE OR QUOTE

-------
32.9

Emond

4,902

5,612 (@ 8,837 hours)

4,727

CADM

-

-

-

71.4

Emond

10,439

11,938 (@ 8,837 hours)

9,985

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

233

256 (@ 6,325 hours)

247

CADM

-

-

-

7.14

Emond

524

570 (@ 4,139 hours)

544

CADM

-

-

-

15.7

Emond

904

980 (@ 8,842 hours)

929

CADM

-

-

-

32.9

Emond

1,533

1,661 (@ 8,841 hours)

1,562

CADM

-

-

-

71.4

Emond

2,749

2,986 (@ 8,840 hours)

2,784

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

36.4

41.2 (@ 6,316 hours)

37.8

CADM

-

-

-

7.14

Emond

96.1

108 (@ 3,964 hours)

96.9

CADM

-

-

-

15.7

Emond

186

210 (@ 8,836 hours)

185

CADM

-

-

-

32.9

Emond

353

399 (@ 8,836 hours)

347

CADM

-

-

-

71.4

Emond

709

801 (@ 8,836 hours)

689

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

7.87

8.84 (@6,316 hours)

8.01

CADM

-

-

-

7.14

Emond

16.2

17.9 (@ 3,964 hours)

16.1

CADM

-

-

-

15.7

Emond

25.1

27.5 (@ 8,836 hours)

24.8

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-145 DRAFT—DO NOT CITE OR QUOTE

-------


CADM

-

-

-

32.9

Emond

36.6

39.7 (@. 8,836 hours)

36.1

CADM

-

-

-

71.4

Emond

51.9

55.4 (@. 8,836 hours)

51.1

CADM

-

-

-

1

2

3	C.3.1.26. NTP (2006) 2 Years

Type:

Rat

Dose:

0, 3, 10, 22, 46, 100 ng/kg-day

Strain:

Sprague Dawley

Route:

Oral gavage

Body
weight:

8 weeks old
(BW=215g)

Regime:

5 days/weeks for 105 weeks

Sex:

Female and male

Simulation time:

17,640 hours* (105 weeks)

4	*The CADM model simulates for 104 weeks only (17,472 hours). As a result, the terminal values from the CADM

5	model may be underestimated compared to the Emond model, which considers the full 105 weeks of exposure.

6

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

2.56

3.47 (@. 17,568 hours)

2.62

CADM

-

-

-

7.14

Emond

5.69

7.97 (@. 17,568 hours)

5.46

CADM

-

-

-

15.7

Emond

9.79

14.6 (@. 17,568 hours)

9.22

CADM

-

-

-

32.9

Emond

16.6

26.4 (@. 17,568 hours)

15.4

CADM

-

-

-

71.4

Emond

29.7

50.8 (@. 17,568 hours)

27.1

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

385

460 (@. 17,572 hours)

403

CADM

632

715

715

7.14

Emond

1,177

1,320 (@. 17,573 hours)

1,135

CADM

2,127

2,387

2,387

15.7

Emond

2,487

2,779 (@. 17,573 hours)

2,361

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-146 DRAFT—DO NOT CITE OR QUOTE

-------


CADM

4,691

5,252

5,252

32.9

Emond

5,051

5,637 (@ 17,573 hours)

4,749

CADM

9,822

10,984

10,984

71.4

Emond

10,734

11,976 (@ 17,573 hours)

10,018

CADM

21,366

23,880

23,880

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

243

271 (@ 17,581 hours)

261

CADM

302

355

277

7.14

Emond

541

575 (@ 17,579 hours)

549

CADM

667

787

611

15.7

Emond

930

985 (@ 17,578 hours)

934

CADM

1,242

1,463

1,138

32.9

Emond

1,574

1,667 (@ 17,577 hours)

1,568

CADM

2,369

2,787

2,173

71.4

Emond

2,821

2,995 (@ 17,576 hours)

2,792

CADM

4,890

5,748

4,489

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

38.1

44.0 (@ 17,572 hours)

40.4

CADM

46.0

48.0

48.0

7.14

Emond

99.5

109 (@ 17,572 hours)

97.9

CADM

125

130

130

15.7

Emond

192

211 (@ 17,572 hours)

186

CADM

257

267

267

32.9

Emond

364

400 (@ 17,572 hours)

348

CADM

520

538

538

71.4

Emond

729

804 (@ 17,572 hours)

691

CADM

1,110

1,149

1,149

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

2.14

Emond

8.17

9.30 (@ 17,572 hours)

8.43

This document is a draft for review purposes only and does not constitute Agency policy.

C-147 DRAFT—DO NOT CITE OR QUOTE

-------


CADM

-

-

-

7.14

Emond

16.6

18.0 (@ 17,572 hours)

16.2

CADM

-

-

-

15.7

Emond

25.6

27.6 (@ 17,572 hours)

24.9

CADM

-

-

-

32.9

Emond

37.3

39.7 (@ 17,572 hours)

36.2

CADM

-

-

-

71.4

Emond

52.7

55.5 (@ 17,572 hours)

51.2

CADM

-

-

-

1

2

3	C.3.1.27. Sewall et al (1995)

Type:

Rat

Dose:

49, 149.8, 490, and 1750 ng/kg every two
weeks or 3.5, 10.7, 35, and 125 ng/kg-day

Strain:

Sprauge-Dawley

Route:

Oral gavage

Body weight:

12 wk old
(BW set to 25 Og)

Regime:

Once every 2 weeks for 30 weeks

Sex:

Female

Simulation time:

5040 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

3.5

Emond

3.29

13.7 (@ 4,704 hours)

2.88

CADM

-

-

-

10.7

Emond

7.11

38.7 (@ 4,704 hours)

5.79

CADM

-

-

-

35

Emond

16.6

120 (@ 4,704 hours)

12.6

CADM

-

-

-

125

Emond

44.7

414 (@ 4,704 hours)

31.4

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

3.5

Emond

550

901 (@4,711 hours)

459

CADM

-

-

-

10.7

Emond

1,605

2,632 (@ 4,712 hours)

1,229

CADM

-

-

-

35

Emond

5,072

8,350 (@ 4,712 hours)

3,618

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-148 DRAFT—DO NOT CITE OR QUOTE

-------


CADM

-

-

-

125

Emond

17,683

29,256 (@ 4,713 hours)

12,011

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

3.5

Emond

310

383 (@ 4,765 hours)

290

CADM

-

-

-

10.7

Emond

670

827 (@ 4,763 hours)

590

CADM

-

-

-

35

Emond

1,569

1,957 (@ 4,760 hours)

1,304

CADM

-

-

-

125

Emond

4,217

5,376 (@ 4,757 hours)

3,303

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

3.5

Emond

51.4

72.5 (@ 4,710 hours)

45.3

CADM

-

-

-

10.7

Emond

130

189 (@ 4,710 hours)

106

CADM

-

-

-

35

Emond

364

546 (@ 4,710 hours)

274

CADM

-

-

-

125

Emond

1,164

1,793 (@ 4,710 hours)

824

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

3.5

Emond

10.2

15.8 ( a 2 hours)

9.18

CADM

-

-

-

10.7

Emond

19.8

34.4 (@ 1 hours)

17.0

CADM

-

-

-

35

Emond

37.0

63.2 (a 1 hours)

31.4

CADM

-

-

-

125

Emond

63.1

90.9 ( a 1 hours)

55.2

CADM

-

-

-

1

2

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-149 DRAFT—DO NOT CITE OR QUOTE

-------
1 C.3.1.28. Shi et ah (2007) Adult Portion

Type:

Rat

Dose:

1, 5, 50 and 200 ng/kg

Strain:

Sprague Dawley

Route:

Oral exposure

Body weight:

BW set to 4.5 g

Regime:

Weekly doses for 11 months

Sex:

Female

Simulation time:

8040 hours

2

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.143

Emond

0.342

0.475 (@. 7,561 hours)

0.380

CADM

-

-

-

0.714

Emond

1.07

1.53 (@. 7,560 hours)

1.09

CADM

-

-

-

7.14

Emond

5.23

9.12 (@. 7,560 hours)

4.86

CADM

-

-

-

28.6

Emond

13.9

29.2 (@. 7,560 hours)

12.4

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.143

Emond

26.1

36.5 (@. 7,564 hours)

29.6

CADM

-

-

-

0.714

Emond

118

159 (@. 7,564 hours)

120

CADM

-

-

-

7.14

Emond

1,068

1,415 (@. 7,565 hours)

970

CADM

-

-

-

28.6

Emond

4,119

5,450 (@. 7,565 hours)

3,574

CADM

-

-

-

FA T CONCENTRA TLONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.143

Emond

32.5

40.0 (@. 7,583 hours)

36.7

CADM

-

-

-

0.714

Emond

102

120 (@. 7,584 hours)

106

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l50 DRAFT—DO NOT CITE OR QUOTE

-------
7.14

Emond

497

571 (@. 7,584 hours)

475

CADM

-

-

-

28.6

Emond

1,322

1,527 (@. 7,584 hours)

1,217

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.143

Emond

3.94

4.99 (@. 7,566 hours)

4.45

CADM

-

-

-

0.714

Emond

14.0

17.2 (@. 7,566 hours)

14.5

CADM

-

-

-

7.14

Emond

90.8

112 (@ 7,566 hours)

84.4

CADM

-

-

-

28.6

Emond

300

374 (@. 7,566 hours)

266

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.143

Emond

1.18

1.60 (@. 7,563 hours)

1.31

CADM

-

-

-

0.714

Emond

3.62

4.75 (@. 7,563 hours)

3.70

CADM

-

-

-

7.14

Emond

15.6

19.7 (@. 7,564 hours)

14.7

CADM

-

-

-

28.6

Emond

33.5

40.7 (@. 7,564 hours)

31.2

CADM

-

-

-

1

2

3	C.3.1.29. Smialowicz et al. (2008)

Type:

Mice

Dose:

0, 1.5, 15, 150, 450 ng/kg-day

Strain:

B6C3F1

Route:

Oral gavage

Body weight:

13 wk old
(BW set to 28g)

Regime:

5 days/week for 13 weeks

Sex:

Female

Simulation time:

2184

4

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-151 DRAFT—DO NOT CITE OR QUOTE

-------
WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.07

Emond

0.438

0.815 (@ 2,112 hours)

0.557

CADM

-

-

-

10.7

Emond

2.46

5.12 (@ 2,112 hours)

2.65

CADM

-

-

-

107

Emond

13.4

36.4 (@ 2,112 hours)

12.7

CADM

-

-

-

321

Emond

31.6

98.6 (@2,112 hours)

28.4

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.07

Emond

67.1

107 (@2,116 hours)

91.5

CADM

59.0

92.0

88.0

10.7

Emond

683

971 (@2,117 hours)

787

CADM

767

1,000

907

107

Emond

6,784

9,010 (@2,117 hours)

7,043

CADM

8,349

10,306

8,998

321

Emond

20,218

26,379 (@2,117 hours)

20,405

CADM

25,344

31,006

26,967

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.07

Emond

156

229 (@ 2,130 hours)

225

CADM

151

210

204

10.7

Emond

885

1,155 (@2,124 hours)

1,111

CADM

689

815

774

107

Emond

4,831

5,979 (@ 2,120 hours)

5,591

CADM

2,771

3,224

2,937

321

Emond

11,420

14,037 (@2,119 hours)

12,920

CADM

6,337

7,509

6,688

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l52 DRAFT—DO NOT CITE OR QUOTE

-------
BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.07

Emond

17.0

25.5 (@ 2,116 hours)

23.9

CADM

21.0

29.0

29.0

10.7

Emond

117

159 (@ 2,116 hours)

141

CADM

119

145

135

107

Emond

852

1,103 (@ 2,116 hours)

923

CADM

727

875

778

321

Emond

2,304

2,958 (@2,116 hours)

2,419

CADM

1,961

2,370

2,080

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1.07

Emond

1.48

2.17 (@2,116 hours)

1.90

CADM

-

-

-

10.7

Emond

7.60

9.86 (@2,116 hours)

8.42

CADM

-

-

-

107

Emond

30.3

36.0 (@2,117 hours)

31.1

CADM

-

-

-

321

Emond

51.1

58.1 (@2,117 hours)

51.8

CADM

-

-

-

1

2

3	C.3.1.30. Toth et al, 1 Year (1979)

Type:

Mice

Dose:

7, 700, 7000 ng/kg/week

Strain:

Swiss/H/Riop

Route:

Oral gavage
In gastric tube

Body weight:

10 weeks old (BW=
27g)

Regime:

1/week for 1 year (365 days)

Sex:

Female and male

Simulation time:

8,760 hours

We did not simulate the scenario using the CADM model because this model can only be run for a maximum of 123
days.

4

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l53 DRAFT—DO NOT CITE OR QUOTE

-------
WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

0.573

1.61 (@ 8,736 hours)

0.682

CADM

-

-

-

100

Emond

14.2

116 (@ 8,736 hours)

15.7

CADM

-

-

-

1,000

Emond

91.2

1,108 (@ 8,736 hours)

99.3

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

94.2

131 (@ 8,743 hours)

123

CADM

-

-

-

100

Emond

7,343

10,134 (@ 8,745 hours)

9,604

CADM

-

-

-

1,000

Emond

70,243

97,658 (@ 8,745 hours)

92,506

CADM

-

-

-

FA T CONCENTRA TLONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

215

247 (@ 8,613 hours)

245

CADM

-

-

-

100

Emond

5,339

5,914 (@ 8,760 hours)

5,914

CADM

-

-

-

1,000

Emond

34,249

38,828 (@ 8,756 hours)

38,807

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

23.4

28.4 (@ 8,742 hours)

27.9

CADM

-

-

-

100

Emond

929

1,189 (@ 8,742 hours)

1,132

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l54 DRAFT—DO NOT CITE OR QUOTE

-------
1,000

Emond

7,569

10,045 (@ 8,742 hours)

9,471

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

1

Emond

1.93

2.65 (@ 8,741 hours)

2.35

CADM

-

-

-

100

Emond

31.8

58.4 (@ 2 hours)

36.7

CADM

-

-

-

1,000

Emond

78.6

103 ( a 2 hours)

84.8

CADM

-

-

-

1

2

3	C.3.1.31. Van Birgelen et ah (1995)

Type:

Rat

Dose:

0, 13.5, 26.4, 46.9, 320, 1024 ng/kg-
day

Strain:

Sprague Dawley

Route:

Oral gavage

Body weight:

150 g

Regime:

Once per day for 13 weeks

Sex:

Female

Simulation time:

2184 hours (13 weeks)

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

13.5

Emond

7.20

11.1 (@ 2,160 hours)

8.47

CADM

-

-

-

26.4

Emond

11.8

18.6 (@ 2,160 hours)

13.5

CADM

-

-

-

46.9

Emond

18.1

29.6 (@ 2,160 hours)

20.5

CADM

-

-

-

320

Emond

86.4

156 (@2,160 hours)

95.4

CADM

-

-

-

1024

Emond

250

470 (@2,160 hours)

275

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-155 DRAFT—DO NOT CITE OR QUOTE

-------
LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

13.5

Emond

1,655

2,208 (@ 2,164 hours)

2,107

CADM

-

-

-

26.4

Emond

3,228

4,216 (@ 2,164 hours)

4,017

CADM

-

-

-

46.9

Emond

5,719

7,366 (@ 2,164 hours)

7,008

CADM

-

-

-

320

Emond

38,484

47,999 (@ 2,164 hours)

45,537

CADM

-

-

-

1024

Emond

121,640

150,410 (@ 2,164 hours)

142,510

CADM

-

-

-

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

13.5

Emond

669

843 (@ 2,167 hours)

835

CADM

-

-

-

26.4

Emond

1,092

1,357 (@ 2,166 hours)

1,342

CADM

-

-

-

46.9

Emond

1,680

2,071 (@2,166 hours)

2,045

CADM

-

-

-

320

Emond

8,027

9,816 (@2,165 hours)

9,639

CADM

-

-

-

1024

Emond

23,234

28,519 (@2,165 hours)

27,954

CADM

-

-

-

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

13.5

Emond

132

173 (@ 2,164 hours)

167

CADM

-

-

-

26.4

Emond

240

308 (@ 2,164 hours)

296

CADM

-

-

-

46.9

Emond

404

513 (@2,164 hours)

492

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l56 DRAFT—DO NOT CITE OR QUOTE

-------
320

Emond

2,437

3,031 (@ 2,164 hours)

2,887

CADM

-

-

-

1024

Emond

7,521

9,310 (@ 2,164 hours)

8,846

CADM

-

-

-

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

13.5

Emond

19.9

24.2 (@ 2,164 hours)

23.4

CADM

-

-

-

26.4

Emond

29.0

34.3 (@ 2,164 hours)

33.2

CADM

-

-

-

46.9

Emond

38.8

45.0 (@ 2,164 hours)

43.7

CADM

-

-

-

320

Emond

79.1

85.2 (@ 2,164 hours)

84.1

CADM

-

-

-

1024

Emond

97.5

101 (@2,164 hours)

101

CADM

-

-

-

1

2

3	C.3.1.32. Vantlen Heuvel et a I. (1994)

Type:

Rat

Dose:

0.05, 0.1, 1, 10, 100, 1000, 10000 ng/kg/d

Strain:

Sprague Dawley

Route:

Oral gavage

Body
weight:

10 weeks old
(BW 225 to 275g, set
to 25 Og)

Regime:

Single dose

Sex:

Female

Simulation
time:

24 hours *

4	* 1 week is the minimum that can be simulated with the C ADM model, so the CADM model was not used.

5

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.05

Emond

0.01

0.011 (@ 0 hours)

0.0039

CADM

-

-

-

0.1

Emond

0.0113

0.022 (@ 0 hours)

0.008

CADM

-

-

-

1

Emond

0.106

0.215 (@ 0 hours)

0.0723

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-157 DRAFT—DO NOT CITE OR QUOTE

-------
10

Emond

0.883

2.15 (@, 0 hours)

0.583

CADM

-

-

-

100

Emond

6.45

21.5 {(cf 0 hours)

3.85

CADM

-

-

-

1000

Emond

48.3

216 ( a 0 hours)

23.9

CADM

-

-

-

10000

Emond

435

2,166 ( a 0 hours)

186

CADM

-

-

-

LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.05

Emond

0.232

0.315 ((ai 3 hours)

0.173

CADM

-

-

0.0140

0.1

Emond

0.469

0.631 ( a 3 hours)

0.353

CADM

-

-

0.0320

1

Emond

5.08

6.42 {(cf 4 hours)

4.08

CADM

-

-

0.950

10

Emond

60.2

68.7 (@, 5 hours)

54.1

CADM

-

-

52.7

100

Emond

730

800 ( a 9 hours)

719

CADM

-

-

1,342

1000

Emond

8,186

8,919 (@. 11 hours)

8,442

CADM

-

-

15,967

10000

Emond

84,254

91,675 (@. 11 hours)

88,230

CADM

-

-

162,773

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.05

Emond

0.138

0.215 (@. 24 hours)

0.215

CADM

-

-

0.780

0.1

Emond

0.274

0.427 (@. 24 hours)

0.427

CADM

-

-

1.57

1

Emond

2.58

3.97 ( a 24 hours)

3.97

CADM

-

-

15.3

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-158 DRAFT—DO NOT CITE OR QUOTE

-------
10

Emond

22.1

32.8 (@. 24 hours)

32.8

CADM

-

-

125

100

Emond

170

235 ( a 24 hours)

235

CADM

-

-

739

1000

Emond

1,348

1,720 (@. 24 hours)

1,720

CADM

-

-

5,779

10000

Emond

12,500

15,265 (@. 24 hours)

15,265

CADM

-

-

55,825

BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.05

Emond

0.0269

0.028 ((ai 9 hours)

0.0283

CADM

-

-

0.0450

0.1

Emond

0.0538

0.057 ((ci\ 9 hours)

0.0565

CADM

-

-

0.0900

1

Emond

0.536

0.568 ((ai 9 hours)

0.562

CADM

-

-

0.900

10

Emond

5.32

5.65 ( a 8 hours)

5.55

CADM

-

-

9.00

100

Emond

52.8

56.3 {(cf 1 hours)

54.4

CADM

-

-

90.0

1000

Emond

525

562 ( a 1 hours)

538

CADM

-

-

900

10000

Emond

5,238

5,610 ( a 1 hours)

5,353

CADM

-

-

9,000

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted Ave

Max

Terminal

0.05

Emond

0.0194

0.027 (@. 3 hours)

0.0142

CADM

-

-

-

0.1

Emond

0.0383

0.054 ((ai 3 hours)

0.0281

CADM

-

-

-

1

Emond

0.353

0.506 ((ai 3 hours)

0.261

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-l59 DRAFT—DO NOT CITE OR QUOTE

-------
10

Emond

2.77

4.24 (@ 2 hours)

2.08

CADM

-

-

-

100

Emond

16.1

26.4 ( a 2 hours)

12.4

CADM

-

-

-

1000

Emond

57.4

80.2 ( a 1 hours)

48.5

CADM

-

-

-

10000

Emond

100

108 ( a 1 hours)

96.1

CADM

-

-

-

1

2

3	C.3.1.33. White et al (1986)

Type:

Mice

Dose:

10, 50, 100, 500, 1000, 2000 ng/kg-day

Strain:

B6C3F1

Route:

Oral gavage

Body weight:

7 weeks old (BW set
to 23g)

Regime:

1/day for 14 days

Sex:

Female

Simulation time:

336 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted

Ave

Max

Terminal

10

Emond

1.09

2.73 (@ 312 hours)

1.42

CADM

-

-

-

50

Emond

4.08

11.6 (@ 312 hours)

4.98

CADM

-

-

-

100

Emond

7.14

21.7 (@ 312 hours)

8.44

CADM

-

-

-

500

Emond

26.8

96.5 (@312 hours)

29.8

CADM

-

-

-

1,000

Emond

48.7

187 (@312 hours)

53.1

CADM

-

-

-

2,000

Emond

90.6

365 (@312 hours)

97.5

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-160 DRAFT—DO NOT CITE OR QUOTE

-------
LIVER CONCENTRATIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted
Ave

Max

Terminal

10

Emond

216

375 (@317 hours)

343

CADM

217

468 (336h)

463

50

Emond

1,279

2,164 (@317 hours)

1,997

CADM

1,775

3,261 (336h)

3,261

100

Emond

2,707

4,525 (@ 317 hours)

4,184

CADM

3,999

6,923 (336h)

6,923

500

Emond

14,802

24,165 (@317 hours)

22,383

CADM

22,705

36,362 (336h)

36,362

1,000

Emond

30,278

49,034 (@317 hours)

45,414

CADM

46,309

73,145 (336h)

73,145

2,000

Emond

61,381

98,703 (@317 hours)

91,363

CADM

93,577

146,695 (336h)

146,695

FA T CONCENTRA TIONS (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted
Ave

Max

Terminal

10

Emond

279

507 (@ 336 hours)

507

CADM

316

537 (336h)

537

50

Emond

1,056

1,846 (@ 336 hours)

1,846

CADM

1,029

1,564 (336h)

1,564

100

Emond

1,854

3,195 (@ 333 hours)

3,195

CADM

1,662

2,470 (336h)

2,470

500

Emond

7,008

11,868 (@324 hours)

11,816

CADM

5,711

8,594 (336h)

8,594

1,000

Emond

12,746

21,566 (@ 323 hours)

21,424

CADM

10,498

15,993 (336h)

15,993

2,000

Emond

23,691

40,177 (@ 322 hours)

39,843

CADM

19,990

30,726 (336h)

30,726

This document is a draft for review purposes only and does not constitute Agency policy.

C-161 DRAFT—DO NOT CITE OR QUOTE

-------
BODY BURDEN (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted

Ave

Max

Terminal

10

Emond

37.7

65.9 (@ 317 hours)

63.8

CADM

47.9

85.9 (336h)

85.9

50

Emond

175

297 (@ 317 hours)

284

CADM

207

342 (336h)

342

100

Emond

338

570 (@ 316 hours)

542

CADM

388

624 (336h)

624

500

Emond

1,597

2,637 (@ 316 hours)

2,480

CADM

1,761

2,754 (336h)

2,754

1,000

Emond

3,137

5,153 (@316 hours)

4,830

CADM

3,455

5,387 (336h)

5,387

2,000

Emond

6,186

10,118 (@316 hours)

9,459

CADM

6,836

10,643 (336h)

10,643

BOUND LIVER (ng/kg)

Dose
(ng/kg-day)
Adjusted dose

Model

Metric

Time-weighted

Ave

Max

Terminal

10

Emond

3.49

5.32 (@316 hours)

4.82

CADM

-

-

-

50

Emond

11.4

16.4 (@317 hours)

15.1

CADM

-

-

-

100

Emond

18.1

25.1 (@317 hours)

23.4

CADM

-

-

-

500

Emond

44.2

56.2 (@317 hours)

53.8

CADM

-

-

-

1,000

Emond

59.3

71.9 (@317 hours)

69.7

CADM

-

-

-

2,000

Emond

74.4

86.1 (@317 hours)

84.3

CADM

-

-

-

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

C-162 DRAFT—DO NOT CITE OR QUOTE

-------
1	C.3.2. Gestational Studies

2	C.3.2.1. Bell et aL (2007)

Type:

Rat

Dose:

2.4, 8, and 46 ng/kg-day with a 0.03 ng/kg-day
background

Strain:

Han/Wistar

Route:

Diet oral dose

Body weight:

6 weeks
(BW= 85g)

Regime:

Once per day for 12 weeks prior to mating, during the
two week mating period, and during gestation

Sex:

Female

Simulation
time:

2,352 hr (98 days) prior to gestation + 504 hr (21 days)
during gestation for a total simulation of 2,856 hours

3	* Time averages are computed during the gestation period only.

4

WHOLE BLOOD CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2.43

2.20

6,295

3.10 (@ 2,352 hours)

2.20

8.03

5.14

14,674

7.31 (@ 2,352 hours)

5.08

46.03

18.4

52,584

28.1 (@ 2,352 hours)

18.1

LIVER CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2.43

320

914,290

437 (@ 2,356 hours)

321

8.03

1,040

2,969,800

1,349 (@2,356
hours)

1,042

46.03

5,892

16,829,000

7,289 (@ 2,356
hours)

6,007

FAT CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2.43

205

585,530

263 (@ 2,336 hours)

211

8.03

478

1,365,100

589 (@ 2,335 hours)

486

46.03

1,713

4,891,500

2,045 (@ 2,334
hours)

1,745

This document is a draft for review purposes only and does not constitute Agency policy.

C-163 DRAFT—DO NOT CITE OR QUOTE

-------
BODY BURDEN (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2.43

33.0

94,390

44.4 (@ 2,836 hours)

43.4

8.03

90.4

258,110

117 (@ 2,836 hours)

114

46.03

422

1,206,500

531 (@ 2,836 hours)

511

FETUS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2.43

3.03

8,648

39.6 (@ 2,530 hours)

6.48

8.03

6.65

18,999

86.7 (@ 2,529 hours)

14.4

46.03

20.9

59,794

272 (@ 2,527 hours)

46.0

BOUND LIVER (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2.43

7.10

20,289

8.98 (@ 2,356 hours)

7.23

8.03

15.1

43,242

18.2 (@ 2,356 hours)

15.4

46.03

39.6

113,070

44.8 (@ 2,356 hours)

40.6

1

2

3	C.3.2.2. Haavisto et al. (2006)

Type:

Rat

Dose:

20, 400, and 1,000 ng/kg

Strain:

Sprague Dawley

Route:

Oral exposure

Body weight

BW = 190 g

Regime:

Single dose on GDI3

Sex:

Female

Simulation time

336 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

2.86

68.9

8.01 (@312 hours)

1.73

400

11.3

273

40.1 (@312 hours)

6.28

1000

46.9

1,129

202 (@312 hours)

22.8

This document is a draft for review purposes only and does not constitute Agency policy.

C-164 DRAFT—DO NOT CITE OR QUOTE

-------
LIVER CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

265

6,371

298 (@319 hours)

244

400

1,497

36,005

1,653 (@ 320 hours)

1,462

1000

8,061

193,860

8,832 (@321 hours)

8,147

FAT CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

56.3

1,354

81.9 (@ 336 hours)

81.9

400

232

5,584

321 (@ 336 hours)

321

1000

1,002

24,084

1,313 (@336 hours)

1,313

BODY BURDEN (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

21.1

508

22.5 (@319 hours)

21.9

400

105

2,528

112 (@319 hours)

108

1000

524

12,612

561 (@319 hours)

538

FETUS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

8.47

203

11.3 (@ 336 hours)

11.3

400

31.2

751

40.3 (@ 336 hours)

40.3

1000

112

2,689

139 (@ 336 hours)

139

BOUND LIVER (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

8.20

197

13.5 (@ 314 hours)

6.03

400

24.9

598

40.8 (@313 hours)

19.1

1000

57.1

1,373

80.1 (@313 hours)

47.7

This document is a draft for review purposes only and does not constitute Agency policy.

C-165 DRAFT—DO NOT CITE OR QUOTE

-------
1 C.3.2.3. Hojo et al. (2002)

Type:

Rat

Dose:

20, 60 and 180 ng/kg

Strain:

Sprague Dawley

Route:

Oral exposure

Body weight

20 ng/kg BW = 27 lg
60 ng/kg BW = 275g
180 ng/kg BW = 262g

Regime:

Single dose on GD8

Sex:

Female

Simulation time

216 hours

2

WHOLE BLOOD CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

1.62

39.1

4.47 (@ 192 hours)

1.02

60

4.17

100

13.3 (@ 192 hours)

2.50

180

10.7

258

40.3 (@ 192 hours)

5.96

LIVER CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

128

20,554

144 (@ 198 hours)

43.2

60

420

72,340

465 (@ 200 hours)

147

180

1,364

250,820

1,497 (@ 201 hours)

497

FAT CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

32.5

17,253

63.0 (@ 281 hours)

49.4

60

86.4

44,093

161 (@ 284 hours)

124

180

226

108,730

398 (@ 286 hours)

301

BODY BURDEN (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

10.6

3,054

11.3 (@ 200 hours)

8.67

60

31.8

8,702

33.8 (@ 199 hours)

23.6

180

95.0

24,747

101 (@ 199 hours)

63.4

This document is a draft for review purposes only and does not constitute Agency policy.

C-166 DRAFT—DO NOT CITE OR QUOTE

-------
FETUS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

15.9

2,334

18.4 (@ 206 hours)

1.64

60

39.8

5,829

45.7 (@ 205 hours)

4.10

180

96.3

13,866

110 (@ 203 hours)

9.72

BOUND LIVER (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

4.88

759

7.74 (@ 194 hours)

1.75

60

11.2

1,848

18.5 (@ 194 hours)

4.26

180

23.6

4,157

38.5 (@ 193 hours)

9.65

1

2

3	C.3.2.4. Ikeda et al. (2005)

Type:

Rat

Dose:

400 ng/kg single dose and 80 ng/kg weekly
maintenance dose

Strain:

Sprague Dawley

Route:

Oral gavage

Body weight:

10 weeks
(BW= 25 Og)

Regime:

400 ng/kg single dose, two weekly maintenance
doses prior to gestation and weekly maintenance
doses during gestation

Sex:

Female

Simulation
time:

504 hr (21 days) prior to gestation + 504 hr (21 days)
during gestation for a total simulation of 1,008 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

16.5

22.9

23,086

101 (@ 144 hours)

10.1

LIVER CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

16.5

7,755

7,817,300

17,016 (@ 150 hours)

2,698

This document is a draft for review purposes only and does not constitute Agency policy.

C-167 DRAFT—DO NOT CITE OR QUOTE

-------
FAT CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

16.5

2,087

2,103,900

3,663 (@ 184 hours)

1,028

BODY BURDEN (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

16.5

548

552,590

1,085 (@ 149 hours)

262

FETUS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

16.5

45.9

46,290

245 (@ 679 hours)

30.2

BOUND LIVER (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

16.5

44.0

44,361

63.8 (@ 149 hours)

26.8

1

2

3	C.3.2.5. Kattainen et al. (2001)

Type:

Rat

Dose:

30, 100, 300, and 1,000 ng/kg

Strain:

Han/Wistar (Kuopio)
and Long/Evans
(Turku/AB) crossing.

Route:

Oral exposure

Body weight:

BW no specify
(BW set to 190g)*

Regime:

Single dose in the GDI5

Sex:

Female

Simulation
time:

360 hours

4	*Derelanko and Hollinger (1995).

5

WHOLE BLOOD CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

2.23

53.7

5.95 (@ 336 hours)

1.36

100

6.25

150

19.8 (@ 336 hours)

3.62

This document is a draft for review purposes only and does not constitute Agency policy.

C-168 DRAFT—DO NOT CITE OR QUOTE

-------
300

16.1

387

59.8 (@ 336 hours)

8.62

1,000

46.9

1,128

200 (@ 336 hours)

22.7

LIVER CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

193

4,648

219 (@ 342 hours)

175

100

713

17,141

793 (@ 344 hours)

680

300

2,298

55,266

2,533 (@ 345 hours)

2,267

1,000

8,055

193,720

8,831 (@345 hours)

8,134

FAT CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

42.8

1,027

62.8 (@ 360 hours)

62.8

100

123

2,964

175 (@ 360 hours)

175

300

327

7,853

446 (@ 360 hours)

446

1,000

981

23,588

1,289 (@ 360 hours)

1,289

BODY BURDEN (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

15.9

382

16.9 (@ 343 hours)

16.4

100

52.7

1,266

56.2 (@ 343 hours)

54.3

300

158

3,791

168 (@ 343 hours)

162

1,000

524

12,612

561 (@ 343 hours)

538

FETUS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

4.86

117

6.66 (@ 360 hours)

6.66

100

13.2

317

17.6 (@ 360 hours)

17.6

300

31.5

758

41.2 (@ 360 hours)

41.2

1,000

82.2

1,975

104 (@ 360 hours)

104

This document is a draft for review purposes only and does not constitute Agency policy.

C-169 DRAFT—DO NOT CITE OR QUOTE

-------
BOUND LIVER (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

6.57

158

10.7 (@ 338 hours)

4.80

100

15.8

381

26.3 (@ 338 hours)

11.9

300

31.6

760

50.6 (@ 337 hours)

24.7

1,000

57.1

1,373

80.1 (@ 337 hours)

47.7

1

2

3	C.3.2.6. Keller et al. (2007)

Type:

Mouse

Dose:

10, 100, and 1000 ng/kg

Strain:

CBA/J and C3H/HeJ

Route:

Oral

Body weight:

Not specified (24 g
used in the simulation)

Regime:

Single dose at gestation day 13

Sex:

Female

Simulation time:

336 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

10

0.537

12.9

1.43 (@312 hours)

0.269

100

4.29

103

14.3 (@312 hours)

1.95

1,000

34.1

820

143 (@312 hours)

12.3

LIVER CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

10

30.6

737

39.8 (@316 hours)

22.2

100

371

8,922

421 (@319 hours)

317

1,000

4,214

101,360

4,697 (@ 321 hours)

3,940

FAT CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

10

22.4

538

33.3 (@ 336 hours)

33.3

100

188

4,523

264 (@ 336 hours)

264

1,000

1,591

38,233

2,080 (@ 336 hours)

2,080

This document is a draft for review purposes only and does not constitute Agency policy.

C-170 DRAFT—DO NOT CITE OR QUOTE

-------
BODY BURDEN (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

10

5.57

134

5.99 (@319 hours)

5.72

100

54.3

1,306

59.0 (@318 hours)

54.7

1,000

530

12,747

581 (@318 hours)

524

FETUS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

10

2.57

61.7

3.80 (@ 336 hours)

3.80

100

21.7

522

30.0 (@ 334 hours)

29.9

1,000

179

4,312

233 (@ 329 hours)

225

BOUND LIVER (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

10

1.74

41.8

3.14 (@315 hours)

1.01

100

11.5

276

23.5 (@314 hours)

6.99

1,000

46.7

1,123

79.8 (@314 hours)

32.9

1

2

3	C.3.2.7. Li et al (2006) 3-Day

Type:

Mouse

Dose:

2, 50, and 100 ng/kg-day

Strain:

NIH

Route:

Oral

Body weight:

25-28 g (used 27 g in
the simulation)

Regime:

Daily exposure from gestation day 1 to
gestation day 8

Sex:

Female

Simulation time:

72 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2

0.159

11.4

0.392 (@ 48 hours)

0.136

50

2.84

205

8.90 (@ 48 hours)

2.38

100

5.12

369

17.3 (@ 48 hours)

4.20

This document is a draft for review purposes only and does not constitute Agency policy.

C-171 DRAFT—DO NOT CITE OR QUOTE

-------
LIVER CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2

8.98

647

15.1 (@ 52 hours)

9.10

50

333

23,971

539 (@ 53 hours)

402

100

718

51,738

1,156 (@ 53 hours)

888

FAT CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2

17.0

1,227

31.1 (@ 72 hours)

31.1

50

315

22,704

548 (@ 72 hours)

548

100

576

41,460

984 (@ 72 hours)

984

BODY BURDEN (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2

2.29

165

3.51 (@ 55 hours)

3.43

50

53.6

3,863

82.2 (@ 54 hours)

77.1

100

105

7,598

162 (@ 53 hours)

150

FETUS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2

0.00

0

0.000 (@ 72 hours)

0.00

50

0.0

0

0.000 (@ 72 hours)

0.00

100

0.0

0

0.000 (@ 72 hours)

0.00

BOUND LIVER (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

2

0.538

38.8

0.864 (@51 hours)

0.498

50

8.24

594

13.5 (@ 2 hours)

8.16

100

13.6

981

23.7 (@ 2 hours)

13.6

1

2

This document is a draft for review purposes only and does not constitute Agency policy.

C-172 DRAFT—DO NOT CITE OR QUOTE

-------
1 C.3.2.8. Markowski et al. (2001)

Type:

Rat

Dose:

20, 60 and 180 ng/kg

Strain:

Holtzman rats

Route:

Oral exposure

Body weight:

BW no specify
(BW set to 190g)*

Regime:

Single dose in the GD18

Sex:

Female

Simulation
time:

432 hours

2	*Derelanko and Hollinger (1995).

3

WHOLE BLOOD CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

1.56

37.5

3.82 (@ 408 hours)

0.958

60

4.03

97.0

11.5 (@ 408 hours)

2.38

180

10.3

248

34.8 (@ 408 hours)

5.72

LIVER CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

123

2,959

141 (@ 414 hours)

109

60

409

9,843

459 (@ 415 hours)

382

180

1,334

32,086

1,479 (@ 416 hours)

1,295

FAT CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

27.9

670

41.6 (@ 432 hours)

41.6

60

74.0

1,778

107 (@ 432 hours)

107

180

195

4,685

273 (@ 432 hours)

273

BODY BURDEN (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

10.6

254

11.2 (@415 hours)

10.9

60

31.7

762

33.8 (@415 hours)

32.7

180

94.7

2,278

101 (@ 415 hours)

97.5

This document is a draft for review purposes only and does not constitute Agency policy.

C-173 DRAFT—DO NOT CITE OR QUOTE

-------
FETUS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

1.26

30.2

1.80 (@ 432 hours)

1.80

60

3.21

77.2

4.49 (@ 432 hours)

4.49

180

7.81

188

10.7 (@ 432 hours)

10.7

BOUND LIVER (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

20

4.74

114

7.59 (@410 hours)

3.43

60

11.0

265

18.2 (@410 hours)

8.16

180

23.2

559

38.1 (@ 409 hours)

17.7

1

2

3	C.3.2.9. Mietinnen et al. (2006)

Type:

Rat

Dose:

30, 100, 300 and 1000 ng/kg

Strain:

cross-breeding of
Han/Wistar and Long-
Evans rats

Route:

Oral exposure

Body weight:

BW 11 weeks
(BW set to 180g)

Regime:

Single dose in the GDI5

Sex:

Female

Simulation
time:

360 hours

4

WHOLE BLOOD CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

2.22

53.4

5.87 (@ 336 hours)

1.36

100

6.23

150

19.6 (@ 336 hours)

3.61

300

16.0

386

59.0 (@ 336 hours)

8.61

1,000

46.6

1,123

198 (@ 336 hours)

22.7

This document is a draft for review purposes only and does not constitute Agency policy.

C-174 DRAFT—DO NOT CITE OR QUOTE

-------
LIVER CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

193

4,631

219 (@ 342 hours)

174

100

711

17,096

791 (@ 344 hours)

677

300

2,294

55,166

2,530 (@ 345 hours)

2,260

1,000

8,042

193,410

8,820 (@ 345 hours)

8,114

FAT CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

43.0

1,034

63.2 (@ 360 hours)

63.2

100

124

2,984

176 (@ 360 hours)

176

300

329

7,905

449 (@ 360 hours)

449

1,000

987

23,729

1,296 (@ 360 hours)

1,296

BODY BURDEN (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

15.9

381

16.9 (@ 343 hours)

16.4

100

52.6

1,266

56.1 (@ 343 hours)

54.3

300

158

3,791

168 (@ 343 hours)

162

1,000

524

12,609

561 (@ 343 hours)

538

FETUS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

4.83

116

6.62 (@ 360 hours)

6.62

100

13.1

315

17.5 (@ 360 hours)

17.5

300

31.3

753

41.0 (@ 360 hours)

41.0

1,000

81.7

1,963

104 (@ 360 hours)

104

BOUND LIVER (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

30

6.56

158

10.7 (@ 338 hours)

4.78

This document is a draft for review purposes only and does not constitute Agency policy.

C-175 DRAFT—DO NOT CITE OR QUOTE

-------
100

15.8

381

26.3 (@ 338 hours)

11.9

300

31.6

760

50.5 (@ 337 hours)

24.6

1,000

57.0

1,372

80.1 (@ 337 hours)

47.6

1

2

3	C.3.2.10. Noharaetal. (2000)

Type:

Rat

Dose:

12.5, 50, 200 or 800 ng TCDD/kg

Strain:

Holtzman rats

Route:

Oral exposure

Body weight:

BW no specify
(BW set to 190g)*

Regime:

Single dose in the GDI5

Sex:

Female

Simulation
time:

360 hours

4	*Derelanko and Hollinger (1995).

5

WHOLE BLOOD CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

1.03

24.8

2.44 (@ 336 hours)

0.645

50

3.45

82.9

9.78 (@ 336 hours)

2.07

200

11.3

271

39.2 (@ 336 hours)

6.25

800

38.1

918

158 (@ 336 hours)

18.9

LIVER CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

73.8

1,776

86.1 (@341 hours)

63.6

50

336

8,084

378 (@ 343 hours)

311

200

1,492

35,890

1,651 (@ 344 hours)

1,454

800

6,389

153,640

7,012 (@ 345 hours)

6,423

FAT CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

19.7

473

29.5 (@ 360 hours)

29.5

50

67.6

1,624

97.8 (@ 360 hours)

97.8

200

229

5,504

317 (@360 hours)

317

800

803

19,292

1,061 (@ 360 hours)

1,061

This document is a draft for review purposes only and does not constitute Agency policy.

C-176 DRAFT—DO NOT CITE OR QUOTE

-------
BODY BURDEN (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

6.62

159

7.04 (@ 343 hours)

6.88

50

26.4

635

28.1 (@ 343 hours)

27.3

200

105

2,528

112 (@343 hours)

108

800

420

10,092

449 (@ 343 hours)

430

FETUS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

2.25

54.0

3.14 (@ 360 hours)

3.14

50

7.43

179

10.1 (@ 360 hours)

10.1

200

22.8

548

30.1 (@ 360 hours)

30.1

800

68.1

1,638

87.0 (@ 360 hours)

87.0

BOUND LIVER (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

3.24

77.9

5.12 (@338 hours)

2.32

50

9.66

232

16.0 (@ 338 hours)

7.12

200

24.8

597

40.7 (@ 337 hours)

19.0

800

51.9

1,248

75.0 (@ 337 hours)

42.7

1

2

3	C.3.2.11. Ohsako et al. (2001)

Type:

Rat

Dose:

12.5, 50, 200, and 800 ng/kg-day

Strain:

Holtzmann

Route:

Oral exposure on GDI5

Body weight

10 weeks (200g)

Regime:

Single dose

Sex:

Female

Simulation time

384 hours

4

This document is a draft for review purposes only and does not constitute Agency policy.

C-177 DRAFT—DO NOT CITE OR QUOTE

-------
WHOLE BLOOD CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

1.04

25.0

2.48 (@ 360 hours)

0.649

50

3.47

83.6

9.93 (@ 360 hours)

2.07

200

11.4

273

39.9 (@ 360 hours)

6.26

800

38.4

925

161 (@ 360 hours)

18.9

LIVER CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

74.3

1,788

86.5 (@ 365 hours)

64.2

50

338

8,126

379 (@ 367 hours)

314

200

1,497

36,006

1,655 (@ 368 hours)

1,461

800

6,402

153,960

7,025 (@ 369 hours)

6,443

FAT CONCENTRATIONS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

19.0

457

28.6 (@ 384 hours)

28.6

50

65.3

1,569

94.7 (@ 384 hours)

94.7

200

221

5,321

307 (@ 384 hours)

307

800

777

18,671

1,029 (@ 384 hours)

1,029

BODY BURDEN (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

6.63

159

7.05 (@ 367 hours)

6.89

50

26.4

635

28.2 (@ 367 hours)

27.3

200

105

2,529

112 (@ 367 hours)

108

800

420

10,093

449 (@ 367 hours)

430

FETUS (ng/kg) and AUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

1.65

39.5

2.33 (@ 384 hours)

2.33

This document is a draft for review purposes only and does not constitute Agency policy.

C-178 DRAFT—DO NOT CITE OR QUOTE

-------
50

5.44

131

7.48 (@ 384 hours)

7.48

200

16.7

401

22.3 (@ 384 hours)

22.3

800

49.9

1,200

64.6 (@ 384 hours)

64.6

BOUND LIVER (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

12.5

3.25

78.3

5.13 (@362 hours)

2.34

50

9.69

233

16.0 (@ 362 hours)

7.16

200

24.9

598

40.7 (@ 361 hours)

19.1

800

51.9

1,249

75.0 (@ 361 hours)

42.8

C.3.2.12. Schantz et al. (1996) and Amin et al. (2000)

Type:

Rat

Dose:

25 and 100 ng/kg-day

Strain:

Sprague Dawley

Route:

Oral exposure

Body weight:

BW not specified
(BW set to 25 Og)

Regime:

Daily doses from GD 10-16

Sex:

Female

Simulation time:

384 hours; time averages are calculated
from the beginning of the dosing

4

WHOLE BLOOD CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

3.38

487

8.63 (@ 360 hours)

4.03

100

10.6

1,522

31.1 (@360 hours)

12.3

LIVER CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

512

73,686

871 (@ 365 hours)

778

100

2,374

341,960

4,012 (@366 hours)

3,665

FAT CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

169

24,323

306 (@ 384 hours)

306

This document is a draft for review purposes only and does not constitute Agency policy.

C-179 DRAFT—DO NOT CITE OR QUOTE

-------
100

532

76,675

950 (@ 384 hours)

950

BODY BURDEN (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

45.1

6,490

76.6 (@ 365 hours)

74.3

100

111

25,438

298 (@ 365 hours)

287

FETUS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

25.2

3,627

30.4 (@ 343 hours)

27.3

100

74.1

10,672

88.1 (@342 hours)

77.9

BOUND LIVER (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

9.99

1,439

14.4 (@ 364 hours)

12.8

100

25.2

3,632

34.2 (@ 364 hours)

31.6

1

2

3	C.3.2.13. Seoetal. (1995)

Type:

Rat

Dose:

25 and 100 ng/kg-day

Strain:

Sprague Dawley

Route:

Oral exposure

Body weight:

BW not specified
(BW set to 190g)

Regime:

Daily doses from GD 10-16

Sex:

Female

Simulation time:

384 hours; time averages are calculated
from the beginning of the dosing

4

WHOLE BLOOD CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

3.33

479

8.25 (@ 360 hours)

4.00

100

10.4

1,498

29.6 (@ 360 hours)

12.2

This document is a draft for review purposes only and does not constitute Agency policy.

C-l 80 DRAFT—DO NOT CITE OR QUOTE

-------
LIVER CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

504

72,592

861 (@ 365 hours)

767

100

2,347

337,970

3,978 (@ 365 hours)

3,627

FAT CONCENTRATIONS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

172

24,807

310 (@384 hours)

310

100

542

78,097

962 (@ 384 hours)

962

BODY BURDEN (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

45.0

6,486

76.5 (@ 365 hours)

74.2

100

176

25,387

298 (@ 365 hours)

287

FETUS (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

24.7

3,551

29.8 (@ 343 hours)

26.8

100

72.6

10,456

86.6 (@ 342 hours)

76.8

BOUND LIVER (ng/kg) andAUC ((ng/kg) • hr)

Dose
(ng/kg-day)
Adjusted dose

Metric

Time-weighted Ave

Area Under the
Curve

Max

Terminal

25

9.90

1,426

14.3 (@ 364 hours)

12.7

100

25.0

3,607

34.1 (@ 364 hours)

31.4

This document is a draft for review purposes only and does not constitute Agency policy.

C-181 DRAFT—DO NOT CITE OR QUOTE

-------
1	Table C-l. Model input parameters potentially addressed by selected

2	articles

3

Articles

Model input parameters potentially addressed

Absorption

Desorption

Distribution

Elimination

Kinetics

Induction CYP1A1

Interspecies
differences

Age Differences

Aryl hydrocarbon
receptor (AhR)

Mode of action

Partition
coefficient

Aylward et al., 2004

•

•

•

•

•













Aylward et al., 2005a, b

•

•

•

•

•













Aylward et al., 2009







•















Bohonowych and Denison, 2007











•

•



•





Boverhof et al., 2005











•

•









Connor and Aylward, 2006













•

•

•





Heinzl et al., 2007





•











•





Irigaray et al., 2005





•







•









Kerger et al., 2006





•



•





•







Kerger et al., 2007















•







Kim et al., 2003





•

















Korenaga et al., 2007











•

•









Korkalainen et al., 2004













•

•







Kransler et al., 2007













•

•







Maruyama et al., 2002

•



•

•















Maruyama et al., 2003

•



•

•















Maruyama and Aoki, 2006

•



•

•















Millbrath et al., 2009





•

•

•



•









Moser and McLachlan, 2002



•



•















Mullerova and Kopecky, 2007





•

















Nadal et al., 2009







•

•













Nohara et al., 2006













•



•





Olsman et al., 2007

















•





Saghir et al., 2005





•

•

•













Schecter et al., 2003







•







•







Staskal et al., 2005











•





•





Toyoshiba et al., 2004





•





•





•





Wilkes et al., 2008











•











4	Partition coefficient estimates and CYP parameter value estimates were derived from Wang et al. (1997, 2000) and

5	Santostefano etal. (1998).

This document is a draft for review purposes only and does not constitute Agency policy.

C-182 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

C.4. RESPONSE SURFACE TABLES

In order to calculate human equivalent doses, the human model must be run with a daily
intake which gives average blood concentrations which match the average concentrations in the
rodent models. However, such calculation can require numerous human model runs with
repeated intake adjustments in order to reach the target blood concentrations. To facilitate this
process, a response surface was created for the human model. In the response surface, numerous
intakes were run and the blood, fat, and body burden average concentrations were recorded.
These tables can then be used to estimate the intake which would give a target blood
concentration. The two closest intakes are found and the intake is estimated by linearly
interpolating between the two doses. Then, this intake is run through the human model to
confirm that the average blood concentration is within a specified tolerance of the target blood
concentration.

For the current analysis, three different response surfaces were created: non-gestational
lifetime to be used with long-term animal bioassays, nongestational five year average runs to be
used with shorter term animal bioassays, and gestationsl to be used with gestational animal
bioassays. All three response sufraces are shown in the following tables.

This document is a draft for review purposes only and does not constitute Agency policy.

C-l 83 DRAFT—DO NOT CITE OR QUOTE

-------
C.4.1. Nongestational Lifetime

O



Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

1.00E-09

2.39E-05

8.58E-06

2.52E-07

1.33E-09

3.18E-05

1.14E-05

3.35E-07

1.67E-09

3.98E-05

1.43E-05

4.19E-07

2.00E-09

4.77E-05

1.72E-05

5.03E-07

2.33E-09

5.57E-05

2.00E-05

5.87E-07

2.67E-09

6.36E-05

2.29E-05

6.70E-07

3.00E-09

7.16E-05

2.57E-05

7.54E-07

3.33E-09

7.95E-05

2.86E-05

8.38E-07

3.67E-09

8.74E-05

3.14E-05

9.22E-07

4.00E-09

9.54E-05

3.43E-05

1.01E-06

4.33E-09

1.03E-04

3.72E-05

1.09E-06

4.67E-09

1.11E-04

4.00E-05

1.17E-06

5.00E-09

1.19E-04

4.29E-05

1.26E-06

5.33E-09

1.27E-04

4.57E-05

1.34E-06

5.67E-09

1.35E-04

4.86E-05

1.42E-06

6.00E-09

1.43E-04

5.14E-05

1.51E-06

6.33E-09

1.51E-04

5.43E-05

1.59E-06

6.67E-09

1.59E-04

5.71E-05

1.68E-06

7.00E-09

1.67E-04

6.00E-05

1.76E-06

7.33E-09

1.75E-04

6.29E-05

1.84E-06

7.67E-09

1.83E-04

6.57E-05

1.93E-06

8.00E-09

1.91E-04

6.86E-05

2.01E-06

8.33E-09

1.99E-04

7.14E-05

2.09E-06

8.67E-09

2.07E-04

7.43E-05

2.18E-06

9.00E-09

2.14E-04

7.71E-05

2.26E-06

9.33E-09

2.22E-04

8.00E-05

2.34E-06

9.67E-09

2.30E-04

8.28E-05

2.43E-06

1.00E-08

2.38E-04

8.57E-05

2.51E-06

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

1.33E-08

3.17E-04

1.14E-04

3.34E-06

1.67E-08

3.96E-04

1.43E-04

4.18E-06

2.00E-08

4.75E-04

1.71E-04

5.01E-06

2.33E-08

5.54E-04

1.99E-04

5.84E-06

2.67E-08

6.33E-04

2.28E-04

6.67E-06

3.00E-08

7.12E-04

2.56E-04

7.50E-06

3.33E-08

7.91E-04

2.85E-04

8.34E-06

3.67E-08

8.70E-04

3.13E-04

9.17E-06

4.00E-08

9.49E-04

3.41E-04

1.00E-05

4.33E-08

1.03E-03

3.70E-04

1.08E-05

4.67E-08

1.11E-03

3.98E-04

1.17E-05

5.00E-08

1.19E-03

4.27E-04

1.25E-05

5.33E-08

1.26E-03

4.55E-04

1.33E-05

5.67E-08

1.34E-03

4.83E-04

1.41E-05

6.00E-08

1.42E-03

5.12E-04

1.50E-05

6.33E-08

1.50E-03

5.40E-04

1.58E-05

6.67E-08

1.58E-03

5.68E-04

1.66E-05

7.00E-08

1.66E-03

5.96E-04

1.75E-05

7.33E-08

1.73E-03

6.25E-04

1.83E-05

7.67E-08

1.81E-03

6.53E-04

1.91E-05

8.00E-08

1.89E-03

6.81E-04

1.99E-05

8.33E-08

1.97E-03

7.10E-04

2.08E-05

8.67E-08

2.05E-03

7.38E-04

2.16E-05

9.00E-08

2.13E-03

7.66E-04

2.24E-05

9.33E-08

2.21E-03

7.94E-04

2.32E-05

9.67E-08

2.28E-03

8.23E-04

2.41E-05

1.00E-07

2.36E-03

8.51E-04

2.49E-05

1.33E-07

3.14E-03

1.13E-03

3.31E-05

1.67E-07

3.92E-03

1.41E-03

4.13E-05

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

2.00E-07

4.70E-03

1.70E-03

4.96E-05

2.33E-07

5.48E-03

1.98E-03

5.78E-05

2.67E-07

6.26E-03

2.26E-03

6.60E-05

3.00E-07

7.04E-03

2.54E-03

7.42E-05

3.33E-07

7.82E-03

2.82E-03

8.24E-05

3.67E-07

8.60E-03

3.10E-03

9.06E-05

4.00E-07

9.38E-03

3.38E-03

9.89E-05

4.33E-07

1.02E-02

3.66E-03

1.07E-04

4.67E-07

1.09E-02

3.95E-03

1.15E-04

5.00E-07

1.17E-02

4.23E-03

1.24E-04

5.33E-07

1.25E-02

4.50E-03

1.31E-04

5.66E-07

1.32E-02

4.78E-03

1.39E-04

5.99E-07

1.40E-02

5.05E-03

1.47E-04

6.33E-07

1.47E-02

5.32E-03

1.55E-04

6.66E-07

1.55E-02

5.60E-03

1.63E-04

6.99E-07

1.63E-02

5.87E-03

1.71E-04

7.32E-07

1.70E-02

6.15E-03

1.79E-04

7.65E-07

1.78E-02

6.42E-03

1.87E-04

7.98E-07

1.85E-02

6.69E-03

1.95E-04

8.32E-07

1.93E-02

6.97E-03

2.03E-04

8.65E-07

2.00E-02

7.24E-03

2.11E-04

8.98E-07

2.08E-02

7.52E-03

2.19E-04

9.31E-07

2.16E-02

7.79E-03

2.27E-04

9.64E-07

2.23E-02

8.07E-03

2.35E-04

9.97E-07

2.31E-02

8.34E-03

2.43E-04

1.01E-06

2.34E-02

8.46E-03

2.47E-04

1.03E-06

2.37E-02

8.59E-03

2.50E-04

1.04E-06

2.41E-02

8.71E-03

2.54E-04

1.06E-06

2.44E-02

8.84E-03

2.58E-04

-------
o



Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

1.07E-06

2.48E-02

8.97E-03

2.61E-04

1.09E-06

2.52E-02

9.10E-03

2.65E-04

1.11E-06

2.55E-02

9.23E-03

2.69E-04

1.12E-06

2.59E-02

9.37E-03

2.73E-04

1.14E-06

2.63E-02

9.51E-03

2.77E-04

1.16E-06

2.67E-02

9.65E-03

2.81E-04

1.17E-06

2.70E-02

9.79E-03

2.85E-04

1.19E-06

2.74E-02

9.93E-03

2.89E-04

1.21E-06

2.78E-02

1.01E-02

2.93E-04

1.23E-06

2.82E-02

1.02E-02

2.98E-04

1.24E-06

2.87E-02

1.04E-02

3.02E-04

1.26E-06

2.91E-02

1.05E-02

3.06E-04

1.28E-06

2.95E-02

1.07E-02

3.11E-04

1.30E-06

2.99E-02

1.08E-02

3.15E-04

1.32E-06

3.04E-02

1.10E-02

3.20E-04

1.34E-06

3.08E-02

1.12E-02

3.25E-04

1.36E-06

3.13E-02

1.13E-02

3.29E-04

1.38E-06

3.17E-02

1.15E-02

3.34E-04

1.40E-06

3.22E-02

1.16E-02

3.39E-04

1.42E-06

3.26E-02

1.18E-02

3.44E-04

1.44E-06

3.31E-02

1.20E-02

3.49E-04

1.46E-06

3.36E-02

1.22E-02

3.54E-04

1.49E-06

3.41E-02

1.24E-02

3.59E-04

1.53E-06

3.51E-02

1.27E-02

3.70E-04

1.58E-06

3.61E-02

1.31E-02

3.81E-04

1.62E-06

3.72E-02

1.35E-02

3.92E-04

1.67E-06

3.83E-02

1.39E-02

4.03E-04

1.72E-06

3.94E-02

1.43E-02

4.15E-04

1.77E-06

4.05E-02

1.47E-02

4.27E-04

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

1.83E-06

4.17E-02

1.51E-02

4.39E-04

1.88E-06

4.29E-02

1.56E-02

4.52E-04

1.94E-06

4.41E-02

1.60E-02

4.65E-04

2.00E-06

4.54E-02

1.65E-02

4.79E-04

2.06E-06

4.67E-02

1.70E-02

4.93E-04

2.12E-06

4.81E-02

1.75E-02

5.07E-04

2.18E-06

4.95E-02

1.80E-02

5.22E-04

2.25E-06

5.09E-02

1.85E-02

5.37E-04

2.32E-06

5.24E-02

1.90E-02

5.52E-04

2.39E-06

5.39E-02

1.96E-02

5.68E-04

2.46E-06

5.55E-02

2.02E-02

5.85E-04

2.53E-06

5.71E-02

2.07E-02

6.02E-04

2.61E-06

5.87E-02

2.13E-02

6.19E-04

2.68E-06

6.04E-02

2.20E-02

6.37E-04

2.76E-06

6.22E-02

2.26E-02

6.55E-04

2.85E-06

6.40E-02

2.33E-02

6.74E-04

2.93E-06

6.58E-02

2.39E-02

6.93E-04

3.02E-06

6.77E-02

2.46E-02

7.13E-04

3.11E-06

6.96E-02

2.53E-02

7.34E-04

3.21E-06

7.16E-02

2.61E-02

7.55E-04

3.30E-06

7.37E-02

2.68E-02

7.76E-04

3.40E-06

7.58E-02

2.76E-02

7.99E-04

3.50E-06

7.80E-02

2.84E-02

8.22E-04

3.61E-06

8.02E-02

2.92E-02

8.45E-04

3.72E-06

8.25E-02

3.01E-02

8.69E-04

3.83E-06

8.48E-02

3.09E-02

8.94E-04

3.94E-06

8.73E-02

3.18E-02

9.20E-04

4.06E-06

8.98E-02

3.27E-02

9.46E-04

4.18E-06

9.23E-02

3.37E-02

9.73E-04

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

4.31E-06

9.49E-02

3.47E-02

1.00E-03

4.44E-06

9.76E-02

3.57E-02

1.03E-03

4.57E-06

1.00E-01

3.67E-02

1.06E-03

4.71E-06

1.03E-01

3.77E-02

1.09E-03

4.85E-06

1.06E-01

3.88E-02

1.12E-03

4.99E-06

1.09E-01

3.99E-02

1.15E-03

5.14E-06

1.12E-01

4.11E-02

1.18E-03

5.30E-06

1.15E-01

4.22E-02

1.22E-03

5.46E-06

1.19E-01

4.34E-02

1.25E-03

5.62E-06

1.22E-01

4.47E-02

1.29E-03

5.79E-06

1.25E-01

4.59E-02

1.32E-03

5.96E-06

1.29E-01

4.73E-02

1.36E-03

6.14E-06

1.33E-01

4.86E-02

1.40E-03

6.33E-06

1.36E-01

5.00E-02

1.44E-03

6.52E-06

1.40E-01

5.14E-02

1.48E-03

6.71E-06

1.44E-01

5.28E-02

1.52E-03

6.91E-06

1.48E-01

5.43E-02

1.56E-03

7.12E-06

1.52E-01

5.58E-02

1.60E-03

7.33E-06

1.56E-01

5.74E-02

1.65E-03

7.55E-06

1.61E-01

5.90E-02

1.69E-03

7.78E-06

1.65E-01

6.06E-02

1.74E-03

8.01E-06

1.70E-01

6.23E-02

1.79E-03

8.25E-06

1.74E-01

6.41E-02

1.84E-03

8.50E-06

1.79E-01

6.59E-02

1.89E-03

8.76E-06

1.84E-01

6.77E-02

1.94E-03

9.02E-06

1.89E-01

6.96E-02

1.99E-03

9.29E-06

1.94E-01

7.15E-02

2.05E-03

9.57E-06

2.00E-01

7.35E-02

2.10E-03

9.86E-06

2.05E-01

7.56E-02

2.16E-03

-------
o



Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

1.02E-05

2.11E-01

7.77E-02

2.22E-03

1.05E-05

2.16E-01

7.98E-02

2.28E-03

1.08E-05

2.22E-01

8.20E-02

2.34E-03

1.11E-05

2.28E-01

8.43E-02

2.41E-03

1.14E-05

2.34E-01

8.67E-02

2.47E-03

1.18E-05

2.41E-01

8.91E-02

2.54E-03

1.21E-05

2.47E-01

9.15E-02

2.61E-03

1.25E-05

2.54E-01

9.41E-02

2.68E-03

1.29E-05

2.61E-01

9.67E-02

2.75E-03

1.32E-05

2.68E-01

9.93E-02

2.82E-03

1.36E-05

2.75E-01

1.02E-01

2.90E-03

1.41E-05

2.83E-01

1.05E-01

2.98E-03

1.45E-05

2.90E-01

1.08E-01

3.06E-03

1.49E-05

2.98E-01

1.11E-01

3.14E-03

1.54E-05

3.06E-01

1.14E-01

3.22E-03

1.58E-05

3.14E-01

1.17E-01

3.31E-03

1.63E-05

3.23E-01

1.20E-01

3.40E-03

1.68E-05

3.31E-01

1.23E-01

3.49E-03

1.73E-05

3.40E-01

1.27E-01

3.58E-03

1.78E-05

3.49E-01

1.30E-01

3.68E-03

1.83E-05

3.58E-01

1.34E-01

3.78E-03

1.89E-05

3.68E-01

1.37E-01

3.88E-03

1.95E-05

3.78E-01

1.41E-01

3.98E-03

2.00E-05

3.88E-01

1.45E-01

4.09E-03

2.06E-05

3.98E-01

1.49E-01

4.20E-03

2.13E-05

4.09E-01

1.53E-01

4.31E-03

2.19E-05

4.20E-01

1.57E-01

4.42E-03

2.25E-05

4.31E-01

1.61E-01

4.54E-03

2.32E-05

4.42E-01

1.66E-01

4.66E-03

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

2.39E-05

4.54E-01

1.70E-01

4.78E-03

2.46E-05

4.66E-01

1.75E-01

4.91E-03

2.54E-05

4.78E-01

1.80E-01

5.04E-03

2.61E-05

4.91E-01

1.84E-01

5.17E-03

2.69E-05

5.04E-01

1.89E-01

5.31E-03

2.77E-05

5.17E-01

1.95E-01

5.45E-03

2.86E-05

5.31E-01

2.00E-01

5.59E-03

2.94E-05

5.45E-01

2.05E-01

5.74E-03

3.03E-05

5.59E-01

2.11E-01

5.89E-03

3.12E-05

5.74E-01

2.16E-01

6.05E-03

3.21E-05

5.89E-01

2.22E-01

6.20E-03

3.31E-05

6.06E-01

2.29E-01

6.38E-03

3.41E-05

6.22E-01

2.35E-01

6.54E-03

3.51E-05

6.38E-01

2.41E-01

6.72E-03

3.62E-05

6.54E-01

2.48E-01

6.89E-03

3.73E-05

6.71E-01

2.54E-01

7.08E-03

3.84E-05

6.89E-01

2.61E-01

7.25E-03

3.95E-05

7.07E-01

2.68E-01

7.45E-03

4.07E-05

7.23E-01

2.74E-01

7.62E-03

4.19E-05

7.41E-01

2.82E-01

7.82E-03

4.32E-05

7.60E-01

2.89E-01

8.01E-03

4.45E-05

7.80E-01

2.97E-01

8.22E-03

4.58E-05

8.00E-01

3.05E-01

8.43E-03

4.72E-05

8.20E-01

3.13E-01

8.64E-03

4.86E-05

8.41E-01

3.21E-01

8.86E-03

5.01E-05

8.63E-01

3.29E-01

9.09E-03

5.16E-05

8.84E-01

3.38E-01

9.32E-03

5.31E-05

9.07E-01

3.47E-01

9.55E-03

5.47E-05

9.30E-01

3.56E-01

9.80E-03

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

5.64E-05

9.53E-01

3.65E-01

1.00E-02

5.81E-05

9.77E-01

3.75E-01

1.03E-02

5.98E-05

1.00E+00

3.84E-01

1.06E-02

6.16E-05

1.03E+00

3.95E-01

1.08E-02

6.34E-05

1.05E+00

4.05E-01

1.11E-02

6.54E-05

1.08E+00

4.15E-01

1.14E-02

6.73E-05

1.11E+00

4.26E-01

1.17E-02

6.93E-05

1.13E+00

4.37E-01

1.19E-02

7.14E-05

1.16E+00

4.48E-01

1.22E-02

7.36E-05

1.19E+00

4.58E-01

1.25E-02

7.58E-05

1.22E+00

4.70E-01

1.28E-02

7.80E-05

1.25E+00

4.82E-01

1.31E-02

8.04E-05

1.28E+00

4.94E-01

1.34E-02

8.28E-05

1.31E+00

5.07E-01

1.38E-02

8.53E-05

1.34E+00

5.20E-01

1.41E-02

8.78E-05

1.37E+00

5.33E-01

1.45E-02

9.05E-05

1.41E+00

5.47E-01

1.48E-02

9.32E-05

1.44E+00

5.61E-01

1.52E-02

9.60E-05

1.48E+00

5.75E-01

1.55E-02

9.89E-05

1.51E+00

5.90E-01

1.59E-02

1.02E-04

1.55E+00

6.05E-01

1.63E-02

1.05E-04

1.59E+00

6.20E-01

1.67E-02

1.08E-04

1.62E+00

6.36E-01

1.71E-02

1.11E-04

1.66E+00

6.52E-01

1.75E-02

1.15E-04

1.70E+00

6.69E-01

1.79E-02

1.18E-04

1.75E+00

6.86E-01

1.84E-02

1.22E-04

1.79E+00

7.03E-01

1.88E-02

1.25E-04

1.83E+00

7.20E-01

1.93E-02

1.29E-04

1.87E+00

7.39E-01

1.97E-02

-------
o



Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

1.33E-04

1.92E+00

7.57E-01

2.02E-02

1.37E-04

1.97E+00

7.76E-01

2.07E-02

1.41E-04

2.01E+00

7.96E-01

2.12E-02

1.45E-04

2.08E+00

8.23E-01

2.19E-02

1.50E-04

2.11E+00

8.36E-01

2.22E-02

1.54E-04

2.16E+00

8.57E-01

2.27E-02

1.59E-04

2.23E+00

8.88E-01

2.35E-02

1.63E-04

2.29E+00

9.10E-01

2.41E-02

1.68E-04

2.32E+00

9.24E-01

2.44E-02

1.73E-04

2.37E+00

9.47E-01

2.50E-02

1.79E-04

2.43E+00

9.71E-01

2.56E-02

1.84E-04

2.49E+00

9.96E-01

2.62E-02

1.89E-04

2.55E+00

1.02E+00

2.68E-02

1.95E-04

2.61E+00

1.05E+00

2.75E-02

2.01E-04

2.67E+00

1.07E+00

2.81E-02

2.07E-04

2.76E+00

1.11E+00

2.91E-02

2.13E-04

2.80E+00

1.13E+00

2.94E-02

2.20E-04

2.86E+00

1.16E+00

3.01E-02

2.26E-04

2.95E+00

1.19E+00

3.11E-02

2.33E-04

3.02E+00

1.22E+00

3.18E-02

2.40E-04

3.09E+00

1.25E+00

3.26E-02

2.47E-04

3.14E+00

1.27E+00

3.30E-02

2.55E-04

3.21E+00

1.31E+00

3.38E-02

2.62E-04

3.29E+00

1.34E+00

3.46E-02

2.70E-04

3.39E+00

1.38E+00

3.57E-02

2.78E-04

3.47E+00

1.42E+00

3.65E-02

2.86E-04

3.55E+00

1.45E+00

3.74E-02

2.95E-04

3.61E+00

1.48E+00

3.80E-02

3.04E-04

3.72E+00

1.53E+00

3.91E-02

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

3.13E-04

3.80E+00

1.56E+00

4.00E-02

3.22E-04

3.89E+00

1.60E+00

4.10E-02

3.32E-04

3.98E+00

1.64E+00

4.19E-02

3.42E-04

4.07E+00

1.68E+00

4.29E-02

3.52E-04

4.16E+00

1.72E+00

4.38E-02

3.63E-04

4.26E+00

1.77E+00

4.48E-02

3.74E-04

4.35E+00

1.81E+00

4.58E-02

3.85E-04

4.45E+00

1.85E+00

4.69E-02

3.97E-04

4.55E+00

1.90E+00

4.80E-02

4.08E-04

4.66E+00

1.94E+00

4.90E-02

4.21E-04

4.76E+00

1.99E+00

5.01E-02

4.33E-04

4.87E+00

2.04E+00

5.13E-02

4.46E-04

4.98E+00

2.09E+00

5.24E-02

4.60E-04

5.09E+00

2.14E+00

5.36E-02

4.74E-04

5.20E+00

2.19E+00

5.48E-02

4.88E-04

5.32E+00

2.24E+00

5.60E-02

5.02E-04

5.43E+00

2.30E+00

5.72E-02

5.17E-04

5.55E+00

2.35E+00

5.85E-02

5.33E-04

5.68E+00

2.41E+00

5.98E-02

5.49E-04

5.80E+00

2.47E+00

6.11E-02

5.65E-04

5.93E+00

2.53E+00

6.24E-02

5.82E-04

6.06E+00

2.59E+00

6.38E-02

6.00E-04

6.19E+00

2.65E+00

6.52E-02

6.18E-04

6.33E+00

2.71E+00

6.66E-02

6.36E-04

6.46E+00

2.78E+00

6.80E-02

6.55E-04

6.60E+00

2.84E+00

6.95E-02

6.75E-04

6.75E+00

2.91E+00

7.10E-02

6.95E-04

6.89E+00

2.98E+00

7.26E-02

7.16E-04

7.04E+00

3.05E+00

7.41E-02

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

7.38E-04

7.20E+00

3.13E+00

7.58E-02

7.60E-04

7.35E+00

3.20E+00

7.74E-02

7.83E-04

7.51E+00

3.28E+00

7.91E-02

8.06E-04

7.61E+00

3.33E+00

8.01E-02

8.30E-04

7.77E+00

3.41E+00

8.19E-02

8.55E-04

7.94E+00

3.49E+00

8.36E-02

8.81E-04

8.11E+00

3.58E+00

8.54E-02

9.07E-04

8.30E+00

3.67E+00

8.74E-02

9.21E-04

8.37E+00

3.70E+00

8.81E-02

9.35E-04

8.46E+00

3.75E+00

8.90E-02

9.49E-04

9.14E+00

4.12E+00

9.62E-02

9.63E-04

9.54E+00

4.33E+00

1.00E-01

9.69E-04

9.70E+00

4.42E+00

1.02E-01

9.77E-04

9.87E+00

4.51E+00

1.04E-01

1.17E-03

1.01E+01

4.58E+00

1.07E-01

1.18E-03

1.02E+01

4.63E+00

1.08E-01

1.20E-03

1.03E+01

4.68E+00

1.09E-01

1.22E-03

1.04E+01

4.73E+00

1.10E-01

1.24E-03

1.05E+01

4.75E+00

1.10E-01

1.26E-03

1.06E+01

4.81E+00

1.11E-01

1.27E-03

1.07E+01

4.86E+00

1.12E-01

1.29E-03

1.08E+01

4.92E+00

1.14E-01

1.31E-03

1.09E+01

4.97E+00

1.15E-01

1.33E-03

1.10E+01

5.03E+00

1.16E-01

1.35E-03

1.11E+01

5.08E+00

1.17E-01

1.37E-03

1.12E+01

5.13E+00

1.18E-01

1.39E-03

1.13E+01

5.18E+00

1.19E-01

1.41E-03

1.14E+01

5.23E+00

1.20E-01

1.43E-03

1.15E+01

5.29E+00

1.21E-01

-------
o



Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

1.46E-03

1.16E+01

5.34E+00

1.22E-01

1.48E-03

1.17E+01

5.40E+00

1.23E-01

1.50E-03

1.18E+01

5.47E+00

1.25E-01

1.52E-03

1.20E+01

5.54E+00

1.26E-01

1.54E-03

1.21E+01

5.61E+00

1.28E-01

1.57E-03

1.22E+01

5.66E+00

1.29E-01

1.59E-03

1.24E+01

5.73E+00

1.30E-01

1.61E-03

1.25E+01

5.82E+00

1.32E-01

1.64E-03

1.27E+01

5.88E+00

1.33E-01

1.66E-03

1.28E+01

5.95E+00

1.35E-01

1.69E-03

1.29E+01

6.02E+00

1.36E-01

1.71E-03

1.31E+01

6.10E+00

1.37E-01

1.74E-03

1.32E+01

6.17E+00

1.39E-01

1.76E-03

1.33E+01

6.24E+00

1.40E-01

1.79E-03

1.35E+01

6.32E+00

1.42E-01

1.82E-03

1.36E+01

6.39E+00

1.43E-01

1.84E-03

1.38E+01

6.46E+00

1.45E-01

1.87E-03

1.40E+01

6.59E+00

1.47E-01

1.90E-03

1.46E+01

6.95E+00

1.54E-01

2.02E-03

1.50E+01

7.16E+00

1.58E-01

2.08E-03

1.51E+01

7.23E+00

1.59E-01

2.14E-03

1.53E+01

7.31E+00

1.61E-01

2.20E-03

1.56E+01

7.47E+00

1.64E-01

2.27E-03

1.59E+01

7.65E+00

1.68E-01

2.34E-03

1.62E+01

7.82E+00

1.71E-01

2.41E-03

1.66E+01

8.00E+00

1.74E-01

2.48E-03

1.69E+01

8.19E+00

1.78E-01

2.55E-03

1.72E+01

8.38E+00

1.81E-01

2.63E-03

1.76E+01

8.57E+00

1.85E-01

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

2.71E-03

1.79E+01

8.77E+00

1.89E-01

2.79E-03

1.83E+01

8.98E+00

1.92E-01

2.87E-03

1.87E+01

9.19E+00

1.96E-01

2.96E-03

1.90E+01

9.41E+00

2.00E-01

3.05E-03

1.94E+01

9.62E+00

2.04E-01

3.14E-03

1.98E+01

9.85E+00

2.08E-01

3.23E-03

2.02E+01

1.01E+01

2.13E-01

3.33E-03

2.06E+01

1.03E+01

2.17E-01

3.43E-03

2.10E+01

1.06E+01

2.21E-01

3.53E-03

2.14E+01

1.08E+01

2.25E-01

3.64E-03

2.18E+01

1.11E+01

2.30E-01

3.75E-03

2.25E+01

1.15E+01

2.37E-01

3.98E-03

2.29E+01

1.17E+01

2.41E-01

4.10E-03

2.32E+01

1.18E+01

2.44E-01

4.22E-03

2.35E+01

1.20E+01

2.48E-01

4.35E-03

2.40E+01

1.23E+01

2.52E-01

4.48E-03

2.44E+01

1.26E+01

2.57E-01

4.61E-03

2.49E+01

1.29E+01

2.63E-01

4.75E-03

2.55E+01

1.33E+01

2.69E-01

4.89E-03

2.61E+01

1.36E+01

2.74E-01

5.04E-03

2.69E+01

1.41E+01

2.83E-01

5.19E-03

2.75E+01

1.45E+01

2.90E-01

5.35E-03

2.83E+01

1.51E+01

2.98E-01

5.51E-03

2.91E+01

1.55E+01

3.06E-01

5.67E-03

2.97E+01

1.59E+01

3.13E-01

5.84E-03

3.03E+01

1.63E+01

3.19E-01

5.93E-03

3.04E+01

1.64E+01

3.20E-01

6.02E-03

3.07E+01

1.65E+01

3.23E-01

6.20E-03

3.15E+01

1.71E+01

3.31E-01

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

6.38E-03

3.22E+01

1.76E+01

3.39E-01

6.57E-03

3.28E+01

1.80E+01

3.46E-01

6.77E-03

3.35E+01

1.84E+01

3.53E-01

6.98E-03

3.42E+01

1.89E+01

3.60E-01

7.18E-03

3.50E+01

1.94E+01

3.68E-01

7.40E-03

3.57E+01

1.99E+01

3.76E-01

7.51E-03

3.61E+01

2.02E+01

3.80E-01

7.62E-03

3.63E+01

2.03E+01

3.82E-01

7.85E-03

3.67E+01

2.06E+01

3.87E-01

8.09E-03

3.70E+01

2.07E+01

3.89E-01

8.33E-03

3.75E+01

2.10E+01

3.94E-01

8.58E-03

3.89E+01

2.21E+01

4.09E-01

8.71E-03

3.93E+01

2.24E+01

4.14E-01

8.84E-03

3.97E+01

2.26E+01

4.18E-01

9.10E-03

4.04E+01

2.31E+01

4.25E-01

9.37E-03

4.13E+01

2.38E+01

4.35E-01

9.66E-03

4.21E+01

2.43E+01

4.44E-01

9.94E-03

4.31E+01

2.50E+01

4.53E-01

1.02E-02

4.39E+01

2.56E+01

4.62E-01

1.06E-02

4.47E+01

2.62E+01

4.71E-01

1.09E-02

4.56E+01

2.68E+01

4.80E-01

1.12E-02

4.66E+01

2.75E+01

4.90E-01

1.15E-02

4.75E+01

2.82E+01

5.00E-01

1.19E-02

4.82E+01

2.87E+01

5.07E-01

1.22E-02

4.91E+01

2.94E+01

5.17E-01

1.26E-02

5.00E+01

3.00E+01

5.26E-01

1.30E-02

5.12E+01

3.09E+01

5.39E-01

1.34E-02

5.24E+01

3.19E+01

5.52E-01

1.38E-02

5.36E+01

3.28E+01

5.65E-01

-------
o



Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

1.42E-02

5.48E+01

3.37E+01

5.77E-01

1.46E-02

5.57E+01

3.44E+01

5.87E-01

1.50E-02

5.68E+01

3.52E+01

5.97E-01

1.55E-02

5.78E+01

3.60E+01

6.08E-01

1.60E-02

5.88E+01

3.67E+01

6.19E-01

1.64E-02

5.97E+01

3.75E+01

6.29E-01

1.69E-02

6.10E+01

3.85E+01

6.42E-01

1.74E-02

6.22E+01

3.95E+01

6.55E-01

1.80E-02

6.34E+01

4.04E+01

6.68E-01

1.85E-02

6.47E+01

4.14E+01

6.81E-01

1.91E-02

6.60E+01

4.25E+01

6.94E-01

1.96E-02

6.73E+01

4.35E+01

7.08E-01

2.02E-02

6.86E+01

4.46E+01

7.22E-01

2.08E-02

7.00E+01

4.57E+01

7.36E-01

2.14E-02

7.13E+01

4.69E+01

7.51E-01

2.21E-02

7.28E+01

4.81E+01

7.66E-01

2.28E-02

7.42E+01

4.93E+01

7.81E-01

2.34E-02

7.57E+01

5.05E+01

7.97E-01

2.41E-02

7.71E+01

5.18E+01

8.12E-01

2.49E-02

7.87E+01

5.31E+01

8.28E-01

2.56E-02

8.02E+01

5.44E+01

8.44E-01

2.64E-02

8.18E+01

5.58E+01

8.61E-01

2.72E-02

8.33E+01

5.71E+01

8.77E-01

2.80E-02

8.50E+01

5.86E+01

8.95E-01

2.88E-02

8.67E+01

6.01E+01

9.12E-01

2.97E-02

8.83E+01

6.16E+01

9.30E-01

3.06E-02

9.03E+01

6.34E+01

9.50E-01

3.15E-02

9.21E+01

6.50E+01

9.69E-01

3.24E-02

9.40E+01

6.67E+01

9.89E-01

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

3.34E-02

9.57E+01

6.83E+01

1.01E+00

3.44E-02

9.74E+01

6.99E+01

1.03E+00

3.54E-02

9.92E+01

7.15E+01

1.04E+00

3.65E-02

1.01E+02

7.32E+01

1.06E+00

3.76E-02

1.03E+02

7.51E+01

1.08E+00

3.87E-02

1.05E+02

7.69E+01

1.10E+00

3.99E-02

1.07E+02

7.89E+01

1.13E+00

4.11E-02

1.09E+02

8.09E+01

1.15E+00

4.23E-02

1.11E+02

8.30E+01

1.17E+00

4.36E-02

1.14E+02

8.53E+01

1.20E+00

4.49E-02

1.16E+02

8.76E+01

1.22E+00

4.63E-02

1.18E+02

8.99E+01

1.24E+00

4.76E-02

1.21E+02

9.22E+01

1.27E+00

4.91E-02

1.23E+02

9.46E+01

1.29E+00

5.05E-02

1.25E+02

9.70E+01

1.32E+00

5.21E-02

1.28E+02

9.95E+01

1.34E+00

5.36E-02

1.30E+02

1.02E+02

1.37E+00

5.52E-02

1.33E+02

1.05E+02

1.40E+00

5.69E-02

1.35E+02

1.07E+02

1.43E+00

5.86E-02

1.38E+02

1.10E+02

1.45E+00

6.03E-02

1.41E+02

1.13E+02

1.48E+00

6.22E-02

1.43E+02

1.16E+02

1.51E+00

6.40E-02

1.46E+02

1.19E+02

1.54E+00

6.59E-02

1.49E+02

1.22E+02

1.57E+00

6.79E-02

1.52E+02

1.25E+02

1.60E+00

7.00E-02

1.55E+02

1.28E+02

1.63E+00

7.21E-02

1.58E+02

1.31E+02

1.66E+00

7.42E-02

1.61E+02

1.35E+02

1.69E+00

7.64E-02

1.64E+02

1.38E+02

1.73E+00

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

7.87E-02

1.67E+02

1.42E+02

1.76E+00

8.11E-02

1.71E+02

1.46E+02

1.80E+00

8.35E-02

1.74E+02

1.50E+02

1.83E+00

8.60E-02

1.78E+02

1.54E+02

1.87E+00

8.86E-02

1.81E+02

1.58E+02

1.90E+00

9.13E-02

1.85E+02

1.62E+02

1.94E+00

9.40E-02

1.88E+02

1.66E+02

1.98E+00

9.68E-02

1.92E+02

1.70E+02

2.02E+00

9.97E-02

1.96E+02

1.75E+02

2.06E+00

1.03E-01

1.99E+02

1.79E+02

2.10E+00

1.06E-01

2.03E+02

1.84E+02

2.14E+00

1.09E-01

2.07E+02

1.89E+02

2.18E+00

1.12E-01

2.11E+02

1.94E+02

2.22E+00

1.16E-01

2.15E+02

1.99E+02

2.27E+00

1.19E-01

2.20E+02

2.04E+02

2.31E+00

1.23E-01

2.24E+02

2.10E+02

2.36E+00

1.26E-01

2.28E+02

2.15E+02

2.40E+00

1.30E-01

2.33E+02

2.21E+02

2.45E+00

1.34E-01

2.38E+02

2.27E+02

2.50E+00

1.38E-01

2.42E+02

2.33E+02

2.55E+00

1.42E-01

2.47E+02

2.39E+02

2.60E+00

1.46E-01

2.52E+02

2.46E+02

2.65E+00

1.51E-01

2.57E+02

2.52E+02

2.70E+00

1.55E-01

2.62E+02

2.59E+02

2.75E+00

1.60E-01

2.67E+02

2.66E+02

2.81E+00

1.65E-01

2.72E+02

2.73E+02

2.86E+00

1.70E-01

2.78E+02

2.80E+02

2.92E+00

1.75E-01

2.83E+02

2.88E+02

2.98E+00

1.80E-01

2.89E+02

2.95E+02

3.04E+00

-------
o



Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

1.86E-01

2.94E+02

3.03E+02

3.10E+00

1.91E-01

3.00E+02

3.12E+02

3.16E+00

1.97E-01

3.06E+02

3.20E+02

3.22E+00

2.03E-01

3.12E+02

3.28E+02

3.28E+00

2.09E-01

3.18E+02

3.37E+02

3.35E+00

2.15E-01

3.25E+02

3.46E+02

3.42E+00

2.22E-01

3.31E+02

3.56E+02

3.48E+00

2.28E-01

3.38E+02

3.65E+02

3.55E+00

2.35E-01

3.44E+02

3.75E+02

3.62E+00

2.42E-01

3.51E+02

3.86E+02

3.70E+00

2.49E-01

3.58E+02

3.96E+02

3.77E+00

2.57E-01

3.65E+02

4.07E+02

3.85E+00

2.65E-01

3.73E+02

4.18E+02

3.92E+00

2.72E-01

3.80E+02

4.29E+02

4.00E+00

2.81E-01

3.88E+02

4.41E+02

4.08E+00

2.89E-01

3.95E+02

4.53E+02

4.16E+00

2.98E-01

4.03E+02

4.65E+02

4.24E+00

3.07E-01

4.11E+02

4.77E+02

4.33E+00

3.16E-01

4.19E+02

4.90E+02

4.41E+00

3.25E-01

4.28E+02

5.04E+02

4.50E+00

3.35E-01

4.36E+02

5.18E+02

4.59E+00

3.45E-01

4.45E+02

5.32E+02

4.68E+00

3.56E-01

4.54E+02

5.47E+02

4.78E+00

3.66E-01

4.63E+02

5.62E+02

4.87E+00

3.77E-01

4.72E+02

5.77E+02

4.97E+00

3.89E-01

4.82E+02

5.93E+02

5.07E+00

4.00E-01

4.91E+02

6.09E+02

5.17E+00

4.12E-01

5.01E+02

6.26E+02

5.28E+00

4.25E-01

5.11E+02

6.43E+02

5.38E+00

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

4.37E-01

5.22E+02

6.61E+02

5.49E+00

4.50E-01

5.32E+02

6.79E+02

5.60E+00

4.64E-01

5.43E+02

6.98E+02

5.71E+00

4.92E-01

5.65E+02

7.37E+02

5.95E+00

5.07E-01

5.76E+02

7.57E+02

6.07E+00

5.22E-01

5.88E+02

7.78E+02

6.19E+00

5.54E-01

6.12E+02

8.22E+02

6.44E+00

5.71E-01

6.25E+02

8.44E+02

6.58E+00

5.88E-01

6.37E+02

8.68E+02

6.71E+00

6.05E-01

6.50E+02

8.92E+02

6.84E+00

6.23E-01

6.64E+02

9.17E+02

6.98E+00

6.61E-01

6.91E+02

9.68E+02

7.27E+00

6.81E-01

7.05E+02

9.95E+02

7.42E+00

7.02E-01

7.20E+02

1.02E+03

7.57E+00

7.23E-01

7.34E+02

1.05E+03

7.73E+00

7.44E-01

7.49E+02

1.08E+03

7.89E+00

7.67E-01

7.65E+02

1.11E+03

8.05E+00

7.90E-01

7.80E+02

1.14E+03

8.21E+00

8.13E-01

7.97E+02

1.17E+03

8.38E+00

8.38E-01

8.13E+02

1.21E+03

8.56E+00

8.63E-01

8.30E+02

1.24E+03

8.73E+00

8.89E-01

8.47E+02

1.28E+03

8.91E+00

9.16E-01

8.65E+02

1.31E+03

9.10E+00

9.43E-01

8.83E+02

1.35E+03

9.29E+00

9.71E-01

9.01E+02

1.39E+03

9.48E+00

1.00E+00

9.20E+02

1.43E+03

9.68E+00

1.06E+00

9.58E+02

1.51E+03

1.01E+01

1.09E+00

9.78E+02

1.55E+03

1.03E+01

1.13E+00

9.99E+02

1.59E+03

1.05E+01

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

1.16E+00

1.02E+03

1.64E+03

1.07E+01

1.19E+00

1.04E+03

1.68E+03

1.10E+01

1.23E+00

1.06E+03

1.73E+03

1.12E+01

1.27E+00

1.09E+03

1.78E+03

1.14E+01

1.31E+00

1.11E+03

1.83E+03

1.17E+01

1.34E+00

1.13E+03

1.88E+03

1.19E+01

1.38E+00

1.16E+03

1.94E+03

1.22E+01

1.43E+00

1.18E+03

1.99E+03

1.24E+01

1.47E+00

1.21E+03

2.05E+03

1.27E+01

1.51E+00

1.23E+03

2.11E+03

1.30E+01

1.56E+00

1.26E+03

2.17E+03

1.32E+01

1.61E+00

1.28E+03

2.23E+03

1.35E+01

1.65E+00

1.31E+03

2.29E+03

1.38E+01

1.70E+00

1.34E+03

2.36E+03

1.41E+01

1.75E+00

1.37E+03

2.42E+03

1.44E+01

1.81E+00

1.40E+03

2.49E+03

1.47E+01

1.86E+00

1.43E+03

2.56E+03

1.50E+01

1.92E+00

1.46E+03

2.64E+03

1.54E+01

1.97E+00

1.49E+03

2.71E+03

1.57E+01

2.03E+00

1.52E+03

2.79E+03

1.60E+01

2.09E+00

1.56E+03

2.87E+03

1.64E+01

2.16E+00

1.59E+03

2.95E+03

1.67E+01

2.22E+00

1.62E+03

3.03E+03

1.71E+01

2.29E+00

1.66E+03

3.12E+03

1.75E+01

2.36E+00

1.70E+03

3.21E+03

1.79E+01

2.43E+00

1.73E+03

3.30E+03

1.82E+01

2.50E+00

1.77E+03

3.40E+03

1.86E+01

2.58E+00

1.81E+03

3.49E+03

1.91E+01

2.65E+00

1.85E+03

3.59E+03

1.95E+01

-------
o



Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

2.73E+00

1.89E+03

3.70E+03

1.99E+01

2.82E+00

1.93E+03

3.80E+03

2.04E+01

2.90E+00

1.98E+03

3.91E+03

2.08E+01

2.99E+00

2.02E+03

4.03E+03

2.13E+01

3.08E+00

2.07E+03

4.14E+03

2.17E+01

3.17E+00

2.11E+03

4.26E+03

2.22E+01

3.26E+00

2.16E+03

4.38E+03

2.27E+01

3.36E+00

2.21E+03

4.51E+03

2.32E+01

3.46E+00

2.26E+03

4.64E+03

2.38E+01

3.57E+00

2.31E+03

4.77E+03

2.43E+01

3.67E+00

2.36E+03

4.91E+03

2.49E+01

3.78E+00

2.42E+03

5.05E+03

2.54E+01

3.90E+00

2.47E+03

5.20E+03

2.60E+01

4.01E+00

2.53E+03

5.35E+03

2.66E+01

4.13E+00

2.58E+03

5.50E+03

2.72E+01

4.26E+00

2.64E+03

5.66E+03

2.78E+01

4.39E+00

2.70E+03

5.83E+03

2.85E+01

4.52E+00

2.77E+03

6.00E+03

2.91E+01

4.65E+00

2.83E+03

6.17E+03

2.98E+01

4.79E+00

2.90E+03

6.35E+03

3.05E+01

4.94E+00

2.96E+03

6.53E+03

3.12E+01

5.08E+00

3.03E+03

6.72E+03

3.19E+01

5.24E+00

3.10E+03

6.92E+03

3.27E+01

5.39E+00

3.18E+03

7.12E+03

3.34E+01

5.56E+00

3.25E+03

7.33E+03

3.42E+01

5.72E+00

3.33E+03

7.54E+03

3.50E+01

5.89E+00

3.41E+03

7.76E+03

3.58E+01

6.07E+00

3.49E+03

7.98E+03

3.67E+01

6.25E+00

3.57E+03

8.22E+03

3.76E+01

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

6.44E+00

3.65E+03

8.45E+03

3.85E+01

6.63E+00

3.74E+03

8.70E+03

3.94E+01

6.83E+00

3.83E+03

8.95E+03

4.03E+01

7.04E+00

3.92E+03

9.21E+03

4.13E+01

7.25E+00

4.02E+03

9.48E+03

4.23E+01

7.47E+00

4.11E+03

9.76E+03

4.33E+01

7.69E+00

4.21E+03

1.00E+04

4.43E+01

7.92E+00

4.32E+03

1.03E+04

4.54E+01

8.16E+00

4.42E+03

1.06E+04

4.65E+01

8.40E+00

4.53E+03

1.10E+04

4.77E+01

8.66E+00

4.64E+03

1.13E+04

4.88E+01

8.92E+00

4.75E+03

1.16E+04

5.00E+01

9.18E+00

4.87E+03

1.19E+04

5.13E+01

9.46E+00

4.99E+03

1.23E+04

5.25E+01

9.74E+00

5.11E+03

1.26E+04

5.38E+01

1.00E+01

5.22E+03

1.30E+04

5.50E+01

1.00E+01

5.24E+03

1.30E+04

5.51E+01

1.34E+01

6.64E+03

1.72E+04

6.99E+01

1.67E+01

8.04E+03

2.14E+04

8.47E+01

2.00E+01

9.45E+03

2.56E+04

9.94E+01

2.33E+01

1.08E+04

2.97E+04

1.14E+02

2.67E+01

1.22E+04

3.39E+04

1.28E+02

3.00E+01

1.36E+04

3.81E+04

1.43E+02

3.33E+01

1.49E+04

4.22E+04

1.57E+02

3.67E+01

1.63E+04

4.63E+04

1.72E+02

4.00E+01

1.77E+04

5.05E+04

1.86E+02

4.33E+01

1.90E+04

5.46E+04

2.00E+02

4.67E+01

2.04E+04

5.87E+04

2.15E+02

5.00E+01

2.17E+04

6.28E+04

2.29E+02

1

Nongestational Lifetime

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden

(ng/kg)

Blood
(ng/kg)

5.33E+01

2.31E+04

6.69E+04

2.43E+02

5.67E+01

2.45E+04

7.10E+04

2.57E+02

6.00E+01

2.58E+04

7.51E+04

2.72E+02

6.33E+01

2.72E+04

7.92E+04

2.86E+02

6.67E+01

2.85E+04

8.32E+04

3.00E+02

7.00E+01

2.99E+04

8.73E+04

3.14E+02

7.33E+01

3.12E+04

9.13E+04

3.29E+02

7.67E+01

3.26E+04

9.54E+04

3.43E+02

8.00E+01

3.39E+04

9.94E+04

3.57E+02

8.33E+01

3.53E+04

1.03E+05

3.71E+02

8.67E+01

3.66E+04

1.07E+05

3.86E+02

9.00E+01

3.80E+04

1.12E+05

4.00E+02

9.33E+01

3.94E+04

1.16E+05

4.14E+02

9.67E+01

4.07E+04

1.20E+05

4.28E+02

1.00E+02

4.21E+04

1.24E+05

4.43E+02

-------
C.4.2. Nongestational 5-Year
Average

o



Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

1.00E-09

5.18E-05

1.87E-05

5.45E-07

1.33E-09

6.90E-05

2.50E-05

7.26E-07

1.67E-09

8.62E-05

3.12E-05

9.07E-07

2.00E-09

1.03E-04

3.74E-05

1.09E-06

2.33E-09

1.21E-04

4.36E-05

1.27E-06

2.67E-09

1.38E-04

4.99E-05

1.45E-06

3.00E-09

1.55E-04

5.61E-05

1.63E-06

3.33E-09

1.72E-04

6.23E-05

1.81E-06

3.67E-09

1.90E-04

6.86E-05

1.99E-06

4.00E-09

2.07E-04

7.48E-05

2.17E-06

4.33E-09

2.24E-04

8.10E-05

2.36E-06

4.67E-09

2.41E-04

8.72E-05

2.54E-06

5.00E-09

2.58E-04

9.35E-05

2.72E-06

5.33E-09

2.76E-04

9.97E-05

2.90E-06

5.67E-09

2.93E-04

1.06E-04

3.08E-06

6.00E-09

3.10E-04

1.12E-04

3.26E-06

6.33E-09

3.27E-04

1.18E-04

3.44E-06

6.67E-09

3.44E-04

1.25E-04

3.62E-06

7.00E-09

3.61E-04

1.31E-04

3.80E-06

7.33E-09

3.79E-04

1.37E-04

3.98E-06

7.67E-09

3.96E-04

1.43E-04

4.16E-06

8.00E-09

4.13E-04

1.49E-04

4.34E-06

8.33E-09

4.30E-04

1.56E-04

4.52E-06

8.67E-09

4.47E-04

1.62E-04

4.70E-06

9.00E-09

4.65E-04

1.68E-04

4.89E-06

9.33E-09

4.82E-04

1.74E-04

5.07E-06

9.67E-09

4.99E-04

1.80E-04

5.25E-06

1.00E-08

5.16E-04

1.87E-04

5.43E-06

1.33E-08

6.87E-04

2.48E-04

7.22E-06

1.67E-08

8.57E-04

3.10E-04

9.01E-06

2.00E-08

1.03E-03

3.72E-04

1.08E-05

2.33E-08

1.20E-03

4.34E-04

1.26E-05

2.67E-08

1.37E-03

4.96E-04

1.44E-05

3.00E-08

1.54E-03

5.57E-04

1.62E-05

3.33E-08

1.71E-03

6.19E-04

1.80E-05

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

3.67E-08

1.88E-03

6.81E-04

1.98E-05

4.00E-08

2.05E-03

7.43E-04

2.16E-05

4.33E-08

2.22E-03

8.04E-04

2.34E-05

4.67E-08

2.39E-03

8.66E-04

2.51E-05

5.00E-08

2.56E-03

9.28E-04

2.69E-05

5.33E-08

2.73E-03

9.89E-04

2.87E-05

5.67E-08

2.90E-03

1.05E-03

3.05E-05

6.00E-08

3.07E-03

1.11E-03

3.23E-05

6.33E-08

3.24E-03

1.17E-03

3.40E-05

6.67E-08

3.41E-03

1.23E-03

3.58E-05

7.00E-08

3.57E-03

1.30E-03

3.76E-05

7.33E-08

3.74E-03

1.36E-03

3.94E-05

7.67E-08

3.91E-03

1.42E-03

4.11E-05

8.00E-08

4.08E-03

1.48E-03

4.29E-05

8.33E-08

4.25E-03

1.54E-03

4.47E-05

8.67E-08

4.42E-03

1.60E-03

4.65E-05

9.00E-08

4.59E-03

1.66E-03

4.82E-05

9.33E-08

4.76E-03

1.72E-03

5.00E-05

9.67E-08

4.93E-03

1.79E-03

5.18E-05

1.00E-07

5.09E-03

1.85E-03

5.36E-05

1.33E-07

6.74E-03

2.45E-03

7.09E-05

1.67E-07

8.39E-03

3.05E-03

8.82E-05

2.00E-07

1.00E-02

3.65E-03

1.06E-04

2.33E-07

1.17E-02

4.25E-03

1.23E-04

2.67E-07

1.33E-02

4.85E-03

1.40E-04

3.00E-07

1.50E-02

5.45E-03

1.57E-04

3.33E-07

1.66E-02

6.05E-03

1.75E-04

3.67E-07

1.83E-02

6.65E-03

1.92E-04

4.00E-07

1.99E-02

7.25E-03

2.09E-04

4.33E-07

2.16E-02

7.85E-03

2.27E-04

4.67E-07

2.32E-02

8.45E-03

2.44E-04

5.00E-07

2.49E-02

9.05E-03

2.61E-04

5.33E-07

2.64E-02

9.63E-03

2.78E-04

5.66E-07

2.80E-02

1.02E-02

2.94E-04

5.99E-07

2.96E-02

1.08E-02

3.11E-04

6.33E-07

3.11E-02

1.14E-02

3.28E-04

6.66E-07

3.27E-02

1.19E-02

3.44E-04

6.99E-07

3.43E-02

1.25E-02

3.61E-04

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

7.32E-07

3.59E-02

1.31E-02

3.77E-04

7.65E-07

3.74E-02

1.37E-02

3.94E-04

7.98E-07

3.90E-02

1.42E-02

4.10E-04

8.32E-07

4.06E-02

1.48E-02

4.27E-04

8.65E-07

4.22E-02

1.54E-02

4.43E-04

8.98E-07

4.37E-02

1.60E-02

4.60E-04

9.31E-07

4.53E-02

1.66E-02

4.77E-04

9.64E-07

4.69E-02

1.71E-02

4.93E-04

9.97E-07

4.85E-02

1.77E-02

5.10E-04

1.01E-06

4.92E-02

1.80E-02

5.17E-04

1.03E-06

4.99E-02

1.82E-02

5.24E-04

1.04E-06

5.06E-02

1.85E-02

5.32E-04

1.06E-06

5.13E-02

1.88E-02

5.40E-04

1.07E-06

5.20E-02

1.90E-02

5.47E-04

1.09E-06

5.28E-02

1.93E-02

5.55E-04

1.11E-06

5.35E-02

1.96E-02

5.63E-04

1.12E-06

5.43E-02

1.99E-02

5.71E-04

1.14E-06

5.51E-02

2.01E-02

5.79E-04

1.16E-06

5.59E-02

2.04E-02

5.88E-04

1.17E-06

5.67E-02

2.07E-02

5.96E-04

1.19E-06

5.75E-02

2.10E-02

6.05E-04

1.21E-06

5.83E-02

2.13E-02

6.13E-04

1.23E-06

5.92E-02

2.16E-02

6.22E-04

1.24E-06

6.00E-02

2.20E-02

6.31E-04

1.26E-06

6.09E-02

2.23E-02

6.40E-04

1.28E-06

6.17E-02

2.26E-02

6.49E-04

1.30E-06

6.26E-02

2.29E-02

6.58E-04

1.32E-06

6.35E-02

2.32E-02

6.68E-04

1.34E-06

6.44E-02

2.36E-02

6.77E-04

1.36E-06

6.53E-02

2.39E-02

6.87E-04

1.38E-06

6.63E-02

2.43E-02

6.97E-04

1.40E-06

6.72E-02

2.46E-02

7.07E-04

1.42E-06

6.82E-02

2.50E-02

7.17E-04

1.44E-06

6.91E-02

2.53E-02

7.27E-04

1.46E-06

7.02E-02

2.57E-02

7.38E-04

1.49E-06

7.12E-02

2.61E-02

7.48E-04

1.53E-06

7.32E-02

2.68E-02

7.70E-04

1.58E-06

7.53E-02

2.76E-02

7.92E-04

-------
o



Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

1.62E-06

7.74E-02

2.84E-02

8.14E-04

1.67E-06

7.96E-02

2.92E-02

8.37E-04

1.72E-06

8.19E-02

3.00E-02

8.61E-04

1.77E-06

8.42E-02

3.09E-02

8.86E-04

1.83E-06

8.66E-02

3.18E-02

9.11E-04

1.88E-06

8.91E-02

3.27E-02

9.37E-04

1.94E-06

9.16E-02

3.36E-02

9.63E-04

2.00E-06

9.42E-02

3.46E-02

9.91E-04

2.06E-06

9.69E-02

3.56E-02

1.02E-03

2.12E-06

9.96E-02

3.66E-02

1.05E-03

2.18E-06

1.02E-01

3.77E-02

1.08E-03

2.25E-06

1.05E-01

3.87E-02

1.11E-03

2.32E-06

1.08E-01

3.98E-02

1.14E-03

2.39E-06

1.11E-01

4.10E-02

1.17E-03

2.46E-06

1.15E-01

4.21E-02

1.20E-03

2.53E-06

1.18E-01

4.33E-02

1.24E-03

2.61E-06

1.21E-01

4.46E-02

1.27E-03

2.68E-06

1.24E-01

4.58E-02

1.31E-03

2.76E-06

1.28E-01

4.71E-02

1.35E-03

2.85E-06

1.32E-01

4.85E-02

1.38E-03

2.93E-06

1.35E-01

4.98E-02

1.42E-03

3.02E-06

1.39E-01

5.13E-02

1.46E-03

3.11E-06

1.43E-01

5.27E-02

1.50E-03

3.21E-06

1.47E-01

5.42E-02

1.54E-03

3.30E-06

1.51E-01

5.57E-02

1.59E-03

3.40E-06

1.55E-01

5.73E-02

1.63E-03

3.50E-06

1.59E-01

5.89E-02

1.68E-03

3.61E-06

1.64E-01

6.05E-02

1.72E-03

3.72E-06

1.68E-01

6.22E-02

1.77E-03

3.83E-06

1.73E-01

6.40E-02

1.82E-03

3.94E-06

1.78E-01

6.58E-02

1.87E-03

4.06E-06

1.83E-01

6.76E-02

1.92E-03

4.18E-06

1.88E-01

6.95E-02

1.97E-03

4.31E-06

1.93E-01

7.15E-02

2.03E-03

4.44E-06

1.98E-01

7.34E-02

2.08E-03

4.57E-06

2.04E-01

7.55E-02

2.14E-03

4.71E-06

2.09E-01

7.76E-02

2.20E-03

4.85E-06

2.15E-01

7.98E-02

2.26E-03

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

4.99E-06

2.21E-01

8.20E-02

2.32E-03

5.14E-06

2.27E-01

8.42E-02

2.39E-03

5.30E-06

2.33E-01

8.66E-02

2.45E-03

5.46E-06

2.39E-01

8.90E-02

2.52E-03

5.62E-06

2.46E-01

9.14E-02

2.59E-03

5.79E-06

2.53E-01

9.39E-02

2.66E-03

5.96E-06

2.59E-01

9.65E-02

2.73E-03

6.14E-06

2.66E-01

9.92E-02

2.80E-03

6.33E-06

2.74E-01

1.02E-01

2.88E-03

6.52E-06

2.81E-01

1.05E-01

2.95E-03

6.71E-06

2.88E-01

1.07E-01

3.03E-03

6.91E-06

2.96E-01

1.10E-01

3.11E-03

7.12E-06

3.04E-01

1.13E-01

3.19E-03

7.33E-06

3.12E-01

1.16E-01

3.28E-03

7.55E-06

3.20E-01

1.19E-01

3.36E-03

7.78E-06

3.28E-01

1.23E-01

3.45E-03

8.01E-06

3.37E-01

1.26E-01

3.54E-03

8.25E-06

3.46E-01

1.29E-01

3.64E-03

8.50E-06

3.55E-01

1.33E-01

3.73E-03

8.76E-06

3.64E-01

1.36E-01

3.83E-03

9.02E-06

3.74E-01

1.40E-01

3.93E-03

9.29E-06

3.84E-01

1.44E-01

4.04E-03

9.57E-06

3.94E-01

1.48E-01

4.15E-03

9.86E-06

4.05E-01

1.52E-01

4.25E-03

1.02E-05

4.15E-01

1.56E-01

4.36E-03

1.05E-05

4.26E-01

1.60E-01

4.48E-03

1.08E-05

4.37E-01

1.64E-01

4.59E-03

1.11E-05

4.48E-01

1.68E-01

4.71E-03

1.14E-05

4.60E-01

1.73E-01

4.83E-03

1.18E-05

4.72E-01

1.78E-01

4.96E-03

1.21E-05

4.84E-01

1.82E-01

5.08E-03

1.25E-05

4.96E-01

1.87E-01

5.21E-03

1.29E-05

5.09E-01

1.92E-01

5.35E-03

1.32E-05

5.22E-01

1.97E-01

5.49E-03

1.36E-05

5.35E-01

2.02E-01

5.63E-03

1.41E-05

5.49E-01

2.08E-01

5.77E-03

1.45E-05

5.63E-01

2.13E-01

5.92E-03

1.49E-05

5.77E-01

2.18E-01

6.07E-03

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

1.54E-05

5.92E-01

2.24E-01

6.23E-03

1.58E-05

6.07E-01

2.30E-01

6.38E-03

1.63E-05

6.23E-01

2.36E-01

6.55E-03

1.68E-05

6.38E-01

2.42E-01

6.71E-03

1.73E-05

6.54E-01

2.49E-01

6.88E-03

1.78E-05

6.71E-01

2.55E-01

7.05E-03

1.83E-05

6.88E-01

2.62E-01

7.23E-03

1.89E-05

7.05E-01

2.69E-01

7.41E-03

1.95E-05

7.23E-01

2.75E-01

7.60E-03

2.00E-05

7.41E-01

2.83E-01

7.79E-03

2.06E-05

7.60E-01

2.90E-01

7.99E-03

2.13E-05

7.79E-01

2.97E-01

8.18E-03

2.19E-05

7.98E-01

3.05E-01

8.39E-03

2.25E-05

8.18E-01

3.13E-01

8.60E-03

2.32E-05

8.38E-01

3.21E-01

8.81E-03

2.39E-05

8.59E-01

3.29E-01

9.03E-03

2.46E-05

8.80E-01

3.38E-01

9.25E-03

2.54E-05

9.02E-01

3.46E-01

9.48E-03

2.61E-05

9.24E-01

3.55E-01

9.71E-03

2.69E-05

9.47E-01

3.64E-01

9.95E-03

2.77E-05

9.70E-01

3.73E-01

1.02E-02

2.86E-05

9.94E-01

3.83E-01

1.04E-02

2.94E-05

1.02E+00

3.92E-01

1.07E-02

3.03E-05

1.04E+00

4.02E-01

1.10E-02

3.12E-05

1.07E+00

4.12E-01

1.12E-02

3.21E-05

1.09E+00

4.23E-01

1.15E-02

3.31E-05

1.12E+00

4.35E-01

1.18E-02

3.41E-05

1.15E+00

4.46E-01

1.21E-02

3.51E-05

1.18E+00

4.57E-01

1.23E-02

3.62E-05

1.21E+00

4.68E-01

1.27E-02

3.73E-05

1.24E+00

4.80E-01

1.30E-02

3.84E-05

1.26E+00

4.92E-01

1.33E-02

3.95E-05

1.29E+00

5.04E-01

1.35E-02

4.07E-05

1.32E+00

5.14E-01

1.39E-02

4.19E-05

1.35E+00

5.26E-01

1.42E-02

4.32E-05

1.38E+00

5.39E-01

1.45E-02

4.45E-05

1.41E+00

5.52E-01

1.49E-02

4.58E-05

1.45E+00

5.66E-01

1.52E-02

-------
o



Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

4.72E-05

1.48E+00

5.80E-01

1.56E-02

4.86E-05

1.52E+00

5.94E-01

1.59E-02

5.01E-05

1.55E+00

6.08E-01

1.63E-02

5.16E-05

1.59E+00

6.23E-01

1.67E-02

5.31E-05

1.62E+00

6.38E-01

1.71E-02

5.47E-05

1.66E+00

6.53E-01

1.75E-02

5.64E-05

1.70E+00

6.69E-01

1.79E-02

5.81E-05

1.74E+00

6.85E-01

1.83E-02

5.98E-05

1.78E+00

7.02E-01

1.87E-02

6.16E-05

1.82E+00

7.19E-01

1.91E-02

6.34E-05

1.86E+00

7.36E-01

1.96E-02

6.54E-05

1.90E+00

7.53E-01

2.00E-02

6.73E-05

1.95E+00

7.71E-01

2.05E-02

6.93E-05

1.99E+00

7.90E-01

2.09E-02

7.14E-05

2.04E+00

8.08E-01

2.14E-02

7.36E-05

2.06E+00

8.18E-01

2.16E-02

7.58E-05

2.11E+00

8.37E-01

2.21E-02

7.80E-05

2.15E+00

8.57E-01

2.26E-02

8.04E-05

2.20E+00

8.77E-01

2.31E-02

8.28E-05

2.25E+00

8.98E-01

2.36E-02

8.53E-05

2.30E+00

9.19E-01

2.42E-02

8.78E-05

2.35E+00

9.40E-01

2.47E-02

9.05E-05

2.40E+00

9.62E-01

2.52E-02

9.32E-05

2.46E+00

9.84E-01

2.58E-02

9.60E-05

2.51E+00

1.01E+00

2.64E-02

9.89E-05

2.57E+00

1.03E+00

2.69E-02

1.02E-04

2.62E+00

1.05E+00

2.75E-02

1.05E-04

2.68E+00

1.08E+00

2.81E-02

1.08E-04

2.74E+00

1.10E+00

2.88E-02

1.11E-04

2.80E+00

1.13E+00

2.94E-02

1.15E-04

2.86E+00

1.15E+00

3.00E-02

1.18E-04

2.92E+00

1.18E+00

3.07E-02

1.22E-04

2.98E+00

1.21E+00

3.13E-02

1.25E-04

3.05E+00

1.24E+00

3.20E-02

1.29E-04

3.11E+00

1.26E+00

3.27E-02

1.33E-04

3.18E+00

1.29E+00

3.34E-02

1.37E-04

3.25E+00

1.32E+00

3.41E-02

1.41E-04

3.32E+00

1.35E+00

3.48E-02

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

1.45E-04

3.45E+00

1.41E+00

3.62E-02

1.50E-04

3.46E+00

1.41E+00

3.63E-02

1.54E-04

3.53E+00

1.45E+00

3.71E-02

1.59E-04

3.67E+00

1.51E+00

3.86E-02

1.63E-04

3.75E+00

1.54E+00

3.94E-02

1.68E-04

3.77E+00

1.55E+00

3.96E-02

1.73E-04

3.86E+00

1.59E+00

4.06E-02

1.79E-04

3.95E+00

1.63E+00

4.15E-02

1.84E-04

4.04E+00

1.67E+00

4.24E-02

1.89E-04

4.13E+00

1.71E+00

4.33E-02

1.95E-04

4.22E+00

1.75E+00

4.43E-02

2.01E-04

4.31E+00

1.79E+00

4.52E-02

2.07E-04

4.44E+00

1.84E+00

4.66E-02

2.13E-04

4.49E+00

1.87E+00

4.72E-02

2.20E-04

4.59E+00

1.92E+00

4.82E-02

2.26E-04

4.72E+00

1.97E+00

4.95E-02

2.33E-04

4.81E+00

2.02E+00

5.05E-02

2.40E-04

4.91E+00

2.06E+00

5.16E-02

2.47E-04

5.00E+00

2.10E+00

5.24E-02

2.55E-04

5.10E+00

2.15E+00

5.35E-02

2.62E-04

5.21E+00

2.19E+00

5.47E-02

2.70E-04

5.33E+00

2.25E+00

5.60E-02

2.78E-04

5.44E+00

2.30E+00

5.71E-02

2.86E-04

5.55E+00

2.35E+00

5.83E-02

2.95E-04

5.66E+00

2.40E+00

5.94E-02

3.04E-04

5.78E+00

2.46E+00

6.07E-02

3.13E-04

5.90E+00

2.51E+00

6.19E-02

3.22E-04

6.02E+00

2.57E+00

6.32E-02

3.32E-04

6.14E+00

2.63E+00

6.44E-02

3.42E-04

6.26E+00

2.68E+00

6.57E-02

3.52E-04

6.39E+00

2.74E+00

6.71E-02

3.63E-04

6.52E+00

2.80E+00

6.84E-02

3.74E-04

6.65E+00

2.87E+00

6.98E-02

3.85E-04

6.78E+00

2.93E+00

7.12E-02

3.97E-04

6.92E+00

3.00E+00

7.26E-02

4.08E-04

7.06E+00

3.06E+00

7.41E-02

4.21E-04

7.20E+00

3.13E+00

7.56E-02

4.33E-04

7.34E+00

3.20E+00

7.71E-02

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

4.46E-04

7.49E+00

3.27E+00

7.86E-02

4.60E-04

7.64E+00

3.34E+00

8.02E-02

4.74E-04

7.79E+00

3.42E+00

8.18E-02

4.88E-04

7.95E+00

3.49E+00

8.34E-02

5.02E-04

8.10E+00

3.57E+00

8.50E-02

5.17E-04

8.26E+00

3.65E+00

8.67E-02

5.33E-04

8.43E+00

3.73E+00

8.84E-02

5.49E-04

8.59E+00

3.81E+00

9.02E-02

5.65E-04

8.76E+00

3.89E+00

9.19E-02

5.82E-04

8.93E+00

3.98E+00

9.37E-02

6.00E-04

9.11E+00

4.07E+00

9.56E-02

6.18E-04

9.29E+00

4.16E+00

9.74E-02

6.36E-04

9.47E+00

4.25E+00

9.94E-02

6.55E-04

9.65E+00

4.34E+00

1.01E-01

6.75E-04

9.84E+00

4.44E+00

1.03E-01

6.95E-04

1.00E+01

4.54E+00

1.05E-01

7.16E-04

1.02E+01

4.64E+00

1.07E-01

7.38E-04

1.04E+01

4.74E+00

1.09E-01

7.60E-04

1.06E+01

4.84E+00

1.12E-01

7.83E-04

1.08E+01

4.95E+00

1.14E-01

8.06E-04

1.10E+01

5.06E+00

1.16E-01

8.30E-04

1.13E+01

5.17E+00

1.18E-01

8.55E-04

1.15E+01

5.28E+00

1.20E-01

8.81E-04

1.17E+01

5.40E+00

1.23E-01

9.07E-04

1.19E+01

5.52E+00

1.25E-01

9.21E-04

1.20E+01

5.58E+00

1.26E-01

9.35E-04

1.22E+01

5.64E+00

1.27E-01

9.49E-04

1.30E+01

6.23E+00

1.37E-01

9.63E-04

1.38E+01

6.92E+00

1.45E-01

9.69E-04

1.43E+01

7.14E+00

1.50E-01

9.77E-04

1.48E+01

7.34E+00

1.55E-01

9.84E-04

1.52E+01

7.50E+00

1.59E-01

9.91E-04

1.55E+01

7.64E+00

1.63E-01

1.37E-03

1.56E+01

7.50E+00

1.63E-01

1.39E-03

1.57E+01

7.58E+00

1.65E-01

1.41E-03

1.59E+01

7.66E+00

1.66E-01

1.43E-03

1.60E+01

7.75E+00

1.68E-01

1.46E-03

1.62E+01

7.83E+00

1.69E-01

-------
o



Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

1.48E-03

1.63E+01

7.92E+00

1.71E-01

1.50E-03

1.65E+01

8.00E+00

1.73E-01

1.52E-03

1.66E+01

8.09E+00

1.74E-01

1.54E-03

1.68E+01

8.18E+00

1.76E-01

1.57E-03

1.69E+01

8.27E+00

1.78E-01

1.59E-03

1.71E+01

8.36E+00

1.79E-01

1.61E-03

1.73E+01

8.46E+00

1.81E-01

1.64E-03

1.74E+01

8.55E+00

1.83E-01

1.66E-03

1.76E+01

8.64E+00

1.84E-01

1.69E-03

1.78E+01

8.74E+00

1.86E-01

1.71E-03

1.79E+01

8.83E+00

1.88E-01

1.74E-03

1.81E+01

8.93E+00

1.90E-01

1.76E-03

1.83E+01

9.03E+00

1.92E-01

1.79E-03

1.84E+01

9.13E+00

1.93E-01

1.82E-03

1.86E+01

9.23E+00

1.95E-01

1.84E-03

1.88E+01

9.33E+00

1.97E-01

1.87E-03

1.91E+01

9.53E+00

2.00E-01

1.90E-03

1.98E+01

1.01E+01

2.08E-01

1.93E-03

2.05E+01

1.08E+01

2.14E-01

1.96E-03

2.05E+01

1.05E+01

2.14E-01

2.27E-03

2.14E+01

1.09E+01

2.25E-01

2.34E-03

2.18E+01

1.11E+01

2.29E-01

2.41E-03

2.22E+01

1.14E+01

2.33E-01

2.48E-03

2.26E+01

1.16E+01

2.37E-01

2.55E-03

2.31E+01

1.19E+01

2.42E-01

2.63E-03

2.35E+01

1.22E+01

2.46E-01

2.71E-03

2.39E+01

1.24E+01

2.51E-01

2.79E-03

2.44E+01

1.27E+01

2.56E-01

2.87E-03

2.49E+01

1.30E+01

2.61E-01

2.96E-03

2.53E+01

1.33E+01

2.66E-01

3.05E-03

2.58E+01

1.36E+01

2.71E-01

3.14E-03

2.63E+01

1.39E+01

2.76E-01

3.23E-03

2.68E+01

1.42E+01

2.81E-01

3.33E-03

2.73E+01

1.45E+01

2.86E-01

3.43E-03

2.78E+01

1.49E+01

2.91E-01

3.53E-03

2.83E+01

1.52E+01

2.97E-01

3.64E-03

2.88E+01

1.55E+01

3.02E-01

3.75E-03

2.96E+01

1.61E+01

3.10E-01

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

3.81E-03

2.99E+01

1.63E+01

3.14E-01

3.86E-03

3.00E+01

1.63E+01

3.14E-01

4.22E-03

3.04E+01

1.66E+01

3.19E-01

4.35E-03

3.10E+01

1.69E+01

3.25E-01

4.48E-03

3.16E+01

1.73E+01

3.31E-01

4.61E-03

3.21E+01

1.77E+01

3.37E-01

4.75E-03

3.28E+01

1.81E+01

3.44E-01

4.89E-03

3.34E+01

1.86E+01

3.50E-01

5.04E-03

3.44E+01

1.94E+01

3.60E-01

5.19E-03

3.57E+01

2.06E+01

3.74E-01

5.35E-03

3.72E+01

2.12E+01

3.90E-01

5.51E-03

3.81E+01

2.17E+01

3.99E-01

5.67E-03

3.88E+01

2.23E+01

4.07E-01

5.84E-03

3.95E+01

2.28E+01

4.14E-01

5.93E-03

3.98E+01

2.30E+01

4.18E-01

6.02E-03

4.00E+01

2.33E+01

4.20E-01

6.20E-03

4.10E+01

2.38E+01

4.30E-01

6.38E-03

4.18E+01

2.44E+01

4.38E-01

6.57E-03

4.26E+01

2.49E+01

4.46E-01

6.77E-03

4.34E+01

2.55E+01

4.55E-01

6.98E-03

4.42E+01

2.61E+01

4.63E-01

7.18E-03

4.50E+01

2.67E+01

4.72E-01

7.40E-03

4.59E+01

2.73E+01

4.81E-01

7.51E-03

4.63E+01

2.77E+01

4.85E-01

7.62E-03

4.66E+01

2.78E+01

4.88E-01

7.85E-03

4.71E+01

2.81E+01

4.94E-01

8.09E-03

4.72E+01

2.79E+01

4.95E-01

8.33E-03

4.74E+01

2.83E+01

4.97E-01

8.58E-03

4.93E+01

2.99E+01

5.17E-01

8.71E-03

4.98E+01

3.03E+01

5.22E-01

8.84E-03

5.03E+01

3.06E+01

5.27E-01

9.10E-03

5.13E+01

3.15E+01

5.38E-01

9.37E-03

5.23E+01

3.22E+01

5.49E-01

9.66E-03

5.33E+01

3.29E+01

5.59E-01

9.94E-03

5.44E+01

3.38E+01

5.70E-01

1.02E-02

5.54E+01

3.46E+01

5.81E-01

1.06E-02

5.64E+01

3.54E+01

5.92E-01

1.09E-02

5.75E+01

3.62E+01

6.03E-01

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

1.12E-02

5.86E+01

3.71E+01

6.14E-01

1.15E-02

5.96E+01

3.79E+01

6.25E-01

1.19E-02

6.05E+01

3.85E+01

6.35E-01

1.22E-02

6.14E+01

3.92E+01

6.43E-01

1.26E-02

6.24E+01

4.01E+01

6.54E-01

1.30E-02

6.38E+01

4.15E+01

6.69E-01

1.34E-02

6.56E+01

4.31E+01

6.87E-01

1.38E-02

6.74E+01

4.42E+01

7.07E-01

1.42E-02

6.87E+01

4.53E+01

7.20E-01

1.46E-02

6.94E+01

4.59E+01

7.28E-01

1.50E-02

7.06E+01

4.69E+01

7.40E-01

1.55E-02

7.19E+01

4.78E+01

7.54E-01

1.60E-02

7.30E+01

4.87E+01

7.66E-01

1.64E-02

7.38E+01

4.96E+01

7.74E-01

1.69E-02

7.55E+01

5.11E+01

7.92E-01

1.74E-02

7.69E+01

5.23E+01

8.07E-01

1.80E-02

7.84E+01

5.36E+01

8.22E-01

1.85E-02

7.99E+01

5.49E+01

8.37E-01

1.91E-02

8.13E+01

5.62E+01

8.53E-01

1.96E-02

8.29E+01

5.75E+01

8.69E-01

2.02E-02

8.44E+01

5.89E+01

8.85E-01

2.08E-02

8.60E+01

6.03E+01

9.01E-01

2.14E-02

8.76E+01

6.18E+01

9.18E-01

2.21E-02

8.92E+01

6.32E+01

9.35E-01

2.28E-02

9.09E+01

6.48E+01

9.53E-01

2.34E-02

9.26E+01

6.63E+01

9.71E-01

2.41E-02

9.44E+01

6.80E+01

9.89E-01

2.49E-02

9.64E+01

6.98E+01

1.01E+00

2.56E-02

9.79E+01

7.13E+01

1.03E+00

2.64E-02

9.98E+01

7.30E+01

1.05E+00

2.72E-02

1.02E+02

7.48E+01

1.07E+00

2.80E-02

1.04E+02

7.66E+01

1.09E+00

2.88E-02

1.06E+02

7.85E+01

1.11E+00

2.97E-02

1.07E+02

8.04E+01

1.13E+00

3.06E-02

1.10E+02

8.28E+01

1.15E+00

3.15E-02

1.12E+02

8.51E+01

1.17E+00

3.24E-02

1.14E+02

8.69E+01

1.20E+00

3.34E-02

1.16E+02

8.88E+01

1.22E+00

-------
o



Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

3.44E-02

1.18E+02

9.08E+01

1.24E+00

3.54E-02

1.20E+02

9.28E+01

1.26E+00

3.65E-02

1.22E+02

9.47E+01

1.28E+00

3.76E-02

1.24E+02

9.73E+01

1.30E+00

3.87E-02

1.27E+02

9.96E+01

1.33E+00

3.99E-02

1.29E+02

1.02E+02

1.35E+00

4.11E-02

1.32E+02

1.04E+02

1.38E+00

4.23E-02

1.34E+02

1.07E+02

1.40E+00

4.36E-02

1.37E+02

1.10E+02

1.43E+00

4.49E-02

1.40E+02

1.13E+02

1.47E+00

4.63E-02

1.43E+02

1.16E+02

1.49E+00

4.76E-02

1.45E+02

1.19E+02

1.52E+00

4.91E-02

1.48E+02

1.22E+02

1.55E+00

5.05E-02

1.51E+02

1.25E+02

1.58E+00

5.21E-02

1.53E+02

1.28E+02

1.61E+00

5.36E-02

1.56E+02

1.31E+02

1.64E+00

5.52E-02

1.59E+02

1.34E+02

1.67E+00

5.69E-02

1.62E+02

1.38E+02

1.70E+00

5.86E-02

1.65E+02

1.41E+02

1.73E+00

6.03E-02

1.69E+02

1.45E+02

1.77E+00

6.22E-02

1.72E+02

1.48E+02

1.80E+00

6.40E-02

1.74E+02

1.52E+02

1.83E+00

6.59E-02

1.78E+02

1.55E+02

1.86E+00

6.79E-02

1.81E+02

1.59E+02

1.90E+00

7.00E-02

1.84E+02

1.63E+02

1.93E+00

7.21E-02

1.88E+02

1.67E+02

1.97E+00

7.42E-02

1.91E+02

1.71E+02

2.01E+00

7.64E-02

1.95E+02

1.76E+02

2.05E+00

7.87E-02

1.99E+02

1.81E+02

2.09E+00

8.11E-02

2.03E+02

1.86E+02

2.13E+00

8.35E-02

2.07E+02

1.90E+02

2.17E+00

8.60E-02

2.11E+02

1.95E+02

2.21E+00

8.86E-02

2.15E+02

2.00E+02

2.25E+00

9.13E-02

2.19E+02

2.05E+02

2.30E+00

9.40E-02

2.23E+02

2.10E+02

2.34E+00

9.68E-02

2.27E+02

2.16E+02

2.38E+00

9.97E-02

2.32E+02

2.22E+02

2.43E+00

1.03E-01

2.36E+02

2.27E+02

2.48E+00

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

1.06E-01

2.41E+02

2.33E+02

2.52E+00

1.09E-01

2.45E+02

2.39E+02

2.57E+00

1.12E-01

2.50E+02

2.44E+02

2.62E+00

1.16E-01

2.55E+02

2.51E+02

2.67E+00

1.19E-01

2.60E+02

2.57E+02

2.72E+00

1.23E-01

2.65E+02

2.64E+02

2.77E+00

1.26E-01

2.70E+02

2.71E+02

2.83E+00

1.30E-01

2.75E+02

2.78E+02

2.88E+00

1.34E-01

2.80E+02

2.86E+02

2.94E+00

1.38E-01

2.86E+02

2.93E+02

3.00E+00

1.42E-01

2.92E+02

3.01E+02

3.06E+00

1.46E-01

2.97E+02

3.09E+02

3.11E+00

1.51E-01

3.03E+02

3.16E+02

3.17E+00

1.55E-01

3.08E+02

3.24E+02

3.23E+00

1.60E-01

3.14E+02

3.33E+02

3.29E+00

1.65E-01

3.20E+02

3.42E+02

3.36E+00

1.70E-01

3.27E+02

3.51E+02

3.42E+00

1.75E-01

3.33E+02

3.60E+02

3.49E+00

1.80E-01

3.39E+02

3.69E+02

3.56E+00

1.86E-01

3.46E+02

3.79E+02

3.63E+00

1.91E-01

3.53E+02

3.89E+02

3.70E+00

1.97E-01

3.60E+02

3.99E+02

3.77E+00

2.03E-01

3.66E+02

4.09E+02

3.84E+00

2.09E-01

3.73E+02

4.20E+02

3.91E+00

2.15E-01

3.81E+02

4.31E+02

3.99E+00

2.22E-01

3.88E+02

4.43E+02

4.07E+00

2.28E-01

3.96E+02

4.55E+02

4.15E+00

2.35E-01

4.03E+02

4.67E+02

4.23E+00

2.42E-01

4.11E+02

4.79E+02

4.31E+00

2.49E-01

4.20E+02

4.92E+02

4.40E+00

2.57E-01

4.28E+02

5.05E+02

4.48E+00

2.65E-01

4.36E+02

5.19E+02

4.57E+00

2.72E-01

4.45E+02

5.32E+02

4.66E+00

2.81E-01

4.53E+02

5.46E+02

4.75E+00

2.89E-01

4.62E+02

5.61E+02

4.84E+00

2.98E-01

4.71E+02

5.75E+02

4.93E+00

3.07E-01

4.80E+02

5.91E+02

5.03E+00

3.16E-01

4.90E+02

6.07E+02

5.13E+00

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

3.25E-01

4.99E+02

6.23E+02

5.23E+00

3.35E-01

5.09E+02

6.40E+02

5.34E+00

3.45E-01

5.19E+02

6.57E+02

5.44E+00

3.56E-01

5.30E+02

6.75E+02

5.55E+00

3.66E-01

5.40E+02

6.93E+02

5.66E+00

3.77E-01

5.51E+02

7.12E+02

5.77E+00

3.89E-01

5.62E+02

7.31E+02

5.89E+00

4.00E-01

5.73E+02

7.51E+02

6.00E+00

4.12E-01

5.84E+02

7.71E+02

6.12E+00

4.25E-01

5.96E+02

7.92E+02

6.25E+00

4.37E-01

6.08E+02

8.13E+02

6.37E+00

4.50E-01

6.20E+02

8.35E+02

6.50E+00

4.64E-01

6.32E+02

8.58E+02

6.63E+00

4.92E-01

6.58E+02

9.05E+02

6.89E+00

5.07E-01

6.71E+02

9.29E+02

7.03E+00

5.22E-01

6.85E+02

9.55E+02

7.17E+00

5.54E-01

7.12E+02

1.01E+03

7.46E+00

5.71E-01

7.27E+02

1.04E+03

7.61E+00

5.88E-01

7.41E+02

1.06E+03

7.77E+00

6.05E-01

7.56E+02

1.09E+03

7.92E+00

6.23E-01

7.71E+02

1.12E+03

8.08E+00

6.61E-01

8.03E+02

1.18E+03

8.41E+00

6.81E-01

8.19E+02

1.22E+03

8.58E+00

7.02E-01

8.36E+02

1.25E+03

8.76E+00

7.23E-01

8.53E+02

1.28E+03

8.94E+00

7.44E-01

8.70E+02

1.32E+03

9.12E+00

7.67E-01

8.88E+02

1.36E+03

9.31E+00

7.90E-01

9.06E+02

1.39E+03

9.50E+00

8.13E-01

9.25E+02

1.43E+03

9.69E+00

8.38E-01

9.44E+02

1.47E+03

9.89E+00

8.63E-01

9.63E+02

1.51E+03

1.01E+01

8.89E-01

9.83E+02

1.55E+03

1.03E+01

9.16E-01

1.00E+03

1.60E+03

1.05E+01

9.43E-01

1.02E+03

1.64E+03

1.07E+01

9.71E-01

1.05E+03

1.69E+03

1.10E+01

1.00E+00

1.07E+03

1.73E+03

1.12E+01

1.06E+00

1.11E+03

1.83E+03

1.16E+01

1.09E+00

1.14E+03

1.88E+03

1.19E+01

-------
o



Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

1.13E+00

1.16E+03

1.94E+03

1.21E+01

1.16E+00

1.18E+03

1.99E+03

1.24E+01

1.19E+00

1.21E+03

2.04E+03

1.27E+01

1.23E+00

1.23E+03

2.10E+03

1.29E+01

1.27E+00

1.26E+03

2.16E+03

1.32E+01

1.31E+00

1.29E+03

2.22E+03

1.35E+01

1.34E+00

1.31E+03

2.28E+03

1.38E+01

1.38E+00

1.34E+03

2.35E+03

1.40E+01

1.43E+00

1.37E+03

2.41E+03

1.43E+01

1.47E+00

1.40E+03

2.48E+03

1.46E+01

1.51E+00

1.43E+03

2.55E+03

1.50E+01

1.56E+00

1.46E+03

2.62E+03

1.53E+01

1.61E+00

1.49E+03

2.69E+03

1.56E+01

1.65E+00

1.52E+03

2.77E+03

1.59E+01

1.70E+00

1.55E+03

2.85E+03

1.63E+01

1.75E+00

1.59E+03

2.93E+03

1.66E+01

1.81E+00

1.62E+03

3.01E+03

1.70E+01

1.86E+00

1.66E+03

3.10E+03

1.74E+01

1.92E+00

1.69E+03

3.18E+03

1.77E+01

1.97E+00

1.73E+03

3.27E+03

1.81E+01

2.03E+00

1.77E+03

3.37E+03

1.85E+01

2.09E+00

1.80E+03

3.46E+03

1.89E+01

2.16E+00

1.84E+03

3.56E+03

1.93E+01

2.22E+00

1.88E+03

3.66E+03

1.97E+01

2.29E+00

1.92E+03

3.76E+03

2.02E+01

2.36E+00

1.97E+03

3.87E+03

2.06E+01

2.43E+00

2.01E+03

3.98E+03

2.11E+01

2.50E+00

2.05E+03

4.09E+03

2.15E+01

2.58E+00

2.10E+03

4.21E+03

2.20E+01

2.65E+00

2.15E+03

4.33E+03

2.25E+01

2.73E+00

2.19E+03

4.45E+03

2.30E+01

2.82E+00

2.24E+03

4.58E+03

2.35E+01

2.90E+00

2.29E+03

4.71E+03

2.40E+01

2.99E+00

2.34E+03

4.85E+03

2.46E+01

3.08E+00

2.40E+03

4.98E+03

2.51E+01

3.17E+00

2.45E+03

5.13E+03

2.57E+01

3.26E+00

2.51E+03

5.27E+03

2.63E+01

3.36E+00

2.56E+03

5.42E+03

2.69E+01

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

3.46E+00

2.62E+03

5.58E+03

2.75E+01

3.57E+00

2.68E+03

5.74E+03

2.81E+01

3.67E+00

2.74E+03

5.90E+03

2.87E+01

3.78E+00

2.80E+03

6.07E+03

2.94E+01

3.90E+00

2.87E+03

6.25E+03

3.01E+01

4.01E+00

2.93E+03

6.42E+03

3.07E+01

4.13E+00

3.00E+03

6.61E+03

3.15E+01

4.26E+00

3.07E+03

6.80E+03

3.22E+01

4.39E+00

3.14E+03

6.99E+03

3.29E+01

4.52E+00

3.22E+03

7.20E+03

3.37E+01

4.65E+00

3.29E+03

7.40E+03

3.45E+01

4.79E+00

3.37E+03

7.62E+03

3.53E+01

4.94E+00

3.45E+03

7.83E+03

3.61E+01

5.08E+00

3.53E+03

8.06E+03

3.69E+01

5.24E+00

3.61E+03

8.29E+03

3.78E+01

5.39E+00

3.69E+03

8.53E+03

3.87E+01

5.56E+00

3.78E+03

8.78E+03

3.96E+01

5.72E+00

3.87E+03

9.03E+03

4.06E+01

5.89E+00

3.96E+03

9.29E+03

4.15E+01

6.07E+00

4.06E+03

9.56E+03

4.25E+01

6.25E+00

4.15E+03

9.84E+03

4.35E+01

6.44E+00

4.25E+03

1.01E+04

4.46E+01

6.63E+00

4.36E+03

1.04E+04

4.56E+01

6.83E+00

4.46E+03

1.07E+04

4.67E+01

7.04E+00

4.57E+03

1.10E+04

4.79E+01

7.25E+00

4.68E+03

1.13E+04

4.90E+01

7.47E+00

4.79E+03

1.17E+04

5.02E+01

7.69E+00

4.91E+03

1.20E+04

5.15E+01

7.92E+00

5.03E+03

1.24E+04

5.27E+01

8.16E+00

5.15E+03

1.27E+04

5.40E+01

8.40E+00

5.28E+03

1.31E+04

5.53E+01

8.66E+00

5.41E+03

1.35E+04

5.67E+01

8.92E+00

5.54E+03

1.39E+04

5.81E+01

9.18E+00

5.68E+03

1.43E+04

5.95E+01

9.46E+00

5.82E+03

1.47E+04

6.10E+01

9.74E+00

5.97E+03

1.51E+04

6.25E+01

1.00E+01

6.10E+03

1.55E+04

6.39E+01

1.00E+01

6.12E+03

1.56E+04

6.41E+01

1

Non-gestational 5-year Average

Intake

(ng/kg/
day)

Fat
(ng/kg)

Body
Burden
(ng/kg)

Blood
(ng/kg)

1.34E+01

7.77E+03

2.05E+04

8.15E+01

1.67E+01

9.43E+03

2.55E+04

9.88E+01

2.00E+01

1.11E+04

3.05E+04

1.16E+02

2.33E+01

1.27E+04

3.54E+04

1.33E+02

2.67E+01

1.43E+04

4.03E+04

1.50E+02

3.00E+01

1.60E+04

4.53E+04

1.67E+02

3.33E+01

1.76E+04

5.02E+04

1.84E+02

3.67E+01

1.92E+04

5.51E+04

2.01E+02

4.00E+01

2.08E+04

6.00E+04

2.18E+02

4.33E+01

2.24E+04

6.49E+04

2.35E+02

4.67E+01

2.40E+04

6.97E+04

2.52E+02

5.00E+01

2.57E+04

7.46E+04

2.69E+02

5.33E+01

2.73E+04

7.94E+04

2.86E+02

5.67E+01

2.89E+04

8.43E+04

3.03E+02

6.00E+01

3.05E+04

8.91E+04

3.19E+02

6.33E+01

3.21E+04

9.39E+04

3.36E+02

6.67E+01

3.37E+04

9.87E+04

3.53E+02

7.00E+01

3.53E+04

1.04E+05

3.70E+02

7.33E+01

3.69E+04

1.08E+05

3.87E+02

7.67E+01

3.85E+04

1.13E+05

4.04E+02

8.00E+01

4.01E+04

1.18E+05

4.20E+02

8.33E+01

4.17E+04

1.23E+05

4.37E+02

8.67E+01

4.33E+04

1.27E+05

4.54E+02

9.00E+01

4.49E+04

1.32E+05

4.71E+02

9.33E+01

4.65E+04

1.37E+05

4.88E+02

9.67E+01

4.81E+04

1.41E+05

5.04E+02

1.00E+02

4.97E+04

1.46E+05

5.21E+02

1.10E+02

5.45E+04

1.60E+05

5.72E+02

1.20E+02

5.94E+04

1.74E+05

6.22E+02

-------
C.4.3. Gestational

o



Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

1.00E-09

2.81E-05

1.11E-05

2.96E-07

1.33E-09

3.74E-05

1.47E-05

3.94E-07

1.67E-09

4.68E-05

1.84E-05

4.92E-07

2.00E-09

5.61E-05

2.21E-05

5.91E-07

2.33E-09

6.55E-05

2.58E-05

6.89E-07

2.67E-09

7.48E-05

2.95E-05

7.88E-07

3.00E-09

8.42E-05

3.32E-05

8.86E-07

3.33E-09

9.35E-05

3.69E-05

9.84E-07

3.67E-09

1.03E-04

4.05E-05

1.08E-06

4.00E-09

1.12E-04

4.42E-05

1.18E-06

4.33E-09

1.22E-04

4.79E-05

1.28E-06

4.67E-09

1.31E-04

5.16E-05

1.38E-06

5.00E-09

1.40E-04

5.53E-05

1.48E-06

5.33E-09

1.50E-04

5.90E-05

1.57E-06

5.67E-09

1.59E-04

6.26E-05

1.67E-06

6.00E-09

1.68E-04

6.63E-05

1.77E-06

6.33E-09

1.78E-04

7.00E-05

1.87E-06

6.67E-09

1.87E-04

7.37E-05

1.97E-06

7.00E-09

1.96E-04

7.74E-05

2.07E-06

7.33E-09

2.06E-04

8.11E-05

2.16E-06

7.67E-09

2.15E-04

8.47E-05

2.26E-06

8.00E-09

2.24E-04

8.84E-05

2.36E-06

8.33E-09

2.34E-04

9.21E-05

2.46E-06

8.67E-09

2.43E-04

9.58E-05

2.56E-06

9.00E-09

2.52E-04

9.95E-05

2.66E-06

9.33E-09

2.62E-04

1.03E-04

2.75E-06

9.67E-09

2.71E-04

1.07E-04

2.85E-06

1.00E-08

2.80E-04

1.11E-04

2.95E-06

1.33E-08

3.73E-04

1.47E-04

3.93E-06

1.67E-08

4.66E-04

1.84E-04

4.91E-06

2.00E-08

5.59E-04

2.21E-04

5.89E-06

2.33E-08

6.52E-04

2.57E-04

6.87E-06

2.67E-08

7.46E-04

2.94E-04

7.85E-06

3.00E-08

8.39E-04

3.31E-04

8.83E-06

3.33E-08

9.32E-04

3.67E-04

9.81E-06

3.67E-08

1.02E-03

4.04E-04

1.08E-05

4.00E-08

1.12E-03

4.41E-04

1.18E-05

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

4.33E-08

1.21E-03

4.78E-04

1.27E-05

4.67E-08

1.30E-03

5.14E-04

1.37E-05

5.00E-08

1.40E-03

5.51E-04

1.47E-05

5.33E-08

1.49E-03

5.88E-04

1.57E-05

5.67E-08

1.58E-03

6.24E-04

1.67E-05

6.00E-08

1.67E-03

6.61E-04

1.76E-05

6.33E-08

1.77E-03

6.97E-04

1.86E-05

6.67E-08

1.86E-03

7.34E-04

1.96E-05

7.00E-08

1.95E-03

7.70E-04

2.05E-05

7.33E-08

2.04E-03

8.07E-04

2.15E-05

7.67E-08

2.14E-03

8.43E-04

2.25E-05

8.00E-08

2.23E-03

8.80E-04

2.35E-05

8.33E-08

2.32E-03

9.17E-04

2.44E-05

8.67E-08

2.41E-03

9.53E-04

2.54E-05

9.00E-08

2.51E-03

9.90E-04

2.64E-05

9.33E-08

2.60E-03

1.03E-03

2.74E-05

9.67E-08

2.69E-03

1.06E-03

2.83E-05

1.00E-07

2.79E-03

1.10E-03

2.93E-05

1.33E-07

3.70E-03

1.46E-03

3.90E-05

1.67E-07

4.62E-03

1.83E-03

4.86E-05

2.00E-07

5.54E-03

2.19E-03

5.83E-05

2.33E-07

6.46E-03

2.55E-03

6.80E-05

2.67E-07

7.37E-03

2.92E-03

7.76E-05

3.00E-07

8.29E-03

3.28E-03

8.73E-05

3.33E-07

9.21E-03

3.64E-03

9.69E-05

3.67E-07

1.01E-02

4.01E-03

1.07E-04

4.00E-07

1.10E-02

4.37E-03

1.16E-04

4.33E-07

1.20E-02

4.74E-03

1.26E-04

4.67E-07

1.29E-02

5.10E-03

1.36E-04

5.00E-07

1.38E-02

5.46E-03

1.45E-04

5.33E-07

1.47E-02

5.82E-03

1.55E-04

5.66E-07

1.56E-02

6.17E-03

1.64E-04

5.99E-07

1.65E-02

6.53E-03

1.73E-04

6.33E-07

1.74E-02

6.88E-03

1.83E-04

6.66E-07

1.83E-02

7.24E-03

1.92E-04

6.99E-07

1.92E-02

7.59E-03

2.02E-04

7.32E-07

2.01E-02

7.95E-03

2.11E-04

7.65E-07

2.09E-02

8.30E-03

2.20E-04

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

7.98E-07

2.18E-02

8.66E-03

2.30E-04

8.32E-07

2.27E-02

9.01E-03

2.39E-04

8.65E-07

2.36E-02

9.37E-03

2.49E-04

8.98E-07

2.45E-02

9.72E-03

2.58E-04

9.31E-07

2.54E-02

1.01E-02

2.67E-04

9.64E-07

2.63E-02

1.04E-02

2.77E-04

9.97E-07

2.72E-02

1.08E-02

2.86E-04

1.01E-06

2.76E-02

1.09E-02

2.90E-04

1.03E-06

2.80E-02

1.11E-02

2.95E-04

1.04E-06

2.84E-02

1.13E-02

2.99E-04

1.06E-06

2.88E-02

1.14E-02

3.03E-04

1.07E-06

2.93E-02

1.16E-02

3.08E-04

1.09E-06

2.97E-02

1.18E-02

3.12E-04

1.11E-06

3.01E-02

1.20E-02

3.17E-04

1.12E-06

3.06E-02

1.21E-02

3.22E-04

1.14E-06

3.10E-02

1.23E-02

3.26E-04

1.16E-06

3.15E-02

1.25E-02

3.31E-04

1.17E-06

3.19E-02

1.27E-02

3.36E-04

1.19E-06

3.24E-02

1.29E-02

3.41E-04

1.21E-06

3.29E-02

1.31E-02

3.46E-04

1.23E-06

3.34E-02

1.32E-02

3.51E-04

1.24E-06

3.38E-02

1.34E-02

3.56E-04

1.26E-06

3.43E-02

1.36E-02

3.61E-04

1.28E-06

3.48E-02

1.38E-02

3.67E-04

1.30E-06

3.54E-02

1.40E-02

3.72E-04

1.32E-06

3.59E-02

1.42E-02

3.77E-04

1.34E-06

3.64E-02

1.45E-02

3.83E-04

1.36E-06

3.69E-02

1.47E-02

3.89E-04

1.38E-06

3.75E-02

1.49E-02

3.94E-04

1.40E-06

3.80E-02

1.51E-02

4.00E-04

1.42E-06

3.86E-02

1.53E-02

4.06E-04

1.44E-06

3.92E-02

1.56E-02

4.12E-04

1.46E-06

3.98E-02

1.58E-02

4.18E-04

1.49E-06

4.03E-02

1.60E-02

4.25E-04

1.53E-06

4.15E-02

1.65E-02

4.37E-04

1.58E-06

4.27E-02

1.70E-02

4.50E-04

1.62E-06

4.40E-02

1.75E-02

4.63E-04

1.67E-06

4.53E-02

1.80E-02

4.76E-04

-------
o



Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

1.72E-06

4.66E-02

1.85E-02

4.90E-04

1.77E-06

4.80E-02

1.91E-02

5.05E-04

1.83E-06

4.94E-02

1.96E-02

5.20E-04

1.88E-06

5.08E-02

2.02E-02

5.35E-04

1.94E-06

5.23E-02

2.08E-02

5.50E-04

2.00E-06

5.38E-02

2.14E-02

5.66E-04

2.06E-06

5.54E-02

2.21E-02

5.83E-04

2.12E-06

5.70E-02

2.27E-02

6.00E-04

2.18E-06

5.87E-02

2.34E-02

6.17E-04

2.25E-06

6.04E-02

2.41E-02

6.35E-04

2.32E-06

6.22E-02

2.48E-02

6.54E-04

2.39E-06

6.40E-02

2.55E-02

6.73E-04

2.46E-06

6.58E-02

2.62E-02

6.93E-04

2.53E-06

6.77E-02

2.70E-02

7.13E-04

2.61E-06

6.97E-02

2.78E-02

7.33E-04

2.68E-06

7.17E-02

2.86E-02

7.55E-04

2.76E-06

7.38E-02

2.94E-02

7.77E-04

2.85E-06

7.60E-02

3.03E-02

8.00E-04

2.93E-06

7.82E-02

3.12E-02

8.22E-04

3.02E-06

8.04E-02

3.21E-02

8.46E-04

3.11E-06

8.27E-02

3.30E-02

8.71E-04

3.21E-06

8.51E-02

3.40E-02

8.96E-04

3.30E-06

8.76E-02

3.50E-02

9.22E-04

3.40E-06

9.01E-02

3.60E-02

9.48E-04

3.50E-06

9.27E-02

3.71E-02

9.76E-04

3.61E-06

9.54E-02

3.81E-02

1.00E-03

3.72E-06

9.82E-02

3.93E-02

1.03E-03

3.83E-06

1.01E-01

4.04E-02

1.06E-03

3.94E-06

1.04E-01

4.16E-02

1.09E-03

4.06E-06

1.07E-01

4.28E-02

1.12E-03

4.18E-06

1.10E-01

4.40E-02

1.16E-03

4.31E-06

1.13E-01

4.53E-02

1.19E-03

4.44E-06

1.16E-01

4.66E-02

1.22E-03

4.57E-06

1.20E-01

4.79E-02

1.26E-03

4.71E-06

1.23E-01

4.93E-02

1.30E-03

4.85E-06

1.27E-01

5.08E-02

1.33E-03

4.99E-06

1.30E-01

5.22E-02

1.37E-03

5.14E-06

1.34E-01

5.37E-02

1.41E-03

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

5.30E-06

1.38E-01

5.53E-02

1.45E-03

5.46E-06

1.42E-01

5.69E-02

1.49E-03

5.62E-06

1.46E-01

5.85E-02

1.53E-03

5.79E-06

1.50E-01

6.02E-02

1.58E-03

5.96E-06

1.54E-01

6.19E-02

1.62E-03

6.14E-06

1.59E-01

6.37E-02

1.67E-03

6.33E-06

1.63E-01

6.55E-02

1.72E-03

6.52E-06

1.68E-01

6.74E-02

1.76E-03

6.71E-06

1.72E-01

6.93E-02

1.81E-03

6.91E-06

1.77E-01

7.13E-02

1.86E-03

7.12E-06

1.82E-01

7.33E-02

1.92E-03

7.33E-06

1.87E-01

7.54E-02

1.97E-03

7.55E-06

1.93E-01

7.75E-02

2.03E-03

7.78E-06

1.98E-01

7.97E-02

2.08E-03

8.01E-06

2.03E-01

8.20E-02

2.14E-03

8.25E-06

2.09E-01

8.43E-02

2.20E-03

8.50E-06

2.15E-01

8.67E-02

2.26E-03

8.76E-06

2.21E-01

8.92E-02

2.33E-03

9.02E-06

2.27E-01

9.17E-02

2.39E-03

9.29E-06

2.34E-01

9.43E-02

2.46E-03

9.57E-06

2.40E-01

9.70E-02

2.53E-03

9.86E-06

2.47E-01

9.97E-02

2.60E-03

1.02E-05

2.54E-01

1.03E-01

2.67E-03

1.05E-05

2.61E-01

1.05E-01

2.74E-03

1.08E-05

2.68E-01

1.08E-01

2.82E-03

1.11E-05

2.75E-01

1.11E-01

2.90E-03

1.14E-05

2.83E-01

1.15E-01

2.98E-03

1.18E-05

2.91E-01

1.18E-01

3.06E-03

1.21E-05

2.99E-01

1.21E-01

3.14E-03

1.25E-05

3.07E-01

1.25E-01

3.23E-03

1.29E-05

3.16E-01

1.28E-01

3.32E-03

1.32E-05

3.24E-01

1.32E-01

3.41E-03

1.36E-05

3.33E-01

1.35E-01

3.51E-03

1.41E-05

3.42E-01

1.39E-01

3.60E-03

1.45E-05

3.52E-01

1.43E-01

3.70E-03

1.49E-05

3.61E-01

1.47E-01

3.80E-03

1.54E-05

3.71E-01

1.51E-01

3.90E-03

1.58E-05

3.81E-01

1.55E-01

4.01E-03

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

1.63E-05

3.92E-01

1.60E-01

4.12E-03

1.68E-05

4.03E-01

1.64E-01

4.23E-03

1.73E-05

4.13E-01

1.69E-01

4.35E-03

1.78E-05

4.25E-01

1.73E-01

4.47E-03

1.83E-05

4.36E-01

1.78E-01

4.59E-03

1.89E-05

4.48E-01

1.83E-01

4.71E-03

1.95E-05

4.60E-01

1.88E-01

4.84E-03

2.00E-05

4.72E-01

1.93E-01

4.97E-03

2.06E-05

4.85E-01

1.98E-01

5.10E-03

2.13E-05

4.98E-01

2.04E-01

5.24E-03

2.19E-05

5.12E-01

2.10E-01

5.38E-03

2.25E-05

5.25E-01

2.15E-01

5.53E-03

2.32E-05

5.40E-01

2.21E-01

5.68E-03

2.39E-05

5.54E-01

2.27E-01

5.83E-03

2.46E-05

5.69E-01

2.34E-01

5.98E-03

2.54E-05

5.84E-01

2.40E-01

6.14E-03

2.61E-05

6.00E-01

2.47E-01

6.31E-03

2.69E-05

6.16E-01

2.53E-01

6.48E-03

2.77E-05

6.32E-01

2.60E-01

6.65E-03

2.86E-05

6.49E-01

2.67E-01

6.82E-03

2.94E-05

6.66E-01

2.75E-01

7.01E-03

3.03E-05

6.84E-01

2.82E-01

7.19E-03

3.12E-05

7.02E-01

2.90E-01

7.38E-03

3.21E-05

7.20E-01

2.98E-01

7.58E-03

3.31E-05

7.42E-01

3.07E-01

7.80E-03

3.41E-05

7.62E-01

3.15E-01

8.01E-03

3.51E-05

7.82E-01

3.24E-01

8.22E-03

3.62E-05

8.03E-01

3.33E-01

8.44E-03

3.73E-05

8.24E-01

3.42E-01

8.68E-03

3.84E-05

8.45E-01

3.51E-01

8.89E-03

3.95E-05

8.68E-01

3.61E-01

9.12E-03

4.07E-05

8.88E-01

3.69E-01

9.34E-03

4.19E-05

9.11E-01

3.79E-01

9.59E-03

4.32E-05

9.35E-01

3.89E-01

9.83E-03

4.45E-05

9.59E-01

4.00E-01

1.01E-02

4.58E-05

9.83E-01

4.10E-01

1.03E-02

4.72E-05

1.01E+00

4.21E-01

1.06E-02

4.86E-05

1.04E+00

4.33E-01

1.09E-02

-------


Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

5.01E-05

1.06E+00

4.44E-01

1.12E-02

5.16E-05

1.09E+00

4.56E-01

1.14E-02

5.31E-05

1.12E+00

4.68E-01

1.17E-02

5.47E-05

1.15E+00

4.81E-01

1.21E-02

5.64E-05

1.18E+00

4.93E-01

1.24E-02

5.81E-05

1.21E+00

5.06E-01

1.27E-02

5.98E-05

1.24E+00

5.20E-01

1.30E-02

6.16E-05

1.27E+00

5.34E-01

1.33E-02

6.34E-05

1.30E+00

5.48E-01

1.37E-02

6.54E-05

1.33E+00

5.62E-01

1.40E-02

6.73E-05

1.37E+00

5.77E-01

1.44E-02

6.93E-05

1.40E+00

5.92E-01

1.47E-02

7.14E-05

1.44E+00

6.08E-01

1.51E-02

7.36E-05

1.47E+00

6.24E-01

1.55E-02

7.58E-05

1.51E+00

6.40E-01

1.59E-02

7.80E-05

1.55E+00

6.57E-01

1.63E-02

8.04E-05

1.59E+00

6.74E-01

1.67E-02

8.28E-05

1.63E+00

6.92E-01

1.71E-02

8.53E-05

1.67E+00

7.10E-01

1.75E-02

8.78E-05

1.71E+00

7.28E-01

1.79E-02

9.05E-05

1.75E+00

7.47E-01

1.84E-02

9.32E-05

1.79E+00

7.66E-01

1.88E-02

9.60E-05

1.84E+00

7.86E-01

1.93E-02

9.89E-05

1.88E+00

8.07E-01

1.98E-02

1.02E-04

1.93E+00

8.28E-01

2.03E-02

1.05E-04

1.98E+00

8.49E-01

2.08E-02

1.08E-04

2.03E+00

8.71E-01

2.13E-02

1.11E-04

2.08E+00

8.93E-01

2.18E-02

1.15E-04

2.13E+00

9.16E-01

2.24E-02

1.18E-04

2.18E+00

9.39E-01

2.29E-02

1.22E-04

2.23E+00

9.63E-01

2.34E-02

1.25E-04

2.28E+00

9.87E-01

2.40E-02

1.29E-04

2.34E+00

1.01E+00

2.46E-02

1.33E-04

2.40E+00

1.04E+00

2.52E-02

1.37E-04

2.45E+00

1.06E+00

2.58E-02

1.41E-04

2.51E+00

1.09E+00

2.64E-02

1.45E-04

2.58E+00

1.12E+00

2.72E-02

1.50E-04

2.63E+00

1.15E+00

2.77E-02

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

1.54E-04

2.70E+00

1.18E+00

2.83E-02

1.59E-04

2.78E+00

1.21E+00

2.92E-02

1.63E-04

2.84E+00

1.24E+00

2.99E-02

1.68E-04

2.89E+00

1.27E+00

3.04E-02

1.73E-04

2.96E+00

1.30E+00

3.11E-02

1.79E-04

3.03E+00

1.33E+00

3.18E-02

1.84E-04

3.10E+00

1.36E+00

3.26E-02

1.89E-04

3.17E+00

1.40E+00

3.33E-02

1.95E-04

3.25E+00

1.43E+00

3.41E-02

2.01E-04

3.32E+00

1.47E+00

3.49E-02

2.07E-04

3.43E+00

1.52E+00

3.61E-02

2.13E-04

3.51E+00

1.56E+00

3.69E-02

2.20E-04

3.57E+00

1.59E+00

3.75E-02

2.26E-04

3.67E+00

1.63E+00

3.85E-02

2.33E-04

3.77E+00

1.68E+00

3.96E-02

2.40E-04

3.86E+00

1.72E+00

4.05E-02

2.47E-04

3.95E+00

1.76E+00

4.15E-02

2.55E-04

4.04E+00

1.81E+00

4.24E-02

2.62E-04

4.13E+00

1.85E+00

4.34E-02

2.70E-04

4.22E+00

1.90E+00

4.44E-02

2.78E-04

4.32E+00

1.94E+00

4.54E-02

2.86E-04

4.42E+00

1.99E+00

4.64E-02

2.95E-04

4.52E+00

2.04E+00

4.75E-02

3.04E-04

4.62E+00

2.09E+00

4.86E-02

3.13E-04

4.73E+00

2.14E+00

4.97E-02

3.22E-04

4.84E+00

2.20E+00

5.08E-02

3.32E-04

4.95E+00

2.25E+00

5.20E-02

3.42E-04

5.06E+00

2.30E+00

5.31E-02

3.52E-04

5.17E+00

2.36E+00

5.43E-02

3.63E-04

5.29E+00

2.42E+00

5.56E-02

3.74E-04

5.41E+00

2.48E+00

5.68E-02

3.85E-04

5.53E+00

2.54E+00

5.81E-02

3.97E-04

5.65E+00

2.60E+00

5.94E-02

4.08E-04

5.78E+00

2.66E+00

6.07E-02

4.21E-04

5.91E+00

2.73E+00

6.20E-02

4.33E-04

6.04E+00

2.79E+00

6.34E-02

4.46E-04

6.17E+00

2.86E+00

6.48E-02

4.60E-04

6.31E+00

2.93E+00

6.63E-02

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

4.74E-04

6.45E+00

3.00E+00

6.77E-02

4.88E-04

6.59E+00

3.07E+00

6.92E-02

5.02E-04

6.74E+00

3.15E+00

7.07E-02

5.17E-04

6.88E+00

3.22E+00

7.23E-02

5.33E-04

7.03E+00

3.30E+00

7.39E-02

5.49E-04

7.19E+00

3.38E+00

7.55E-02

5.65E-04

7.34E+00

3.46E+00

7.71E-02

5.82E-04

7.50E+00

3.54E+00

7.88E-02

6.00E-04

7.67E+00

3.63E+00

8.05E-02

6.18E-04

7.83E+00

3.71E+00

8.22E-02

6.36E-04

8.00E+00

3.80E+00

8.40E-02

6.55E-04

8.17E+00

3.89E+00

8.58E-02

6.75E-04

8.35E+00

3.98E+00

8.77E-02

6.95E-04

8.53E+00

4.08E+00

8.95E-02

7.16E-04

8.70E+00

4.17E+00

9.14E-02

7.38E-04

8.89E+00

4.27E+00

9.33E-02

7.60E-04

9.08E+00

4.37E+00

9.53E-02

7.83E-04

9.27E+00

4.47E+00

9.74E-02

8.06E-04

9.47E+00

4.58E+00

9.94E-02

8.30E-04

9.67E+00

4.69E+00

1.02E-01

8.55E-04

9.88E+00

4.80E+00

1.04E-01

8.81E-04

1.01E+01

4.91E+00

1.06E-01

9.07E-04

1.03E+01

5.03E+00

1.08E-01

9.21E-04

1.04E+01

5.09E+00

1.09E-01

9.35E-04

1.05E+01

5.14E+00

1.10E-01

9.49E-04

1.26E+01

6.31E+00

1.32E-01

1.37E-03

1.38E+01

6.99E+00

1.45E-01

1.39E-03

1.40E+01

7.07E+00

1.46E-01

1.41E-03

1.41E+01

7.15E+00

1.48E-01

1.43E-03

1.42E+01

7.23E+00

1.49E-01

1.46E-03

1.44E+01

7.31E+00

1.51E-01

1.48E-03

1.45E+01

7.39E+00

1.52E-01

1.50E-03

1.46E+01

7.47E+00

1.54E-01

1.52E-03

1.48E+01

7.55E+00

1.55E-01

1.54E-03

1.49E+01

7.64E+00

1.57E-01

1.57E-03

1.51E+01

7.73E+00

1.58E-01

1.59E-03

1.52E+01

7.82E+00

1.60E-01

1.61E-03

1.54E+01

7.91E+00

1.62E-01

-------


Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

1.64E-03

1.56E+01

8.00E+00

1.63E-01

1.69E-03

1.59E+01

8.19E+00

1.67E-01

1.71E-03

1.60E+01

8.28E+00

1.68E-01

1.74E-03

1.62E+01

8.38E+00

1.70E-01

1.76E-03

1.64E+01

8.47E+00

1.72E-01

1.79E-03

1.65E+01

8.57E+00

1.73E-01

1.82E-03

1.67E+01

8.67E+00

1.75E-01

1.84E-03

1.69E+01

8.77E+00

1.77E-01

1.87E-03

1.74E+01

9.10E+00

1.83E-01

2.34E-03

1.98E+01

1.06E+01

2.08E-01

2.41E-03

2.02E+01

1.08E+01

2.12E-01

2.48E-03

2.06E+01

1.11E+01

2.16E-01

2.55E-03

2.10E+01

1.13E+01

2.21E-01

2.63E-03

2.14E+01

1.16E+01

2.25E-01

2.71E-03

2.19E+01

1.18E+01

2.30E-01

2.79E-03

2.23E+01

1.21E+01

2.34E-01

2.87E-03

2.28E+01

1.24E+01

2.39E-01

2.96E-03

2.32E+01

1.27E+01

2.44E-01

3.05E-03

2.37E+01

1.30E+01

2.48E-01

3.14E-03

2.41E+01

1.33E+01

2.53E-01

3.23E-03

2.46E+01

1.36E+01

2.58E-01

3.33E-03

2.51E+01

1.39E+01

2.63E-01

3.43E-03

2.56E+01

1.42E+01

2.69E-01

3.53E-03

2.61E+01

1.46E+01

2.74E-01

3.64E-03

2.66E+01

1.49E+01

2.79E-01

4.22E-03

2.83E+01

1.60E+01

2.96E-01

4.35E-03

2.88E+01

1.63E+01

3.02E-01

4.48E-03

2.93E+01

1.67E+01

3.08E-01

4.61E-03

2.99E+01

1.71E+01

3.14E-01

4.75E-03

3.05E+01

1.75E+01

3.20E-01

4.89E-03

3.11E+01

1.79E+01

3.26E-01

5.04E-03

3.30E+01

1.92E+01

3.46E-01

5.19E-03

3.41E+01

2.00E+01

3.58E-01

5.35E-03

3.49E+01

2.05E+01

3.66E-01

5.51E-03

3.55E+01

2.10E+01

3.73E-01

5.67E-03

3.62E+01

2.14E+01

3.80E-01

5.84E-03

3.69E+01

2.19E+01

3.87E-01

5.93E-03

3.73E+01

2.22E+01

3.91E-01

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

6.02E-03

3.77E+01

2.25E+01

3.95E-01

6.20E-03

3.84E+01

2.30E+01

4.03E-01

6.38E-03

3.92E+01

2.36E+01

4.12E-01

6.57E-03

4.00E+01

2.42E+01

4.20E-01

6.77E-03

4.08E+01

2.48E+01

4.28E-01

6.98E-03

4.16E+01

2.54E+01

4.37E-01

7.18E-03

4.25E+01

2.60E+01

4.45E-01

7.40E-03

4.33E+01

2.66E+01

4.54E-01

7.51E-03

4.37E+01

2.69E+01

4.58E-01

8.33E-03

4.46E+01

2.76E+01

4.68E-01

8.58E-03

4.66E+01

2.91E+01

4.89E-01

8.71E-03

4.74E+01

2.97E+01

4.97E-01

8.84E-03

4.79E+01

3.00E+01

5.02E-01

9.10E-03

4.86E+01

3.06E+01

5.10E-01

9.37E-03

4.95E+01

3.13E+01

5.19E-01

9.66E-03

5.07E+01

3.22E+01

5.32E-01

9.94E-03

5.17E+01

3.30E+01

5.42E-01

1.02E-02

5.27E+01

3.38E+01

5.53E-01

1.06E-02

5.37E+01

3.46E+01

5.63E-01

1.09E-02

5.46E+01

3.53E+01

5.73E-01

1.12E-02

5.58E+01

3.63E+01

5.85E-01

1.15E-02

5.67E+01

3.69E+01

5.94E-01

1.19E-02

5.74E+01

3.75E+01

6.02E-01

1.22E-02

5.85E+01

3.84E+01

6.13E-01

1.26E-02

5.96E+01

3.93E+01

6.25E-01

1.30E-02

6.19E+01

4.12E+01

6.49E-01

1.34E-02

6.32E+01

4.23E+01

6.63E-01

1.38E-02

6.45E+01

4.33E+01

6.76E-01

1.42E-02

6.57E+01

4.44E+01

6.89E-01

1.46E-02

6.64E+01

4.49E+01

6.96E-01

1.50E-02

6.78E+01

4.61E+01

7.11E-01

1.55E-02

6.83E+01

4.66E+01

7.16E-01

1.60E-02

6.96E+01

4.76E+01

7.29E-01

1.64E-02

7.09E+01

4.88E+01

7.43E-01

1.69E-02

7.26E+01

5.02E+01

7.61E-01

1.74E-02

7.40E+01

5.14E+01

7.76E-01

1.80E-02

7.54E+01

5.27E+01

7.90E-01

1.85E-02

7.68E+01

5.39E+01

8.06E-01

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

1.91E-02

7.83E+01

5.52E+01

8.21E-01

1.96E-02

7.98E+01

5.66E+01

8.37E-01

2.02E-02

8.13E+01

5.79E+01

8.53E-01

2.08E-02

8.29E+01

5.94E+01

8.69E-01

2.14E-02

8.45E+01

6.08E+01

8.86E-01

2.21E-02

8.61E+01

6.23E+01

9.03E-01

2.28E-02

8.78E+01

6.38E+01

9.20E-01

2.34E-02

8.95E+01

6.54E+01

9.38E-01

2.41E-02

9.12E+01

6.70E+01

9.56E-01

2.49E-02

9.29E+01

6.86E+01

9.75E-01

2.56E-02

9.47E+01

7.03E+01

9.93E-01

2.64E-02

9.65E+01

7.20E+01

1.01E+00

2.72E-02

9.84E+01

7.37E+01

1.03E+00

2.80E-02

1.00E+02

7.55E+01

1.05E+00

2.88E-02

1.02E+02

7.74E+01

1.07E+00

2.97E-02

1.04E+02

7.93E+01

1.09E+00

3.06E-02

1.07E+02

8.20E+01

1.12E+00

3.15E-02

1.09E+02

8.38E+01

1.14E+00

3.24E-02

1.11E+02

8.57E+01

1.16E+00

3.34E-02

1.13E+02

8.76E+01

1.18E+00

3.44E-02

1.15E+02

8.96E+01

1.20E+00

3.54E-02

1.16E+02

9.15E+01

1.22E+00

3.65E-02

1.18E+02

9.35E+01

1.24E+00

3.76E-02

1.21E+02

9.61E+01

1.27E+00

3.87E-02

1.23E+02

9.84E+01

1.29E+00

3.99E-02

1.26E+02

1.01E+02

1.32E+00

4.11E-02

1.28E+02

1.03E+02

1.34E+00

4.23E-02

1.31E+02

1.06E+02

1.37E+00

4.36E-02

1.34E+02

1.09E+02

1.40E+00

4.49E-02

1.36E+02

1.12E+02

1.43E+00

4.63E-02

1.39E+02

1.15E+02

1.45E+00

4.76E-02

1.42E+02

1.18E+02

1.48E+00

4.91E-02

1.44E+02

1.21E+02

1.51E+00

5.05E-02

1.47E+02

1.24E+02

1.54E+00

5.21E-02

1.50E+02

1.27E+02

1.57E+00

5.36E-02

1.52E+02

1.30E+02

1.60E+00

5.52E-02

1.55E+02

1.33E+02

1.63E+00

5.69E-02

1.59E+02

1.37E+02

1.66E+00

-------


Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

5.86E-02

1.62E+02

1.40E+02

1.69E+00

6.03E-02

1.64E+02

1.43E+02

1.72E+00

6.22E-02

1.67E+02

1.46E+02

1.75E+00

6.40E-02

1.70E+02

1.50E+02

1.79E+00

6.59E-02

1.74E+02

1.54E+02

1.82E+00

6.79E-02

1.77E+02

1.58E+02

1.86E+00

7.00E-02

1.80E+02

1.62E+02

1.89E+00

7.21E-02

1.84E+02

1.66E+02

1.93E+00

7.42E-02

1.87E+02

1.70E+02

1.96E+00

7.64E-02

1.91E+02

1.75E+02

2.00E+00

7.87E-02

1.95E+02

1.79E+02

2.05E+00

8.11E-02

1.99E+02

1.84E+02

2.09E+00

8.35E-02

2.03E+02

1.89E+02

2.13E+00

8.60E-02

2.07E+02

1.93E+02

2.17E+00

8.86E-02

2.11E+02

1.98E+02

2.21E+00

9.13E-02

2.15E+02

2.03E+02

2.25E+00

9.40E-02

2.19E+02

2.08E+02

2.29E+00

9.68E-02

2.23E+02

2.14E+02

2.34E+00

9.97E-02

2.28E+02

2.20E+02

2.39E+00

1.03E-01

2.32E+02

2.25E+02

2.43E+00

1.06E-01

2.36E+02

2.31E+02

2.48E+00

1.09E-01

2.40E+02

2.36E+02

2.52E+00

1.12E-01

2.45E+02

2.42E+02

2.57E+00

1.16E-01

2.50E+02

2.49E+02

2.62E+00

1.19E-01

2.55E+02

2.55E+02

2.67E+00

1.23E-01

2.60E+02

2.62E+02

2.72E+00

1.26E-01

2.65E+02

2.69E+02

2.78E+00

1.30E-01

2.70E+02

2.76E+02

2.83E+00

1.34E-01

2.75E+02

2.83E+02

2.89E+00

1.38E-01

2.81E+02

2.91E+02

2.95E+00

1.42E-01

2.87E+02

2.99E+02

3.00E+00

1.46E-01

2.92E+02

3.06E+02

3.06E+00

1.51E-01

2.97E+02

3.14E+02

3.12E+00

1.55E-01

3.03E+02

3.22E+02

3.18E+00

1.60E-01

3.09E+02

3.30E+02

3.24E+00

1.65E-01

3.15E+02

3.39E+02

3.30E+00

1.70E-01

3.21E+02

3.48E+02

3.37E+00

1.75E-01

3.27E+02

3.57E+02

3.43E+00

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

1.80E-01

3.34E+02

3.67E+02

3.50E+00

1.86E-01

3.40E+02

3.76E+02

3.57E+00

1.91E-01

3.47E+02

3.86E+02

3.64E+00

1.97E-01

3.54E+02

3.96E+02

3.71E+00

2.03E-01

3.60E+02

4.06E+02

3.78E+00

2.09E-01

3.68E+02

4.17E+02

3.85E+00

2.15E-01

3.75E+02

4.28E+02

3.93E+00

2.22E-01

3.82E+02

4.40E+02

4.01E+00

2.28E-01

3.90E+02

4.52E+02

4.09E+00

2.35E-01

3.98E+02

4.64E+02

4.17E+00

2.42E-01

4.05E+02

4.76E+02

4.25E+00

2.49E-01

4.13E+02

4.89E+02

4.33E+00

2.57E-01

4.22E+02

5.02E+02

4.42E+00

2.65E-01

4.30E+02

5.15E+02

4.51E+00

2.72E-01

4.38E+02

5.29E+02

4.60E+00

2.81E-01

4.47E+02

5.42E+02

4.68E+00

2.89E-01

4.55E+02

5.56E+02

4.77E+00

2.98E-01

4.64E+02

5.71E+02

4.87E+00

3.07E-01

4.73E+02

5.86E+02

4.96E+00

3.16E-01

4.83E+02

6.03E+02

5.06E+00

3.25E-01

4.92E+02

6.19E+02

5.16E+00

3.35E-01

5.02E+02

6.35E+02

5.26E+00

3.45E-01

5.13E+02

6.53E+02

5.37E+00

3.56E-01

5.23E+02

6.70E+02

5.48E+00

3.66E-01

5.33E+02

6.88E+02

5.59E+00

3.77E-01

5.44E+02

7.07E+02

5.70E+00

3.89E-01

5.55E+02

7.26E+02

5.81E+00

4.00E-01

5.66E+02

7.46E+02

5.93E+00

4.12E-01

5.77E+02

7.66E+02

6.05E+00

4.25E-01

5.88E+02

7.86E+02

6.17E+00

4.37E-01

6.00E+02

8.08E+02

6.29E+00

4.50E-01

6.12E+02

8.30E+02

6.42E+00

4.64E-01

6.24E+02

8.52E+02

6.54E+00

4.92E-01

6.50E+02

8.99E+02

6.81E+00

5.07E-01

6.63E+02

9.23E+02

6.95E+00

5.22E-01

6.76E+02

9.49E+02

7.09E+00

5.54E-01

7.04E+02

1.00E+03

7.38E+00

5.71E-01

7.18E+02

1.03E+03

7.53E+00

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

5.88E-01

7.32E+02

1.06E+03

7.68E+00

6.05E-01

7.47E+02

1.08E+03

7.83E+00

6.23E-01

7.62E+02

1.11E+03

7.99E+00

6.61E-01

7.94E+02

1.18E+03

8.32E+00

6.81E-01

8.10E+02

1.21E+03

8.49E+00

7.02E-01

8.27E+02

1.24E+03

8.67E+00

7.23E-01

8.43E+02

1.28E+03

8.84E+00

7.44E-01

8.61E+02

1.31E+03

9.02E+00

7.67E-01

8.78E+02

1.35E+03

9.21E+00

7.90E-01

8.96E+02

1.38E+03

9.40E+00

8.13E-01

9.15E+02

1.42E+03

9.59E+00

8.38E-01

9.33E+02

1.46E+03

9.78E+00

8.63E-01

9.53E+02

1.50E+03

9.99E+00

9.16E-01

9.93E+02

1.59E+03

1.04E+01

9.43E-01

1.01E+03

1.63E+03

1.06E+01

9.71E-01

1.03E+03

1.68E+03

1.08E+01

1.00E+00

1.06E+03

1.72E+03

1.11E+01

1.06E+00

1.10E+03

1.82E+03

1.15E+01

1.09E+00

1.12E+03

1.87E+03

1.18E+01

1.13E+00

1.15E+03

1.92E+03

1.20E+01

1.16E+00

1.17E+03

1.98E+03

1.23E+01

1.19E+00

1.20E+03

2.03E+03

1.25E+01

1.23E+00

1.22E+03

2.09E+03

1.28E+01

1.27E+00

1.25E+03

2.15E+03

1.31E+01

1.31E+00

1.27E+03

2.21E+03

1.33E+01

1.34E+00

1.30E+03

2.27E+03

1.36E+01

1.38E+00

1.33E+03

2.33E+03

1.39E+01

1.43E+00

1.35E+03

2.40E+03

1.42E+01

1.47E+00

1.38E+03

2.46E+03

1.45E+01

1.51E+00

1.41E+03

2.53E+03

1.48E+01

1.56E+00

1.44E+03

2.60E+03

1.51E+01

1.61E+00

1.47E+03

2.68E+03

1.55E+01

1.65E+00

1.51E+03

2.75E+03

1.58E+01

1.70E+00

1.54E+03

2.83E+03

1.61E+01

1.75E+00

1.57E+03

2.91E+03

1.65E+01

1.81E+00

1.61E+03

2.99E+03

1.68E+01

1.86E+00

1.64E+03

3.08E+03

1.72E+01

1.92E+00

1.68E+03

3.16E+03

1.76E+01

-------


Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

1.97E+00

1.71E+03

3.25E+03

1.79E+01

2.03E+00

1.75E+03

3.34E+03

1.83E+01

2.09E+00

1.79E+03

3.44E+03

1.87E+01

2.16E+00

1.83E+03

3.54E+03

1.91E+01

2.22E+00

1.87E+03

3.64E+03

1.96E+01

2.29E+00

1.91E+03

3.74E+03

2.00E+01

2.36E+00

1.95E+03

3.85E+03

2.04E+01

2.43E+00

1.99E+03

3.95E+03

2.09E+01

2.50E+00

2.04E+03

4.07E+03

2.13E+01

2.58E+00

2.08E+03

4.18E+03

2.18E+01

2.65E+00

2.13E+03

4.30E+03

2.23E+01

2.73E+00

2.17E+03

4.42E+03

2.28E+01

2.82E+00

2.22E+03

4.55E+03

2.33E+01

2.90E+00

2.27E+03

4.68E+03

2.38E+01

2.99E+00

2.32E+03

4.81E+03

2.44E+01

3.08E+00

2.38E+03

4.95E+03

2.49E+01

3.17E+00

2.43E+03

5.09E+03

2.55E+01

3.26E+00

2.48E+03

5.24E+03

2.60E+01

3.36E+00

2.54E+03

5.39E+03

2.66E+01

3.46E+00

2.60E+03

5.54E+03

2.72E+01

3.57E+00

2.66E+03

5.70E+03

2.79E+01

3.67E+00

2.72E+03

5.86E+03

2.85E+01

3.78E+00

2.78E+03

6.03E+03

2.91E+01

3.90E+00

2.84E+03

6.20E+03

2.98E+01

4.01E+00

2.91E+03

6.38E+03

3.05E+01

4.13E+00

2.98E+03

6.56E+03

3.12E+01

4.26E+00

3.04E+03

6.75E+03

3.19E+01

4.39E+00

3.12E+03

6.95E+03

3.27E+01

4.52E+00

3.19E+03

7.15E+03

3.34E+01

4.65E+00

3.26E+03

7.35E+03

3.42E+01

4.79E+00

3.34E+03

7.56E+03

3.50E+01

4.94E+00

3.42E+03

7.78E+03

3.58E+01

5.08E+00

3.50E+03

8.01E+03

3.66E+01

5.24E+00

3.58E+03

8.24E+03

3.75E+01

5.39E+00

3.66E+03

8.47E+03

3.84E+01

5.56E+00

3.75E+03

8.72E+03

3.93E+01

5.72E+00

3.84E+03

8.97E+03

4.02E+01

5.89E+00

3.93E+03

9.23E+03

4.12E+01

Gestational

Intake

Fat

Body Burden

Blood

(ng/kg/ day)

(ng/kg)

(ng/kg)

(ng/kg)

6.07E+00

4.02E+03

9.50E+03

4.22E+01

6.25E+00

4.12E+03

9.77E+03

4.32E+01

6.44E+00

4.22E+03

1.01E+04

4.42E+01

6.63E+00

4.32E+03

1.03E+04

4.53E+01

6.83E+00

4.42E+03

1.06E+04

4.64E+01

7.04E+00

4.53E+03

1.10E+04

4.75E+01

7.25E+00

4.64E+03

1.13E+04

4.86E+01

7.47E+00

4.75E+03

1.16E+04

4.98E+01

7.69E+00

4.87E+03

1.19E+04

5.10E+01

7.92E+00

4.99E+03

1.23E+04

5.23E+01

8.16E+00

5.11E+03

1.26E+04

5.36E+01

8.40E+00

5.24E+03

1.30E+04

5.49E+01

8.66E+00

5.37E+03

1.34E+04

5.62E+01

8.92E+00

5.50E+03

1.38E+04

5.76E+01

9.18E+00

5.63E+03

1.42E+04

5.91E+01

9.46E+00

5.77E+03

1.46E+04

6.05E+01

9.74E+00

5.92E+03

1.50E+04

6.20E+01

1.00E+01

6.05E+03

1.54E+04

6.34E+01

1.00E+01

6.07E+03

1.54E+04

6.36E+01

1.34E+01

7.71E+03

2.04E+04

8.08E+01

1.67E+01

9.35E+03

2.53E+04

9.80E+01

2.00E+01

1.10E+04

3.02E+04

1.15E+02

2.33E+01

1.26E+04

3.52E+04

1.32E+02

2.67E+01

1.42E+04

4.01E+04

1.49E+02

3.00E+01

1.58E+04

4.50E+04

1.66E+02

3.33E+01

1.74E+04

4.98E+04

1.83E+02

3.67E+01

1.90E+04

5.47E+04

2.00E+02

4.00E+01

2.07E+04

5.96E+04

2.17E+02

4.33E+01

2.23E+04

6.44E+04

2.33E+02

4.67E+01

2.39E+04

6.93E+04

2.50E+02

5.00E+01

2.54E+04

7.41E+04

2.67E+02

-------
1 C.5. REFERENCES

2	Amin, S; Moore, RW; Peterson, RE; et al. (2000) Gestational and lactational exposure to TCDD or coplanar PCBs

3	alters adult expression of saccharin preference behavior in female rats. Neurotoxicol Teratol 22(5):675-682.

4	Aylward, LL; Brunet, RC; Carrier, G; et al. (2005a) Concentration-dependent TCDD elimination kinetics in

5	humans: toxicokinetic modeling for moderately to highly exposed adults from Seveso, Italy, and Vienna, Austria,

6	and impact on dose estimates for the NIOSH cohort. J Exp Anal Environ Epidemiol 15:51-65.

7	Aylward, LL; Brunet, RC; Starr, TB; et al. (2005b) Exposure reconstruction for the TCDD-exposed NIOSH cohort

8	using a concentration- and age-dependent model of elimination. Risk Anal 25(4):945-956.

9	Aylward, LL; Bodner, KM; Collins, JJ; et al. (2009) TCDD exposure estimation for workers at a New Zealand

10	2,4,5-T manufacturing facility based on serum sampling data. J Expo Sci Environ Epidemiol. 3 June;

11	doi:10.1038/jes.2009.31. Available online at http://www.nature.com/jes/journal/vaop/ncurrent/full/jes200931a.html.

12	Bell, DR; Clode, S; Fan, MQ; et al. (2007) Toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin in the developing male

13	Wistar(Han) rat. I: No decrease in epididymal sperm count after a single acute dose. Toxicol Sci (l):214-23.

14	Bohonowych, JE; Denison, MS. (2007) Persistent binding of ligands to the aryl hydrocarbon receptor. Toxicological

15	Sciences 98(1):99-109.

16	Boverhof, DR; Burgoon, LD; Tashiro, C; et al. (2005) Temporal and dose-dependent hepatic gene expression

17	patterns in mice provide new insights into TCDD-mediated hepatotoxicity. Toxicol Sci 85(2): 1048-1063.

18	Cantoni, L; Salmona, M; Rizzardini, M. (1981) Porphyrogenic effect of chronic treatment with 2,3,7,8-

19	tetrachlorodibenzo-p-dioxin in female rats. Dose-effect relationship following urinary excretion of porphyrins.

20	Toxicol Appl Pharmacol 57:156-163.

21	Carrier, G; Brunet, RC; Brodeur, J. (1995a) Modeling of the toxicokinetics of polychlorinated dibenzo-p-dioxins

22	and dibenzofuranes in mammalians, including humans: II kinetics of absorption and disposition of PCDDs/PCDFs.

23	Toxicol Appl Pharmacol 131(267):276.

24	Carrier, G; Brunet, RC; Brodeur, J. (1995b) Modeling of the toxicokinetics of polychlorinated dibenzo-p-dioxins

25	and dibenzofurans in mammalians, including humans. Toxicol Appl Pharmacol 131:253-266.

26	Chu, I; Valli, VE; Rousseaux, CG. (2007) Combined effects of 2,3,7,8-tetrachlorodibenzo-pdioxin and

27	polychlorinated biphenyl congeners in rats. Toxicol Environ Chem 89(l):71-87.

28	Connor, KT; Aylward, LL. (2006) Human response to dioxin: aryl hydrocarbon receptor (AHR) molecular structure,

29	function, and dose-response data for enzyme induction indicate an impaired human AhR. J Toxicol Environ Health

30	Part B: Critical Reviews 9(2):147-171.

31	Crofton, KM; Craft, ES; Hedge, JM; et al. (2005) Thyroid-hormone-disrupting chemicals: evidence for dose-

32	dependent additivity or synergism. Environ Health Perspect 113(11): 1549-1554.

33	Derelanko, MJ; Hollinger, MA. (1995) CRC Handbook of Toxicology, pp.1-948. New York, NY.

34	Diliberto, JJ; Burgin, D; Birnbaum, LS. (1997) Role of CYP1A2 in hepatic sequestration of dioxin: studies using

35	CYP1A2 knock-out mice. BiochemBiophys Res Commun236(2):431-433.

36	Emond, C; Birnbaum, LS; DeVito, M. (2004) Physiologically based pharmacokinetic model for developmental

37	exposures to TCDD in the rat. Toxicol Sci 80(1): 115-133.

3 8	Emond, C; Michalek, JE; Birnbaum, LS; et al. (2005) Comparison of the use of a physiologically based

3 9	pharmacokinetic model and a classical pharmacokinetic model for dioxin exposure assessments. Environ Health

40	Perspect 113(12): 1666-1668.

This document is a draft for review purposes only and does not constitute Agency policy.

C-204 DRAFT—DO NOT CITE OR QUOTE

-------
1	Emond, C; Birnbaum, LS; Devito, MJ. (2006) Use of a physiologically based pharmacokinetic model for rats to

2	study the influence of body fat mass and induction of CYP1A2 on the pharmacokinetics of TCDD. Environ Health

3	Perspect 114(9): 1394-1400.

4	Fattore, EE; Trossvik, C; Hakansson, H. (2000) Relative potency values derived from hepatic vitamin A reduction in

5	male and female Sprague-Dawley rats following subchronic dietary exposure to individual polychlorinated dibenzo-

6	p-dioxin and dibenzofuran congeners and a mixture thereof. Toxicol Appl Pharmacol 165(3): 184-194.

7	Haddad, S; Beliveau, M; Tardif, R; Krishnan, K. (2001) A PBPK modeling-based approach to account for

8	interactions in the health risk assessment of chemical mixtures. Toxicol Sci 63:125-131.

9	Hassoun, EA; Wilt, SC; Devito, MJ; et al. (1998) Induction of oxidative stress in brain tissues of mice after

10	subchronic exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Toxicol Sci 42:23-27.

11	Hassoun, EA; Li, F; Abushaban, A; et al. (2000) The relative abilities of TCDD and its congeners to induce

12	oxidative stress in the hepatic and brain tissues of rats after subchronic exposure. Toxicology 145:103-113.

13	Heinzl, H; Mittlback, M; Edler, L. (2007) On the translation of uncertainty from toxicokinetic to toxicodynamic

14	models - the TCDD example. Chemosphere 67(9):S365-S374.

15	Hojo, R; Stern, S; Zareba, G; et al. (2002) Sexually dimorphic behavioral responses to prenatal dioxin exposure.

16	Environ Health Perspect 110:247-254.

17	Huh, C; Bloch, WE. (2003) A review of U.S. anthropometric reference data (1971-2000) with comparisons to both

18	stylized and tomographic anatomic models. Physics in Medicine and Biology 48(20):3411-3429.

19	Ikeda, M; Mitsui, T; Setani, K; et al. (2005) In utero and lactational exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin

20	in rats disrupts brain sexual differentiation. Toxicol Appl Pharmacol 205(1):98-105.

21	ILSI (International Life Sciences Institute). (1994) Physiological parameter values for PBPK models. Washington,

22	DC: Risk Science Institute.

23	Irigaray, P; Mejean, L; Laurent, F. (2005) Behaviour of dioxin in pig adipocytes. Food Chem Toxicol

24	43(3):457-460.

25	Kattainen, H; Tuukkanen, J; Simanainen, U; et al. (2001) In utero/lactational 2,3,7,8-tetrachlorodibenzo-p-dioxin

26	exposure impairs molar tooth development in rats. Toxicol Appl Pharmacol 174(3):216-224.

27	Keller, JM; Huet-Hudson, YM; Leamy, LJ. (2007) Qualitative effects of dioxin on molars vary among inbred mouse

28	strains. Arch Oral Biol 52(5):450-454.

29	Kerger, BD; Leung, HW; Scott, P; et al. (2006) Age- and concentration-dependent elimination half-life of 2,3,7,8-

30	tetrachlorodibenzo-p-dioxin in Seveso children. Environ Health Perspect 114(10):1596-1602.

31	Kerger, BD; Leung, HW; Scott, PK; et al. (2007) Refinements on the age-dependent half-life model for estimating

32	child body burdens of polychlorodibenzodioxins and dibenzofurans. Chemosphere 67(9):S272-S278.

33	Kim, AH; Kohn, MC; Nyska, A; et al. (2003) Area under the curve as a dose metric for promotional responses

34	following 2,3,7,8-tetrachlorodibenzo-p-dioxin exposure. Toxicol Appl Pharmacol 191 (1): 12—21.

35	Kitchin, KT; Woods, JS. (1979) 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) effects on hepatic microsomal

36	cytochrome P-448-mediated enzyme activities. Toxicol Appl Pharmacol 47:537-546.

37	Kociba, RJ; Keeler, PA; Park, GN; et al. (1976) 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD): results of a 13-week

38	oral toxicity study in rats. Toxicol Appl Pharmacol 35:553-574.

39	Kociba, RJ; Keyes, DG; Beyer, JE; et al. (1978) Results of a two-year chronic toxicity and oncogenicity study of

40	2,3,7,8-tetrachlorodibenzo-p-dioxin in rats. Toxicol Appl Pharmacol 46(2):279-303.

This document is a draft for review purposes only and does not constitute Agency policy.

C-205 DRAFT—DO NOT CITE OR QUOTE

-------
1	Kociba, RJ; Keyes, DG; Beyer, JE; et al. (1979) Long-term toxicological studies of 2,3,7,8-tetrachlorodibenzo-p-

2	dioxin (TCDD) in laboratory animals. Ann NY Acad Sci 320: 397-404.

3	Korenaga, T; Fukusato, T; Ohta, M; et al. (2007) Long-term effects of subcutaneously injected 2,3,7,8-

4	tetrachlorodibenzo-p-dioxin on the liver of rhesus monkeys. Chemosphere 67(9):S399-S404.

5	Korkalainen, M; Tuomisto, J; Pohjanvirta, R. (2004) Primary structure and inducibility by 2,3,7,8-

6	tetrachlorodibenzo-p-dioxin (TCDD) of aryl hydrocarbon receptor repressor in a TCDD-sensitive and a TCDD-

7	resistant rat strain. Biochem Biophys Res Commun 315(1):123—131.

8	Kransler, KM; McGarrigle, BP; Olson, JR. (2007) Comparative developmental toxicity of 2,3,7,8-

9	tetrachlorodibenzo-p-dioxin in the hamster, rat, and guinea pig. Toxicology 229(3):214-225.

10	Krishnan, D. (2007) Neurobehavioral and neuroendocrine assessment of rats perinatally exposed to polychlorinated

11	biphenyls: a possible model for autism. Bowling Green State University.

12	Krishnan, K. (2008) Physiologically based pharmacokinetic modelling in toxicology. In: Hayes, AW; ed. Principles

13	and methods of toxicology. 5th ed. New York, NY: CRC Press, pp. 231-291.

14	Latchoumycandane, C; Mathur, PP. (2002) Effects of vitamin E on reactive oxygen species-mediated 2,3,7,8-

15	tetrachlorodi-benzo-p-dioxin toxicity in rat testis. J Appl Toxicol 22(5):345-351

16	Latchoumycandane, C; Chitra, KC; Mathur, PP. (2002) The effect of 2,3,7,8-tetrachlorodibenzo-p-dioxin on the

17	antioxidant system in mitochondrial and microsomal fractions of rat testis. Toxicology 171(2-3): 127-135.

18	Leung, HW; Poland, A; Paustenbach, D; et al. (1990) Pharmacokinetics of (125-I)-2-lodo-3,7,8-trichlorodibenzo-p-

19	dioxininmice: analysis with a physiological modeling approach. Toxicol Appl Pharmacol 103:411-419.

20	Li, B; Liu, H; Dai, L; et al. (2006) The early embryo loss caused by 2,3,7,8-tetrachlorodibenzo-p-dioxin may be

21	related to the accumulation of this compound in the uterus. Reprod Toxicol 21(3):301-306.

22	Luecke, RH; Pearce, BA; Wosilait, WD; et al. (2007) Postnatal growth considerations for PBPK modeling. J

23	Toxicol Environ Health A 70(12): 1027-37.

24	Markowski, VP; Zareba, G; Stern, S; et al. (2001) Altered operant responding for motor reinforcement and the

25	determination of benchmark doses following perinatal exposure to low-level 2,3,7,8-tetrachlorodibenzo-p-dioxin.

26	Environ Health Perspect 109(6):621-627.

27	Maruyama, W; Aoki, Y. (2006) Estimated cancer risk of dioxins to humans using a bioassay and physiologically

28	based pharmacokinetic model. Toxicol Appl Pharmacol 214(2):188-198.

29	Maruyama, W; Yoshida, K; Tanaka, T; et al. (2002) Determination of tissue-blood partition coefficients for a

30	physiological model for humans, and estimation of dioxin concentration in tissues. Chemosphere 46(7):975-985.

31	Maruyama, W; Yoshida, K; Tanaka, T; et al. (2003) Simulation of dioxin accumulation in human tissues and

32	analysis of reproductive risk. Chemosphere 53(4):301—313.

33	Miettinen, HM; Sovari, R; Alaluusua, S; et al. (2006) The effect of perinatal TCDD exposure on caries susceptibility

34	in rats. Toxicol Sci 91(2):568-575.

35	Milbrath, MO; Wenger, Y; Chang, CW; et al. (2009) Apparent half-lives of dioxins, furans, and polychlorinated

36	biphenyls as a function of age, body fat, smoking status, and breast-feeding. Environ Health Perspect

37	-117(3):417425.

3 8	Moser, GA; McLachlan, MS. (2002) Modeling digestive tract absorption and desorption of lipophilic organic

39	contaminants in humans. Environ Sci Technol 36(15):3318-25.

40	Mullerova, D; Kopecky, J. (2007) White adipose tissue: storage and effector site for environmental pollutants.

41	Physiol Res 56(4):375-381.

This document is a draft for review purposes only and does not constitute Agency policy.

C-206 DRAFT—DO NOT CITE OR QUOTE

-------
1	Murray, FJ; Smith, FA; Nitschke, KD; et al. (1979) Three-generation reproduction study of rats given 2,3,7,8-

2	tetrachlorodibenzo-p-dioxin (TCDD) in the diet. Toxicol Appl Pharmacol 50:241-252.

3	Murray, TJ; Yang, X; Sherr, D. (2006) Growth of a human mammary tumor cell line is blocked by galangi, a

4	naturally occurring bioflavonoid, and is accompanied by down-regulation of cyclins D3, E, and A. Breast Cancer

5	Res 8:R17 (doi:10.1186/bcrl391)

6	Nadal, M; Perello, G; Schuhmacher, M; et al. (2008) Concentrations of PCDD/PCDFs in plasma of subjects living

7	in the vicinity of a hazardous waste incinerator: Follow-up and modeling validation. Chemosphere 73(6):901-906.

8	Nadal, M; Domingo, JL; Garcia, F; et al. (2009) Levels of PCDD/F in adipose tissue on non-occupationally exposed

9	subjects living near a hazardous waste incinerator in Catalonia, Spain. Chemosphere 74(11):1471-1476.

10	NAS (National Academy of Sciences). (2006) Health risks from dioxin and related compounds: evaluation of the

11	EPA reassessment. Washington, DC: National Academies Press. Available online at

12	http://www.nap.edu/catalog.php?record_id=l 1688.

13	Nohara, K; Fujimaki, H; Tsukumo, S; et al. (2000) The effects of perinatal exposure to low doses of 2,3,7,8-

14	tetrachlorodibenxo-p-dioxin on immune organs of rats. Toxicology 154(1—3): 123—133

15	Nohara, K; Ao, K; Miyamoto, Y; et al. (2006) Comparison of the 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)-

16	induced CYP1A1 gene expression profile in lymphocytes from mice, rats, and humans: Most potent induction in

17	humans. Toxicology 225(2-3):204-213.

18	NTP (National Toxicology Program). (1982) Bioassay of 2,3,7,8-tetrachlorodibenzo-p-dioxin for possible

19	carcinogenicity (gavage study). Tech. Rept. Ser. No. 201. U.S. Department of Health and Human Services, Public

20	Health Service, Research Triangle Park, NC.

21	NTP (National Toxicology Program). (2006a) NTP technical report on the toxicology and carcinogenesis studies of

22	2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) (CAS No. 1746-01-6) in female Harlan Sprague-Dawley rats (Gavage

23	Studies). Natl Toxicol ProgramTech Rep 521. Public Health Service, National Institute of Health, U.S. Department

24	of Health and Human Services, Research Triangle Park, NC.

25	O'Flaherty, EJ. (1992) Modeling bone mineral metabolism, with special reference to calcium and lead.

26	Neurotoxicity 13(4):789-797.

27	Ohsako, S; Miyabara, Y; Nishimura, N; et al. (2001) Maternal exposure to a low dose of 2,3,7,8-tetrachlorodibenzo-

28	p-dioxin (TCDD) suppressed the development of reproductive organs of male rats: dose-dependent increase of

29	mRNA levels of 5alpha-reductase type 2 in contrast to decrease of androgen receptor in the pubertal ventral prostate.

30	Toxicol Sci 60(l):132-43.

31	Olsman, H; Engwall, M; Kammann, U; et al. (2007) Relative differences in aryl hydrocarbon receptor-mediated

32	response for 18 polybrominated and mixed halogenated dibenzo-p-dioxins and -furans in cell lines from four

33	different species. Environ Toxicol Chem 26(11):2448-2454.

34	Pelekis, M; Gephart, LA; Lerman, SE. (2001) Physiological-model-based derivation of the adult and child
3 5	pharmacokinetic intraspecies uncertainty factors for volatile organic compounds. Reg Toxicol Pharmacol

36	33(1): 12-20.

37	Poulin, P; Theil, FP. (2000) A priori prediction of tissue:plasma partition coefficients of drugs to facilitate the use of
3 8	physiologically-based pharmacokinetic models in drug discovery. J Pharm Sci 89:16-35.

39	Saghir, SA; Lebofsky, M; Pinson, DM; et al. (2005) Validation of Haber's Rule (dose X time = constant) in rats and

40	mice for monochloroacetic acid and 2,3,7,8-tetrachlorodibenzo-p-dioxin under conditions of kinetic steady state.

41	Toxicology 215(l-2):48-56.

This document is a draft for review purposes only and does not constitute Agency policy.

C-207 DRAFT—DO NOT CITE OR QUOTE

-------
1	Santostefano, MJ; Wang, X; Richardson, VM; et al. (1998) A pharmacodynamic analysis of TCDD-induced

2	cytochrome P450 gene expression in multiple tissues: dose- and time-dependent effects. Toxicol Appl Pharmacol

3	151:294-310.

4	Schantz, SL; Seo, BW; Moshtaghian, J; et al. (1996) Effects of gestational and lactational exposure to TCDD or

5	coplanar PCBs on spatial learning. Neurotoxicol Teratol 18(3):305—313.

6	Schecter, A; Pavuk, M; Papke, O; et al. (2003) Dioxin, dibenzofuran, and coplanar PCB levels in Laotian blood and

7	milk from agent orange-sprayed and nonsprayed areas, 2001. J Toxicol Environ Health Part A: Current Issues

8	66(21):2067-2075.

9	Sewall, CH; Flagler, N; Vanden Heuvel, JP; et al. (1995) Alterations in thyroid function in female Sprague-Dawley

10	rats following chronic treatment with 2,3,7,8-tetrachlorodibenzo-p-dioxin. Toxicol Appl Pharmacol 132:237-244.

11	Shi, Z; Valdez, K; Ting, A; et al. (2007) Ovarian endocrine disruption underlies premature reproductive senescence

12	following environmentally relevant chronic exposure to the aryl hydrocarbon receptor agonist 2,3,7,8-

13	tetrachlorodibenzo-p-dioxin. Biol Reprod 76(2): 198-202.

14	Smialowicz, RJ; DeVito, MJ; Williams, WC; et al. (2008) Relative potency based on hepatic enzyme induction

15	predicts immunosuppressive effects of a mixture of PCDDS/PCDFS and PCBS. Toxicol Appl Pharmacol

16	227(3):477-484.

17	Staskal, DF; Diliberto, JJ; Devito, MJ; et al. (2005) Inhibition of human and rat CYP1A2 by TCDD and dioxin-like

18	chemicals. Toxicol Sci 84(2):225-231.

19	Toth, K; Somfai-Relle, S; Sugar, J; et al. (1979) Carcinogenicity testing of herbicide 2,4,5-trichlorophenoxyethanol

20	containing dioxin and of pure dioxin in Swiss mice. Nature 278:548-549.

21	Toyoshiba, H; Walker, NJ; Bailer, AJ; et al. (2004) Evaluation of toxic equivalency factors for induction of

22	cytochromes P450 CYP1A1 and CYP1A2 enzyme activity by dioxin-like compounds. Toxicol Appl Pharmacol

23	194(2): 156-168.

24	U.S. EPA (Environmental Protection Agency). (2003) Exposure and human health reassessment of 2,3,7,8-

25	tetrachlorodibenzo-p-dioxin (TCDD) and related compounds [NAS review draft]. Volumes 1-3. National Center

26	for Environmental Assessment, Washington, DC; EPA/600/P-00/001 Cb, Volume 1. Available online at

27	http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/.

28	Van Birgelen, AP; Van der Kolk, J; Fase, KM; et al. (1995) Subchronic dose-response study of 2,3,7,8-

29	tetrachlorodibenzo-p-dioxin in female Sprague-Dawley rats. Toxicol Appl Pharmacol 132:1-13.

30	Vanden Heuvel, JP; Clark, GC; Tritscher, A; et al. (1994) Accumulation of polychlorinated dibenzo-p-dioxins and

31	dibenzofurans in liver of control laboratory rats. Fundam Appl Toxicol 23:465-469.

32	Wang, X; Santostefano, MJ; Evans, MV; et al. (1997) Determination of parameters responsible for pharmacokinetic

33	behavior of TCDD in female Sprague-Dawley Rats. Toxicol Appl Pharmacol 147:151-168.

34	Wang, X; Santostefano, MJ; Devito, MJ; et al. (2000) Extrapolation of a PBPK model for dioxins across dosage

35	regimen, gender, strain, and species. Toxicol Sci 56(l):49-60.

36	White, KL, Jr; Lysy, HH; McCay, JA; et al. (1986) Modulation of serum complement levels following exposure to

37	polychlorinated dibenzo-p-dioxins. Toxicol Appl Pharmacol 84:209-219.

3 8	Wilkes, JG; Hass, BS; Buzatu, DA; et al. (2008) Modeling and assaying dioxin-like biological effects for both

39	dioxin-like and certain non-dioxin-like compounds. Toxicol Sci 102(1): 187—195.

This document is a draft for review purposes only and does not constitute Agency policy.

C-208 DRAFT—DO NOT CITE OR QUOTE

-------
[This page intentionally left blank.]

-------
DRAFT

DO NOT CITE OR QUOTE

May 2010
External Review Draft

APPENDIX D
Epidemiological Kinetic Modeling

NOTICE

THIS DOCUMENT IS AN EXTERNAL REVIEW DRAFT. It has not been formally released
by the U.S. Environmental Protection Agency and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its technical accuracy and policy
implications.

National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH

-------
CONTENTS—APPENDIX D: Epidemiological Kinetic Modeling

LIST OF TABLES	D-iii

APPENDIX D. EPIDEMIOLOGICAL KINETIC MODELING	D-1

D.l. BACCARELLIET AL. (2008) MODELING	D-l

D. 1.1. Input File for Exposure During Pregnancy	D-l

D.1.2. Table of Results for Baccarelli et al. (2008)	D-l

D.2. MOCARELLI ET AL. (2008) MODELING	D-2

D.2.1. Input File for Exposure for Pulse to Measurement 0.5 Years After the

Seveso Pulse Dose	D-2

D.2.2. Input File for Exposure from Pulse to the End of the Critical Window

3.8 Years After the Seveso Pulse Dose	D-2

D.2.3. Input File for Continuous Exposure for 10 Years	D-3

D.2.4. Tables of Results for Mocarelli et al. (2008)	D-4

D.3. ALALUUSUA ET AL. (2004) MODELING	D-4

D.3.1. Input File for Exposure for Pulse to Measurement 0.5 Years After the

Seveso Pulse Dose	D-4

D.3.2. Input File for Exposure from Pulse to the End of the Critical Window

2.5 Years After the Seveso Pulse Dose	D-5

D.3.3. Input File for Continuous Exposure for 5 Years	D-5

D.3.4. Tables of Results for Alaluusua et al. (2004)	D-6

D.4. ESKANAZI ET AL. (2002) MODELING	D-7

D.4.1. Input File for Exposure for Pulse to Measurement 0.5 Years After the

Seveso Pulse Dose	D-7

D.4.2. Input File for Exposure from Pulse to the End of the Critical Window

6.7 Years After the Seveso Pulse Dose	D-8

D.4.3. Input File for Continuous Exposure for 13 Years	D-8

D.4.4. Tables of Results for Eskanazi et al. (2002)	D-9

D.5. REFERENCES	D-10

This document is a draft for review purposes only and does not constitute Agency policy.

D-ii	DRAFT—DO NOT CITE OR QUOTE

-------
LIST OF TABLES

D-l. Estimated continuous intake corresponding to maternal serum concentration in

Figure 2A	D-l

D-2. Estimated maximum intake corresponding to maternal serum concentration in

Figure 2A	D-2

D-3. Matching critical window average after pulse to critical window average for

continuous intake run	D-4

D-4. Matching critical window peak after pulse to peak critical window concentration

for continuous intake run	D-4

D-5. Matching critical window average after pulse to critical window average for

continuous intake run	D-6

D-6. Matching critical window peak after pulse to peak critical window

concentration for continuous intake run	D-7

D-7. Matching critical window average after pulse to critical window average for

continuous intake run	D-9

D-8. Matching critical window peak after pulse to peak critical window

concentration for continuous intake run	D-9

This document is a draft for review purposes only and does not constitute Agency policy.

D-iii	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

APPENDIX D. EPIDEMIOLOGICAL KINETIC MODELING

D.l. BACCARELLI ET AL. (2008) MODELING

D.l.l. Input File for Exposure During Pregnancy

CINT = 1 % 168 %100	%integration time

%Exposure scenario

= 0 % delay before begin exposure (HOUR)

= 401190 %TIME EXPO SURE STOP (HOUR)
= 24 %TIME
= 401190 %DELAY BEFORE BACKGROUND EXP (HOUR)
%TIME OF BACKGROUND EXP STOP (HOUR)

EXPTIMEON
EXPTIMEOFF
DAY CYCLE
B CKTIMEON
B CKTIMEOFF =401190
IVLACK =401190
IVPERIOD =401190

%GESTATION CONTROL
MATTING = 262800 % BEGINNING MATTING (HOUR)at 30 years old
TIMELIMIT = 269184 %SIMULATION LIMIT TIME (HOUR)

TRANSTIME ON = 264312 % EXCHANGE MOTHER FETUS 1512 HOUR POST
MATTING

%Exposure dose
MSTOT	=0.021 % ng of TCDD /kg of BW

MSTOTBCKGR =0. %0.1 % ORAL BACKGROUND EXPOSURE DOSE (nG/KG)
DOSEIV = 0. %10
DOSEIVLATE = 0. %10

% TRANFER MOTHER TO FETUS CLEARANCE
CLPLAFET = 0.001 % MOTHER TO FETUS TRANFERT CLE ARAN CE(L/HR)

D.1.2. Table of Results for Baccarelli et al. (2008)

Table D-l. Estimated continuous intake corresponding to maternal serum
concentration in Figure 2A

Variable

Value

Notes

Infant b-TSH

5 uU/mL

BMR

Maternal lipid adjusted serum

270 ng/kg

From Figure 2A

Intake

0.024 ng/kg-day

From Emond model, pregnancy at 30
years

This document is a draft for review purposes only and does not constitute Agency policy.

D-1	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

Table D-2. Estimated maximum intake corresponding to maternal serum
concentration in Figure 2A

Variable

Value

Notes

Infant b-TSH

—

—

Maternal lipid adjusted serum

309.5 ng/kg

Maximum from Figure 2A

Intake

0.030 ng/kg-day

From Emond model, pregnancy at 30
years

D.2. MOCARELLI ET AL. (2008) MODELING

D.2.1. Input File for Exposure for Pulse to Measurement 0.5 Years After the Seveso Pulse
Dose

CINT = 1. %

EXPTIMEON = 54312. % Delay before begin exposure (HOUR) 6.2 years
EXPTIMEOFF = 54335. %324120 % HOUR/YEAR !TIME EXPOSURE STOP
(HOUR) 6.2 years + 23 hours
DAY CYCLE = 24. % TIME

BCK TIME ON = 0. % DELAY BEFORE BACKGROUND EXP (HOUR)

BCK TIME OFF = 613200 % TIME OF BACKGROUND EXP STOP (HOUR)

TIMELIMIT = 58692. % half a year (July 1976 until January 1977) past 6.2 years
MSTOTBCKGR = 3.7E-4 % ORAL BACKGROUND EXPOSURE DOSE (UG/KG)

% oral dose oral dose oral dose

MSTOT = 232.4 % Seveso, ORAL DAILY EXPOSURE DOSE (NG/KG)

DOSEIV = 0 % 40 %50 %5 %0.5 %0.3 %0.2 %0.1%0.05%0.3 %NG/KG
% oral dose oral dose oral dose

MEANLIPID = 731 % 711 %664 %778 %468 %671 %730 %662 %592%615%730%
PAS_INDUC= 1 % NON INDUCTION (0) CONTROLE DE L'INDUCTION

%human variable parameter
MALE = 1.

FEMALE = 0.

Y0 = 0.	% 0 years old at the beginning of the simulation

D.2.2. Input File for Exposure from Pulse to the End of the Critical Window 3.8 Years
After the Seveso Pulse Dose

CINT = 1. %

EXP TIME ON = 54312. % Delay before begin exposure (HOUR) 6.2 years
EXP TIME OFF = 54335. %324120 % HOUR/YEAR !TIME EXPOSURE STOP
(HOUR) 6.2 years + 23 hours
DAY CYCLE = 24. % TIME

This document is a draft for review purposes only and does not constitute Agency policy.

D-2	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

BCKTIMEON = 0. % DELAY BEFORE BACKGROUND EXP (HOUR)
BCKTIMEOFF = 613200. % TIME OF BACKGROUND EXP STOP (HOUR)
TIMELIMIT = 87600. % 10 years

MSTOTBCKGR = 3.7e-4 % ORAL BACKGROUND EXPOSURE DOSE (UG/KG)
% oral dose oral dose oral dose

MSTOT = 232.5 % Serveso, ORAL DAILY EXPOSURE DOSE (NG/KG)

DOSEIV = 0 % 40 %50 %5 %0.5 %0.3 %0.2 %0.1%0.05%0.3 %NG/KG
% oral dose oral dose oral dose

MEANLIPID = 730 % 711 %664 %778 %468 %671 %730 %662 %592%615%730%
PAS_INDUC= 1 % NON INDUCTION (0) CONTROLE DE L'INDUCTION

%human variable parameter
MALE = 1.

FEMALE = 0.

Y0 = 0.	% 0 years old at the beginning of the simulation

D.2.3. Input File for Continuous Exposure for 10 Years

CINT = 1. %

EXP TIME ON =0. % Delay before begin exposure (HOUR)

EXP TIME OFF = 87600. % HOUR/YEAR !TIME EXPOSURE STOP (HOUR)

DAY CYCLE = 24. % TIME

BCK TIME ON = 0. %324120 % DELAY BEFORE BACKGROUND EXP (HOUR)
BCKTIMEOFF = 613200 %324120 % TIME OF BACKGROUND EXP STOP (HOUR)
TIMELIMIT = 87600. % 10 years

MSTOTBCKGR = 0. %3.35E-4 % ORAL BACKGROUND EXPOSURE DOSE (UG/KG)
% oral dose oral dose oral dose

MSTOT = 3.903 % Seveso, ORAL DAILY EXPOSURE DOSE (NG/KG)

DOSEIV = 0 % 40 %50 %5 %0.5 %0.3 %0.2 %0.1%0.05%0.3 %NG/KG
% oral dose oral dose oral dose

MEANLIPID = 730 % 711 %664 %778 %468 %671 %730 %662 %592%615%730%
PAS_INDUC= 1 % NON INDUCTION (0) CONTROLE DE L'INDUCTION

%human variable parameter
MALE = 1.

FEMALE = 0.

Y0 = 0.	% 0 years old at the beginning of the simulation

This document is a draft for review purposes only and does not constitute Agency policy.

D-3	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

D.2.4. Tables of Results for Mocarelli et al. (2008)

Table D-3. Matching critical window average after pulse to critical window
average for continuous intake run

Person modeled,
beginning at age 0

Lipid adjusted
serum (1976) ng/kg
from Figure 3E

Pulse dose, 0.5
year lag time
(ng/kg)

Average lipid
adjusted serum
3.8 years after
incident (ng/kg)

Continuous intake
for 10 years
(ng/kg-day)

Boy, lstquartile

68

8.135

57.72

0.008024

Boy, 4th quartile

733

232.5

580.5

0.2128

Table D-4. Matching critical window peak after pulse to peak critical
window concentration for continuous intake run

Person modeled,
beginning at age 0

Lipid adjusted
serum (1976) ng/kg
from Figure 3E

Pulse dose, 0.5
year lag time
(ng/kg)

Peak lipid
adjusted serum
after incident
(ng/kg)

Continuous intake
for 10 years
(ng/kg-day)

Boy, lstquartile

68

8.135

248.0

0.03194

Boy, 4th quartile

733

232.5

6674

3.904

D.3. ALALUUSUA ET AL. (2004) MODELING

D.3.1. Input File for Exposure for Pulse to Measurement 0.5 Years After the Seveso Pulse
Dose

CINT = 1. %

EXPTIMEON =21900. % Delay before begin exposure (HOUR) 2.5 years
EXPTIMEOFF = 21923. % 21900+23 % HOUR/YEAR ! TIME EXPOSURE STOP
(HOUR) 2.5 years and 23 hours
DAY CYCLE = 24. % TIME

BCK TIME ON = 0. % DELAY BEFORE BACKGROUND EXP (HOUR)

BCK TIME OFF = 613200. % TIME OF BACKGROUND EXP STOP (HOUR)

TIMELIMIT = 26280. % half a year (July 1976 until January 1977) past 2.5 years
MSTOTBCKGR = 3.7e-4 % ORAL BACKGROUND EXPOSURE DOSE (UG/KG)

% oral dose oral dose oral dose

MSTOT = 24.22 % Seveso, ORAL DAILY EXPOSURE DOSE (NG/KG)

DOSEIV = 0 % 40 %50 %5 %0.5 %0.3 %0.2 %0.1%0.05%0.3 %NG/KG
% oral dose oral dose oral dose

This document is a draft for review purposes only and does not constitute Agency policy.

D-4	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

MEANLIPID = 730 % 711 %664 %778 %468 %671 %730 %662 %592%615%730%
PAS_INDUC= 1 % NON INDUCTION (0) CONTROLE DE L'INDUCTION

%human variable parameter
MALE = 1.

FEMALE = 0.

Y0 = 0.	% 0 years old at the beginning of the simulation

D.3.2. Input File for Exposure from Pulse to the End of the Critical Window 2.5 Years
After the Seveso Pulse Dose

CINT = 1. %

EXPTIMEON =21900. % Delay before begin exposure (HOUR) 2.5 years
EXPTIMEOFF = 21923. % 324120 % HOUR/YEAR ! TIME EXPOSURE STOP
(HOUR) 2.5 years and 23 hours
DAY CYCLE = 24. % TIME

BCK TIME ON = 0. % 324120 % DELAY BEFORE BACKGROUND EXP (HOUR)
BCKTIMEOFF = 613200. % 324120 % TIME OF BACKGROUND EXP STOP (HOUR)
TIMELIMIT =43800. % 5 years

MSTOTBCKGR = 3.7e-4 % ORAL BACKGROUND EXPOSURE DOSE (UG/KG)

% oral dose oral dose oral dose

MSTOT = 24.22 % Seveso, ORAL DAILY EXPOSURE DOSE (NG/KG)

DOSEIV = 0 % 40 %50 %5 %0.5 %0.3 %0.2 %0.1%0.05%0.3 %NG/KG
% oral dose oral dose oral dose

MEANLIPID = 730 % 711 %664 %778 %468 %671 %730 %662 %592%615%730%
PAS_INDUC= 1 % NON INDUCTION (0) CONTROLE DE L'INDUCTION

%human variable parameter
MALE = 1.

FEMALE = 0.

Y0 = 0.	% 0 years old at the beginning of the simulation

D.3.3. Input File for Continuous Exposure for 5 Years

CINT = 1. %

EXP TIME ON =0. % Delay before begin exposure (HOUR)

EXP TIME OFF = 43800. % 324120 % HOUR/YEAR !TIME EXPOSURE STOP (HOUR)
DAY CYCLE = 24. % TIME

BCK TIME ON = 0. % 324120 % DELAY BEFORE BACKGROUND EXP (HOUR)
BCK TIME OFF = 613200. % 324120 % TIME OF BACKGROUND EXP STOP (HOUR)
TIMELIMIT =43800. % End of critical window (5 years)

MSTOTBCKGR = 0. % ORAL BACKGROUND EXPOSURE DOSE (UG/KG)

% oral dose oral dose oral dose

MSTOT = 0.03486 % Seveso, ORAL DAILY EXPOSURE DOSE (NG/KG)

This document is a draft for review purposes only and does not constitute Agency policy.

D-5	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

DOSEIV = 0 % 40 %50 %5 %0.5 %0.3 %0.2 %0.1%0.05%0.3 %NG/KG
% oral dose oral dose oral dose

MEANLIPID = 730 % 711 %664 %778 %468 %671 %730 %662 %592%615%730%
PAS_INDUC= 1 % NON INDUCTION (0) CONTROLE DE L'INDUCTION

%human variable parameter
MALE = 1.

FEMALE = 0.

Y0 = 0.	% 0 years old at the beginning of the simulation

D.3.4. Tables of Results for Alaluusua et al. (2004)

Table D-5. Matching critical window average after pulse to critical window
average for continuous intake run

Person modeled,
beginning at age 0

Lipid adjusted
serum (1976) ng/kg
estimated from
tertile bins3

Pulse dose,
0.5 year lag
time (ng/kg)

Average lipid
adjusted serum 2.5
years after
incident (ng/kg)

Continuous
intake for 5
years (ng/kg-
day)

Boy, 1st tertile

130

24.22

110.8

0.03486

Boy, 2nd tertile

383

108.9

322.7

0.1578

Boy, 3rd tertile

1830

1041

1538

1.511

Girl, 1st tertile

130

23.03

110.8

0.03211

Girl, 2nd tertile

383

105.3

324.4

0.1481

Girl, 3rd tertile

1830

1015

1546

1.427

Boy and girl, averaged,
1st tertile

130

-

-

0.03349

Boy and girl, averaged,
2nd tertile

383

-

-

0.1530

Boy and girl, averaged,
3rd tertile

1830

-

-

1.469

aMean of tertile bin assuming a lognormal distribution of serum concentrations.

This document is a draft for review purposes only and does not constitute Agency policy.

D-6	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

Table D-6. Matching critical window peak after pulse to peak critical
window concentration for continuous intake run

Person modeled,
beginning at age 0

Lipid adjusted
serum (1976) ng/kg
estimated from
tertile bins

Pulse dose,
0.5 year lag
time (ng/kg)

Peak lipid
adjusted serum
after incident
(ng/kg)

Continuous
intake for 5
years (ng/kg-
day)

Boy, 1st tertile

130

24.22

618.8

0.2113

Boy, 2nd tertile

383

108.9

2700

1.783

Boy, 3rd tertile

1830

1041

24706

31.35

Girl, 1st tertile

130

23.02

588.0

0.1882

Girl, 2nd tertile

383

105.3

2610

1.642

Girl, 3rd tertile

1830

1015

24113

29.52

Boy and girl, averaged,
1st tertile

130

-

-

0.1998

Boy and girl, averaged,
2nd tertile

383

-

-

1.713

Boy and girl, averaged,
3rd tertile

1830

-

-

30.44

aMean of tertile bin assuming a lognormal distribution of serum concentrations.

D.4. ESKANAZI ET AL. (2002) MODELING

D.4.1. Input File for Exposure for Pulse to Measurement 0.5 Years After the Seveso Pulse
Dose

CINT = 1. %

EXPTIMEON = 58692. % Delay before begin exposure (HOUR) 6.7 years
EXPTIMEOFF = 58715. % HOUR/YEAR !TIME EXPOSURE STOP (HOUR) 6.7 years +
23 hours

DAY CYCLE = 24. % TIME

BCK TIME ON = 0. %324120 % DELAY BEFORE BACKGROUND EXP (HOUR)
BCK TIME OFF = 613200. %324120 % TIME OF BACKGROUND EXP STOP (HOUR)
TIMELIMIT = 63072. % half a year (July 1976 until January 1977) past 6.7 years
MSTOTBCKGR = 3.7e-4 % ORAL BACKGROUND EXPOSURE DOSE (UG/KG)

% oral dose oral dose oral dose

MSTOT = 7193 % Seveso, ORAL DAILY EXPOSURE DOSE (NG/KG)

DOSEIV = 0 % 40 %50 %5 %0.5 %0.3 %0.2 %0.1%0.05%0.3 %NG/KG
% oral dose oral dose oral dose

MEANLIPID = 730 % 711 %664 %778 %468 %671 %730 %662 %592%615%730%

This document is a draft for review purposes only and does not constitute Agency policy.

D-7	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

PAS_INDUC= 1 % NON INDUCTION (0) CONTROLE DE L'INDUCTION

%human variable parameter
MALE = 0.

FEMALE = 1.

Y0 = 0.	% 0 years old at the beginning of the simulation

D.4.2. Input File for Exposure from Pulse to the End of the Critical Window 6.7 Years
After the Seveso Pulse Dose

CINT = 1. %

EXPTIMEON = 58692. % Delay before begin exposure (HOUR) 6.7 years
EXPTIMEOFF = 58715. %324120 % HOUR/YEAR !TIME EXPOSURE STOP
(HOUR) 6.7 years + 23 hours
DAY CYCLE = 24. % TIME

BCKTIMEON = 0. %324120 % DELAY BEFORE BACKGROUND EXP (HOUR)
BCKTIMEOFF = 613200 %324120 % TIME OF BACKGROUND EXP STOP (HOUR)
TIMELIMIT = 113880. % 13 years

MSTOTBCKGR = 3.7e-4 % ORAL BACKGROUND EXPOSURE DOSE (UG/KG)

% oral dose oral dose oral dose

MSTOT =7193 % Seveso, ORAL DAILY EXPOSURE DOSE (NG/KG)

DOSEIV = 0 % 40 %50 %5 %0.5 %0.3 %0.2 %0.1%0.05%0.3 %NG/KG
% oral dose oral dose oral dose

MEANLIPID = 730 % 711 %664 %778 %468 %671 %730 %662 %592%615%730%
PAS_INDUC= 1 % NON INDUCTION (0) CONTROLE DE L'INDUCTION

%human variable parameter
MALE = 0.

FEMALE = 1.

Y0 = 0.	% 0 years old at the beginning of the simulation

D.4.3. Input File for Continuous Exposure for 13 Years

CINT = 1. %

EXP TIME ON =0. % Delay before begin exposure (HOUR)

EXP TIME OFF = 113880. %324120 % HOUR/YEAR !TIME EXPOSURE STOP

(HOUR) 13 years

DAY CYCLE = 24. % TIME

BCK TIME ON = 0. %324120 % DELAY BEFORE BACKGROUND EXP (HOUR)
BCK TIME OFF = 613200. %324120 % TIME OF BACKGROUND EXP STOP (HOUR)
TIMELIMIT = 113880. % 13 years

MSTOTBCKGR = 0. %3.35E-4 % ORAL BACKGROUND EXPOSURE DOSE (UG/KG)
% oral dose oral dose oral dose

MSTOT = 166 % Seveso, ORAL DAILY EXPOSURE DOSE (NG/KG)

This document is a draft for review purposes only and does not constitute Agency policy.

D-8	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

DOSEIV = 0 % 40 %50 %5 %0.5 %0.3 %0.2 %0.1%0.05%0.3 %NG/KG
% oral dose oral dose oral dose

MEANLIPID = 730 % 711 %664 %778 %468 %671 %730 %662 %592%615%730%
PAS_INDUC= 1 % NON INDUCTION (0) CONTROLE DE L'INDUCTION

%human variable parameter
MALE = 0.

FEMALE = 1.

Y0 = 0.	% 0 years old at the beginning of the simulation

D.4.4. Tables of Results for Eskanazi et al. (2002)

Table D-7. Matching critical window average after pulse to critical window
average for continuous intake run

Person modeled,
beginning at age 0

Lipid adjusted serum
(adjusted to 1976-
1977 levels) ng/kg
from Figure 1A

Pulse dose,
0.5 year lag
time (ng/kg)

Average lipid
adjusted serum 6.7
years after incident
(ng/kg)

Continuous
intake for 13
years (ng/kg-day)

Girl, estrous cycle
28.5 days

166

28.40

114.0

0.01660

Girl, estrous cycle
29 days

693

215.5

455.1

0.1224

Girl, estrous cycle
29.5 days

2020

1008

1295

0.5693

Girl, estrous cycle
30 days

8450

7193

5179

4.054

Table D-8. Matching critical window peak after pulse to peak critical
window concentration for continuous intake run

Person modeled,
beginning at age 0

Lipid adjusted serum
(adjusted to 1976-
1977 levels) ng/kg
from Figure 1A

Pulse dose,
0.5 year lag
time (ng/kg)

Peak lipid
adjusted serum
after incident
(ng/kg)

Continuous intake
for 13 years
(ng/kg-day)

Girl, estrous cycle
28.5 days

166

28.40

838.2

0.1800

Girl, estrous cycle
29 days

693

215.5

6183

3.148

Girl, estrous cycle
29.5 days

2020

1008

28316

20.86

Girl, estrous cycle
30 days

8450

7193

198240

166.6

This document is a draft for review purposes only and does not constitute Agency policy.

D-9	DRAFT—DO NOT CITE OR QUOTE

-------
1

2	D.5. REFERENCES

3	Alaluusua, S; Calderara, P; Gerthoux, PM; et al. (2004) Developmental dental aberrations after the dioxin accident

4	inSeveso. Environ Health Perspect 112(13): 1313-1318.

5	Baccarelli, A; Giacomini, SM; Corbetta, C; et al. (2008) Neonatal thyroid function in Seveso 25 years after maternal

6	exposure to dioxin. PLoS Med 5(7):el61.

7	Eskenazi, B; Mocarelli, P; Warner, M; et al. (2002). Serum dioxin concentrations and endometriosis: a cohort study

8	in Seveso, Italy. Environ Health Perspect 110(7): 629-634.

9	Mocarelli, P; Gerthoux, PM; Patterson, DG, Jr.; et al. (2008) Dioxin exposure, from infancy through puberty,

10	produces endocrine disruption and affects human semen quality. Environ Health Perspect 116(l):70-77.

11

This document is a draft for review purposes only and does not constitute Agency policy.

D-10 DRAFT—DO NOT CITE OR QUOTE

-------
[This page intentionally left blank.]

-------
DRAFT

DO NOT CITE OR QUOTE

May 2010
External Review Draft

APPENDIX E
Noncancer Benchmark Dose Modeling

NOTICE

THIS DOCUMENT IS AN EXTERNAL REVIEW DRAFT. It has not been formally released
by the U.S. Environmental Protection Agency and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its technical accuracy and policy
implications.

National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH

-------
CONTENTS—APPENDIX E: Noncancer Benchmark Dose Modeling

APPENDIX E. NONCANCER BENCHMARK DOSE MODELING	E-1

E.l. BMDS INPUT TABLES	E-l

E.l.l. Amin et al. (2000)	E-l

E.l.2. Bell et al. (2007)	E-l

E.l.3. Cantoni et al. (1981)	E-2

E.l.4. Crofton et al. (2005)	E-2

E.l.5. DeCaprio et al. (1986)	E-3

E.l.6. Franc et al. (2001)	E-4

E.l.7. Hojoetal. (2002)	E-4

E.l.8. Kattainen et al. (2001)	E-5

E. 1.9. Keller et al. (2007, 2008a, b)	E-5

E.l.10. Kocibaetal. (1978)	E-6

E. 1.11. Latchoumycandane and Mathur (2002)	E-6

E.l. 12. Li etal. (1997)	E-7

E.l.13. Li et al. (2006)	E-7

E.l. 14. Markowski et al. (2001)	E-8

E.l.15. Miettinen et al. (2006)	E-8

E.l. 16. National Toxicology Program (1982)	E-9

E.l. 17. National Toxicology Program (2006)	E-10

E.l.18. Ohsako etal. (2001)	E-ll

E.l. 19. Shi et al. (2007)	E-ll

E.l.20. Smialowicz et al. (2008)	E-12

E.l.21. Toth etal. (1979)	E-12

E.l.22. Van Birgelen et al. (1995)	I>13

E.l.23. White etal. (1986)	I>13

E.2. ALTERNATE DOSE: WHOLE BLOOD BMDS RESULTS	11-14

E.2.1. Amin et al., 2000: 0.25% Saccharin Consumed, Female	E-14

E.2.1.1. Summary Table of BMDS Modeling Results	E-14

E.2.1.2. Output for Selected Model: Linear	E-14

E.2.1.3. Figure for Selected Model: Linear	E-17

E.2.1.4. Output for Additional Model Presented: Power, Unrestricted....E-l7

E.2.1.5. Figure for Additional Model Presented: Power, Unrestricted	E-20

E.2.2. Amin et al., 2000: 0.25% Saccharin Preference Ratio, Female	E-20

E.2.2.1. Summary Table of BMDS Modeling Results	E-20

E.2.2.2. Output for Selected Model: Linear	E-21

E.2.2.3. Figure for Selected Model: Linear	E-23

E.2.3. Amin et al., 2000: 0.50% Saccharin Consumed, Female	E-24

E.2.3.1. Summary Table of BMDS Modeling Results	E-24

E.2.3.2. Output for Selected Model: Linear	E-24

E.2.3.3. Figure for Selected Model: Linear	E-27

E.2.3.4. Output for Additional Model Presented: Power, Unrestricted ....E-27
E.2.3.5. Figure for Additional Model Presented: Power, Unrestricted	E-30

This document is a draft for review purposes only and does not constitute Agency policy.

E-ii DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.2.4. Amin et al., 2000: 0.50% Saccharin Preference Ratio, Female	E-31

E.2.4.1. Summary Table of BMDS Modeling Results	E-31

E.2.4.2. Output for Selected Model: Linear	E-31

E.2.4.3. Figure for Selected Model: Linear	E-34

E.2.4.4. Output for Additional Model Presented: Power, Unrestricted ....E-34

E.2.4.5. Figure for Additional Model Presented: Power, Unrestricted	E-37

E.2.5. Bell et al., 2007a: Balano-Preputial Separation, Postnatal Day 49	E-38

E.2.5.1. Summary Table of BMDS Modeling Results	E-38

E.2.5.2. Output for Selected Model: Log-Logistic	E-38

E.2.5.3. Figure for Selected Model: Log-Logistic	E-40

E. 2.5.4. Output for Additi onal Model Pre sented: Log-Logi sti c,

Unrestricted	E-41

E.2.5.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-43

E.2.6. Cantoni et al., 1981: Urinary Coproporhyrins, 3 Months	E-44

E.2.6.1. Summary Table of BMDS Modeling Results	E-44

E.2.6.2. Output for Selected Model: Exponential (M4)	E-44

E.2.6.3. Figure for Selected Model: Exponential (M4)	E-47

E.2.6.4. Output for Additional Model Presented: Power, Unrestricted ....E-47

E.2.6.5. Figure for Additional Model Presented: Power, Unrestricted	E-50

E.2.7. Cantoni et al., 1981: Urinary Porphyrins	E-51

E.2.7.1. Summary Table of BMDS Modeling Results	E-51

E.2.7.2. Output for Selected Model: Exponential (M2)	E-51

E.2.7.3. Figure for Selected Model: Exponential (M2)	E-54

E.2.8. Crofton et al., 2005: Serum, T4	E-55

E.2.8.1. Summary Table of BMDS Modeling Results	E-55

E.2.8.2. Output for Selected Model: Exponential (M4)	E-55

E.2.8.3. Figure for Selected Model: Exponential (M4)	E-58

E.2.9. Franc et al., 2001: S-D Rats, Relative Liver Weight	E-59

E.2.9.1. Summary Table of BMDS Modeling Results	E-59

E.2.9.2. Output for Selected Model: Power	E-59

E.2.9.3. Figure for Selected Model: Power	E-62

E.2.10. Franc et al., 2001: L-E Rats, Relative Liver Weight	E-63

E.2.10.1. Summary Table of BMDS Modeling Results	E-63

E.2.10.2. Output for Selected Model: Hill	E-63

E.2.10.3. Figure for Selected Model: Hill	E-66

E.2.10.4. Output for Additional Model Presented: Hill, Unrestricted	E-66

E.2.10.5. Figure for Additional Model Presented: Hill, Unrestricted	E-69

E.2.11. Franc et al., 2001: S-D Rats, Relative Thymus Weight	E-70

E.2.11.1. Summary Table of BMDS Modeling Results	E-70

E.2.11.2. Output for Selected Model: Exponential (M4)	E-70

E.2.11.3. Figure for Selected Model: Exponential (M4)	E-73

E.2.11.4. Output for Additional Model Presented: Polynomial, 3-degree..E-73

This document is a draft for review purposes only and does not constitute Agency policy.

E-iii DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.2.11.5. Figure for Additional Model Presented: Polynomial, 3-degree ..E-76

E.2.12. Franc et al., 2001: L-E Rats, Relative Thymus Weight	E-77

E.2.12.1. Summary Table of BMDS Modeling Results	E-77

E.2.12.2. Output for Selected Model: Exponential (M4)	E-77

E.2.12.3. Figure for Selected Model: Exponential (M4)	E-80

E.2.13. Franc et al., 2001: Ft/W Rats, Relative Thymus Weight	E-81

E.2.13.1. Summary Table of BMDS Modeling Results	E-81

E.2.13.2. Output for Selected Model: Exponential (M2)	E-81

E.2.13.3. Figure for Selected Model: Exponential (M2)	E-84

E.2.14. Hojo et al., 2002: DRL Reinforce Per Minute	E-85

E.2.14.1. Summary Table of BMDS Modeling Results	E-85

E.2.14.2. Output for Selected Model: Exponential (M4)	E-85

E.2.14.3. Figure for Selected Model: Exponential (M4)	E-88

E.2.15. Hojo et al., 2002: DRL Response Per Minute	E-89

E.2.15.1. Summary Table of BMDS Modeling Results	E-89

E.2.15.2. Output for Selected Model: Exponential (M4)	E-89

E.2.15.3. Figure for Selected Model: Exponential (M4)	E-92

E.2.16. Kattainen et al., 2001: 3rd Molar Eruption, Female	E-93

E.2.16.1. Summary Table of BMDS Modeling Results	E-93

E.2.16.2. Output for Selected Model: Log-Logistic	E-93

E.2.16.3. Figure for Selected Model: Log-Logistic	E-95

E.2.16.4. Output for Additional Model Presented: Log-Logistic,

Unrestricted	E-95

E.2.16.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-97

E.2.17. Kattainen et al., 2001: 3rd Molar Length, Female	E-98

E.2.17.1. Summary Table of BMDS Modeling Results	E-98

E.2.17.2. Output for Selected Model: Hill	E-98

E.2.17.3. Figure for Selected Model: Hill	E-101

E.2.17.4. Output for Additional Model Presented: Hill, Unrestricted	E-101

E.2.17.5. Figure for Additional Model Presented: Hill, Unrestricted	E-104

E.2.18. Keller etal., 2007: Missing Mandibular Molars, CBA J	E-105

E.2.18.1. Summary Table of BMDS Modeling Results	E-105

E.2.18.2. Output for Selected Model: Multistage, 1-Degree	E-105

E.2.18.3. Figure for Selected Model: Multistage, 1-Degree	E-107

E.2.19. Kociba et al., 1978: Urinary Coproporphyrin, Females	E-108

E.2.19.1. Summary Table of BMDS Modeling Results	E-108

E.2.19.2. Output for Selected Model: Exponential (M4)	E-108

E.2.19.3. Figure for Selected Model: Exponential (M4)	E-l 11

E.2.20. Kociba et al., 1978: Uroporphyrin per Creatinine, Female	E-l 12

E.2.20.1. Summary Table of BMDS Modeling Results	E-l 12

E.2.20.2. Output for Selected Model: Linear	E-l 12

E.2.20.3. Figure for Selected Model: Linear	E-l 15

This document is a draft for review purposes only and does not constitute Agency policy.

E-iv DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.2.21. Latchoumycandane and Mathur, 2002: Sperm Production	E-l 16

E.2.21.1. Summary Table of BMDS Modeling Results	E-l 16

E.2.21.2. Output for Selected Model: Hill	E-l 16

E.2.21.3. Figure for Selected Model: Hill	E-l 19

E.2.21.4. Output for Additional Model Presented: Hill, Unrestricted	E-l 19

E.2.21.5. Figure for Additional Model Presented: Hill, Unrestricted	E-122

E.2.22. Li et aL 1997: FSH	E-123

E.2.22.1. Summary Table of BMDS Modeling Results	E-123

E.2.22.2. Output for Selected Model: Power	E-123

E.2.22.3. Figure for Selected Model: Power	E-126

E.2.22.4. Output for Additional Model Presented: Power, Unrestricted ..E-126
E.2.22.5. Figure for Additional Model Presented: Power, Unrestricted...E-l29

E.2.23. Li et al., 2006: Estradiol, 3-Day	E-130

E.2.23.1. Summary Table of BMDS Modeling Results	E-130

E.2.23.2. Output for Selected Model: Linear	E-130

E.2.23.3. Figure for Selected Model: Linear	E-133

E.2.24. Li et al., 2006: Progesterone, 3-Day	E-134

E.2.24.1. Summary Table of BMDS Modeling Results	E-134

E.2.24.2. Output for Selected Model: Hill	E-134

E.2.24.3. Figure for Selected Model: Hill	E-137

E.2.25. Markowski et al., 2001: FR10 Run Opportunities	E-138

E.2.25.1. Summary Table of BMDS Modeling Results	E-138

E.2.25.2. Output for Selected Model: Exponential (M2)	E-138

E.2.25.3. Figure for Selected Model: Exponential (M2)	E-141

E.2.26. Markowski et al., 2001: FR2 Revolutions	E-142

E.2.26.1. Summary Table of BMDS Modeling Results	E-142

E.2.26.2. Output for Selected Model: Hill	E-142

E.2.26.3. Figure for Selected Model: Hill	E-145

E.2.26.4. Output for Additional Model Presented: Power, Unrestricted ..E-145
E.2.26.5. Figure for Additional Model Presented: Power, Unrestricted...E-l48

E.2.27. Markowski et al., 2001: FR5 Run Opportunities	E-149

E.2.27.1. Summary Table of BMDS Modeling Results	E-149

E.2.27.2. Output for Selected Model: Hill	E-149

E.2.27.3. Figure for Selected Model: Hill	E-152

E.2.27.4. Output for Additional Model Presented: Power, Unrestricted ..E-152
E.2.27.5. Figure for Additional Model Presented: Power, Unrestricted...E-155

E.2.28. Miettinen et al., 2006: Cariogenic Lesions, Pups	E-156

E.2.28.1. Summary Table of BMDS Modeling Results	E-156

E.2.28.2. Output for Selected Model: Log-Logistic	E-156

E.2.28.3. Figure for Selected Model: Log-Logistic	E-158

E.2.28.4. Output for Additional Model Presented: Log-Logistic,

Unrestricted	E-158

This document is a draft for review purposes only and does not constitute Agency policy.

E-v DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.2.28.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-160

E.2.29. Murray et al., 1979: Fertility in F2 Generation	E-161

E.2.29.1. Summary Table of BMDS Modeling Results	E-161

E.2.29.2. Output for Selected Model: Multistage, 2-Degree	E-161

E.2.29.3. Figure for Selected Model: Multistage, 2-Degree	E-163

E.2.30. National Toxicology Program, 1982: Toxic Hepatitis, Male Mice	E-164

E.2.30.1. Summary Table of BMDS Modeling Results	E-164

E.2.30.2. Output for Selected Model: Multistage, 3-Degree	E-164

E.2.30.3. Figure for Selected Model: Multistage, 3-Degree	E-166

E.2.31. National Toxicology Program, 2006: Alveolar Metaplasia	E-167

E.2.31.1. Summary Table of BMDS Modeling Results	E-167

E.2.31.2. Output for Selected Model: Log-Logistic	E-167

E.2.31.3. Figure for Selected Model: Log-Logistic	E-169

E.2.32. National Toxicology Program, 2006: Eosinophilic Focus, Liver	E-170

E.2.32.1. Summary Table of BMDS Modeling Results	E-170

E.2.32.2. Output for Selected Model: Probit	E-170

E.2.32.3. Figure for Selected Model: Probit	E-172

E.2.33. National Toxicology Program, 2006: Fatty Change Diffuse, Liver	E-173

E.2.33.1. Summary Table of BMDS Modeling Results	E-173

E.2.33.2. Output for Selected Model: Weibull	E-173

E.2.33.3. Figure for Selected Model: Weibull	E-175

E.2.34. National Toxicology Program, 2006: Gingival Hyperplasia, Squamous, 2

Years	E-176

E.2.34.1. Summary Table of BMDS Modeling Results	E-176

E.2.34.2. Output for Selected Model: Log-Logistic	E-176

E.2.34.3. Figure for Selected Model: Log-Logistic	E-178

E.2.34.4. Output for Additional Model Presented: Log-Logistic,

Unrestricted	E-178

E.2.34.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-180

E.2.35. National Toxicology Program, 2006: Hepatocyte Hypertrophy,

2 Years	11-181

E.2.35.1. Summary Table of BMDS Modeling Results	E-181

E.2.35.2. Output for Selected Model: Multistage, 5-Degree	E-181

E.2.35.3. Figure for Selected Model: Multistage, 5-Degree	E-183

E.2.36. National Toxicology Program, 2006: Necrosis, Liver	E-184

E.2.36.1. Summary Table of BMDS Modeling Results	E-184

E.2.36.2. Output for Selected Model: Log-Probit, Unrestricted	E-184

E.2.36.3. Figure for Selected Model: Log-Probit, Unrestricted	E-186

E.2.37. National Toxicology Program, 2006: Oval Cell Hyperplasia	E-187

E.2.37.1. Summary Table of BMDS Modeling Results	E-187

E.2.37.2. Output for Selected Model: Probit	E-187

This document is a draft for review purposes only and does not constitute Agency policy.

E-vi DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.2.37.3. Figure for Selected Model: Probit	E-189

E.2.37.4. Output for Additional Model Presented: Weibull	E-189

E.2.37.5. Figure for Additional Model Presented: Weibull	E-191

E.2.38. National Toxicology Program, 2006: Pigmentation, Liver	E-192

E.2.38.1. Summary Table of BMDS Modeling Results	E-192

E.2.38.2. Output for Selected Model: Log-Probit	E-192

E.2.38.3. Figure for Selected Model: Log-Probit	E-194

E.2.39. National Toxicology Program, 2006: Toxic Hepatopathy	E-195

E.2.39.1. Summary Table of BMDS Modeling Results	E-195

E.2.39.2. Output for Selected Model: Multistage, 5-Degree	E-195

E.2.39.3. Figure for Selected Model: Multistage, 5-Degree	E-197

E.2.40. Ohsako et al., 2001: Ano-Genital Length, PND 120	E-198

E.2.40.1. Summary Table of BMDS Modeling Results	E-198

E.2.40.2. Output for Selected Model: Hill	E-198

E.2.40.3. Figure for Selected Model: Hill	E-201

E.2.40.4. Output for Additional Model Presented: Hill, Unrestricted	E-201

E.2.40.5. Figure for Additional Model Presented: Hill, Unrestricted	E-204

11.2.41.	Sewall et al., 1995: T4 In Serum	11-205

E.2.41.1. Summary Table of BMDS Modeling Results	E-205

E.2.41.2. Output for Selected Model: Hill	E-205

E.2.41.3. Figure for Selected Model: Hill	E-208

E.2.41.4. Output for Additional Model Presented: Hill, Unrestricted	E-208

E.2.41.5. Figure for Additional Model Presented: Hill, Unrestricted	E-211

11.2.42.	Shi et al., 2007: Estradiol 17B, PE9	11-212

E.2.42.1. Summary Table of BMDS Modeling Results	E-212

E.2.42.2. Output for Selected Model: Exponential (M4)	E-212

E.2.42.3. Figure for Selected Model: Exponential (M4)	E-215

11.2.43.	Smialowicz et al., 2008: PFC per 10 6 Cells	11-216

E.2.43.1. Summary Table of BMDS Modeling Results	E-216

E.2.43.2. Output for Selected Model: Power, Unrestricted	E-216

E.2.43.3. Figure for Selected Model: Power, Unrestricted	E-219

E.2.44. Smialowicz et al., 2008: PFC per Spleen	E-220

E.2.44.1. Summary Table of BMDS Modeling Results	E-220

E.2.44.2. Output for Selected Model: Power, Unrestricted	E-220

E.2.44.3. Figure for Selected Model: Power, Unrestricted	E-223

E.2.45. Toth et al., 1979: Amyloidosis	E-224

E.2.45.1. Summary Table of BMDS Modeling Results	E-224

E.2.45.2. Output for Selected Model: Log-Logistic	E-224

E.2.45.3. Figure for Selected Model: Log-Logistic	E-226

E.2.45.4. Output for Additional Model Presented: Log-Logistic,

Unrestricted	E-226

E.2.45.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-228

This document is a draft for review purposes only and does not constitute Agency policy.

E-vii DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.2.46. Toth et al., 1979: Skin Lesions	E-229

E.2.46.1. Summary Table of BMDS Modeling Results	E-229

E.2.46.2. Output for Selected Model: Log-Logistic	E-229

E.2.46.3. Figure for Selected Model: Log-Logistic	E-231

E.2.46.4. Output for Additional Model Presented: Log-Logistic,

Unrestricted	E-231

E.2.46.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-23 3

E.2.47. Van Birgelen et al., 1995a: Hepatic Retinol	E-234

E.2.47.1. Summary Table of BMDS Modeling Results	E-234

E.2.47.2. Output for Selected Model: Exponential (M4)	E-234

E.2.47.3. Figure for Selected Model: Exponential (M4)	E-237

E.2.47.4. Output for Additional Model Presented: Power, Unrestricted ..E-237
E.2.47.5. Figure for Additional Model Presented: Power, Unrestricted...E-240

E.2.48. Van Birgelen et al., 1995a: Hepatic Retinol Palmitate	E-241

E.2.48.1. Summary Table of BMDS Modeling Results	E-241

E.2.48.2. Output for Selected Model: Exponential (M4)	E-241

E.2.48.3. Figure for Selected Model: Exponential (M4)	E-244

E.2.48.4. Output for Additional Model Presented: Power, Unrestricted ..E-244
E.2.48.5. Figure for Additional Model Presented: Power, Unrestricted...E-247

E.2.49. White et al., 1986: CH50	1>248

E.2.49.1. Summary Table of BMDS Modeling Results	E-248

E.2.49.2. Output for Selected Model: Hill	E-248

E.2.49.3. Figure for Selected Model: Hill	E-251

E.2.49.4. Output for Additional Model Presented: Hill, Unrestricted	E-251

E.2.49.5. Figure for Additional Model Presented: Hill, Unrestricted	E-254

E.3. ADMINISTERED DOSE BMDS RESULTS	E-255

E.3.1. Amin et al., 2000: 0.25% Saccharin Consumed, Female	E-255

E.3.1.1. Summary Table of BMDS Modeling Results	E-255

E.3.1.2. Output for Selected Model: Linear	E-255

E.3.1.3. Figure for Selected Model: Linear	E-258

E.3.1.4. Output for Additional Model Presented: Power, Unrestricted..E-258
E.3.1.5. Figure for Additional Model Presented: Power, Unrestricted...E-261

E.3.2. Amin et al., 2000: 0.25% Saccharin Preference Ratio, Female	E-262

E.3.2.1. Summary Table of BMDS Modeling Results	E-262

E.3.2.2. Output for Selected Model: Linear	E-262

E.3.2.3. Figure for Selected Model: Linear	E-265

E.3.3. Amin et al., 2000: 0.50% Saccharin Consumed, Female	E-266

E.3.3.1. Summary Table of BMDS Modeling Results	E-266

E.3.3.2. Output for Selected Model: Linear	E-266

E.3.3.3. Figure for Selected Model: Linear	E-269

E.3.3.4. Output for Additional Model Presented: Power, Unrestricted..E-269
E.3.3.5. Figure for Additional Model Presented: Power, Unrestricted...E-272

This document is a draft for review purposes only and does not constitute Agency policy.

E-viii DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.3.4. Amin et al., 2000: 0.50% Saccharin Preference Ratio, Female	E-273

E.3.4.1. Summary Table of BMDS Modeling Results	E-273

E.3.4.2. Output for Selected Model: Linear	E-273

E.3.4.3. Figure for Selected Model: Linear	E-276

E.3.4.4. Output for Additional Model Presented: Power, Unrestricted..E-276
E.3.4.5. Figure for Additional Model Presented: Power, Unrestricted...E-279

E.3.5. Bell et al., 2007a: Balano-Preputial Separation, Postnatal Day 49	E-280

E.3.5.1. Summary Table of BMDS Modeling Results	E-280

E.3.5.2. E-280

E.3.5.3. Output for Selected Model: Log-Logistic	E-280

E.3.5.4. Figure for Selected Model: Log-Logistic	E-282

E. 3.5.5. Output for Additi onal Model Pre sented: Log-Logi sti c,

Unrestricted	E-283

E.3.5.6. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-285

E.3.6. Cantoni et al., 1981: Urinary Coproporhyrins, 3 Months	E-286

E.3.6.1. Summary Table of BMDS Modeling Results	E-286

E.3.6.2. Output for Selected Model: Exponential (M4)	E-286

E.3.6.3. Figure for Selected Model: Exponential (M4)	E-289

E.3.6.4. Output for Additional Model Presented: Power, Unrestricted..E-289
E.3.6.5. Figure for Additional Model Presented: Power, Unrestricted...E-292

E.3.7. Cantoni et al., 1981: Urinary Porphyrins	E-293

E.3.7.1. Summary Table of BMDS Modeling Results	E-293

E.3.7.2. Output for Selected Model: Exponential (M2)	E-293

E.3.7.3. Figure for Selected Model: Exponential (M2)	E-296

E.3.8. Crofton et al., 2005: Serum, T4	E-297

E.3.8.1. Summary Table of BMDS Modeling Results	E-297

E.3.8.2. Output for Selected Model: Exponential (M4)	E-297

E.3.8.3. Figure for Selected Model: Exponential (M4)	E-300

E.3.9. Franc et al., 2001: S-D Rats, Relative Liver Weight	E-301

E.3.9.1. Summary Table of BMDS Modeling Results	E-301

E.3.9.2. Output for Selected Model: Power	E-301

E.3.9.3. Figure for Selected Model: Power	E-304

E.3.9.4. Output for Additional Model Presented: Power, Unrestricted..E-304
E.3.9.5. Figure for Additional Model Presented: Power, Unrestricted...E-307

E.3.10. Franc et al., 2001: L-E Rats, Relative Liver Weight	E-308

E.3.10.1. Summary Table of BMDS Modeling Results	E-308

E.3.10.2. Output for Selected Model: Hill	E-308

E.3.10.3. Figure for Selected Model: Hill	E-311

E.3.10.4. Output for Additional Model Presented: Hill, Unrestricted	E-311

E.3.10.5. Figure for Additional Model Presented: Hill, Unrestricted	E-314

E.3.11. Franc et al., 2001: S-D Rats, Relative Thymus Weight	E-315

E.3.11.1. Summary Table of BMDS Modeling Results	E-315

This document is a draft for review purposes only and does not constitute Agency policy.

E-ix DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.3.11.2. Output for Selected Model: Exponential (M4)	E-315

E.3.11.3. Figure for Selected Model: Exponential (M4)	E-318

E.3.11.4. Output for Additional Model Presented: Polynomial,

3-Degree	E-318

E.3.11.5. Figure for Additional Model Presented: Polynomial,

3-Degree	E-321

E.3.12. Franc et al., 2001: L-E Rats, Relative Thymus Weight	E-322

E.3.12.1. Summary Table of BMDS Modeling Results	E-322

E.3.12.2. Output for Selected Model: Exponential (M4)	E-322

E.3.12.3. Figure for Selected Model: Exponential (M4)	E-325

E.3.13. Franc et al., 2001: Ft/W Rats, Relative Thymus Weight	E-326

E.3.13.1. Summary Table of BMDS Modeling Results	E-326

E.3.13.2. Output for Selected Model: Exponential (M2)	E-326

E.3.13.3. Figure for Selected Model: Exponential (M2)	E-329

E.3.13.4. Output for Additional Model Presented: Exponential (M4)	E-329

E.3.13.5. Figure for Additional Model Presented: Exponential (M4)	E-332

E.3.14. Hojo et al., 2002: DRL Reinforce Per Minute	E-333

E.3.14.1. Summary Table of BMDS Modeling Results	E-333

E.3.14.2. Output for Selected Model: Linear	E-333

E.3.14.3. Figure for Selected Model: Linear	E-336

E.3.14.4. Output for Additional Model Presented: Exponential (M4)	E-336

E.3.14.5. Figure for Additional Model Presented: Exponential (M4)	E-339

E.3.15. Hojo et al., 2002: DRL Response Per Minute	E-340

E.3.15.1. Summary Table of BMDS Modeling Results	E-340

E.3.15.2. Output for Selected Model: Exponential (M4)	E-340

E.3.15.3. Figure for Selected Model: Exponential (M4)	E-343

E.3.16. Kattainen et al., 2001: 3rd Molar Eruption, Female	E-344

E.3.16.1. Summary Table of BMDS Modeling Results	E-344

E.3.16.2. Output for Selected Model: Log-Logistic	E-344

E.3.16.3. Figure for Selected Model: Log-Logistic	E-346

E.3.16.4. Output for Additional Model Presented: Log-Logistic,

Unrestricted	E-346

E.3.16.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-348

E.3.17. Kattainen et al., 2001: 3rd Molar Length, Female	E-349

E.3.17.1. Summary Table of BMDS Modeling Results	E-349

E.3.17.2. Output for Selected Model: Hill	E-349

E.3.17.3. Figure for Selected Model: Hill	E-352

E.3.17.4. Output for Additional Model Presented: Hill, Unrestricted	E-352

E.3.17.5. Figure for Additional Model Presented: Hill, Unrestricted	E-355

E.3.18. Keller etal., 2007: Missing Mandibular Molars, CBA J	E-356

E.3.18.1. Summary Table of BMDS Modeling Results	E-356

E.3.18.2. Output for Selected Model: Multistage, 1-Degree	E-356

This document is a draft for review purposes only and does not constitute Agency policy.

E-x DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.3.18.3. Figure for Selected Model: Multistage, 1-Degree	E-358

E.3.19. Kociba et al., 1978: Urinary Coproporphyrin, Females	E-359

E.3.19.1. Summary Table of BMDS Modeling Results	E-359

E.3.19.2. Output for Selected Model: Exponential (M4)	E-359

E.3.19.3. Figure for Selected Model: Exponential (M4)	E-362

E.3.20. Kociba et al., 1978: Uroporphyrin per Creatinine, Female	E-363

E.3.20.1. Summary Table of BMDS Modeling Results	E-363

E.3.20.2. Output for Selected Model: Linear	E-363

E.3.20.3. Figure for Selected Model: Linear	E-366

E.3.21. Latchoumycandane and Mathur, 2002: Sperm Production	E-367

E.3.21.1. Summary Table of BMDS Modeling Results	E-367

E.3.21.2. Output for Selected Model: Hill	E-367

E.3.21.3. Figure for Selected Model: Hill	E-370

E.3.21.4. Output for Additional Model Presented: Hill, Unrestricted	E-370

E.3.21.5. Figure for Additional Model Presented: Hill, Unrestricted	E-373

E.3.22. Li et al., 1997: FSH	E-374

E.3.22.1. Summary Table of BMDS Modeling Results	E-374

E.3.22.2. Output for Selected Model: Power	E-374

E.3.22.3. Figure for Selected Model: Power	E-377

E.3.22.4. Output for Additional Model Presented: Power, Unrestricted ..E-377
E.3.22.5. Figure for Additional Model Presented: Power, Unrestricted...E-380

E.3.23. Li et al., 2006: Estradiol, 3-Day	E-381

E.3.23.1. Summary Table of BMDS Modeling Results	E-381

E.3.23.2. Output for Selected Model: Linear	E-381

E.3.23.3. Figure for Selected Model: Linear	E-384

E.3.24. Li et al., 2006: Progesterone, 3-Day	E-385

E.3.24.1. Summary Table of BMDS Modeling Results	E-385

E.3.24.2. Output for Selected Model: Exponential (M4)	E-385

E.3.24.3. Figure for Selected Model: Exponential (M4)	E-388

E.3.24.4. Output for Additional Model Presented: Hill, Unrestricted	E-388

E.3.24.5. Figure for Additional Model Presented: Hill, Unrestricted	E-391

E.3.25. Markowski et al., 2001: FR10 Run Opportunities	E-392

E.3.25.1. Summary Table of BMDS Modeling Results	E-392

E.3.25.2. Output for Selected Model: Exponential (M2)	E-392

E.3.25.3. Figure for Selected Model: Exponential (M2)	E-395

E.3.26. Markowski et al., 2001: FR2 Revolutions	E-396

E.3.26.1. Summary Table of BMDS Modeling Results	E-396

E.3.26.2. Output for Selected Model: Hill	E-396

E.3.26.3. Figure for Selected Model: Hill	E-399

E.3.26.4. Output for Additional Model Presented: Power, Unrestricted ..E-399
E.3.26.5. Figure for Additional Model Presented: Power, Unrestricted...E-402

E.3.27. Markowski et al., 2001: FR5 Run Opportunities	E-403

E.3.27.1. Summary Table of BMDS Modeling Results	E-403

This document is a draft for review purposes only and does not constitute Agency policy.

E-xi DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.3.27.2. Output for Selected Model: Hill	E-403

E.3.27.3. Figure for Selected Model: Hill	E-406

E.3.27.4. Output for Additional Model Presented: Power, Unrestricted ..E-407
E.3.27.5. Figure for Additional Model Presented: Power, Unrestricted...E-409

E.3.28. Miettinen et al., 2006: Cariogenic Lesions, Pups	E-410

E.3.28.1. Summary Table of BMDS Modeling Results	E-410

E.3.28.2. Output for Selected Model: Log-Logistic	E-410

E.3.28.3. Figure for Selected Model: Log-Logistic	E-412

E.3.28.4. Output for Additional Model Presented: Log-Logistic,

Unrestricted	E-412

E.3.28.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-414

E.3.29. Murray et al., 1979: Fertility in F2 Generation	E-415

E.3.29.1. Summary Table of BMDS Modeling Results	E-415

E.3.29.2. Output for Selected Model: Multistage, 2-Degree	E-415

E.3.29.3. Figure for Selected Model: Multistage, 2-Degree	E-417

E.3.30. National Toxicology Program, 1982: Toxic Hepatitis, Male Mice	E-418

E.3.30.1. Summary Table of BMDS Modeling Results	E-418

E.3.30.2. Output for Selected Model: Multistage, 3-Degree	E-418

E.3.30.3. Figure for Selected Model: Multistage, 3-Degree	E-420

E.3.31. National Toxicology Program, 2006: Alveolar Metaplasia	E-421

E.3.31.1. Summary Table of BMDS Modeling Results	E-421

E.3.31.2. Output for Selected Model: Log-Logistic	E-421

E.3.31.3. Figure for Selected Model: Log-Logistic	E-423

E. 3.31.4. Output for Additi onal Model Pre sented: Log-Logi sti c,

Unrestricted	E-423

E.3.31.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-425

E.3.32. National Toxicology Program, 2006: Eosinophilic Focus, Liver	E-426

E.3.32.1. Summary Table of BMDS Modeling Results	E-426

E.3.32.2. Output for Selected Model: Probit	E-426

E.3.32.3. Figure for Selected Model: Probit	E-428

E.3.33. National Toxicology Program, 2006: Fatty Change Diffuse, Liver	E-429

E.3.33.1. Summary Table of BMDS Modeling Results	E-429

E.3.33.2. Output for Selected Model: Weibull	E-429

E.3.33.3. Figure for Selected Model: Weibull	E-431

E.3.34. National Toxicology Program, 2006: Gingival Hyperplasia, Squamous, 2

Years	E-432

E.3.34.1. Summary Table of BMDS Modeling Results	E-432

E.3.34.2. Output for Selected Model: Log-Logistic	E-432

E.3.34.3. Figure for Selected Model: Log-Logistic	E-434

E.3.34.4. Output for Additional Model Presented: Log-Logistic,

Unrestricted	E-434

This document is a draft for review purposes only and does not constitute Agency policy.

E-xii DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.3.34.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-436

E.3.35. National Toxicology Program, 2006: Hepatocyte Hypertrophy,

2 Years	E-437

E.3.35.1. Summary Table of BMDS Modeling Results	E-437

E.3.35.2. Output for Selected Model: Multistage, 5-Degree	E-437

E.3.35.3. Figure for Selected Model: Multistage, 5-Degree	E-439

E.3.36. National Toxicology Program, 2006: Necrosis, Liver	E-440

E.3.36.1. Summary Table of BMDS Modeling Results	E-440

E.3.36.2. Output for Selected Model: Log-Probit, Unrestricted	E-440

E.3.36.3. Figure for Selected Model: Log-Probit, Unrestricted	E-442

E.3.37. National Toxicology Program, 2006: Oval Cell Hyperplasia	E-443

E.3.37.1. Summary Table of BMDS Modeling Results	E-443

E.3.37.2. Output for Selected Model: Probit	E-443

E.3.37.3. Figure for Selected Model: Probit	E-445

E.3.37.4. Output for Additional Model Presented: Weibull	E-445

E.3.37.5. Figure for Additional Model Presented: Weibull	E-447

E.3.38. National Toxicology Program, 2006: Pigmentation, Liver	E-448

E.3.38.1. Summary Table of BMDS Modeling Results	E-448

E.3.38.2. Output for Selected Model: Log-Probit	E-448

E.3.38.3. Figure for Selected Model: Log-Probit	E-450

E.3.39. National Toxicology Program, 2006: Toxic Hepatopathy	E-451

E.3.39.1. Summary Table of BMDS Modeling Results	E-451

E.3.39.2. Output for Selected Model: Multistage, 5-Degree	E-451

E.3.39.3. Figure for Selected Model: Multistage, 5-Degree	E-453

E.3.40. Ohsako et al., 2001: Ano-Genital Length, PND 120	E-454

E.3.40.1. Summary Table of BMDS Modeling Results	E-454

E.3.40.2. Output for Selected Model: Hill	E-454

E.3.40.3. Figure for Selected Model: Hill	E-457

E.3.40.4. Output for Additional Model Presented: Hill, Unrestricted	E-457

E.3.40.5. Figure for Additional Model Presented: Hill, Unrestricted	E-460

11.3.41.	Sewall et al., 1995: T4 In Serum	11-461

E.3.41.1. Summary Table of BMDS Modeling Results	E-461

E.3.41.2. Output for Selected Model: Hill	E-461

E.3.41.3. Figure for Selected Model: Hill	E-464

E.3.41.4. Output for Additional Model Presented: Hill, Unrestricted	E-464

E.3.41.5. Figure for Additional Model Presented: Hill, Unrestricted	E-467

11.3.42.	Shi et al., 2007: Estradiol 17B, PE9	11-468

E.3.42.1. Summary Table of BMDS Modeling Results	E-468

E.3.42.2. Output for Selected Model: Exponential (M4)	E-468

E.3.42.3. Figure for Selected Model: Exponential (M4)	E-471

11.3.43.	Smialowicz et al., 2008: PFC per 10 6 Cells	11-472

E.3.43.1. Summary Table of BMDS Modeling Results	E-472

This document is a draft for review purposes only and does not constitute Agency policy.

E-xiii DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.3.43.2. Output for Selected Model: Power, Unrestricted	E-472

E.3.43.3. Figure for Selected Model: Power, Unrestricted	E-475

E.3.43.4. Output for Additional Model Presented: Power	E-475

E.3.43.5. Figure for Additional Model Presented: Power	E-478

E.3.44. Smialowicz et al., 2008: PFC per Spleen	E-479

E.3.44.1. Summary Table of BMDS Modeling Results	E-479

E.3.44.2. Output for Selected Model: Power, Unrestricted	E-479

E.3.44.3. Figure for Selected Model: Power, Unrestricted	E-482

E.3.44.4. Output for Additional Model Presented: Power	E-482

E.3.44.5. Figure for Additional Model Presented: Power	E-485

E.3.45. Toth et al., 1979: Amyloidosis	E-486

E.3.45.1. Summary Table of BMDS Modeling Results	E-486

E.3.45.2. Output for Selected Model: Log-Logistic	E-486

E.3.45.3. Figure for Selected Model: Log-Logistic	E-488

E.3.45.4. Output for Additional Model Presented: Log-Logistic,

Unrestricted	E-488

E.3.45.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-490

E.3.46. Toth et al., 1979: Skin Lesions	E-491

E.3.46.1. Summary Table of BMDS Modeling Results	E-491

E.3.46.2. Output for Selected Model: Logistic	E-491

E.3.46.3. Figure for Selected Model: Logistic	E-493

E.3.46.4. Output for Additional Model Presented: Log-Logistic,

Unrestricted	E-493

E.3.46.5. Figure for Additional Model Presented: Log-Logistic,

Unrestricted	E-495

E.3.47. Van Birgelen et al., 1995a: Hepatic Retinol	E-496

E.3.47.1. Summary Table of BMDS Modeling Results	E-496

E.3.47.2. Output for Selected Model: Exponential (M4)	E-496

E.3.47.3. Figure for Selected Model: Exponential (M4)	E-499

E.3.47.4. Output for Additional Model Presented: Power, Unrestricted ..E-499
E.3.47.5. Figure for Additional Model Presented: Power, Unrestricted...E-502

E.3.48. Van Birgelen et al., 1995a: Hepatic Retinol Palmitate	E-503

E.3.48.1. Summary Table of BMDS Modeling Results	E-503

E.3.48.2. Output for Selected Model: Linear	E-503

E.3.48.3. Figure for Selected Model: Linear	E-506

E.3.48.4. Output for Additional Model Presented: Power, Unrestricted ..E-506
E.3.48.5. Figure for Additional Model Presented: Power, Unrestricted...E-509

E.3.49. White et al., 1986: CH50	E-510

E.3.49.1. Summary Table of BMDS Modeling Results	E-510

E.3.49.2. Output for Selected Model: Hill	E-510

E.3.49.3. Figure for Selected Model: Hill	E-513

E.3.49.4. Output for Additional Model Presented: Hill, Unrestricted	E-513

This document is a draft for review purposes only and does not constitute Agency policy.

E-xiv DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

E.3.49.5. Figure for Additional Model Presented: Hill, Unrestricted	E-516

E.4. REFERENCES	E-517

E-xv DRAFT—DO NOT CITE OR QUOTE

This document is a draft for review purposes only and does not constitute Agency policy.

-------
1	APPENDIX E. NONCANCER BENCHMARK DOSE MODELING

2

3

4	E.l. BMDS INPUT TABLES

5	E.l.l. Amin et al. (2000)

Endpointc

Administered Dose (ng/kg-day)

0

25 a

100

Internal Dose (ng/kg blood) b

0

3.38

10.57

(n = 10)

(n = 10)

(n = 10)

Saccharin consumed, female rats (0.25%)
(ml saccharin solution/100 g body weight)0

31.67 ±6.53

24.60 ±3.79

10.70 ± 1.68

Saccharin consumed, female rats (0.50%)
(ml saccharin solution/100 g body weight)0

22.40 ±5.05

11.38 ±2.42

4.54 ± 1.05

Saccharin preference ratio, female rats (0.25%) (ratio
of saccharin solution consumed to total fluid
consumed) d

82.14 ±4.22

58.12 ± 10.71

54.87 ±6.17

Saccharin preference ratio, female rats (0.50%) (ratio
of saccharin solution consumed to total fluid
consumed) d

72.73 ± 7.79

44.48 ± 10.39

33.77 ±7.79

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 Values are the mean ± SE. Data obtained from Figure 2 in Amin et al. 2000.
d Values are the ratio ± SE. Data obtained from Figure 3 in Amin et al. 2000.

6

7

8	E.1.2. Bell et al. (2007)



Administered Dose (ng/kg-day)



0

2.4 a

8

46



Internal Dose (ng/kg blood) b



0

2.20

5.14

18.41

Endpoint

(n = 30)

(n = 30)

(n = 30)

(n = 30)

Proportion of male rat pups that had not
undergone balano-preputial separation on
PND 49 0

1/30 (3%)

5/30 (17%)

6/30 (20%)

15/30 (50%)

aLOAEL identified.

b From the Emond PBPK model described in 3.3.
0 Data obtained from Figure 2 in Bell et al. 2007.

This document is a draft for review purposes only and does not constitute Agency policy.

E-l	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.3. Cantoni et al. (1981)

Endpoint

Administered Dose (ng/kg-day)

0

1.43 a

14.3

143

Internal Dose (ng/kg blood) b

0

1.85

8.84

50.05

(n = 4)

(n = 4)

(n = 3)

(n = 3)

Urinary coproporphyria in female
rats (|ig coproporphyrin methyl
ester/24 hr) at 3 months0

0.74 ±0.17

1.81 ± 0.42 d

2.73 ± 0.75 e

3.00 ± 1.30 e

Urinary porphyrins in rats
(nmol/24 hr) after 45 weeks 0

2.27 ±0.49

5.55 ± 0.85 d

7.62 ± 1.79 d

196.89 ± 63.14 e

aLOAEL identifed.

bFrom the Emond PBPK model described in 3.3.

0 Values are the mean ± SE. Data for urinary coproporphyria and urinary porphyrins obtained from Figure 1 and
Table 1, respectively, in Cantoni et al. 1981.
d Statistically significant as compared to control (p < 0.05).
e Statistically significant as compared to control (p < 0.01).

2

3

4	E.1.4. Crofton et al. (2005)



Administered Dose (ng/kg-day)



0

0.1

3

10

30 a

100"

300

1,000

3,000

10,000



Internal Dose (ng/kg blood) c



0

0.02

0.49

1.38

3.46

9.26

23.07

65.65

180.90

583.48

Endpoint

(n = 14)

(n = 6)

(n = 12)

(n = 6)

(n = 6)

(n = 6)

(n = 6)

(n = 6)

(n = 6)

(n = 4)

Serum T4 in female
rats

(% control) d

100.00 ±

96.27 ±

98.57 ±

99.76 ±

93.32 ±

70.94 ±

62.52 ±

52.68 ±

54.66 ±

49.15 ±

15.44

14.98

18.11

19.04

12.11

12.74

14.75

22.73

19.71

11.15

aNOAEL identifed.
bLOAEL identifed.

0 From the Emond PBPK model described in 3.3.

d Values are the mean ± SD. Data were obtained from a Crofton et al. supplemental file, available at
http://ehp.niehs.nih.gov/docs/2005/8195/supplemental.pdf.

This document is a draft for review purposes only and does not constitute Agency policy.

E-2	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.5. DeCaprio et al. (1986)

Endpoint

Administered Dose (ng/kg-day)

0

0.12

0.61 a

4.9 b

26

Internal Dose (ng/kg blood)c

n/a

n/a

n/a

n/a

n/a

(n = 10)

(n = 10)

(n = 11)

(n = 10)

(n = 4)

Absolute kidney weight (g),
males d

5.49 ±0.17

5.14 ± 0.12

4.71 ±0.12

4.3 ± 0.15 f

-

Absolute thymus weight (g),
males d

0.56 ±0.050

0.45 ±0.022

0.44 ±0.034

0.35 ± 0.167 g

-

Body weight (g), males e

713 ± 15

682 ± 16

651 ± 19

603 ± 20 f

433 ± 38 h

Relative brain weight, males d

0.54 ±0.015

0.56 ±0.016

0.6 ±0.016

0.65 ± 0.016 f

-

Relative liver weight,
males d

4.54 ±0.23

4.1 ±0.14

5.36 ±0.61

5.63±0.29 f

-

Relative thymus weight, males d

0.078 ±
0.006

0.066 ± 0.003

0.068 ± 0.004

0.06±0.003 f

-

Endpoint

Administered Dose (ng/kg-day)

0

0.12

0.68

4.86

31

Internal Dose (ng/kg blood)c

0

n/a

n/a

n/a

n/a

(n = 8)

(n = 10)

(n = 9)

(n = 10)

(n = 4)

Body weight (g), females e

602 ± 12

583 ± 22

570 ± 22

531± 14f

351± 49 h

Relative liver weight, females d

4.3 ±0.26

4.49 ±0.35

4.27 ±0.16

5.54 ±0.43 f

-

aNOAEL identified.
bLOAEL identified.

0 Internal dose not calculated using the Emond PBPK (guinea pigs).

dOrgan weight data in guinea pigs obtained from Table 2 of DeCaprio et al. 1986. Values are the mean ± SE.

Relative organs weights were calculated as organ weight (g) / body weight (g) X 100.

eBody weight data in guinea pigs obtained from Table 1 of DeCaprio et al. 1986. Values are the mean ± SE.

Statistically significant as compared to control (p < 0.05).

8 Statistically significant as compared to control (p < 0.01).

h Statistically significant as compared to control (p < 0.001).

This document is a draft for review purposes only and does not constitute Agency policy.

E-3	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.6. Franc et al. (2001)

Endpoint

Administered Dose (ng/kg-day)

0

10 a

30 b

100

Internal Dose (ng/kg blood)c

0

6.59

14.48

36.43

(n = 8)

(n = 8)

(n = 8)

(n = 8)

S-D rats, relative liver weightd

100.0 ±5.0

108.1 ± 6.0 e

116.8 ± 9.2 e

155.3 ± 10.9 e

L-E rats, relative liver weightd

100.0 ±3.5

106.3 ±6.3

116.8 ± 3.2 e

122.2 ± 7.0 e

S-D rats, relative thymus weightd

100.2 ±29.4

91.2 ± 17.0

51.4 ± 15.4 e

22.8 ± 10.6 e

L-E rats, relative thymus weightd

103.4 ± 19.3

95.4 ±24.9

38.7 ± 17.0 e

35.0 ± 27.6 e

HAV rats, relative thymus weightd

101.2 ± 12.7

97.5 ± 11.7.0

71.0 ± 8.5 e

49.3 ± 15.4 e

aNOAEL identified.
bLOAEL identified.

0 From the Emond PBPK model described in 3.3.

d Values are the mean ± SE. Data obtained from Figure 5 in Franc et al. 2001.
e Statistically significant as compared to control (p < 0.05).

2

3

4	E.1.7. Hojo et al. (2002)

Endpoint

Administered Dose (ng/kg-day)

0

20 a

60

180

Internal Dose (ng/kg blood) b

0

1.62

4.17

10.70

(n = 5)

(n = 5)

(n = 6)

(n = 5)

DRL reinforcements/min, rat litters0

-0.814 ±0.45

-0.364 ± 0.82

0.374 ±0.54

-0.163 ±0.44

DRL responses/min, rat litters 0

18.44 ±7.99

-0.99 ± 10.96

-4.52 ±7.19

-0.41 ± 15.23

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 DRL = differential reinforcement of low rate. Values are the mean ± SD. Data obtained from Table 5 in Hojo et
al. 2002.

5

This document is a draft for review purposes only and does not constitute Agency policy.

E-4	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.8. Kattainen et al. (2001)

Endpoint

Administered Dose (ng/kg-day)

0

30 a

100

300

1,000

Internal Dose (ng/kg blood) b

0

2.23

6.25

16.08

46.86

(n = 16)

(n = 17)

(n = 15)

(n = 12)

(n = 19)

3rd molar mesio-distal
length in female rat
offspring (molar
development) (mm)0

1.86 ±0.017

1.58 ± 0.045 e

1.6 ± 0.069 e

1.5 ± 0.064 e

1.35 ± 0.118 e

Proportion of female rat
offspring without 3rd molar
eruption on PND 35 d

1/16 (10%)

3/17 (20%)

4/15 (30%)

6/12 (50%)e

13/19 (70%)e

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 Values are the mean ± SE. Data were obtained from Figure 3 in Kattainen et al. 2001.
d Data were obtained from Figure 2 in Kattainen et al. 2001.
e Statistically significant as compared to control (p < 0.05).

2

3

4	E.1.9. Keller et al. (2007, 2008a, b)

Endpoint

Administered Dose (ng/kg-day)

0

10 a

100

1,000

Internal Dose (ng/kg blood) b

0

0.54

4.29

34.06

Frequency of missing 3rd mandibular molars in CBA J
mice 0

0/29 (0%)

2/23 (10%)

6/29 (20%)

30/30 (100%)

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.
0 Data obtained from Table 1 in Keller et al. 2007.

This document is a draft for review purposes only and does not constitute Agency policy.

E-5	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.10. Kociba et al. (1978)

Endpoint

Administered Dose (ng/kg-day)

0

la

10 b

100

Internal Dose (ng/kg blood)c

0

1.55

7.15

38.56

(n = 5)

(n = 5)

(n = 5)

(n = 5)

Urinary coproporphyrin
(Hg/48 h), female rats d

9.8 ± 1.3

8.6 ±2

16.4 ± 4.7 e

17.4 ± 4e

|ig uroporphyrin per mg creatinine,
female rats d

0.157 ±0.05

0.143 ±0.037

0.181 ±0.053

0.296 ± 0.074e

aNOAEL identified.
bLOAEL identified.

0 From the Emond PBPK model described in 3.3.

d Values are the mean ± SD. Data obtained from Table 2 in Kociba et al. 1978.
e Statistically significant as compared to control (p < 0.05).

2

3

4	E.l.ll. Latchoumycandane and Mathur (2002)



Administered Dose (ng/kg-day)



0

la

10

100



Internal Dose (ng/kg blood) b



0

0.78

4.65

27.27

Endpoint

(n = 6)

(n = 6)

(n = 6)

(n = 6)

Daily sperm production (/106) in adult
male rats (mg)0

22.19 ±2.67

15.67 ±2.65 d

13.65 ± 2.19 d

13.1 ± 3.16 d

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 Values are the mean ± SD. Data obtained from Table 1 in Latchoumycandane and Mathur 2002.
d Statistically significant as compared to control (p < 0.05).

5

6

This document is a draft for review purposes only and does not constitute Agency policy.

E-6	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.12. Li et al. (1997)

Endpoint

Administered Dose (ng/kg-day)

0

3a

10 b

30

100

300

1,000

3,000

10,000

30,000

Internal Dose (ng/kg blood) c

0

0.27

0.80

2.1

5.87

15

43.33

119.94

385.96

1171.90

(n = 10)

(n =10)

(n = 10)

(n = 10)

(n = 10)

(n = 10)

(n = 10)

(n = 10)

(n = 10)

(n = 10)

Serum FSH (ng/ml)
in female rats d

23.86
±
9.38

22.16
±
15.34

85.23
±
29.83

73.30 ±
15.34

126.14 ±
50.28

132.10 ±
36.65

116.76 ±
16.19

304.26 ±
48.58

346.88 ±
47.73

455.11 ±
90.34

aNOAEL identified.
bLOAEL identified.

0 From the Emond PBPK model described in 3.3.

d Values are the mean ± SE. Data obtained from Figure 3 in Li et al. 1997.

2

3

4	E.1.13. Li et al. (2006)

Endpoint

Administered Dose (ng/kg-day)

0

2a

50

100

Internal Dose (ng/kg blood) b

0

0.16

2.84

5.12

(n = 10)

(n = 10)

(n = 10)

(n = 10)

Serum estradiol/(pg-ml)_1 in female

mice

(l~3d)c

10.17 ±3.85

19.91 ±6.31

24.72 ± 4.60

18.09 ±5.57

Serum progesterone (ng-ml)"1 in
female mice
(l~3d)c

61.74 ±3.51

30.56 ±
12.80 d

16.93 ± 10.53

11.36 ± 13.83

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 Values are the mean ± SE. Data obtained from Figures 3 (estradiol) and 4 (progesterone) in Li et al. 2006.
d Statistically significant as compared to control (p < 0.01).

This document is a draft for review purposes only and does not constitute Agency policy.

E-7	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.14. Markowski et al. (2001)

Endpoint

Administered Dose (ng/kg-day)

0

20 a

60

180

Internal Dose (ng/kg blood) b

0

1.56

4.03

10.32

(n = 7)

(n = 4)

(n = 6)

(n = 7)

FR10 earned run opportunities, adult
female offspring0

13.29 ±8.65

11.25 ±5.56

5.75 ±3.53

7 ±6.01

FR2 total revolutions, adult female
offspring0

119.29 ±69.9

108.5 ±61

56.5 ±31.21

68.14 ±33.23

FR5 earned run opportunities, adult
female offspring0

26.14 ± 12.28

23.5 ±7.04

12.8 ±6.17

13.14 ± 7.14

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 Values are the mean ± SD. Data obtained from Table 3 in Markowski et al. 2001.

2

3

4	E.1.15. Miettinen et al. (2006)

Endpoint

Administered Dose (ng/kg-day)

0

30 a

100

300

1,000

Internal Dose (ng/kg blood) b

0

2.22

6.23

16.01

46.64

(n = 42)

II

(n = 15)

(n = 24)

II

Cariogenic lesions in rat pups 0

25/42 (60%)

23/29 (79%) d

19/25 (76%)

20/24
(83%) d

29/32
(91%) d

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 Data obtained from Table 2 in Miettinen et al. 2006.
d Statistically significant as compared to control (p < 0.05).

5

6

This document is a draft for review purposes only and does not constitute Agency policy.

E-8	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.16. National Toxicology Program (1982)



Administered Dose (ng/kg-day)



0

1.43 a

7.14

71.4



Internal Dose (ng/kg blood) b



0

0.77

2.27

11.24

Endpoint

r-
II

3,

(n = 49)

(n = 49)

(n = 50)

Numbers of male mice with toxic
hepatitis 0

1/73 (1.4%)

5/49 (10%)

3/49 (6.1%)

44/50 (88%)

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.
0 Data obtained from Table 11 inNTP 1982.

This document is a draft for review purposes only and does not constitute Agency policy.

E-9	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.17. National Toxicology Program (2006)

Endpointe

Administered Dose (ng/kg-day)

0

2.14 a

7.14

15.7

32.9

71.4

Internal Dose (ng/kg blood) b

0

2.56

5.69

9.79

16.57

29.70

(n = 10)

(n = 10)

(n = 10)

(n = 10)

(n = 10)

(n = 10)

Gingival squamous hyperplasia

1/53 (2%)

7/54
(13%)d

14/53
(26%)°

13/53
(25%)°

15/53
(28%)°

16/53
(30%)°

Liver, hepatocyte hypertrophy

0/53
(0%)

19/54
(40%)°'

19/53
(40%)°

42/53
(80%)°

41/53
(80%)°

52/53
(100%)°

Heart, cardiomyopathy

10/53
(19%)

12/54
(22%)

22/53°
(42%)

25/52°
(48%)

32/53°
(60%)

36/52°
(69%)

Liver, eosinophilic focus, multiple

3/53
(6%)

8/54
(15%)

14/53
(26%)

17/53
(32%)

22/53
(42%)

42/53
(79%)

Liver, fatty change, diffuse

0/53
(0%)

2/54
(4%)

12/53°
(23%)

17/53°
(32%)

30/53°
(57%)

48/53°
(91%)

Liver, necrosis

1/53
(2%)

4/54
(7%)

4/53
(8%)

8/5 3 d
(15%)

10/53°
(19%)

17/53°
(32%)

Liver, pigmentation

4/53
(8%)

9/54
(17%)

34/53°
(64%)

48/53°
(91%)

52/53°
(98%)

53/53°
(100%)

Liver, toxic hepatopathy

0/53
(0%)

2/54
(4%)

8/53
(15%)

30/53
(57%)

45/50
(85%)

53/53
(100%)

Oval cell hyperplasia

0/53
(0%)

4/54
(10%)d

3/53
(10%)

20/53
(40%)°

38/53
(70%)d

53/53
(100%)°

Lung, alveolar to bronchiolar
epithelial metaplasia (Alveolar
epithelium, metaplasia, bronchiolar)

2/53
(4%)

19/54 c
(35%)

33/53°
(62%)

35/52°
(67%)

45/53°
(85%)

46/52°
(89%)

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 Statistically significant as compared to control (p < 0.01).
d Statistically significant as compared to control (p < 0.05).

e Data are for female rats in 2-year gavage study. Data for all endpoints obtained from Table A5b in NTP 2006.

This document is a draft for review purposes only and does not constitute Agency policy.

E-10	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.18. Ohsako et al. (2001)

Endpoint

Administered Dose (ng/kg-day)

0

12.5 a

50 b

200

800

Internal Dose (ng/kg blood)c

0

1.04

3.47

11.36

38.42

(n = 12)

(n = 10)

(n = 10)

(n = 10)

(n = 12)

Anogenital distance (mm) in male
rat offspring, PND120 d

28.91 ±0.90

27.94 ±0.79

25.17 ± 1.02 e

26.01 ±0.90 f

23.80 ±0.45 e

aNOAEL for selected endpoint.
bLOAEL for selected endpoint.

0 From the Emond PBPK model described in 3.3.

d Values are the mean ± SE. Data obtained from Figure 7 in Ohsako et al. 2001.
e Statistically significant as compared to control (p < 0.01).

Statistically significant as compared to control (p < 0.05).

2

3

4	E.1.19. Shi et al. (2007)

Endpoint

Administered Dose (ng/kg-day)

0

0.143 a

0.714 b

7.14

28.6

Internal Dose (ng/kg blood)c

0

0.34

1.07

5.23

13.91

(n = 10)

(n = 10)

(n = 10)

(n = 10)

(n = 10)

Serum estradiol - 17(3 at proestrus
9 in female rats at 9 mo.of age
(pg/ml) d

102.86 ± 13.10

86.19 ±6.19

63.33 ± 9.29 e

48.1 ± 5.95 e

38.57 ± 7.14 e

aNOAEL identified.
bLOAEL identified.

0 From the Emond PBPK model described in 3.3.

d Values are the mean ± SE. Data obtained from Figure 4 in Shi et al. 2007.
e Statistically significant as compared to control (p < 0.05).

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 1	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.20. Smialowicz et al. (2008)

Endpoint

Administered Dose (ng/kg-day)

0

1.07 a

10.7

107

321

Internal Dose (ng/kg blood) b

0

0.44

2.46

13.40

31.65

(n = 15)

(n = 14)

(n = 15)

(n = 15)

(n = 8)

PFC per 106 cells in female
mice 0

1491 ±716

1129±171d

945±516d

677 ± 465 d

161± 117d

PFC x 104 per spleen in female
mice 0

27.8 ± 13.4

21 ± 13.6 d

17.6 ± 9.4 d

12.6 ± 8.7 d

3.0 ±3.1 d

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 Values are the mean ± SD. Data obtained from Table 4 in Smialowicz et al. 2008.
d Statistically significant as compared to control (p < 0.05).

2

3

4	E.1.21. Toth et al. (1979)

Endpoint

Administered Dose (ng/kg-day)

0

la

100

1,000

Internal Dose (ng/kg blood) b

0

0.57

14.21

91.21

(n =38)

(n = 44)

(n = 44)

(n = 43)

Number with amyloidosis plus skin
lesions in mice 0

0/38 (0%)

5/44(11%)

10/44 (23%)

17/43 (40%)

Number with skin lesions in mice 0

0/38 (0%)

5/44(11%)

13/44 (30%)

25/43 (58%)

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.
0 Data obtained from Table 2 in Toth et al. 1979.

5

6

This document is a draft for review purposes only and does not constitute Agency policy.

E-12	DRAFT—DO NOT CITE OR QUOTE

-------
l E.1.22. Van Birgelen et al. (1995)

Endpoint

Administered Dose (ng/kg-day)

0

14

26

47

320

1,024

Internal Dose (ng/kg blood) b

0

7.20

11.76

18.09

86.41

250.16

n = 8

n = 8

n = 8

n = 8

n = 8

n = 8

Hepatic retinol (mg/g liver) in
female rats0

14.9 ±3.1

8.4 ± 1.2 d

8.2 ± 0.8 d

5.1 ± 0.3 d

2.2 ± 0.3 d

0.6 ± 0.2 d

Hepatic retinol palmitate
(mg/g liver) in female rats 0

472 ± 96

94 ± 24 d

107 ± 27 d

74 ± 14 d

22 ± 8 d

3 ± 1d

Plasma FT4 (pmol/liter) in
female rats0

23.4 ± 1.1

24.5 ±2.0

22.4 ± 1.0

19.3 ±3.3

16.3 ± 1.5 d

10.3 ± 1.7 d

Plasma TT4 (nmol/liter) in
female rats0

40.9 ±2.4

41.4 ± 1.9

41.4 ±2.3

32.3 ± 2.6 d

33.6 ± 2.2 d

25.5 ± 2.7 d

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 Values are the mean ± SE. Data obtained from Table 3 in Van Birgelen et al. 1995.
d Statistically significant as compared to control (p < 0.05).

2

3

4	E.1.23. White et al. (1986)

Endpoint

Administered Dose (ng/kg-day)

0

10 a

50

100

500

1,000

2,000

Internal Dose (ng/kg blood) b

0

1.09

4.08

7.14

26.81

48.72

90.56

(n = 8)

(n = 8)

(n = 8)

(n = 8)

(n = 8)

(n = 8)

(n = 8)

CH50 (U/ml) in
female mice 0

91 ± 5

54 ± 3 d

63 ± 4d

56 ± 9 d

41 ± 6 d

32 ±6 d

17 ± 6 d

aLOAEL identified.

bFrom the Emond PBPK model described in 3.3.

0 Values are the mean ± SE. Data obtained from Table 1 in White et al. 1986.
d Statistically significant as compared to control (p < 0.05).

This document is a draft for review purposes only and does not constitute Agency policy.

E-13	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

E.2. ALTERNATE DOSE: WHOLE BLOOD BMDS RESULTS
E.2.1. Amin et al., 2000: 0.25% Saccharin Consumed, Female
E.2.1.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

xV

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

linear b

1

0.551

179.214

9.147E+00

6.094E+00



polynomial, 2-
degree

1

0.551

179.214

9.147E+00

6.094E+00



power

1

0.551

179.214

9.147E+00

6.094E+00

power bound hit (power =1)

power,
unrestricted0

0

N/A

180.858

8.367E+00

3.419E+00

unrestricted (power = 0.736)

a Non-constant variance model selected (p = 0.0005)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.1.2. Output for Selected Model: Linear

Amin et al., 2000: 0.25% Saccharin Consumed, Female

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\l\Blood\l_Amin_2000_25_SC_Linear_l.(d)

Gnuplot Plotting File: C:\l\Blood\l_Amin_2 000_25_SC_Linear_l.plt

Mon Feb 08 10:44:22 2010

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose

Signs of the polynomial coefficients are not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial
lalpha =
rho =
beta_0 =
beta 1 =

Parameter Values
5.29482
0

31. 5112
-1.97726

This document is a draft for review purposes only and does not constitute Agency policy.

E-14	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
beta_0
beta 1

lalpha
1

-0. 99
-0.029
0. 044

rho

-0. 99
1

0. 026
-0. 04

beta_0

-0.029
0. 026
1

-0. 94

beta_l

0. 044
-0. 04
-0. 94
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
beta_0
beta 1

Estimate

-2.54215
2 .40985
31.2644
-1.9414

Std. Err.

1.65048
0.541771
4 .1929
0. 436071

Lower Conf. Limit
-5.77702
1.34799
23.0464
-2.79609

Upper Conf. Limit

0.692726
3. 4717
39.4823
-1. 08672

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10
3.378 10
10.57 10

31. 7

24 . 6
10.7

31. 3

24 . 7
10.8

20.6
12
5.33

17 . 8
13. 4
4 . 91

0.0727
-0.0264
-0.0362

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Model A2:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

Mu(i) + e(ij ;

Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi

Var{e(i)}

Mu + e(i)
Sigma/N2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
-92.841935
-85.255316
-85.429148
-85.606998
-98.136607

Param's

4
6

5
4
2

AIC

193.683870
182.510632
180.858295
179.213995

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

This document is a draft for review purposes only and does not constitute Agency policy.

E-15	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test -2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

25.7626
15.1732
0.347663
0.3557

4
2
1
1

<.0001
0.0005072

0.5554
0.5509

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation

Specified effect

1

Risk Type

Estimated standard deviations from the control mean

Confidence level

0. 95

BMD

9.14709

BMDL

6.09414

This document is a draft for review purposes only and does not constitute Agency policy.

E-16	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.1.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

10:44 02/08 2010

E.2.1.4. Output for Additional Model Presented: Power, Unrestricted
Amin et al., 2000: 0.25% Saccharin Consumed, Female

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\l_Amin_2000_25_SC_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\l_Amin_2 000_25_SC_Pwr_U_l.plt

Mon Feb 08 10:44:22 2010

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 7	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	5.29482

rho =	0

control =	31.6727

slope =	-2.2195

power =	0.952715

Asymptotic Correlation Matrix of Parameter Estimates

lalpha	rho	control	slope	power

lalpha 1	-0.99	0.34	-0.17	-0.061

rho -0.99	1	-0.42	0.19	0.068

control 0.34	-0.42	1	-0.72	-0.56

slope -0.17	0.19	-0.72	1	0.97

power -0.061	0.068	-0.56	0.97	1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate
-2 .48291
2 . 38455
32 . 99
-3.91099
0 .735877

Std. Err.

2 . 08669
0. 692047
5. 40754
3.83883
0.350669

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-6.57274
1.02817
22 .3914
-11.435
0.0485775

1. 60693
3.74094
43.5886
3. 61299
1.42318

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10 31.7 33	20.6 18.7 -0.223

3.378 10 24.6 23.4	12 12.4 0.302

10.57 10 10.7 10.8	5.33 4.94 -0.08

Warning: Likelihood for fitted model	larger than the Likelihood for model A3.

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

This document is a draft for review purposes only and does not constitute Agency policy.

E-18	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

Model R:	Yi = Mu + e(i

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1	-92.841935	4	193.683870

A2	-85.255316	6	182.510632

A3	-85.429148	5	180.858295

fitted	-85.429148	5	180.858295

R	-98.136607	2	200.273213

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	25.7626	4	<.0001

Test 2	15.1732	2	0.0005072

Test 3	0.347663	1	0.5554

Test 4	-8.2423e-013	0	NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 8.3667 8

BMDL = 3.41906

This document is a draft for review purposes only and does not constitute Agency policy.

E-19	DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.1.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

2	10:44 02/08 2010

3

4

5	E.2.2. Amin et al., 2000: 0.25% Saccharin Preference Ratio, Female

6	E.2.2.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

xV

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

linear b

1

0.002

227.807

1.162E+01

5.572E+00



polynomial, 2-
degree

1

0.002

227.807

1.162E+01

5.572E+00



power

1

0.002

227.807

1.162E+01

5.572E+00

power bound liit (power = 1)

a Non-constant variance model selected (p = 0.0135)
b Best-fitting model, BMDS output presented in this appendix

7

8

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-20	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

E.2.2.2. Output for Selected Model: Linear

Amin et al., 2000: 0.25% Saccharin Preference Ratio, Female

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\l\Blood\2_Amin_2000_25_SP_Linear_l.(d)

Gnuplot Plotting File: C:\l\Blood\2_Amin_2 000_25_SP_Linear_l.plt

Mon Feb 08 10:44:49 2010

The form of the response function is:

Y[dose] = beta 0 + beta l^dose + beta 2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose

Signs of the polynomial coefficients are not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	6.34368

rho =	0

beta_0 =	7 5.4888

beta~l =	-2.24733

Asymptotic Correlation Matrix of Parameter	Estimates

lalpha rho beta 0	beta 1

lalpha 1 -1 0.22	-0.31

rho -1 1 -0.22	0.31

beta_0 0.22 -0.22 1	-0.77

beta_l -0.31 0.31 -0.77	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf. Limit	Upper Conf. Limit

lalpha 3.00523 9.2122	-15.0503	21.0608

rho 0.797764 2.21138	-3.53646	5.13199

beta_0 75.1087 6.74312	61.8924	88.3249

beta_l -2.16469 1.00825	-4.14082	-0.188553

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

This document is a draft for review purposes only and does not constitute Agency policy.

E-21	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

0 10	82.1

3.378 10	58.1

10.57 10	54.9

75.1	13.3
67.8 33.9

52.2	19.5

25.2	0.884

24.2	-1.27

21.8	0.383

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(ij

Var{e(ij)}	= Sigma^2

Model A2:	Yij	= Mu(i) + e(ij

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij = Mu(i) + e(1j)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -108.574798	4	225.149597

A2 -104.269377	6	220.538754

A3 -105.147952	5	220.295903

fitted -109.903705	4	227.807410

R -112.382522	2	228.765045

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

16.2263
8.61084
1.75715
9.51151

0.00273
0.0135
0.185
0.002042

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

This document is a draft for review purposes only and does not constitute Agency policy.

E-22	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

Risk Type	=

Confidence level =
BMD =

Estimated standard deviations from the control mean
0. 95
11.6241

BMDL =	5.57215

E.2.2.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

10:44 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-23	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

E.2.3. Amin et al., 2000: 0.50% Saccharin Consumed, Female
E.2.3.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

linear b

1

0.060

158.591

1.016E+01

6.567E+00



polynomial, 2-
degree

1

0.060

158.591

1.016E+01

6.567E+00



power

1

0.060

158.591

1.016E+01

6.567E+00

power bound hit (power =1)

power,
unrestricted0

0

N/A

157.060

6.567E+00

1.155E+00

unrestricted (power = 0.396)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.3.2. Output for Selected Model: Linear

Amin et al., 2000: 0.50% Saccharin Consumed, Female

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\l\Blood\3_Amin_2000_50_SC_Linear_l.(d)

Gnuplot Plotting File: C:\l\Blood\3_Amin_2 000_50_SC_Linear_l.plt

Mon Feb 08 10:45:20 2010

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose

Signs of the polynomial coefficients are not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial
lalpha =
rho =
beta_0 =
beta 1 =

Parameter Values

4 . 68512
0

20.0631
-1.57142

Asymptotic Correlation Matrix of Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

E-24	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

lalpha
rho
beta_0
beta 1

lalpha
1

-0. 96
0. 019
-0.0016

rho

-0. 96
1

-0.031
0. 015

beta_0

0. 019
-0.031
1

-0. 96

beta_l

-0.0016
0. 015
-0. 96
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
beta_0
beta 1

Estimate

-0.982115
2.11808
18.6171
-1.33226

Std. Err.

0.982262
0.401166
3.1782
0.322037

Lower Conf. Limit

-2.90731
1. 33181
12.3879
-1.96344

Upper Conf. Limit

0.943084
2.90435
24.8462
-0.70108

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10
3.378 10
10.57 10

22 . 4
11. 4
4 . 54

18 . 6
14 .1
4 . 54

16

7 . 66
3.33

13.5

10.1
3. 04

0. 873
-0.856
-0.00339

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -83.696404	4	175.392808

A2 -73.511830	6	159.023660

A3 -73.530233	5	157.060467

fitted -75.295363	4	158.590726

R -90.294746	2	184.589492

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

This document is a draft for review purposes only and does not constitute Agency policy.

E-25	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

Tests of Interest

Test -2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

0. 0368066

33.5658
20.3691

3.53026

4
2
1
1

0.06026

<.0001
<.0001
0.8479

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation

Specified effect

1

Risk Type

Estimated standard deviations from the control mean

Confidence level

0. 95

BMD

10.1633

BMDL

6.56742

This document is a draft for review purposes only and does not constitute Agency policy.

E-26	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.3.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

10:45 02/08 2010

E.2.3.4. Output for Additional Model Presented: Power, Unrestricted
Amin et al., 2000: 0.50% Saccharin Consumed, Female

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\3_Amin_2000_50_SC_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\3_Amin_2 000_50_SC_Pwr_U_l.plt

Mon Feb 08 10:45:20 2010

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-27	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha =	4.68512

rho =	0

control =	22.3564

slope =	-6.53901

power =	0.425213

Asymptotic Correlation Matrix of Parameter Estimates

lalpha	rho	control	slope	power

lalpha 1	-0.96	0.34	-0.31	-0.15

rho -0.96	1	-0.47	0.36	0.15

control 0.34	-0.47	1	-0.81	-0.52

slope -0.31	0.36	-0.81	1	0.92

power -0.15	0.15	-0.52	0.92	1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate
-0.708629
1.96142
22.6293
-7.10123
0.395571

Std. Err.
1.298
0.529653
4.48416
4.04394
0.168677

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-3.25267

0.923323
13.8405
-15.0272
0.0649698

1.83541

2.99953
31.4181
0. 824743
0.726173

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10
3.378 10
10.57 10

22 . 4
11. 4
4 . 54

22 . 6
11.1
4 . 58

16

7 . 66
3.33

15
7.46
3.12

-0.0577
0.105
-0.0475

Degrees of freedom for Test A3 vs fitted

0

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

This document is a draft for review purposes only and does not constitute Agency policy.

E-28	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

Model R:	Yi = Mu + e(i

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1	-83.696404	4	175.392808

A2	-73.511830	6	159.023660

A3	-73.530233	5	157.060467

fitted	-73.530233	5	157.060467

R	-90.294746	2	184.589492

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	33.5658	4	<.0001

Test 2	20.3691	2	<.0001

Test 3	0.0368066	1	0.8479

Test 4	0	0	NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 6.56719

BMDL = 1.154 7 6

This document is a draft for review purposes only and does not constitute Agency policy.

E-29	DRAFT—DO NOT CITE OR QUOTE

-------
E.2.3.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

10:45 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-30	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

E.2.4. Amin et al., 2000: 0.50% Saccharin Preference Ratio, Female
E.2.4.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

linear b

1

0.135

234.250

8.144E+00

5.105E+00



polynomial, 2-
degree

1

0.135

234.250

8.144E+00

5.105E+00



power

1

0.135

234.250

8.144E+00

5.105E+00

power bound hit (power =1)

power,
unrestricted0

0

N/A

234.020

2.598E+00

1.057E-14

unrestricted (power = 0.282)

a Constant variance model selected (p = 0.5593)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.4.2. Output for Selected Model: Linear

Amin et al., 2000: 0.50% Saccharin Preference Ratio, Female

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\l\Blood\4_Amin_2000_50_SP_LinearCV_l.(d)

Gnuplot Plotting File: C:\l\Blood\4_Amin_2 000_50_SP_LinearCV_l.plt

Mon Feb 08 10:45:50 2010

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Signs of the polynomial coefficients are not restricted
A constant variance model is fit

Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial P.

alpha =
rho =
beta_0 =
beta 1 =

rameter Values
764.602

0 Specified
65.8627
-3.34297

This document is a draft for review purposes only and does not constitute Agency policy.

E-31	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha	beta 0	beta 1

alpha 1	2.6e-008	2.1e-009

beta_0 2.6e-008	1	-0.73

beta 1 2. le-009	-0.73	1

Parameter Estimates

Variable
alpha
beta_0
beta 1

Estimate

741.255
65.8627
-3.34297

95.0% Wald Confidence Interval
Std. Err.	Lower Conf. Limit Upper Conf. Limit

191.391	366.135	1116.38

7.22524	51.7015	80.0239

1.12815	-5.55412	-1.13183

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10
3.378 10
10.57 10

12. 7
44.5
33. 8

65. 9

54 . 6
30.5

24 . 6
32 . 9
24 . 6

27 . 2
27 . 2
27 . 2

0.191
-1 . 17
0.375

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi

Var{e(i)}

Mu + e(i)
Sigma/N2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-113.009921	4	234.019841

A2	-112.428886	6	236.857773

A3	-113.009921	4	234.019841

fitted	-114.125184	3	234.250368

R	-117.976057	2	239.952114

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

This document is a draft for review purposes only and does not constitute Agency policy.

E-32	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test -2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

11.0943
1.16207
1.16207
2.23053

4
2
2
1

0.02552
0.5593
0.5593
0.1353

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation

Specified effect

1

Risk Type

Estimated standard deviations from the control mean

Confidence level

0. 95

BMD

8 .14425

BMDL

5.10523

This document is a draft for review purposes only and does not constitute Agency policy.

E-33	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.4.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

10:45 02/08 2010

E.2.4.4. Output for Additional Model Presented: Power, Unrestricted
Amin et al., 2000: 0.50% Saccharin Preference Ratio, Female

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\4_Amin_2000_50_SP_PwrCV_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\4_Amin_2 000_50_SP_PwrCV_U_l.plt

Mon Feb 08 10:45:50 2010

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-34	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho
control
slope
power

764.602

0

72.7273
-20.0402
0.281985

Specified

Asymptotic Correlation Matrix of Parameter Estimates

alpha
control

The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

-1. 2e-009

control
-1. 2e-009
1

slope
-1.2e-009
-0.51

power
-2 . 2e-010
-0.22

slope -1.2e-009	-0.51	1	0.92

power -2.2e-010	-0.22	0.92	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
control
slope
power

Estimate
688.142
72.7273
-20.0402
0.281985

Std. Err.

177.677
8.29543
15.0576
0.325861

Lower Conf. Limit

339. 9
56.4686
-49.5526
-0.35669

Upper Conf. Limit
1036.38
88 . 986
9. 47219
0. 920661

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10 72.7	72.7	24.6	26.2	4.67e-009

3.378 10 44.5	44.5	32.9	26.2	1.52e-008

10.57 10 33.8	33.8	24.6	26.2	1.77e-008

Warning: Likelihood for	fitted model	larger than	the Likelihood	for model A3.

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-3 5	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-113.009921	4	234.019841

A2	-112.428886	6	236.857773

A3	-113.009921	4	234.019841

fitted	-113.009921	4	234.019841

R	-117.976057	2	239.952114

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	11.0943	4	0.02552

Test 2	1.16207	2	0.5593

Test 3	1.16207	2	0.5593

Test 4	-2.84217e-014	0	NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 2.59831

BMDL = 1.05661e-014

This document is a draft for review purposes only and does not constitute Agency policy.

E-36	DRAFT—DO NOT CITE OR QUOTE

-------
E.2.4.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

10:45 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-3 7	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.2.5. Bell et al., 2007a: Balano-Preputial Separation, Postnatal Day 49
E.2.5.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

2

0.684

112.136

2.867E+00

1.943E+00

power bound hit (power = 1)

logistic

2

0.342

113.915

6.159E+00

4.746E+00

negative intercept (intercept =
-2.246)

log-logistica

2

0.777

111.908

2.246E+00

1.394E+00

slope bound hit (slope = 1)

log-probit

2

0.269

114.254

5.322E+00

3.512E+00

slope bound hit (slope =1)

multistage, 3-
degree

2

0.684

112.136

2.867E+00

1.943E+00

final B = 0

probit

2

0.367

113.713

5.715E+00

4.422E+00



Weibull

2

0.684

112.136

2.867E+00

1.943E+00

power bound hit (power = 1)

gamma,
unrestricted

1

0.566

113.746

1.862E+00

1.829E-01

unrestricted (power = 0.741)

log-logistic,
unrestricted b

1

0.501

113.871

1.998E+00

2.795E-01

unrestricted (slope = 0.93)

log-probit,
unrestricted

1

0.456

113.977

2.038E+00

3.250E-01

unrestricted (slope = 0.54)

Weibull,
unrestricted

1

0.551

113.771

1.914E+00

2.346E-01

unrestricted (power = 0.795)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.2.5.2. Output for Selected Model: Log-Logistic

Bell et al., 2007a: Balano-Preputial Separation, Postnatal Day 49

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\5_Bell_2007_BPS_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\Blood\5_Bell_2007_BPS_LogLogistic_l.plt

Mon Feb 08 10:46:18 2010

0

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

This document is a draft for review purposes only and does not constitute Agency policy.

E-3 8	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Slope parameter is restricted as slope >= 1

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values

background = 0.0333333
intercept =	-2.99896

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept

background	1	-0.49

intercept	-0.49	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf.	Limit Upper Conf. Limit

background 0.038005 *	*	*

intercept -3.00658 *	*	*

slope 1 *	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-53.7077	4

Fitted model -53.954 2	0.492596 2	0.7817

Reduced model -63.9797 1	20.544 3	0.0001309

AIC:	111.908

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0380 1.140 1.000	30	-0.134

2.2040 0.1326 3.977 5.000	30	0.551

5.1378 0.2329 6.988 6.000	30	-0.427

18.4110 0.4965 14.895 15.000	30	0.038

Chi/N2 = 0.50	d.f. = 2	P-value = 0.7769

Benchmark Dose Computation
Specified effect =	0.1

This document is a draft for review purposes only and does not constitute Agency policy.

E-39	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

Risk Type	=

Confidence level =
BMD =

Extra risk
0. 95
2 .24647

BMDL =	1.39385

E.2.5.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

dose

10:46 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-40	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

E.2.5.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
Bell et al., 2007a: Balano-Preputial Separation, Postnatal Day 49

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\5_Bell_2007_BPS_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\5_Bell_2007_BPS_LogLogistic_U_l.plt

Mon Feb 08 10:46:18 2010

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose)

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values

background = 0.0333333
intercept =	-2.68464

slope =	0.858398

Asymptotic Correlation Matrix of Parameter Estimates
background intercept	slope

background	1	-0.48	0.35

intercept	-0.48	1	-0.94

slope	0.35	-0.94	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.0353402	*	*	*

intercept	-2.84051	*	*	*

slope	0.929645	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood) # Param's Deviance Test d.f. P-value

Full model	-53.7077	4

This document is a draft for review purposes only and does not constitute Agency policy.

E-41	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

Fitted model	-53.9354	3	0.455534	1	0.4997

Reduced model	-63.9797	1	20.544	3	0.0001309

AIC:	113.871

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.0353

1. 060

1. 000

30

-0.060

2.2040

0.1400

4 .201

5. 000

30

0.420

5.1378

0.2389

7 .166

6. 000

30

-0.499

18.4110

0.4858

14.573

15.000

30

0.156

Chi^2 = 0.45	d.f. = 1	P-value = 0.5005

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	1.99765

BMDL =	0.27 9534

This document is a draft for review purposes only and does not constitute Agency policy.

E-42	DRAFT—DO NOT CITE OR QUOTE

-------
E.2.5.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

dose

10:46 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-43	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.2.6. Cantoni et al., 1981: Urinary Coproporhyrins, 3 Months
E.2.6.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

0.003

32.882

3.209E+01

1.567E+01



exponential (M3)

2

0.003

32.882

3.209E+01

1.567E+01

power hit bound (d = 1)

exponential
(M4)b

1

0.486

23.459

5.339E-01

1.803E-01



exponential (M5)

1

0.486

23.459

5.339E-01

1.803E-01

power hit bound (d = 1)

Hill

1

0.788

23.047

4.333E-01

error

n lower bound hit (n = 1)

linear

2

0.005

31.595

1.464E+01

2.753E+00



polynomial, 3-
degree

2

0.005

31.595

1.464E+01

2.753E+00



power

2

0.005

31.595

1.464E+01

2.753E+00

power bound hit (power =1)

power,
unrestricted0

1

0.610

23.235

2.766E-02

2.031E-05

unrestricted (power = 0.304)

Hill, unrestricted

0

N/A

24.974

2.602E-01

error

unrestricted (n = 0.739)

a Non-constant variance model selected (p = 0.0039)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.6.2. Output for Selected Model: Exponential (M4)

Cantoni et al., 1981: Urinary Coproporhyrins, 3 Months

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\6_Cantoni_l981_UriCopro_Exp_l.(d)

Gnuplot Plotting File:

Mon Feb 08 10:46:46 2010

Figurel-UrinaryCoproporphyrin 3months

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;

This document is a draft for review purposes only and does not constitute Agency policy.

E-44	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	-1.50063

rho	2.6097 9

a	0.704303

b	0.0604961

c	4.47268

d	1

Parameter Estimates
Variable	Model 4

lnalpha	-1.75302

rho	2.6322

a	0.761218

b	0.241561

c	4.15597

d	1

Table of Stats From Input Data
Dose	N	Obs Mean	Obs Std Dev

0	4	0.7414	0.3475

1.847	4	1.807	0.8341

8.839	4	2.734	1.506

50.05	4	3	2.6

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	0.7612	0.2907	-0.1366

1.847	1.626	0.7892	0.4588

8.839	2.88	1.674	-0.1743

50.05	3.164	1.895	-0.1725

Other models for which likelihoods are calculated:

This document is a draft for review purposes only and does not constitute Agency policy.

E-45	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma^2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-12.90166	5	35.80333

A2	-6.203643	8	28.40729

A3	-6.487204	6	24.97441

R	-15.73713	2	35.47427

4	-6.729737	5	23.45947

Additive constant for all log-likelihoods =	-14.7. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Does response

and/or variances differ among Dose levels? (A2 vs. R)

Test

2 :

Are Variances

Homogeneous? (A2 vs. A1)

Test

3:

Are variances

adequately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

19. 07
13. 4
0.5671
0.4851

D. F.

p-value

0.004052
0.003854
0.7531
0.4861

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

This document is a draft for review purposes only and does not constitute Agency policy.

E-46	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

EMD =

0.533855

E'.MDL =	0.18 02 93

E.2.6.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

10:46 02/08 2010

E.2.6.4. Output for Additional Model Presented: Power, Unrestricted
Cantoni et al., 1981: Urinary Coproporhyrins, 3 Months

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\6_Cantoni_l981_UriCopro_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\6_Cantoni_1981_UriCopro_Pwr_U_l.plt

Mon Feb 08 10:46:47 2010

Figurel-UrinaryCoproporphyrin_3months

The form of the response function is:
Y[dose] = control + slope * doseApower

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-47	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean

Independent variable = Dose

The power is not restricted

The variance is to be modeled as Var(i)

exp(lalpha + log(mean(i))

rho)

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha =	0.90039

rho =	0

control =	0.741372

slope =	0.93685

power =	0.224 904

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope
power

lalpha
1

-0. 62
-0.53
-0.036
0. 024

rho

-0. 62
1

0.43
-0.2
-0.16

control

-0.53
0.43
1

-0.28
0. 086

slope
-0.036
-0.2

-0.28
1

-0.77

power

0. 024
-0.16
0. 086
-0.77
1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate
-1.78125
2.64332
0.75678
0. 845767
0.304211

Std. Err.

0. 617807
0.744946
0.139979
0.324854
0.135053

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-2.99213
1.18325
0. 482426
0.209065
0. 0395119

-0.570373
4 .10338
1. 03113
1.48247
0.568909

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	4	0.741	0.757	0.348	0.284	-0.109

1.847	4	1.81	1.78	0.834	0.877	0.0705

8.839	4	2.73	2.4	1.51	1.3	0.515

50.05	4	3	3.54	2.6	2.18	-0.493

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-48	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-12.901663	5	35.803325

A2	-6.203643	8	28.407287

A3	-6.487204	6	24.974409

fitted	-6.617347	5	23.234694

R	-15.737135	2	35.474269

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	19.067	6	0.004052

Test 2	13.396	3	0.003854

Test 3	0.567122	2	0.7531

Test 4	0.260285	1	0.6099

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 0.0276599

BMDL = 2.03143e-005

This document is a draft for review purposes only and does not constitute Agency policy.

E-49	DRAFT—DO NOT CITE OR QUOTE

-------
E.2.6.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

10:46 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-50	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

E.2.7. Cantoni et al., 1981: Urinary Porphyrins
E.2.7.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)b

2

<0.001

55.465

3.760E+00

2.762E+00



exponential (M3)

2

<0.001

55.465

3.760E+00

2.762E+00

power hit bound (d = 1)

exponential (M4)

1

<0.0001

59.187

2.484E-01

1.448E-01



exponential (M5)

0

N/A

61.084

2.878E-01

1.461E-01



Hill

0

N/A

62.199

6.233E+00

3.341E+00



linear

2

<0.001

57.187

2.484E-01

1.448E-01



polynomial, 3-
degree

1

<0.0001

10.000

error

error



power

1

<0.0001

59.084

2.878E-01

1.461E-01



a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix

E.2.7.2. Output for Selected Model: Exponential (M2)

Cantoni et al., 1981: Urinary Porphyrins

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\7_Cantoni_l981_UriPor_Exp_l.(d)

Gnuplot Plotting File:

Mon Feb 08 10:47:24 2010

Table 1, dose converted to ng per kg per day

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean

This document is a draft for review purposes only and does not constitute Agency policy.

E-51	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 2

lnalpha	-3.57509

rho	2.23456

a	3.36453

b	0.0819801

c	0

d	1

Parameter Estimates
Variable	Model 2

lnalpha	-1.85879

rho	1.82273

a	3.57896

b	0.0803347

c	0

d	1



Table of

Stats From Input

Data

Dose

N

Obs Mean

Obs Std

0

4

2 . 27

0.49

1.847

4

5.55

o

CO
CJ1

8 . 839

3

7 . 62

1.79

en
o

o
CJ1

3

196. 9

63.14

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	3.579	1.262	-2.074

1.847	4.152	1.445	1.936

8.839	7.28	2.41	0.2441

50.05	199.5	49.25	-0.09069

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + log(mean(i)) * rho)

This document is a draft for review purposes only and does not constitute Agency policy.

E-52	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Model R:	Yij = Mu + e(i

Var{e(ij)} = Sigma^2

Likelihoods of Interest
Model	Log(likelihood)	DF	AIC

A1	-51.42175	5	112.8435

A2	-15.31211	8	46.62422

A3	-15.66963	6	43.33925

R	-68.75058	2	141.5012

2	-23.73254	4	55.46509

Additive constant for all log-likelihoods =	-12.87. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Does response and/or variances differ among Dose levels? (A2

Are Variances Homogeneous? (A2 vs. A1)

Are variances adeguately modeled? (A2 vs. A3)

Does Model 2 fit the data? (A3 vs. 2)

Tests of Interest
Test	-2*log(Likelihood Ratio)	D. F.	p-value

Test 1	106.9	6	< 0.0001

Test 2	72.22	3	< 0.0001

Test 3	0.715	2	0.6994

Test 4	16.13	2	0.000315

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 4 is less than .1. Model 2 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	3.75968

BMDL =	2.76247

This document is a draft for review purposes only and does not constitute Agency policy.

E-53	DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.7.3. Figure for Selected Model: Exponential (M2)

Exponential Model 2 with 0.95 Confidence Level

dose

2	10:47 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-54	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.2.8. Crofton et al., 2005: Serum, T4

E.2.8.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

8

<0.0001

516.356

1.144E+02

6.239E+01



exponential (M3)

8

<0.0001

516.356

1.144E+02

6.239E+01

power hit bound (d = 1)

exponential
(M4)b

7

0.942

476.449

5.190E+00

3.029E+00



exponential (M5)

6

0.912

478.234

5.757E+00

3.094E+00



Hill

6

0.972

477.450

5.724E+00

3.024E+00



linear

8

<0.0001

522.460

2.406E+02

1.761E+02



polynomial, 8-
degree

8

<0.0001

522.460

2.406E+02

1.761E+02



power

8

<0.0001

522.460

2.406E+02

1.761E+02

power bound hit (power =1)

power,
unrestricted

7

0.018

491.101

2.449E+00

3.307E-01

unrestricted (power = 0.243)

a Constant variance model selected (p = 0.7647)
b Best-fitting model, BMDS output presented in this appendix

E.2.8.2. Output for Selected Model: Exponential (M4)

Crofton et al., 2005: Serum, T4

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\8_Crofton_2005_T4_ExpCV_l.(d)

Gnuplot Plotting File:

Mon Feb 08 10:48:04 2010

0

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-55	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 10

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values

Variable

Model 4

lnalpha

rho(S)

5.47437

0

104.999
0.00641895
0.445764

1

(S)

Specified

Parameter Estimates

Variable

lnalpha
rho

Model 4

5.50623

0

100.332
0. 076678
0.523626

1

Table of Stats From Input Data

Dose

0

0.0202

0.4882

1.	384
3. 455
9.257
23. 07
65. 65
180. 9
583.5

N
14

6
12

Obs Mean

100
96.27
98 . 57
99.76
93.32
70. 94
62 . 52
52 . 68
54 . 66
49.15

Obs Std Dev

15. 44
14 . 98
18 .11
19. 04
12 .11

12.74

14.75
22 .73
19.71
11.15

Estimated Values of Interest

Dose

0

0.0202
0.4882
1. 384
3. 455

Est Mean

100.3
100.3
98 . 58
95.52
89.21

Est Std

15. 69
15. 69
15. 69
15. 69
15. 69

Scaled Residual

-0.07952
-0.6231
-0.000744
0.6614
0.6422

This document is a draft for review purposes only and does not constitute Agency policy.

E-56	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

9.257	76.04	15.69	-0.7962

23.07	60.69	15.69	0.2854

65.65	52.85	15.69	-0.02621

180.9	52.54	15.69	0.3319

583.5	52.54	15.69	-0.4323

Other models for which	likelihoods are calculated:

Model A1:	Yij	=	Mu(i) + e(1j)

Var{e(ij)}	=	Sigma^2

Model A2:	Yij	=	Mu(i) + e(1j)

Var{e(ij)}	=	Sigma(i)^2

Model A3:	Yij	=	Mu(i) + e(1j)

Var{e(ij)}	=	exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	=	Mu + e(1)

Var{e(ij)}	=	Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-233.0774	11	488.1549

A2	-230.2028	20	500.4056

A3	-233.0774	11	488.1549

R	-268.4038	2	540.8076

4	-234.2243	4	476.4486

Additive constant for all log-likelihoods =	-66.16. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

76.4
5.749
5.749
2.294

D. F.

18
9
9
7

p-value

< 0.0001
0.7647
0.7647
0.9418

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

This document is a draft for review purposes only and does not constitute Agency policy.

E-57	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	5.18983

BMDL =	3.02894

E.2.8.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

10:48 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-58	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

E.2.9. Franc et al., 2001: S-D Rats, Relative Liver Weight
E.2.9.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

z2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

0.968

234.369

7.800E+00

6.040E+00



exponential (M3)

1

0.880

236.327

9.201E+00

6.051E+00



exponential (M4)

1

0.580

236.610

6.365E+00

4.512E+00



exponential (M5)

0

N/A

238.346

9.474E+00

4.425E+00



Hill

0

N/A

238.346

9.479E+00

3.004E+00



linear

2

0.858

234.610

6.365E+00

4.512E+00



polynomial, 3-
degree

1

0.935

236.311

8.946E+00

4.598E+00



power b

1

0.839

236.346

9.474E+00

4.587E+00



a Constant variance model selected (p = 0.107)
b Best-fitting model, BMDS output presented in this appendix

E.2.9.2. Output for Selected Model: Power
Franc et al., 2001: S-D Rats, Relative Liver Weight

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\88_Franc_2001_SD_RelLivWt_PowerCV_l.(d)

Gnuplot Plotting File: C:\l\Blood\88_Franc_2001_SD_RelLivWt_PowerCV_l.plt

Thu Apr 15 11:46:32 2010

Figure 5, SD rats, relative liver weight

The form of the response function is:

Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

The power is restricted to be greater than or egual to 1
A constant variance model is fit

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-59	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Default Initial Parameter Values

alpha
rho
control
slope
power

527 .447

0

100
0.947018
1.13144

Specified

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha	control	slope	power

alpha	1 -6.3e-009	5.4e-009 -4.7e-009

control -6.3e-009	1	-0.74	0.71

slope	5.4e-009	-0.74	1	-1

power -4.7e-009	0.71	-1	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
control
slope
power

Estimate

462.113
100.494
0.593276
1. 25841

Std. Err.
115.528
7.31114
1.31535
0.597816

Lower Conf. Limit

235.682
86.1645
-1. 98476
0.086712

Upper Conf. Limit

688.544
114 . 824
3.17131
2.43011

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	8	100	100	14	21.5	-0.065

6.587	8	108	107	16.9	21.5	0.158

14.48	8	117	118	25.9	21.5	-0.109

36.43	8	155	155	30.9	21.5	0.0157

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-60	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-114.152281	5	238.304562

A2	-111.103649	8	238.207299

A3	-114.152281	5	238.304562

fitted	-114.172940	4	236.345880

R	-125.052064	2	254.104127

Test

1

Test

2

Test

3

Test

4

(Note:

Explanation of Tests

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adequately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

27 .8968
6. 09726
6.09726
0. 0413179

<.0001
0.107
0.107
0. 8389

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Relative risk

Confidence level =	0.95

BMD = 9.47 4 08

BMDL = 4.587 3

This document is a draft for review purposes only and does not constitute Agency policy.

E-61	DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.9.3. Figure for Selected Model: Power

Power Model with 0.95 Confidence Level

dose

2	11:46 04/15 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-62	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.2.10. Franc et al., 2001: L-E Rats, Relative Liver Weight
E.2.10.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

z2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

0.441

208.974

1.708E+01

1.098E+01



exponential (M3)

2

0.441

208.974

1.708E+01

1.098E+01

power hit bound (d = 1)

exponential (M4)

1

0.785

209.408

7.997E+00

2.601E+00



exponential (M5)

1

0.785

209.408

7.997E+00

2.601E+00

power hit bound (d = 1)

Hill b

1

0.829

209.381

7.725E+00

1.225E+00

n lower bound hit (n = 1)

linear

2

0.499

208.725

1.570E+01

9.619E+00



polynomial, 3-
degree

1

<0.0001

10.000

8.604E+00

error



power

2

0.499

208.725

1.570E+01

9.619E+00

power bound hit (power =1)

Hill, unrestricted

C

0

N/A

211.337

7.217E+00

1.147E+00

unrestricted (n = 0.545)

power,
unrestricted

1

0.965

209.336

7.193E+00

error

unrestricted (power = 0.524)

a Non-constant variance model selected (p = 0.0632)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.10.2. Output for Selected Model: Hill

Franc et al., 2001: L-E Rats, Relative Liver Weight

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\89_Franc_2001_LE_RelLivWt_Hill_l.(d)

Gnuplot Plotting File: C:\l\Blood\8 9_Franc_2001_LE_RelLivWt_Hill_l.plt

Thu Apr 15 11:48:14 2010

Figure 5, L-E rats, relative liver weight

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose

Power parameter restricted to be greater than 1

This document is a draft for review purposes only and does not constitute Agency policy.

E-63	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho

intercept

5. 41581

0

100
22 . 225
0. 443155
18.746

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
rho

intercept

lalpha
1
-1

-0.21
0.33
0.18

rho
-1

1

0.21
-0.33
-0.18

intercept

-0.21
0.21
1

0. 028
0.35

v

0.33
-0.33
0. 028
1

0. 91

k

0.18
-0.18
0.35
0. 91
1

Parameter Estimates

Variable
lalpha
rho

intercept

k

Estimate

-17.2754
4 .77884
99.5348
36.3963
1

20.5223

Std. Err.

17 . 3066
3. 67625
3.61286
24 .1862
NA

28 . 2566

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-51.1957
-2 .42648
92.4538
-11.0079

-34.8596

16.6449

11.9842
106.616
83.8004

75.9042

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	8	100	99.5	10	10.5	0.125

6.584	8	106	108	17.9	12.9	-0.455

14.47	8	117	115	8.97	14.8	0.426

36.41	8	122	123	19.9	17.4	-0.0954

Model Descriptions for likelihoods calculated

This document is a draft for review purposes only and does not constitute Agency policy.

E-64	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij	= Mu(i) + e(ij

Var{e(ij)}	= Sigma^2

Model A2 :	Yij	= Mu(i) + e(ij

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model
A1
A2
A3

fitted
R

Log(likelihood)
-100.516456
-96.870820
-99.666984
-99.690373
-105.717087

Param1s
5

AIC
211.032912
209.741641
211.333969
209.380746
215.434174

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

17.6925
7.29127
5.59233
0.0467774

0.007048
0.06317
0.06104
0.8288

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Relative risk

Confidence level =	0.95

BMD =	7.724 92

BMDL =	1.22451

This document is a draft for review purposes only and does not constitute Agency policy.

E-65	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.2.10.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

a>

CO
c
o

Q.

CO
CD

a:

c
m

CD

140

130

120

110

100

90

Hill



BMDL

BMD

10

15

20

25

30

35

dose

11:48 04/15 2010

E.2.10.4. Output for Additional Model Presented: Hill, Unrestricted
Franc et al., 2001: L-E Rats, Relative Liver Weight

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\89_Franc_2001_LE_RelLivWt_Hill_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\8 9_Franc_2001_LE_RelLivWt_Hill_U_l.plt

Thu Apr 15 11:48:50 2010

Figure 5, L-E rats, relative liver weight

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
Power parameter is not restricted
The variance is to be modeled as Var(i)

Total number of dose groups = 4

exp(lalpha + rho

In(mean(i)

This document is a draft for review purposes only and does not constitute Agency policy.

E-66	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha =	5.41581

rho =	0

intercept =	100

v =	22.225

n =	0.443155

k =	18.746

Asymptotic Correlation Matrix of Parameter Estimates

lalpha rho	intercept v n k

lalpha 1 -1	-0.22 -0.14 0.24 -0.15

rho -1 1	0.22 0.14 -0.24 0.15

intercept -0.22 0.22	1 0.022 0.11 0.013

v -0.14 0.14	0.022 1 -0.9 1

n 0.24 -0.24	0.11 -0.9 1 -0.92

k -0.15 0.15	0.013 1 -0.92 1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

Estimate
-19.2405
5.19575
99.5348
440.285
0.544741
7266.27

Std. Err.

18 . 21
3. 86861
3.51796
13708.5
0.730981
485402

Lower Conf. Limit

-54.9315
-2 . 38657
92 . 6398
-26427.9
-0.887956
-944104

Upper Conf. Limit

16.4505
12 .7781
106.43
27308.5
1.97744
958637

Table of Data and Estimated Values

Dose	N Obs Mean	Est Mean

0	8	100	99.5

6.584	8	106	109

14.47	8	117	114

36.41	8	122	123

of Interest

Scaled Res.

0.128

-0.589
0.558
-0.0957

Obs Std Dev Est Std Dev

10

17 . 9
8 . 97
19. 9

10.3
13
14 . 6
17 . 8

Degrees of freedom for Test A3 vs fitted <= 0

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-67	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-100.516456	5	211.032912

A2 -96.870820	8	209.741641

A3 -99.666984	6	211.333969

fitted -99.668321	6	211.336641

R	-105.717087	2	215.434174

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

17.6925
7.29127
5.59233
0.00267242

0.007048
0.06317
0.06104
NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Relative risk

Confidence level =	0.95

BMD =	7.21718

BMDL =	1.147 42

This document is a draft for review purposes only and does not constitute Agency policy.

E-68	DRAFT—DO NOT CITE OR QUOTE

-------
E.2.10.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

11:48 04/15 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-69	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.2.11. Franc et al., 2001: S-D Rats, Relative Thymus Weight
E.2.11.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

z2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

0.814

285.107

2.478E+00

1.535E+00



exponential (M3)

1

0.016

292.452

3.173E+01

1.007E+00



exponential
(M4)b

1

0.720

286.825

1.878E+00

9.221E-01



exponential (M5)

0

N/A

288.696

3.296E+00

9.365E-01



Hill

0

N/A

288.696

3.625E+00

6.199E-01



linear

2

0.404

286.508

4.783E+00

3.893E+00



polynomial, 3-
degree0

2

0.404

286.508

4.783E+00

3.893E+00



power

2

0.404

286.508

4.783E+00

3.893E+00

power bound hit (power = 1)

power,
unrestricted

1

0.483

287.189

6.795E-01

3.271E-03

unrestricted (power = 0.515)

a Non-constant variance model selected (p = 0.0320)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.11.2. Output for Selected Model: Exponential (M4)

Franc et al., 2001: S-D Rats, Relative Thymus Weight

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\91_Franc_2001_SD_RelThyWt_Exp_l.(d)

Gnuplot Plotting File:

Thu Apr 15 11:51:19 2010

Figure 5, SD rats, relative thymus weight

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.

This document is a draft for review purposes only and does not constitute Agency policy.

E-70	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model 3 is nested within Model 5.
Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	3.35464

rho	1.08199

a	105

b	0.0569979

c	0.108531

d	1

Parameter Estimates
Variable	Model 4

lnalpha	2.4312

rho	1.28672

a	110.959

b	0.0663498

c	0.146486

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	8	100	83.2

6.587	8	91.17	47.97

14.48	8	51.41	43.48

36.43	8	22.79	29.98

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	111	69.78	-0.4442

6.587	77.43	55.36	0.7019

14.48	52.49	43.11	-0.0709

36.43	24.7	26.54	-0.2031

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-71	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-141.9834	5	293.9669

A2	-137.5818	8	291.1637

A3	-138.3482	6	288.6964

R	-146.9973	2	297.9946

4	-138.4123	5	286.8245

Additive constant for all log-likelihoods =	-29.41. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

18 . 83
8 .803
1. 533
0.1282

p-value

0.004459
0.03203
0.4647
0.7203

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 0.100000

Risk Type = Relative deviation
Confidence Level = 0.950000

BMD =	1.87814

BMDL =	0.922136

This document is a draft for review purposes only and does not constitute Agency policy.

E-72	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.2.11.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

150

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha	=	8.007 5

rho	=	0

beta_0	=	100

beta_l	=	0

beta~2	=	-0.475283

beta 3	=	0

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
beta_0
beta 1

The model parameter(s) -beta 2 -beta 3

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
1

-0. 99
0. 018
0.0095

rho

-0. 99
1

-0.022
-0.0024

beta_0

0. 018
-0.022
1

-0. 87

beta_l

0.0095
-0.0024
-0. 87
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

beta_0
beta_l
beta_2
beta 3

Estimate

2 . 8315
1.19884
94.5944
-1.97776
0
0

Std. Err.
1.71297
0. 416889
14.6685
0.509904
NA
NA

Lower Conf. Limit
-0.525852
0.381756
65.8446
-2 . 97715

Upper Conf. Limit
6.18885
2.01593
123.344
-0.978362

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	8	100	94.6	83.2	63	0.243

6.587	8	91.2	81.6	48	57.6	0.471

14.48	8	51.4	66	43.5	50.7	-0.811

36.43	8	22.8	22.5	30	26.7	0.0269

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij|

This document is a draft for review purposes only and does not constitute Agency policy.

E-74	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-141.983433	5	293.966865

A2	-137.581833	8	291.163667

A3	-138.348184	6	288.696368

fitted	-139.254163	4	286.508326

R	-146.997301	2	297.994602

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

18.8309
8.8032
1.5327
1.81196

0.004459
0.03203
0.4647
0.4041

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Relative risk

Confidence level =	0.95

BMD =	4.78292

BMDL =	3.8 932

This document is a draft for review purposes only and does not constitute Agency policy.

E-75	DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.11.5. Figure for Additional Model Presented: Polynomial, 3-degree

Polynomial Model with 0.95 Confidence Level

dose

2	11:51 04/15 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-76	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.2.12. Franc et al., 2001: L-E Rats, Relative Thymus Weight
E.2.12.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

z2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

0.440

301.449

2.726E+00

1.212E+00



exponential (M3)

2

0.440

301.449

2.726E+00

1.212E+00

power hit bound (d = 1)

exponential
(M4)b

1

0.227

303.266

2.084E+00

5.926E-01



exponential (M5)

0

N/A

303.805

7.859E+00

9.801E-01



Hill

0

N/A

303.805

7.480E+00

7.512E-01



linear

2

0.304

302.186

5.045E+00

3.349E+00



polynomial, 3-
degree

2

0.304

302.186

5.045E+00

3.349E+00



power

2

0.304

302.186

5.045E+00

3.349E+00

power bound hit (power = 1)

power,
unrestricted

1

0.168

303.710

1.374E+00

9.032E-09

unrestricted (power = 0.601)

a Constant variance model selected (p = 0.5063)
b Best-fitting model, BMDS output presented in this appendix

E.2.12.2. Output for Selected Model: Exponential (M4)

Franc et al., 2001: L-E Rats, Relative Thymus Weight

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\92_Franc_2001_LE_RelThyWt_ExpCV_l.(d)

Gnuplot Plotting File:

Thu Apr 15 11:53:37 2010

Figure 5, L-E rats, relative thymus weight

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-77	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

Specified

lnalpha	8.1814

rho(S)	0

a	105

b	0.0506168

c	0.166582

d	1

Parameter Estimates
Variable	Model 4

lnalpha	8.22706

rho	0

a	105.977

b	0.0660042

c	0.221786

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	8	100	54.72

6.584	8	95.41	70.46

14.47	8	38.69	47.97

36.41	8	34.98	77.96

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	106	61.16	-0.2764

6.584	76.91	61.16	0.8555

14.47	55.24	61.16	-0.765

36.41	30.96	61.16	0.186

Other models for which

Model A1:	Yij

Var{e(ij)}

likelihoods are calculated:

= Mu(i) + e(ij)

= Sigma/N2

This document is a draft for review purposes only and does not constitute Agency policy.

E-78	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-146.9024	5	303.8049

A2	-145.7361	8	307.4723

A3	-146.9024	5	303.8049

R	-150.6049	2	305.2098

4	-147.6329	4	303.2658

Additive constant for all log-likelihoods =	-29.41. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

9.738
2 . 333
2 . 333
1. 461

p-value

0.1362
0.5063
0.5063
0.2268

The p-value for Test 1 is greater than .05. There may not be a
diffence between responses and/or variances among the dose levels
Modelling the data with a dose/response curve may not be appropriate.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 0.100000

Risk Type = Relative deviation
Confidence Level = 0.950000

BMD =	2.0837 9

BMDL =	0.592601

This document is a draft for review purposes only and does not constitute Agency policy.

E-79	DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.12.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

dose

2	11:53 04/15 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-80	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.2.13. Franc et al., 2001: HAV Rats, Relative Thymus Weight
E.2.13.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

z2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential
(M2)b

2

0.698

261.646

5.094E+00

3.132E+00



exponential (M3)

1

0.407

263.616

5.944E+00

3.140E+00



exponential (M4)

1

0.396

263.646

5.063E+00

1.864E+00



exponential (M5)

0

N/A

264.927

9.945E+00

2.127E+00



Hill

0

N/A

264.927

9.638E+00

1.853E+00



linear

2

0.645

261.804

6.874E+00

5.006E+00



polynomial, 3-
degree

2

0.645

261.804

6.874E+00

5.006E+00



power

2

0.645

261.804

6.874E+00

5.006E+00

power bound hit (power = 1)

power,
unrestricted

1

0.363

263.755

5.487E+00

2.573E-01

unrestricted (power = 0.881)

a Constant variance model selected (p = 0.4331)
b Best-fitting model, BMDS output presented in this appendix

E.2.13.2. Output for Selected Model: Exponential (M2)

Franc et al., 2001: HAV Rats, Relative Thymus Weight

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\93_Franc_2001_HW_RelThyWt_ExpCV_l.(d)

Gnuplot Plotting File:

Thu Apr 15 11:55:55 2010

Figure 5, H/W rats, relative thymus weight

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-81	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 2

Specified

lnalpha	6.96647

rho(S)	0

a	56.9433

b	0.0204806

c	0

d	1

Parameter Estimates
Variable	Model 2

lnalpha	6.98 8 95

rho	0

a	103.047

b	0.0206828

c	0

d	1

Table of Stats From Input Data
Dose	N	Obs Mean	Obs Std Dev

0	8	100	35.98

6.588	8	97.53	32.98

14.48	8	71.02	23.99

36.44	8	49.29	43.48

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	103	32.93	-0.2617

6.588	89.92	32.93	0.6532

14.48	76.38	32.93	-0.4596

36.44	48.49	32.93	0.06871

Other models for which

Model A1:	Yij

Var{e(ij)}

likelihoods are calculated:

= Mu(i) + e(ij)

= Sigma/N2

This document is a draft for review purposes only and does not constitute Agency policy.

E-82	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-127.4636	5	264.9271

A2	-126.0925	8	268.185

A3	-127.4636	5	264.9271

R	-132.935	2	269.87

2	-127.8231	3	261.6463

Additive constant for all log-likelihoods =	-29.41. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Test

2 :

Test

3:

Test

4 :

Does response and/or variances differ among Dose levels:
Are Variances Homogeneous? (A2 vs. A1)

Are variances adeguately modeled? (A2 vs. A3)

Does Model 2 fit the data? (A3 vs. 2)

(A2 vs. R)

Test

Test 1
Test 2
Test 3
Test 4

Tests of Interest

-2*log(Likelihood Ratio)

13. 69
2.742
2.742
0.7192

p-value

0.03336
0.4331
0.4331
0. 698

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 4 is greater than .1. Model 2 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 0.100000

Risk Type = Relative deviation
Confidence Level = 0.950000

BMD =	5.09411

BMDL =	3.13214

This document is a draft for review purposes only and does not constitute Agency policy.

E-83	DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.13.3. Figure for Selected Model: Exponential (M2)

Exponential_beta Model 2 with 0.95 Confidence Level

dose

2	11:55 04/15 2010

3

4

5

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-84	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.2.14. Hojo et al., 2002: DRL Reinforce Per Minute
E.2.14.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

Hill

1

0.101

4.465

1.667E+00

6.209E-08

n upper bound hit (n = 18)

linear

2

0.009

9.124

1.352E+01

6.020E+00



polynomial, 3-
degree

2

0.009

9.124

1.352E+01

6.020E+00



power

2

0.009

9.124

1.352E+01

6.020E+00

power bound hit (power =1)

power,
unrestricted

1

0.025

6.780

2.428E-01

1.070E-14

unrestricted (power = 0.103)

exponential (M2)

2

0.007

9.612

1.623E+01

8.673E+00



exponential (M3)

2

0.007

9.612

1.623E+01

8.673E+00

power hit bound (d = 1)

exponential
(M4)b

1

0.054

5.488

1.316E+00

2.367E-03



exponential (M5)

0

N/A

6.465

1.728E+00

9.452E-03



a Constant variance model selected (p = 0.4321)
b Best-fitting model, BMDS output presented in this appendix

E.2.14.2. Output for Selected Model: Exponential (M4)

Hojo et al., 2002: DRL Reinforce Per Minute

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\21_Hojo_2002_DRLrein_ExpCV_l.(d)

Gnuplot Plotting File:

Mon Feb 08 10:49:08 2010

Table 5, values adjusted by a constant to allow exponential model

The form of the response function by Model:

Model 2:	Y[dose] = a * exp{sign * b * dose}

Model 3:	Y[dose] = a * exp{sign * (b * dosej^d}

Model 4:	Y[dose] = a * [c-(c-l) * exp{-b * dose}]

Model 5:	Y[dose] = a * [c-(c-l) * exp{-(b * dosej^d}]

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.

Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-85	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

Specified

lnalpha	-1.29672

rho(S)	0

a	0.0817

b	0.15642

c	16.3733

d	1

Parameter Estimates
Variable	Model 4

lnalpha	-1.11961

rho	0

a	0.0547452

b	0.708154

c	18.214

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	5	0.086	0.448

1.625	5	0.536	0.821

4.169	6	1.274	0.54

10.7	5	0.737	0.443

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	0.05475	0.5713	0.1223

1.625	0.6989	0.5713	-0.6375

4.169	0.9479	0.5713	1.398

10.7	0.9966	0.5713	-1.016

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-86	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	3.11555	5	3.7689

A2	4.489557	8	7.020886

A3	3.11555	5	3.7689

R	-2.435087	2	8.870174

4	1.255891	4	5.488219

Additive constant for all log-likelihoods =	-19.3. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

13. 85
2.748
2.748
3.719

p-value

0.03137
0.4321
0.4321
0.05379

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is less than .1. Model 4 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	1.31616

This document is a draft for review purposes only and does not constitute Agency policy.

E-87	DRAFT—DO NOT CITE OR QUOTE

-------
1	E'.MDL = 0.00236664

2

3

4	E.2.14.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

5	10:49 02/08 2010

6

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-88	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.2.15. Hojo et al., 2002: DRL Response Per Minute

E.2.15.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

Hill

0

N/A

126.353

1.373E+00

1.070E-14



linear

2

0.006

132.243

1.064E+01

5.340E+00



polynomial, 3-
degree

2

0.006

132.243

1.064E+01

5.340E+00



power

2

0.006

132.243

1.064E+01

5.340E+00

power bound hit (power = 1)

power,
unrestricted

2

0.741

122.455

1.070E+03

error

unrestricted (power = 0)

exponential (M2)

2

0.570

122.980

5.027E-01

error



exponential (M3)

2

0.570

122.980

5.027E-01

error

power hit bound (d = 1)

exponential
(M4)b

1

0.477

124.360

3.813E-01

1.553E-02



exponential (M5)

0

N/A

126.353

8.430E-01

2.221E-02



a Constant variance model selected (p = 0.3004)
b Best-fitting model, BMDS output presented in this appendix

E.2.15.2. Output for Selected Model: Exponential (M4)

Hojo et al., 2002: DRL Response Per Minute

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\23_Hojo_2002_DRLresp_ExpCV_l.(d)

Gnuplot Plotting File:

Mon Feb 08 10:50:10 2010

Table 5, values adjusted by a constant to allow exponential model

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-89	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

Specified

lnalpha	4.51689

rho(S)	0

a	24.6362

b	0.379327

c	0.0184785

d	1

Parameter Estimates
Variable	Model 4

lnalpha	4.54096

rho	0

a	23.4674

b	1.61185

c	0.101317

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	5	23.46	7.986

1.625	5	4.013	10.96

4.169	6	0.478	7.194

10.7	5	4.594	15.23

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	23.47	9.684	-0.001008

1.625	3.915	9.684	0.02265

4.169	2.403	9.684	-0.4869

10.7	2.378	9.684	0.5118

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-90	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-57.92733	5	125.8547

A2	-56.09669	8	128.1934

A3	-57.92733	5	125.8547

R	-64.49611	2	132.9922

4	-58.1801	4	124.3602

Additive constant for all log-likelihoods =	-19.3. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

16.8
3. 661
3. 661
0.5056

p-value

0.01005
0.3004
0.3004
0.4771

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	0.381347

This document is a draft for review purposes only and does not constitute Agency policy.

E-91	DRAFT—DO NOT CITE OR QUOTE

-------
1	E'.MDL = 0.0155267

2

3

4	E.2.15.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

5	10:50 02/08 2010

6

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-92	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

E.2.16. Kattainen et al., 2001: 3rd Molar Eruption, Female
E.2.16.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

logistic

3

0.360

88.508

9.223E+00

6.671E+00

negative intercept (intercept =
-1.586)

log-logistica

3

0.982

85.227

2.399E+00

1.328E+00

slope bound hit (slope = 1)

log-probit

3

0.522

87.424

7.346E+00

4.561E+00

slope bound hit (slope =1)

probit

3

0.379

88.352

8.802E+00

6.549E+00

negative intercept (intercept =
-0.975)

multistage, 4-
degree

3

0.781

86.155

4.042E+00

2.626E+00

final B = 0

log-logistic,
unrestricted b

2

0.949

87.162

1.931E+00

1.840E-01

unrestricted (slope = 0.91)

log-probit,
unrestricted

2

0.941

87.181

2.075E+00

2.395E-01

unrestricted (slope = 0.549)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.2.16.2. Output for Selected Model: Log-Logistic

Kattainen et al., 2001: 3rd Molar Eruption, Female

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\24_Katt_2001_Erup_LogLogistic_BMRl.(d)

Gnuplot Plotting File: C:\l\Blood\24_Katt_2001_Erup_LogLogistic_BMRl.plt

Mon Feb 08 10:50:39 2010

Figure 2

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

Slope parameter is restricted as slope >= 1
Total number of observations = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

This document is a draft for review purposes only and does not constitute Agency policy.

E-93	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

Default Initial Parameter Values
background =	0.0625

intercept =	-3.07535

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept

background	1	-0.53

intercept	-0.53	1

Parameter Estimates

Variable
background
intercept
slope

Estimate

0.0699339
-3.07219
1

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.







Analysis

of

Deviance Table







Model

Log(likelihooc

) #

Param's Deviance Test

d

f. P-value

Full

model



-40.5286



5







Fitted

model



-40.6137



2 0.170195

3

0.9823

Reduced

model



-50.7341



1 20

411

4

0.0004142



AIC:



85.2274



















Goodness of Fit























Scaled

Dose

Est

. Prob

Expected

Observed

Size



Residual

0.0000

0.

0 6 9 9

1.

119

1. 000

16



-0.117

2 2297

0.

1570

2 .

6 6 9

3. 000

17



0.221

6.2523

0.

2788

4 .

182

4 . 000

15



-0.105

16.0824

0.

4670

5.

604

6. 000

12



0.229

46.8576

0.

7066

13.

426

13.000

19



-0.215

Chi ^2 = 0

. 17

d. f.

= 3

P

-value = 0.9820







Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	2.3987 9

BMDL =	1.32815

This document is a draft for review purposes only and does not constitute Agency policy.

E-94	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.16.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

0.8

"O

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-006
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background =	0.0625

intercept =	-2.7659

slope =	0.901885

Asymptotic Correlation Matrix of Parameter Estimates
background intercept	slope

background	1	-0.52	0.38

intercept	-0.52	1	-0.94

slope	0.38	-0.94	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.0630045	*	*	*

intercept	-2.79616	*	*	*

slope	0.910333	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-40.5286	5

Fitted model -40.5811 3	0.105049 2	0.9488

Reduced model -50.7341 1	20.411 4	0.0004142

AIC:	87.1622

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0

0630

1. 008

1

000

16

-0

008

2.2297

0

1683

2 .862

3

000

17

0

090

6.2523

0

2922

4 . 383

4

000

15

-0

217

16.0824

0

4692

5. 631

6

000

12

0

214

46.8576

0

6903

13.116

13

000

19

-0

058

Chi/N2 = 0.10	d.f. = 2	P-value = 0.94 91

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

This document is a draft for review purposes only and does not constitute Agency policy.

E-96	DRAFT—DO NOT CITE OR QUOTE

-------
BMD =	1.9307 9

BMDL =	0.184 03

E.2.16.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

"O

(D

-t—<

o

CD
!t=
<
c
o
¦+-'
o
m

0.6

0.4

0.2

0 :
BMDL

10:50 02/08 2010

dose

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-97	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.2.17. Kattainen et al., 2001: 3rd Molar Length, Female
E.2.17.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

3

<0.0001

124.866

1.669E+01

9.933E+00



exponential (M3)

3

<0.0001

124.866

1.669E+01

9.933E+00

power hit bound (d = 1)

exponential (M4)

2

0.002

147.120

4.237E-01

2.530E-01



exponential (M5)

2

0.002

147.120

4.237E-01

2.530E-01

power hit bound (d = 1)

Hill b

2

0.022

152.239

3.132E-01

1.679E-01

n lower bound hit (n = 1)

linear

3

<0.0001

124.024

1.982E+01

1.277E+01



polynomial, 4-
degree

3

<0.0001

124.024

1.982E+01

1.277E+01



power

3

<0.0001

124.024

1.982E+01

1.277E+01

power bound hit (power =1)

Hill, unrestricted0

1

<0.0001

130.856

1.215E-02

error

unrestricted (n = 13.042)

power,
unrestricted

2

0.263

157.201

1.964E-03

8.002E-06

unrestricted (power = 0.195)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.17.2. Output for Selected Model: Hill

Kattainen et al., 2001: 3rd Molar Length, Female

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\25_Katt_2001_Length_Hill_l.(d)

Gnuplot Plotting File: C:\l\Blood\25_Katt_2001_Length_Hill_l.plt

Mon Feb 08 10:51:09 2010

Figure 3 female only

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose

Power parameter restricted to be greater than 1

This document is a draft for review purposes only and does not constitute Agency policy.

E-98	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho

intercept

-2 . 37155

0

1. 85591
-0.507874
0. 845932
2 . 03129

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
rho

intercept

lalpha
1

-0. 98
-0.16
0.84
-0.38

rho

-0. 98
1

0.2
-0.79
0.4

intercept

-0.16
0.2
1

-0.3
-0.11

v

0.84
-0.79
-0.3
1

-0.52

k

-0.38
0.4
-0.11
-0.52
1

Parameter Estimates

Variable
lalpha
rho

intercept

k

Estimate
3.31084
-14.2657
1. 85483
-0.453667
1

1. 91219

Std. Err.
1.404
2.62739
0.0159477
0.0620227
NA

0.624785

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

0.559057
-19.4153
1.82357
-0.575229

0. 687636

6. 06262
-9.11612

1.88609
-0.332105

3.13675

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

16

i—1

CO
G\

i—1

CO
CJ1

0.0661

0. 0639

0.0674

2 . 23

17

1 . 58

1 . 61

0.185

0.175

-0.789

6.252

15

1 . 6

1 . 51

0.265

0.28

1. 22

i—1

G\

O
CO

12

1 . 5

1 .45

0.221

0.371

o
en

i—1

46.86

19

1. 35

1 .42

0.515

0. 431

-0.716

Model

Descriptions for

likelihoods

calculated





This document is a draft for review purposes only and does not constitute Agency policy.

E-99	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
56.758717
85.856450
84.934314
81.119648
45.373551

10

7
5
2

AIC
-101.517434
-151.712901
-155.868628
-152.239295
-86.747101

Test

1

Test

2

Test

3

Test

4

(Note:

Explanation of Tests

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

80.9658
58 .1955
1.84427
7 . 62933

<.0001
<.0001
0.6053
0.02205

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.313211

BMDL =	0.167922

This document is a draft for review purposes only and does not constitute Agency policy.

E-100 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.17.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

1.9

1.8

1.7

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha =	-2.37155

rho =	0

intercept =	1.85591

v =	-0.507874

n =	0.845932

k =	2 . 03129

Asymptotic Correlation Matrix of Parameter Estimates

lalpha	rho	intercept	v	n	k

lalpha 1	-0.98	-0.16	0.84	1.4e-016	3.3e-017

rho -0.98	1	0.22	-0.77	-2.2e-016	-5.1e-017

intercept -0.16	0.22	1	-0.35	6e-017	1.4e-017

v 0.84	-0.77	-0.35	1	-2.6e-016	-6.2e-017

n 1.4e-016	-2.2e-016	6e-017	-2.6e-016	1	1

k 3.3e-017	-5.1e-017	1.4e-017	-6.2e-017	1	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

Estimate

4 . 25154
-15.7639
1. 85591
-0.357293
13.0417
0.0136512

Std. Err.
1.5913
2.90127
0.0160104
0.0463784
4.64308e+013
2.57737e+011

Lower Conf. Limit

1.13265
-21.4503
1. 82453
-0.448193
-9.10027e+013
-5.05155e+011

Upper Conf. Limit

7.37044
-10.0776
1.88729
-0.266393
9.10027e+013
5. 05155e + 011

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

2 . 23
6.252
16. 08
46.86

16

17
15
12
19

1.86
1. 58
1. 6
1. 5
1. 35

1.86
1. 5
1. 5
1. 5
1. 5

0.0661

0.185
0.265
0.221
0.515

0. 064
0.345
0.345
0.345
0.345

2.09e-009
0. 937
1.09
0.0534
-1. 9

Model Descriptions for likelihoods calculated

Model A1:	Yij =	Mu(i) + e(ij)

Var{e(ij)} =	Sigma/'2

Model A2:	Yij =	Mu(i) + e(ij)

Var{e(ij)} =	Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-102 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	56.758717	6	-101.517434

A2	85.856450	10	-151.712901

A3	84.934314	7	-155.868628

fitted	71.427978	6	-130.855955

R	45.373551	2	-86.747101

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	80.9658	8	<.0001

Test 2	58.1955	4	<.0001

Test 3	1.84427	3	0.6053

Test 4	27.0127	1	<.0001

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.012148

BMDL computation failed.

This document is a draft for review purposes only and does not constitute Agency policy.

E-103 DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.17.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model

CD
CO
c
o

Q.

CO
CD

a:

c
m

CD

19 r _

1.7

1.6

1.5

1.4

1.3

1.2

1.1

dose

10:51 02/08 2010

This document is a draft for review purposes only and does not constitute Agency policy.

E-104 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

E.2.18. Keller et al., 2007: Missing Mandibular Molars, CBA J
E.2.18.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

1

0.105

52.510

3.342E+00

8.986E-01



logistic

2

0.335

49.984

3.069E+00

2.212E+00

negative intercept (intercept =
-3.414)

log-logistic

1

0.105

52.524

4.009E+00

2.411E+00



log-probit

1

0.105

52.524

3.845E+00

2.421E+00



multistage, 1-
degreea

3

0.255

50.425

1.091E+00

7.624E-01



multistage, 2-
degree

1

0.122

51.391

1.916E+00

9.654E-01



multistage, 3-
degree

1

0.150

50.853

1.713E+00

9.584E-01



probit

2

0.342

49.904

2.927E+00

2.053E+00

negative intercept (intercept =
-1.873)

Weibull

1

0.108

52.219

2.744E+00

9.350E-01



a Best-fitting model, BMDS output presented in this appendix

E.2.18.2. Output for Selected Model: Multistage, 1-Degree

Keller et al., 2007: Missing Mandibular Molars, CBA J

Multistage Model. (Version: 3.0; Date: 05/16/2008)

Input Data File: C:\l\Blood\26_Keller_2007_Molars_Multil_l.(d)

Gnuplot Plotting File: C:\l\Blood\26_Keller_2007_Molars_Multil_l.plt

Mon Feb 08 10:51:47 2010

Table 1 using mandibular molars only

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0

This document is a draft for review purposes only and does not constitute Agency policy.

E-105 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) = 3.03988e+018

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.096571	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-21.5798	4

Fitted model -24.2126 1	5.26564 3	0.1533

Reduced model -71.326 1	99.4926 3	<.0001

AIC:	50.4251

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.0000

0. 000

0. 000

29

0. 000

0.5374

0.0506

1.163

2 . 000

23

0.796

4.2881

0.3391

9. 833

6. 000

29

-1.504

34.0560

0.9627

28.881

30.000

30

1. 078

Chi/N2 = 4.06	d.f. = 3	P-value = 0.2554

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	1.09102

BMDL =	0.762404

This document is a draft for review purposes only and does not constitute Agency policy.

E-106 DRAFT—DO NOT CITE OR QUOTE

-------
1

2	E'.MDU =	1.564 96

3

4	Taken together, (0.762404, 1.56496) is a 90	% two-sided confidence

5	interval for the EMD

6

7

8	E.2.18.3. Figure for Selected Model: Multistage, 1-Degree

Multistage Model with 0.95 Confidence Level

dose

9	10:51 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-107 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.2.19. Kociba et al., 1978: Urinary Coproporphyrin, Females
E.2.19.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

<0.0001

82.975

2.378E+01

1.340E+01



exponential (M3)

2

<0.0001

82.975

2.378E+01

1.340E+01

power hit bound (d = 1)

exponential
(M4)b

1

0.006

73.823

1.566E+00

7.180E-01



exponential (M5)

0

N/A

69.047

6.225E+00

1.586E+00



Hill

0

N/A

69.047

5.473E+00

error



linear

2

<0.001

82.233

1.790E+01

3.862E+00



polynomial, 3-
degree

2

<0.001

82.233

1.790E+01

3.862E+00



power

2

<0.001

82.233

1.790E+01

3.862E+00

power bound hit (power =1)

power,
unrestricted

1

<0.001

78.691

1.148E+00

8.984E-09

unrestricted (power = 0.416)

a Non-constant variance model selected (p = 0.0298)
b Best-fitting model, BMDS output presented in this appendix

E.2.19.2. Output for Selected Model: Exponential (M4)

Kociba et al., 1978: Urinary Coproporphyrin, Females

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\29_Kociba_l978_Copro_Exp_l.(d)

Gnuplot Plotting File:

Mon Feb 08 10:52:47 2010

Table2-UrinaryCoproporphyrin

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-108 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	-5.58269

rho	2.98472

a	8.17

b	0.0692478

c	2.23623

d	1

Parameter Estimates
Variable	Model 4

lnalpha	-4.90852

rho	2.80743

a	8.91071

b	0.15304

c	1.97526

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	5	9.8	1.3

1.547	5	8.6	2

7.155	5	16.4	4.7

38.56	5	17.4	4

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	8.911	1.852	1.074

1.547	10.74	2.407	-1.991

7.155	14.69	3.736	1.021

38.56	17.58	4.805	-0.08246

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-109 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest
Model	Log(likelihood)	DF	AIC

A1

-31.69739

5

73

39478

A2

-27 . 21541

8

70

43081

A3

-28.16434

6

68

32868

R

-41. 73188

2

87

46376

4

-31.91136

5

73

82272

Additive constant for all log-likelihoods =	-18.38. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Does response

and/or variances differ among Dose levels? (A2 vs. R)

Test

2 :

Are Variances

Homogeneous? (A2 vs. A1)

Test

3:

Are variances

adequately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

29. 03
8 . 964
1.898
7 . 494

p-value

C 0.0001
0.02977
0.3872
0.00619

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is less than .1. Model 4 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	1.56562

BMDL =	0.718033

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 10 DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.19.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

2	10:52 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-lll DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

E.2.20. Kociba et al., 1978: Uroporphyrin per Creatinine, Female
E.2.20.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

0.755

-93.828

1.641E+01

1.259E+01



exponential (M3)

2

0.755

-93.828

1.641E+01

1.259E+01

power hit bound (d = 1)

exponential (M4)

1

0.499

-91.935

1.216E+01

3.958E+00



exponential (M5)

0

N/A

-90.190

7.542E+00

4.128E+00



Hill

0

N/A

-90.190

7.607E+00

3.966E+00



linear b

2

0.793

-93.928

1.306E+01

9.287E+00



polynomial, 3-
degree

2

0.793

-93.928

1.306E+01

9.287E+00



power

1

0.497

-91.928

1.326E+01

9.287E+00



a Constant variance model selected (p = 0.4919)
b Best-fitting model, BMDS output presented in this appendix

E.2.20.2. Output for Selected Model: Linear

Kociba et al., 1978: Uroporphyrin per Creatinine, Female

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\l\Blood\28_Kociba_l978_Uropor_LinearCV_l.(d)

Gnuplot Plotting File: C:\l\Blood\28_Kociba_197 8_Uropor_LinearCV_l.plt

Mon Feb 08 10:52:17 2010

Table 2

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Signs of the polynomial coefficients are not restricted
A constant variance model is fit

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 12 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Default Initial Parameter Values
alpha = 0.0030385

rho =	0 Specified

beta_0 =	0.149139

beta 1 = 0.00381789

Asymptotic Correlation Matrix of Parameter Estimates

The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
beta_0
beta 1

alpha
1

1. 9e-009
-2.6e-009

beta_0

1.9e-009
1

-0.6

beta_l
-2 . 6e-009
-0.6
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
beta_0
beta 1

Estimate

0. 00248773
0.149139
0.00381789

Std. Err.

0. 000786688
0.0139684
0.000711776

Lower Conf. Limit

0.000945846
0.121761
0.00242284

Upper Conf. Limit

0. 00402961
0.176517
0.00521295

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	5

1.547	5

7.155	5

38.56	5

0.157

0.143
0.181
0.296

0.149

0.155
0.176
0.296

0. 05
0. 037
0. 053
0. 074

0. 0499

0. 0499
0. 0499
0. 0499

0.352
-0.54
0.204
-0.0161

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

Log(likelihood)
50.195349

Param's
5

AIC
-90.390697

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 13 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

A2	51.400051	8	-86.800103

A3	50.195349	5	-90.390697

fitted	49.963863	3	-93.927727

R	41.049755	2	-78.099510

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	20.7006	6	0.002076

Test 2	2.40941	3	0.4919

Test 3	2.40941	3	0.4919

Test 4	0.46297	2	0.7934

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	13.064

BMDL =	9.2 8715

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 14 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.20.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

10:52 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-115 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.2.21. Latchoumycandane and Mathur, 2002: Sperm Production
E.2.21.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

<0.0001

93.831

1.739E+01

9.432E+00



exponential (M3)

2

<0.0001

93.831

1.739E+01

9.432E+00

power hit bound (d = 1)

exponential (M4)

1

0.700

75.261

1.912E-01

7.976E-02



exponential (M5)

0

N/A

77.263

2.925E-01

7.970E-02



Hill b

1

0.962

75.115

1.171E-01

1.324E-02

n lower bound hit (n = 1)

linear

2

<0.0001

94.250

1.995E+01

1.212E+01



polynomial, 3-
degree

2

<0.0001

94.250

1.995E+01

1.212E+01



power

2

<0.0001

94.250

1.995E+01

1.212E+01

power bound hit (power = 1)

Hill, unrestricted0

0

N/A

77.113

9.955E-02

1.228E-09

unrestricted (n = 0.916)

power,
unrestricted

1

0.501

75.566

6.921E-06

6.921E-06

unrestricted (power = 0.087)

a Constant variance model selected (p = 0.8506)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.21.2. Output for Selected Model: Hill

Latchoumycandane and Mathur, 2002: Sperm Production

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\30_Latch_2002_Sperm_HillCV_l.(d)

Gnuplot Plotting File: C:\l\Blood\30_Latch_2002_Sperm_HillCV_l.plt

Mon Feb 08 10:53:26 2010

(xlO^) Table 1 without Vitamin E

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 16 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Power parameter restricted to be greater than 1
A constant variance model is fit

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho

intercept

7 . 23328

0

22.19
-9.09
1. 93059
0.546864

Specified

Asymptotic Correlation Matrix of Parameter Estimates

alpha
intercept

The model parameter(s) -rho -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

-2 . 2e-009
-3.7e-008
-5.9e-009

intercept
-2 . 2e-009
1

-0.76
-0.23

v

-3.7e-008
-0.76
1

-0.24

k

-5.9e-009
-0.23
-0.24
1

Parameter Estimates

Variable
alpha
intercept

k

Estimate

6.0283
22.1894
-9.16715
1

0.320198

Std. Err.
1.74022
1.00236
1.30966
NA

0.220443

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

2.61753
20.2248
-11.734

-0.111862

9. 43907
24 .154
-6.60026

0.752259

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	6	22.2	22.2	2.67	2.46	0.000631

0.7845 6	15.7 15.7	2.65 2.46	-0.00931

4.651	6	13.7	13.6	2.19	2.46	0.0372

27.27	6	13.1	13.1	3.16	2.46	-0.0285

Model Descriptions for likelihoods calculated
Model A1:	Yij = Mu(i) + e(ij)

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 17 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

Var{e(ij)} = Sigma^2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model

Log(likelihood)

# Param's

AIC

A1

-33.556444

5

77 . 112888

A2

-33.158811

8

82.317623

A3

-33.556444

5

77 . 112888

fitted

-33.557588

4

75.115176

R

-47 . 392394

2

98 .784788

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

28.4672
0.795266
0.795266
0.00228746

<.0001
0.8506
0.8506
0.9619

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.117131

BMDL =	0.0132353

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 18 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.21.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

10:53 02/08 2010

E.2.21.4. Output for Additional Model Presented: Hill, Unrestricted
Latchoumycandane and Mathur, 2002: Sperm Production

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\30_Latch_2002_Sperm_HillCV_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\30_Latch_2002_Sperm_HillCV_U_l.plt

Mon Feb 08 10:53:26 2010

(xl0^6) Table 1 without Vitamin E

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 19 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha =	7.23328

rho =	0 Specified

intercept =	22.19

v =	-9.09

n =	1.93059

k =	0.546864

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
intercept

alpha
1

-9.8e-009
1.6e-007
1.6e-007
1.2e-007

intercept
-9.8e-009
1

-0.5

-0.015
-0.13

v

1.6e-007
-0.5

1

0.76
0.56

n

1.6e-007
-0.015
0.76
1

0.86

k

1. 2e-007
-0.13
0.56
0.86
1

Parameter Estimates

Variable
alpha
intercept

Estimate

6.02773
22.19
-9.23667
0. 916265
0.301742

Std

Err.
1.74006
1.00231
2.03204
1.66287
0.440535

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

2.61728
20.2255
-13.2194
-2 . 34291
-0.561692

9. 43818
24 .1545
-5.25394
4 .17544
1.16518

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	6 22.2 22.2 2.67	2.46	3.4e-008

0.7845	6 15.7 15.7 2.65 2.46	-1.51e-007

4.651	6 13.7 13.6 2.19	2.46	2.62e-007

27.27	6 13.1 13.1 3.16	2.46	-5.45e-007

Degrees	of freedom for Test A3 vs fitted <= 0

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-120 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model

Log(likelihood)

# Param's

AIC

A1

-33.556444

5

77 . 112888

A2

-33.158811

8

82.317623

A3

-33.556444

5

77 . 112888

fitted

-33.556444

5

77 . 112888

R

-47 . 392394

2

98 .784788

Explanation of Tests

Test 1:

Test 2
Test 3
Test 4
(Note:

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

28.4672
0.795266
0.795266
>.96332e-013

<.0001
0.8506
0.8506
NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than
model appears to be appropriate here

.1. A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.0995543

BMDL = 1.22818e-009

This document is a draft for review purposes only and does not constitute Agency policy.

E-121 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.21.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

10:53 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-122 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.2.22. Li et al., 1997: FSH

E.2.22.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

8

<0.0001

1095.292

5.222E+02

4.121E+02



exponential (M3)

8

<0.0001

1095.292

5.222E+02

4.121E+02

power hit bound (d = 1)

exponential (M4)

7

<0.0001

1059.480

3.432E+01

9.930E+00



exponential (M5)

6

<0.0001

1066.195

1.019E+02

8.583E-01



Hill

7

<0.0001

1056.459

5.423E+00

error

n lower bound hit (n = 1)

linear

8

<0.0001

1077.695

2.003E+02

1.357E+02



polynomial, 8-
degree

9

<0.0001

1155.670

error

1.916E+02



power b

8

<0.0001

1077.695

2.003E+02

1.357E+02

power bound hit (power =

1)

Hill, unrestricted

6

0.001

1039.481

2.204E-01

error

unrestricted (n = 0.32)

power,
unrestricted0

7

0.002

1037.474

1.963E-01

2.484E-02

unrestricted (power = 0.305)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.22.2. Output for Selected Model: Power

Li et al., 1997: FSH

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\72_Li_l997_FSH_Pwr_l.(d)

Gnuplot Plotting File: C:\l\Blood\72_Li_1997_FSH_Pwr_l.plt

Mon Feb 08~13:36:35 2010

Figure 3: FSH in female S-D rats 24hr after dosing, 22 day old rats

The form of the response function is:

Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose

The power is restricted to be greater than or egual to 1

This document is a draft for review purposes only and does not constitute Agency policy.

E-123 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

Total number of dose groups = 10

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho
control
slope
power

9.8191

0

22.1591
52 .284
0.294106

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope

The model parameter(s) -power

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
1

-0. 99
-0.29
-0.033

rho

-0. 99
1

0.2
0. 033

control

-0.29
0.2
1

-0.36

slope

-0.033
0. 033
-0.36
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
control
slope
power

Estimate
3.50054
1. 27087
87 .4348
0. 492306
1

Std. Err.

1. 225
0.241869
12.9347
0.0919718
NA

Lower Conf. Limit
1.09958
0.796814
62.0833
0.312044

Upper Conf. Limit

5.9015
1.74492
112 .786
0. 672567

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

10

23. 9

CO
-j

29.6

CO
cxi

-2 . 04

0.266

10

22 . 2

87 . 6

48.5

98 . 7

-2 .1

0.7988

10

85.2

87 . 8

94 . 3

98 . 9

-0.0832

2 .097

10

73.3

CO
CO

CJ1

48.5

99.4

-0.483

5. 867

10

126

90.3

159

101

1.12

15

10

132

CO
cxi

116

104

1.14

43.33

10

117

109

51. 2

113

0.223

119. 9

10

304

146

154

137

3. 65

386

10

347

277

151

205

1. 07

1172

10

455

664

286

358

i—1

CO
CJ1

This document is a draft for review purposes only and does not constitute Agency policy.

E-124 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)

-535.687163
-496.367061
-502.709623
-534.847518
-574.835246

Param's

11

20

12
4
2

AIC
1093.374327
1032.734122
1029.419246
1077.695035
1153.670492

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

156.936

78 . 6402
12 . 6851
64 . 2758

18
9

<.0001
<.0001
0.1232
<.0001

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 200.314

This document is a draft for review purposes only and does not constitute Agency policy.

E-125 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E'.MDL = 135.673

E.2.22.3. Figure for Selected Model: Power

Power Model with 0.95 Confidence Level

0	200	400	600	800	1000	1200

dose

13:36 02/08 2010

E.2.22.4. Output for Additional Model Presented: Power, Unrestricted
Li et al., 1997: FSH

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\72_Li_l997_FSH_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\72_Li_1997_FSH_Pwr_U_l.plt

Mon Feb 08 13:36:46 2010

Figure 3: FSH in female S-D rats 24hr after dosing, 22 day old rats

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-126 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

Total number of dose groups = 10

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho
control
slope
power

9.8191

0

22.1591
52 .284
0.294106

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope
power

lalpha
1

-0. 99
-0.69
-0. 06
0.26

rho

-0. 99
1

0. 65
0. 0089
-0.23

control

-0.69
0. 65
1

-0.23
0. 029

slope

-0. 06
0. 0089
-0.23
1

-0. 85

power

0.26
-0.23
0. 029
-0. 85
1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate

3.67487
1.17882
15.8201
52.528
0.304867

Std. Err.

1.12134
0.221526
6. 87715
9. 46821
0.0336805

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

1.47708
0.744632
2.34113
33.9706
0.238855

5. 87265
1. 613

29.299
71. 0853
0.37088

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

0.266
0.7988
2 .097
5. 867
15
43.33
119. 9
386
1172

10
10

10
10
10
10
10
10
10
10

23. 9

22 . 2

85.2
73.3
126
132
117
304
347
455

15. 8

50. 9

64 . 9
81. 7
106
136
182
242
339
469

29.6
48.5

94 . 3
48.5
159
116
51. 2
154
151
286

32
63.7

73.5
84 .1
98 .1
114
135
160
195
236

0.795
-1.43
0. 876
-0.314
0. 652
-0.102
-1. 52
1.24
0.134
-0.182

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

This document is a draft for review purposes only and does not constitute Agency policy.

E-127 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Var{e(ij)} = Sigma^2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1	-535.687163	11	1093.374327

A2	-496.367061	20	1032.734122

A3	-502.709623	12	1029.419246

fitted	-513.737215	5	1037.474431

R	-574.835246	2	1153.670492

Test

1

Test

2

Test

3

Test

4

Explanation of Tests

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test -2*log(Likelihood Ratio) Test df	p-value

Test 1	156.936	18	<.0001

Test 2	78.6402	9	<.0001

Test 3	12.6851	8	0.1232

Test 4	22.0552	7	0.002485

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 0.196278

BMDL = 0.0248364

This document is a draft for review purposes only and does not constitute Agency policy.

E-128 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.22.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

13:36 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-129 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.23. Li et al., 2006: Estradiol, 3-Day

E.2.23.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

0.156

269.027

1.416E+01

5.544E+00



exponential (M3)

2

0.156

269.027

1.416E+01

5.544E+00

power hit bound (d = 1)

exponential (M4)

1

0.341

268.212

error

error



exponential (M5)

0

N/A

270.212

error

error



Hill

0

N/A

270.212

error

error



linear b

2

0.162

268.952

1.606E+01

5.379E+00



polynomial, 3-
degree

2

0.162

268.952

1.606E+01

5.379E+00



power

2

0.162

268.952

1.606E+01

5.379E+00

power bound hit (power =1)

Hill, unrestricted

0

N/A

270.265

9.273E+12

9.273E+12

unrestricted (n = 0.03)

power,
unrestricted

1

0.328

268.265

9.455E+10

error

unrestricted (power = 0.015)

a Constant variance model selected (p = 0.4372)
b Best-fitting model, BMDS output presented in this appendix

E.2.23.2. Output for Selected Model: Linear

Li et al., 2006: Estradiol, 3-Day

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\l\Blood\31_Li_2006_Estra_LinearCV_l.(d)

Gnuplot Plotting File: C:\l\Blood\31_Li_2006_Estra_LinearCV_l.plt

Mon Feb 08 10:54:00 2010

Figure 3, 3-day estradiol

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/'2 + ...

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Signs of the polynomial coefficients are not restricted
A constant variance model is fit

This document is a draft for review purposes only and does not constitute Agency policy.

E-130 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
alpha =	267 . 211

rho =	0 Specified

beta_0 =	16.1705

beta 1 =	1.0106

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha	beta 0	beta 1

alpha 1	2.le-012	5e-014

beta_0 2.le-012	1	-0.69

beta 1 5e-014	-0.69	1

Parameter Estimates

Variable
alpha
beta_0
beta 1

Estimate
263.435
16.1705
1.0106

95.0% Wald Confidence Interval
Std. Err.	Lower Conf. Limit Upper Conf. Limit

58.9057	147.981	378.888

3.55949	9.19407	23.147

1.2148	-1.37037	3.39156

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	10	10.2	16.2	12.2	16.2	-1.17

0.1588 10 19.9	16.3 20 16.2	0.697

2.839	10	24.7	19	14.6	16.2	1.11

5.124	10	18.1	21.3	17.6	16.2	-0.635

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-131 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-129.653527	5	269.307054

A2	-128.294657	8	272.589314

A3	-129.653527	5	269.307054

fitted	-131.476097	3	268.952193

R	-131.819169	2	267.638338

Explanation of Tests

Test

1

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adequately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Test

Tests of Interest
-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

7.04902
2 .71774
2 .71774
3.64514

0.3163

0.4372
0.4372
0.1616

The p-value for Test 1 is greater than .05. There may not be a
diffence between responses and/or variances among the dose levels
Modelling the data with a dose/response curve may not be appropriate

The p-value for Test 2 is greater than .1.
model appears to be appropriate here

A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	16.0605

BMDL =	5.37 8 95

This document is a draft for review purposes only and does not constitute Agency policy.

E-132 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.23.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

10:54 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-133 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.2.24. Li et al., 2006: Progesterone, 3-Day
E.2.24.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

<0.001

329.928

2.619E+00

error



exponential (M3)

2

0.001

328.101

1.340E-01

error

power hit bound (d = 1)

exponential (M4)

1

0.384

315.734

1.074E-02

6.633E-03



exponential (M5)

0

N/A

317.734

4.301E-02

4.272E-03



Hill b

1

0.386

315.728

9.461E-04

8.006E-11

n lower bound hit (n = 1)

linear

2

<0.001

330.729

3.891E+00

2.626E+00



polynomial, 3-
degree

2

<0.001

330.729

3.891E+00

2.626E+00



power

2

<0.001

330.729

3.891E+00

2.626E+00

power bound hit (power =1)

power,
unrestricted

1

0.404

315.673

2.812E-59

2.812E-59

unrestricted (power = 0.01)

a Non-constant variance model selected (p = 0.0013)
b Best-fitting model, BMDS output presented in this appendix

E.2.24.2. Output for Selected Model: Hill

Li et al., 2006: Progesterone, 3-Day

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\32_Li_2006_Progest_Hill_l.(d)

Gnuplot Plotting File: C:\l\Blood\32_Li_2006_Progest_Hill_l.plt

Wed Feb 10~10:57:14 2 010

Figure 4, 3-day progesterone

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose

Power parameter restricted to be greater than 1

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

This document is a draft for review purposes only and does not constitute Agency policy.

E-134 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Relative Function
Parameter

has been set to: le-008
has been set to: le-008

Default Initial Parameter Values

lalpha
rho

intercept

7 . 08699

0

61. 7404
-50.3835
1.47286
0.128302

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
rho

intercept

lalpha
1

-0. 99
-0.093
0. 82
0.22

rho

-0. 99
1

0.12
-0.79
-0.2

intercept

-0.093
0.12
1

-0.43
0. 014

v

0. 82
-0.79
-0.43
1

0. 035

k

0.22
-0.2

0. 014
0. 035
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

k

Estimate

14.0902
-2 . 27438
61. 7488
-42 .1007
1

0. 00282851

Std. Err.
3.36095
0.861553
3.3373
7 .70852
NA

0.020619

Lower Conf. Limit

7.50284
-3.963
55.2078
-57.2091

-0.037584

Upper Conf. Limit

20.6775
-0.585772
68.2898
-26.9922

0.0432411

NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	10	61.7	61.7	11.1	10.6	-0.00251

0.1588 10	30.6 20.4	40.5 37.2	0.865

2.839	10	16.9	19.7	33.3	38.7	-0.225

5.124	10	11.4	19.7	43.7	38.8	-0.678

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij|

This document is a draft for review purposes only and does not constitute Agency policy.

E-135 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -159.632675	5	329.265349

A2 -151.812765	8	319.625529

A3 -152.488175	6	316.976349

fitted -152.863841	5	315.727683

R -165.698875	2	335.397750

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	27.7722	6	0.0001037

Test 2	15.6398	3	0.001344

Test 3	1.35082	2	0.5089

Test 4	0.751333	1	0.3861

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 0.000946102
BMDL = 8.00639e-011

This document is a draft for review purposes only and does not constitute Agency policy.

E-136 DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.24.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

Hill

60

CD
CO
c
o

Q.

CO
CD

a:

c
m

CD

40

20

-20

iVIDL

BMD

10:57 02/10 2010

dose

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-137 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.2.25. Markowski et al., 2001: FRIO Run Opportunities
E.2.25.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential
(M2)b

2

0.304

117.150

8.570E+00

2.887E+00



exponential (M3)

2

0.304

117.150

8.570E+00

2.887E+00

power hit bound (d = 1)

exponential (M4)

1

0.371

117.570

3.452E+00

1.299E-02



exponential (M5)

0

N/A

118.918

2.315E+00

1.391E-02



Hill

0

N/A

118.918

1.801E+00

1.274E-09



linear

2

0.226

117.744

1.106E+01

5.741E+00



polynomial, 3-
degree

2

0.226

117.744

1.106E+01

5.741E+00



power

2

0.226

117.744

1.106E+01

5.741E+00

power bound hit (power = 1)

power,
unrestricted

1

0.239

118.158

5.768E+00

1.032E-14

unrestricted (power = 0.276)

a Constant variance model selected (p = 0,1719)
b Best-fitting model, BMDS output presented in this appendix

E.2.25.2. Output for Selected Model: Exponential (M2)

Markowski et al., 2001: FR10 Run Opportunities

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\33_Mark_2001_FR10opp_ExpCV_l.(d)

Gnuplot Plotting File:

Mon Feb 08 10:55:13 2010

Table 3

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-138 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 2

Specified

lnalpha	3.5321

rho(S)	0

a	6.77975

b	0.0581937

c	0

d	1

Parameter Estimates
Variable	Model 2

lnalpha	3.63127

rho	0

a	12.2901

b	0.0808832

c	0

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	7	13.29	8.65

1.557	4	11.25	5.56

4.03	6	5.75	3.53

10.32	7	7	6.01

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	12.29	6.145	0.4305

1.557	10.84	6.145	0.1347

4.03	8.871	6.145	-1.244

10.32	5.335	6.145	0.717

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-139 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-54.38526	5	118.7705

A2	-51.88568	8	119.7714

A3	-54.38526	5	118.7705

R	-57.45429	2	118.9086

2	-55.57522	3	117.1504

Additive constant for all log-likelihoods =	-22.05. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Test

2 :

Test

3:

Test

4 :

Does response and/or variances differ among Dose levels:
Are Variances Homogeneous? (A2 vs. A1)

Are variances adeguately modeled? (A2 vs. A3)

Does Model 2 fit the data? (A3 vs. 2)

(A2 vs. R)

Test

Test 1
Test 2
Test 3
Test 4

Tests of Interest
-2*log(Likelihood Ratio)

11.14

4.999
4.999
2 . 38

p-value

0.08423
0.1719
0.1719
0.3042

The p-value for Test 1 is greater than .05. There may not be a
diffence between responses and/or variances among the dose levels
Modelling the data with a dose/response curve may not be appropriate.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 4 is greater than .1. Model 2 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	8.56961

BMDL =	2.887 08

This document is a draft for review purposes only and does not constitute Agency policy.

E-140 DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.25.3. Figure for Selected Model: Exponential (M2)

Exponential Model 2 with 0.95 Confidence Level

20

(D
CO
c
o
Q.
CO
(D

(Z
c
(0
(D

15

10

10:55 02/08 2010

10

dose

This document is a draft for review purposes only and does not constitute Agency policy.

E-141 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.2.26. Markowski et al., 2001: FR2 Revolutions

E.2.26.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

0.236

217.219

8.486E+00

3.232E+00



exponential (M3)

2

0.236

217.219

8.486E+00

3.232E+00

power hit bound (d = 1)

exponential (M4)

1

0.263

217.583

3.413E+00

1.766E-02



exponential (M5)

0

N/A

218.532

2.415E+00

9.313E-01



Hill b

1

0.654

216.532

1.840E+00

5.992E-01

n upper bound hit (n = 18)

linear

2

0.180

217.764

1.058E+01

5.602E+00



polynomial, 3-
degree

2

0.180

217.764

1.058E+01

5.602E+00



power

2

0.180

217.764

1.058E+01

5.602E+00

power bound hit (power =1)

power,
unrestricted0

1

0.161

218.294

5.739E+00

1.032E-14

unrestricted (power = 0.318)

a Constant variance model selected (p = 0.1092)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.26.2. Output for Selected Model: Hill

Markowski et al., 2001: FR2 Revolutions

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\34_Mark_2001_FR2rev_HillCV_l.(d)

Gnuplot Plotting File: C:\l\Blood\34_Mark_2001_FR2rev_HillCV_l.plt

Mon Feb 08 10:55:47 2010

Table 3

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter restricted to be greater than 1
A constant variance model is fit

This document is a draft for review purposes only and does not constitute Agency policy.

E-142 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha =	2598 . 74

rho =	0 Specified

intercept =	119.29

v =	-62.79

n =	2 .13752

k =	2.53662

Asymptotic Correlation Matrix of Parameter Estimates

alpha
intercept

The model parameter(s) -rho -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

1. 2e-008
le-009
3 . 5e-008

intercept
1. 2e-008
1

-0. 81
-0.52

v

le-009
-0. 81
1

0.37

k

3 . 5e-008
-0.52
0.37
1

Parameter Estimates

Variable
alpha
intercept

k

Estimate
2183.85
119.29
-56.5223
18

1.68653

Std. Err.
630.425
17 . 6629
21.9082
NA

0.295154

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

948.245
84 . 6713
-99.4615

1.10804

3419.46

153.909
-13.5831

2 .26502

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

1. 557
4 . 03
10.32

119

109
56.5
68 .1

119

108
62 . 8
62 . 8

6 9.9
61
31. 2
33.2

46.7
46.7
46.7
46.7

-2 . 41e-007
2 . 29e-007
-0.329
0.304

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij|

This document is a draft for review purposes only and does not constitute Agency policy.

E-143 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -104.165520	5	218.331040

A2 -101.140174	8	218.280349

A3 -104.165520	5	218.331040

fitted -104.266162	4	216.532324

R -107.599268	2	219.198536

Explanation of Tests

Test

1

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

Test 1
Test 2
Test 3
Test 4

-2*log(Likelihood Ratio)

12.9182
6.05069
6.05069
0.201284

Test df

p-value

0.04435
0.1092
0.1092
0.6537

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1.
model appears to be appropriate here

A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	1.83952

BMDL =	0.599228

This document is a draft for review purposes only and does not constitute Agency policy.

E-144 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.26.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

200

150

CD
CO
c
o
Q.
CO
(D
££
c
(0
(D

100

50

10:55 02/08 2010

10

dose

E.2.26.4. Output for Additional Model Presented: Power, Unrestricted
Markowski et al., 2001: FR2 Revolutions

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\34_Mark_2001_FR2rev_PowerCV_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\34_Mark_2001_FR2rev_PowerCV_U_l.plt

Mon Feb 08 10:55:49 2010

Table 3

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-145 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial
alpha =
rho =
control =
slope =
power =

Parameter Values

2598.74

0 Specified
119.29
-10.3599
0. 824761

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha	control	slope	power

alpha 1	-3e-010	6.9e-010	9.9e-010

control -3e-010	1	-0.63	-0.28

slope 6.9e-010	-0.63	1	0.87

power 9.9e-010	-0.28	0.87	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

alpha	2350.22	678.449	1020.48	3679.95

control	120.082	18.0782	84.6491	155.514

slope	-27.8164	24.2447	-75.3352	19.7023

power	0.317923	0.350841	-0.369713	1.00556

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	7	119	120	69.9	48.5	-0.0432

1.557	4	109	88.1	61	48.5	0.843

4.03	6	56.5	76.8	31.2	48.5	-1.02

10.32	7	68.1	61.7	33.2	48.5	0.353

Model Descriptions for likelihoods calculated

Model A1:

Yij

Var{e ( ij ;

Mu(i) + e(ij ;
Sigma/N2

Model A2:

Yij

Var{e ( ij ;

Mu(i) + e(ij ;

Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that

This document is a draft for review purposes only and does not constitute Agency policy.

E-146 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -104.165520	5	218.331040

A2 -101.140174	8	218.280349

A3 -104.165520	5	218.331040

fitted -105.147159	4	218.294317

R -107.599268	2	219.198536

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	12.9182	6	0.04435

Test 2	6.05069	3	0.1092

Test 3	6.05069	3	0.1092

Test 4	1.96328	1	0.1612

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 5.73906

BMDL = 1.03181e-014

This document is a draft for review purposes only and does not constitute Agency policy.

E-147 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.26.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

10:55 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-148 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.2.27. Markowski et al., 2001: FR5 Run Opportunities
E.2.27.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

2

0.205

133.193

5.078E+00

2.439E+00



exponential (M3)

2

0.205

133.193

5.078E+00

2.439E+00

power hit bound (d = 1)

exponential (M4)

1

0.254

133.328

2.160E+00

6.854E-01



exponential (M5)

0

N/A

134.032

2.124E+00

9.667E-01



Hill b

1

0.939

132.032

1.723E+00

9.085E-01

n upper bound hit (n = 18)

linear

2

0.122

134.229

7.234E+00

4.430E+00



polynomial, 3-
degree

2

0.122

134.229

7.234E+00

4.430E+00



power

2

0.122

134.229

7.234E+00

4.430E+00

power bound hit (power =1)

power,
unrestricted0

1

0.134

134.268

2.666E+00

1.032E-14

unrestricted (power = 0.392)

a Constant variance model selected (p = 0.2262)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.27.2. Output for Selected Model: Hill

Markowski et al., 2001: FR5 Run Opportunities

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\35_Mark_2001_FR5opp_HillCV_l.(d)

Gnuplot Plotting File: C:\l\Blood\35_Mark_2001_FR5opp_HillCV_l.plt

Mon Feb 08 10:56:24 2010

Table 3

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter restricted to be greater than 1
A constant variance model is fit

This document is a draft for review purposes only and does not constitute Agency policy.

E-149 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha =	77.4849

rho =	0 Specified

intercept =	26.14

v =	-13.34

n =	2.77257

k =	2 .48811

Asymptotic Correlation Matrix of Parameter Estimates

alpha
intercept

The model parameter(s) -rho -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

-3.2e-009
1. 9e-008
6 . 2e-008

intercept
-3.2e-009
1

-0. 81
-0.51

v

1.9e-008
-0. 81
1

0.36

k

. 2e-008
-0.51
0.36
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
intercept

k

Estimate

64 . 5863
26.14
-13.1569
18

1.68073

Std. Err.

18 . 6445
3.03753
3.7676
NA

0.208677

Lower Conf. Limit

28.0438
20.1865
-20.5413

1. 27173

Upper Conf. Limit

101.129
32 .0935
-5.77257

2 . 08973

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	7	26.1	26.1	12.3	8.04	-1.9e-008

1.557	4	23.5	23.5	7.04	8.04	-1.94e-007

4.03	6	12.8	13	6.17	8.04	-0.0558

10.32	7	13.1	13	7.14	8.04	0.0517

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij|

This document is a draft for review purposes only and does not constitute Agency policy.

E-150 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -62.013133	5	134.026266

A2 -59.839035	8	135.678070

A3 -62.013133	5	134.026266

fitted -62.016025	4	132.032049

R -67.530040	2	139.060081

Explanation of Tests

Test

1

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio)

Test df

p-value

Test 1
Test 2
Test 3
Test 4

15.382
4.3482
4.3482
0.00578335

0.01748
0.2262
0.2262
0.9394

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1.
model appears to be appropriate here

A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	1.72335

BMDL =	0.9084 91

This document is a draft for review purposes only and does not constitute Agency policy.

E-151 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.27.3. Figure for Selected Model: Hill

40

35
30

O
Cl
CO
0)

(H

§	20

CD

2

15

10
5

10:56 02/08

E.2.27.4. Output for Additional Model Presented: Power, Unrestricted
Markowski et al., 2001: FR5 Run Opportunities

Hill Model with 0.95 Confidence Level

1 1 1—'—'—I—'—'—'—'—'—1 1 1 1 I	

0

2010

10

dose

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\35_Mark_2001_FR5opp_PwrCV_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\35_Mark_2 001_FR5opp_PwrCV_U_l.plt

Mon Feb 08 10:56:24 2010

Table 3

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 52 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho
control
slope
power

77.4849

0 Specified
26.14
-2.3827
0.844532

Asymptotic Correlation Matrix of Parameter Estimates

alpha
control

The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

-9.3e-009

control
-9.3e-009
1

slope
1. 4e-008
-0. 64

power
9.3e-009
-0.34

slope	1.4e-008	-0.64	1	0.9

power	9.3e-009	-0.34	0.9	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
control
slope
power

Estimate

70.8926
26.3582
-5.73309
0.391903

Std. Err.

20.4649
3.12902
4.02937
0.281862

Lower Conf. Limit
30.7821
20.2254
-13.6305
-0.160536

Upper Conf. Limit

111.003
32.4909
2.16433
0.944342

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	7	26.1	26.4	12.3	8.42	-0.0686

1.557	4	23.5	19.5	7.04	8.42	0.941

4.03	6	12.8	16.5	6.17	8.42	-1.06

10.32	7	13.1	12	7.14	8.42	0.343

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2
Model A3 uses any fixed variance parameters that

This document is a draft for review purposes only and does not constitute Agency policy.

E-153 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -62.013133	5	134.026266

A2 -59.839035	8	135.678070

A3 -62.013133	5	134.026266

fitted -63.134001	4	134.268002

R -67.530040	2	139.060081

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	15.382	6	0.01748

Test 2	4.3482	3	0.2262

Test 3	4.3482	3	0.2262

Test 4	2.24174	1	0.1343

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 2.66625

BMDL = 1.03181e-014

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 54 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.27.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

10:56 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 55 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.2.28. Miettinen et al., 2006: Cariogenic Lesions, Pups
E.2.28.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

3

0.410

162.280

3.401E+00

1.889E+00

power bound hit (power = 1)

logistic

3

0.371

162.518

4.108E+00

2.450E+00



log-logistica

3

0.602

161.292

1.428E+00

5.175E-01

slope bound hit (slope = 1)

log-probit

3

0.300

163.040

6.321E+00

3.127E+00

slope bound hit (slope =1)

multistage, 4-
degree

3

0.410

162.280

3.401E+00

1.889E+00

final B = 0

probit

3

0.350

162.656

4.548E+00

2.889E+00



Weibull

3

0.410

162.280

3.401E+00

1.889E+00

power bound hit (power = 1)

gamma,
unrestricted

2

0.798

161.801

3.374E-03

8.884E-
242

unrestricted (power = 0.215)

log-logistic,
unrestricted b

2

0.728

161.983

4.942E-02

error

unrestricted (slope = 0.465)

log-probit,
unrestricted

2

0.732

161.972

6.495E-02

error

unrestricted (slope = 0.289)

Weibull,
unrestricted

2

0.766

161.884

1.792E-02

error

unrestricted (power = 0.324)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.2.28.2. Output for Selected Model: Log-Logistic

Miettinen et al., 2006: Cariogenic Lesions, Pups

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\36_Miet_2006_Cariogenic_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\Blood\36_Miet_2006_Cariogenic_LogLogistic_l.plt

Mon Feb 08 10:56:59 2010

Table 2 converting the percentage into the number of animals, and control is Control II from the
study. Dose is in ng per kg and is from Table 1

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff

This document is a draft for review purposes only and does not constitute Agency policy.

E-156 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Independent variable = Dose

Slope parameter is restricted as slope >= 1
Total number of observations = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values

background =	0.595238

intercept =	-2.494

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept

background	1	-0.66

intercept	-0.66	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.644165	*	*	*

intercept	-2.55354	*	*	*

slope	1	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-77.6769	5

Fitted model -78.646 2	1.93832 3	0.5853

Reduced model -83.2067 1	11.0597 4	0.0259

AIC:	161.292

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0

6442

27.055

25.000

42

-0

662

2.2195

0

6 9 6 6

20.200

23.000

29

1

131

6.2259

0

7603

19.007

19.000

25

-0

003

16.0142

0

8416

20.198

20.000

24

-0

111

46.6355

0

9231

29.540

29.000

32

-0

358

H-

K>

II

1—1

CO



d.f.

= 3 P

-value = 0.6024







Benchmark Dose Computation

This document is a draft for review purposes only and does not constitute Agency policy.

E-157 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

Specified effect =
Risk Type	=

Confidence level =

0.1
Extra risk
0. 95

BMD =	1. 42805

BMDL =	0.517 4 95

E.2.28.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

dose

10:56 02/08 2010

E.2.28.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
Miettinen et al., 2006: Cariogenic Lesions, Pups

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\36_Miet_2006_Cariogenic_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\36_Miet_2006_Cariogenic_LogLogistic_U_l.plt

Mon Feb 08 10:56:59 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 58 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Table 2 converting the percentage into the number of animals, and control is Control II from the
study. Dose is in ng per kg and is from Table 1

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values

background =	0.595238

intercept = -0.739403
slope =	0.442847

Asymptotic Correlation Matrix of Parameter Estimates
background intercept	slope

background	1	-0.51	0.24

intercept	-0.51	1	-0.89

slope	0.24	-0.89	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.597745	*	*	*

intercept	-0.798024	*	*	*

slope	0.465259	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-77.6769	5

Fitted model -77.9915 3	0.629204 2	0.7301

Reduced model -83.2067 1	11.0597 4	0.0259

AIC:	161.983

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000	0.5977	25.105 25.000	42	-0.033

2.2195	0.7566	21.940 23.000	29	0.458

This document is a draft for review purposes only and does not constitute Agency policy.

E-159 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

6.2259	0.8042

16.0142	0.8474

4 6.6355	0.8 910

20.105 19.000
20.338 20.000
28.512 29.000

Chi' '2 = 0.63

d . f . = 2

P-value = 0.7281

25
24
32

-0.557
-0.192
0.277

E'.enchmark Dose Computation
Specified effect =	0 .1

Risk Typ'e	=	E::tra risk

Cc'nfidence level =	0.95

BMD =	0.04 94 22

E'.enchmark duse cO'mp'UtatiO'ri failed. Lower limit includes :eru.

E.2.28.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model

"O

(D

-t—<

o

CD
!t=
<

0.9

0.7

0.6

0.5

0.4

10:57 02/08 2010

10

20

30

dose

40

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-160 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

E.2.29. Murray et al., 1979: Fertility in F2 Generation
E.2.29.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

0

N/A

61.729

4.481E+00

1.590E+00



logistic

1

0.051

61.318

2.420E+00

1.722E+00

negative intercept (intercept =
-2.567)

log-logistic

0

N/A

61.729

4.971E+00

1.565E+00



multistage, 1-
degree

1

0.031

63.154

1.598E+00

8.747E-01



multistage, 2-
degreea

1

0.079

60.464

2.733E+00

1.366E+00



probit

1

0.048

61.544

2.250E+00

1.590E+00

negative intercept (intercept =
-1.459)

Weibull

0

N/A

61.729

5.042E+00

1.604E+00



log-probit,
unrestricted

0

N/A

61.729

4.244E+00

1.506E+00

unrestricted (slope = 3.182)

a Best-fitting model, BMDS output presented in this appendix

E.2.29.2. Output for Selected Model: Multistage, 2-Degree

Murray et al., 1979: Fertility in F2 Generation

Multistage Model. (Version: 3.0; Date: 05/16/2008)

Input Data File: C:\l\Blood\Murray_l979_fert_index_f2_Multi2_l.(d)

Gnuplot Plotting File: C:\l\Blood\Murray_197 9_fert_index_f2_Multi2_l.plt

Wed_Feb 10 16:06:28 2010~

Table 1 but expressed as number of dams who do not produce offspring

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal*dose/Nl-beta2*dose/N2 ) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 3

Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2

This document is a draft for review purposes only and does not constitute Agency policy.

E-161 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0567204
Beta(1) =	0

Beta(2) = 0.0155037

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Beta(l)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Background
Beta(2)

Background
1

-0.45

Beta(2)

-0.45
1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0780188	*	*	*

Beta(1)	0	*	*	*

Beta(2)	0.0141051	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-25.8194	3

Fitted model -28.2318 2	4.82474 1	0.02805

Reduced model -34.0009 1	16.363 2	0.0002798

AIC:	60.4 636

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000

0.0780

2.497

4 . 000

32

0. 991

1.1242

0.0943

1.886

0. 000

20

-1.443

5.8831

0.4341

8 . 683

9. 000

20

0.143

Chi ^2 = 3.08	d.f. = 1	P-value = 0.0790

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	2.73307

This document is a draft for review purposes only and does not constitute Agency policy.

E-162 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

E'.MDL =

1.36619

E'.MDU =	4.10938

Taken together, (1.36619, 4.10938) is a 90	% two-sided confidence

interval for the EMD

E.2.29.3. Figure for Selected Model: Multistage, 2-Degree

Multistage Model with 0.95 Confidence Level

dose

16:06 02/10 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-163 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

E.2.30. National Toxicology Program, 1982: Toxic Hepatitis, Male Mice
E.2.30.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

1

0.027

113.103

3.823E+00

2.005E+00



logistic

2

0.092

110.352

3.108E+00

2.465E+00

negative intercept (intercept =
-3.388)

log-logistic

1

0.026

113.089

3.797E+00

2.141E+00



log-probit

1

0.027

113.111

3.565E+00

2.294E+00



multistage, 3-
degreea

1

0.036

112.045

2.782E+00

1.343E+00



probit

2

0.082

110.512

2.763E+00

2.241E+00

negative intercept (intercept =
-1.894)

Weibull

1

0.025

113.044

3.967E+00

1.704E+00



a Best-fitting model, BMDS output presented in this appendix

E.2.30.2. Output for Selected Model: Multistage, 3-Degree

National Toxicology Program, 1982: Toxic Hepatitis, Male Mice

pit
2010

0

Multistage Model. (Version: 3.0; Date: 05/16/2008)

Input Data File: C:\l\Blood\37_NTP_1982_ToxHep_Multi3_l.(d)

Gnuplot Plotting File: C:\l\Blood\37_NTP_1982_ToxHep_Multi3_l.

Mon Feb 08 10:57:32

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/Nl-beta2*dose/N2-beta3*dose/N3) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-164 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Parameter

has been set to: le-008

Default Initial
Background =
Beta(1) =
Beta(2) =
Beta(3) =

Parameter Values

0.0471757
0.00749116
0

0.00139828

Asymptotic Correlation Matrix of Parameter Estimates

Background
Beta (1)
Beta(3)

The model parameter(s) -Beta(2)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Background
1

-0.77
0.69

Beta(1)

-0.77
1

-0. 95

Beta(3)

0.69
-0. 95
1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)
Beta(3)

Estimate

0. 0267933
0. 0283198
0

0.0012342

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-51.0633
-53.0224
-121.743

Param's

4
3
1

Deviance Test d.f.

3.91812
141.358

P-value

0.04777
<.0001

AIC:

112 . 045

Dose

Goodness of Fit

Est. Prob.

Expected

Observed

Scaled
Residual

0.0000
0.7665
2.2711
11.2437

0268
0482
1005
8775

1. 956

2 . 363
4 . 925
43.877

1. 000
5. 000
3. 000
44.000

73

49

49

50

-0.693
1.759
-0.915
0. 053

Chi ^2

4 .41

d.f.

P-value

0.0357

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

This document is a draft for review purposes only and does not constitute Agency policy.

E-165 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

BMD =

2 .78201

EMDL =	1.34 308

BMDLJ =	4.5214

Taken together, (1.34308,	4.5214 )
interval for the EMD

is a 90	% two'-sided confidenoe

E.2.30.3. Figure for Selected Model: Multistage, 3-Degree

Multistage Model with 0.95 Confidence Level

dose

10:57 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-166 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

E.2.31. National Toxicology Program, 2006: Alveolar Metaplasia
E.2.31.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

4

0.010

320.093

9.886E-01

8.393E-01

power bound hit (power =1)

logistic

4

<0.001

343.283

2.389E+00

2.052E+00

negative intercept (intercept =
-1.059)

log-logistica

3

0.723

312.558

6.497E-01

3.751E-01



log-probit

4

0.024

318.680

1.566E+00

1.318E+00

slope bound hit (slope =1)

multistage, 5-
degree

4

0.010

320.093

9.886E-01

8.393E-01

final B = 0

probit

4

<0.001

347.071

2.542E+00

2.219E+00

negative intercept (intercept =
-0.599)

Weibull

4

0.010

320.093

9.886E-01

8.393E-01

power bound hit (power =1)

gamma,
unrestricted

3

0.426

314.011

1.642E-01

1.874E-02

unrestricted (power = 0.503)

log-probit,
unrestricted

3

0.696

312.677

6.818E-01

2.740E-01

unrestricted (slope = 0.677)

Weibull,
unrestricted

3

0.522

313.492

2.644E-01

6.947E-02

unrestricted (power = 0.661)

a Best-fitting model, BMDS output presented in this appendix

E.2.31.2. Output for Selected Model: Log-Logistic

National Toxicology Program, 2006: Alveolar Metaplasia

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\40_NTP_2006_AlvMeta_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\Blood\4 0_NTP_2006_AlvMeta_LogLogistic_l.plt

Mon Feb 08 10:58:58 2010

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

Slope parameter is restricted as slope >= 1
Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

This document is a draft for review purposes only and does not constitute Agency policy.

E-167 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Relative Function Convergence has been set to: le-006
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values

background = 0.0377358
intercept =	-1.69494

slope =	1.12282

Asymptotic Correlation Matrix of Parameter Estimates
background intercept	slope

background	1	-0.21	0.1

intercept	-0.21	1	-0.93

slope	0.1	-0.93	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.0373462	*	*	*

intercept	-1.70923	*	*	*

slope	1.13164	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-152.615	6

Fitted model	-153.279	3	1.32728	3	0.7227

Reduced model	-216.802	1	128.374	5	<.0001

AIC:	312.558

Goodness of Fit

Scaled

Dose

Est

. Prob.

Expected

Observed

Size

Residual

0.0000

0.

0373

1. 979

2 . 000

53

0

015

2.5565

0.

3682

19.881

19.000

54

-0

249

5.6937

0.

5807

30.776

33.000

53

0

619

9.7882

0.

7162

37 .243

35.000

52

-0

690

16.5688

0.

8197

43.446

45.000

53

0

555

29.6953

0.

8976

46.674

46.000

52

-0

308

Chi ^2

1. 33

d.f.

P-value

0.7232

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

This document is a draft for review purposes only and does not constitute Agency policy.

E-168 DRAFT—DO NOT CITE OR QUOTE

-------
BMD =	0.64971

BMDL =	0.375051

E.2.31.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

"O

(D

-t—<

o

CD
!t=
<
c
o
¦+-'
o
m

0.6

0.4

0.2

0 -

10:58 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-169 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

E.2.32. National Toxicology Program, 2006: Eosinophilic Focus, Liver
E.2.32.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

3

0.293

331.902

3.573E+00

2.225E+00



logistic

4

0.405

330.400

5.949E+00

5.137E+00

negative intercept (intercept =
-2.043)

log-logistic

3

0.152

333.515

4.139E+00

2.077E+00



log-probit

4

0.192

332.312

4.889E+00

3.980E+00

slope bound hit (slope =1)

multistage, 5-
degree

3

0.752

329.328

3.393E+00

2.466E+00



probita

4

0.459

329.945

5.583E+00

4.864E+00

negative intercept (intercept
= -1.235)

Weibull

3

0.324

331.628

3.770E+00

2.249E+00



log-probit,
unrestricted

3

0.116

334.150

4.146E+00

2.152E+00

unrestricted (slope = 0.895)

a Best-fitting model, BMDS output presented in this appendix

E.2.32.2. Output for Selected Model: Probit

National Toxicology Program, 2006: Eosinophilic Focus, Liver

Probit Model. (Version: 3.1; Date: 05/16/2008)

Input Data File: C:\l\Blood\45_NTP_2006_LivEosFoc_Probit_l.(d)

Gnuplot Plotting File: C:\l\Blood\45_NTP_2006_LivEosFoc_Probit_l.plt

Mon Feb 08 11:00:54 2010

The form of the probability function is:

P[response] = CumNorm(Intercept + Slope* Dose),

where CumNorm(.) is the cumulative normal distribution function

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-170 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Default Initial	(and Specified) Parameter Values

background =	0 Specified

intercept =	-1.28017

slope =	0.0712441

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

intercept	slope

intercept	1	-0.77

slope	-0.77	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

intercept	-1.23453	0.125132	-1.47979	-0.989279

slope	0.0688678	0.00823346	0.0527305	0.085005

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-161.07	6

Fitted model -162.972 2	3.80461 4	0.4331

Reduced model -202.816 1	83.4925 5	<.0001

AIC:	329.945

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0

1085

5.751

3

000

53

-1

215

2.5565

0

1449

7 . 826

8

000

54

0

067

5.6937

0

1998

10.588

14

000

53

1

172

9.7882

0

2876

15.242

17

000

53

0

533

16.5688

0

4628

24.526

22

000

53

-0

6 9 6

29.6953

0

7912

41.932

42

000

53

0

023

Chi/N2 = 3.62	d.f. = 4	P-value = 0.4593

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	5.5830 9

BMDL =	4 . 86394

This document is a draft for review purposes only and does not constitute Agency policy.

E-171 DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.32.3. Figure for Selected Model: Probit

Probit Model with 0.95 Confidence Level

dose

2	11:00 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 72 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

E.2.33. National Toxicology Program, 2006: Fatty Change Diffuse, Liver
E.2.33.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

4

0.659

252.348

4.028E+00

2.923E+00



logistic

4

0.056

262.132

5.890E+00

5.042E+00

negative intercept (intercept =
-2.825)

log-logistic

4

0.359

254.413

4.254E+00

3.228E+00



log-probit

4

0.367

254.428

4.204E+00

3.277E+00



multistage, 5-
degree

3

0.581

254.045

3.524E+00

2.234E+00



probit

4

0.075

260.915

5.567E+00

4.784E+00

negative intercept (intercept =
-1.665)

Weibull3

4

0.724

251.989

3.917E+00

2.856E+00



a Best-fitting model, BMDS output presented in this appendix

E.2.33.2. Output for Selected Model: Weibull

National Toxicology Program, 2006: Fatty Change Diffuse, Liver

Weibull Model using Weibull Model (Version: 2.12; Date: 05/16/2008)
Input Data File: C:\l\Blood\47_NTP_2006_LivFatDiff_Weibull_l.(d)

Gnuplot Plotting File: C:\l\Blood\47_NTP_2006_LivFatDiff_Weibull_l.plt

Mon Feb 08 11:01:56 2010

NTP liver fatty change diffuse

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(-slope*doseApower)]

Dependent variable = DichEff
Independent variable = Dose

Power parameter is restricted as power >=1
Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial (and Specified) Parameter Values

Background = 0.00925926
Slope = 0.00721355

This document is a draft for review purposes only and does not constitute Agency policy.

E-173 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Power =

1.69678

Asymptotic Correlation Matrix of Parameter Estimates

Slope
Power

The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Slope
1

-0. 98

Power

-0. 98
1

Parameter Estimates

Variable
Background
Slope
Power

Estimate

0

0.0135075
1. 50444

Std. Err.

NA

0. 00640459
0.168981

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

0.00095478
1.17324

0.0260603
1. 83564

NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.

Analysis of Deviance Table

Model

Log(likelihood) #

Param's Deviance

Test

d

f. P-value

Full

model



- . 992

6







Fitted

model



-123.995

2 2.00444



4

0.7349

Reduced

model



-204.846

1 163.708



5

<.0001



AIC:



251.989















Goodness of Fit





















Scaled

Dose

Est

. Prob

Expected

Observed Si

ze



Residual

0.0000

0.

0000

0. 000

0. 000

53



0. 000

2.5565

0.

0539

2 . 912

2 . 000

54



-0.550

5.6937

0.

1688

8. 949

12.000

53



1.119

9.7882

0.

3415

18 .102

17.000

53



-0.319

16.5688

0.

6024

31.929

30.000

53



-0.542

29.6953

0.

8913

47 . 238

48 . 000

53



0.336

Chi ^2 = 2

. 06

d. f.

= 4 P

-value = 0.7243







Benchmark Dose

Computation









Specified

effect

=

0.1









Risk Type



=

Extra risk









Confidence

level

=

0. 95











BMD

=

3.91723











BMDL



2 . 85566









This document is a draft for review purposes only and does not constitute Agency policy.

E-174 DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.33.3. Figure for Selected Model: Weibull

Weibull Model with 0.95 Confidence Level

dose

2	11:01 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 75 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.2.34. National Toxicology Program, 2006: Gingival Hyperplasia, Squamous, 2 Years
E.2.34.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

4

0.036

314.985

7.743E+00

5.166E+00

power bound hit (power = 1)

logistic

4

0.016

318.602

1.392E+01

1.056E+01

negative intercept (intercept =
-1.859)

log-logistica

4

0.055

313.351

5.850E+00

3.730E+00

slope bound hit (slope = 1)

log-probit

4

0.005

321.426

1.535E+01

1.038E+01

slope bound hit (slope =1)

multistage, 5-
degree

4

0.036

314.985

7.743E+00

5.166E+00

final B = 0

probit

4

0.018

318.240

1.318E+01

9.924E+00

negative intercept (intercept =
-1.123)

Weibull

4

0.036

314.985

7.743E+00

5.166E+00

power bound hit (power = 1)

gamma,
unrestricted

3

0.633

307.618

5.309E-01

9.859E-07

unrestricted (power = 0.282)

log-logistic,
unrestricted b

3

0.655

307.507

7.049E-01

1.260E-05

unrestricted (slope = 0.374)

log-probit,
unrestricted

3

0.668

307.444

8.357E-01

4.796E-05

unrestricted (slope = 0.22)

Weibull,
unrestricted

3

0.644

307.562

6.143E-01

3.872E-06

unrestricted (power = 0.325)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.2.34.2. Output for Selected Model: Log-Logistic

National Toxicology Program, 2006: Gingival Hyperplasia, Squamous, 2 Years

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\42_NTP_2006_GingHypSq_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\Blood\42_NTP_2006_GingHypSq_LogLogistic_l.plt

Mon Feb 08 10:59:57 2010

[insert study notes]

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

This document is a draft for review purposes only and does not constitute Agency policy.

E-176 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Slope parameter is restricted as slope >= 1
Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values

background = 0.0188679
intercept =	-3.75308

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept

background	1	-0.79

intercept	-0.79	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf.	Limit Upper Conf. Limit

background 0.0671812 *	*	*

intercept -3.96371 *	*	*

slope 1 *	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-149.95	6

Fitted model -154.675 2	9.45085 4	0.05077

Reduced model -162.631 1	25.3627 5	0.0001186

AIC:	313.351

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.0672

3.561

1.

000

53

-1. 405

2.5565

0.1104

5. 960

7 .

000

54

0. 452

5.6937

0.1582

8 . 385

14 .

000

53

2 .113

9.7882

0.2134

11.311

13.

000

53

0.566

16.5688

0.2905

15.394

15.

000

53

-0.119

29.6953

0.4036

21.389

16.

000

53

-1.509

i^2 = 9.26

d.f.

= 4 P

-value

= 0.0550





Benchmark Dose Computation

This document is a draft for review purposes only and does not constitute Agency policy.

E-177 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Specified effect =	0.1

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	5.85026

BMDL =	3.7 2 96

E.2.34.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

0	5	10	15	20	25	30

dose

10:59 02/08 2010

E.2.34.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
National Toxicology Program, 2006: Gingival Hyperplasia, Squamous, 2 Years

t i i i i i i i i r

Log-Logistic

BMDL

BMD

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\42_NTP_2006_GingHypSq_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\42_NTP_2006_GingHypSq_LogLogistic_U_l.plt

Mon Feb 08 10:59:57 2010

[insert study notes]

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 78 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values

background = 0.0188679
intercept =	-2.2

slope =	0.424326

Asymptotic Correlation Matrix of	Parameter Estimates

background intercept	slope

background	1	-0.27	0.11

intercept	-0.27	1	-0.93

slope	0.11	-0.93	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.0185138	*	*	*

intercept	-2.06653	*	*	*

slope	0.373721	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-149.95	6

Fitted model -150.753 3	1.60697 3	0.657e

Reduced model -162.631 1	25.3627 5	0.000118^

AIC:	307.507

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0185 0.981 1.000	53 0.019

2.5565 0.1681 9.078 7.000	54	-0.756

5.6937 0.2101 11.136 14.000	53 0.966

9.7882 0.2433 12.893 13.000	53 0.034

This document is a draft for review purposes only and does not constitute Agency policy.

E-179 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

16.5688	0.27 92	14.7 95 15.000

29.6953	0.3230	17.117 16.000

53
53

0. 0 63
-0.328

Chi''2 = 1.62	d.f. = 3	P-value = 0.6554

E'.enchmark Dose Computation
Specified effect =	0 .1

Risk Typ'e	=	E::tra risk

Cc'nfidence level =	0.95

BMD =	0.7 04 8 98

EMDL = 1. 2 6 0 3 4 e - 0 0 5

E.2.34.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

BMD

15
dose

10:59 02/08 2010

"i i 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1 1 1 r

Log-Logistic

777/5 document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 80 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

E.2.35. National Toxicology Program, 2006: Hepatocyte Hypertrophy, 2 Years
E.2.35.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

5

0.034

273.875

9.091E-01

7.868E-01

power bound hit (power =1)

logistic

4

<0.001

297.895

2.475E+00

2.122E+00

negative intercept (intercept =
-1.685)

log-logistic

4

0.006

279.210

1.137E+00

6.491E-01



log-probit

5

0.006

277.800

1.530E+00

1.321E+00



multistage, 5-
degreea

4

0.018

275.693

9.272E-01

7.906E-01



probit

4

<0.001

299.731

2.453E+00

2.137E+00

negative intercept (intercept =
-0.985)

Weibull

5

0.034

273.875

9.091E-01

7.868E-01

power bound hit (power =1)

gamma,
unrestricted

4

0.027

275.270

error

error

unrestricted (power = 0.844)

log-probit,
unrestricted

4

0.008

278.360

1.191E+00

7.038E-01

unrestricted (slope = 0.864)

Weibull,
unrestricted

4

0.024

275.439

7.345E-01

3.588E-01

unrestricted (power = 0.92)

a Best-fitting model, BMDS output presented in this appendix

E.2.35.2. Output for Selected Model: Multistage, 5-Degree

National Toxicology Program, 2006: Hepatocyte Hypertrophy, 2 Years

Multistage Model. (Version: 3.0; Date: 05/16/2008)

Input Data File: C:\l\Blood\43_NTP_2006_HepHyper_Multi5_l.(d)

Gnuplot Plotting File: C:\l\Blood\43_NTP_2006_HepHyper_Multi5_l.plt

Mon Feb 08 11:00:25 2010

[insert study notes]

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/Nl-beta2*dose/N2-beta3*dose/N3-beta4*dose/N4-beta5*dose/N5) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 6

This document is a draft for review purposes only and does not constitute Agency policy.

E-181 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of records with missing values = 0
Total number of parameters in model = 6
Total number of specified parameters = 0
Degree of polynomial = 5

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background
Beta(1)
Beta(2)
Beta(3)
Beta ( 4 )
Beta(5)

0.112745

0.0950808
0
0
0

4.39515e-008

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(2)	-Beta(3)	-Beta(4)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)	Beta(5)

Beta(1)	1	-0.5

Beta(5)	-0.5	1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)
Beta(3)
Beta(4)
Beta(5)

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-129.986
-135.847
-219.97

Param's Deviance Test d.f.

P-value

11.7216
179.968

0. 01955
<.0001

AIC:

275.693

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000

0

0000

0. 000

0. 000

53

0

000

2.5565

0

2521

13.614

19.000

54

1

688

5.6937

0

4764

25.251

19.000

53

-1

719

9.7882

0

6717

35.599

42.000

53

1

872

16.5688

0

8510

45.106

41.000

53

-1

584

29.6953

0

9769

51. 778

52.000

53

0

203

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 82 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

Chi' 2 = 11.86	d.f. = 4	P-value = 0.0184

E'.enchmark Dose Computation

Specified effect	=	0 .1

Risk Typ'e	=	E::tra risk

Cc'nfidence level	=	0.95

BMD	=	0.92721

EMDL	=	0.7 90637

BMDLJ	=	1.14523

Taken together, (0.7 90637, 1.14523) is a 90	% two-sided confidence

interval for the EMD

E.2.35.3. Figure for Selected Model: Multistage, 5-Degree

Multistage Model with 0.95 Confidence Level

dose

11:00 02/08 2010

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 83 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

E.2.36. National Toxicology Program, 2006: Necrosis, Liver
E.2.36.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

4

0.939

234.400

8.655E+00

6.340E+00

power bound hit (power =1)

logistic

4

0.601

236.742

1.484E+01

1.240E+01

negative intercept (intercept
= -2.818)

log-logistic

4

0.943

234.382

7.928E+00

5.605E+00

slope bound hit (slope =1)

log-probit

4

0.572

236.863

1.333E+01

1.024E+01

slope bound hit (slope =1)

multistage, 5-
degree

4

0.939

234.400

8.655E+00

6.340E+00

final B = 0

probit

4

0.666

236.293

1.393E+01

1.154E+01

negative intercept (intercept
= -1.626)

Weibull

4

0.939

234.400

8.655E+00

6.340E+00

power bound hit (power =1)

gamma,
unrestricted

3

0.883

236.290

7.726E+00

3.453E+00

unrestricted (power = 0.87)

log-logistic,
unrestricted

3

0.860

236.377

7.733E+00

3.536E+00

unrestricted (slope = 0.974)

log-probit,
unrestricteda

3

0.805

236.598

7.501E+00

3.504E+00

unrestricted (slope = 0.517)

Weibull,
unrestricted

3

0.879

236.302

7.763E+00

3.508E+00

unrestricted (power = 0.895)

a Best-fitting model, BMDS output presented in this appendix

E.2.36.2. Output for Selected Model: Log-Probit, Unrestricted

National Toxicology Program, 2006: Necrosis, Liver

Probit Model. (Version: 3.1; Date: 05/16/2008)

Input Data File: C:\l\Blood\50_NTP_2006_LivNec_LogProbit_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\50_NTP_2006_LivNec_LogProbit_U_l.plt

Mon Feb 08 11:29:30 2010

NTP liver necrosis

The form of the probability function is:

P[response] = Background

+ (1-Background) * CumNorm(Intercept + Slope^Log(Dose) ) ,

where CumNorm(.) is the cumulative normal distribution function

Dependent variable = DichEff

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 84 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial	(and Specified) Parameter Values

background =	0.0188679

intercept =	-2.16223

slope =	0.457376

Asymptotic Correlation Matrix of	Parameter Estimates

background intercept	slope
background 1 -0.65 0.55

intercept	-0.65	1	-0.97

slope	0.55	-0.97	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.0221151	0.0221351	-0.0212689	0.065499

intercept	-2.32352	0.556343	-3.41393	-1.23311

slope	0.517104	0.185064	0.154385	0.879823

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-114.813	6

Fitted model -115.299 3	0.972184 3	0.808

Reduced model -127.98 1	26.3331 5	<.0001

AIC:	236.598

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0

0221

1.172

1.

000

53

-0.161

2.5565

0

0544

2 . 938

4 .

000

54

0. 637

5.6937

0

0976

5.174

4 .

000

53

-0.543

9.7882

0

1457

7 .720

8 .

000

53

0.109

16.5688

0

2096

11.106

10.

000

53

-0.373

29.6953

0

3002

15.908

17 .

000

53

0.327

i-2 =0.99



d.f.

= 3 P

-value

= 0.8048





Benchmark Dose Computation
Specified effect =	0.1

This document is a draft for review purposes only and does not constitute Agency policy.

E-185 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

Risk Type
Confidence level
BMD
BMDL

Extra risk
0. 95
7.50077
3.5039

E.2.36.3. Figure for Selected Model: Log-Probit, Unrestricted

LogProbit Model with 0.95 Confidence Level

0.5 —~—

0.4

"O

CD

-t—'

O
CD
!t=
<

0.3

0.2

0.1

10

15
dose

20

25

30

11:29 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 86 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

E.2.37. National Toxicology Program, 2006: Oval Cell Hyperplasia
E.2.37.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

3

0.074

199.468

6.739E+00

5.074E+00



logistic

4

0.171

196.803

6.064E+00

5.145E+00

negative intercept (intercept =
-3.834)

log-logistic

3

0.042

201.659

6.936E+00

5.604E+00



log-probit

3

0.072

200.121

7.090E+00

5.931E+00



multistage, 5-
degree

3

0.207

195.962

4.785E+00

3.105E+00



probita

4

0.227

195.448

5.673E+00

4.793E+00

negative intercept (intercept
= -2.19)

Weibullb

3

0.077

198.375

5.718E+00

4.088E+00



a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.2.37.2. Output for Selected Model: Probit

National Toxicology Program, 2006: Oval Cell Hyperplasia

Probit Model. (Version: 3.1; Date: 05/16/2008)

Input Data File: C:\l\Blood\53_NTP_2006_OvalHyper_Probit_l.(d)

Gnuplot Plotting File: C:\l\Blood\53_NTP_2006_OvalHyper_Probit_l.plt

Mon_Feb 08 13725:23 2010

The form of the probability function is:

P[response] = CumNorm(Intercept + Slope* Dose),

where CumNorm(.) is the cumulative normal distribution function

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-l 87 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Default Initial	(and Specified) Parameter Values

background =	0 Specified

intercept =	-2.29925

slope =	0.169545

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

intercept
slope

intercept
1

-0. 87

slope

-0. 87
1

Parameter Estimates

Variable
intercept
slope

Estimate
-2 .18988
0.172453

95.0% Wald Confidence Interval
Std. Err.	Lower Conf. Limit Upper Conf. Limit

0.208021	-2.5976	-1.78217

0.0182446	0.136694	0.208211

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table

Param's Deviance Test d.f.

Log(likelihood)
-92.4898
-95.7242
-210.191

195.448

6.46873
235.402

P-value

0.1668
C.0001

Goodness of Fit

Scaled

Dose

Est

. Prob.

Expected

Observed

Size

Residual

0.0000

0.

0143

0.756

0

000

53

-0

876

2.5565

0.

0401

2 .168

4

000

54

1

270

5.6937

0.

1135

6. 017

3

000

53

-1

306

9.7882

0.

3079

16.317

20

000

53

1

096

16.5688

0.

7478

39.631

38

000

53

-0

516

29.6953

0.

9983

52 911

53

000

53

0

299

Chi ^2

5. 64

d.f.

P-value

0.2274

Benchmark Dose Computation

Specified effect
Risk Type
Confidence level
BMD
BMDL

0.1
Extra risk

0. 95
5.67298
4.79341

This document is a draft for review purposes only and does not constitute Agency policy.

E-188 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.37.3. Figure for Selected Model: Probit

Probit Model with 0.95 Confidence Level

dose

13:25 02/08 2010

E.2.37.4. Output for Additional Model Presented: Weibull
National Toxicology Program, 2006: Oval Cell Hyperplasia

Weibull Model using Weibull Model (Version: 2.12; Date: 05/16/2008)
Input Data File: C:\l\Blood\53_NTP_2006_OvalHyper_Weibull_l.(d)

Gnuplot Plotting File: C:\l\Blood\53_NTP_2 006_OvalHyper_Weibull_l.plt

Mon Feb 08 13:25:23 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(-slope*doseApower)]

Dependent variable = DichEff

Independent variable = Dose

Power parameter is restricted as power >=1

Total number of observations = 6

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-l 89 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial	(and Specified) Parameter Values

Background =	0.00925926

Slope =	0.00296825

Power =	2.17092

Asymptotic Correlation Matrix of	Parameter Estimates

Background Slope	Power

Background 1 -0.72	0.7

Slope -0.72 1	-0.99

Power 0.7 -0.99	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0164137	0.0221488	-0.0269971	0.0598245

Slope	0.00162074	0.00202897	-0.00235596	0.00559745

Power	2.39427	0.455116	1.50226	3.28628

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-92.4898	6

Fitted model -96.1875 3	7.3953 3	0.06031

Reduced model -210.191 1	235.402 5	<.0001

AIC:	198.375

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.0164

0. 870

0

000

53

-0

940

2.5565

0.0314

1.695

4

000

54

1

799

5.6937

0.1138

6. 034

3

000

53

-1

312

9.7882

0.3285

17.411

20

000

53

0

757

16.5688

0.7440

39.431

38

000

53

-0

450

29.6953

0.9957

52 . 774

53

000

53

0

476

Chi/N2 = 6.85	d.f. = 3	P-value = 0.0770

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	5.71754

BMDL =	4.08 823

This document is a draft for review purposes only and does not constitute Agency policy.

E-190 DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.37.5. Figure for Additional Model Presented: Weibull

Weibull Model with 0.95 Confidence Level

dose

2	13:25 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-191 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

E.2.38. National Toxicology Program, 2006: Pigmentation, Liver
E.2.38.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

3

0.552

196.971

2.172E+00

1.493E+00



logistic

4

0.247

197.066

1.853E+00

1.521E+00

negative intercept (intercept =
-2.51)

log-logistic

3

0.984

195.530

2.566E+00

1.937E+00



log-probita

3

0.962

195.526

2.463E+00

1.890E+00



multistage, 5-
degree

3

0.058

199.955

1.822E+00

9.916E-01

final B = 0

probit

4

0.004

200.504

1.710E+00

1.430E+00

negative intercept (intercept =
-1.392)

Weibull

3

0.219

199.007

1.756E+00

1.190E+00



a Best-fitting model, BMDS output presented in this appendix

E.2.38.2. Output for Selected Model: Log-Probit

National Toxicology Program, 2006: Pigmentation, Liver

Probit Model. (Version: 3.1; Date: 05/16/2008)

Input Data File: C:\l\Blood\54_NTP_2006_Pigment_LogProbit_l.(d)

Gnuplot Plotting File: C:\l\Blood\54_NTP_2006_Pigment_LogProbit_l.plt

Mon Feb 08 13:25:55 2010

0

The form of the probability function is:

P[response] = Background

+ (1-Background) * CumNorm(Intercept + Slope^Log(Dose) ) ,

where CumNorm(.) is the cumulative normal distribution function

Dependent variable = DichEff
Independent variable = Dose

Slope parameter is restricted as slope >= 1
Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-192 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

User has chosen the log transformed model

Default Initial	(and Specified) Parameter Values

background =	0.0754717

intercept =	-2.48683

slope =	1.53221

Asymptotic Correlation Matrix of Parameter Estimates

background
intercept
slope

background
1

-0.42
0.33

intercept

-0.42
1

-0. 96

slope

0.33
-0. 96
1

Parameter Estimates

Variable
background
intercept
slope

Estimate

0.0725473
-2.93268
1. 83184

Std. Err.

0.0338856
0. 487158
0.246868

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

0.00613263
-3.8875
1.34798

0.138962
-1 . 97787
2 . 31569

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table

Param's Deviance Test d.f.

Log(likelihood)

-94.6177
-94.7632
-210.717

0.291072
232.198

P-value

0.9617
C.0001

195.526

Goodness of Fit

Est. Prob.

Expected

Observed

Scaled
Residual

0.0000 0.

0725

. :

4 .

000

53

0. 082

2.5565 0.

1769

9.553

9.

000

54

-0.197

5.6937 0.

6291

33.342

34 .

000

53

0.187

9.7882 0.

9013

47 .771

48 .

000

53

0.105

16.5688 0.

9874

52.334

52 .

000

53

-0.412

29.6953 0.

9 9 95

52.974

53.

000

53

0.160

Chi ^2 = 0.29

d.f.

= 3 P-

-value

= 0.9624





Benchmark Dose

Computation









Specified effect

=

0.1









Risk Type

=

Extra risk









Confidence level

=

0. 95









BMD

=

2.46293









BMDL



1.88981









This document is a draft for review purposes only and does not constitute Agency policy.

E-193 DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.38.3. Figure for Selected Model: Log-Probit

LogProbit Model with 0.95 Confidence Level

dose

2	13:25 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-194 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

E.2.39. National Toxicology Program, 2006: Toxic Hepatopathy
E.2.39.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

4

0.754

185.763

4.302E+00

3.463E+00



logistic

4

0.159

191.136

4.833E+00

4.068E+00

negative intercept (intercept =
-3.756)

log-logistic

3

0.391

189.577

4.697E+00

3.818E+00



log-probit

3

0.394

189.580

4.972E+00

3.780E+00



multistage, 5-
degreea

4

0.693

185.924

3.980E+00

3.059E+00

final 15 = 0

probit

4

0.231

189.820

4.621E+00

3.860E+00

negative intercept (intercept =
-2.172)

Weibull

4

0.716

185.785

4.089E+00

3.215E+00



a Best-fitting model, BMDS output presented in this appendix

E.2.39.2. Output for Selected Model: Multistage, 5-Degree

National Toxicology Program, 2006: Toxic Hepatopathy

Multistage Model. (Version: 3.0; Date: 05/16/2008)

Input Data File: C:\l\Blood\55_NTP_2006_ToxHepa_Multi5_l.(d)

Gnuplot Plotting File: C:\l\Blood\55_NTP_2006_ToxHepa_Multi5_l.plt

Mon_Feb 08 13:26:28 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/Nl-beta2*dose/N2-beta3*dose/N3-beta4*dose/N4-beta5*dose/N5) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 6
Total number of specified parameters = 0
Degree of polynomial = 5

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-195 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Parameter

has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) =	0

Beta(2) =	0

Beta(3) =	0

Beta(4) =	0

Beta(5) = 4.36963e+012

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l)	-Beta(4)	-Beta(5)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(2)	Beta(3)

Beta(2)	1	-0.95

Beta(3)	-0.95	1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)
Beta(3)
Beta(4)
Beta(5)

Estimate

0
0

0. 00639021
5.5404e-005
0
0

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-89.8076
-90.9619
-218.207

Param's Deviance Test d.f.

2.30853
256.799

P-value

0.6792
C.0001

AIC:

185.924

Goodness of Fit

Scaled

Dose

Est

. Prob.

Expected

Observed

Size

Residual

0.0000

0.

0000

0. 000

0

000

53

0

000

2.5565

0.

0420

2 .265

2

000

54

-0

180

5.6937

0.

1969

10.434

8

000

53

-0

841

9.7882

0.

4 901

25.976

30

000

53

1

106

16.5688

0.

8715

46.189

45

000

53

-0

488

29.6953

0.

9994

52 966

53

000

53

0

185

Chi ^2

2 . 23

d.f.

P-value

0.6928

Benchmark Dose Computation
Specified effect =	0.1

This document is a draft for review purposes only and does not constitute Agency policy.

E-196 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

Risk Type

Extra risk

Confidence level =	0.95

BMD =	3.98025

BMDL =	3.05855

BMDU =	4.897 35

Taken together, (3.05855,	4.89735) is a 90	% two-sided confidence

interval for the BMD

E.2.39.3. Figure for Selected Model: Multistage, 5-Degree

Multistage Model with 0.95 Confidence Level

dose

13:26 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-197 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.2.40. Ohsako et al., 2001: Ano-Genital Length, PND 120
E.2.40.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

3

0.027

171.073

2.592E+01

1.750E+01



exponential (M3)

3

0.027

171.073

2.592E+01

1.750E+01

power hit bound (d = 1)

exponential (M4)

2

0.106

168.392

2.248E+00

8.445E-01



exponential (M5)

1

0.049

169.789

2.193E+00

9.382E-01



Hill b

2

0.154

167.647

2.879E+00

8.028E-01

n lower bound hit (n = 1)

linear

3

0.025

171.258

2.700E+01

1.881E+01



polynomial, 4-
degree

3

0.025

171.258

2.700E+01

1.881E+01



power

3

0.025

171.258

2.700E+01

1.881E+01

power bound hit (power =1)

Hill, unrestricted0

1

0.056

169.555

3.494E+00

3.046E-01

unrestricted (n = 0.591)

power,
unrestricted

2

0.153

167.654

4.151E+00

2.395E-01

unrestricted (power = 0.291)

a Constant variance model selected (p = 0.165)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.40.2. Output for Selected Model: Hill

Ohsako et al., 2001: Ano-Genital Length, PND 120

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\56_Ohsako_2001_Anogen_HillCV_l.(d)

Gnuplot Plotting File: C:\l\Blood\56_Ohsako_2001_Anogen_HillCV_l.plt

Mon Feb 08 13727:02 2010

Figure 7

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

This document is a draft for review purposes only and does not constitute Agency policy.

E-198 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Power parameter restricted to be greater than 1
A constant variance model is fit

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho

intercept

7 . 27386

0

28.905
-5.1065
1.57046
2.4317

Specified

Asymptotic Correlation Matrix of Parameter Estimates

alpha
intercept

The model parameter(s) -rho -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

4 . 4e-008
-9.8e-008
7.2e-008

intercept
4.4e-008
1

-0.57
-0.52

v

-9.8e-008
-0.57
1

-0.23

k

7 . 2e-008
-0.52
-0.23
1

Parameter Estimates

Variable
alpha
intercept

k

Estimate

7.07394
28.9732
-5.02686
1

2 . 56203

Std

Err.

1. 36138
0.74996
1.05086
NA

2 .11462

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

4.40568

27 . 5034
-7 . 08651

-1. 58255

9.7422
30.4431
-2.9672

6.70661

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	12	28.9	29	3.13	2.66	-0.0889

1.04	10	27.9	27.5	2.5	2.66	0.495

3.471	10	25.2	26.1	3.21	2.66	-1.09

11.36	10	26	24.9	2.85	2.66	1.35

38.42	12	23.8	24.3	1.56	2.66	-0.602

Model Descriptions for likelihoods calculated

This document is a draft for review purposes only and does not constitute Agency policy.

E-199 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)

-77 . 952340
-74.703868
-77 . 952340
-7 9.823277
-89.824703

Param's

10

AIC

167.904680
169.407736
167.904680
167.646555
183.649405

Explanation of Tests

Test 1:

Test 2
Test 3
Test 4
(Note:

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

30.2417

6 . 4 9 6 9 4
6 . 4 9 6 9 4
3.74187

0. 0001916
0.165
0.165
0.154

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than
model appears to be appropriate here

.1. A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	2.87863

BMDL =	0.802782

This document is a draft for review purposes only and does not constitute Agency policy.

E-200 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.40.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

13:27 02/08 2010

E.2.40.4. Output for Additional Model Presented: Hill, Unrestricted
Ohsako et al., 2001: Ano-Genital Length, PND 120

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\56_Ohsako_2001_Anogen_HillCV_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\56_Ohsako_2 001_Anogen_HillCV_U_l.plt

Mon Feb 08 13:27:04 2010

Figure 7

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-201 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha =	7.27386

rho =	0 Specified

intercept =	28.905

v =	-5.1065

n =	1. 57046

k =	2 .4317

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
intercept

alpha
1

-3.le-008
7.5e-009
1. 7e-008
-8 . 8e-009

intercept
-3.le-008
1

0. 001
0.0016
-0.13

v

7.5e-009
0. 001
1

0. 98
-0. 99

n

1. 7e-008
0. 0016
0. 98
1

-0. 97

k

. 8e-009
-0.13
-0. 99
-0. 97
1

Parameter Estimates

Variable
alpha
intercept

Estimate

7 . 06192
28 . 9618
-6.82284
0.591421
7 .47064

Std. Err.

I.	35907
0.754441

II.1104
1. 04

48.002

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

4.3982
27.4831
-28.5989
-1.44695
-86.6115

9.72564

30.4404
14.9532
2.62979
101.553

Table of Data and Estimated Values of Interest
5	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	12	28.9	29	3.13	2.66	-0.074

1.04	10	27.9	27.3	2.5	2.66	0.71

3.471	10	25.2	26.3	3.21	2.66	-1.36

11.36	10	26	25.1	2.85	2.66	1.04

38.42	12	23.8	24	1.56	2.66	-0.284

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij|

This document is a draft for review purposes only and does not constitute Agency policy.

E-202 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -77.952340	6	167.904680

A2 -74.703868	10	169.407736

A3 -77.952340	6	167.904680

fitted -79.777354	5	169.554709

R -89.824703	2	183.649405

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

30.2417
6.4 9 6 9 4
6.4 9 6 9 4
3.65003

0.0001916
0.165
0.165
0.05607

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	3.49389

BMDL =	0.304602

This document is a draft for review purposes only and does not constitute Agency policy.

E-203 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.40.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

13:27 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-204 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

E.2.41. Sewall et al., 1995: T4 In Serum
E.2.41.1. Summary Table of BMDS Modeling Results

Model3

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

3

0.722

204.495

1.869E+01

1.243E+01



exponential (M3)

3

0.722

204.495

1.869E+01

1.243E+01

power hit bound (d = 1)

exponential (M4)

2

0.854

205.483

1.106E+01

4.650E+00



exponential (M5)

2

0.854

205.483

1.106E+01

4.650E+00

power hit bound (d = 1)

Hillb

2

0.898

205.382

1.031E+01

3.603E+00

n lower bound hit (n = 1)

linear

3

0.576

205.150

2.238E+01

1.619E+01



polynomial, 4-
degree

3

0.576

205.150

2.238E+01

1.619E+01



power

3

0.576

205.150

2.238E+01

1.619E+01

power bound hit (power =1)

Hill, unrestricted0

1

0.864

207.196

9.706E+00

1.973E+00

unrestricted (n = 0.569)

power, unrestricted

2

0.985

205.197

9.726E+00

1.914E+00

unrestricted (power = 0.538)

a Constant variance model selected (p = 0.4078)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.41.2. Output for Selected Model: Hill

Sewall et al., 1995: T4 In Serum

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\58_Sewall_l995_T4_HillCV_l.(d)

Gnuplot Plotting File: C:\l\Blood\58_Sewall_1995_T4_HillCV_l.plt

Mon Feb 08 13:28:15 2010

Figure 1, Saline noninitiated

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter restricted to be greater than 1

This document is a draft for review purposes only and does not constitute Agency policy.

E-205 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

A constant variance model is fit

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho

intercept

33.0913

0

30.6979
-12 .2937
0.950815
12 . 5808

Specified

Asymptotic Correlation Matrix of Parameter Estimates

alpha
intercept

The model parameter(s) -rho -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

-1. 2e-009
-1. 8e-008
1. 5e-008

intercept
-1.2e-009
1

0.3
-0. 65

v

-1.8e-008
0.3
1

-0.89

k

1. 5e-008
-0. 65
-0.89
1

Parameter Estimates

Variable
alpha
intercept

k

Estimate
29.5556
30.3957
-18.2488
1

24.2883

Std. Err.
6.23087
1.68747
7.72836
NA
26.743

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

17.3433

27 . 0883
-33.3961

-28 .127

41.7679
33.7031
-3.10154

76.7035

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

9

30

7

o

CO

4

6 6

5.44

0.167

3.291

9

27

9

28 . 2

7

17

5.44

-0.188

7 .107

9

25

9

26.3

6

81

5.44

-0.204

16. 63

9

23

6

23

5

38

5.44

0.319

44.66

9

18

4

18 . 6

4

12

5.44

-0.0942

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

This document is a draft for review purposes only and does not constitute Agency policy.

E-206 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

Var{e(ij)} = Sigma^2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model
A1
A2
A3

fitted
R

Log(likelihood)
-98.583448
-96.590204
-98.583448
-98.691143
-109.013252

Param1s

10

AIC
209.166896
213.180407
209.166896
205.382286
222.026503

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

24.8461
3.98649
3.98649
0.21539

0.001651
0.4078
0.4078
0. 897 9

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	10.306

BMDL =	3.60269

This document is a draft for review purposes only and does not constitute Agency policy.

E-207 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.41.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

13:28 02/08 2010

E.2.41.4. Output for Additional Model Presented: Hill, Unrestricted
Sewall et al., 1995: T4 In Serum

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\58_Sewall_l995_T4_HillCV_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\58_Sewall_1995_T4_HillCV_U_l.plt

Mon Feb 08 13:28:15 2010

Figure 1, Saline noninitiated

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-208 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho

intercept

33.0913

0

30.6979
-12 .2937
0.950815
12 . 5808

Specified

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
intercept

alpha
1

-3.9e-005
0.00022
0.00021
-0.00022

intercept
-3.9e-005
1

-0.17
-0.31
0.18

v

0.00022
-0.17
1

0. 97
-1

n

0.00021
-0.31
0. 97
1

-0. 98

k

-0.00022
0.18
-1

-0. 98
1

Parameter Estimates

Variable
alpha
intercept

Estimate

29.4337
30.7096
-143.244
0.569063
2856.29

Std. Err.
6.20518
1.79801
3972.28
0.947248
171186

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

17 . 2718

27 .1855
-7928.78
-1.28751
-332662

41. 5957

34 . 2336
7642.29
2 .42564
338374

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

9

30

7

30

7

4

6 6

5.43

-0.00646

291

9

27

9

27

7

7

17

5.43

0.0842

107

9

25

9

26

1

6

81

5.43

-0.134

. 63

9

23

6

23

4

5

38

5.43

0.0657

. 6 6

9

18

4

18

4

4

12

5.43

-0.00948

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

This document is a draft for review purposes only and does not constitute Agency policy.

E-209 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model
A1
A2
A3

fitted
R

Log(likelihood)
-98.583448
-96.590204
-98.583448
-98.598183
-109.013252

10

AIC
209.166896
213.180407
209.166896
207.196367
222.026503

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

24.8461
3.98649
3.98649
0.0294713

0.001651
0.4078
0.4078
0.8637

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	9.70574

BMDL =	1.97319

This document is a draft for review purposes only and does not constitute Agency policy.

E-210 DRAFT—DO NOT CITE OR QUOTE

-------
l E.2.41.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

2	13:28 02/08 2010

3

4

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-211 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.42. Shi et al., 2007: Estradiol 17B, PE9
E.2.42.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

3

0.010

391.638

6.976E+00

3.761E+00



exponential (M3)

3

0.010

391.638

6.976E+00

3.761E+00

power hit bound (d = 1)

exponential
(M4)b

2

0.690

382.969

8.068E-01

3.544E-01



exponential (M5)

2

0.690

382.969

8.068E-01

3.544E-01

power hit bound (d = 1)

Hill

2

0.975

382.278

7.239E-01

error

n lower bound hit (n = 1)

linear

3

0.003

394.308

9.841E+00

6.687E+00



polynomial, 4-
degree

3

0.003

394.308

9.841E+00

6.687E+00



power

3

0.003

394.308

9.841E+00

6.687E+00

power bound hit (power = 1)

Hill, unrestricted

1

0.897

384.243

7.086E-01

error

unrestricted (n = 0.875)

power,
unrestricted

2

0.506

383.590

6.280E-01

3.304E-02

unrestricted (power = 0.222)

a Non-constant variance model selected (p = 0.0521)
b Best-fitting model, BMDS output presented in this appendix

E.2.42.2. Output for Selected Model: Exponential (M4)

Shi et al., 2007: Estradiol 17B, PE9

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\59_Shi_2007_Estradiol_Exp_l.(d)

Gnuplot Plotting File:

Mon Feb 08 13:28:52 2010

Figure 4 PE9 only

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

This document is a draft for review purposes only and does not constitute Agency policy.

E-212 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	2.65881

rho	0.913414

a	108

b	0.277637

c	0.340136

d	1

Parameter Estimates
Variable	Model 4

lnalpha	1.66773

rho	1.15314

a	103.146

b	1.00685

c	0.418742

d	1

Table of Stats From Input Data
Dose	N	Obs Mean	Obs Std Dev

0	10	102.9	41.41

0.3418	10 86.19	19.58

1.075	10	63.33	29.36

5.23	10 48.1	18.82

13.91	10	38.57	22.59

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	103.1	33.35	-0.02738

0.3418	85.69	29.96	0.05296

1.075	63.51	25.21	-0.02238

5.23	43.5	20.27	0.7167

13.91	43.19	20.19	-0.7237

Other models for which likelihoods are calculated:

This document is a draft for review purposes only and does not constitute Agency policy.

E-213 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma^2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-188.3615	6	388.7231

A2	-183.667	10	387.3339

A3	-186.1132	7	386.2263

R	-203.3606	2	410.7211

4	-186.4844	5	382.9687

Additive constant for all log-likelihoods =	-45.95. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Does response

and/or variances differ among Dose levels? (A2 vs. R)

Test

2 :

Are Variances

Homogeneous? (A2 vs. A1)

Test

3:

Are variances

adequately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

39.39
9.389
4 . 892
0.7424

D. F.

p-value

< 0.0001
0.05208
0.1798
0.6899

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

This document is a draft for review purposes only and does not constitute Agency policy.

E-214 DRAFT—DO NOT CITE OR QUOTE

-------
BMD =	0.806817

BMDL =	0.354366

E.2.42.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

a>

CO
c
o

Q.

CO
CD

a:

c
m

CD

140

120

100

80

60

40

20

Exponential

10

12

14

dose

13:28 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-215 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.43. Smialowicz et al., 2008: PFC per 10A6 Cells
E.2.43.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

3

0.101

901.897

8.343E+00

5.064E+00



exponential (M3)

3

0.101

901.897

8.343E+00

5.064E+00

power hit bound (d = 1)

exponential (M4)

2

0.044

903.897

8.325E+00

1.465E+00



exponential (M5)

2

0.044

903.897

8.325E+00

1.465E+00

power hit bound (d = 1)

Hill

2

0.063

903.192

3.669E+00

6.970E-01

n lower bound hit (n = 1)

linear

3

0.048

903.585

1.373E+01

1.053E+01



polynomial, 4-
degree

3

0.048

903.585

1.374E+01

1.053E+01



power

3

0.048

903.585

1.373E+01

1.053E+01

power bound hit (power =1)

Hill, unrestricted

1

0.213

901.219

1.928E+00

2.208E-01

unrestricted (n = 0.35)

power,
unrestricted b

2

0.481

899.130

1.902E+00

2.158E-01

unrestricted (power = 0.333)

a Constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix

E.2.43.2. Output for Selected Model: Power, Unrestricted

Smialowicz et al., 2008: PFC per 10A6 Cells

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\60_Smial_2008_PFCcells_PwrCV_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\60_Smial_2008_PFCcells_PwrCV_U_l.plt

Mon Feb 08 13:29:38 2010

Anti Response to SRBCs, PFC per 10to6 cells, Table 4

The form of the response function is:

Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

This document is a draft for review purposes only and does not constitute Agency policy.

E-216 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho
control
slope
power

232385

0

1491
-491.716
0.288021

Specified

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha	control	slope	power

alpha	1 -3.4e-009	1.8e-009 -1.2e-010

control -3.4e-009	1	-0.82	-0.65

slope	1.8e-009	-0.82	1	0.94

power -1.2e-010	-0.65	0.94	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
control
slope
power

Estimate

219793
1470.48
-378.406
0.333124

Std. Err.
37974.5
123.73
157.002
0.113501

Lower Conf. Limit

145365
1227.98
-686.125
0.110666

Upper Conf. Limit

294222
1712.99
-70.6872
0.555581

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

0. 438
2.464
13. 4
31. 65

15

1. 4 9e + 003

1. 47e + 003

716

469

0.169

14

1.13e + 003

1.18e + 003

171

469

-0.431

15

945

959

516

469

-0.12

15

677

572

465

469

0. 867

8

161

274

117

469

-0.684

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-217 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -444.832859	6	901.665718

A2 -425.402825	10	870.805651

A3 -444.832859	6	901.665718

fitted -445.564823	4	899.129647

R -463.753685	2	931.507371

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	76.7017	8	<.0001

Test 2	38.8601	4	<.0001

Test 3	38.8601	4	<.0001

Test 4	1.46393	2	0.481

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. Consider running a
non-homogeneous variance model

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 1.9024 9

BMDL = 0.215843

This document is a draft for review purposes only and does not constitute Agency policy.

E-218 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.43.3. Figure for Selected Model: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

13:29 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-219 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.2.44. Smialowicz et al., 2008: PFC per Spleen

E.2.44.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

3

0.124

377.565

1.334E+01

8.593E+00



exponential (M3)

2

0.069

379.138

1.536E+01

8.895E+00



exponential (M4)

3

0.124

377.565

1.334E+01

8.593E+00



exponential (M5)

1

0.021

381.138

1.536E+01

8.895E+00



Hill

2

0.116

378.108

1.568E+01

error

n lower bound hit (n = 1)

linear

3

0.126

377.522

2.055E+01

1.624E+01



polynomial, 4-
degree

3

0.126

377.522

2.055E+01

1.624E+01



power

3

0.126

377.522

2.055E+01

1.624E+01

power bound hit (power =1)

Hill, unrestricted

1

0.103

378.463

1.202E+01

error

unrestricted (n = 0.544)

power,
unrestricted b

2

0.270

376.420

1.187E+01

3.762E+00

unrestricted (power = 0.531)

a Non-constant variance model selected (p = 0.0011)
b Best-fitting model, BMDS output presented in this appendix

E.2.44.2. Output for Selected Model: Power, Unrestricted

Smialowicz et al., 2008: PFC per Spleen

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\61_Smial_2008_PFCspleen_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\61_Smial_2008_PFCspleen_Pwr_U_l.plt

Mon Feb 0 8 13:3 0:16~2 010

Anti Response to SRBCs - PFC x 10 to the 4 per spleen, Table 4

The form of the response function is:

Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

This document is a draft for review purposes only and does not constitute Agency policy.

E-220 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	4.76607

rho =	0

control =	27.8

slope =	-9.21898

power =	0.286443

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope
power

lalpha
1

-0. 98
0.25
-0.28
-0.22

rho

-0. 98
1

-0.3

0.28
0.22

control
0.25
-0.3

1

-0. 83
-0.74

slope

-0.28
0.28
-0. 83
1

0. 99

power

-0.22
0.22
-0.74
0. 99
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
control
slope
power

Estimate
0.746922
1. 36826
25.3816
-3.5662
0.531216

Std. Err.

1.02058
0.355827
2.96691
2.52558
0.175728

Lower Conf. Limit

-1.25337
0. 67085
19.5666
-8 . 51626
0.186796

Upper Conf. Limit
2 .74721
2.06567
31.1967
1. 38385
0. 875637

Table of Data and Estimated Values of Interest

0

0. 438
2.464
13. 4
31. 65

N

Obs Mean

Est Mean

Obs Std Dev

Est Std Dev

Scaled Res

15

27 . 8

25. 4

13

4

13.3

0.706

14

21

23.1

13

6

12 . 4

-0.626

15

17 . 6

19.6

9

4

11.1

-0.704

15

12 . 6

11 2

8

7

7 . 6

0.702

8

3

CO

o

CO

3

1

3.1

-0.0313

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

This document is a draft for review purposes only and does not constitute Agency policy.

E-221 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

Model R:	Yi = Mu + e(i

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1 -190.565019	6	393.130038

A2 -181.476284	10	382.952569

A3 -181.900030	7	377.800059

fitted -183.210137	5	376.420274

R -204.636496	2	413.272993

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	46.3204	8	<.0001

Test 2	18.1775	4	0.001139

Test 3	0.84749	3	0.8381

Test 4	2.62021	2	0.2698

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 11.8748

BMDL = 3.76161

This document is a draft for review purposes only and does not constitute Agency policy.

E-222 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.44.3. Figure for Selected Model: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

13:30 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-223 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.2.45. Toth et al., 1979: Amyloidosis

E.2.45.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

2

0.040

149.120

1.965E+01

1.283E+01

power bound hit (power = 1)

logistic

2

0.019

151.340

3.701E+01

2.858E+01

negative intercept (intercept =
-2.16)

log-logistica

2

0.053

148.269

1.503E+01

8.747E+00

slope bound hit (slope = 1)

log-probit

2

0.009

152.855

3.782E+01

2.502E+01

slope bound hit (slope =1)

multistage, 3-
degree

2

0.040

149.120

1.965E+01

1.283E+01

final B = 0

probit

2

0.021

151.115

3.467E+01

2.657E+01

negative intercept (intercept =
-1.276)

Weibull

2

0.040

149.120

1.965E+01

1.283E+01

power bound hit (power = 1)

gamma,
unrestricted

2

0.959

140.119

4.349E-01

2.891E-03

unrestricted (power = 0.254)

log-logistic,
unrestricted b

2

0.903

140.240

4.843E-01

5.312E-03

unrestricted (slope = 0.326)

log-probit,
unrestricted

2

0.870

140.315

4.960E-01

7.292E-03

unrestricted (slope = 0.186)

Weibull,
unrestricted

2

0.933

140.174

4.641E-01

4.069E-03

unrestricted (power = 0.289)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.2.45.2. Output for Selected Model: Log-Logistic

Toth et al., 1979: Amyloidosis

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\62_Toth_l979_Amylyr_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\Blood\62_Toth_197 9_Amylyr_LogLogistic_l.plt

Mon Feb 08 13:30:54 2010

Table 2

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

This document is a draft for review purposes only and does not constitute Agency policy.

E-224 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Slope parameter is restricted as slope >= 1

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background =	0

intercept =	-4.54593

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept

background	1	-0.49

intercept	-0.49	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf.	Limit Upper Conf. Limit

background 0.0699918 *	*	*

intercept -4.90704 *	*	*

slope 1 *	*	*

Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-68.017	4

Fitted model	-72.1346	2	8.23525	2	0.0162e

Reduced model	-82.0119	1	27.99	3	<.0001

AIC:	148.269

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0700 2.660 0.000	38	-1.691

0.5732 0.0739 3.252 5.000	44	1.007

14.2123 0.1584 6.971 10.000	44	1.251

91.2070 0.4446 19.117 17.000	43	-0.650

Chi/N2 = 5.86	d.f. = 2	P-value = 0.0534

Benchmark Dose Computation
Specified effect =	0.1

This document is a draft for review purposes only and does not constitute Agency policy.

E-225 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Risk Type	=

Confidence level =

Extra risk
0. 95

BMD =	15.0264

BMDL =	8.7 4 665

E.2.45.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

dose

13:30 02/08 2010

E.2.45.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
Toth et al., 1979: Amyloidosis

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\62_Toth_l979_Amylyr_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\62_Toth_197 9_Amylyr_LogLogistic_U_l.plt

Mon Feb 08 13:30:54 2010

Table 2

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-226 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background =	0

intercept =	-1.92722

slope =	0.314472

Asymptotic Correlation Matrix of Parameter Estimates

The model parameter(s) -background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

intercept	slope

intercept	1	-0.84

slope	-0.84	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0	*	*	*

intercept	-1.96073	*	*	*

slope	0.326156	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-68.017	4

Fitted model	-68.1201	2	0.206341	2	0.902

Reduced model	-82.0119	1	27.99	3	<.0001

AIC:	140.24

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0000 0.000 0.000	38	0.000

0.5732 0.1051 4.623 5.000	44	0.186

14.2123 0.2507 11.029 10.000	44	-0.358

91.2070 0.3802 16.348 17.000	43	0.205

This document is a draft for review purposes only and does not constitute Agency policy.

E-227 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

Chi' 2 = 0.20	d.f. = 2

P-value = 0.9028

E'.enchmark Dose Computation
Specified effect =	0 .1

Risk Typ'e	=	E::tra risk

Cc'nfidence level =	0.95

BMD =	0.4 84272

EMDL =	0.00531211

E.2.45.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

"O

(D

-t—<

o

(U
!t=
<

20

40

60

80

dose

13:30 02/08 2010

This document is a draft for review purposes only and does not constitute Agency policy.

E-228 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.2.46. Toth et al., 1979: Skin Lesions

E.2.46.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

gamma

2

0.032

156.346

1.037E+01

7.470E+00

power bound hit (power = 1)

logistic

2

0.005

161.421

2.487E+01

1.982E+01

negative intercept (intercept =
-1.999)

log-logistica

2

0.078

153.963

6.413E+00

4.025E+00

slope bound hit (slope = 1)

log-probit

2

0.003

161.788

1.887E+01

1.280E+01

slope bound hit (slope =1)

multistage, 3-
degree

2

0.032

156.346

1.037E+01

7.470E+00

final B = 0

probit

2

0.006

160.991

2.309E+01

1.858E+01

negative intercept (intercept =
-1.198)

Weibull

2

0.032

156.346

1.037E+01

7.470E+00

power bound hit (power = 1)

gamma,
unrestricted

2

0.945

147.148

error

error

unrestricted (power = 0.341)

log-logistic,
unrestricted b

2

0.744

147.631

5.969E-01

6.773E-02

unrestricted (slope = 0.48)

log-probit,
unrestricted

2

0.670

147.844

5.939E-01

8.147E-02

unrestricted (slope = 0.279)

Weibull,
unrestricted

2

0.866

147.324

5.539E-01

5.181E-02

unrestricted (power = 0.405)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.2.46.2. Output for Selected Model: Log-Logistic

Toth et al., 1979: Skin Lesions

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\63_Toth_l979_SkinLes_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\Blood\63_Toth_197 9_SkinLes_LogLogistic_l.plt

~~Wed Feb 10 14:47:53 2 010~

Table 2

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

This document is a draft for review purposes only and does not constitute Agency policy.

E-229 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Slope parameter is restricted as slope >= 1

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background =	0

intercept =	-3.94312

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept

background	1	-0.43

intercept	-0.43	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf.	Limit Upper Conf. Limit

background 0.0564562 *	*	*

intercept -4.05558 *	*	*

slope 1 *	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-71.5177	4

Fitted model -74.9813 2	6.92722 2	0.03132

Reduced model -95.8498 1	48.6642 3	<.0001

AIC:	153.963

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0565 2.145 0.000	38	-1.508

0.5732 0.0657 2.892 5.000	44	1.282

14.2123 0.2429 10.687 13.000	44	0.813

91.2070 0.6343 27.275 25.000	43	-0.720

Chi ^2 = 5.10	d.f. = 2	P-value = 0.0782

Benchmark Dose Computation
Specified effect =	0.1

This document is a draft for review purposes only and does not constitute Agency policy.

E-230 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Risk Type	=

Confidence level =

Extra risk
0. 95

BMD =	6.4132

BMDL =	4.02 4 9

E.2.46.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

dose

14:47 02/10 2010

E.2.46.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
Toth et al., 1979: Skin Lesions

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\l\Blood\63_Toth_l979_SkinLes_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\63_Toth_197 9_SkinLes_LogLogistic_U_l.plt

Wed Feb 10 14:47:54 2010

Table 2

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-231 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background =	0

intercept =	-1.87608

slope =	0.458888

Asymptotic Correlation Matrix of Parameter Estimates

The model parameter(s) -background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

intercept	slope

intercept	1	-0.86

slope	-0.86	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0	*	*	*

intercept	-1.94946	*	*	*

slope	0.4802	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-71.5177	4

Fitted model -71.8153 2	0.59526 2	0.742£

Reduced model -95.8498 1	48.6642 3	<.0001

AIC:	147.631

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0000 0.000 0.000	38 0.000

0.5732 0.0983 4.323 5.000	44 0.343

14.2123 0.3374 14.845 13.000	44	-0.588

91.2070 0.5542 23.832 25.000	43 0.358

This document is a draft for review purposes only and does not constitute Agency policy.

E-232 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

Chi' 2 = 0.59	d.f. = 2

E'-value =

0.7438

E'.enchmark Dose Computation
Specified effect =	0 .1

Risk Typ'e	=	E::tra risk

Cc'nfidence level =	0.95

BMD =	0.596932

BMDL =	0.06773

E.2.46.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

20	40	60	80

dose

14:47 02/10 2010

T 1 '' ' • • • ' ' I • • • ' ' • • • • I ' ' • • • ' ' • • I 1 1 ' ' • • • ' ' I ' • • • ' ^

Log-Logistic 	

T

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-233 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.2.47. Van Birgelen et al., 1995a: Hepatic Retinol
E.2.47.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

4

<0.0001

159.735

7.790E+00

4.150E+00



exponential (M3)

4

<0.0001

3222.700

5.542E+01

error

power hit bound (d = 1)

exponential
(M4)b

3

<0.001

141.454

2.488E+01

3.363E+00



exponential (M5)

3

<0.001

141.454

2.488E+01

3.363E+00

power hit bound (d = 1)

Hill

3

0.239

124.865

5.316E+00

error

n lower bound hit (n = 1)

linear

4

<0.0001

176.828

1.877E+02

1.437E+02



polynomial, 5-
degree

4

<0.0001

176.828

1.877E+02

1.437E+02



power

4

<0.0001

176.828

1.877E+02

1.437E+02

power bound hit (power =1)

Hill, unrestricted

2

0.241

125.495

3.595E+00

error

unrestricted (n = 0.763)

power,
unrestricted0

3

0.011

131.771

3.802E-01

1.393E-02

unrestricted (power = 0.14)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.47.2. Output for Selected Model: Exponential (M4)

Van Birgelen et al., 1995a: Hepatic Retinol

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\65_VanB_l995a_HepRet_Exp_l.(d)

Gnuplot Plotting File:

Mon Feb 08 13:32:00 2010

Tbl3, hepatic retinol

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;

This document is a draft for review purposes only and does not constitute Agency policy.

E-234 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	-1.16065

rho	1.53688

a	15.645

b	0.0254351

c	0.0365247

d	1

Parameter Estimates
Variable	Model 4

lnalpha	-0.92683

rho	1.77262

a	11.5049

b	0.0286598

c	0.0653043

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	8	14.9	8.768

7.204	8	8.4	3.394

11.76	8	8.2	2.263

18.09	8	5.1	0.8485

86.41	8	2.2	0.8485

250.2	8	0.6	0.5657

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	11.5	5.483	1.751

7.204	9.499	4.627	-0.6719

11.76	8.428	4.161	-0.1552

18.09	7.154	3.599	-1.615

86.41	1.655	0.9832	1.568

250.2	0.7596	0.4931	-0.9155

This document is a draft for review purposes only and does not constitute Agency policy.

E-23 5 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Other models for which	likelihoods are calculated:

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-87.1567	7	188.3134

A2	-47.28742	12	118.5748

A3	-55.32422	8	126.6484

R	-109.967	2	223.934

4	-65.72714	5	141.4543

Additive constant for all log-likelihoods =	-44.11. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Does response

and/or variances differ among Dose levels? (A2 vs. R)

Test

2 :

Are Variances

Homogeneous? (A2 vs. A1)

Test

3:

Are variances

adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

125. 4
79.74
16. 07
20. 81

10
5
4
3

p-value

<	0.0001

<	0.0001
0. 002922

0.0001155

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is less than .1. You may want to
consider a different variance model.

The p-value for Test 6a is less than .1. Model 4 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

This document is a draft for review purposes only and does not constitute Agency policy.

E-236 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	24.8811

BMDL =	3.36281

E.2.47.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

13:32 02/08 2010

E.2.47.4. Output for Additional Model Presented: Power, Unrestricted
Van Birgelen et al., 1995a: Hepatic Retinol

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\65_VanB_l995a_HepRet_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\65_VanB_1995a_HepRet_Pwr_U_l.plt

Mon Feb 08 13:32:03 2010

Tbl3, hepatic retinol

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-23 7 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean

Independent variable = Dose

The power is not restricted

The variance is to be modeled as Var(i)

exp(lalpha + log(mean(i))

rho)

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	2.76506

rho =	0

control =	14.9

slope =	-3.98831

power =	0.231232

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope
power

lalpha
1

-0.8

-0.042
0. 038
0. 063

rho
-0.8

1

-0.089
0.0044
-0.1

control

-0.042
-0.089
1

-0. 95
-0. 81

slope

0. 038
0.0044
-0. 95
1

0. 95

power
0. 063
-0.1

-0. 81
0. 95
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
control
slope
power

Estimate

-0.986251
1.67858
16.9266
-7.51118
0.139871

Std. Err.
0.394722
0.202896
2.23237
2.04379
0.0269576

Lower Conf. Limit
-1.75989
1.28091
12 . 5513
-11. 5169
0. 0870351

Upper Conf. Limit
-0.212609
2.07625
21.302
-3.50543
0.192707

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

7 .204
11.76
18.09
86.41
250.2

14 . 9

8 . 4
8 . 2
5.1
2 . 2
0.6

16. 9

7 . 03
6.32
5. 67
2 . 91
0.666

8 .77

3.39
2 .26
0.849
0.849
0.566

6.56

3.14
2 . 87
2 . 62
1. 5
0. 434

-0.874
1.24
1. 85
-0.611
-1. 34
-0.427

Model Descriptions for likelihoods calculated

This document is a draft for review purposes only and does not constitute Agency policy.

E-23 8 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)

-87 .156698
-47.287416
-55.324218
-60.885746
-109.967018

Param's

7
12

8
5
2

AIC

188.313395
118.574833
126.648436
131.771493
223.934036

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

125.359
79.7386
16.0736
11.1231

10
5
4
3

<.0001
<.0001
0.002922
0.01108

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 0.380208

BMDL = 0.013927

This document is a draft for review purposes only and does not constitute Agency policy.

E-239 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.47.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

13:32 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-240 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.2.48. Van Birgelen et al., 1995a: Hepatic Retinol Palmitate
E.2.48.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

4

<0.0001

460.282

error

error



exponential (M3)

4

<0.0001

460.282

error

error

power hit bound (d = 1)

exponential
(M4)b

3

<0.0001

446.995

1.415E+02

3.647E+01



exponential (M5)

3

<0.0001

446.995

1.415E+02

3.647E+01

power hit bound (d = 1)

Hill

3

0.009

416.233

3.657E+00

error

n lower bound hit (n = 1)

linear

4

<0.0001

486.375

3.487E+02

2.412E+02



polynomial, 5-
degree

0

N/A

584.170

error

5.617E+02



power

4

<0.0001

486.375

3.487E+02

2.412E+02

power bound hit (power = 1)

Hill, unrestricted

3

<0.0001

527.310

6.875E-14

6.875E-14

unrestricted (n = 0.613)

power,
unrestricted0

3

0.239

408.982

5.262E-02

5.889E-05

unrestricted (power = 0.064)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.48.2. Output for Selected Model: Exponential (M4)

Van Birgelen et al., 1995a: Hepatic Retinol Palmitate

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\l\Blood\66_VanB_l995a_HepRetPalm_Exp_l.(d)

Gnuplot Plotting File:

Mon Feb 08 13:32:41 2010

Tbl3, hepatic retinol palmitate

The form of the response function by Model:

Model 2:	Y[dose] = a * exp{sign * b * dose}

Model 3:	Y[dose] = a * exp{sign * (b * dosej^d}

Model 4:	Y[dose] = a * [c-(c-l) * exp{-b * dose}]

Model 5:	Y[dose] = a * [c-(c-l) * exp{-(b * dosej^d}]

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;

This document is a draft for review purposes only and does not constitute Agency policy.

E-241 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	0.284674

rho	1.77158

a	4 95.6

b	0.0337826

c	0.00576502

d	1

Parameter Estimates
Variable	Model 4

No Convergence

lnalpha	-0.241601

rho	2.03456

a	223.848

b	0.0300737

c	0.0129253

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	8	472	271.5

7.204	8	94	67.88

11.76	8	107	76.37

18.09	8	74	39.6

86.41	8	22	22.63

250.2	8	3	2.828

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	223.8	217.8	3.222

7.204	180.8	175.3	-1.401

11.76	158	152.9	-0.9443

18.09	131.1	126.4	-1.278

86.41	19.33	18.03	0.4197

This document is a draft for review purposes only and does not constitute Agency policy.

E-242 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

250.2

3. 013

2 .721

-0.01317

Other models for which	likelihoods are calculated:

Model A1:	Yij	=	Mu(i) + e(1j)

Var{e(ij)}	=	Sigma^2

Model A2:	Yij	=	Mu(i) + e(1j)

Var{e(ij)}	=	Sigma(i)^2

Model A3:	Yij	=	Mu(i) + e(1j)

Var{e(ij)}	=	exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	=	Mu + e(1)

Var{e(ij)}	=	Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-250.5548	7	515.1096

A2	-196.7557	12	417.5115

A3	-197.3832	8	410.7663

R	-276.7896	2	557.5793

4	-218.4977	5	446.9954

Additive constant for all log-likelihoods =	-44.11. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

160.1
107 . 6
1. 255
42 . 23

10
5
4
3

p-value

<	0.0001

<	0.0001

0.869

<	0.0001

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is less than .1. Model 4 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

This document is a draft for review purposes only and does not constitute Agency policy.

E-243 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Specified Effect	=	1.000000

Risk Type	=	Estimated standard deviations from contrui

Confidence Level	=	0.950000

EMD	=	141.528

E'.MDL	=	36.4 7 21

E.2.48.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

13:32 02/08 2010

E.2.48.4. Output for Additional Model Presented: Power, Unrestricted
Van Birgelen et al., 1995a: Hepatic Retinol Palmitate

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\l\Blood\66_VanB_l995a_HepRetPalm_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\66_VanB_19 95a_HepRetPalm_Pwr_U_l.plt

Mon Feb 08 13:32:47 2010

Tbl3, hepatic retinol palmitate

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-244 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	9.57332

rho =	0

control =	472

slope =	-320.514

power = 0.0711173

Asymptotic Correlation Matrix of Parameter Estimates

lalpha	rho	control	slope	power

lalpha 1	-0.95	0.3	-0.31	-0.3

rho -0.95	1	-0.41	0.39	0.29

control 0.3	-0.41	1	-0.98	-0.82

slope -0.31	0.39	-0.98	1	0.9

power -0.3	0.29	-0.82	0.9	1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate

0. 0640168
1. 81132
464.29
-324.216
0. 0639088

Std. Err.

0. 859472
0.197468
87 . 5705
83.3327
0. 0139778

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-1.62052
1.42429
292.655
-487.545
0.0365129

1.74855
2 .19835
635.925
-160.887
0.0913048

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

7 .204
11.76
18.09
86.41
250.2

472

464

272

269

0.0812

94

96.5

67 . 9

64 . 7

-0.108

107

CO
CO

76.4

57 . 6

1.09

74

74.2

39.6

51

-0.00941

22

33.2

22 . 6

24 . 6

-1.28

3

2 .86

2 . 83

2 . 68

0.145

This document is a draft for review purposes only and does not constitute Agency policy.

E-245 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
-250.554817
-196.755746
-197.383174
-199.490808
-276.789644

Param's

7
12

8
5
2

AIC
515.109634
417.511491
410.766347
408.981615
557.579287

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

160.068
107.598
1. 25486
4 . 21527

10
5
4
3

<.0001
<.0001
0.869
0.2391

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 0.0526247

This document is a draft for review purposes only and does not constitute Agency policy.

E-246 DRAFT—DO NOT CITE OR QUOTE

-------
BMDL = 5.8 8 8 83e-005

E.2.48.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

a>

CO
c
o

Q.

CO
CD

a:

c
m

CD

700

600

500

400

300

200

100

Power

S/IDLBMD

13:32 02/08 2010

50

100

150

200

250

dose

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-247 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.2.49. White et al., 1986: CH50

E.2.49.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

5

0.002

389.664

1.957E+01

1.261E+01



exponential (M3)

5

0.002

389.664

1.957E+01

1.261E+01

power hit bound (d = 1)

exponential (M4)

4

0.001

390.632

1.411E+01

5.177E+00



exponential (M5)

4

0.001

390.632

1.411E+01

5.177E+00

power hit bound (d = 1)

Hill b

4

0.002

389.601

8.632E+00

1.498E+00

n lower bound hit (n = 1)

linear

5

<0.001

394.446

3.497E+01

2.568E+01



polynomial, 6-
degree

5

<0.001

394.446

3.497E+01

2.568E+01



power

5

<0.001

394.446

3.497E+01

2.568E+01

power bound hit (power = 1)

Hill, unrestricted0

3

0.071

381.520

1.481E-01

4.351E-03

unrestricted (n = 0.246)

power,
unrestricted

4

0.148

379.265

1.211E-01

1.225E-03

unrestricted (power = 0.227)

a Non-constant variance model selected (p = 0.0871)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.2.49.2. Output for Selected Model: Hill

White etal., 1986: CH50

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\71_White_198 6_CH50_Hill_l
Gnuplot Plotting File: C:\l\Blood\71_White_198 6_CH50

Mon Feb 08

[insert study notes]

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose

Power parameter restricted to be greater than 1

This document is a draft for review purposes only and does not constitute Agency policy.

E-248 DRAFT—DO NOT CITE OR QUOTE

• (d)

_Hill_l.pit
~13:35:56 2010

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))
Total number of dose groups = 7

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho

intercept

5.60999
0
91
-74
0.118036
1.094

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
rho

intercept

lalpha
1

-0. 99
0.27
0.23
-0.32

rho

-0. 99
1

-0.28
-0.24
0.33

intercept

0.27
-0.28
1

0.39
-0.78

v

0.23
-0.24
0.39
1

-0. 85

k

-0.32
0.33
-0.78
-0. 85
1

Parameter Estimates

Variable
lalpha
rho

intercept

k

Estimate

4 . 581
0.31293
74 . 6365
-66.2096
1

20.8286

Std. Err.

1.66273
0. 431616
6.33673
14.7876
NA

21.3237

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

1.32211
-0.533022
62.2167
-95.1928

-20.965

7 . 83989
1.15888
87.0562
-37.2264

62.6223

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 8

91

74.6

14 .1

19

4

2 .39

1.094 8

54

71 . 3

8.49

19

3

-2 . 54

4.085 8

63

63. 8

11. 3

18

9

-0.117

7.14 8

56

57 . 7

25.5

18

6

-0.263

26.81 8

41

37 . 4

17

17

4

0.589

48.72 8

32

28 . 3

17

16

7

0. 636

90.56 8

17

20.8

17

15

9

-0.678

This document is a draft for review purposes only and does not constitute Agency policy.

E-249 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
-181.340979
-175.820265
-181. 238690
-189.800288
-212 . 367055

14

9
5
2

AIC

378 . 681959
379.640529
380.477380
389.600575
428.734109

Explanation of Tests

Test

1

Test

2

Test

3

Test

4

(Note:

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

73.0936
11. 0414
10.8369
17 .1232

12

<.0001
0.0871
0.05471
0.001829

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	8.63239

This document is a draft for review purposes only and does not constitute Agency policy.

E-250 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E'.MDL =

1.49823

E.2.49.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

13:35 02/08 2010

E.2.49.4. Output for Additional Model Presented: Hill, Unrestricted
White etal., 1986: CH50

pit
2010

[insert study notes]

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\l\Blood\7l_White_l986_CH50_Hill_U_l.(d)

Gnuplot Plotting File: C:\l\Blood\71_White_1986_CH50_Hill_U_l.

Mon Feb 08 13:35:57

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-251 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Independent variable = Dose

Power parameter is not restricted

The variance is to be modeled as Var(i)

exp(lalpha + rho * ln(mean(i)

Total number of dose groups = 7

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho

intercept

5.60999
0
91
-74
0.118036
1.094

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho

intercept

lalpha
1
-1
0.16
0.19
-0.4
-0.014

rho
-1

1

-0.16
-0.19
0.4
0. 011

intercept

0.16
-0.16
1

0.15
-0.58
0. 015

v

0.19
-0.19
0.15
1

-0. 02
-0. 93

n

-0.4

0.4
-0.58
-0. 02
1

-0.35

k

-0.014
0. 011
0. 015
-0. 93
-0.35
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

Estimate
6.54093
-0.245847
89.6302
-628.486
0.246409
493877

Std. Err.
2.08879
0.541645
5.59428
727.973
0. 058636
2 . 74838e + 006

Lower Conf. Limit
2 .44698
-1.30745
78 . 6656
-2055.29
0.131484
-4.89284e+006

Upper Conf. Limit

10.6349
0. 815757
100.595
798.315
0.361333
5. 88059e + 006

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

1.094
4 . 085
7 .14
26. 81
48.72
90.56

91

54
63
56
41
32
17

89.6

65.2

56.3
51. 7
38 . 3
30. 9
22 . 3

14 .1
8.49
11. 3
25.5
17
17
17

15.1

15. 8
16

16.2
16.8
17 . 3

18

0.256
-2 . 01
1.17
0.746
0. 453
0.175
-0.831

Model Descriptions for likelihoods calculated

This document is a draft for review purposes only and does not constitute Agency policy.

E-252 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(1)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
-181.340979
-175.820265
-181. 238690
-184.759769
-212 . 367055

Param's

14
9

AIC

378 . 681959
379.640529
380.477380
381.519538
428.734109

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

73.0936
11. 0414
10.8369
7 . 04216

12

<.0001
0.0871
0.05471
0.07057

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.148074

BMDL = 0.00435112

This document is a draft for review purposes only and does not constitute Agency policy.

E-253 DRAFT—DO NOT CITE OR QUOTE

-------
E.2.49.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

13:35 02/08 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-254 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

E.3. ADMINISTERED DOSE BMDS RESULTS

E.3.1. Amin et al., 2000: 0.25% Saccharin Consumed, Female

E.3.1.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

linear b

1

0.358

179.702

8.816E+01

5.890E+01



polynomial, 2-
degree

1

0.358

179.702

8.816E+01

5.890E+01



power

1

0.358

179.702

8.816E+01

5.890E+01

power bound hit (power =1)

power,
unrestricted0

0

N/A

180.858

7.530E+01

2.537E+01

unrestricted (power = 0.605)

a Non-constant variance model selected (p = 0.0005)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.1.2. Output for Selected Model: Linear

Amin et al., 2000: 0.25% Saccharin Consumed, Female

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\1\l_Amin_2000_25_SC_Linear_l.(d)

Gnuplot Plotting File: C:\l\l_Amin_2 000_25_SC_Linear_l.plt

Tue_Feb 16~17:22:16 2010

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose

Signs of the polynomial coefficients are not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	5.29482

rho =	0

beta_0 =	30.8266

beta~l = -0.204134

This document is a draft for review purposes only and does not constitute Agency policy.

E-255 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
beta_0
beta 1

lalpha
1

-0. 99
-0.016
0. 03

rho

-0. 99
1

0. 013
-0.026

beta_0

-0.016
0. 013
1

-0. 94

beta_l

0. 03
-0.026
-0. 94
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
beta_0
beta 1

Estimate

-2.55843
2 .42056
30.3968
-0.196699

Std. Err.

1.66185
0.545617
4 . 03582
0.0443352

Lower Conf. Limit

-5.8156
1.35117
22.4868
-0.283594

Upper Conf. Limit

0.698746
3. 48995
38 . 3069
-0.109803

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10
25 10
100 10

31. 7

24 . 6
10.7

30. 4
25.5
10.7

20.6
12
5.33

17 . 3
14
4 . 92

0.233
-0.2

-0.0204

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Model A2:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

Mu(i) + e(ij ;

Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi

Var{e(i)}

Mu + e(i)
Sigma/N2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
-92.841935
-85.255316
-85.429148
-85.851107
-98.136607

Param's

4
6

5
4
2

AIC

193.683870
182.510632
180.858295
179.702213

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

This document is a draft for review purposes only and does not constitute Agency policy.

E-256 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test -2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

25.7626
15.1732
0.347663
0. 843918

4
2
1
1

<.0001
0.0005072

0.5554
0.3583

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation

Specified effect

1

Risk Type

Estimated standard deviations from the control mean

Confidence level

0. 95

BMD

88 .1623

BMDL

58.9029

This document is a draft for review purposes only and does not constitute Agency policy.

E-257 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.1.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

17:22 02/16 2010

E.3.1.4. Output for Additional Model Presented: Power, Unrestricted
Amin et al., 2000: 0.25% Saccharin Consumed, Female

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\l_Amin_2000_25_SC_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\l_Amin_2 000_25_SC_Pwr_U_l.plt

Tue Feb 16 17:22:17 2010

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-258 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho
control
slope
power

5.29482

0

31.6727
-0.567889
0.783745

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope
power

lalpha
1

-0. 99
0.34
-0.14
-0.061

rho

-0. 99
1

-0.42
0.15
0. 068

control

0.34
-0.42
1

-0. 67
-0.56

slope

-0.14
0.15
-0. 67
1

0. 99

power

-0.061
0. 068
-0.56
0. 99
1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate
-2 .48291
2 . 38455
32 . 99
-1.36469
0.605364

Std. Err.

2.08669
0.692047
5.40754
2.01258
0.288476

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-6.57274

1. 02817
22 .3914
-5.30927
0.0399625

1.60693
3.74094
43.5886
2.5799
1 . 17077

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10
25 10
100 10

31. 7

24 . 6
10.7

33

23. 4
10.8

20.6
12
5.33

18 . 7
12 . 4
4 . 94

-0.223
0.302
-0. 08

Warning: Likelihood for fitted model larger than the Likelihood for model A3.

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Model A2:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

Mu(i) + e(ij ;

Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that

This document is a draft for review purposes only and does not constitute Agency policy.

E-259 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -92.841935	4	193.683870

A2 -85.255316	6	182.510632

A3 -85.429148	5	180.858295

fitted -85.429148	5	180.858295

R -98.136607	2	200.273213

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	25.7626	4	<.0001

Test 2	15.1732	2	0.0005072

Test 3	0.347663	1	0.5554

Test 4	-8.2423e-013	0	NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 7 5.2994

BMDL = 25.3717

This document is a draft for review purposes only and does not constitute Agency policy.

E-260 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.1.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

17:22 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-261 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

E.3.2. Amin et al., 2000: 0.25% Saccharin Preference Ratio, Female
E.3.2.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

linear b

1

0.002

228.094

1.264E+02

6.128E+01



polynomial, 2-
degree

1

0.002

228.094

1.264E+02

6.128E+01



power

1

0.002

228.094

1.264E+02

6.128E+01

power bound hit (power = 1)

a Non-constant variance model selected (p = 0.0135)
b Best-fitting model, BMDS output presented in this appendix

E.3.2.2. Output for Selected Model: Linear

Amin et al., 2000: 0.25% Saccharin Preference Ratio, Female

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\1\2_Amin_2000_25_SP_Linear_l.(d)

Gnuplot Plotting File: C:\l\2_Amin_2 000_25_SP_Linear_l.plt

Tue_Feb 16~17:22:44 2010

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose

Signs of the polynomial coefficients are not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	6.34368

rho =	0

beta_0 =	74.2008

beta 1 = -0.219781

Asymptotic Correlation Matrix of Parameter Estimates
lalpha	rho	beta 0	beta 1

lalpha	1	-1	0.2	-0.28

This document is a draft for review purposes only and does not constitute Agency policy.

E-262 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

rho
beta_0
beta 1

-1

0.2
-0.28

1

-0.19
0.28

-0.19
1

-0.76

0.28
-0.76
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
beta_0
beta 1

Estimate
0.338774
1.43998
73.6633
-0.207175

Std. Err.
9.23768
2 . 21674
6.6623
0.101074

Lower Conf. Limit
-17.7667
-2.90476
60.6054
-0.405276

Upper Conf. Limit
18.4443
5.78472
86.7211
-0.00907442

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10
25 10
100 10

82 .1
58 .1
54 . 9

73 . 7
68 . 5
52 . 9

13.3

33. 9
19.5

26.2

24 . 8
20.6

1. 02
-1. 32
0.295

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-108.574798	4	225.149597

A2	-104.269377	6	220.538754

A3	-105.147952	5	220.295903

fitted	-110.046917	4	228.093834

R	-112.382522	2	228.765045

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

This document is a draft for review purposes only and does not constitute Agency policy.

E-263 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

16.2263

1.75715
9.79793

8.61084

4
2
1
1

0.00273
0.0135
0.185
0.001747

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation

Specified effect

1

Risk Type

Estimated standard deviations from the control mean

Confidence level

0. 95

BMD

126.365

BMDL

61.2812

This document is a draft for review purposes only and does not constitute Agency policy.

E-264 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.2.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

17:22 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-265 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

E.3.3. Amin et al., 2000: 0.50% Saccharin Consumed, Female
E.3.3.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

linear b

1

0.031

159.737

9.874E+01

6.417E+01



polynomial, 2-
degree

1

0.031

159.737

9.874E+01

6.417E+01



power

1

0.031

159.737

9.874E+01

6.417E+01

power bound hit (power =1)

power,
unrestricted0

0

N/A

157.060

5.610E+01

6.781E+00

unrestricted (power = 0.325)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.3.2. Output for Selected Model: Linear

Amin et al., 2000: 0.50% Saccharin Consumed, Female

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\1\3_Amin_2000_50_SC_Linear_l.(d)

Gnuplot Plotting File: C:\l\3_Amin_2 000_50_SC_Linear_l.plt

Tue_Feb 16~17:23:14 2010

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose

Signs of the polynomial coefficients are not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha =	4.68512

rho =	0

beta_0 =	19.3484

beta 1 = -0.158141

Asymptotic Correlation Matrix of Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

E-266 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

lalpha
rho
beta_0
beta 1

lalpha
1

-0. 97
0. 018
-0.0021

rho

-0. 97
1

-0.027
0. 014

beta_0

0. 018
-0.027
1

-0. 95

beta_l

-0.0021
0. 014
-0. 95
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
beta_0
beta 1

Estimate

-0.997428
2 .13634
18 .1144
-0.135736

Std. Err.

0.992786
0. 404 989
3.10302
0.0331501

Lower Conf. Limit

-2.94325
1.34257
12.0326
-0.200709

Upper Conf. Limit

0. 948397
2.9301
24.1962
-0.0707631

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10
25 10
100 10

22 . 4
11. 4
4 . 54

18 .1
14 . 7
4 . 54

16

7 . 66
3.33

13. 4

10.7
3. 06

1

-0.983
-0.00393

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -83.696404	4	175.392808

A2 -73.511830	6	159.023660

A3 -73.530233	5	157.060467

fitted -75.868688	4	159.737377

R -90.294746	2	184.589492

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

This document is a draft for review purposes only and does not constitute Agency policy.

E-267 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

(Note:

When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test -2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

0. 0368066

33.5658
20.3691

4.67691

4
2
1
1

<.0001
<.0001
0.8479
0.03057

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation

Specified effect

1

Risk Type

Estimated standard deviations from the control mean

Confidence level

0. 95

BMD

98.7409

BMDL

64.169

This document is a draft for review purposes only and does not constitute Agency policy.

E-268 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.3.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

17:23 02/16 2010

E.3.3.4. Output for Additional Model Presented: Power, Unrestricted
Amin et al., 2000: 0.50% Saccharin Consumed, Female

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\3_Amin_2000_50_SC_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\3_Amin_2 000_50_SC_Pwr_U_l.plt

Tue Feb 16 17:23:15 2010

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-269 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha =	4.68512

rho =	0

control =	22.3564

slope =	-3.5587 4

power =	0.34 97 99

Asymptotic Correlation Matrix of Parameter Estimates

lalpha	rho	control	slope	power

lalpha 1	-0.96	0.34	-0.26	-0.15

rho -0.96	1	-0.47	0.3	0.15

control 0.34	-0.47	1	-0.73	-0.52

slope -0.26	0.3	-0.73	1	0.96

power -0.15	0.15	-0.52	0.96	1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate
-0.708629
1.96142
22.6293
-4.03215
0.325414

Std. Err.
1.298
0.529653
4.48416
3.21302
0.138761

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-3.25267

0.923323
13.8405
-10.3296
0.053447

1.83541

2.99953
31.4181
2 .26526
0.597381

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10	22.4	22.6	16	15	-0.0577

25 10	11.4	11.1	7.66	7.46	0.105

100 10	4.54	4.58	3.33	3.12	-0.0475

Warning: Likelihood for fitted model larger than the Likelihood for model A3.

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

This document is a draft for review purposes only and does not constitute Agency policy.

E-270 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

Model R:	Yi = Mu + e(i

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1	-83.696404	4	175.392808

A2	-73.511830	6	159.023660

A3	-73.530233	5	157.060467

fitted	-73.530233	5	157.060467

R	-90.294746	2	184.589492

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	33.5658	4	<.0001

Test 2	20.3691	2	<.0001

Test 3	0.0368066	1	0.8479

Test 4	-2.84217e-014	0	NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 56.0 967

BMDL = 6.7 8112

This document is a draft for review purposes only and does not constitute Agency policy.

E-271 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.3.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

17:23 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-272 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

E.3.4. Amin et al., 2000: 0.50% Saccharin Preference Ratio, Female
E.3.4.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

linear b

1

0.088

234.936

8.278E+01

5.100E+01



polynomial, 2-
degree

1

0.088

234.936

8.278E+01

5.100E+01



power

1

0.088

234.936

8.278E+01

5.100E+01

power bound hit (power =1)

power,
unrestricted0

0

N/A

234.020

1.817E+01

1.000E-13

unrestricted (power = 0.232)

a Constant variance model selected (p = 0.5593)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.4.2. Output for Selected Model: Linear

Amin et al., 2000: 0.50% Saccharin Preference Ratio, Female

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\1\4_Amin_2000_50_SP_LinearCV_l.(d)

Gnuplot Plotting File: C:\l\4_Amin_2 000_50_SP_LinearCV_l.plt

Tue_Feb 16 17:23:43 2010

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Signs of the polynomial coefficients are not restricted
A constant variance model is fit

Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
alpha =	764.602

rho =	0 Specified

beta_0 =	64.1858

beta~l = -0.332668

This document is a draft for review purposes only and does not constitute Agency policy.

E-273 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Asymptotic Correlation Matrix of Parameter Estimates

alpha
beta_0
beta 1

The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

2e-008
1. 4e-009

beta_0

2e-008
1

-0.7

beta_l
1. 4e-009
-0.7
1

Parameter Estimates

Variable
alpha
beta_0
beta 1

Estimate
758.396
64 .1858
-0.332668

95.0% Wald Confidence Interval
Std. Err.	Lower Conf. Limit Upper Conf. Limit

195.817	374.602	1142.19

7.04184	50.3841	77.9876

0.118327	-0.564584	-0.100752

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 10
25 10
100 10

12. 7
44.5
33. 8

64 . 2
55. 9
30. 9

24 . 6
32 . 9
24 . 6

27 . 5
27 . 5
27 . 5

0. 981
-1. 31
0.327

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi

Var{e(i)}

Mu + e(i)
Sigma/N2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)

-113.009921
-112 .428886
-113.009921
-114.468091
-117.976057

Param's

4
6
4
3
2

AIC

234.019841
236.857773
234.019841
234.936183
239.952114

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

This document is a draft for review purposes only and does not constitute Agency policy.

E-274 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test -2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

11.0943
1.16207
1.16207
2.91634

4
2
2
1

0.02552
0.5593
0.5593

0.08769

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation

Specified effect

1

Risk Type

Estimated standard deviations from the control mean

Confidence level

0. 95

BMD

82 .7823

BMDL

50.9971

This document is a draft for review purposes only and does not constitute Agency policy.

E-275 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.4.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

17:23 02/16 2010

E.3.4.4. Output for Additional Model Presented: Power, Unrestricted
Amin et al., 2000: 0.50% Saccharin Preference Ratio, Female

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\4_Amin_2000_50_SP_PwrCV_U_l.(d)

Gnuplot Plotting File: C:\l\4_Amin_2 000_50_SP_PwrCV_U_l.plt

Tue Feb 16 17:23:44 2010

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-276 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 3

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho
control
slope
power

764.602

0

72.7273
-13.387
0.231973

Specified

Asymptotic Correlation Matrix of Parameter Estimates

alpha
control

The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

-1. 3e-008

control
-1.3e-008
1

slope
5.9e-009
-0.4

power
2.5e-009
-0.22

slope	5.9e-009	-0.4	1	0.97

power	2.5e-009	-0.22	0.97	1

Parameter Estimates

Variable
alpha
control
slope
power

Estimate
688.142
72.7273
-13.387
0.231973

Std. Err.

177.677
8.29543
15.9957
0.268067

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

339. 9

56.4686
-44.738
-0.293429

1036.38

88 . 986
17 . 9639
0.757376

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	10 72.7	72.7	24.6	26.2	5.16e-008

25	10 44.5	44.5	32.9	26.2	-1.27e-008

100	10 33.8	33.8	24.6	26.2	-2e-008

Degrees	of freedom for	Test A3 vs fitted	<= 0

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-277 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-113.009921	4	234.019841

A2	-112.428886	6	236.857773

A3	-113.009921	4	234.019841

fitted	-113.009921	4	234.019841

R	-117.976057	2	239.952114

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	11.0943	4	0.02552

Test 2	1.16207	2	0.5593

Test 3	1.16207	2	0.5593

Test 4	0	0	NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 18.1732

BMDL = le-013

This document is a draft for review purposes only and does not constitute Agency policy.

E-278 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.4.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

17:23 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-279 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.3.5. Bell et al., 2007a: Balano-Preputial Separation, Postnatal Day 49
E.3.5.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

2

0.369

113.514

7.332E+00

4.687E+00

power bound hit (power = 1)

logistic

2

0.237

114.853

1.501E+01

1.137E+01

negative intercept (intercept =
-2.07)

log-logistica

2

0.456

112.952

5.209E+00

2.870E+00

slope bound hit (slope = 1)

log-probit

2

0.178

115.488

1.428E+01

9.138E+00

slope bound hit (slope =1)

multistage, 3-
degree

2

0.369

113.514

7.332E+00

4.687E+00

final B = 0

probit

2

0.248

114.723

1.399E+01

1.061E+01

negative intercept (intercept =
-1.23)

Weibull

2

0.369

113.514

7.332E+00

4.687E+00

power bound hit (power = 1)

gamma,
unrestricted

1

0.566

113.746

1.894E+00

7.609E-02

unrestricted (power = 0.506)

log-logistic,
unrestricted b

1

0.484

113.908

2.127E+00

1.363E-01

unrestricted (slope = 0.67)

log-probit,
unrestricted

1

0.439

114.021

2.179E+00

1.671E-01

unrestricted (slope = 0.389)

Weibull,
unrestricted

1

0.534

113.802

2.007E+00

1.075E-01

unrestricted (power = 0.574)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix
E.3.5.2.

E.3.5.3. Output for Selected Model: Log-Logistic

Bell et al., 2007a: Balano-Preputial Separation, Postnatal Day 49

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\5_Bell_2007_BPS_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\5_Bell_2007_BPS_LogLogistic_l.plt

Tue Feb 16 17724:10 2010

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

Slope parameter is restricted as slope >= 1

This document is a draft for review purposes only and does not constitute Agency policy.

E-280 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values

background = 0.0333333
intercept =	-3.75371

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept

background	1	-0.58

intercept	-0.58	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.0635251	*	*	*

intercept	-3.84765	*	*	*

slope	1	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-53.7077	4

Fitted model -54.476 2	1.53661 2	0.4638

Reduced model -63.9797 1	20.544 3	0.0001309

AIC:	112.952

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000
2.4000
8.0000
46.0000

0.0635
0.1091
0.2000
0.5273

1

15

906
3.274
6. 001
19

1. 000

5.	000

6.	000
15.000

30
30
30
30

-0.678
1. 011

-0.000
-0.300

Chi ^2

1. 57

d.f.

P-value

0.4559

Benchmark Dose Computation
Specified effect =	0.1

This document is a draft for review purposes only and does not constitute Agency policy.

E-281 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	5.20918

BMDL =	2.86991

E.3.5.4. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

dose

17:24 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-282 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

E.3.5.5. Output for Additional Model Presented: Log-Logistic, Unrestricted
Bell et al., 2007a: Balano-Preputial Separation, Postnatal Day 49

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\5_Bell_2007_BPS_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\5_Bell_2007_BPS_LogLogistic_U_l.plt

Tue Feb 16 17721:10 2010

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose)

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial	Parameter Values

background =	0.0333333

intercept =	-2.54947

slope =	0.615936

Asymptotic Correlation Matrix of Parameter Estimates
background intercept	slope

background	1	-0.49	0.35

intercept	-0.49	1	-0.93

slope	0.35	-0.93	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.0354714	*	*	*

intercept	-2.70296	*	*	*

slope	0.670238	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood) # Param's Deviance Test d.f. P-value

Full model	-53.7077	4

This document is a draft for review purposes only and does not constitute Agency policy.

E-283 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

Fitted model	-53.9541	3	0.492844	1	0.4827

Reduced model	-63.9797	1	20.544	3	0.0001309

AIC:	113.908

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.0355

1. 064

1. 000

30

-0.063

2.4000

0.1392

4.176

5. 000

30

0. 435

8.0000

0.2405

7 . 216

6. 000

30

-0.520

46.0000

0.4848

14.544

15.000

30

0.167

Chi^2 = 0.49	d.f. = 1	P-value = 0.4836

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	2.12667

BMDL =	0.13633

This document is a draft for review purposes only and does not constitute Agency policy.

E-284 DRAFT—DO NOT CITE OR QUOTE

-------
l E.3.5.6. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-285 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.3.6. Cantoni et al., 1981: Urinary Coproporhyrins, 3 Months
E.3.6.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

0.002

33.792

1.101E+02

5.318E+01



exponential (M3)

2

0.002

33.792

1.101E+02

5.318E+01

power hit bound (d = 1)

exponential
(M4)b

1

0.341

23.881

3.741E-01

1.253E-01



exponential (M5)

1

0.341

23.881

3.741E-01

1.253E-01

power hit bound (d = 1)

Hill

1

0.535

23.359

3.273E-01

error

n lower bound hit (n = 1)

linear

2

0.002

33.301

7.734E+01

1.975E+01



polynomial, 3-
degree

2

0.002

33.301

7.734E+01

1.975E+01



power

2

0.002

33.301

7.734E+01

1.975E+01

power bound hit (power =1)

power,
unrestricted0

1

0.665

23.162

4.637E-03

8.796E-08

unrestricted (power = 0.22)

Hill, unrestricted

0

N/A

24.974

7.264E-02

1.656E-04

unrestricted (n = 0.48)

a Non-constant variance model selected (p = 0.0039)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.6.2. Output for Selected Model: Exponential (M4)

Cantoni et al., 1981: Urinary Coproporhyrins, 3 Months

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\6_Cantoni_l981_UriCopro_Exp_l.(d)

Gnuplot Plotting File:

Tue Feb 16 17:24:39 2010

Figurel-UrinaryCoproporphyrin 3months

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;

This document is a draft for review purposes only and does not constitute Agency policy.

E-286 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	-1.50063

rho	2.6097 9

a	0.704303

b	0.0205927

c	4.47268

d	1

Parameter Estimates
Variable	Model 4

lnalpha	-1.74154

rho	2.66803

a	0.755982

b	0.3715

c	3.93845

d	1

Table of Stats From Input Data
Dose	N	Obs Mean	Obs Std Dev

0	4	0.7414	0.3475

1.43	4	1.807	0.8341

14.3	4	2.734	1.506

143	4	3	2.6

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	0.756	0.2882	-0.1014

1.43	1.671	0.8307	0.3265

14.3	2.966	1.786	-0.2607

143	2.977	1.794	0.02532

Other models for which likelihoods are calculated:

This document is a draft for review purposes only and does not constitute Agency policy.

E-287 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma^2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-12.90166	5	35.80333

A2	-6.203643	8	28.40729

A3	-6.487204	6	24.97441

R	-15.73713	2	35.47427

4	-6.940389	5	23.88078

Additive constant for all log-likelihoods =	-14.7. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Does response

and/or variances differ among Dose levels? (A2 vs. R)

Test

2 :

Are Variances

Homogeneous? (A2 vs. A1)

Test

3:

Are variances

adequately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

19. 07
13. 4
0.5671
0.9064

D. F.

p-value

0.004052
0.003854
0.7531
0.3411

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

This document is a draft for review purposes only and does not constitute Agency policy.

E-288 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

EMD =
E'.MDL =

0.374114
0.125287

E.3.6.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

a>

CO
c
o

Q.

CO
CD

a:

c
m

CD

5
4

3
2
1

0
-1

Exponential

BplDLBMD

20

40

60

80

100

120

dose

17:24 02/16 2010

E.3.6.4. Output for Additional Model Presented: Power, Unrestricted
Cantoni et al., 1981: Urinary Coproporhyrins, 3 Months

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\6_Cantoni_l981_UriCopro_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\6_Cantoni_1981_UriCopro_Pwr_U_l.plt

Tue Feb 16 17:24:41 2010

Figurel-UrinaryCoproporphyrin_3months

The form of the response function is:
Y[dose] = control + slope * doseApower

140

This document is a draft for review purposes only and does not constitute Agency policy.

E-289 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean

Independent variable = Dose

The power is not restricted

The variance is to be modeled as Var(i)

exp(lalpha + log(mean(i))

rho)

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha =	0.90039

rho =	0

control =	0.741372

slope =	1.00533

power =	0.163111

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope
power

lalpha
1

-0. 62
-0.53
-0.038
0. 027

rho

-0. 62
1

0.43
-0.24
-0.16

control

-0.53
0.43
1

-0.3

0.09

slope
-0.038
-0.24
-0.3
1

-0.72

power

0. 027
-0.16
0.09
-0.72
1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate
-1.78404
2 . 6428
0.757242
0.927009
0.220276

Std. Err.

0. 61698
0.74449
0.139966
0.325923
0 . 0964599

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-2.9933
1.18363
0. 482915
0.288212
0. 031218

-0.57478
4 .10197
1.03157
1. 56581
0.409334

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	4

1.43	4

14.3	4

143	4

0.741
1. 81

2 .73
3

0.757
1.76
2 .42
3.52

0.348
0. 834
1. 51
2 . 6

0.284
0. 865
1. 32
2 .16

-0.112
0.108
0.471
-0.483

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-290 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-12.901663	5	35.803325

A2	-6.203643	8	28.407287

A3	-6.487204	6	24.974409

fitted	-6.580755	5	23.161510

R	-15.737135	2	35.474269

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	19.067	6	0.004052

Test 2	13.396	3	0.003854

Test 3	0.567122	2	0.7531

Test 4	0.187101	1	0.6653

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 0.00463746

BMDL = 8.7 9634e-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-291 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.6.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

17:24 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-292 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.3.7. Cantoni et al., 1981: Urinary Porphyrins
E.3.7.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential
(M2)b

2

<0.0001

58.753

1.223E+01

9.037E+00



exponential (M3)

2

<0.0001

58.753

1.223E+01

9.037E+00

power hit bound (d = 1)

exponential (M4)

1

<0.0001

63.138

2.227E-01

1.137E-01



exponential (M5)

1

<0.0001

63.138

2.227E-01

1.137E-01

power hit bound (d = 1)

Hill

0

N/A

62.356

9.363E+00

4.664E+00



linear

2

<0.0001

62.487

7.732E-01

2.816E-01



polynomial, 3-
degree

1

<0.0001

10.000

error

error



power

2

<0.0001

62.487

7.732E-01

2.816E-01

power bound hit (power =1)

power,
unrestricted

1

<0.0001

59.914

1.025E-01

2.389E-02

unrestricted (power = 0.746)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix

E.3.7.2. Output for Selected Model: Exponential (M2)

Cantoni et al., 1981: Urinary Porphyrins

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\7_Cantoni_l981_UriPor_Exp_l.(d)

Gnuplot Plotting File:

Tue Feb 16 17:25:14 2010

Table 1, dose converted to ng per kg per day

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-293 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 2

lnalpha	-3.57509

rho	2.23456

a	3.83141

b	0.0277822

c	0

d	1

Parameter Estimates
Variable	Model 2

lnalpha	-1.55886

rho	1.77962

a	4.17268

b	0.0270415

c	0

d	1

Table of Stats From Input	Data

Dose	N	Obs Mean	Obs Std Dev

0	4	2.27	0.49

1.43	4	5.55	0.85

14.3	3	7.62	1.79

143	3	196.9	63.14

Estimated Values of	Interest
Dose Est Mean Est Std Scaled Residual

0	4.173	1.635	-2.327

1.43	4.337	1.692	1.433

14.3	6.143	2.307	1.109

143	199.4	51.04	-0.08645

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-294 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest
Model	Log(likelihood)	DF	AIC

A1	-51.42175	5	112.8435

A2	-15.31211	8	46.62422

A3	-15.66963	6	43.33925

R	-68.75058	2	141.5012

2	-25.37651	4	58.75302

Additive constant for all log-likelihoods =	-12.87. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Test

2 :

Test

3:

Test

4 :

Does response and/or variances differ among Dose levels? (A2 vs. R)

Are Variances Homogeneous? (A2 vs. A1)

Are variances adeguately modeled? (A2 vs. A3)

Does Model 2 fit the data? (A3 vs. 2)

Test 1
Test 2
Test 3
Test 4

Tests of Interest

-2*log(Likelihood Ratio)

106. 9
12. 22
0.715
19.41

p-value

0.0001
0.0001
0.6 9 9 4
0.0001

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 4 is less than .1. Model 2 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	12.227 2

BMDL =	9.03732

This document is a draft for review purposes only and does not constitute Agency policy.

E-295 DRAFT—DO NOT CITE OR QUOTE

-------
l E.3.7.3. Figure for Selected Model: Exponential (M2)

Exponential_beta Model 2 with 0.95 Confidence Level

dose

2	17:25 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-296 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.3.8. Crofton et al., 2005: Serum, T4

E.3.8.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

8

<0.0001

518.241

2.136E+03

1.157E+03



exponential (M3)

8

<0.0001

518.241

2.136E+03

1.157E+03

power hit bound (d = 1)

exponential
(M4)b

7

0.957

476.204

5.633E+01

3.006E+01



exponential (M5)

7

0.957

476.204

5.633E+01

3.006E+01

power hit bound (d = 1)

Hill

6

0.973

477.434

5.564E+01

2.590E+01



linear

8

<0.0001

523.518

4.246E+03

3.086E+03



polynomial, 8-
degree

8

<0.0001

523.518

4.246E+03

3.086E+03



power

8

<0.0001

523.518

4.246E+03

3.086E+03

power bound hit (power =1)

power,
unrestricted

7

0.030

489.670

2.179E+01

2.271E+00

unrestricted (power = 0.217)

a Constant variance model selected (p = 0.7647)
b Best-fitting model, BMDS output presented in this appendix

E.3.8.2. Output for Selected Model: Exponential (M4)

Crofton et al., 2005: Serum, T4

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\8_Crofton_2005_T4_ExpCV_l.(d)

Gnuplot Plotting File:

Tue Feb 16 17:26:01 2010

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-297 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 10

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values

Variable

Model 4

lnalpha

rho(S)

5.47437

0

104.999
0.000371694
0.445764

1

(S)

Specified

Parameter Estimates

Variable

lnalpha
rho

Model 4

5.50283

0

99.776
0.00728387
0.533516

1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	14	100	15.44

0.1 6	96.27	14.98

3	12	98.57	18.11

10 6	99.76	19.04

30 6	93.32	12.11

100 6	70.94	12.74

300 6	62.52	14.75

1000 6	52.68	22.73

3000 6	54.66	19.71

le+004 4	49.15	11.15

Estimated Values of Interest

Dose Est	Mean	Est Std	Scaled Residual

0

99

78

15. 66

0.05325

0.1

99

74

15. 66

-0.5434

3

98

77

15. 66

-0. 04357

10

96

51

15. 66

0.5085

30

90

64

15. 66

0.4195

This document is a draft for review purposes only and does not constitute Agency policy.

E-298 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

100	75.7	15.66	-0.744

300	58.47	15.66	0.6334

1000	53.26	15.66	-0.09133

3000	53.23	15.66	0.2237

le+004	53.23	15.66	-0.5218

Other models for which	likelihoods are calculated:

Model A1:	Yij	=	Mu(i) + e(1j)

Var{e(ij)}	=	Sigma^2

Model A2:	Yij	=	Mu(i) + e(1j)

Var{e(ij)}	=	Sigma(i)^2

Model A3:	Yij	=	Mu(i) + e(1j)

Var{e(ij)}	=	exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	=	Mu + e(1)

Var{e(ij)}	=	Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-233.0774	11	488.1549

A2	-230.2028	20	500.4056

A3	-233.0774	11	488.1549

R	-268.4038	2	540.8076

4	-234.1019	4	476.2038

Additive constant for all log-likelihoods =	-66.16. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

76.4
5.749
5.749
2.049

D. F.

18
9
9
7

p-value

< 0.0001
0.7647
0.7647
0.9571

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

This document is a draft for review purposes only and does not constitute Agency policy.

E-299 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	56.3321

BMDL =	30.0635

E.3.8.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

a>

CO
c
o

Q.

CO
CD

a:

c
m

CD

120

100

80

60

40

20

Exponential

BMDLBMD

2000

4000

6000

8000

10000

dose

17:26 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-300 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.3.9. Franc et al., 2001: S-D Rats, Relative Liver Weight
E.3.9.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Hill

1

0.797

236.371

1.826E+01

5.463E+00

n lower bound hit (n = 1)

exponential (M2)

2

0.935

234.440

2.262E+01

1.757E+01



exponential (M3)

2

0.935

234.440

2.262E+01

1.757E+01

power hit bound (d = 1)

exponential (M4)

1

0.797

236.371

1.827E+01

6.112E+00



exponential (M5)

1

0.797

236.371

1.827E+01

6.112E+00

power hit bound (d = 1)

linear

2

0.967

234.372

1.861E+01

1.339E+01



polynomial, 3-
degree

2

0.967

234.372

1.861E+01

1.339E+01



power b

2

0.967

234.372

1.861E+01

1.339E+01

power bound hit (power =1)

Hill, unrestricted

0

N/A

238.366

1.726E+01

2.022E+00

unrestricted (n = 0.965)

power,
unrestricted0

1

0.805

236.365

1.725E+01

2.003E+00

unrestricted (power = 0.962)

a Constant variance model selected (p = 0.107)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.9.2. Output for Selected Model: Power
Franc et al., 2001: S-D Rats, Relative Liver Weight

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\88_Franc_2001_SD_RelLivWt_PowerCV_l.(d)

Gnuplot Plotting File: C:\l\88_Franc_2001_SD_RelLivWt_PowerCV_l.plt

Fri Apr 16 16:28:45 2010

Figure 5, SD rats, relative liver weight

The form of the response function is:

Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

This document is a draft for review purposes only and does not constitute Agency policy.

E-301 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The power is restricted to be greater than or equal to 1
A constant variance model is fit

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial
alpha =
rho =
control =
slope =
power =

Parameter Values

527 .447

0 Specified
100
1.15946
0. 839423

Asymptotic Correlation Matrix of Parameter Estimates

alpha
control
slope

The model parameter(s) -rho -power

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

1. 3e-012
-6.2e-013

control
1.3e-012
1

-0. 67

slope
). 2e-013
-0. 67
1

Parameter Estimates

Variable
alpha
control
slope
power

Estimate

462.485
101.047
0.542984
1

Std. Err.

115.621
5.10511
0. 0973507
NA

NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

235.872
91. 0415
0.352181

68 9.0 9 9
111.053
0.733788

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0
10
30
100

100
108
117
155

101
106
117
155

14

16. 9
25. 9
30. 9

21. 5

21. 5
21. 5
21. 5

-0.138
0.208
-0.0702
0.000298

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-302 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
-114 .152281
-111.103649
-114.152281
-114.185827
-125.052064

Param's

5
8
5
3
2

AIC

238.304562
238.207299
238.304562
234.371654
254.104127

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

27 .8968

6. 09726
6.09726
0.0670927

C. 0001
0.107
0.107
0. 967

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Relative risk

Confidence level =	0.95

BMD = 18.6096

BMDL = 13.387 9

This document is a draft for review purposes only and does not constitute Agency policy.

E-303 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.3.9.3. Figure for Selected Model: Power

Power Model with 0.95 Confidence Level

dose

16:28 04/16 2010

E.3.9.4. Output for Additional Model Presented: Power, Unrestricted
Franc et al., 2001: S-D Rats, Relative Liver Weight

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\88_Franc_2001_SD_RelLivWt_PowerCV_U_l.(d)

Gnuplot Plotting File: C:\l\88_Franc_2001_SD_RelLivWt_PowerCV_U_l.plt

Fri Apr 16 16:28:46 2010

Figure 5, SD rats, relative liver weight

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-304 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho
control
slope
power

527 .447

0

100
1.15946
0. 839423

Specified

Asymptotic Correlation Matrix of Parameter Estimates

alpha
control

The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

le-009

control
le-009
1

slope
). 2e-010
-0.74

power
4 . 7e-010
0.71

slope -6.2e-010	-0.74	1	-1

power	4.7e-010	0.71	-1	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
control
slope
power

Estimate
462.394
100.636
0. 650456
0.961853

Std. Err.
115.598
7 .29156
1.43713
0. 465182

Lower Conf. Limit

235.825
86.3448
-2 .16627
0. 0501134

Upper Conf. Limit

688.963
114.927
3.46718
1.87359

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	8	100	101	14	21.5	-0.0836

10	8	108	107	16.9	21.5	0.192

30	8	117	118	25.9	21.5	-0.128

100	8	155	155	30.9	21.5	0.0192

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

This document is a draft for review purposes only and does not constitute Agency policy.

E-305 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

Model R:	Yi

Var{e(i)}

Mu + e(i
Sigma/N2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
-114 .152281
-111.103649
-114.152281
-114.182670
-125.052064

Param's

5
8
5
4
2

AIC

238.304562
238.207299
238.304562
236.365340
254.104127

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

27 .8968

6. 09726
6.09726
0.0607785

<.0001
0.107
0.107
0.8053

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Relative risk

Confidence level =	0.95

BMD = 17.24 69

BMDL = 2.00336

This document is a draft for review purposes only and does not constitute Agency policy.

E-306 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.9.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

16:28 04/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-307 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.3.10. Franc et al., 2001: L-E Rats, Relative Liver Weight
E.3.10.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

0.245

210.148

5.143E+01

3.188E+01



exponential (M3)

2

0.245

210.148

5.143E+01

3.188E+01

power hit bound (d = 1)

exponential (M4)

1

0.607

209.599

1.476E+01

3.702E+00



exponential (M5)

1

0.607

209.599

1.476E+01

3.702E+00

power hit bound (d = 1)

Hill b

1

0.703

209.480

1.321E+01

1.591E+00

n lower bound hit (n = 1)

linear

2

0.273

209.933

4.753E+01

2.788E+01



polynomial, 3-
degree

1

<0.0001

10.000

1.505E+01

error



power

2

0.273

209.933

4.753E+01

2.788E+01

power bound hit (power =1)

Hill, unrestricted

C

0

N/A

211.341

1.163E+01

9.756E-01

unrestricted (n = 0.418)

power,
unrestricted

1

0.940

209.340

1.155E+01

1.513E-02

unrestricted (power =0.394)

a Non-constant variance model selected (p = 0.0632)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.10.2. Output for Selected Model: Hill

Franc et al., 2001: L-E Rats, Relative Liver Weight

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\89_Franc_2001_LE_RelLivWt_Hill_l.(d)

Gnuplot Plotting File: C:\l\8 9_Franc_2001_LE_RelLivWt_Hill_l.plt

Fri Apr 16 16:29:20 2010

Figure 5, L-E rats, relative liver weight

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose

Power parameter restricted to be greater than 1

This document is a draft for review purposes only and does not constitute Agency policy.

E-308 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho

intercept

5. 41581

0

100
22 . 225
0.329526
40.8403

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
rho

intercept

lalpha
1
-1

-0.18
0.38
0.2

rho
-1

1

0.17
-0.38
-0.2

intercept

-0.18
0.17
1

-0.13
0.39

v

0.38
-0.38
-0.13
1

0 .77

k

0.2
-0.2

0.39
0 .77
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

k

Estimate
-15.3958
4.38043
99.5667
28 . 8965
1

25.1273

Std. Err.

17 . 0376
3. 61867
3.7178
12.6477
NA
30.138

Lower Conf. Limit
-48.7889
-2 .71204
92 .28
4 .10739

-33.9421

Upper Conf. Limit

17 . 9973
11.4729
106.853
53.6856

84.1966

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	8	100	99.6	10	10.8	0.114

10	8	106	108	17.9	12.8	-0.329

30	8	117	115	8.97	14.9	0.288

100	8	122	123	19.9	17	-0.0723

Model Descriptions for likelihoods calculated

This document is a draft for review purposes only and does not constitute Agency policy.

E-309 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij	= Mu(i) + e(ij

Var{e(ij)}	= Sigma^2

Model A2 :	Yij	= Mu(i) + e(ij

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model
A1
A2
A3

fitted
R

Log(likelihood)
-100.516456
-96.870820
-99.666984
-99.739888
-105.717087

Param1s
5

AIC
211.032912
209.741641
211.333969
209.479776
215.434174

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

17.6925
7.29127
5.59233
0.145807

0.007048
0.06317
0.06104
0.7026

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Relative risk

Confidence level =	0.95

BMD =	13.2 0 94

BMDL =	1.59127

This document is a draft for review purposes only and does not constitute Agency policy.

E-310 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.3.10.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

16:29 04/16 2010

E.3.10.4. Output for Additional Model Presented: Hill, Unrestricted
Franc et al., 2001: L-E Rats, Relative Liver Weight

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\89_Franc_2001_LE_RelLivWt_Hill_U_l.(d)

Gnuplot Plotting File: C:\l\8 9_Franc_2 001_LE_RelLivWt_Hill_U_l.plt

Fri Apr 16 16:29:27 2010

Figure 5, L-E rats, relative liver weight

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
Power parameter is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + rho * In(mean(i)))

Total number of dose groups = 4

This document is a draft for review purposes only and does not constitute Agency policy.

E-311 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha =	5.41581

rho =	0

intercept =	100

v =	22.225

n =	0.329526

k =	40.8403

Asymptotic Correlation Matrix of Parameter Estimates

lalpha rho	intercept v n k

lalpha 1 -1	-0.21 -0.099 0.23 -0.13

rho -1 1	0.21 0.099 -0.23 0.13

intercept -0.21 0.21	1 0.023 0.14 0.011

v -0.099 0.099	0.023 1 -0.84 1

n 0.23 -0.23	0.14 -0.84 1 -0.88

k -0.13 0.13	0.011 1 -0.88 1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

Estimate

-18 . 8355
5.1098
99.526
286.422
0. 418159
32981.9

Std. Err.

18 . 0637
3.83743
3.53402
4487.2
0. 457476
1.52481e+006

Lower Conf. Limit

-54.2397
-2 .41144
92.5994
-8508.33
-0.478477
-2.95559e+006

Upper Conf. Limit
16.5688
12.631
106.453
9081.17
1.31479
3.02155e+006

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0
10
30
100

100
106
117
122

99.5

109
114
123

10
17 . 9
8 . 97
19. 9

10.3
13
14 . 6
17 . 7

0.13

-0.563
0.529
-0.0942

Degrees of freedom for Test A3 vs fitted

0

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-312 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-100.516456	5	211.032912

A2 -96.870820	8	209.741641

A3 -99.666984	6	211.333969

fitted -99.670736	6	211.341472

R	-105.717087	2	215.434174

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

17.6925
7.29127
5.59233
0.00750301

0.007048
0.06317
0.06104
NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Relative risk

Confidence level =	0.95

BMD =	11.6342

BMDL =	0.975601

This document is a draft for review purposes only and does not constitute Agency policy.

E-313 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.10.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

16:29 04/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-314 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.3.11. Franc et al., 2001: S-D Rats, Relative Thymus Weight
E.3.11.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

0.551

285.890

6.730E+00

3.627E+00



exponential (M3)

1

<0.0001

303.995

3.858E+02

6.615E-01



exponential
(M4)b

1

0.972

286.698

3.559E+00

1.714E+00



exponential (M5)

0

N/A

288.696

3.796E+00

1.714E+00



Hill

0

N/A

288.696

4.299E+00

9.311E-01



linear

2

0.252

287.456

1.330E+01

1.062E+01



polynomial, 3-
degree0

2

0.252

287.456

1.330E+01

1.062E+01



power

2

0.252

287.456

1.330E+01

1.062E+01

power bound hit (power =1)

power,
unrestricted

1

0.510

287.131

5.049E-01

4.411E-04

unrestricted (power = 0.388)

a Non-constant variance model selected (p = 0.0320)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.11.2. Output for Selected Model: Exponential (M4)

Franc et al., 2001: S-D Rats, Relative Thymus Weight

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\91_Franc_2001_SD_RelThyWt_Exp_l.(d)

Gnuplot Plotting File:

Fri Apr 16 16:30:07 2010

Figure 5, SD rats, relative thymus weight

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.

This document is a draft for review purposes only and does not constitute Agency policy.

E-315 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model 3 is nested within Model 5.
Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	3.35464

rho	1.08199

a	105

b	0.0424361

c	0.206726

d	1

Parameter Estimates
Variable	Model 4

lnalpha	2.54324

rho	1.25901

a	108.904

b	0.0379343

c	0.208146

d	1

Table of Stats From Input	Data

Dose	N	Obs Mean	Obs Std Dev

0	8	100	83.2

10	8	91.17	47.97

30	8	51.41	43.48

100	8	22.79	29.98

Estimated Values of	Interest
Dose Est Mean Est Std Scaled Residual

0	108.9	68.33	-0.3686

10	81.68	57.01	0.4706

30	50.3	42.02	0.0748

100	24.61	26.79	-0.192

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-316 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-141.9834	5	293.9669

A2	-137.5818	8	291.1637

A3	-138.3482	6	288.6964

R	-146.9973	2	297.9946

4	-138.3488	5	286.6976

Additive constant for all log-likelihoods =	-29.41. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

18 . 83
8 .803
1. 533
0.001216

p-value

0.004459
0.03203
0.4647
0.9722

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 0.100000

Risk Type = Relative deviation
Confidence Level = 0.950000

BMD =	3.55883

BMDL =	1.71399

This document is a draft for review purposes only and does not constitute Agency policy.

E-317 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.3.11.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

150

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial	Parameter Values

lalpha	=	8.007 5

rho	=	0

beta_0	=	100

beta~l	=	-0.352259

beta_2	=	-0.0585481

beta 3	=	0

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -beta 2 -beta 3

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha	rho	beta 0	beta 1

lalpha 1	-0.99	0.031	-0.016

rho -0.99	1	-0.034	0.022

beta_0 0.031	-0.034	1	-0.84

beta 1 -0.016	0.022	-0.84	1

Parameter Estimates

Variable
lalpha
rho

beta_0
beta_l
beta_2
beta 3

Estimate

2.92328
1.18295
89.841
-0.675682
0
0

Std. Err.
1.7394
0. 423359
13.7418
0.175538
NA
NA

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-0.485884
0.353177
62.9076
-1.01973

6.33243
2.01271
116.774
-0.331634

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	8	100	89.8	83.2	61.7	0.466

10	8	91.2	83.1	48	58.9	0.388

30	8	51.4	69.6	43.5	53	-0.968

100	8	22.8	22.3	30	27	0.0543

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij|

This document is a draft for review purposes only and does not constitute Agency policy.

E-319 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-141.983433	5	293.966865

A2	-137.581833	8	291.163667

A3	-138.348184	6	288.696368

fitted	-139.728204	4	287.456407

R	-146.997301	2	297.994602

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

18.8309
8.8032
1.5327
2.76004

0.004459
0.03203
0.4647
0.2516

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Relative risk

Confidence level =	0.95

BMD =	13.2963

BMDL =	10.6163

This document is a draft for review purposes only and does not constitute Agency policy.

E-320 DRAFT—DO NOT CITE OR QUOTE

-------
l E.3.11.5. Figure for Additional Model Presented: Polynomial, 3-Degree

Polynomial Model with 0.95 Confidence Level

dose

2	16:30 04/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-321 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.3.12. Franc et al., 2001: L-E Rats, Relative Thymus Weight
E.3.12.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

0.394

301.666

6.406E+00

2.122E+00



exponential (M3)

2

0.394

301.666

6.406E+00

2.122E+00

power hit bound (d = 1)

exponential
(M4)b

1

0.317

302.808

3.520E+00

1.067E+00



exponential (M5)

0

N/A

303.805

1.280E+01

1.450E+00



Hill

0

N/A

303.805

1.195E+01

9.965E-01



linear

2

0.236

302.690

1.429E+01

9.087E+00



polynomial, 3-
degree

2

0.236

302.690

1.429E+01

9.087E+00



power

2

0.236

302.690

1.429E+01

9.087E+00

power bound hit (power = 1)

power,
unrestricted

1

0.175

303.643

1.297E+00

2.703E-08

unrestricted (power = 0.454)

a Constant variance model selected (p = 0.5063)
b Best-fitting model, BMDS output presented in this appendix

E.3.12.2. Output for Selected Model: Exponential (M4)

Franc et al., 2001: L-E Rats, Relative Thymus Weight

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\92_Franc_2001_LE_RelThyWt_ExpCV_l.(d)

Gnuplot Plotting File:

Fri Apr 16 16:30:58 2010

Figure 5, L-E rats, relative thymus weight

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-322 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

Specified

lnalpha	8.1814

rho(S)	0

a	105

b	0.0413945

c	0.3173

d	1

Parameter Estimates
Variable	Model 4

lnalpha	8.21275

rho	0

a	106.57

b	0.0425967

c	0.28189

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	8	100	54.72

10	8	95.41	70.46

30	8	38.69	47.97

100	8	34.98	77.96

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	106.6	60.73	-0.306

10	80.03	60.73	0.7164

30	51.36	60.73	-0.5902

100	31.12	60.73	0.1798

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-323 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-146.9024	5	303.8049

A2	-145.7361	8	307.4723

A3	-146.9024	5	303.8049

R	-150.6049	2	305.2098

4	-147.404	4	302.8079

Additive constant for all log-likelihoods =	-29.41. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

9.738
2 . 333
2 . 333
1. 003

p-value

0.1362
0.5063
0.5063
0.3166

The p-value for Test 1 is greater than .05. There may not be a
diffence between responses and/or variances among the dose levels
Modelling the data with a dose/response curve may not be appropriate.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 0.100000

Risk Type = Relative deviation
Confidence Level = 0.950000

BMD =	3.52038

This document is a draft for review purposes only and does not constitute Agency policy.

E-324 DRAFT—DO NOT CITE OR QUOTE

-------
1	E'.MDL =	1.06729

2	E.3.12.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

dose

16:30 04/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-325 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.3.13. Franc et al., 2001: HAV Rats, Relative Thymus Weight
E.3.13.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

C

2

0.682

261.694

1.366E+01

8.014E+00



exponential (M3)

2

0.682

261.694

1.366E+01

8.014E+00

power hit bound (d = 1)

exponential
(M4)b

1

0.512

263.358

8.820E+00

3.219E+00



exponential (M5)

0

N/A

264.927

1.776E+01

3.500E+00



Hill

0

N/A

264.927

1.701E+01

2.729E+00



linear

2

0.543

262.148

1.919E+01

1.373E+01



polynomial, 3-
degree

2

0.543

262.148

1.919E+01

1.373E+01



power

2

0.543

262.148

1.919E+01

1.373E+01

power bound hit (power = 1)

power,
unrestricted

1

0.381

263.694

8.127E+00

1.406E-01

unrestricted (power = 0.665)

a Constant variance model selected (p = 0.4331)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.13.2. Output for Selected Model: Exponential (M2)

Franc et al., 2001: HAV Rats, Relative Thymus Weight

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\93_Franc_2001_HW_RelThyWt_ExpCV_l.(d)

Gnuplot Plotting File:

Fri Apr 16 16:31:40 2010

Figure 5, H/W rats, relative thymus weight

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.

This document is a draft for review purposes only and does not constitute Agency policy.

E-326 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model 3 is nested within Model 5.
Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 2

Specified

lnalpha	6.96647

rho(S)	0

a	59.5084

b	0.00715458

c	0

d	1

Parameter Estimates
Variable	Model 2

lnalpha	6.99043

rho	0

a	99.7761

b	0.00771341

c	0

d	1

Table of Stats From Input Data
Dose	N	Obs Mean	Obs Std Dev

0	8	100	35.98

10	8	97.53	32.98

30	8	71.02	23.99

100	8	49.29	43.48

Estimated Values of Interest
Est Mean	Est Std	Scaled Residual

0	99.78	32.96	0.01921

10	92.37	32.96	0.4426

30	79.16	32.96	-0.6986

100	46.14	32.96	0.271

Other models for which likelihoods are calculated:

This document is a draft for review purposes only and does not constitute Agency policy.

E-327 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma^2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-127.4636	5	264.9271

A2	-126.0925	8	268.185

A3	-127.4636	5	264.9271

R	-132.935	2	269.87

2	-127.8469	3	261.6939

Additive constant for all log-likelihoods =	-29.41. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Test

2 :

Test

3:

Test

4 :

Does response and/or variances differ among Dose levels? (A2 vs. R)

Are Variances Homogeneous? (A2 vs. A1)

Are variances adeguately modeled? (A2 vs. A3)

Does Model 2 fit the data? (A3 vs. 2)

Test

Test 1
Test 2
Test 3
Test 4

Tests of Interest

-2*log(Likelihood Ratio)

13. 69
2.742
2.742
0.7668

D. F.

p-value

0.03336
0.4331
0.4331
0.6815

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 4 is greater than .1. Model 2 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 0.100000

Risk Type = Relative deviation
Confidence Level = 0.950000

BMD =	13.6594

This document is a draft for review purposes only and does not constitute Agency policy.

E-328 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E'.MDL =	8.0137 3

E.3.13.3. Figure for Selected Model: Exponential (M2)

Exponential_beta Model 2 with 0.95 Confidence Level

a>

CO
c
o

Q.

CO
CD

a:

c
m

CD

140

120

100

80

60

40

20

Exponential

BMD

20

40

60

80

100

dose

16:31 04/16 2010

E.3.13.4. Output for Additional Model Presented: Exponential (M4)
Franc et al., 2001: H/W Rats, Relative Thymus Weight

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\93_Franc_2001_HW_RelThyWt_ExpCV_l.(d)

Gnuplot Plotting File:

Fri Apr 16 16:31:40 2010

Figure 5, H/W rats, relative thymus weight

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign
exp{sign
[c—(c—1)
[c—(c—1)

b * dose}
(b * dose)^d}
exp{-b * dose}]
exp{-(b * dose)^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

This document is a draft for review purposes only and does not constitute Agency policy.

E-329 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

Specified

lnalpha	6.96647

rho(S)	0

a	105

b	0.03169

c	0.447105

d	1

Parameter Estimates
Variable	Model 4

lnalpha	6.97993

rho	0

a	103.091

b	0.02048

c	0.394904

d	1

Table of Stats From Input	Data

use	N	Obs Mean	Obs Std Dev

0	8	100	35.98

10	8	97.53	32.98

30	8	71.02	23.99

100	8	49.29	43.48

Estimated Values of	Interest
Dose Est Mean Est Std Scaled Residual

0	103.1	32.78	-0.2667

10	91.54	32.78	0.5166

30	74.46	32.78	-0.2961

100	48.76	32.78	0.04621

This document is a draft for review purposes only and does not constitute Agency policy.

E-330 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Other models for which	likelihoods are calculated:

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-127.4636	5	264.9271

A2	-126.0925	8	268.185

A3	-127.4636	5	264.9271

R	-132.935	2	269.87

4	-127.6789	4	263.3577

Additive constant for all log-likelihoods =	-29.41. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

13. 69
2.742
2.742
0.4306

D. F.

p-value

0.03336
0.4331
0.4331
0.5117

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 0.100000

Risk Type = Relative deviation

This document is a draft for review purposes only and does not constitute Agency policy.

E-331 DRAFT—DO NOT CITE OR QUOTE

-------
Confidence Level = 0.950000

BMD =	8.82023

BMDL =	3.21928

E.3.13.5. Figure for Additional Model Presented: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

dose

16:31 04/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-332 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

E.3.14. Hojo et al., 2002: DRL Reinforce Per Minute
E.3.14.1. Summary Table of BMDS Modeling Results

Model3

Degrees of
Freedom

X2 P-
Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Hill

0

N/A

6.465

2.060E+01

1.713E-05



linear b

2

0.008

9.552

2.677E+02

1.100E+02



polynomial, 3-
degree

2

0.008

9.552

2.677E+02

1.100E+02



power

2

0.008

9.552

2.677E+02

1.100E+02

power bound hit (power =1)

power, unrestricted

1

0.025

6.780

2.187E+00

4.612E-08

unrestricted (power = 0.089)

exponential (M2)

2

0.006

9.894

3.043E+02

1.505E+02



exponential (M3)

2

0.006

9.894

3.043E+02

1.505E+02

power hit bound (d = 1)

exponential (M4)0

1

0.062

5.241

1.734E+01

3.827E-02



exponential (M5)

0

N/A

6.465

2.140E+01

1.240E-05



a Constant variance model selected (p = 0.4321)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.14.2. Output for Selected Model: Linear

Hojo et al., 2002: DRL Reinforce Per Minute

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\1\20_Hojo_2002_DRLrein_LinearCV_l.(d)

Gnuplot Plotting File: C:\l\20_Hojo_2002_DRLrein_LinearCV_l.plt

Tue Feb 16 17:29:42 2010

Table 5

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/'2 + ...

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Signs of the polynomial coefficients are not restricted
A constant variance model is fit

Total number of dose groups = 4

This document is a draft for review purposes only and does not constitute Agency policy.

E-333 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
alpha =	0.337763

rho =	0 Specified

beta_0 =	-0.4 04

beta~l = 0.00249615

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha	beta 0	beta 1

alpha 1	-1.4e-008	2.2e-008

beta_0 -1.4e-008	1	-0.69

beta 1 2.2e-008	-0.69	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
beta_0
beta 1

Estimate

0. 435671
-0.372098
0.00246548

Std. Err.
0.134451
0.198702
0. 00211361

Lower Conf. Limit
0.172152
-0.761547
-0.00167711

Upper Conf. Limit

0 . 6 9 919
0. 017352
0. 00660807

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	5	-0.814	-0.372	0.448	0.66	-1.5

20	5	-0.364	-0.323	0.821	0.66	-0.14

60	6	0.374	-0.224	0.54	0.66	2.22

180	5	-0.163	0.0717	0.443	0.66	-0.795

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-334 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	3.115550	5	3.768900

A2	4.489557	8	7.020886

A3	3.115550	5	3.768900

fitted	-1.775882	3	9.551763

R	-2.435087	2	8.870174

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adequately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	13.8493	6	0.03137

Test 2	2.74801	3	0.4321

Test 3	2.74801	3	0.4321

Test 4	9.78286	2	0.007511

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	267.718

BMDL =	110.032

This document is a draft for review purposes only and does not constitute Agency policy.

E-335 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.14.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

17:29 02/16 2010

E.3.14.4. Output for Additional Model Presented: Exponential (M4)
Hojo et al., 2002: DRL Reinforce Per Minute

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\21_Hojo_2002_DRLrein_ExpCV_l.(d)

Gnuplot Plotting File:

Tue Feb 16 17:30:21 2010

Table 5, values adjusted by a constant to allow exponential model

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign * b * dose}
exp{sign * (b * dose)^d}
[c-(c-l) * exp{-b * dose}]
[c-(c-l) * exp{-(b * dose)^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-336 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

Specified

lnalpha	-1.29672

rho(S)	0

a	0.0817

b	0.00880867

c	16.3733

d	1

Parameter Estimates
Variable	Model 4

lnalpha	-1.13136

rho	0

a	0.0542868

b	0.0525016

c	18.5072

d	1

Table of Stats From Input Data
Dose	N	Obs Mean	Obs Std Dev

0	5	0.086	0.448

20	5	0.536	0.821

60	6	1.274	0.54

180	5	0.737	0.443

Estimated Values of Interest
Est Mean	Est Std	Scaled Residual

0	0.05429	0.568	0.1249

20	0.6721	0.568	-0.5359

60	0.964	0.568	1.337

180	1.005	0.568	-1.054

Other models for which likelihoods are calculated:

This document is a draft for review purposes only and does not constitute Agency policy.

E-337 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma^2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	3.11555	5	3.7689

A2	4.489557	8	7.020886

A3	3.11555	5	3.7689

R	-2.435087	2	8.870174

4	1.379312	4	5.241376

Additive constant for all log-likelihoods =	-19.3. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

13. 85
2.748
2.748
3.472

D. F.

p-value

0.03137
0.4321
0.4321
0.0624

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is less than .1. Model 4 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

This document is a draft for review purposes only and does not constitute Agency policy.

E-338 DRAFT—DO NOT CITE OR QUOTE

-------
BMD =	17.3391

BMDL = 0.0382689

E.3.14.5. Figure for Additional Model Presented: Exponential (M4)

(D
CO
c
o
Q.
CO
(D

(Z
c
(0
(D

Exponential_beta Model 4 with 0.95 Confidence Level

i	1	1	1	1—

80 100
dose

120

140

160

180

17:30 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-339 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.3.15. Hojo et al., 2002: DRL Response Per Minute

E.3.15.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Hill

0

N/A

126.353

1.646E+01

1.800E-13



linear

2

0.004

132.825

2.067E+02

9.757E+01



polynomial, 3-
degree

2

0.004

132.825

2.067E+02

9.757E+01



power

2

0.004

132.825

2.067E+02

9.757E+01

power bound hit (power = 1)

power,
unrestricted

2

0.741

122.455

1.800E+04

error

unrestricted (power = 0)

exponential (M2)

2

0.568

122.985

6.184E+00

error



exponential (M3)

2

0.568

122.985

6.184E+00

error

power hit bound (d = 1)

exponential
(M4)b

1

0.479

124.356

4.775E+00

2.704E-01



exponential (M5)

0

N/A

126.353

1.118E+01

2.127E-01



a Constant variance model selected (p = 0.3004)
b Best-fitting model, BMDS output presented in this appendix

E.3.15.2. Output for Selected Model: Exponential (M4)

Hojo et al., 2002: DRL Response Per Minute

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\23_Hojo_2002_DRLresp_ExpCV_l.(d)

Gnuplot Plotting File:

Tue Feb 16 17:31:24 2010

Table 5, values adjusted by a constant to allow exponential model

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-340 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

Specified

lnalpha	4.51689

rho(S)	0

a	24.6362

b	0.0212679

c	0.0184785

d	1

Parameter Estimates
Variable	Model 4

lnalpha	4.54075

rho	0

a	23.465

b	0.12859

c	0.100615

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	5	23.46	7.986

20	5	4.013	10.96

60	6	0.478	7.194

180	5	4.594	15.23

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	23.47	9.683	-0.0004677

20	3.973	9.683	0.009182

60	2.37	9.683	-0.4787

180	2.361	9.683	0.5157

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-341 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-57.92733	5	125.8547

A2	-56.09669	8	128.1934

A3	-57.92733	5	125.8547

R	-64.49611	2	132.9922

4	-58.17787	4	124.3557

Additive constant for all log-likelihoods =	-19.3. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

16.8
3. 661
3. 661
0.5011

p-value

0.01005
0.3004
0.3004
0.479

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	4.77493

This document is a draft for review purposes only and does not constitute Agency policy.

E-342 DRAFT—DO NOT CITE OR QUOTE

-------
1	E'.MDL =	0.27 04 4 7

2

3

4	E.3.15.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

dose

5	17:31 02/16 2010

6

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-343 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

E.3.16. Kattainen et al., 2001: 3rd Molar Eruption, Female
E.3.16.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

logistic

3

0.292

89.060

1.941E+02

1.390E+02

negative intercept (intercept =
-1.508)

log-logistica

3

0.923

85.535

4.763E+01

2.481E+01

slope bound hit (slope = 1)

log-probit

3

0.390

88.231

1.574E+02

9.512E+01

slope bound hit (slope =1)

probit

3

0.306

88.919

1.858E+02

1.370E+02

negative intercept (intercept =
-0.927)

multistage, 4-
degree

3

0.641

86.798

8.677E+01

5.520E+01

final B = 0

log-logistic,
unrestricted b

2

0.952

87.157

2.599E+01

1.730E+00

unrestricted (slope = 0.794)

log-probit,
unrestricted

2

0.941

87.179

2.813E+01

2.334E+00

unrestricted (slope = 0.478)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.3.16.2. Output for Selected Model: Log-Logistic

Kattainen et al., 2001: 3rd Molar Eruption, Female

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\24_Katt_2001_Erup_LogLogistic_BMRl.(d)

Gnuplot Plotting File: C:\l\24_Katt_2001_Erup_LogLogistic_BMRl.plt

Tue_Feb 16 17:31:52 2010

Figure 2

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

Slope parameter is restricted as slope >= 1
Total number of observations = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

This document is a draft for review purposes only and does not constitute Agency policy.

E-344 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

Default Initial Parameter Values
background =	0.0625

intercept =	-6.063

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept

background	1	-0.56

intercept	-0.56	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf. Limit Upper Conf. Limit

background 0.0846785 *	*	*

intercept -6.06063 *	*	*

slope 1 *	*	*

Indicates that this value is not calculated.

Model
Full model
Fitted model

Analysis of Deviance Table

Deviance

Log(likelihood
-40.5286
-40.7674

Param's

5
2

Test d.f.

0. 477533

P-value

0.9238

Reduced

model

50.7341

1



20.411

4 0.0004142



AIC:

85.5347













Goodness

of

Fit















Scaled

Dose

Est. Prob.

Expected

Observed

Size

Residual

0.0000

0.0847

1. 355

1.

000

16

-0.319

30.0000

0.1445

2 .457

3.

000

17

0.374

100.0000

0.2578

3. 867

4 .

000

15

0. 078

300.0000

0. 4615

5.538

6 .

000

12

0.267

1000.0000

0.7254

13.782

13.

000

19

-0.402

Chi ^2

0. 4E

d. f.

3

P-value

0.9231

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	47.6274

BMDL =	24.8121

This document is a draft for review purposes only and does not constitute Agency policy.

E-345 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.16.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

"O

(D

-t—<

o

CD
!t=
<

0.6

0.4

0.2

200

400	600

dose

800

1000

17:31 02/16 2010

E.3.16.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
Kattainen et al., 2001: 3rd Molar Eruption, Female

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\24_Katt_2001_Erup_LogLogistic_U_BMRl.(d)

Gnuplot Plotting File: C:\l\24_Katt_2001_Erup_LogLogistic_U_BMRl.plt

Tue Feb 16 17:31:53 2010

Figure 2

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose)

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-346 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of observations = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background =	0.0625

intercept =	-4.71231

slope =	0.782659

Asymptotic Correlation Matrix of Parameter Estimates
background intercept	slope

background	1	-0.48	0.39

intercept	-0.48	1	-0.98

slope	0.39	-0.98	1

Parameter Estimates

Variable
background
intercept
slope

Estimate

0.0633217
-4.78282
0.793723

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)
-40.5286
-40.5783
-50.7341

Param's

5
3
1

Deviance Test d.f.

0.0994416
20.411

P-value

0. 9515
0.0004142

AIC:

87 .1566

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000

0.0633

1. 013

1.

000

16

-0.013

30.0000

0.1670

2.840

3.

000

17

0.104

100.0000

0.2924

4 . 387

4 .

000

15

-0.219

300.0000

0. 4721

5.666

6 .

000

12

0.193

1000.0000

0.6892

13.095

13.

000

19

-0.047

Chi ^2 = 0.10

d.f.

= 2 P

-value

= 0.9518





Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

This document is a draft for review purposes only and does not constitute Agency policy.

E-347 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

Confidence level =	0.95

BMD =	25.986

BMDL =	1.73001

E.3.16.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

dose

17:31 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-348 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

E.3.17. Kattainen et al., 2001: 3rd Molar Length, Female
E.3.17.1. Summary Table of BMDS Modeling Results

Model3

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

3

<0.0001

-122.954

4.027E+02

2.366E+02



exponential (M3)

3

<0.0001

-122.954

4.027E+02

2.366E+02

power hit bound (d = 1)

exponential (M4)

2

<0.0001

-80.747

error

error



exponential (M5)

1

<0.0001

-78.747

error

error



Hillb

2

0.013

-151.152

4.052E+00

2.144E+00

n lower bound hit (n = 1)

linear

3

<0.0001

-122.325

4.659E+02

2.963E+02



polynomial, 4-
degree

3

<0.0001

-122.325

4.659E+02

2.963E+02



power

3

<0.0001

-122.325

4.659E+02

2.963E+02

power bound hit (power =1)

Hill, unrestricted0

1

0.087

-154.939

1.913E-02

1.928E-04

unrestricted (n = 0.197)

power, unrestricted

2

0.250

-157.093

9.098E-03

9.097E-03

unrestricted (power = 0.169)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.17.2. Output for Selected Model: Hill

Kattainen et al., 2001: 3rd Molar Length, Female

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\25_Katt_2001_Length_Hill_l.(d)

Gnuplot Plotting File: C:\l\25_Katt_2001_Length_Hill_l.plt

Tue Feb 16~17:32:21 2010

Figure 3 female only

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose

Power parameter restricted to be greater than 1

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))

This document is a draft for review purposes only and does not constitute Agency policy.

E-349 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho

intercept

-2 . 37155

0

1. 85591
-0.507874
0. 826204
27 . 3305

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
rho

intercept

lalpha
1

-0. 98
-0.16
0.84
-0.37

rho

-0. 98
1

0.2
-0.79
0.39

intercept

-0.16
0.2
1

-0.31
-0.11

v

0.84
-0.79
-0.31
1

-0.48

k

-0.37
0.39
-0.11
-0.48
1

Parameter Estimates

Variable
lalpha
rho

intercept

k

Estimate
3.34561
-14.3325
1. 8548
-0.441166
1

24.0343

Std. Err.
1.40443
2 . 62129
0.0159017
0.058852
NA

7.84495

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

0.592981

-19.4701
1.82364
-0.556513

8.65852

6.09824
-9.19484
1.88597
-0.325818

39.4101

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	16	1.86	1.85	0.0661	0.0637	0.0692

30	17	1.58	1.61	0.185	0.176	-0.768

100	15	1.6	1.5	0.265	0.293	1.28

300	12	1.5	1.45	0.221	0.378	0.527

1000	19	1.35	1.42	0.515	0.423	-0.783

Model Descriptions for likelihoods calculated

This document is a draft for review purposes only and does not constitute Agency policy.

E-350 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
56.758717
85.856450
84.934314
80.575940
45.373551

10

7
5
2

AIC
-101.517434
-151.712901
-155.868628
-151.151880
-86.747101

Test

1

Test

2

Test

3

Test

4

(Note:

Explanation of Tests

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

80.9658
58 .1955
1.84427
8 .71675

<.0001
<.0001
0.6053
0.0128

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	4 . 05231

BMDL =	2.14357

This document is a draft for review purposes only and does not constitute Agency policy.

E-3 51 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.17.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

1.9

1.8

1.7

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho

intercept

-2 . 37155

0

1. 85591
-0.507874
0. 826204
27 . 3305

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho

intercept

lalpha
1

-0. 98
-0.18
0.18
-0.28
-0.011

rho

-0. 98
1

0.22
-0.18
0.29
0. 011

intercept

-0.18
0.22
1

-0.025
-0.059
0.0019

v

0.18
-0.18
-0.025
1

0.51
-0. 96

n

-0.28
0.29
-0.059
0.51
1

-0.71

k

-0.011
0. 011
0.0019
-0. 96
-0.71
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

Estimate
3.21882
-14.0862
1. 85564
-2 .48572
0.196925
1. 92 967e + 006

Std. Err.
1.4221
2.68292
0.0160224
2 .89658
0.0499318
1.60869e+007

Lower Conf. Limit

0. 431563
-19.3446
1.82424
-8.16291
0.0990606
-2.96e+007

Upper Conf. Limit

6.00607

82777
88704
19148
2947 9

3.34593e+007

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0
30
100
300
1000

16

17
15
12
19

1.86
1. 58
1. 6
1. 5
1. 35

1.86
1. 6
1. 54
1.48
1. 4

0.0661

0.185
0.265
0.221
0.515

0.0643
0.18
0.234
0.316
0.471

0.0164

-0.598
0. 857
0.259
-0.466

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-353 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	56.758717	6	-101.517434

A2	85.856450	10	-151.712901

A3	84.934314	7	-155.868628

fitted	83.469680	6	-154.939361

R	45.373551	2	-86.747101

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	80.9658	8	<.0001

Test 2	58.1955	4	<.0001

Test 3	1.84427	3	0.6053

Test 4	2.92927	1	0.08699

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.0191282

BMDL =	0.0001928

This document is a draft for review purposes only and does not constitute Agency policy.

E-3 54 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.17.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

17:32 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-355 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

E.3.18. Keller et al., 2007: Missing Mandibular Molars, CBA J
E.3.18.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

1

0.105

52.490

7.293E+01

2.027E+01



logistic

2

0.320

50.095

7.168E+01

5.142E+01

negative intercept (intercept =
-3.372)

log-logistic

1

0.105

52.524

9.278E+01

5.273E+01



log-probit

1

0.105

52.524

8.849E+01

5.297E+01



multistage, 1-
degreea

3

0.276

49.409

2.778E+01

1.884E+01



multistage, 2-
degree

1

0.126

51.515

4.619E+01

2.214E+01



multistage, 3-
degree

1

0.141

51.222

4.253E+01

2.212E+01



probit

2

0.325

50.032

6.848E+01

4.775E+01

negative intercept (intercept =
-1.851)

Weibull

1

0.108

52.216

6.079E+01

2.078E+01



a Best-fitting model, BMDS output presented in this appendix

E.3.18.2. Output for Selected Model: Multistage, 1-Degree

Keller et al., 2007: Missing Mandibular Molars, CBA J

Multistage Model. (Version: 3.0; Date: 05/16/2008)

Input Data File: C:\1\26_Keller_2007_Molars_Multil_l.(d)

Gnuplot Plotting File: C:\l\26_Keller_2007_Molars_Multil_l.plt

Tue Feb_16 17:32:56 2010

Table 1 using mandibular molars only

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0

This document is a draft for review purposes only and does not constitute Agency policy.

E-356 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) = 1.02909e+017

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.00379264	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-21.5798	4

Fitted model -23.7044 1	4.24924 3	0.2358

Reduced model -71.326 1	99.4926 3	<.0001

AIC:	49.4088

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.0000

0. 000

0. 000

29

0. 000

10.0000

0.0372

0. 856

2 . 000

23

1.260

100.0000

0.3156

9.153

6. 000

29

-1.260

1000.0000

0.9775

29.324

30.000

30

0. 832

Chi ^2 = 3.87	d.f. = 3	P-value = 0.2762

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	27.7803

BMDL =	18.84 47

This document is a draft for review purposes only and does not constitute Agency policy.

E-357 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

E'.MDU =

41.7256

Taken together, (18.8447, 41.7256) is a 90	% two-sided confidence

interval for the EMD

E.3.18.3. Figure for Selected Model: Multistage, 1-Degree

Multistage Model with 0.95 Confidence Level

-i—i—		—i—i—¦ | '

Multista

1

"O

(D

-t—<

o

CD
!t=
<
c
o

o
m

0.6

0.4

0.2

400	600

dose

17:32 02/16 2010

This document is a draft for review purposes only and does not constitute Agency policy.

E-358 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.3.19. Kociba et al., 1978: Urinary Coproporphyrin, Females
E.3.19.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

<0.0001

84.006

7.054E+01

4.341E+01



exponential (M3)

2

<0.0001

84.006

7.054E+01

4.341E+01

power hit bound (d = 1)

exponential
(M4)b

1

0.040

70.556

1.625E+00

7.300E-01



exponential (M5)

0

N/A

69.092

3.128E+00

1.024E+00



Hill

0

N/A

69.047

6.677E+00

error



linear

2

<0.0001

83.713

6.195E+01

3.112E+01



polynomial, 3-
degree

2

<0.0001

83.713

6.195E+01

3.112E+01



power

2

<0.0001

83.713

6.195E+01

3.112E+01

power bound hit (power =1)

power,
unrestricted

1

0.001

78.260

7.808E-01

1.693E-08

unrestricted (power = 0.306)

a Non-constant variance model selected (p = 0.0298)
b Best-fitting model, BMDS output presented in this appendix

E.3.19.2. Output for Selected Model: Exponential (M4)

Kociba et al., 1978: Urinary Coproporphyrin, Females

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\29_Kociba_l978_Copro_Exp_l.(d)

Gnuplot Plotting File:

Tue Feb 16 17:34:45 2010

Table2-UrinaryCoproporphyrin

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-359 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	-5.58269

rho	2.98472

a	8.17

b	0.0259469

c	2.23623

d	1

Parameter Estimates
Variable	Model 4

lnalpha	-4.94473

rho	2.76088

a	8.93039

b	0.136554

c	1.9753

d	1

Table of Stats From Input	Data

use N Obs Mean	Obs Std Dev

0 5 9.8	1.3

15 8.6	2

10 5 16.4	4.7

100 5 17.4	4

Estimated Values of	Interest

Dose Est Mean Est Std	Scaled Residual

0	8.93	1.733	1.122

1	10.04	2.038	-1.582
10	15.42	3.683	0.5967

100	17.64	4.436	-0.1211

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-360 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest
Model	Log(likelihood)	DF	AIC

A1

-31.69739

5

73

39478

A2

-27 . 21541

8

70

43081

A3

-28.16434

6

68

32868

R

-41. 73188

2

87

46376

4

-30.27804

5

70

55608

Additive constant for all log-likelihoods =	-18.38. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Does response

and/or variances differ among Dose levels? (A2 vs. R)

Test

2 :

Are Variances

Homogeneous? (A2 vs. A1)

Test

3:

Are variances

adequately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

29. 03
8 . 964
1.898
4 . 227

p-value

C 0.0001
0.02977
0.3872
0.03978

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is less than .1. Model 4 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	1.62505

BMDL =	0.729987

This document is a draft for review purposes only and does not constitute Agency policy.

E-361 DRAFT—DO NOT CITE OR QUOTE

-------
l E.3.19.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

dose

2	17:34 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-362 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.3.20. Kociba et al., 1978: Uroporphyrin per Creatinine, Female
E.3.20.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

0.661

-93.561

4.357E+01

3.328E+01



exponential (M3)

2

0.661

-93.561

4.357E+01

3.328E+01

power hit bound (d = 1)

exponential (M4)

1

0.576

-92.078

1.719E+01

5.516E+00



exponential (M5)

0

N/A

-90.190

1.080E+01

5.613E+00



Hill

0

N/A

-90.190

1.099E+01

5.088E+00



linear b

2

0.720

-93.735

3.522E+01

2.500E+01



polynomial, 3-
degree

2

0.720

-93.735

3.522E+01

2.500E+01



power

2

0.720

-93.735

3.522E+01

2.500E+01

power bound hit (power =1)

power,
unrestricted

1

0.515

-91.967

2.274E+01

3.334E+00

unrestricted (power = 0.731)

a Constant variance model selected (p = 0.4919)
b Best-fitting model, BMDS output presented in this appendix

E.3.20.2. Output for Selected Model: Linear

Kociba et al., 1978: Uroporphyrin per Creatinine, Female

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\1\28_Kociba_l978_Uropor_LinearCV_l.(d)

Gnuplot Plotting File: C:\l\28_Kociba_197 8_Uropor_LinearCV_l.plt

Tue Feb_16 17:347l2 2010

Table 2

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/'2 + ...

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Signs of the polynomial coefficients are not restricted
A constant variance model is fit

Total number of dose groups = 4

Total number of records with missing values = 0

This document is a draft for review purposes only and does not constitute Agency policy.

E-363 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
alpha = 0.0030385

rho =	0 Specified

beta_0 =	0.154759

beta 1 = 0.0014231

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha	beta 0	beta 1

alpha 1	-2.2e-009	3.5e-009

beta_0 -2.2e-009	1	-0.55

beta 1 3.5e-009	-0.55	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
beta_0
beta 1

Estimate

0. 00251184
0.154759
0.0014231

Std. Err.

0.000794315
0.0134422
0.000267497

Lower Conf. Limit

0.000955015
0.128413
0. 000898818

Upper Conf. Limit

0. 00406867
0.181105
0. 00194739

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	5	0.157	0.155	0.05	0.0501	0.1

1	5	0.143	0.156	0.037	0.0501	-0.588
10	5	0.181	0.169	0.053	0.0501	0.536

100	5	0.296	0.297	0.074	0.0501	-0.0477

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

This document is a draft for review purposes only and does not constitute Agency policy.

E-364 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

Model	Log(likelihood)	# Param's	AIC

A1	50.195349	5	-90.390697

A2	51.400051	8	-86.800103

A3	50.195349	5	-90.390697

fitted	49.867385	3	-93.734769

R	41.049755	2	-78.099510

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	20.7006	6	0.002076

Test 2	2.40941	3	0.4919

Test 3	2.40941	3	0.4919

Test 4	0.655928	2	0.7204

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	35.2176

BMDL =	25.0024

This document is a draft for review purposes only and does not constitute Agency policy.

E-365 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.20.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

17:34 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-366 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

E.3.21. Latchoumycandane and Mathur, 2002: Sperm Production

E.3.21.1. Summary Table of BMDS Modeling Results

Model3

Degrees of
Freedom

z2 P-
Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

<0.0001

95.106

7.640E+01

3.992E+01



exponential (M3)

2

<0.0001

95.106

7.640E+01

3.992E+01

power hit bound (d = 1)

exponential (M4)

1

0.699

75.263

2.435E-01

1.016E-01



exponential (M5)

0

N/A

77.263

3.697E-01

1.016E-01



Hillb

1

0.859

75.144

1.450E-01

1.559E-02

n lower bound hit (n = 1)

linear

2

<0.0001

95.308

8.275E+01

4.852E+01



polynomial, 3-
degree

2

<0.0001

95.308

8.275E+01

4.852E+01



power

2

<0.0001

95.308

8.275E+01

4.852E+01

power bound hit (power =1)

Hill, unrestricted0

0

N/A

77.113

6.943E-02

2.060E-06

unrestricted (n = 0.709)

power, unrestricted

1

0.499

75.570

2.706E-07

2.706E-07

unrestricted (power = 0.067)

a Constant variance model selected (p = 0.8506)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.21.2. Output for Selected Model: Hill

Latchoumycandane and Mathur, 2002: Sperm Production

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\30_Latch_2002_Sperm_HillCV_l.(d)

Gnuplot Plotting File: C:\l\30_Latch_2002_Sperm_HillCV_l.plt

_Tue Feb 16 18:13:20 2010

(xlO^) Table 1 without Vitamin E

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter restricted to be greater than 1

This document is a draft for review purposes only and does not constitute Agency policy.

E-367 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

A constant variance model is fit

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho

intercept

7 . 23328

0

22.19
-9.09
1.80484
0. 697086

Specified

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
intercept

alpha
1

. 3e-010
3e-008
. 3e-009

intercept
6.3e-010
1

-0.78
-0.23

v

3e-008
-0.78
1

-0.17

k

. 3e-009
-0.23
-0.17
1

Parameter Estimates

Variable
alpha
intercept

k

Estimate

6.03567
22.1885
-9.00869
1

0.386669

Std. Err.
1.74235
1.00316
1.26801
NA

0.265663

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

9.45061
24 .1547
-6.52343

-0.134021

0.907359

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

1
10

100

22 . 2
15.7
13.7
13.1

22 . 2
15.7
13.5
13.2

2 . 67
2 . 65
2.19
3.16

2.46
2.46
2.46
2.46

0.00151
-0.0218
0.134
-0.114

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

This document is a draft for review purposes only and does not constitute Agency policy.

E-368 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model

Log(likelihood)

# Param's

AIC

A1

-33.556444

5

77 . 112888

A2

-33.158811

8

82.317623

A3

-33.556444

5

77 . 112888

fitted

-33.572245

4

75.144490

R

-47 . 392394

2

98 .784788

Explanation of Tests

Test 1:

Test 2
Test 3
Test 4
(Note:

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

28.4672
0.795266
0.795266
0.031602

<.0001
0.8506
0.8506
0.8589

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than
model appears to be appropriate here

. 1.

homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.144988

BMDL =	0.0155926

This document is a draft for review purposes only and does not constitute Agency policy.

E-369 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.21.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

18:13 02/16 2010

E.3.21.4. Output for Additional Model Presented: Hill, Unrestricted
Latchoumycandane and Mathur, 2002: Sperm Production

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\30_Latch_2002_Sperm_HillCV_U_l.(d)

Gnuplot Plotting File: C:\l\30_Latch_2002_Sperm_HillCV_U_l.plt

Tue Feb 16 18:13:21 2010

(xl0^6) Table 1 without Vitamin E

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-3 70 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha =	7.23328

rho =	0 Specified

intercept =	22.19

v =	-9.09

n =	1.80484

k =	0.697086

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
intercept

alpha
1

-7 . 6e-009
8e-008
5e-008
1. 9e-008

intercept
-7.6e-009
1

-0.5

-0.015
-0.13

v

8e-008
-0.5

1

0.75
0.55

n

5e-008
-0.015
0.75
1

0.86

k

1.9e-008
-0.13
0.55
0.86
1

Parameter Estimates

Variable
alpha
intercept

Estimate

6.02773
22.19
-9.23433
0.709305
0.290697

Std.
1.
1.

Err.

74006
00231

2.02073
1.28329

0.548737

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

2.61728
20.2255
-13.1949
-1.8059
-0.784807

9. 43818
24 .1545
-5.27378
3.22451
1. 3662

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	6	22.2	22.2	2.67	2.46	2.62e-008

1	6	15.7	15.7	2.65	2.46	-1.5e-008
10	6 13.7	13.7	2.19	2.46	-4.56e-008

100	6 13.1	13.1	3.16	2.46	-3.52e-007

Degrees	of freedom for Test A3 vs fitted <= 0

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-371 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model

Log(likelihood)

# Param's

AIC

A1

-33.556444

5

77 . 112888

A2

-33.158811

8

82.317623

A3

-33.556444

5

77 . 112888

fitted

-33.556444

5

77 . 112888

R

-47 . 392394

2

98 .784788

Explanation of Tests

Test 1:

Test 2
Test 3
Test 4
(Note:

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

28.4672
0.795266
0.795266
2.84217e-014

<.0001
0.8506
0.8506
NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than
model appears to be appropriate here

.1. A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.0694325

BMDL = 2.06007e-006

This document is a draft for review purposes only and does not constitute Agency policy.

E-372 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.21.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

18:13 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-3 73 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.3.22. Li et al., 1997: FSH

E.3.22.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

8

<0.0001

1095.240

1.340E+04

1.060E+04



exponential (M3)

8

<0.0001

1095.240

1.340E+04

1.060E+04

power hit bound (d = 1)

exponential (M4)

7

<0.0001

1061.243

1.031E+03

4.015E+02



exponential (M5)

7

<0.0001

1061.243

1.031E+03

4.015E+02

power hit bound (d = 1)

Hill

7

<0.0001

1059.547

6.645E+02

error

n lower bound hit (n = 1)

linear

8

<0.0001

1078.221

5.287E+03

3.602E+03



polynomial, 8-
degree

9

<0.0001

1155.670

error

error



power b

8

<0.0001

1078.221

5.287E+03

3.602E+03

power bound hit (power = 1)

Hill, unrestricted

6

0.001

1039.902

2.809E+00

6.602E-01

unrestricted (n = 0.291)

power,
unrestricted0

7

0.002

1037.821

2.508E+00

2.525E-01

unrestricted (power = 0.279)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.22.2. Output for Selected Model: Power

Li et al., 1997: FSH

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\72_Li_l997_FSH_Pwr_l.(d)

Gnuplot Plotting File: C:\l\72_Li_1997_FSH_Pwr_l.plt

Tue Feb 16 20:07:31 2010

Figure 3: FSH in female S-D rats 24hr after dosing, 22 day old rats

The form of the response function is:

Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose

The power is restricted to be greater than or egual to 1

This document is a draft for review purposes only and does not constitute Agency policy.

E-374 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

Total number of dose groups = 10

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho
control
slope
power

9.8191

0

22.1591
26.1213
0.264963

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope

The model parameter(s) -power

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
1

-0. 99
-0.29
-0.023

rho

-0. 99
1

0.2
0. 023

control

-0.29
0.2
1

-0.35

slope

-0.023
0. 023
-0.35
1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate
3.5473
1.26137
88.9479
0. 0188972
1

Std. Err.

1. 23656
0.244246
12 . 9114
0. 00351723
NA

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

1.12369
0.782659
63.6419
0.0120035

5. 9709
1.74009
114.254
0. 0257908

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

10

23. 9

cxi

CO
CO

29.6

9 9.9

-2 . 06

3

10

22 . 2

89

48.5

9 9.9

-2 .12

10

10

85.2

89.1

94 . 3

100

-0.124

30

10

73.3

89.5

48.5

100

-0.511

100

10

126

90. 8

159

101

1.1

300

10

132

94 . 6

116

104

1.14

1000

10

117

108

51. 2

113

0.25

3000

10

304

146

154

136

3. 68

le+004

10

347

278

151

205

1. 06

3e+004

10

455

656

286

352

-1. 8

This document is a draft for review purposes only and does not constitute Agency policy.

E-375 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)

-535.687163
-496.367061
-502.709623
-535.110448
-574.835246

Param's

11

20

12
4
2

AIC
1093.374327
1032.734122
1029.419246
1078.220896
1153.670492

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

156.936

78 . 6402
12 . 6851
64 .8016

18
9

<.0001
<.0001
0.1232
<.0001

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 5286.67

This document is a draft for review purposes only and does not constitute Agency policy.

E-376 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E'.MDL = 3 6 01.91

E.3.22.3. Figure for Selected Model: Power

Power Model with 0.95 Confidence Level

0	5000	10000	15000	20000	25000	30000

dose

20:07 02/16 2010

E.3.22.4. Output for Additional Model Presented: Power, Unrestricted
Li et al„ 1997: FSH

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\72_Li_l997_FSH_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\72_Li_1997_FSH_Pwr_U_l.plt

Tue Feb 16 20:07:33 2010

Figure 3: FSH in female S-D rats 24hr after dosing, 22 day old rats

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-3 77 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

Total number of dose groups = 10

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho
control
slope
power

9.8191

0

22.1591
26.1213
0.264963

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope
power

lalpha
1

-0. 99
-0.69
-0.15
0.28

rho

-0. 99
1

0. 65
0.11
-0.26

control

-0.69
0. 65
1

-0.17
0. 024

slope

-0.15
0.11
-0.17
1

-0. 93

power

0.28
-0.26
0. 024
-0. 93
1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate
3.72156
1.17032
15.7412
24.963
0.278637

Std. Err.

1.13117
0.223249
6.97367
6. 42976
0. 0312355

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

1. 5045
0.732758
2 . 07307
12.3609
0.217417

5.93861

1. 60788
29.4094
37 . 5651
0.339857

Table of Data and Estimated Values of Interest

Table ¦

Dose



0

10

3

10

10

10

30

10

100

10

300

10

1000

10

3000

10

le+004
3e+004

10
10

Obs Mean

23. 9

22 . 2

85.2

73.3
126
132
117
304

347
455

Est Mean Obs Std Dev Est Std Dev Scaled Res.

15.7

49.6
63.2
80.1
106
138
187
248
341
457

29.6
48.5
94 . 3
48.5
159
116
51. 2
154
151
286

32 . 3

63.2
72 . 7
83. 6
98 . 4
115
137
162
195
232

0.796
-1. 38
0. 96
-0.259
0. 654
-0.164
-1. 62
1.1
0 . 0 9 9 9
-0.0271

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

This document is a draft for review purposes only and does not constitute Agency policy.

E-378 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Var{e(ij)} = Sigma/N2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)

-535.687163
-496.367061
-502.709623
-513.910636
-574.835246

Param's

11

20

12
5
2

AIC
1093.374327
1032.734122
1029.419246
1037.821272
1153.670492

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

156.936

78 . 6402
12 . 6851
22.402

18
9

<.0001
<.0001
0.1232
0.002165

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 2.5083 9

BMDL = 0.252541

This document is a draft for review purposes only and does not constitute Agency policy.

E-379 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.22.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

20:07 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-3 80 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.23. Li et al., 2006: Estradiol, 3-Day

E.3.23.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

0.147

269.146

3.044E+02

1.108E+02



exponential (M3)

2

0.147

269.146

3.044E+02

1.108E+02

power hit bound (d = 1)

exponential (M4)

1

0.341

268.212

error

error



exponential (M5)

0

N/A

270.212

error

error



Hill

0

N/A

270.212

error

error



linear b

2

0.151

269.084

3.471E+02

1.082E+02



polynomial, 3-
degree

2

0.151

269.084

3.471E+02

1.082E+02



power

2

0.151

269.084

3.471E+02

1.082E+02

power bound hit (power =1)

Hill, unrestricted

0

N/A

270.266

1.059E+17

1.059E+17

unrestricted (n = 0.025)

power,
unrestricted

1

0.327

268.266

3.727E+14

error

unrestricted (power = 0.012)

a Constant variance model selected (p = 0.4372)
b Best-fitting model, BMDS output presented in this appendix

E.3.23.2. Output for Selected Model: Linear

Li et al., 2006: Estradiol, 3-Day

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\1\31_Li_2006_Estra_LinearCV_l.(d)

Gnuplot Plotting File: C:\l\31_Li_2006_Estra_LinearCV_l.plt

Tue Feb 16 18:13:56 2010

Figure 3, 3-day estradiol

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/'2 + ...

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Signs of the polynomial coefficients are not restricted
A constant variance model is fit

This document is a draft for review purposes only and does not constitute Agency policy.

E-3 81 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
alpha =	267 . 211

rho =	0 Specified

beta_0 =	16.4428

beta 1 = 0.0468351

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha	beta 0	beta 1

alpha 1	-2.6e-013	-4.5e-015

beta_0 -2.6e-013	1	-0.68

beta 1 -4.5e-015	-0.68	1

Parameter Estimates

Variable
alpha
beta_0
beta 1

Estimate

264.303
16.4428
0.0468351

95.0% Wald Confidence Interval
Std. Err.	Lower Conf. Limit Upper Conf. Limit

59.1	148.469	380.137

3.50431	9.57445	23.3111

0.062677	-0.0760095	0.16968

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	10

2	10

50	10

100	10

10.2
19. 9
24 . 7
18 .1

16.4

16.5
18 . 8
21.1

12 . 2
20
14 . 6
17 . 6

16.3
16.3
16.3
16.3

-1. 22
0. 656
1.16
-0.591

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-3 82 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-129.653527	5	269.307054

A2	-128.294657	8	272.589314

A3	-129.653527	5	269.307054

fitted	-131.541911	3	269.083823

R	-131.819169	2	267.638338

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adequately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.'

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	7.04902	6	0.3163

Test 2	2.71774	3	0.4372

Test 3	2.71774	3	0.4372

Test 4	3.77677	2	0.1513

The p-value for Test 1 is greater than .05. There may not be a
diffence between responses and/or variances among the dose levels
Modelling the data with a dose/response curve may not be appropriate

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	34 7.12

BMDL =	108.17 3

This document is a draft for review purposes only and does not constitute Agency policy.

E-3 83 DRAFT—DO NOT CITE OR QUOTE

-------
l E.3.23.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-3 84 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.3.24. Li et al., 2006: Progesterone, 3-Day
E.3.24.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

<0.001

330.234

5.252E+01

error



exponential (M3)

2

<0.001

330.234

5.252E+01

error

power hit bound (d = 1)

exponential (M4)

b

1

0.384

315.734

1.353E-01

8.351E-02



exponential (M5)

0

N/A

317.734

5.225E-01

7.503E-02



Hill

1

0.386

315.729

1.135E-02

1.161E-05

n lower bound hit (n = 1)

linear

2

<0.001

331.121

7.765E+01

5.264E+01



polynomial, 3-
degree

2

<0.001

331.121

7.765E+01

5.264E+01



power

2

<0.001

331.121

7.765E+01

5.264E+01

power bound hit (power =1)

power,
unrestricted

1

0.405

315.670

1.066E-63

1.066E-63

unrestricted (power = 0.009)

a Non-constant variance model selected (p = 0.0013)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.24.2. Output for Selected Model: Exponential (M4)

Li et al., 2006: Progesterone, 3-Day

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\32_Li_2006_Progest_Exp_l.(d)

Gnuplot Plotting File:

Tue Feb 16 18:14:31 2010

Figure 4, 3-day progesterone

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.

This document is a draft for review purposes only and does not constitute Agency policy.

E-385 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model 3 is nested within Model 5.
Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	11.3313

rho	-1.44835

a	64.8274

b	0.0456906

c	0.166844

d	1

Parameter Estimates
Variable	Model 4

lnalpha	14.074

rho	-2.27065

a	61.7474

b	2.13327

c	0.318566

d	1

Table of Stats From Input	Data

Dose	N	Obs Mean	Obs Std Dev

0	10	61.74	11.1

2	10	30.56	40.48

50	10	16.93	33.3

100	10	11.36	43.75

Estimated Values of	Interest
Dose Est Mean Est Std Scaled Residual

0	61.75	10.55	-0.002085

2	20.26	37.38	0.8713

50	19.67	38.66	-0.224

100	19.67	38.66	-0.6801

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-3 86 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:

Yij = Mu + e(i)
Var{e(ij)} = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-159.6327	5	329.2653

A2	-151.8128	8	319.6255

A3	-152.4882	6	316.9763

R	-165.6989	2	335.3978

4	-152.8668	5	315.7335

Additive constant for all log-likelihoods =	-36.76. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Test

2 :

Test

3:

Does response and/or variances differ among Dose levels:

Are Variances Homogeneous? (A2 vs. A1)

Are variances adeguately modeled? (A2 vs. A3)

(A2 vs. R)

Test

Does Model 4 fit the data? (A3 vs 4)

Tests of Interest
Test	-2*log(Likelihood Ratio)	D. F.	p-value

Test 1	27.77	6	0.0001037

Test 2	15.64	3	0.001344

Test 3	1.351	2	0.5089

Test 6a	0.7572	1	0.3842

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	0.135296

BMDL = 0.0835054

This document is a draft for review purposes only and does not constitute Agency policy.

E-3 87 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.24.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

60

40

20

-20

Exponential

VIDL

BMD

20

40

60

80

100

dose

18:14 02/16 2010

E.3.24.4. Output for Additional Model Presented: Hill, Unrestricted

Li et al., 2006: Progesterone, 3-Day

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\32_Li_2006_Progest_Hill_U_l.(d)

Gnuplot Plotting File: C:\l\32_Li_2 006_Progest_Hill_U_l.plt

Tue Feb 16 18:14:41 2010

Figure 4, 3-day progesterone

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
Power parameter is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + rho * In(mean(i)))

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-388 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha =	7.08699

rho =	0

intercept =	61.7404

v =	-50.3835

n =	1.43997

k =	1.6159

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho

intercept

The model parameter(s) -k

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
1

-0. 99
-0.097
0.84

NA

rho

-0. 99
1

0.13
-0. 81
NA

intercept

-0.097
0.13
1

-0.43
NA

v

0.84
-0. 81
-0.43
1

NA

NA

NA
NA
NA

NA - This parameter's variance has been estimated as zero or less.
THE MODEL HAS PROBABLY NOT CONVERGED!!!

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

k

Estimate

13.9863
-2 . 25026
61. 7404
-42 .1239
2.02774
le-013

Std.

Err.

NA
NA
NA
NA
NA
NA

Lower Conf.

Limit

NA
NA
NA
NA
NA

Upper Conf.

Limit

NA
NA
NA
NA
NA

At least some variance estimates are negative.
THIS USUALLY MEANS THE MODEL HAS NOT CONVERGED!
Try again from another starting point.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0
2
50
100

10
10
10
10

61. 7

30. 6
16. 9
11. 4

61. 7

19.6
19.6
19.6

11.1
40.5
33.3
43.7

10.5

38 . 3
38 . 3
38 . 3

9 . 7 4e-008

0. 905
-0.222
-0.683

This document is a draft for review purposes only and does not constitute Agency policy.

E-3 89 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)

-159.632675
-151.812765
-152.488175
-152.873643
-165.698875

Param's
5

AIC
329.265349
319.625529
316.976349
315.747285
335.397750

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

0.0001037
0.001344
0.5089
0.3799

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 5.817 03e-014

This document is a draft for review purposes only and does not constitute Agency policy.

E-390 DRAFT—DO NOT CITE OR QUOTE

-------
BMDL = 5.817 03e-014

5 E.3.24.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

a>

CO
c
o
Q.
CO
(D

(Z
c
(0
(D

18:14 02/16 2010

This document is a draft for review purposes only and does not constitute Agency policy.

E-391 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E.3.25. Markowski et al., 2001: FRIO Run Opportunities
E.3.25.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential
(M2)b

2

0.248

117.557

1.653E+02

5.025E+01



exponential (M3)

2

0.248

117.557

1.653E+02

5.025E+01

power hit bound (d = 1)

exponential (M4)

1

0.412

117.445

4.742E+01

1.729E-01



exponential (M5)

0

N/A

118.918

3.178E+01

3.967E-05



Hill

0

N/A

118.918

2.348E+01

6.728E-06



linear

2

0.190

118.089

2.081E+02

1.051E+02



polynomial, 3-
degree

2

0.190

118.089

2.081E+02

1.051E+02



power

2

0.190

118.089

2.081E+02

1.051E+02

power bound hit (power = 1)

power,
unrestricted

1

0.238

118.164

9.153E+01

5.911E-07

unrestricted (power = 0.237)

a Constant variance model selected (p = 0,1719)
b Best-fitting model, BMDS output presented in this appendix

E.3.25.2. Output for Selected Model: Exponential (M2)

Markowski et al., 2001: FR10 Run Opportunities

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\33_Mark_2001_FR10opp_ExpCV_l.(d)

Gnuplot Plotting File:

Tue Feb 16 18:15:26 2010

Table 3

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

E-392 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 2

Specified

lnalpha	3.5321

rho(S)	0

a	6.98169

b	0.00309891

c	0

d	1

Parameter Estimates
Variable	Model 2

lnalpha	3.64 823

rho	0

a	11.9443

b	0.0044262

c	0

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	7	13.29	8.65

20	4	11.25	5.56

60	6	5.75	3.53

180	7	7	6.01

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	11.94	6.197	0.5745

20	10.93	6.197	0.1025

60	9.158	6.197	-1.347

180	5.385	6.197	0.6897

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

E-393 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-54.38526	5	118.7705

A2	-51.88568	8	119.7714

A3	-54.38526	5	118.7705

R	-57.45429	2	118.9086

2	-55.77871	3	117.5574

Additive constant for all log-likelihoods =	-22.05. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Test

2 :

Test

3:

Test

4 :

Does response and/or variances differ among Dose levels:
Are Variances Homogeneous? (A2 vs. A1)

Are variances adeguately modeled? (A2 vs. A3)

Does Model 2 fit the data? (A3 vs. 2)

(A2 vs. R)

Test

Test 1
Test 2
Test 3
Test 4

Tests of Interest
-2*log(Likelihood Ratio)

11.14

4.999
4.999
2.787

p-value

0.08423
0.1719
0.1719
0.2482

The p-value for Test 1 is greater than .05. There may not be a
diffence between responses and/or variances among the dose levels
Modelling the data with a dose/response curve may not be appropriate.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 4 is greater than .1. Model 2 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	165.284

BMDL =	5 0.24 88

This document is a draft for review purposes only and does not constitute Agency policy.

E-394 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.25.3. Figure for Selected Model: Exponential (M2)

Exponential_beta Model 2 with 0.95 Confidence Level

dose

18:15 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-395 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.3.26. Markowski et al., 2001: FR2 Revolutions

E.3.26.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

0.192

217.636

1.627E+02

5.807E+01



exponential (M3)

2

0.192

217.636

1.627E+02

5.807E+01

power hit bound (d = 1)

exponential (M4)

1

0.298

217.415

4.668E+01

1.965E-01



exponential (M5)

0

N/A

218.532

3.308E+01

1.193E+01



Hill b

0

N/A

218.532

2.364E+01

7.336E+00

n upper bound hit (n = 18)

linear

2

0.150

218.129

1.989E+02

1.025E+02



polynomial, 3-
degree

2

0.150

218.129

1.989E+02

1.025E+02



power

2

0.150

218.129

1.989E+02

1.025E+02

power bound hit (power =1)

power,
unrestricted0

1

0.160

218.302

9.101E+01

1.800E-13

unrestricted (power = 0.272)

a Constant variance model selected (p = 0.1092)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.26.2. Output for Selected Model: Hill

Markowski et al., 2001: FR2 Revolutions

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\34_Mark_2001_FR2rev_HillCV_l.(d)

Gnuplot Plotting File: C:\l\34_Mark_2001_FR2rev_HillCV_l.plt

Tue Feb 16 18:16:03 2010

Table 3

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter restricted to be greater than 1
A constant variance model is fit

This document is a draft for review purposes only and does not constitute Agency policy.

E-396 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha =	2598 . 74

rho =	0 Specified

intercept =	119.29

v =	-62.79

n =	1.80602

k =	35.85

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
intercept

alpha
1

-8 . le-009
4 . 5e-008
-3e-005
3e-005

intercept
-8 . le-009
1

-0. 81
-0.00013
-0.0022

v

4 . 5e-008
-0. 81
1

0.0002
0.0014

n

-3e-005
-0. 00013
0.0002
1
-1

k

3e-005
-0.0022
0.0014
-1
1

Parameter Estimates

Variable
alpha
intercept

Estimate
2183.85
119.29
-56.5223
18

21.6708

Std. Err.
630.425
17 . 6629
21.9082
8854.08
855.263

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

948.245
84.6713
-99.4615
-17335.7
-1654.61

3419.46

153.909
-13.5831
17371.7
1697.95

Table of Data and Estimated Values of Interest
5	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	7 119	119 69.9	46.7	2.74e-008

20	4 109	108 61	46.7	8.42e-010

60	6 56.5	62.8 31.2	46.7	-0.329

180	7 68.1	62.8 33.2	46.7	0.304

Degrees	of freedom for Test A3 vs fitted <= 0

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

This document is a draft for review purposes only and does not constitute Agency policy.

E-397 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -104.165520	5	218.331040

A2 -101.140174	8	218.280349

A3 -104.165520	5	218.331040

fitted -104.266162	5	218.532324

R -107.599268	2	219.198536

Explanation of Tests

Test

1

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

Test 1
Test 2
Test 3
Test 4

-2*log(Likelihood Ratio)

12.9182
6.05069
6.05069
0.201283

Test df

p-value

0.04435
0.1092
0.1092
NA

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1.
model appears to be appropriate here

A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

NA - Degrees of freedom for Test 4 are less than or egual to 0. The Chi-Sguare
test for fit is not valid

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	23.6366

BMDL =	7.33648

This document is a draft for review purposes only and does not constitute Agency policy.

E-398 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.26.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

200

150

100

50

20

40

60

80 100
dose

120

140

160

180

18:16 02/16 2010

E.3.26.4. Output for Additional Model Presented: Power, Unrestricted
Markowski et al., 2001: FR2 Revolutions

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\34_Mark_2001_FR2rev_PowerCV_U_l.(d)

Gnuplot Plotting File: C:\l\34_Mark_2001_FR2rev_PowerCV_U_l.plt

Tue Feb 16 18:16:04 2010

Table 3

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-399 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho
control
slope
power

2598.74

0 Specified
119.29
-1.79436
0.708231

Asymptotic Correlation Matrix of Parameter Estimates

alpha
control

The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

9 . 7e-009

control
9 . 7e-009
1

slope
-1. 9e-008
-0.49

power
-1. 6e-008
-0.28

slope -1.9e-008	-0.49	1	0.96

power -1.6e-008	-0.28	0.96	1

Parameter Estimates

Variable
alpha
control
slope
power

Estimate

2351
120.074
-14.1965
0.27229

Std. Err.

678.674
18 . 0837
22 .2073
0.301344

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

1020.82
84.6305
-57.722
-0.318334

3681.17
155.517
29.329
0. 862913

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	7	119	120	69.9	48.5	-0.0428

20	4	109	88	61	48.5	0.846

60	6	56.5	76.8	31.2	48.5	-1.02

180	7	68.1	61.7	33.2	48.5	0.352

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2
Model A3 uses any fixed variance parameters that

This document is a draft for review purposes only and does not constitute Agency policy.

E-400 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -104.165520	5	218.331040

A2 -101.140174	8	218.280349

A3 -104.165520	5	218.331040

fitted -105.151136	4	218.302271

R -107.599268	2	219.198536

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	12.9182	6	0.04435

Test 2	6.05069	3	0.1092

Test 3	6.05069	3	0.1092

Test 4	1.97123	1	0.1603

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 91.0145

BMDL = 1.8e-013

This document is a draft for review purposes only and does not constitute Agency policy.

E-401 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.26.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

18:16 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-402 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.3.27. Markowski et al., 2001: FR5 Run Opportunities
E.3.27.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

2

0.149

133.830

9.491E+01

4.324E+01



exponential (M3)

2

0.149

133.830

9.491E+01

4.324E+01

power hit bound (d = 1)

exponential (M4)

1

0.303

133.087

2.961E+01

9.356E+00



exponential (M5)

0

N/A

134.032

2.871E+01

1.226E+01



Hill b

1

0.939

132.032

2.214E+01

1.117E+01

n upper bound hit (n = 18)

linear

2

0.091

134.825

1.349E+02

8.118E+01



polynomial, 3-
degree

2

0.091

134.825

1.349E+02

8.118E+01



power

2

0.091

134.825

1.349E+02

8.118E+01

power bound hit (power =1)

power,
unrestricted0

1

0.133

134.281

3.721E+01

1.439E-07

unrestricted (power = 0.336)

a Constant variance model selected (p = 0.2262)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.27.2. Output for Selected Model: Hill

Markowski et al., 2001: FR5 Run Opportunities

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\35_Mark_2001_FR5opp_HillCV_l.(d)

Gnuplot Plotting File: C:\l\35_Mark_2001_FR5opp_HillCV_l.plt

Tue Feb 16 18:16:39 2010

Table 3

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter restricted to be greater than 1
A constant variance model is fit

This document is a draft for review purposes only and does not constitute Agency policy.

E-403 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha =	77.4849

rho =	0 Specified

intercept =	26.14

v =	-13.34

n =	2.36002

k =	35.0654

Asymptotic Correlation Matrix of Parameter Estimates

alpha
intercept

The model parameter(s) -rho -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

-3.6e-009
9 . 8e-009
3. 6e-008

intercept
-3.6e-009
1

-0. 81
-0.51

v

9 . 8e-009
-0. 81
1

0.36

k

3. 6e-008
-0.51
0.36
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
intercept

k

Estimate

64 . 5863
26.14
-13.1569
18

21.5963

Std. Err.

18 . 6445
3. 03753
3.7676
NA

2 . 68136

Lower Conf. Limit

28.0438
20.1865
-20.5413

16.3409

Upper Conf. Limit

101.129
32 .0935
-5.77257

26.8517

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	7	26.1	26.1	12.3	8.04	1.02e-008

20	4	23.5	23.5	7.04	8.04	-1.39e-007

60	6	12.8	13	6.17	8.04	-0.0558

180	7	13.1	13	7.14	8.04	0.0517

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij|

This document is a draft for review purposes only and does not constitute Agency policy.

E-404 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -62.013133	5	134.026266

A2 -59.839035	8	135.678070

A3 -62.013133	5	134.026266

fitted -62.016024	4	132.032049

R -67.530040	2	139.060081

Explanation of Tests

Test

1

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio)

Test df

p-value

Test 1
Test 2
Test 3
Test 4

15.382
4.3482
4.3482
0.0057833

0.01748
0.2262
0.2262
0.9394

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1.
model appears to be appropriate here

A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	22.14 4

BMDL =	11.165

This document is a draft for review purposes only and does not constitute Agency policy.

E-405 DRAFT—DO NOT CITE OR QUOTE

-------
l E.3.27.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

a>

CO
c
o

Q.

CO
CD

a:

c
m

CD

40

35

30

25

20

15 - -L

10

20

40

60

80 100
dose

120

140

160

180

18:16 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-406 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

E.3.27.4. Output for Additional Model Presented: Power, Unrestricted
Markowski et al., 2001: FR5 Run Opportunities

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\35_Mark_2001_FR5opp_PwrCV_U_l.(d)

Gnuplot Plotting File: C:\l\35_Mark_2001_FR5opp_PwrCV_U_l.plt

Tue Feb 16 187l6:40 2010

Table 3

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

Total number of dose groups = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial
alpha =
rho =
control =
slope =
power =

Parameter Values
77.4849

0 Specified
26.14
-0.39517
0.725538

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
control

alpha
1

7 . 4e-009

control
7 . 4e-009
1

slope
4.3e-008
-0.51

power
4 . 8e-008
-0.34

slope	4.3e-008	-0.51	1	0.97

power	4.8e-008	-0.34	0.97	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf. Limit	Upper Conf. Limit

alpha 70.9323 20.4764	30.7993	111.065

control 26.3567 3.13032	20.2213	32.492

slope -2.49841 3.16984	-8.71118	3.71437

power 0.336003 0.242031	-0.138368	0.810375

This document is a draft for review purposes only and does not constitute Agency policy.

E-407 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0
20
60
180

26.1

23.5
12 . 8
13.1

26.4

19.5
16.5
12 .1

12 . 3
7 . 04
6.17
7 .14

} .42

J .42
J .42
} .42

-0.0681
0. 945
-1. 07
0.341

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi

Var{e(i)}

Mu + e(i)
Sigma/N2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -62.013133	5	134.026266

A2 -59.839035	8	135.678070

A3 -62.013133	5	134.026266

fitted -63.140714	4	134.281428

R -67.530040	2	139.060081

Explanation of Tests

Test 1:

Test 2
Test 3
Test 4
(Note:

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

15.382
4.3482
4.3482
2 . 25516

0.01748
0.2262
0.2262
0.1332

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than
model appears to be appropriate here

.1. A homogeneous variance

The p-value for Test 3 is greater than
to be appropriate here

.1. The modeled variance appears

This document is a draft for review purposes only and does not constitute Agency policy.

E-408 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 37.2131

BMDL = 1.4 3 92 6e-007

E.3.27.5. Figure for Additional Model Presented: Power, Unrestricted

CD
CO
c
o

Q.

CO
CD

a:

c
m

CD

Power Model with 0.95 Confidence Level

20

40

60

80 100
dose

120

140

160

180

18:16 02/16 2010

This document is a draft for review purposes only and does not constitute Agency policy.

E-409 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.3.28. Miettinen et al., 2006: Cariogenic Lesions, Pups
E.3.28.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

3

0.345

162.699

7.505E+01

4.086E+01

power bound hit (power = 1)

logistic

3

0.315

162.909

8.991E+01

5.250E+01



log-logistica

3

0.506

161.767

3.130E+01

1.054E+01

slope bound hit (slope = 1)

log-probit

3

0.257

163.393

1.390E+02

6.729E+01

slope bound hit (slope =1)

multistage, 4-
degree

3

0.345

162.699

7.505E+01

4.086E+01

final B = 0

probit

3

0.299

163.031

9.941E+01

6.208E+01



Weibull

3

0.345

162.699

7.505E+01

4.086E+01

power bound hit (power = 1)

gamma,
unrestricted

2

0.797

161.805

1.591E-02

1.335E-
240

unrestricted (power = 0.184)

log-logistic,
unrestricted b

2

0.723

161.998

3.713E-01

error

unrestricted (slope = 0.403)

log-probit,
unrestricted

2

0.726

161.987

5.098E-01

error

unrestricted (slope = 0.25)

Weibull,
unrestricted

2

0.761

161.897

1.174E-01

error

unrestricted (power = 0.281)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.3.28.2. Output for Selected Model: Log-Logistic

Miettinen et al., 2006: Cariogenic Lesions, Pups

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\36_Miet_2006_Cariogenic_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\36_Miet_2006_Cariogenic_LogLogistic_l.plt

Tue Feb 16 18:17:16 2010

Table 2 converting the percentage into the number of animals, and control is Control II from the
study. Dose is in ng per kg and is from Table 1

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff

This document is a draft for review purposes only and does not constitute Agency policy.

E-410 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Independent variable = Dose

Slope parameter is restricted as slope >= 1
Total number of observations = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background =	0.595238

intercept =	-5.52519

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept
background	1	-0.64

intercept	-0.64	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.658158	*	*	*

intercept	-5.64068	*	*	*

slope	1	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-77.6769	5

Fitted model -78.8837 2	2.41374 3 0.4911

Reduced model -83.2067 1	11.0597 4 0.0259

AIC:	161.767

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0

6582

27 . 643

25.000

42

-0

860

30.0000

0

6911

20.041

23.000

29

1

189

100.0000

0

7477

18.693

19.000

25

0

141

300.0000

0

8345

20.027

20.000

24

-0

015

1000.0000

0

924 9

29.596

29.000

32

-0

400

Chi ^2 = 2.33



d.f.

= 3 P

-value = 0.5062







Benchmark Dose Computation

This document is a draft for review purposes only and does not constitute Agency policy.

E-411 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

Specified effect =
Risk Type	=

Confidence level =

0.1
Extra risk
0. 95

BMD =	31.2951

BMDL =	10.5354

E.3.28.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

dose

18:17 02/16 2010

E.3.28.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
Miettinen et al., 2006: Cariogenic Lesions, Pups

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\36_Miet_2006_Cariogenic_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\36_Miet_2006_Cariogenic_LogLogistic_U_l.plt

Tue Feb 16 18:17:18 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-412 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Table 2 converting the percentage into the number of animals, and control is Control II from the
study. Dose is in ng per kg and is from Table 1

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values

background =	0.595238

intercept =	-1.68849

slope =	0.382632

Asymptotic Correlation Matrix of Parameter Estimates
background intercept	slope

background	1	-0.41	0.24

intercept	-0.41	1	-0.96

slope	0.24	-0.96	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.597778	*	*	*

intercept	-1.79836	*	*	*

slope	0.402606	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-77.6769	5

Fitted model -77.9988 3	0.643944 2	0.7247

Reduced model -83.2067 1	11.0597 4	0.0259

AIC:	161.998

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000	0.5978	25.107 25.000	42	-0.034

30.0000	0.7564	21.936 23.000	29	0.460

This document is a draft for review purposes only and does not constitute Agency policy.

E-413 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

100.0000	0.8045

300.0000	0.8480

1000.0000	0.8905

Chi ^2

0. 65

d. f.

2

20.112 19.000
20.351 20.000
28.495 29.000

P-value = 0.7227

25
24
32

-0.561
-0.200
0.286

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	0.371315

Benchmark dose computation failed. Lower limit includes zero.

E.3.28.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model

"O

(D

-t—<

o

CD
!t=
<

0.9

0.7

0.6

0.5

0.4

18:17 02/16 2010

200

400

600

800

1000

dose

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-414 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

E.3.29. Murray et al., 1979: Fertility in F2 Generation
E.3.29.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

X2 P-
Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

0

N/A

61.729

7.016E+00

1.698E+00



logistic

1

0.072

60.497

4.007E+00

2.836E+00

negative intercept (intercept = -
2.53)

log-logistic

0

N/A

61.729

7.902E+00

1.584E+00



multistage, 1-
degree

1

0.053

61.644

2.380E+00

1.320E+00



multistage, 2-
degreea

1

0.094

59.935

4.548E+00

1.635E+00



probit

1

0.070

60.613

3.707E+00

2.615E+00

negative intercept (intercept = -
1.446)

Weibull

0

N/A

61.729

8.115E+00

1.698E+00



log-probit,
unrestricted

0

N/A

61.729

6.373E+00

1.503E+00

unrestricted (slope = 2.306)

a Best-fitting model, BMDS output presented in this appendix

E.3.29.2. Output for Selected Model: Multistage, 2-Degree

Murray et al., 1979: Fertility in F2 Generation

Multistage Model. (Version: 3.0; Date: 05/16/2008)

Input Data File: C:\l\Murray_l979_fert_index_f2_Multi2_l.(d)

Gnuplot Plotting File: C:\l\Murray_197 9_fert_index_f2_Multi2_l.plt

Tue Feb 16 20:08:06 2010

Table 1 but expressed as number of dams who do not produce offspring

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal*dose/Nl-beta2*dose/N2 ) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 3

Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2

This document is a draft for review purposes only and does not constitute Agency policy.

E-415 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0624181
Beta(1) =	0

Beta(2) = 0.00532688

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Beta(l)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Background	Beta(2)

Background	1	-0.44

Beta(2)	-0.44	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0772201	*	*	*

Beta (1)	0	*	*	*

Beta(2)	0.00509404	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-25.8194	3

Fitted model -27.9673 2	4.29584 1 0.03821

Reduced model -34.0009 1	16.363 2 0.0002798

AIC:	59.9347

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000

0.0772

2 .471

4 . 000

32

1. 013

1.0000

0.0819

1. 638

0. 000

20

-1. 336

10.0000

0.4455

8 . 911

9. 000

20

0. 040

Chi ^2 = 2.81	d.f. = 1	P-value = 0.0936

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	4.54787

This document is a draft for review purposes only and does not constitute Agency policy.

E-416 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

E'.MDL =

1.63487

E'.MDU =	6.7 9105

Taken together, (1.63487, 6.79105) is a 90	% two-sided confidence

interval for the EMD

E.3.29.3. Figure for Selected Model: Multistage, 2-Degree

Multistage Model with 0.95 Confidence Level

dose

20:08 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-417 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

E.3.30. National Toxicology Program, 1982: Toxic Hepatitis, Male Mice
E.3.30.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

1

0.026

113.097

1.552E+01

5.155E+00



logistic

2

0.093

110.712

1.769E+01

1.383E+01

negative intercept (intercept =
-3.087)

log-logistic

1

0.027

113.093

1.499E+01

6.628E+00



log-probit

1

0.027

113.111

1.360E+01

7.237E+00



multistage, 3-
degreea

1

0.028

112.555

1.488E+01

4.676E+00



probit

2

0.088

110.696

1.564E+01

1.261E+01

negative intercept (intercept =
-1.731)

Weibull

1

0.026

113.056

1.619E+01

4.903E+00



a Best-fitting model, BMDS output presented in this appendix

E.3.30.2. Output for Selected Model: Multistage, 3-Degree

National Toxicology Program, 1982: Toxic Hepatitis, Male Mice



Multistage Model.

. (Version: 3.0;

: Date: 05/16/2008)



Input Data File:

C:\l\37 NTP 19E

32 ToxHep Multi3 1.(d)



Gnuplot Plotting

File: C:\l\37

NTP 1982 ToxHep Multi3 l.plt

Tue Feb 16 18:17:51 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/Nl-beta2*dose/N2-beta3*dose/N3) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-418 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Parameter

has been set to: le-008

Default Initial Parameter Values

Background = 0.0525767
Beta(1) = 0.00243254
Beta(2) =	0

Beta(3) = 5.29052e-006

Asymptotic Correlation Matrix of Parameter Estimates

Background
Beta(1)
Beta(3)

The model parameter(s) -Beta(2)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Background
1

-0.69
0. 66

Beta(1)

-0.69
1

-0. 98

Beta(3)

0. 66
-0. 98
1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)
Beta(3)

Estimate

0.0383474
0.00605732
0

.60855e-006

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-51.0633
-53.2776
-121.743

Param's

4
3
1

Deviance Test d.f.

4.42854
141.358

P-value

0.03534
<.0001

AIC:

112.555

Dose

Goodness of Fit

Est. Prob.

Expected

Observed

Scaled
Residual

0.0000
1.4000
7.1000
71.0000

0.0383
0.0465
0.0803
0.8798

2.799

2.278
3. 937
43.990

1. 000
5. 000
3. 000
44.000

73

49

49

50

-1.097

1.847
-0.492
0. 004

Chi ^2

4 .86

d.f.

P-value

0. 0275

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

This document is a draft for review purposes only and does not constitute Agency policy.

E-419 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

BMD =

14.8848

E'.MDL =	4.67636

BMDLJ =	28.8293

Taken together, (4.67636,	28.8293) is a 90
interval for the EMD

two'-sided oorifidenoe

E.3.30.3. Figure for Selected Model: Multistage, 3-Degree

Multistage Model with 0.95 Confidence Level

"O

(D

-t—<

o

(U
!t=
<

0.6

0.4

0.2

0 -

18:17 02/16 2010

This document is a draft for review purposes only and does not constitute Agency policy.

E-420 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.3.31. National Toxicology Program, 2006: Alveolar Metaplasia
E.3.31.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

4

<0.001

340.127

2.240E+00

1.791E+00

power bound hit (power =1)

logistic

4

<0.001

358.346

4.997E+00

4.149E+00

negative intercept (intercept =
-0.687)

log-logistica

4

0.409

312.970

6.644E-01

5.041E-01

slope bound hit (slope = 1)

log-probit

4

<0.001

340.296

3.291E+00

2.517E+00

slope bound hit (slope =1)

multistage, 5-
degree

4

<0.001

340.127

2.240E+00

1.791E+00

final B = 0

probit

4

<0.001

362.181

5.656E+00

4.810E+00

negative intercept (intercept =
-0.381)

Weibull

4

<0.001

340.127

2.240E+00

1.791E+00

power bound hit (power =1)

gamma,
unrestricted

3

0.407

314.135

2.211E-02

8.081E-04

unrestricted (power = 0.297)

log-logistic,
unrestricted b

3

0.739

312.487

3.062E-01

7.972E-02

unrestricted (slope = 0.785)

log-probit,
unrestricted

3

0.727

312.543

3.316E-01

8.968E-02

unrestricted (slope = 0.471)

Weibull,
unrestricted

3

0.586

313.176

9.000E-02

1.341E-02

unrestricted (power = 0.465)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.3.31.2. Output for Selected Model: Log-Logistic

National Toxicology Program, 2006: Alveolar Metaplasia

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\40_NTP_2006_AlvMeta_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\4 0_NTP_2006_AlvMeta_LogLogistic_l.plt

Tue Feb 16 18:19:30 2010

0

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

This document is a draft for review purposes only and does not constitute Agency policy.

E-421 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Slope parameter is restricted as slope >= 1
Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values

background = 0.0377358
intercept =	-2.03745

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept
background	1	-0.4

intercept	-0.4	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.0448753	*	*	*

intercept	-1.78837	*	*	*

slope	1	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-152.615	6

Fitted model -154.485 2	3.7393 4	0.4424

Reduced model -216.802 1	128.374 5	<.0001

AIC:	312.97

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0

044 9

2.318

2 . 000

53

-0

251

2.1400

0

2966

16.017

19.000

54

0

889

7.1400

0

5647

29.928

33.000

53

0

851

15.7000

0

7366

38.301

35.000

52

-1

039

32.9000

0

8531

45.214

45.000

53

-0

083

71.4000

0

9262

48.162

46.000

52

-1

147

Chi/N2 = 3.98	d.f. = 4	P-value = 0.4088

Benchmark Dose Computation

This document is a draft for review purposes only and does not constitute Agency policy.

E-422 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Specified effect
Risk Type
Confidence level
BMD
BMDL

0.1
Extra risk
0. 95
0.664411
0.504109

E.3.31.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

"O

(D

-t—<

o

CD
!t=
<

0.6

0.4

0.2

0 -

18:19 02/16 2010

10

20

30

40

50

60

70

dose

E.3.31.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
National Toxicology Program, 2006: Alveolar Metaplasia

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\40_NTP_2006_AlvMeta_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\4 0_NTP_2006_AlvMeta_LogLogistic_U_l.plt

Tue Feb 16 18:19:31 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-423 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial	Parameter Values

background =	0.0377358

intercept =	-1.26694

slope =	0.784484

Asymptotic Correlation Matrix of Parameter Estimates
background intercept	slope

background	1	-0.24	0.11

intercept	-0.24	1	-0.9

slope	0.11	-0.9	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.0375286	*	*	*

intercept	-1.26811	*	*	*

slope	0.785033	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-152.615	6

Fitted model -153.244 3	1.2566 3	0.7395

Reduced model -216.802 1	128.374 5	<.0001

AIC:	312.487

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0375 1.989 2.000	53	0.008

2.1400 0.3631 19.609 19.000	54	-0.172

7.1400 0.5845 30.980 33.000	53	0.563

15.7000 0.7205 37.468 35.000	52	-0.763

This document is a draft for review purposes only and does not constitute Agency policy.

E-424 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

32.9000	0.8207	43.4 98 45.000

71.4000	0.8934	46.455 46.000

53
52

0.538
-0.204

Chi' '2 = 1.26 d.f.	= 3 P-value = 0.738

E'.enchmark Dose Computation

Specified effect =	0 .1

Risk Typ'e =	E::tra risk

Cc'nfidence level =	0.95

BMD =	0.306194

EMDL =	0.07 97 223

E.3.31.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

dose

18:19 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-425 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

E.3.32. National Toxicology Program, 2006: Eosinophilic Focus, Liver
E.3.32.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

4

0.367

330.457

5.676E+00

4.532E+00

power bound hit (power = 1)

logistic

4

0.167

333.343

1.258E+01

1.071E+01

negative intercept (intercept =
-1.747)

log-logistic

3

0.117

334.148

4.727E+00

2.867E+00



log-probit

4

0.084

334.683

1.078E+01

8.514E+00



multistage, 5-
degree

3

0.313

331.771

6.568E+00

4.666E+00



probita

4

0.187

332.962

1.196E+01

1.031E+01

negative intercept (intercept
= -1.061)

Weibull

4

0.367

330.457

5.675E+00

4.532E+00

power bound hit (power = 1)

log-probit,
unrestricted

3

0.087

334.849

4.750E+00

1.757E+00

unrestricted (slope = 0.643)

a Best-fitting model, BMDS output presented in this appendix

E.3.32.2. Output for Selected Model: Probit

National Toxicology Program, 2006: Eosinophilic Focus, Liver

pit

56 2010
0

Probit Model. (Version: 3.1; Date: 05/16/2008)

Input Data File: C:\1\45_NTP_2006_LivEosFoc_Probit_l.(d)

Gnuplot Plotting File: C:\l\45_NTP_2006_LivEosFoc_Probit_l.

Tue Feb 16 18:25:

The form of the probability function is:

P[response] = CumNorm(Intercept + Slope* Dose),

where CumNorm(.) is the cumulative normal distribution function

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-426 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Default Initial	(and Specified) Parameter Values

background =	0 Specified

intercept =	-1.11935

slope =	0.0279665

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

intercept	slope

intercept	1	-0.69

slope	-0.69	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

intercept	-1.06148	0.109177	-1.27546	-0.847497

slope	0.0269279	0.00327788	0.0205034	0.0333525

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)
-161.07
-164.481
-202.816

Param's Deviance Test d.f.

6.8221
83.4925

P-value

0.1456
C. 0001

332.962

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000

0

1442

7 . 645

3

000

53

-1

816

2.1400

0

1577

8 . 517

8

000

54

-0

193

7.1400

0

1924

10.195

14

000

53

1

326

15.7000

0

2615

13.860

17

000

53

0

982

32.9000

0

4303

22 .807

22

000

53

-0

224

71.4000

0

8054

42.688

42

000

53

-0

239

Chi ^2 = 6.16	d.f. = 4	P-value = 0.1873

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	11.9584

BMDL =	10.3075

This document is a draft for review purposes only and does not constitute Agency policy.

E-427 DRAFT—DO NOT CITE OR QUOTE

-------
l E.3.32.3. Figure for Selected Model: Probit

Probit Model with 0.95 Confidence Level

dose

2	18:25 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-428 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

E.3.33. National Toxicology Program, 2006: Fatty Change Diffuse, Liver
E.3.33.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

4

0.668

252.294

4.224E+00

3.166E+00



logistic

4

0.005

269.825

1.092E+01

9.292E+00

negative intercept (intercept =
-2.298)

log-logistic

4

0.292

255.082

4.697E+00

3.153E+00



log-probit

4

0.118

257.548

6.236E+00

5.204E+00

slope bound hit (slope =1)

multistage, 5-
degree

4

0.808

251.545

4.021E+00

3.250E+00



probit

4

0.005

269.430

1.052E+01

9.068E+00

negative intercept (intercept =
-1.36)

Weibull3

4

0.679

252.218

4.252E+00

3.174E+00



log-probit,
unrestricted

4

0.282

255.258

4.581E+00

3.193E+00

unrestricted (slope = 0.824)

a Best-fitting model, BMDS output presented in this appendix

E.3.33.2. Output for Selected Model: Weibull

National Toxicology Program, 2006: Fatty Change Diffuse, Liver

Weibull Model using Weibull Model (Version: 2.12; Date: 05/16/2008)
Input Data File: C:\1\47_NTP_2006_LivFatDiff_Weibull_l.(d)

Gnuplot Plotting File: C:\l\47_NTP_2006_LivFatDiff_Weibull_l.plt

Tue Feb 16 18:26757 2010

NTP liver fatty change diffuse

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(-slope*doseApower)]

Dependent variable = DichEff
Independent variable = Dose

Power parameter is restricted as power >=1
Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-429 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Default Initial	(and Specified) Parameter Values

Background =	0.00925926

Slope =	0.00962604

Power =	1.28042

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Slope	Power

Slope	1	-0.97

Power	-0.97	1

Parameter Estimates

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

0.0037073	0.0409874

0.831952	1.31071

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Variable
Background
Slope
Power

Estimate

0

0.0223474
1.07133

Std. Err.

NA

0.00951041
0.122134

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-122.992	6

Fitted model -124.109 2	2.23388 4	0.692e

Reduced model -204.846 1	163.708 5	<.0001

AIC:	252.218

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0

0000

0. 000

0

000

53

0

000

2.1400

0

04 92

2 . 659

2

000

54

-0

414

7.1400

0

1677

8.889

12

000

53

1

144

15.7000

0

3475

18.420

17

000

53

-0

409

32.9000

0

6107

32.365

30

000

53

-0

6 6 6

71.4000

0

8851

4 6.909

48

000

53

0

470

Chi ^2 = 2.31	d.f. = 4	P-value = 0.6785

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	4.25219

BMDL =	3.17375

This document is a draft for review purposes only and does not constitute Agency policy.

E-430 DRAFT—DO NOT CITE OR QUOTE

-------
1

2	E.3.33.3. Figure for Selected Model: Weibull

Weibull Model with 0.95 Confidence Level

dose

3	18:26 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-431 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.3.34. National Toxicology Program, 2006: Gingival Hyperplasia, Squamous, 2 Years
E.3.34.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

4

0.012

318.867

2.295E+01

1.417E+01

power bound hit (power = 1)

logistic

4

0.008

320.908

3.594E+01

2.564E+01

negative intercept (intercept =
-1.711)

log-logistica

4

0.015

317.969

1.838E+01

1.044E+01

slope bound hit (slope = 1)

log-probit

4

0.003

323.633

4.313E+01

2.794E+01

slope bound hit (slope =1)

multistage, 5-
degree

4

0.012

318.867

2.295E+01

1.417E+01

final B = 0

probit

4

0.008

320.687

3.436E+01

2.425E+01

negative intercept (intercept =
-1.034)

Weibull

4

0.012

318.867

2.295E+01

1.417E+01

power bound hit (power = 1)

gamma,
unrestricted

3

0.651

307.529

2.480E-01

5.096E-09

unrestricted (power = 0.199)

log-logistic,
unrestricted b

3

0.675

307.416

3.710E-01

1.505E-07

unrestricted (slope = 0.265)

log-probit,
unrestricted

3

0.688

307.354

4.688E-01

8.851E-07

unrestricted (slope = 0.156)

Weibull,
unrestricted

3

0.663

307.471

3.076E-01

3.210E-08

unrestricted (power = 0.23)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.3.34.2. Output for Selected Model: Log-Logistic

National Toxicology Program, 2006: Gingival Hyperplasia, Squamous, 2 Years

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\42_NTP_2006_GingHypSq_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\42_NTP_2006_GingHypSq_LogLogistic_l.plt

Tue Feb~16 18:20:29~2010

[insert study notes]

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

This document is a draft for review purposes only and does not constitute Agency policy.

E-432 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Slope parameter is restricted as slope >= 1

Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background = 0.0188679
intercept =	-4.5509

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept

background	1	-0.71

intercept	-0.71	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf.	Limit Upper Conf. Limit

background 0.117717 *	*	*

intercept -5.10866 *	*	*

slope 1 *	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-149.95	6

Fitted model -156.985 2	14.0696 4	0.00707^

Reduced model -162.631 1	25.3627 5	0.000118^

AIC:	317.969

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.1177

6.239

1

000

53

-2 . 233

2.1400

0.1290

6. 965

7

000

54

0. 014

7.1400

0.1542

8.174

14

000

53

2 . 216

15.7000

0.1942

10.292

13

000

53

0. 940

32.9000

0.2641

13.995

15

000

53

0.313

71.4000

0.3837

20.335

16

000

53

-1. 225

Chi ^2 = 12.38	d.f. = 4	P-value = 0.0147

Benchmark Dose Computation

This document is a draft for review purposes only and does not constitute Agency policy.

E-433 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Specified effect =	0.1

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	18.3832

BMDL =	10.4 35 9

E.3.34.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

dose

18:20 02/16 2010

E.3.34.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
National Toxicology Program, 2006: Gingival Hyperplasia, Squamous, 2 Years

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\42_NTP_2006_GingHypSq_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\42_NTP_2006_GingHypSq_LogLogistic_U_l.plt

Tue Feb 16 18:20:29 2010

[insert study notes]

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-434 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial	Parameter Values

background =	0.0188679

intercept =	-2.04571

slope =	0.299277

Asymptotic Correlation Matrix of Parameter Estimates
background intercept	slope

background	1	-0.3	0.12

intercept	-0.3	1	-0.91

slope	0.12	-0.91	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0.0185126	*	*	*

intercept	-1.93464	*	*	*

slope	0.264795	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-149.95	6

Fitted model -150.708 3	1.5163 3	0.6785

Reduced model -162.631 1	25.3627 5	0.0001186

AIC:	307.416

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0185 0.981 1.000	53 0.019

2.1400 0.1659 8.959 7.000	54	-0.717

7.1400 0.2105 11.155 14.000	53 0.959

15.7000 0.2447 12.972 13.000	53 0.009

This document is a draft for review purposes only and does not constitute Agency policy.

E-43 5 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

32.9000	0.2806

71.4000	0.3219

14 . 873
17.059

15.000
16.000

53
53

0. 03 9
-0.311

Chi' '2

1. 53

d. f.

E'-value

0.6750

E'.enchmark Dose Computation
Specified effect =	0 .1

Risk Typ'e	=	E::tra risk

Cc'nfidence level =	0.95

BMD =	0.37 0 95 8

BMDL = 1. 5 0 4 9 4 e - 0 0 7

E.3.34.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

dose

18:20 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-436 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

E.3.35. National Toxicology Program, 2006: Hepatocyte Hypertrophy, 2 Years
E.3.35.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

4

<0.001

290.365

1.647E+00

1.340E+00

power bound hit (power = 1)

logistic

4

<0.001

310.492

4.315E+00

3.650E+00

negative intercept (intercept =
-1.237)

log-logistic

5

0.010

278.082

6.978E-01

5.454E-01

slope bound hit (slope =1)

log-probit

4

<0.001

297.168

2.930E+00

2.267E+00

slope bound hit (slope =1)

multistage, 5-
degreea

4

<0.001

290.365

1.647E+00

1.340E+00

final 15 = 0

probit

4

<0.001

313.841

4.564E+00

3.923E+00

negative intercept (intercept =
-0.714)

Weibull

4

<0.001

290.365

1.647E+00

1.340E+00

power bound hit (power = 1)

gamma,
unrestricted

4

0.029

275.042

error

error

unrestricted (power = 0.478)

log-logistic,
unrestricted

4

0.005

280.068

6.672E-01

2.939E-01

unrestricted (slope = 0.984)

log-probit,
unrestricted

4

0.006

279.204

7.167E-01

3.322E-01

unrestricted (slope = 0.594)

Weibull,
unrestricted

4

0.019

275.967

3.709E-01

1.315E-01

unrestricted (power = 0.64)

a Best-fitting model, BMDS output presented in this appendix

E.3.35.2. Output for Selected Model: Multistage, 5-Degree

National Toxicology Program, 2006: Hepatocyte Hypertrophy, 2 Years

Multistage Model. (Version: 3.0; Date: 05/16/2008)

Input Data File: C:\1\43_NTP_2006_HepHyper_Multi5_l.(d)

Gnuplot Plotting File: C:\l\43_NTP_2006_HepHyper_Multi5_l.plt

Tue Feb 16 18721:00 2010

[insert study notes]

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/Nl-beta2*dose/N2-beta3*dose/N3-beta4*dose/N4-beta5*dose/N5) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff

This document is a draft for review purposes only and does not constitute Agency policy.

E-437 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Independent variable = Dose
Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 6
Total number of specified parameters = 0
Degree of polynomial = 5

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.232262

Beta(1) =	0.045074

Beta(2) =	0

Beta(3) =	0

Beta(4) =	0

Beta(5) = 2.59945e-010

Asymptotic Correlation Matrix of Parameter Estimates

(	The model parameter(s) -Beta(2)	-Beta(3)	-Beta(4)	-Beta(5)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Background	Beta(1)

Background	1	-0.64

Beta(1)	-0.64	1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)
Beta(3)
Beta(4)
Beta(5)

Estimate

0.0541647
0.0639585
0
0
0
0

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-129.986
-143.183
-219.97

Param's Deviance Test d.f.

26.3932
179.968

P-value

2.63 61629e-0 05
<.0001

AIC:

290.365

Dose

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000
2.1400
7.1400

0.0542
0.1752
0.4009

2 . 871
9. 458
21.248

0. 000
19.000
19.000

53

54
53

-1.742

3.416
-0.630

This document is a draft for review purposes only and does not constitute Agency policy.

E-43 8 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

15.7000	0.6535	34.635 42.000	53	2.126

32.9000	0.8847	46.887 41.000	53	-2.532

71.4000	0.9902	52.479 52.000	53	-0.667

Chi' '2 = 26.48 d.f.	= 4	P-value = 0.0000

E'.enchmark Dose Computation

Specified effect =	0 .1

Risk Typ'e =	E::tra risk

Cc'nfidence level =	0.95

BMD =	1.64733

EMDL =	1.34 007

BMDLJ =	2.05 81

Taken together, (1.34 007, 2.0581 ) is a 90	% two-sided confidence

interval for the EMD

E.3.35.3. Figure for Selected Model: Multistage, 5-Degree

Multistage Model with 0.95 Confidence Level

10	20	30	40	50	60	70

dose

18:21 02/16 2010

Multistage

This document is a draft for review purposes only and does not constitute Agency policy.

E-439 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

E.3.36. National Toxicology Program, 2006: Necrosis, Liver
E.3.36.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

logistic

4

0.397

238.314

3.484E+01

2.842E+01

negative intercept (intercept =
-2.601)

log-logistic

4

0.810

235.265

1.791E+01

1.194E+01

slope bound hit (slope =1)

log-probit

4

0.290

239.107

3.205E+01

2.382E+01

slope bound hit (slope =1)

multistage, 5-
degree

4

0.763

235.581

2.019E+01

1.419E+01

final B = 0

probit

4

0.445

237.888

3.266E+01

2.637E+01

negative intercept (intercept =
-1.508)

Weibull

4

0.763

235.581

2.019E+01

1.419E+01

power bound hit (power =1)

gamma,
unrestricted

3

0.869

236.344

1.114E+01

3.487E+00

unrestricted (power = 0.599)

log-logistic,
unrestricted

3

0.833

236.483

1.112E+01

3.581E+00

unrestricted (slope = 0.695)

log-probit,
unrestricteda

3

0.768

236.742

1.061E+01

3.498E+00

unrestricted (slope = 0.367)

Weibull,
unrestricted

3

0.856

236.393

1.117E+01

3.554E+00

unrestricted (power = 0.64)

a Best-fitting model, BMDS output presented in this appendix

E.3.36.2. Output for Selected Model: Log-Probit, Unrestricted

National Toxicology Program, 2006: Necrosis, Liver

Probit Model. (Version: 3.1; Date: 05/16/2008)

Input Data File: C:\1\50_NTP_2006_LivNec_LogProbit_U_l.(d)

Gnuplot Plotting File: C:\l\50_NTP_2006_LivNec_LogProbit_U_l.plt

Tue Feb 16 18:34731 2010

NTP liver necrosis

The form of the probability function is:

P[response] = Background

+ (1-Background) * CumNorm(Intercept + Slope^Log(Dose) ) ,

where CumNorm(.) is the cumulative normal distribution function

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

This document is a draft for review purposes only and does not constitute Agency policy.

E-440 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial	(and Specified) Parameter Values

background =	0.0188679

intercept =	-1.98094

slope =	0.316942

Asymptotic Correlation Matrix of Parameter Estimates
background intercept	slope

background	1	-0.69	0.59

intercept	-0.69	1	-0.97

slope	0.59	-0.97	1

Parameter Estimates

Variable
background
intercept
slope

Estimate

0.0228339
-2 .14844
0.367034

95.0% Wald Confidence Interval
Std. Err.	Lower Conf. Limit Upper Conf. Limit

0.0230818	-0.0224057	0.0680734

0.527256	-3.18184	-1.11503

0.139055	0.0944904	0.639577

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)
-114 . 813
-115.371
-127.98

Param's Deviance Test d.f.

1.1157
26.3331

P-value

0.7733
C. 0001

236.742

Goodness of Fit

Scaled

Dose

Est

. Prob.

Expected

Observed

Size

Residual

0.0000

0.

0228

1 210

1

000

53

-0

193

2.1400

0.

0529

2 858

4

000

54

0

694

7.1400

0.

0979

5.187

4

000

53

-0

549

15.7000

0.

1475

7.819

8

000

53

0

070

32.9000

0.

2116

11 215

10

000

53

-0

409

71.4000

0.

2968

15.729

17

000

53

0

382

Chi ^2

1.14

d.f.

P-value

0.7678

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

This document is a draft for review purposes only and does not constitute Agency policy.

E-441 DRAFT—DO NOT CITE OR QUOTE

-------
BMD =	10.6107

BMDL =	3.4 97 91

E.3.36.3. Figure for Selected Model: Log-Probit, Unrestricted

LogProbit Model with 0.95 Confidence Level

0.5 	

0.4

"O

CD

-t—'

O
CD
!t=
<
c
o

o
m

0.3

0.2

0.1

0 -

dose

18:34 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-442 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

E.3.37. National Toxicology Program, 2006: Oval Cell Hyperplasia
E.3.37.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

3

0.072

199.446

8.970E+00

5.499E+00



logistic

4

0.069

199.875

9.792E+00

8.245E+00

negative intercept (intercept =
-3.116)

log-logistic

3

0.039

202.012

9.708E+00

7.247E+00



log-probit

3

0.068

200.421

9.968E+00

7.758E+00



multistage, 5-
degree

2

0.066

198.641

5.424E+00

3.514E+00



probita

4

0.112

198.166

9.103E+00

7.701E+00

negative intercept (intercept
= -1.821)

Weibullb

3

0.075

198.690

7.712E+00

4.692E+00



a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.3.37.2. Output for Selected Model: Probit

National Toxicology Program, 2006: Oval Cell Hyperplasia

pit

52 2010

0

Probit Model. (Version: 3.1; Date: 05/16/2008)

Input Data File: C:\1\53_NTP_2006_OvalHyper_Probit_l.(d)

Gnuplot Plotting File: C:\l\53_NTP_2006_OvalHyper_Probit_l.

Tue Feb 16 19:51:

The form of the probability function is:

P[response] = CumNorm(Intercept + Slope* Dose),

where CumNorm(.) is the cumulative normal distribution function

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-443 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Default Initial	(and Specified) Parameter Values

background =	0 Specified

intercept =	-1.92612

slope =	0.0670004

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

intercept	slope

intercept	1	-0.8

slope	-0.8	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

intercept	-1.82129	0.16954	-2.15359	-1.489

slope	0.0767832	0.00835175	0.060414	0.0931523

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)
-92.4898
-97.0832
-210.191

Param's Deviance Test d.f.

9.18683
235.402

P-value

0.0566
C.0001

198.166

Dose

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000

0

0343

1. 817

0

000

53

-1

372

2.1400

0

0488

2 . 633

4

000

54

0

864

7.1400

0

1015

5.379

3

000

53

-1

082

15.7000

0

2690

14 . 258

20

000

53

1

77 9

32.9000

0

7596

40.256

38

000

53

-0

725

71.4000

0

9 9 9 9

52.993

53

000

53

0

082

Chi ^2 = 7.50	d.f. = 4	P-value = 0.1119

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	9.1026

BMDL =	7.7011

This document is a draft for review purposes only and does not constitute Agency policy.

E-444 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.37.3. Figure for Selected Model: Probit

Probit Model with 0.95 Confidence Level

dose

19:51 02/16 2010

E.3.37.4. Output for Additional Model Presented: Weibull
National Toxicology Program, 2006: Oval Cell Hyperplasia

Weibull Model using Weibull Model (Version: 2.12; Date: 05/16/2008)
Input Data File: C:\1\53_NTP_2006_OvalHyper_Weibull_l.(d)

Gnuplot Plotting File: C:\l\53_NTP_2006_OvalHyper_Weibull_l.plt

Tue Feb 16 19:51:53 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(-slope*doseApower)]

Dependent variable = DichEff

Independent variable = Dose

Power parameter is restricted as power >=1

Total number of observations = 6

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-445 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial	(and Specified) Parameter Values

Background =	0.00925926

Slope =	0.00444 52

Power =	1.63009

Asymptotic Correlation Matrix of Parameter Estimates
Background	Slope	Power

Background	1	-0.63	0.61

Slope	-0.63	1	-0.99

Power	0.61	-0.99	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.021258	0.0198428	-0.0176332	0.0601492

Slope	0.0028715	0.00303327	-0.0030736	0.0088166

Power	1.76359	0.309457	1.15706	2.37011

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-92.4898	6

Fitted model	-96.3448	3	7.70998	3	0.0524

Reduced model	-210.191	1	235.402	5	<.0001

AIC:	198.69

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.0213

1.127

0

000

53

-1.073

2.1400

0.0320

1.725

4

000

54

1.760

7.1400

0.1073

5. 685

3

000

53

-1.192

15.7000

0.3234

17 .138

20

000

53

0.840

32.9000

0.7490

39.698

38

000

53

-0.538

71.4000

0.9953

52.750

53

000

53

0.501

Chi/N2 = 6.92	d.f. = 3	P-value = 0.0746

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	7.71171

BMDL =	4 . 69152

This document is a draft for review purposes only and does not constitute Agency policy.

E-446 DRAFT—DO NOT CITE OR QUOTE

-------
1

2	E.3.37.5. Figure for Additional Model Presented: Weibull

Weibull Model with 0.95 Confidence Level

dose

3	19:51 02/16 2010

4

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-447 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

E.3.38. National Toxicology Program, 2006: Pigmentation, Liver
E.3.38.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

3

0.385

197.655

1.547E+00

8.055E-01



logistic

4

<0.001

203.517

2.259E+00

1.872E+00

negative intercept (intercept =
-1.925)

log-logistic

3

0.978

195.600

2.212E+00

1.452E+00



log-probita

3

0.980

195.450

2.072E+00

1.399E+00



multistage, 5-
degree

3

0.210

199.850

9.396E-01

7.079E-01

final B = 0

probit

4

<0.001

210.309

2.259E+00

1.916E+00

negative intercept (intercept =
-1.057)

Weibull

3

0.290

198.489

1.280E+00

7.518E-01



a Best-fitting model, BMDS output presented in this appendix

E.3.38.2. Output for Selected Model: Log-Probit

National Toxicology Program, 2006: Pigmentation, Liver

Probit Model. (Version: 3.1; Date: 05/16/2008)

Input Data File: C:\1\54_NTP_2006_Pigment_LogProbit_l.(d)

Gnuplot Plotting File: C:\l\54_NTP_2006_Pigment_LogProbit_l.plt

Tue Feb 16 19:52:19 2 010

The form of the probability function is:

P[response] = Background

+ (1-Background) * CumNorm(Intercept + Slope^Log(Dose) ) ,

where CumNorm(.) is the cumulative normal distribution function

Dependent variable = DichEff
Independent variable = Dose

Slope parameter is restricted as slope >= 1
Total number of observations = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-448 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

User has chosen the log transformed model

Default Initial	(and Specified) Parameter Values

background =	0.0754717

intercept =	-1.91144

slope =	1.07 385

Asymptotic Correlation Matrix of Parameter Estimates

background
intercept
slope

background
1

-0.45
0.35

intercept

-0.45
1

-0. 94

slope

0.35
-0. 94
1

Parameter Estimates

Variable
background
intercept
slope

Estimate

0.0735956
-2 .19294
1. 25068

Std. Err.

0.0343284
0. 400053
0.169731

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

0. 00631316
-2 . 97703
0.918012

0.140878
-1.40885
1.58335

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table

Param's Deviance Test d.f.

Log(likelihood)

-94.6177
-94.7248
-210.717

195.45

0.214232
232.198

P-value

0. 9753
C.0001

Goodness of Fit

Est. Prob.

Expected

Observed

Scaled
Residual

0.0000

0

0736

3. 901

4

000

53

0

052

2.1400

0

1729

9.338

9

000

54

-0

122

7.1400

0

6338

33.591

34

000

53

0

117

15.7000

0

9023

47.822

48

000

53

0

082

32.9000

0

9863

52 . 275

52

000

53

-0

325

71.4000

0

9992

52.959

53

000

53

0

202

Chi ^2

0.18

d.f.

P-value

0.9801

Benchmark Dose Computation

Specified effect
Risk Type
Confidence level
BMD
BMDL

0.1
Extra risk

0. 95
2.07241
1.39932

This document is a draft for review purposes only and does not constitute Agency policy.

E-449 DRAFT—DO NOT CITE OR QUOTE

-------
l E.3.38.3. Figure for Selected Model: Log-Probit

LogProbit Model with 0.95 Confidence Level

dose

2	19:52 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-450 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

E.3.39. National Toxicology Program, 2006: Toxic Hepatopathy
E.3.39.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

4

0.772

185.634

4.668E+00

3.317E+00



logistic

4

0.012

198.445

7.070E+00

5.925E+00

negative intercept (intercept =
-2.925)

log-logistic

3

0.362

190.061

5.676E+00

4.040E+00



log-probit

3

0.378

189.858

6.061E+00

4.079E+00



multistage, 5-
degreea

4

0.577

186.521

4.163E+00

2.701E+00

final 15 = 0

probit

4

0.019

197.159

6.784E+00

5.712E+00

negative intercept (intercept =
-1.724)

Weibull

4

0.745

185.657

4.454E+00

3.159E+00



a Best-fitting model, BMDS output presented in this appendix

E.3.39.2. Output for Selected Model: Multistage, 5-Degree

National Toxicology Program, 2006: Toxic Hepatopathy

Multistage Model. (Version: 3.0; Date: 05/16/2008)

Input Data File: C:\1\55_NTP_2006_ToxHepa_Multi5_l.(d)

Gnuplot Plotting File: C:\l\55_NTP_2006_ToxHepa_Multi5_l.plt

Tue Feb 16 19:52:49 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/Nl-beta2*dose/N2-beta3*dose/N3-beta4*dose/N4-beta5*dose/N5) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 6
Total number of specified parameters = 0
Degree of polynomial = 5

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

E-451 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Parameter

has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) =	0

Beta(2) =	0

Beta(3) =	0

Beta(4) =	0

Beta(5) = 5.40983e+010

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(3)	-Beta(4)	-Beta(5)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)	Beta(2)

Beta(1)	1	-0.91

Beta(2)	-0.91	1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)
Beta(3)
Beta(4)
Beta(5)

Estimate

0

0. 019656
0. 00135796
0
0
0

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-89.8076
-91.2606
-218.207

Param's Deviance Test d.f.

2.90597
256.799

P-value

0.5737
C.0001

AIC:

186.521

Goodness of Fit

Scaled

Dose

Est

. Prob.

Expected

Observed

Size

Residual

0.0000

0.

0000

0. 000

0

000

53

0

000

2.1400

0.

0471

2 545

2

000

54

-0

350

7.1400

0.

1891

10.021

8

000

53

-0

709

15.7000

0.

4745

25.146

30

000

53

1

335

32.9000

0.

87 96

46.616

45

000

53

-0

682

71.4000

0.

9998

52 987

53

000

53

0

113

Chi ^2

2.89

d.f.

P-value

0.5771

Benchmark Dose Computation
Specified effect =	0.1

This document is a draft for review purposes only and does not constitute Agency policy.

E-452 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	4.16294

BMDL =	2.70063

BMDU =	6.0018 6

Taken together, (2.70063, 6.00186) is a 90	% two-sided confidence

interval for the BMD

E.3.39.3. Figure for Selected Model: Multistage, 5-Degree

Multistage Model with 0.95 Confidence Level

dose

19:52 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-453 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.3.40. Ohsako et al., 2001: Ano-Genital Length, PND 120
E.3.40.1. Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

3

0.019

171.804

5.650E+02

3.785E+02



exponential (M3)

3

0.019

171.804

5.650E+02

3.785E+02

power hit bound (d = 1)

exponential (M4)

2

0.117

168.204

2.854E+01

1.054E+01



exponential (M5)

1

0.049

169.789

2.948E+01

1.135E+01



Hill b

2

0.148

167.727

3.722E+01

9.752E+00

n lower bound hit (n = 1)

linear

3

0.018

171.954

5.852E+02

4.047E+02



polynomial, 4-
degree

3

0.018

171.954

5.852E+02

4.047E+02



power

3

0.018

171.954

5.852E+02

4.047E+02

power bound hit (power =1)

Hill, unrestricted0

1

0.055

169.600

5.101E+01

3.066E+00

unrestricted (n = 0.502)

power,
unrestricted

2

0.151

167.689

6.200E+01

2.291E+00

unrestricted (power = 0.252)

a Constant variance model selected (p = 0.165)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.40.2. Output for Selected Model: Hill

Ohsako et al., 2001: Ano-Genital Length, PND 120

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\56_Ohsako_2001_Anogen_HillCV_l.(d)

Gnuplot Plotting File: C:\l\56_Ohsako_2001_Anogen_HillCV_l.plt

Tue Feb_16 19:53:25 2010

Figure 7

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

This document is a draft for review purposes only and does not constitute Agency policy.

E-454 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Power parameter restricted to be greater than 1
A constant variance model is fit

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho

intercept

7 . 27386

0

28.905
-5.1065
1.40226
33.9669

Specified

Asymptotic Correlation Matrix of Parameter Estimates

alpha
intercept

The model parameter(s) -rho -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

-2 . 2e-009
-2 . 4e-008
-7 . 2e-009

intercept
-2.2e-009
1

-0. 66
-0.5

v

-2.4e-008
-0. 66
1

-0.11

k

-7 . 2e-009
-0.5
-0.11
1

Parameter Estimates

Variable
alpha
intercept

k

Estimate

7.08444
28 . 9809
-4.79692
1

29.8628

Std. Err.

1.3634
0.745637
0.983318
NA

24 . 4463

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

4 .41223

27 . 5195
-6.72418

-18.0511

9.75666

30.4423
-2.86965

77 .7767

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	12	28.9	29	3.13	2.66	-0.0988

12.5	10	27.9	27.6	2.5	2.66	0.442

50	10	25.2	26	3.21	2.66	-0.963

200	10	26	24.8	2.85	2.66	1.42

800	12	23.8	24.4	1.56	2.66	-0.726

Model Descriptions for likelihoods calculated

This document is a draft for review purposes only and does not constitute Agency policy.

E-455 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)

-77 . 952340
-74.703868
-77 . 952340
-79.863340
-89.824703

Param's

10

AIC

167.904680
169.407736
167.904680
167.726680
183.649405

Explanation of Tests

Test 1:

Test 2
Test 3
Test 4
(Note:

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

30.2417

6 . 4 9 6 9 4
6 . 4 9 6 9 4
3. 822

0.0001916
0.165
0.165
0.1479

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than
model appears to be appropriate here

.1. A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	37.224 9

BMDL =	9.75249

This document is a draft for review purposes only and does not constitute Agency policy.

E-456 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.40.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

19:53 02/16 2010

E.3.40.4. Output for Additional Model Presented: Hill, Unrestricted
Ohsako et al., 2001: Ano-Genital Length, PND 120

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\56_Ohsako_2001_Anogen_HillCV_U_l.(d)

Gnuplot Plotting File: C:\l\56_Ohsako_2 001_Anogen_HillCV_U_l.plt

Tue Feb 16 19:53:26 2010

Figure 7

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-457 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha =	7.27386

rho =	0 Specified

intercept =	28.905

v =	-5.1065

n =	1.40226

k =	33.9669

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
intercept

alpha
1

2 . le-009
-1. 8e-008
-1. 7e-008
1. 6e-008

intercept
2 . le-009
1

0. 012
0.0075
-0.13

v

-1.8e-008
0. 012
1

0. 98
-0. 99

n

-1. 7e-008
0. 0075
0. 98
1

-0. 97

k

1. 6e-008
-0.13
-0. 99
-0. 97
1

Parameter Estimates

Variable
alpha
intercept

Estimate

7 . 06785
28 . 9608
-6.94236
0.501942
131.957

Std. Err.

1.36021
0.755363
12 . 2514
0.915162
1071.9

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

4 .40189

27.4803
-30.9547
-1.29174
-1968.92

9.73381

30.4413
17 . 07
2.29563
2232.84

Table of Data and Estimated Values of Interest
5	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	12	28.9	29	3.13	2.66	-0.0727

12.5	10	27.9	27.3	2.5	2.66	0.72

50	10	25.2	26.3	3.21	2.66	-1.37

200	10	26	25.1	2.85	2.66	1.04

800	12	23.8	24	1.56	2.66	-0.287

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij|

This document is a draft for review purposes only and does not constitute Agency policy.

E-458 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -77.952340	6	167.904680

A2 -74.703868	10	169.407736

A3 -77.952340	6	167.904680

fitted -79.800035	5	169.600070

R -89.824703	2	183.649405

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	30.2417	8	0.0001916

Test 2	6.49694	4	0.165

Test 3	6.49694	4	0.165

Test 4	3.69539	1	0.05456

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	51.0107

BMDL =	3.06631

This document is a draft for review purposes only and does not constitute Agency policy.

E-459 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.40.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

19:53 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-460 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

E.3.41. Sewall et al., 1995: T4 In Serum
E.3.41.1. Summary Table of BMDS Modeling Results

Model3

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

3

0.424

205.966

5.762E+01

3.783E+01



exponential (M3)

3

0.424

205.966

5.762E+01

3.783E+01

power hit bound (d = 1)

exponential (M5)

2

0.611

206.152

2.523E+01

8.442E+00

power hit bound (d = 1)

Hillb

2

0.702

205.875

2.071E+01

5.164E+00

n lower bound hit (n = 1)

linear

3

0.332

206.584

6.788E+01

4.858E+01



polynomial, 4-
degree

3

0.332

206.584

6.788E+01

4.858E+01



power

3

0.332

206.584

6.788E+01

4.858E+01

power bound hit (power =1)

Hill, unrestricted0

1

0.844

207.205

1.657E+01

1.903E+00

unrestricted (n = 0.427)

power, unrestricted

2

0.983

205.200

1.658E+01

1.820E+00

unrestricted (power = 0.403)

a Constant variance model selected (p = 0.4078)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.41.2. Output for Selected Model: Hill

Sewall et al., 1995: T4 In Serum

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\58_Sewall_l995_T4_HillCV_l.(d)

Gnuplot Plotting File: C:\l\58_Sewall_1995_T4_HillCV_l.plt

Tue_Feb 16~19:54:30 2010

Figure 1, Saline noninitiated

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter restricted to be greater than 1
A constant variance model is fit

Total number of dose groups = 5

Total number of records with missing values = 0

This document is a draft for review purposes only and does not constitute Agency policy.

E-461 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho

intercept

33.0913

0

30.6979
-12 .2937
0.695384
24 . 6674

Specified

Asymptotic Correlation Matrix of Parameter Estimates

alpha
intercept

The model parameter(s) -rho -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

1. 2e-008
4 . le-008
-2 . 4e-008

intercept
1. 2e-008
1

0.14
-0. 66

v

4 . le-008
0.14
1

-0.76

k

-2 . 4e-008
-0. 66
-0.76
1

Parameter Estimates

Variable
alpha
intercept

k

Estimate

29.8807
2 9.9609
-14.2338
1

33.2198

Std. Err.
6.29941
1.64749
4 . 35645
NA

37.0852

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

17.5341
26.7319
-22 .7723

-39.4658

42 . 2274
33.1899
-5.69537

105.905

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

9

30

7

30

4

6 6

5.47

0

404

3.5

9

27

9

28 . 6

7

17

5.47

-0

399

10.7

9

25

9

26.5

6

81

5.47

-0

328

35

9

23

6

22 . 7

5

38

5.47

0

493

125

9

18

4

i—1

CO

-j

4

12

5.47

-0

171

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-462 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -98.583448	6	209.166896

A2 -96.590204	10	213.180407

A3 -98.583448	6	209.166896

fitted -98.937315	4	205.874631

R -109.013252	2	222.026503

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	24.8461	8	0.001651

Test 2	3.98649	4	0.4078

Test 3	3.98649	4	0.4078

Test 4	0.707735	2	0.702

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	2 0.7117

BMDL =	5.16405

This document is a draft for review purposes only and does not constitute Agency policy.

E-463 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.41.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

19:54 02/16 2010

E.3.41.4. Output for Additional Model Presented: Hill, Unrestricted
Sewall et al., 1995: T4 In Serum

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\58_Sewall_l995_T4_HillCV_U_l.(d)

Gnuplot Plotting File: C:\l\58_Sewall_1995_T4_HillCV_U_l.plt

Tue Feb 16 19:54:31 2010

Figure 1, Saline noninitiated

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-464 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho

intercept

33.0913

0

30.6979
-12 .2937
0.695384
24 . 6674

Specified

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
intercept

alpha
1

-0.0004
0.0059
0.0048
-0.0059

intercept

-0.0004
1

-0.026
-0.44
0. 07

v

0. 0059
-0.026
1

0 .77
-1

n

0.0048
-0.44
0 .77
1

-0. 82

k

-0.0059
0. 07
-1
-0. 82
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
intercept

Estimate

29.4396
30.6757
-141.324
0.426599
31487

Std. Err.
6.20653
1.77521
1202.4
0.262207
770429

Lower Conf. Limit

17 . 2751
27 .1963
-2497.98
-0.0873175
-1. 47853e + 006

Upper Conf. Limit

41.6042
34.155
2215.33
0.940515
1.5415e+006

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

9

30

7

30.7

4

6 6

5.43

0.0123

3.5

9

27

9

27 . 8

7

17

5.43

0.0279

10.7

9

25

9

26.1

6

81

5.43

-0.137

35

9

23

6

23.3

5

38

5.43

0.132

125

9

18

4

18 . 5

4

12

5.43

-0.0354

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

This document is a draft for review purposes only and does not constitute Agency policy.

E-465 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model
A1
A2
A3

fitted
R

Log(likelihood)
-98.583448
-96.590204
-98.583448
-98.602701
-109.013252

10

AIC
209.166896
213.180407
209.166896
207.205403
222.026503

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

24.8461
3.98649
3.98649
0.0385071

0.001651
0.4078
0.4078
0.8444

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	16.568 9

BMDL =	1.90347

This document is a draft for review purposes only and does not constitute Agency policy.

E-466 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.41.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

19:54 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-467 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

E.3.42. Shi et al., 2007: Estradiol 17B, PE9
E.3.42.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

3

0.001

395.701

1.729E+01

8.956E+00



exponential (M3)

3

0.001

395.701

1.729E+01

8.956E+00

power hit bound (d = 1)

exponential (M4)b

2

0.494

383.635

5.559E-01

2.236E-01



exponential (M5)

2

0.494

383.635

5.559E-01

2.236E-01

power hit bound (d = 1)

Hill

2

0.773

382.743

4.434E-01

error

n lower bound hit (n = 1)

linear

3

0.001

397.484

2.243E+01

1.523E+01



polynomial, 4-
degree

3

0.001

397.484

2.243E+01

1.523E+01



power

3

0.001

397.484

2.243E+01

1.523E+01

power bound hit (power = 1)

Hill, unrestricted

1

0.874

384.251

3.998E-01

error

unrestricted (n = 0.616)

power, unrestricted

2

0.506

383.589

3.409E-01

5.002E-03

unrestricted (power = 0.155)

a Non-constant variance model selected (p = 0.0521)
b Best-fitting model, BMDS output presented in this appendix

E.3.42.2. Output for Selected Model: Exponential (M4)

Shi et al., 2007: Estradiol 17B, PE9

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\59_Shi_2007_Estradiol_Exp_l.(d)

Gnuplot Plotting File:

Tue Feb 16 19:55:06 2010

Figure 4 PE9 only

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.

This document is a draft for review purposes only and does not constitute Agency policy.

E-468 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model 3 is nested within Model 5.
Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	2.65881

rho	0.913414

a	108

b	0.136287

c	0.340136

d	1

Parameter Estimates
Variable	Model 4

lnalpha	1.81331

rho	1.12126

a	100.526

b	1.53823

c	0.431796

d	1

Table of Stats From Input	Data

Dose	N	Obs Mean	Obs Std Dev

0	10	102.9	41.41

0.143	10	86.19	19.58

0.714	10	63.33	29.36

7.14	10	48.1	18.82

28.6	10	38.57	22.59

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	100.5	32.83	0.2245

0.143	89.25	30.71	-0.3147

0.714	62.45	25.14	0.1108

7.14	43.41	20.5	0.723

28.6	43.41	20.5	-0.7458

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

This document is a draft for review purposes only and does not constitute Agency policy.

E-469 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Var{e(ij)}	= Sigma^2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i) ) 'k rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-188.3615	6	388.7231

A2	-183.667	10	387.3339

A3	-186.1132	7	386.2263

R	-203.3606	2	410.7211

4	-186.8176	5	383.6352

Additive constant for all log-likelihoods =	-45.95. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Does response

and/or variances differ among Dose levels? (A2 vs. R)

Test

2 :

Are Variances

Homogeneous? (A2 vs. A1)

Test

3:

Are variances

adequately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

39.39
9.389
4 . 892
1.409

D. F.

p-value

< 0.0001
0.05208
0.1798
0.4944

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	0.555948

This document is a draft for review purposes only and does not constitute Agency policy.

E-470 DRAFT—DO NOT CITE OR QUOTE

-------
0.223612

5 E.3.42.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

a>

CO
c
o

Q.

CO
CD

a:

c
m

CD

140

120

100

80

60

40

20

E:MDL

BMD

19:55 02/16 2010

Exponential

10

15
dose

20

25

30

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-471 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.3.43. Smialowicz et al., 2008: PFC per 10A6 Cells
E.3.43.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

3

0.048

903.586

8.234E+01

4.833E+01



exponential (M3)

3

0.048

903.586

8.234E+01

4.833E+01

power hit bound (d = 1)

exponential (M4)

2

0.019

905.578

8.032E+01

6.220E+00



exponential (M5)

2

0.019

905.578

8.032E+01

6.220E+00

power hit bound (d = 1)

Hill

2

0.026

904.975

1.617E+01

2.214E+00

n lower bound hit (n = 1)

linear

3

0.016

905.992

1.450E+02

1.102E+02



polynomial, 4-
degree

2

<0.0001

1198.471

1.375E+03

3.331E+01



power0

3

0.016

905.992

1.450E+02

1.102E+02

power bound hit (power =1)

Hill, unrestricted

1

0.183

901.442

8.297E+00

4.172E-01

unrestricted (n = 0.266)

power,
unrestricted b

2

0.446

899.282

7.676E+00

4.087E-01

unrestricted (power = 0.249)

a Constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.43.2. Output for Selected Model: Power, Unrestricted

Smialowicz et al., 2008: PFC per 10A6 Cells

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\60_Smial_2008_PFCcells_PwrCV_U_l.(d)

Gnuplot Plotting File: C:\l\60_Smial_2008_PFCcells_PwrCV_U_l.plt

_Tue Feb 16 19:55753 2010

Anti Response to SRBCs, PFC per 10to6 cells, Table 4

The form of the response function is:

Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

This document is a draft for review purposes only and does not constitute Agency policy.

E-472 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The power is not restricted
A constant variance model is fit

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha
rho
control
slope
power

232385

0

1491
-384.362
0.215085

Specified

Asymptotic Correlation Matrix of Parameter Estimates

alpha
control

The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

-1. 5e-009

control
-1.5e-009
1

slope
J. 2e-009
-0.79

power
-1. le-008
-0. 65

slope -8.2e-009	-0.79	1	0.96

power -1.le-008	-0.65	0.96	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
control
slope
power

Estimate

220294
1470.38
-282.777
0.248621

Std. Err.

38061.1
124.07
145.113
0.0856348

Lower Conf. Limit

145696
1227.21
-567.193
0.0807799

Upper Conf. Limit

294893
1713.55
1. 64025
0. 416462

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

15

1. 4 9e + 003

1. 47e + 003

716

469

0.17

1 . 07

14

1.13e + 003

1.18e + 003

171

469

-0.429

10.7

15

945

961

516

469

-0.129

107

15

677

567

465

469

0. 91

321

8

161

283

117

469

-0.735

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-473 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1 -444.832859	6	901.665718

A2 -425.402825	10	870.805651

A3 -444.832859	6	901.665718

fitted -445.641102	4	899.282205

R -463.753685	2	931.507371

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	76.7017	8	<.0001

Test 2	38.8601	4	<.0001

Test 3	38.8601	4	<.0001

Test 4	1.61649	2	0.4456

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. Consider running a
non-homogeneous variance model

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 7.67 564

BMDL = 0.408661

This document is a draft for review purposes only and does not constitute Agency policy.

E-474 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.43.3. Figure for Selected Model: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

19:55 02/16 2010

E.3.43.4. Output for Additional Model Presented: Power
Smialowicz et al., 2008: PFC per 10A6 Cells

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\60_Smial_2008_PFCcells_PwrCV_l.(d)

Gnuplot Plotting File: C:\l\60_Smial_2008_PFCcells_PwrCV_l.plt

Tue Feb 16 19:55:53 2010

Anti Response to SRBCs, PFC per 10to6 cells, Table 4

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

The power is restricted to be greater than or egual to 1
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-475 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha =	232385

rho =	0 Specified

control =	1491

slope =	-2925.99

power = -0.136613

Asymptotic Correlation Matrix of Parameter Estimates

alpha
control
slope

The model parameter(s) -rho -power

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

3.6e-0 0 9
-1.2e-008

control

3.6e-0 0 9
1

-0.53

slope
-1.2e-008
-0.53
1

Parameter Estimates

Variable
alpha
control
slope
power

Estimate

250878
1176.24
-3.45384
1

Std. Err.
43345.1
72 .2586
0.592114
NA

NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

165923
1034.61
-4.61436

335833
1317.86
-2 .29332

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

15

1. 4 9e + 003

1.18e + 003

716

501

2 .43

1 . 07

14

1.13e + 003

1.17e + 003

171

501

-0.325

10.7

15

945

1.14e + 003

516

501

-1. 5

107

15

677

807

465

501

-1

321

8

161

67 . 6

117

501

0.528

Model Descriptions for likelihoods calculated

Model A1:

Yij

Var{e ( ij ;

Mu(i) + e(ij ;
Sigma/N2

Model A2:

Yij

Var{e ( ij ;

Mu(i) + e(ij ;

Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

E-476 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma^2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1 -444.832859	6	901.665718

A2 -425.402825	10	870.805651

A3 -444.832859	6	901.665718

fitted -449.996183	3	905.992366

R -463.753685	2	931.507371

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	76.7017	8	<.0001

Test 2	38.8601	4	<.0001

Test 3	38.8601	4	<.0001

Test 4	10.3266	3	0.01598

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. Consider running a
non-homogeneous variance model

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 145.02

BMDL = 110.161

This document is a draft for review purposes only and does not constitute Agency policy.

E-477 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.43.5. Figure for Additional Model Presented: Power

Power Model with 0.95 Confidence Level

dose

19:55 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-478 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.3.44. Smialowicz et al., 2008: PFC per Spleen

E.3.44.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

3

0.133

377.395

1.320E+02

8.431E+01



exponential (M3)

3

0.133

377.395

1.320E+02

8.431E+01

power hit bound (d = 1)

exponential (M4)

3

0.133

377.395

1.320E+02

8.184E+01



exponential (M5)

2

0.061

379.395

1.320E+02

8.184E+01

power hit bound (d = 1)

Hill

2

0.069

379.150

1.401E+02

error

n lower bound hit (n = 1)

linear

3

0.044

379.895

2.151E+02

1.704E+02



polynomial, 4-
degree

3

0.044

379.895

2.151E+02

1.704E+02



power0

3

0.044

379.895

2.151E+02

1.704E+02

power bound hit (power =1)

Hill, unrestricted

2

<0.0001

441.885

7.545E-23

error

unrestricted (n = 0.038)

power,
unrestricted b

2

0.230

376.738

9.374E+01

2.088E+01

unrestricted (power = 0.418)

a Non-constant variance model selected (p = 0.0011)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.44.2. Output for Selected Model: Power, Unrestricted

Smialowicz et al., 2008: PFC per Spleen

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\61_Smial_2008_PFCspleen_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\61_Smial_2008_PFCspleen_Pwr_U_l.plt

_Tue Feb 16 19756:26 2010

Anti Response to SRBCs - PFC x 10 to the 4 per spleen, Table 4

The form of the response function is:

Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
The power is not restricted

This document is a draft for review purposes only and does not constitute Agency policy.

E-479 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	4.76607

rho =	0

control =	27.8

slope =	-7.21601

power =	0.213905

Asymptotic Correlation Matrix of Parameter Estimates

lalpha	rho	control	slope	power

lalpha 1	-0.98	0.25	-0.27	-0.23

rho -0.98	1	-0.31	0.28	0.23

control 0.25	-0.31	1	-0.81	-0.74

slope -0.27	0.28	-0.81	1	0.99

power -0.23	0.23	-0.74	0.99	1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate
0.747155
1. 36972
25.1733
-1.98465
0. 417867

Std. Err.

1. 0244
0.357098
2.93169
1. 82113
0.141932

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-1.26063

0.66982
19.4273
-5.554
0.139686

2.75494

2.06962
30.9193
1.5847
0.696048

Table of Data and Estimated Values of Interest

Dose

N

Obs Mean

Est Mean

Obs Std Dev

Est Std Dev

Scaled Res

0

15

27 . 8

25.2

13

4

13.2

0.769

1. 07

14

21

23.1

13

6

12 . 5

-0.639

10.7

15

17 . 6

19.8

9

4

11. 2

-0.768

107

15

12 . 6

11 2

8

7

7 .59

0.721

321

8

3

3. 04

3

1

3.11

-0.0353

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))

This document is a draft for review purposes only and does not constitute Agency policy.

E-480 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -190.565019	6	393.130038

A2 -181.476284	10	382.952569

A3 -181.900030	7	377.800059

fitted -183.369059	5	376.738118

R -204.636496	2	413.272993

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	46.3204	8	<.0001

Test 2	18.1775	4	0.001139

Test 3	0.84749	3	0.8381

Test 4	2.93806	2	0.2301

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 93.7416

BMDL = 20.8758

This document is a draft for review purposes only and does not constitute Agency policy.

E-481 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.44.3. Figure for Selected Model: Power, Unrestricted

Power Model with 0.95 Confidence Level

35

30

25

CD
CO
c
o

Q.

CO
CD

cr

c
ro

CD

20

15

10

5

0

0	50	100	150	200	250	300

dose

19:56 02/16 2010

E.3.44.4. Output for Additional Model Presented: Power
Smialowicz et al., 2008: PFC per Spleen

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\61_Smial_2008_PFCspleen_Pwr_l.(d)

Gnuplot Plotting File: C:\l\61_Smial_2008_PFCspleen_Pwr_l.plt

Tue Feb 16 19:56:25 2010

Anti Response to SRBCs - PFC x 10 to the 4 per spleen, Table 4

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose

The power is restricted to be greater than or egual to 1

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-482 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 5

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	4.76607

rho =	0

control =	27.8

slope =	-54.5244

power = -0.136501

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -power

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha	rho	control	slope

lalpha	1	-0.98	0.16	-0.48

rho	-0.98	1	-0.25	0.54

control
slope

o.ie

-0. 4E

-0.25
0.54

-0. I

-OA

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
control
slope
power

Estimate

0. 474614
1.48709
21.3571
-0.0574184
1

Std. Err.
1.09569
0.385029
1.69233
0. 00632057
NA

Lower Conf. Limit

-1.6729
0.732449
18.0402
-0.0698064

Upper Conf. Limit

2.62213
2 .24173
24.674
-0.0450303

NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

15

27 . 8

21. 4

13

4

12 . 3

2 . 02

1 . 07

14

21

21. 3

13

6

12 . 3

-0.0898

10.7

15

17 . 6

20.7

9

4

12 .1

-1. 01

107

15

12 . 6

15.2

8

7

9.6

-1. 05

321

8

3

2 . 93

3

1

2 . 82

0.0745

Model

Descriptions for

likelihoods

calculated







Model

A1:

Yij =

Mu(i) + e(ij

)









Var {e

ij ) } =

Sigma/N2









Model A2:	Yij = Mu(i) + e(ij)

This document is a draft for review purposes only and does not constitute Agency policy.

E-483 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -190.565019	6	393.130038

A2 -181.476284	10	382.952569

A3 -181.900030	7	377.800059

fitted -185.947278	4	379.894555

R -204.636496	2	413.272993

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1
Test 2
Test 3
Test 4

46.3204
18.1775
0.84749
8.0945

<.0001
0.001139
0.8381
0.0441

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 215.073

BMDL = 17 0.412

This document is a draft for review purposes only and does not constitute Agency policy.

E-484 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.44.5. Figure for Additional Model Presented: Power

Power Model with 0.95 Confidence Level

dose

19:56 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-485 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.3.45. Toth et al., 1979: Amyloidosis

E.3.45.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

2

0.022

150.666

2.296E+02

1.460E+02

power bound hit (power =1)

logistic

2

0.013

152.187

4.088E+02

3.125E+02

negative intercept (intercept = -
2.098)

log-logistica

2

0.028

149.984

1.759E+02

9.729E+01

slope bound hit (slope = 1)

log-probit

2

0.007

153.479

4.402E+02

2.965E+02

slope bound hit (slope =1)

multistage, 3-
degree

2

0.022

150.666

2.296E+02

1.460E+02

final B = 0

probit

2

0.014

152.040

3.846E+02

2.911E+02

negative intercept (intercept = -
1.238)

Weibull

2

0.022

150.666

2.296E+02

1.460E+02

power bound hit (power =1)

gamma,
unrestricted

2

0.917

140.208

7.687E-01

7.637E-04

unrestricted (power =0.187)

log-logistic,
unrestricted b

2

0.847

140.370

8.465E-01

1.565E-03

unrestricted (slope = 0.238)

log-probit,
unrestricted

2

0.811

140.458

8.545E-01

2.334E-03

unrestricted (slope = 0.135)

Weibull,
unrestricted

2

0.882

140.287

8.179E-01

1.140E-03

unrestricted (power = 0.212)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.3.45.2. Output for Selected Model: Log-Logistic

Toth et al., 1979: Amyloidosis

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\62_Toth_l979_Amylyr_LogLogistic_l.(d)

Gnuplot Plotting File: C:\l\62_Toth_197 9_Amylyr_LogLogistic_l.plt

Tue Feb 16 19:56:59 2010

Table 2

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose

This document is a draft for review purposes only and does not constitute Agency policy.

E-486 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Slope parameter is restricted as slope >= 1

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background =	0

intercept =	-6.90711

slope =	1

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -slope

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

background intercept

background	1	-0.47

intercept	-0.47	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf. Limit Upper Conf. Limit

background 0.0848984 *	*	*

intercept -7.36716 *	*	*

slope 1 *	*	*

Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-68.017	4

Fitted model	-72.9918	2	9.9496	2	0.00691

Reduced model	-82.0119	1	27.99	3	<.0001

AIC:	149.984

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0849 3.226 0.000	38	-1.878

1.0000 0.0855 3.761 5.000	44	0.668

100.0000 0.1393 6.128 10.000	44	1.686

1000.0000 0.4392 18.884 17.000	43	-0.579

Chi ^2 = 7.15	d.f. = 2	P-value = 0.0280

Benchmark Dose Computation
Specified effect =	0.1

This document is a draft for review purposes only and does not constitute Agency policy.

E-487 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Risk Type
Confidence level
BMD
BMDL

Extra risk
0. 95
175.903
97.2899

E.3.45.3. Figure for Selected Model: Log-Logistic

Log-Logistic Model with 0.95 Confidence Level

"O

(D

-t—<

o

CD
!t=
<

0.6

0.5

0.4

0.3

0.2

0.1

200

400

600

800

1000

dose

19:56 02/16 2010

E.3.45.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
Toth et al., 1979: Amyloidosis

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\62_Toth_l979_Amylyr_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\62_Toth_197 9_Amylyr_LogLogistic_U_l.plt

Tue Feb 16 19:57:00 2010

Table 2

This document is a draft for review purposes only and does not constitute Agency policy.

E-488 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff
Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background =	0

intercept =	-2.10894

slope =	0.227921

Asymptotic Correlation Matrix of Parameter Estimates

The model parameter(s) -background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

intercept	slope

intercept	1	-0.89

slope	-0.89	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0	*	*	*

intercept	-2.15753	*	*	*

slope	0.238304	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-68.017	4

Fitted model	-68.1848	2	0.33571	2	0.8455

Reduced model	-82.0119	1	27.99	3	<.0001

AIC:	140.37

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0000 0.000 0.000	38 0.000

1.0000 0.1036 4.560 5.000	44 0.218

100.0000 0.2573 11.321 10.000	44	-0.456

1000.0000 0.3749 16.119 17.000	43 0.277

This document is a draft for review purposes only and does not constitute Agency policy.

E-489 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

Chi' 2 = 0.33	d.f. = 2

E'-value =

0.8471

E'.enchmark Dose Computation
Specified effect =	0 .1

Risk Typ'e	=	E::tra risk

Cc'nfidence level =	0.95

BMD =	0.846547

BMDL =	0.0 015 653 4

E.3.45.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

"O

(D

-t—<

o

CD
!t=
<

200

400	600

dose

800

1000

19:57 02/16 2010

This document is a draft for review purposes only and does not constitute Agency policy.

E-490 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

E.3.46. Toth et al., 1979: Skin Lesions

E.3.46.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

gamma

2

0.009

159.223

1.181E+02

8.308E+01

power bound hit (power =1)

logistica

2

0.002

162.974

2.709E+02

2.147E+02

negative intercept (intercept =
-2.098)

log-logistic

2

0.029

156.567

6.750E+01

4.057E+01

slope bound hit (slope =1)

log-probit

2

0.001

164.598

2.446E+02

1.626E+02

slope bound hit (slope =1)

multistage, 3-
degree

2

0.009

159.223

1.181E+02

8.308E+01

final B = 0

probit

2

0.003

162.684

2.522E+02

2.015E+02

negative intercept (intercept = -
1.238)

Weibull

2

0.009

159.223

1.181E+02

8.308E+01

power bound hit (power =1)

gamma,
unrestricted

2

0.882

147.287

error

error

unrestricted (power = 0.251)

log-logistic,
unrestricted b

2

0.630

147.969

1.137E+00

5.477E-02

unrestricted (slope = 0.351)

log-probit,
unrestricted

2

0.558

148.218

1.096E+00

6.847E-02

unrestricted (slope = 0.202)

Weibull,
unrestricted

2

0.762

147.581

1.077E+00

4.080E-02

unrestricted (power = 0.3)

a Best-fitting model, BMDS output presented in this appendix
b Alternate model, BMDS output also presented in this appendix

E.3.46.2. Output for Selected Model: Logistic

Toth et al., 1979: Skin Lesions

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\63_Toth_l979_SkinLes_Logistic_l.(d)

Gnuplot Plotting File: C:\l\63_Toth_197 9_SkinLes_Logistic_l.plt

Tue Feb 16 19:57:29 2010

Table 2

The form of the probability function is:

P[response] = 1/[1+EXP(-intercept-slope^dose)]

Dependent variable = DichEff
Independent variable = Dose

This document is a draft for review purposes only and does not constitute Agency policy.

E-491 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Slope parameter is not restricted
Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

background =	0 Specified

intercept =	-2.53484

slope = 0.00299511

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

intercept	slope

intercept	1	-0.67

slope	-0.67	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

intercept	-1.91768	0.26892	-2.44475	-1.39061

slope	0.00230499	0.000419329	0.00148312	0.00312686

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-71.5177	4

Fitted model -79.487 2	15.9387 2	0.0003459

Reduced model -95.8498 1	48.6642 3	<.0001

AIC:	162.974

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.1281

4 .869

0. 000

38

-2.363

1.0000

0.1284

5. 649

5. 000

44

-0.292

100.0000

0.1561

6. 870

13.000

44

2.546

1000.0000

0.5956

25.612

25.000

43

-0.190

Chi ^2 = 12.19	d.f. = 2	P-value = 0.0023

Benchmark Dose Computation
Specified effect =	0.1

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	270.917

This document is a draft for review purposes only and does not constitute Agency policy.

E-492 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E'.MDL =

214.66

E.3.46.3. Figure for Selected Model: Logistic

Logistic Model with 0.95 Confidence Level

200	400	600	800	1000

dose

19:57 02/16 2010

E.3.46.4. Output for Additional Model Presented: Log-Logistic, Unrestricted
Toth et al., 1979: Skin Lesions

			 1 ¦ i ¦ ¦ ¦ 1 1 ¦ ¦ ¦ ¦ i ¦ 1 			 1 			 1 			 1 1 1 r

Logistic 	

BMDL

j	i	i	L

BMD

Logistic Model. (Version: 2.12; Date: 05/16/2008)

Input Data File: C:\1\63_Toth_l979_SkinLes_LogLogistic_U_l.(d)

Gnuplot Plotting File: C:\l\63_Toth_197 9_SkinLes_LogLogistic_U_l.plt

Tue Feb 16 20:01:56 2010

Table 2

The form of the probability function is:

P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = DichEff

This document is a draft for review purposes only and does not constitute Agency policy.

E-493 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Independent variable = Dose
Slope parameter is not restricted

Total number of observations = 4

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User has chosen the log transformed model

Default Initial Parameter Values
background =	0

intercept =	-2.14055

slope =	0.332 4 0 9

Asymptotic Correlation Matrix of Parameter Estimates

The model parameter(s) -background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

intercept	slope

intercept	1	-0.9

slope	-0.9	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

background	0	*	*	*

intercept	-2.24241	*	*	*

slope	0.350932	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-71.5177	4

Fitted model -71.9844 2	0.93345 2	0.6271

Reduced model -95.8498 1	48.6642 3	<.0001

AIC:	147.969

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000
1.0000
100.0000
1000.0000

0.0000
0.0960
0.3483
0.5453

0. 000
4 . 224
15.327
23.448

0. 000
5. 000
13.000
25.000

38
44
44
43

0. 000
0.397
-0.736
0.475

Chi ^2

0. 93

d.f.

P-value

0.6295

Benchmark Dose Computation

This document is a draft for review purposes only and does not constitute Agency policy.

E-494 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

Specified effect =	0.1

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	1.137 4

BMDL =	0.0547689

E.3.46.5. Figure for Additional Model Presented: Log-Logistic, Unrestricted

Log-Logistic Model with 0.95 Confidence Level

200	400	600	800	1000

dose

20:01 02/16 2010

			 1 1 i 1 1 1 1 1 1 1 1 1 i 1 1 			 1 			 1 			 1 1 1 r

Log-Logistic 	

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-495 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.3.47. Van Birgelen et al., 1995a: Hepatic Retinol
E.3.47.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

4

<0.0001

164.340

2.912E+02

error



exponential (M3)

4

<0.0001

164.340

2.912E+02

error

power hit bound (d = 1)

exponential (M4)b

3

<0.0001

148.052

1.151E+02

7.098E+01



exponential (M5)

3

<0.0001

148.052

1.151E+02

7.098E+01

power hit bound (d = 1)

Hill

3

0.044

128.757

1.314E+01

error

n lower bound hit (n = 1)

linear

4

<0.0001

178.734

7.815E+02

5.997E+02



polynomial, 5-degree

0

N/A

283.606

2.481E+03

error



power

4

<0.0001

178.734

7.815E+02

5.997E+02

power bound hit (power =1)

Hill, unrestricted

2

0.269

125.273

5.561E+00

error

unrestricted (n = 0.571)

power, unrestricted0

3

0.025

129.990

4.205E-01

8.504E-03

unrestricted (power = 0.118)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.47.2. Output for Selected Model: Exponential (M4)

Van Birgelen et al., 1995a: Hepatic Retinol

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\1\65_VanB_l995a_HepRet_Exp_l.(d)

Gnuplot Plotting File:

Tue Feb 16 20:03:05 2010

Tbl3, hepatic retinol

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;

This document is a draft for review purposes only and does not constitute Agency policy.

E-496 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

lnalpha	-1.16065

rho	1.53688

a	15.645

b	0.00625117

c	0.0365247

d	1

Parameter Estimates
Variable	Model 4

lnalpha	-0.882225

rho	1.82707

a	10.5294

b	0.00720346

c	0.0688661

d	1

Table of Stats From Input Data
Dose	N	Obs Mean	Obs Std Dev

0	8	14.9	8.768

14	8	8.4	3.394

26	8	8.2	2.263

47	8	5.1	0.8485

320	8	2.2	0.8485

1024	8	0.6	0.5657

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	10.53	5.526	2.237

14	9.589	5.073	-0.6628

26	8.855	4.717	-0.3926

47	7.714	4.159	-1.778

320	1.703	1.046	1.343

1024	0.7313	0.4833	-0.7681

This document is a draft for review purposes only and does not constitute Agency policy.

E-497 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Other models for which	likelihoods are calculated:

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-87.1567	7	188.3134

A2	-47.28742	12	118.5748

A3	-55.32422	8	126.6484

R	-109.967	2	223.934

4	-69.02619	5	148.0524

Additive constant for all log-likelihoods =	-44.11. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Does response

and/or variances differ among Dose levels? (A2 vs. R)

Test

2 :

Are Variances

Homogeneous? (A2 vs. A1)

Test

3:

Are variances

adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

125. 4
79.74
16. 07
27 . 4

10
5
4
3

p-value

<	0.0001

<	0.0001
0. 002922

<	0.0001

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is less than .1. You may want to
consider a different variance model.

The p-value for Test 6a is less than .1. Model 4 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

This document is a draft for review purposes only and does not constitute Agency policy.

E-498 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	115.128

BMDL =	70.981

E.3.47.3. Figure for Selected Model: Exponential (M4)

Exponential_beta Model 4 with 0.95 Confidence Level

dose

20:03 02/16 2010

E.3.47.4. Output for Additional Model Presented: Power, Unrestricted
Van Birgelen et al., 1995a: Hepatic Retinol

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\65_VanB_l995a_HepRet_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\65_VanB_19 95a_HepRet_Pwr_U_l.plt

Tue Feb 16 20:03:11 2010

Tbl3, hepatic retinol

The form of the response function is:

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-499 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Y[dose]

control + slope * doseApower

Dependent variable = Mean

Independent variable = Dose

The power is not restricted

The variance is to be modeled as Var(i)

exp(lalpha + log(mean(i))

rho)

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	2.76506

rho =	0

control =	14.9

slope =	-3.78637

power =	0.191713

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope
power

lalpha
1

-0.8

-0.047
0. 042
0. 065

rho
-0.8

1

-0.085
-0.0029
-0.11

control

-0.047
-0.085
1

-0. 95
-0. 81

slope

0. 042
-0.0029
-0. 95
1

0. 96

power

0. 065
-0.11
-0. 81
0. 96
1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate

-1.02622
1. 68421
16.9577
-7 .19097
0.117935

Std. Err.
0.389164
0.199212
2.21133
1.99708
0.0225396

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-1.78897

1. 29376
12.6235
-11.1052
0.0737578

-0.263475

2 . 07466
21.2918
-3.27676
0.162111

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	8	14.9	17	8.77	6.4 9	-0.896

14	8	8.4	7.14	3.39	3.13	1.14

26	8	8.2	6.4	2.26	2.86	1.78

47	8	5.1	5.63	0.849	2.57	-0.588

320	8	2.2	2.76	0.849	1.41	-1.12

1024	8	0.6	0.672	0.566	0.428	-0.475

Model Descriptions for likelihoods calculated

This document is a draft for review purposes only and does not constitute Agency policy.

E-500 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)

-87 .156698
-47.287416
-55.324218
-59.994980
-109.967018

Param's

7
12

8
5
2

AIC

188.313395
118.574833
126.648436
129.989960
223.934036

Test

1

Test

2

Test

3

Test

4

(Note:

Explanation of Tests

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

125.359
79.7386
16.0736
9.34152

10
5
4
3

<.0001

<.0001
0.002922
0.02508

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 0.420475

BMDL = 0.00850422

This document is a draft for review purposes only and does not constitute Agency policy.

E-501 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.47.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

20:03 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-502 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.3.48. Van Birgelen et al., 1995a: Hepatic Retinol Palmitate
E.3.48.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

4

<0.0001

467.446

error

error



exponential (M3)

4

<0.0001

467.446

error

error

power hit bound (d = 1)

exponential (M4)

3

<0.0001

454.087

error

error



exponential (M5)

3

<0.0001

454.087

error

error

power hit bound (d = 1)

Hill

3

<0.0001

563.579

error

error



linear b

4

<0.0001

488.446

1.420E+03

9.889E+02



polynomial, 5-
degree

0

N/A

573.977

error

error



power

4

<0.0001

488.446

1.420E+03

9.889E+02

power bound hit (power =1)

Hill, unrestricted

3

<0.0001

522.322

2.418E-12

2.418E-12

unrestricted (n = 0.452)

power,
unrestricted0

3

0.348

408.062

3.765E-02

1.208E-05

unrestricted (power = 0.054)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.48.2. Output for Selected Model: Linear

Van Birgelen et al., 1995a: Hepatic Retinol Palmitate

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\1\66_VanB_l995a_HepRetPalm_Linear_l.(d)

Gnuplot Plotting File: C:\l\66_VanB_1995a_HepRetPalm_Linear_l.plt

_Tue Feb 16~20:03:l6 2010

Tbl3, hepatic retinol palmitate

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose

Signs of the polynomial coefficients are not restricted

This document is a draft for review purposes only and does not constitute Agency policy.

E-503 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The variance is to be modeled as Var(i)

exp(lalpha + log(mean(i))

rho)

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	9.57332

rho =	0

beta_0 =	177 . 506

beta 1 = -0.204775

Asymptotic Correlation Matrix of Parameter	Estimates

lalpha rho beta 0	beta 1

lalpha 1 -0.95 -0.017	0.022

rho -0.95 1 0.00019	-0.0048

beta_0 -0.017 0.00019 1	-1

beta_l 0.022 -0.0048 -1	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
beta_0
beta 1

Estimate
-0.723216
2 .26615
150.535
-0.143931

Std. Err.

0.638291
0.140196
31.5457
0.0308317

Lower Conf. Limit

-1.97424
1.99137
88.7064
-0.20436

Upper Conf. Limit
0.527811
2 . 54093
212.363
-0.0835018

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0
14
26
47
320
1024

472

94
107
74
22
3

151

149
147
144
104
3.15

272

67 . 9
76.4
39.6
22 . 6
2 . 83

204
201
199
194
135
2 . 56

4 .45

-0.766
-0.567
-1. 02
-1.73
-0.166

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

This document is a draft for review purposes only and does not constitute Agency policy.

E-504 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

Model R:	Yi = Mu + e(i

Var{e(i)) = Sigma^2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1 -250.554817	7	515.109634

A2 -196.755746	12	417.511491

A3 -197.383174	8	410.766347

fitted -240.223107	4	488.446215

R -276.789644	2	557.579287

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?
(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	160.068	10	<.0001

Test 2	107.598	5	<.0001

Test 3	1.25486	4	0.869

Test 4	85.6799	4	<.0001

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	1419.81

BMDL =	988.945

This document is a draft for review purposes only and does not constitute Agency policy.

E-505 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E.3.48.3. Figure for Selected Model: Linear

Linear Model with 0.95 Confidence Level

dose

20:03 02/16 2010

E.3.48.4. Output for Additional Model Presented: Power, Unrestricted
Van Birgelen et al., 1995a: Hepatic Retinol Palmitate

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\1\66_VanB_l995a_HepRetPalm_Pwr_U_l.(d)

Gnuplot Plotting File: C:\l\66_VanB_1995a_HepRetPalm_Pwr_U_l.plt

Tue Feb 16 20:03:50 2010

Tbl3, hepatic retinol palmitate

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-506 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho
control
slope
power

9.57332

0

472
-315.054
0.0586881

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope
power

lalpha
1

-0. 95
0.29
-0.31
-0.3

rho

-0. 95
1

-0.4

0.39
0.29

control
0.29
-0.4

1

-0. 98
-0. 82

slope

-0.31
0.39
-0. 98
1

0. 91

power
-0.3

0.29
-0. 82
0. 91
1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate

0.0734958
1.80632
465.497
-318.06
0. 0540573

Std. Err.

0. 849559
0.194602
86.914
82.4127
0.0117709

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-1.59161
1.42491
295.149
-479.586
0.0309869

1.7386
2 .18774
635.845
-156.534
0. 0771278

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0
14
26
47
320
1024

472

94
107
74
22
3

465

98 . 7
86.2
73.8
31.1
2 .86

272

67 . 9
76.4
39.6
22 . 6
2 . 83

266
65. 6
58 .1
50.5
23.1
2 . 68

0. 069
-0.201
1. 01
0.0086
-1.11
0.145

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Model A2:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

Mu(i) + e(ij ;

Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that

This document is a draft for review purposes only and does not constitute Agency policy.

E-507 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1 -250.554817	7	515.109634

A2 -196.755746	12	417.511491

A3 -197.383174	8	410.766347

fitted -199.031154	5	408.062307

R -276.789644	2	557.579287

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	160.068	10	<.0001

Test 2	107.598	5	<.0001

Test 3	1.25486	4	0.869

Test 4	3.29596	3	0.3482

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 0.0376489

BMDL = 1.207 69e-005

This document is a draft for review purposes only and does not constitute Agency policy.

E-508 DRAFT—DO NOT CITE OR QUOTE

-------
E.3.48.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

20:03 02/16 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-509 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

E.3.49. White et al., 1986: CH50

E.3.49.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

5

0.001

391.472

4.480E+02

2.844E+02



exponential (M3)

5

0.001

391.472

4.480E+02

2.844E+02

power hit bound (d = 1)

exponential (M4)

4

0.001

392.128

3.126E+02

1.140E+02



exponential (M5)

4

0.001

392.128

3.126E+02

1.140E+02

power hit bound (d = 1)

Hillb

4

0.001

391.223

2.042E+02

3.585E+01

n lower bound hit (n = 1)

linear

5

<0.0001

396.430

8.065E+02

5.899E+02



polynomial, 6-
degree

3

<0.0001

643.059

9.600E+02

error



power

5

<0.0001

396.430

8.065E+02

5.899E+02

power bound hit (power =1)

Hill, unrestricted0

3

0.058

381.943

9.677E-01

1.900E-01

unrestricted (n = 0.211)

power,
unrestricted

4

0.131

379.574

7.186E-01

1.157E-02

unrestricted (power = 0.188)

a Non-constant variance model selected (p = 0.0871)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

E.3.49.2. Output for Selected Model: Hill

White etal., 1986: CH50

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\7l_White_l986_CH50_Hill_l.(d)

Gnuplot Plotting File: C:\l\71_White_1986_CH50_Hill_l.plt

_Tue Feb 16 20:06:45 2010

[insert study notes]

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose

Power parameter restricted to be greater than 1

This document is a draft for review purposes only and does not constitute Agency policy.

E-510 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))
Total number of dose groups = 7

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho

intercept

5.60999

0
91
-74

0.0969998
10

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
rho

intercept

lalpha
1

-0. 99
0.19
0.13
-0.22

rho

-0. 99
1

-0.2

-0.14
0.23

intercept
0.19
-0.2

1

0.33
-0.7

v

0.13
-0.14
0.33
1

-0.86

k

-0.22
0.23
-0.7

-0.86
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

k

Estimate

4 . 34761
0.381496
71.6585
-62.7464
1

441.016

Std. Err.
1.59601
0. 413764
5.38454
14 . 9646
NA

460.151

Lower Conf. Limit
1.21948
-0.429467
61.105
-92.0765

-460.864

Upper Conf. Limit
7 .47574
1.19246
82 . 212
-33.4163

1342.9

NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

CO

o

91

71.

. 7

14 .1

19.

. 9

2 .75

10 8

54

70.

. 3

8.49

19.

. 8

-2 . 33

50 8

63

65.

. 3

11. 3

19.

. 5

-0.329

100 8

56

60.

. 1

25.5

19.

. 2

-0.598

500 8

41

38 .

. 3

17

17 .

, 6

0.43

1000 8

32

28 .

. 1

17

16.

, 6

0. 661

2000 8

17

20.

. 2

17

15.

. 6

-0.589

This document is a draft for review purposes only and does not constitute Agency policy.

E-511 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
-181.340979
-175.820265
-181. 238690
-190.611743
-212 . 367055

14

9
5
2

AIC

378 . 681959
379.640529
380.477380
391.223485
428.734109

Explanation of Tests

Test

1

Test

2

Test

3

Test

4

(Note:

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

73.0936
11. 0414
10.8369
18 . 7461

12

<.0001
0.0871
0.05471
0.0008815

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	2 04.214

This document is a draft for review purposes only and does not constitute Agency policy.

E-512 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

E'.MDL =

35.8504

E.3.49.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

20:06 02/16 2010

E.3.49.4. Output for Additional Model Presented: Hill, Unrestricted
White etal., 1986: CH50

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\1\7l_White_l986_CH50_Hill_U_l.(d)

Gnuplot Plotting File: C:\l\71_White_1986_CH50_Hill_U_l.plt

Tue Feb 16 20:06:46 2010

[insert study notes]

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-513 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Independent variable = Dose

Power parameter is not restricted

The variance is to be modeled as Var(i)

exp(lalpha + rho * ln(mean(i)

Total number of dose groups = 7

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

lalpha
rho

intercept

5.60999

0
91
-74

0.0969998
10

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho

intercept

lalpha
1
-1

0.17
0.22
-0.42
-0.022

rho
-1

1

-0.17
-0.22
0.42
0. 019

intercept

0.17
-0.17
1

0.16
-0.58
0. 0069

v

0.22
-0.22
0.16
1

-0.048
-0. 91

n

-0.42
0.42
-0.58
-0.048
1

-0.35

k

-0.022
0. 019
0.0069
-0. 91
-0.35
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

Estimate

6. 62767
-0.266376
89.579
-458.615
0.210614
9. 00638e + 006

Std. Err.
2 .14235
0.555274
5.61106
402.837
0.0503369
. 61231e + 007

Lower Conf. Limit
2 .42875
-1.35469
78 . 5815
-1248.16
0.111956
-8 .13933e + 007

Upper Conf. Limit

10.8266
0. 821941
100.576
330.93
0.309273
9.94 061e + 007

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0
10
50
100
500
1000
2000

91

54
63
56
41
32
17

89.6

65. 4
56.3
51. 5
37
30
22

9

9

14 .1
8.49
11. 3
25.5
17
17
17

15.1

15.	8
16.1
16.3

16.	9

17	. 4

18	.1

0.266
-2 . 04
1.18
0 .777
0.516
0.191
-0.927

Model Descriptions for likelihoods calculated

This document is a draft for review purposes only and does not constitute Agency policy.

E-514 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2 :	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(1)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
-181.340979
-175.820265
-181. 238690
-184.971691
-212 . 367055

Param's

14
9

AIC

378 . 681959
379.640529
380.477380
. 943382
.734109

381.
428 .

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

73.0936
11. 0414
10.8369
7.466

12

<.0001
0.0871
0.05471
0.05844

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.967 68 9

BMDL =	0.18 9992

This document is a draft for review purposes only and does not constitute Agency policy.

E-515 DRAFT—DO NOT CITE OR QUOTE

-------
l E.3.49.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

2	20:06 02/16 2010

3

4

5

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

E-516 DRAFT—DO NOT CITE OR QUOTE

-------
1	E.4. REFERENCES

2	Amin, S; Moore, RW; Peterson, RE; et al. (2000) Gestational and lactational exposure to TCDD or coplanar PCBs

3	alters adult expression of saccharin preference behavior in female rats. Neurotoxicol Teratol 22(5):675-682.

4	Bell, DR; Clode, S; Fan, MQ; et al. (2007a) Toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin in the developing male

5	Wistar(Han) rat. II: Chronic dosing causes developmental delay. Toxicol Sci 99(l):224-233.

6	Bell, DR; Clode, S; Fan, MQ; et al. (2007b) Relationships between tissue levels of 2,3,7,8-tetrachlorodibenzo-

7	p-dioxin (TCDD), mRNAs, and toxicity in the developing male Wistar (Han) rat. Toxicol Sci 99(2):591-604.

8	Cantoni, L; Salmona, M; Rizzardini, M. (1981) Porphyrogenic effect of chronic treatment with

9	2,3,7,8-tetrachlorodibenzo-p-dioxin in female rates. Dose-effect relationship following urinary excretion of

10	porphyrins. Toxicol Appl Pharmacol 57:156-157.

11	Crofton, KM; Craft, ES; Hedge, JM; et al. (2005) Thyroid-hormone-disrupting chemicals: evidence for dose-

12	dependent additivity or synergism. Environ Health Perspect 113(11): 1549-1554.

13	DeCaprio, AP; McMartin, DN; O'Keefe, PE; et al. (1986) Subchronic oral toxicity of 2,3,7,8-tetrachlorodibenzo-

14	p-dioxin in the guinea pig: comparisons with a PCB-containing transformer fluid pyrolysate. Fund Appl Toxicol

15	6:454-463.

16	Franc, MA; Pohjanvirta, R; Tuomisto, J; et al. (2001) Persistent, low-dose 2,3,7,8-tetrachlorodibenzo-p-dioxin

17	exposure: effect on aryl hydrocarbon receptor expression in a dioxin-resistance model. Toxicol Appl Pharmacol

18	175:43-53.

19	Hojo, R; Stern, S; Zareba, G; et al. (2002) Sexually dimorphic behavioral responses to prenatal dioxin exposure.

20	Environ Health Perspect 110(3):247-254.

21	Kattainen, H; Tuukanan, J; Simanainen, U; et al. (2001) In utero/lactational 2,3,7,8-tetrachlorodibenzo-p-dioxin

22	exposure impairs molar tooth development in rats. Toxicol Appl Pharmacol 17:216-224.

23	Keller, JM; Huet-Hudson, YM; Leamy, LJ. (2007) Qualitative effects of dioxin on molars vary among inbred mouse

24	strains. Arch Oral Biol 52:450-454.

25	Keller, JM; Zelditch, ML; Huet, YM; et al. (2008a) Genetic differences in sensitivity to alterations of mandible

26	structure caused by the teratogen 2.3.7.8-tctrachlorodibenzo-/?-dio.\in. Toxicol Pathol 36:1006-1013.

27	Keller, JM; Huet-Hudson, Y; Leamy, LJ. (2008b) Effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin on molar

28	development among non-resistant inbred strains of mice: a geometric morphometric analysis. Growth Devel Aging

29	71:3-16.

30	Kociba, RJ; Keyes, DG; Beyer, JE; et al. (1978) Results of a two-year chronic toxicity and oncogenicity study of

31	2,3,7,8-tetrachlorodibenzo-p-dioxin in rats. Toxicol Appl Pharmacol 46(2):279-303.

32	Latchoumycandane, C; Mathur, PP. (2002) Effects of vitamin E on reactive oxygen species-mediated

33	2,3,7,8-tetrachlorodi-benzo-p-dioxin toxicity in rat testis. J Appl Toxicol 22(5):345-351.

34	Li, B; Liu, H-Y; Dai, L-J; et al. (2006) The early embryo loss caused by 2,3,7,8-tetrachlorodibenzo-p-dioxin may be

35	related to the accumulation of this compound in the uterus. Reprod Toxicol 21:301-306.

36	Markowski, VP; Zareba, G; Stern, S; et al. (2001) Altered operant responding for motor reinforcement and the

37	determination of benchmark doses following perinatal exposure to low-level 2,3,7,8-tetrachlorodibenzo-p-dioxin.

38	Environ Health Perspect 109(6):621-627.

This document is a draft for review purposes only and does not constitute Agency policy.

E-517 DRAFT—DO NOT CITE OR QUOTE

-------
1	Miettinen, HM; Sorvari, R; Alaluusua, S; et al. (2006) The Effect of perinatal TCDD exposure on caries

2	susceptibility in rats. Toxicol Sci 91(2):568-575.

3	NTP (National Toxicology Program). (1982) NTP Technical Report on carcinogenesis bioassay of

4	2,3,7,8-tetrachlorodibenzo-p-dioxin in Osborne-Mendel rats and B6C3F1 mice (gavage study). Public Health

5	Service, U.S. Department of Health and Human Services; NTP TR 209. Available from the National Institute of

6	Environmental Health Sciences, Research Triangle Park, NC.

7	NTP (National Toxicology Program). (2006) NTP technical report on the toxicology and carcinogenesis studies of

8	2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) (CAS No. 1746-01-6) in female Harlan Sprague-Dawley rats (Gavage

9	Studies). Natl Toxicol ProgramTech Rep 521. Public Health Service, National Institute of Health, U.S. Department

10	of Health and Human Services, Research Triangle Park, NC.

11	Ohsako, S; Miyabara, Y; Nishimura, N; et al. (2001) Maternal exposure to a low dose of

12	2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) suppressed the development of reproductive organs of male rats: dose-

13	dependent increase of mRNA levels of 5a-reductase type 2 in contrast to decrease of androgren receptor in the

14	pubertal ventral prostate. Toxicol Sci 60:132-143.

15	Shi, Z; Valdez, KE; Ying, AY; et al. (2007) Ovarian endocrine disruption underlies premature reproduction

16	senescence following environmentally relevant chronic exposure to aryl hydrocarbon receptor agonist

17	2,3,7,8-tetrachlorodibenzo-p-dioxin. Biol Reprod 30(4):293-342.

18	Smialowicz, RJ; DeVito, MJ; Williams, WC; et al. (2008) Relative potency based on hepatic enzyme induction

19	predicts immunosuppressive effects of a mixture of PCDDS/PCDFS and PCBS. Toxicol Appl Pharmacol

20	227:477-484.

21	Toth, KJ; Sugar, S; Somfai-Relle, S; et al. (1978) Carcinogenic bioassay of the herbicide 2,4,5-trichlorphenoxy

22	ethanol (TCPE) with Swiss mice. Prog BiochemPharmacol 14:82-93.

23	Toth, L; Somfai-Relle, S; Sugar, J; et al. (1979) Carcinogenicity testing of herbicide 2,4,5-trichlorophenoxyethanol

24	containing dioxin and of pure dioxin in Swiss mice. Nature 278:548-549.

25	Van Birgelen, AP; Van der Kolk, J; Fase, KM; et al. (1995) Subchronic dose-response study of

26	2,3,7,8-tetrachlorodibenzo-p-dioxin in female Sprague-Dawley rats. Toxicol Appl Pharmacol 132:1-13.

27	White, KL, Jr; Lysy, HH; McCay, JA; et al. (1986) Modulation of serum complement levels following exposure to

28	polychlorinated dibenzo-p-dioxins. Toxicol Appl Pharmacol 84:209-219.

This document is a draft for review purposes only and does not constitute Agency policy.

E-518 DRAFT—DO NOT CITE OR QUOTE

-------
[This page intentionally left blank.]

-------
DRAFT

DO NOT CITE OR QUOTE

May 2010
External Review Draft

APPENDIX F
Cancer Benchmark Dose Modeling

NOTICE

THIS DOCUMENT IS AN EXTERNAL REVIEW DRAFT. It has not been formally released
by the U.S. Environmental Protection Agency and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its technical accuracy and policy
implications.

National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH

-------
CONTENTS—APPENDIX F: Cancer Benchmark Dose Modeling

APPENDIX F. CANCER BENCHMARK DOSE MODELING	F-l

F.l. BLOOD BMDS RESULTS	F-l

F.l.l. Kociba et al., 1978: Stratified squamous cell carcinoma of hard

palate or nasal turbinates	F-l

F.l. 1.1. Summary Table of BMDS Modeling Results	F-l

F.l. 1.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-l

F.l.1.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-3

F.1.2. Kociba et al., 1978: Stratified squamous cell carcinoma of tongue	F-4

F.l.2.1. Summary Table of BMDS Modeling Results	F-4

F. 1.2.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-4

F. 1.2.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-6

F.1.3. Kociba et al., 1978: Adenoma of adrenal cortex	F-7

F.1.3.1. Summary Table of BMDS Modeling Results	F-7

F. 1.3.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-7

F.1.3.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-9

F.1.4. Kociba et al., 1978: Hepatocellular adenoma(s) or carcinoma(s)	F-10

F.l.4.1. Summary Table of BMDS Modeling Results	F-10

F. 1.4.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-10

F. 1.4.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-12

F.1.5. Kociba et al., 1978: Stratified squamous cell carcinoma of hard

palate or nasal turbinates	F-l3

F.l.5.1. Summary Table of BMDS Modeling Results	F-13

F.1.5.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-13

F.1.5.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-15

F.1.6. Kociba et al., 1978: Keratinizing squamous cell carcinoma of lung	F-16

F.l.6.1. Summary Table of BMDS Modeling Results	F-16

F. 1.6.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-16

F. 1.6.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-18

F.1.7. National Toxicology Program, 1982: Subcutaneous Tissue:

Fibrosarcoma	F-19

F.l.7.1. Summary Table of BMDS Modeling Results	F-19

F. 1.7.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-19

F. 1.7.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-21

F.1.8. National Toxicology Program, 1982: Liver: Neoplastic Nodule or

Hepatocellular Carcinoma	F-22

F.l.8.1. Summary Table of BMDS Modeling Results	F-22

F.1.8.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-22

F.1.8.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-24

F.1.9. National Toxicology Program, 1982: Adrenal: Cortical Adenoma, or

Carcinoma or Adenoma, NOS	F-25

F.l.9.1. Summary Table of BMDS Modeling Results	F-25

F. 1.9.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-25

F. 1.9.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-27

This document is a draft for review purposes only and does not constitute Agency policy.

F-ii	DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

F.1.10. National Toxicology Program, 1982: Thyroid: Follicular-Cell

Adenoma	F-28

F. 1.10.1. Summary Table ofBMDS Modeling Results	F-28

F.1.10.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-28

F.1.10.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-30

F. 1.11. National Toxicology Program, 1982: Liver: Neoplastic Nodule or

Hepatocellular Carcinoma	F-31

F. 1.11.1. Summary Table ofBMDS Modeling Results	F-31

F. 1.11.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-31

F.1.11.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-33

F.1.12. National Toxicology Program, 1982: Thyroid: Follicular-Cell

Adenoma or Carcinoma	F-34

F. 1.12.1. Summary Table ofBMDS Modeling Results	F-34

F.1.12.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-34

F.1.12.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-36

F.1.13. National Toxicology Program, 1982: Adrenal cortex: Adenoma	F-37

F.1.13.1. Summary Table ofBMDS Modeling Results	F-37

F.1.13.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-37

F.1.13.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-39

F.1.14. National Toxicology Program, 1982: Subcutaneous Tissue:

Fibrosarcoma	F-40

F. 1.14.1. Summary Table ofBMDS Modeling Results	F-40

F.1.14.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-40

F.1.14.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-42

F.1.15. National Toxicology Program, 1982: Hematopoietio System:

Lymphoma or Leukemia	F-43

F.1.15.1. Summary Table ofBMDS Modeling Results	F-43

F.1.15.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-43

F.1.15.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-45

F.1.16. National Toxicology Program, 1982: Liver: Hepatooellular

Adenoma or Carcinoma	F-46

F. 1.16.1. Summary Table ofBMDS Modeling Results	F-46

F.1.16.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-46

F.1.16.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-48

F.1.17. National Toxicology Program, 1982: Subcutaneous Tissue:

Fibrosarcoma	F-49

F. 1.17.1. Summary Table ofBMDS Modeling Results	F-49

F.1.17.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-49

F.1.17.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-51

F.1.18. National Toxicology Program, 1982: Lung: Alveolar/Bronchiolar

Adenoma or Carcinoma	F-52

F.1.18.1. Summary Table ofBMDS Modeling Results	F-52

F.1.18.2. Output for Selected Model: Multistage Cancer, 2-Degree	F-52

This document is a draft for review purposes only and does not constitute Agency policy.

F-iii	DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

F. 1.18.3. Figure for Selected Model: Multistage Cancer, 2-Degree	F-54

F.1.19. National Toxicology Program, 1982: Liver: Hepatocellular Adenoma

or Carcinoma	F-55

F. 1.19.1. Summary Table ofBMDS Modeling Results	F-55

F.1.19.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-55

F.1.19.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-57

F.1.20. National Toxicology Program, 2006: Liver: Cholangiocarcinoma	F-58

F. 1.20.1. Summary Table ofBMDS Modeling Results	F-58

F.1.20.2. Output for Selected Model: Multistage Cancer, 3-Degree	F-58

F. 1.20.3. Figure for Selected Model: Multistage Cancer, 3-Degree	F-60

F. 1.21. National Toxicology Program, 2006: Liver: Hepatocellular adenoma	F-61

F. 1.21.1. Summary Table ofBMDS Modeling Results	F-61

F. 1.21.2. Output For Selected Model: Multistage Cancer, 3-Degree	F-61

F.1.21.3. Figure For Selected Model: Multistage Cancer, 3-Degree	F-63

F.1.22. National Toxicology Program, 2006: Oral mucosa: squamous cell

carcinoma	F-64

F. 1.22.1. Summary Table ofBMDS Modeling Results	F-64

F. 1.22.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-64

F. 1.22.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-66

F.1.23. National Toxicology Program, 2006: Pancreas: adenoma or

carcinoma	F-67

F.1.23.1. Summary Table ofBMDS Modeling Results	F-67

F. 1.23.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-67

F.1.23.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-69

F.1.24. National Toxicology Program, 2006: Lung: Cystic keratinizing

epithelioma	F-70

F. 1.24.1. Summary Table ofBMDS Modeling Results	F-70

F. 1.24.2. Output for Selected Model: Multistage Cancer, 2-Degree	F-70

F. 1.24.3. Figure for Selected Model: Multistage Cancer, 2-Degree	F-72

F.1.25. Tothetal., 1979: Liver: Tumors	F-73

F. 1.25.1. Summary Table ofBMDS Modeling Results	F-73

F. 1.25.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-73

F.1.25.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-75

F.1.26. DellaPorta et al., 1987: Table 4, B6C3 mice, male, hepatocellular

carcinoma	F-76

F. 1.26.1. Summary Table ofBMDS Modeling Results	F-76

F. 1.26.2. Output for Selected Model: Multistage Cancer, 2-Degree	F-76

F. 1.26.3. Figure for Selected Model: Multistage Cancer, 2-Degree	F-78

F.1.27. DellaPorta et al., 1987: Table 4, B6C3 mice, female, hepatocellular

adenoma	F-79

F. 1.27.1. Summary Table ofBMDS Modeling Results	F-79

F. 1.27.2. Output for Selected Model: Multistage Cancer, 2-Degree	F-79

F. 1.27.3. Figure for Selected Model: Multistage Cancer, 2-Degree	F-81

This document is a draft for review purposes only and does not constitute Agency policy.

F-iv	DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

F.1.28. DellaPorta et al., 1987: Table 4, B6C3 mice, female, hepatocellular

carcinoma	F-82

F. 1.28.1. Summary Table ofBMDS Modeling Results	F-82

F.1.28.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-82

F.1.28.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-84

F.2. ADMINISTERED DOSE BMDS RESULTS	1-85

F.2.1. Kociba et al., 1978: Stratified squamous cell carcinoma of hard

palate or nasal turbinates	F-85

F.2.1.1. Summary Table ofBMDS Modeling Results	F-85

F.2.1.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-85

F.2.1.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-87

F.2.2. Kociba et al., 1978: Stratified squamous cell carcinoma of tongue	F-88

F.2.2.1. Summary Table ofBMDS Modeling Results	F-88

F.2.2.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-88

F.2.2.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-90

F.2.3. Kociba et al., 1978: Adenoma of adrenal cortex	F-91

F.2.3.1. Summary Table ofBMDS Modeling Results	F-91

F.2.3.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-91

F.2.3.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-93

F.2.4. Kociba et al., 1978: Hepatocellular adenoma(s) or carcinoma(s)	F-94

F.2.4.1. Summary Table ofBMDS Modeling Results	F-94

F.2.4.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-94

F.2.4.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-96

F.2.5. Kociba et al., 1978: Stratified squamous cell carcinoma of hard

palate or nasal turbinates	F-97

F.2.5.1. Summary Table ofBMDS Modeling Results	F-97

F.2.5.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-97

F.2.5.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-99

F.2.6. Kociba et al., 1978: Keratinizing squamous cell carcinoma of lung	F-100

F.2.6.1. Summary Table ofBMDS Modeling Results	F-100

F.2.6.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-100

F.2.6.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-102

F.2.7. National Toxicology Program, 1982: Subcutaneous Tissue:

Fibrosarcoma	F-103

F.2.7.1. Summary Table ofBMDS Modeling Results	F-103

F.2.7.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-103

F.2.7.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-105

F.2.8. National Toxicology Program, 1982: Liver: Neoplastic Nodule or

Hepatocellular Carcinoma	F-106

F.2.8.1. Summary Table ofBMDS Modeling Results	F-106

F.2.8.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-106

F.2.8.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-108

This document is a draft for review purposes only and does not constitute Agency policy.

F-v	DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

F.2.9. National Toxicology Program, 1982: Adrenal: Cortical Adenoma, or

Carcinoma or Adenoma, NOS	F-109

F.2.9.1. Summary Table ofBMDS Modeling Results	F-109

F.2.9.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-109

F.2.9.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-l 11

F.2.10. National Toxicology Program, 1982: Thyroid: Follicular-Cell

Adenoma	F-l 12

F.2.10.1. Summary Table ofBMDS Modeling Results	F-l 12

F.2.10.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-l 12

F.2.10.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-l 14

F.2.11. National Toxicology Program, 1982: Liver: Neoplastic Nodule or

Hepatocellular Carcinoma	F-l 15

F.2.11.1. Summary Table ofBMDS Modeling Results	F-l 15

F.2.11.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-l 15

F.2.11.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-l 17

F.2.12. National Toxicology Program, 1982: Thyroid: Follicular-Cell

Adenoma or Carcinoma	F-l 18

F.2.12.1. Summary Table ofBMDS Modeling Results	F-118

F.2.12.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-118

F.2.12.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-120

F.2.13. National Toxicology Program, 1982: Adrenal cortex: Adenoma	F-121

F.2.13.1. Summary Table ofBMDS Modeling Results	F-121

F.2.13.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-121

F.2.13.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-123

F.2.14. National Toxicology Program, 1982: Subcutaneous Tissue:

Fibrosarcoma	F-124

F.2.14.1. Summary Table ofBMDS Modeling Results	F-124

F.2.14.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-124

F.2.14.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-126

F.2.15. National Toxicology Program, 1982: Hematopoietio System:

Lymphoma or Leukemia	F-127

F.2.15.1. Summary Table ofBMDS Modeling Results	F-127

F.2.15.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-127

F.2.15.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-129

F.2.16. National Toxicology Program, 1982: Liver: Hepatooellular

Adenoma or Carcinoma	F-130

F.2.16.1. Summary Table ofBMDS Modeling Results	F-130

F.2.16.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-130

F.2.16.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-132

F.2.17. National Toxicology Program, 1982: Subcutaneous Tissue:

Fibrosarcoma	F-133

F.2.17.1. Summary Table ofBMDS Modeling Results	F-133

F.2.17.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-133

This document is a draft for review purposes only and does not constitute Agency policy.

F-vi	DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

F.2.17.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-135

F.2.18. National Toxicology Program, 1982: Lung: Alveolar/Bronchiolar

Adenoma or Carcinoma	F-136

F.2.18.1. Summary Table ofBMDS Modeling Results	F-136

F.2.18.2. Output for Selected Model: Multistage Cancer, 2-Degree	F-136

F.2.18.3. Figure for Selected Model: Multistage Cancer, 2-Degree	F-138

F.2.19. National Toxicology Program, 1982: Liver: Hepatocellular Adenoma

or Carcinoma	F-139

F.2.19.1. Summary Table ofBMDS Modeling Results	F-139

F.2.19.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-139

F.2.19.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-141

F.2.20. National Toxicology Program, 2006: Liver: Cholangiocarcinoma	F-142

F.2.20.1. Summary Table ofBMDS Modeling Results	F-142

F.2.20.2. Output for Selected Model: Multistage Cancer, 3-Degree	F-142

F.2.20.3. Figure for Selected Model: Multistage Cancer, 3-Degree	F-144

F.2.21. National Toxicology Program, 2006: Liver: Hepatocellular adenoma	F-145

F.2.21.1. Summary Table ofBMDS Modeling Results	F-145

F.2.21.2. Output for Selected Model: Multistage Cancer, 3-Degree	F-145

F.2.21.3. Figure for Selected Model: Multistage Cancer, 3-Degree	F-147

F.2.22. National Toxicology Program, 2006: Oral mucosa: squamous cell

carcinoma	F-148

F.2.22.1. Summary Table ofBMDS Modeling Results	F-148

F.2.22.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-148

F.2.22.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-150

F.2.23. National Toxicology Program, 2006: Pancreas: adenoma or

carcinoma	F-151

F.2.23.1. Summary Table ofBMDS Modeling Results	F-151

F.2.23.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-151

F.2.23.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-153

F.2.24. National Toxicology Program, 2006: Lung: Cystic keratinizing

epithelioma	F-154

F.2.24.1. Summary Table ofBMDS Modeling Results	F-154

F.2.24.2. Output for Selected Model: Multistage Cancer, 2-Degree	F-154

F.2.24.3. Figure for Selected Model: Multistage Cancer, 2-Degree	F-156

F.2.25. Tothetal., 1979: Liver: Tumors	F-157

F.2.25.1. Summary Table ofBMDS Modeling Results	F-157

F.2.25.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-157

F.2.25.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-159

F.2.26. Delia Porta et al., 1987: Table 4, B6C3 mice, male, hepatocellular

carcinoma	F-160

F.2.26.1. Summary Table ofBMDS Modeling Results	F-160

F.2.26.2. Output for Selected Model: Multistage Cancer, 2-Degree	F-160

F.2.26.3. Figure for Selected Model: Multistage Cancer, 2-Degree	F-162

This document is a draft for review purposes only and does not constitute Agency policy.

F-vii	DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

F.2.27. Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular

adenoma	F-163

F.2.27.1. Summary Table ofBMDS Modeling Results	F-163

F.2.27.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-163

F.2.27.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-165

F.2.28. Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular

carcinoma	F-166

F.2.28.1. Summary Table ofBMDS Modeling Results	F-166

F.2.28.2. Output for Selected Model: Multistage Cancer, 1-Degree	F-166

F.2.28.3. Figure for Selected Model: Multistage Cancer, 1-Degree	F-168

F.3. REFERENCES	F-169

This document is a draft for review purposes only and does not constitute Agency policy.

F-viii DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

APPENDIX F.

CANCER BENCHMARK DOSE MODELING

F.l. BLOOD BMDS RESULTS

F.l.l. Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal
turbinates

F. 1.1.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

3

0.815

31.564

5.763E+00

2.795E+00



Multistage Cancer,
2-Degree

3

0.985

30.170

1.369E+01

3.416E+00



Multistage Cancer,
3-Degree

3

0.999

29.930

1.917E+01

3.578E+00



a Best-fitting model, BMDS output presented in this appendix

F. 1.1.2. Output for Selected Model: Multistage Cancer, 1 -Degree

Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal turbinates

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\l mscl IPerc palate nasal.(d)

Gnuplot Plotting File: C:\4\Blood\l mscl IPerc palate nasal.pit

Thu Apr 01 15:56:03 2010

Source - Table 4

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

F-l	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Default Initial Parameter Values
Background =	0

Beta(1) = 0.00226154

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.0017438	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-13.9385	4

Fitted model	-14.7819	1	1.68696	3	0.6398

Reduced model	-20.2589	1	12.6409	3	0.005481

AIC:	31.5639

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0000 0.000 0.000	85	0.000

1.5617 0.0027 0.136 0.000	50	-0.369

7.1600 0.0124 0.620 0.000	50	-0.793

38.7212 0.0653 3.265 4.000	50	0.421

Chi/N2 = 0.94	d.f. = 3	P-value = 0.8153

Benchmark Dose Computation

Specified effect	=	0.01

Risk Type	=	Extra risk

Confidence level	=	0.95

BMD	=	5.7 6347

BMDL	=	2.7 94 85

BMDU	=	14 . 9396

Taken together, (2.79485, 14.9396) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.003578

This document is a draft for review purposes only and does not constitute Agency policy.

F-2	DRAFT—DO NOT CITE OR QUOTE

-------
1 F.l.1.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	14:56 04/01 2010

3

4	Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal turbinates

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-3	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.1.2. Kociba et al., 1978: Stratified squamous cell carcinoma of tongue
F. 1.2.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

Up-
value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degree a

2

0.472

47.933

6.091E+00

2.600E+00



Multistage Cancer,
2-Degree

2

0.472

47.933

6.091E+00

2.600E+00

final B=0

Multistage Cancer,
3-Degree

2

0.472

47.933

6.091E+00

2.600E+00

final B=0

a Best-fitting model, BMDS output presented in this appendix

F. 1.2.2. Output for Selected Model: Multistage Cancer, 1-Degree
Kociba et al., 1978: Stratified squamous cell carcinoma of tongue

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\2_mscl_lPerc_tongue.(d)

Gnuplot Plotting File: C:\4\Blood\2_mscl_lPerc_tongue.plt

Thu Apr 01 15:56:35 2010

Source - Table 4

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0092514
Beta (1) = 0.00137224

This document is a draft for review purposes only and does not constitute Agency policy.

F-4	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.58

Beta(1)	-0.58	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.00510501	*	*	*

Beta (1)	0.00165011	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-21.1523	4

Fitted model -21.9667 2	1.62881 2	0.4429

Reduced model -24.1972 1	6.08976 3	0.1073

AIC:	47.9334

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0051 0.434 0.000	85	-0.660

1.5617 0.0077 0.383 1.000	50	1.000

7.1600 0.0168 0.840 1.000	50	0.177

38.7212 0.0667 3.334 3.000	50	-0.189

Chi ^2 = 1.50	d.f. = 2	P-value = 0.4716

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	6.0907

BMDL =	2.6004 9

BMDU =	519124

Taken together, (2.60049, 519124 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00384542

This document is a draft for review purposes only and does not constitute Agency policy.

F-5	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.2.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	14:56 04/01 2010

3

4	Kociba et al., 1978: Stratified squamous cell carcinoma of tongue

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-6	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.1.3. Kociba et al., 1978: Adenoma of adrenal cortex
F. 1.3.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

3

0.779

52.488

3.254E+00

1.852E+00



Multistage Cancer,
2-Degree

3

0.779

52.488

3.254E+00

1.852E+00

final fi=0

Multistage Cancer,
3-Degree

3

0.779

52.488

3.254E+00

1.852E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F. 1.3.2. Output for Selected Model: Multistage Cancer, 1-Degree
Kociba et al., 1978: Adenoma of adrenal cortex

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\3 mscl IPerc adre adenoma.(d)

Gnuplot Plotting File: C:\4\Blood\3 mscl IPerc adre adenoma.pit

Thu Apr 01 15:57:07 2010

Source - Table 5

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.00493756
Beta(1) = 0.0026639

This document is a draft for review purposes only and does not constitute Agency policy.

F-7	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.00308883	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-24.6514	4

Fitted model -25.2438 1	1.18487 3	0.756£

Reduced model -31.4904 1	13.6781 3	0.00337e

AIC:	52.4876

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.5617
7.1600
38.7212

0.0000
0.0048
0.0219
0.1127

0. 000
0.241
1.094
5. 636

0. 000
0. 000
2 . 000
5. 000

85

50
50
50

0. 000
-0.492
0. 876
-0.285

Chi ^2

1.09

d.f.

P-value

0.77 93

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	3.25376

BMDL =	1.85162

BMDU =	6.585 95

Taken together, (1.85162, 6.58595) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00540067

This document is a draft for review purposes only and does not constitute Agency policy.

F-8	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.3.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	14:57 04/01 2010

3

4	Kociba et al., 1978: Adenoma of adrenal cortex

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-9	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.4. Kociba et al., 1978: Hepatocellular adenoma(s) or carcinoma(s)
F.l.4.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.245

143.261

7.010E-01

5.013E-01



Multistage Cancer,
2-Degree

2

0.245

143.261

7.010E-01

5.013E-01

final fi=0

Multistage Cancer,
3-Degree

2

0.245

143.261

7.010E-01

5.013E-01

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.l.4.2. Output for Selected Model: Multistage Cancer, 1-Degree
Kociba et al., 1978: Hepatocellular adenoma(s) or carcinoma(s)

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\4 mscl IPerc liver ad carc.(d)

Gnuplot Plotting File: C:\4\Blood\4 mscl IPerc liver ad carc.pit

Thu Apr 01_15T57:41 2010

Source - Table 1 in Goodman and Sauer 1992

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0400263
Beta(1) = 0.0124752

This document is a draft for review purposes only and does not constitute Agency policy.

F-10	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background
Beta(1)

1

-0.51

-0.51
1

Variable
Background
Beta(1)

Parameter Estimates

Estimate

0. 0221468
0.0143372

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)	# Param'

-68.2561	4

-69.6304	2

-89.1983	1

Deviance Test d.f.

P-value

2.74857
41.8843

0.253
0001

143.261

Dose

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000
1.5473
7 .1546
38 . 5608

0.0221
0.0436
0.1175
0.4374

1. 905

2 .180
5. 874
19.685

2 . 000
1. 000
9. 000
18 . 000

86
50
50
45

0.	070
-0.817

1.	373
-0.506

Chi ^2

2 . 81

d.f.

P-value

0.2449

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	0.700996

BMDL =	0.501345

BMDU =	1.04839

Taken together, (0.501345, 1.04839) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0199463

This document is a draft for review purposes only and does not constitute Agency policy.

F-l 1	DRAFT—DO NOT CITE OR QUOTE

-------
1 F.l.4.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	14:57 04/01 2010

3

4	Kociba et al., 1978: Hepatocellular adenoma(s) or carcinoma(s)

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-12	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.5. Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal
turbinates

F.l.5.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

3

0.815

31.564

5.763E+00

2.795E+00



Multistage Cancer,
2-Degree

3

0.985

30.170

1.369E+01

3.416E+00



Multistage Cancer,
3-Degree

3

0.999

29.930

1.917E+01

3.578E+00



a Best-fitting model, BMDS output presented in this appendix

F.l.5.2. Output for Selected Model: Multistage Cancer, 1-Degree

Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal turbinates

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\5_mscl_lPerc_nasal.(d)

Gnuplot Plotting File: C:\4\Blood\5_mscl_lPerc_nasal.plt

Thu Apr 01 15:58:14 2010

Source - Table 5

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 7.10818e-005
Beta (1) = 0.00222324

This document is a draft for review purposes only and does not constitute Agency policy.

F-13	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.0022294	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-18.7562	4

Fitted model -18.9547 1	0.397012 3	0.9409

Reduced model -24.1972 1	10.882 3	0.01238

AIC:	39.9093

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.5473
7 .1546
38 . 5608

0.0000
0.0034
0.0158
0.0824

0. 000
0.172
0.791
4 . 036

0. 000

0.	000

1.	000
4 . 000

50
50
49

0. 000
-0.416
0.237
-0.019

Chi ^2

0.23

d.f.

P-value

0.97 2 E

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	4.50809

BMDL =	2.34012

BMDU =	10.4588

Taken together, (2.34012, 10.4588) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00427329

This document is a draft for review purposes only and does not constitute Agency policy.

F-14	DRAFT—DO NOT CITE OR QUOTE

-------
F.l.5.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

14:58 04/01 2010

Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal turbinates

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-l 5	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.1.6. Kociba et al., 1978: Keratinizing squamous cell carcinoma of lung
F.l.6.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

3

0.626

45.298

3.140E+00

1.786E+00



Multistage Cancer,
2-Degree

3

0.964

42.736

1.004E+01

2.707E+00



Multistage Cancer,
3-Degree

3

0.997

42.291

1.556E+01

3.135E+00



a Best-fitting model, BMDS output presented in this appendix

F.l.6.2. Output for Selected Model: Multistage Cancer, 1-Degree
Kociba et al., 1978: Keratinizing squamous cell carcinoma of lung

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\6 mscl IPerc kera carc.(d)

Gnuplot Plotting File: C:\4\Blood\6 mscl IPerc kera carc.pit

Thu Apr 01 15:58:49 2010

Source - Table 5

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) = 0.00419802

This document is a draft for review purposes only and does not constitute Agency policy.

F-16	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.00320098	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-20.0957	4

Fitted model -21.6489 1	3.10639 3	0.3755

Reduced model -31.4904 1	22.7894 3	<.0001

AIC:	45.2978

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.5473
7 .1546
38 . 5608

0.0000
0.0049
0.0226
0.1161

0. 000
0.247
1.132
90

5

0. 000
0. 000
0. 000
7 . 000

50
50
49

0. 000
-0.498
-1.076
0.584

Chi ^2

1.75

d.f.

P-value

0.6263

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	3.13977

BMDL =	1.78648

BMDU =	6.28288

Taken together, (1.78648, 6.28288) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0055976

This document is a draft for review purposes only and does not constitute Agency policy.

F-17	DRAFT—DO NOT CITE OR QUOTE

-------
1 F.l.6.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	14:58 04/01 2010

3

4	Kociba et al., 1978: Keratinizing squamous cell carcinoma of lung

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-l 8	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.7. National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma
F.l.7.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degreea

2

0.179

75.385

3.127E+00

1.380E+00



Multistage Cancer,
2-Degree

2

0.179

75.385

3.127E+00

1.380E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.179

75.385

3.127E+00

1.380E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.l.7.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\7_mscl_lPerc_sub_fibro.(d)

Gnuplot Plotting File: C:\4\Blood\7_mscl_lPerc_sub_fibro.plt

Thu Apr 01 15:59:25 2010

Source - Table 10

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0268183
Beta (1) = 0.00211524

This document is a draft for review purposes only and does not constitute Agency policy.

F-19	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.63

Beta(1)	-0.63	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0149841	*	*	*

Beta(1)	0.00321423	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-33.5998	4

Fitted model -35.6923 2	4.18508 2	0.1234

Reduced model -37.7465 1	8.29346 3	0.04032

AIC:	75.3847

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.9574
5.6942
29.7519

0.0150

0.0212
0.0328
0.1048

1.124

058
642

5.136

0. 000
2 . 000
3. 000
4 . 000

75

50
50
49

-1.068

0.	926

1.	077
-0.530

Chi ^2

3.44

d.f.

P-value

0.1792

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	3.12 683

BMDL =	1.38 04 7

BMDU = 2 .18232e + 006

Taken together, (1.38047, 2.18232e+006) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00724391

This document is a draft for review purposes only and does not constitute Agency policy.

F-20	DRAFT—DO NOT CITE OR QUOTE

-------
1 F.l.7.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	14:59 04/01 2010

3

4	National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-21	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.8. National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular
Carcinoma

F.l.8.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.218

135.190

1.169E+00

7.375E-01



Multistage Cancer,
2-Degree

2

0.491

133.447

5.578E+00

8.771E-01



Multistage Cancer,
3-Degree

1

0.239

135.435

7.204E+00

8.786E-01



a Best-fitting model, BMDS output presented in this appendix

F.l.8.2. Output for Selected Model: Multistage Cancer, 1-Degree

National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\8_mscl_lPerc_liver_nod.(d)

Gnuplot Plotting File: C:\4\Blood\8 mscl IPerc liver nod.pit

Thu Apr 01_16:00:00 2010

Source - Table 10

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0261097
Beta (1) = 0.0102165

This document is a draft for review purposes only and does not constitute Agency policy.

F-22	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.52

Beta(1)	-0.52	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0424738	*	*	*

Beta (1)	0.00859382	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-63.9149	4

Fitted model	-65.5949	2	3.36005	2	0.1864

Reduced model	-74.0195	1	20.2092	3	0.0001536

AIC:	135.19

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.9574
5.6942
29.7519

0.0425
0.0584
0.0882
0.2585

3.1

2 . 8

4 .
12 .

864
410
667

5. 000
1. 000
3. 000
14.000

75

49

50
49

1. 039
-1.135
-0.703
0. 435

Chi ^2

3. 05

d.f.

P-value

0.2175

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	1.1694 8

BMDL =	0.737535

BMDU =	2.17 906

Taken together, (0.737535, 2.17906) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0135587

This document is a draft for review purposes only and does not constitute Agency policy.

F-23	DRAFT—DO NOT CITE OR QUOTE

-------
F.l.8.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

15:00 04/01 2010

National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-24	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

F.1.9. National Toxicology Program, 1982: Adrenal: Cortical Adenoma, or Carcinoma or
Adenoma, NOS

F.l.9.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.337

203.824

1.611E+00

8.140E-01



Multistage Cancer,
2-Degree

2

0.470

203.033

6.652E+00

8.904E-01



Multistage Cancer,
3-Degree

2

0.505

202.868

1.091E+01

9.100E-01



a Best-fitting model, BMDS output presented in this appendix

F.l.9.2. Output for Selected Model: Multistage Cancer, 1-Degree

National Toxicology Program, 1982: Adrenal: Cortical Adenoma, or Carcinoma or Adenoma,
NOS

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\9 mscl IPerc adre cort ad carc.(d)

Gnuplot Plotting File: C:\4\Blood\9 mscl IPerc adre cort ad carc.pit

Thu Apr 01 16:06:15 2010

Source - Table 10

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

This document is a draft for review purposes only and does not constitute Agency policy.

F-25	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Background =
Beta(1) =

0.134165
0.0069662

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.54

Beta(1)	-0.54	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.139854	*	*	*

Beta (1)	0.00623778	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-98.7282	4

Fitted model -99.912 2	2.36764 2	0.3061

Reduced model -102.201 1	6.94636 3	0.07363

AIC:	203.824

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.1399 10.209 11.000	73	0.267

1.9574 0.1503 7.364 9.000	49	0.654

5.6942 0.1699 8.324 5.000	49	-1.264

29.7519 0.2855 13.135 14.000	46	0.282

Chi ^2 = 2.18	d.f. = 2	P-value = 0.3367

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	1.6112

BMDL =	0.814 04

BMDU =	370555

Taken together, (0.81404, 370555 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0122844

This document is a draft for review purposes only and does not constitute Agency policy.

F-26	DRAFT—DO NOT CITE OR QUOTE

-------
1 F.l.9.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	15:06 04/01 2010

3

4	National Toxicology Program, 1982: Adrenal: Cortical Adenoma, or Carcinoma or Adenoma,

5	NOS

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-27	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.10. National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma
F. 1.10.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degreea

2

0.568

92.411

3.376E+00

1.553E+00



Multistage Cancer,
2-Degree

2

0.735

91.749

9.526E+00

1.690E+00



Multistage Cancer,
3-Degree

2

0.773

91.626

1.385E+01

1.720E+00



a Best-fitting model, BMDS output presented in this appendix

F. 1.10.2. Output for Selected Model: Multistage Cancer, 1 -Degree
National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\10_mscl_lPerc_thy_ad.(d)

Gnuplot Plotting File: C:\4\Blood\10_mscl_lPerc_thy_ad.plt

~Thu Apr 01 16:06:53 2010

Source - Table 10

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0283212
Beta (1) = 0.00346762

This document is a draft for review purposes only and does not constitute Agency policy.

F-28	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.54

Beta(1)	-0.54	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0332432	*	*	*

Beta (1)	0.00297726	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-43.5264	4

Fitted model -44.2053 2	1.35778 2	0.5072

Reduced model -46.2299 1	5.40699 3	0.1443

AIC:	92.4106

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0332 2.427 3.000	73	0.374

1.9574 0.0389 1.749 2.000	45	0.194

5.6942 0.0495 2.425 1.000	49	-0.939

29.7519 0.1152 5.414 6.000	47	0.268

Chi ^2 = 1.13	d.f. = 2	P-value = 0.5682

Benchmark Dose Computation

Specified effect	=	0.01

Risk Type	=	Extra risk

Confidence level	=	0.95

BMD	=	3.3757

BMDL	=	1.55287

BMDU	=	306341

Taken together, (1.55287, 306341 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00643967

This document is a draft for review purposes only and does not constitute Agency policy.

F-29	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.10.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	15:06 04/01 2010

3

4	National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-30	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.l.ll. National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular
Carcinoma

F. 1.11.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degreea

2

0.218

135.190

1.169E+00

7.375E-01



Multistage Cancer,
2-Degree

2

0.491

133.447

5.578E+00

8.771E-01



Multistage Cancer,
3-Degree

1

0.239

135.435

7.204E+00

8.786E-01



a Best-fitting model, BMDS output presented in this appendix

F. 1.11.2. Output for Selected Model: Multistage Cancer, 1 -Degree

National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\ll_mscl_lPerc_liver_nod.(d)

Gnuplot Plotting File: C:\4\Blood\ll mscl IPerc liver nod.pit

~Thu Apr 01 16:07:28 2010

Source - Table 9

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) = 0.00219894

This document is a draft for review purposes only and does not constitute Agency policy.

F-31	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.00163808	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-11.3484	4

Fitted model -12.0522 1	1.40767 3	0.7037

Reduced model -15.9189 1	9.14109 3	0.02747

AIC:	26.1044

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.9569
5.7027
29.8723

0.0000
0.0032
0.0093
0.0478

0. 000
0.160
0. 465
2 . 388

0. 000
0. 000
0. 000
3. 000

74

50
50
50

0. 000
-0.401
-0.685
0. 406

Chi ^2

0.79

d.f.

P-value

0. 8507

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	6.13543

BMDL =	2.70101

BMDU =	18 . 9354

Taken together, (2.70101, 18.9354) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00370232

This document is a draft for review purposes only and does not constitute Agency policy.

F-32	DRAFT—DO NOT CITE OR QUOTE

-------
F. 1.11.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

15:07 04/01 2010

National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-33	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.12. National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma or
Carcinoma

F.l.12.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.057

149.263

1.208E+00

6.984E-01



Multistage Cancer,
2-Degree

2

0.057

149.263

1.208E+00

6.984E-01

final fi=0

Multistage Cancer,
3-Degree

2

0.057

149.263

1.208E+00

6.984E-01

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.l.12.2. Output for Selected Model: Multistage Cancer, 1-Degree

National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma or Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\4\Blood\12_mscl_lPerc_thyroid.(d)

Gnuplot Plotting File: C:\4\Blood\12_mscl_lPerc_thyroid.plt

~Thu Apr 01 16:08:03 2010

Source - Table 9

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0768555
Beta (1) = 0.00606248

This document is a draft for review purposes only and does not constitute Agency policy.

F-34	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.62

Beta(1)	-0.62	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0529006	*	*	*

Beta(1)	0.00831706	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-69.5946	4

Fitted model	-72.6315	2	6.07383	2	0.04798

Reduced model	-77.5267	1	15.8643	3	0.001209

AIC:	149.263

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.9569
5.7027
29.8723

0.0529
0.0682
0.0968
0.2613

3. 650

3.273
4 . 839
13.063

1

11

000
5. 000
8 . 000
000

69

48
50
50

-1.425

0.	989

1.	512
-0.664

Chi ^2

5.74

d.f.

P-value

0. 056

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	1.2084

BMDL =	0.698436

BMDU =	2 . 8 910 9

Taken together, (0.698436, 2.89109) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0143177

This document is a draft for review purposes only and does not constitute Agency policy.

F-3 5	DRAFT—DO NOT CITE OR QUOTE

-------
F.l.12.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

15:08 04/01 2010

National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma or Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-36	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.13. National Toxicology Program, 1982: Adrenal cortex: Adenoma
F.l.13.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.062

199.309

3.977E+00

1.223E+00



Multistage Cancer,
2-Degree

2

0.062

199.309

3.977E+00

1.223E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.062

199.309

3.977E+00

1.223E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.l.13.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 1982: Adrenal cortex: Adenoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\13_mscl_lPerc_adre_cort.(d)

Gnuplot Plotting File: C:\l\Blood\13 mscl IPerc adre cort.plt

Fri Apr 02 10:53:16 2010

Source - Table 9

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.163685

Beta (1) = 0.00144687

This document is a draft for review purposes only and does not constitute Agency policy.

F-37	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.6

Beta(1)	-0.6	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.146079	*	*	*

Beta(1)	0.00252696	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-94.8672	4

Fitted model	-97.6546	2	5.57468	2	0.06158

Reduced model	-98.0432	1	6.35197	3	0.09569

AIC:	199.309

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.1461

10.518

6. 000

72

-1.507

1.9569

0.1503

7 . 515

9. 000

50

0.588

5.7027

0.1583

7 .756

12.000

49

1. 661

29.8723

0.2082

10.200

9. 000

49

-0.422

Chi/N2 = 5.55	d.f. = 2	P-value = 0.0622

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	3.97724

BMDL =	1.222 8 6

BMDU did not converge for BMR = 0.010000
BMDU calculation failed
BMDU = Inf

This document is a draft for review purposes only and does not constitute Agency policy.

F-3 8	DRAFT—DO NOT CITE OR QUOTE

-------
1 F.l.13.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	09:53 04/02 2010

3

4	National Toxicology Program, 1982: Adrenal cortex: Adenoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-39	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.14. National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma
F. 1.14.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degreea

2

0.179

75.385

3.127E+00

1.380E+00



Multistage Cancer,
2-Degree

2

0.179

75.385

3.127E+00

1.380E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.179

75.385

3.127E+00

1.380E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F. 1.14.2. Output for Selected Model: Multistage Cancer, 1 -Degree
National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\14_mscl_lPerc_subcu_fibro.(d)

Gnuplot Plotting File: C:\l\Blood\14 mscl IPerc subcu fibro.plt

Fri Apr 02 10:59:38 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.010477

Beta(1) = 0.00314237

This document is a draft for review purposes only and does not constitute Agency policy.

F-40	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.55

Beta(1)	-0.55	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0124357	*	*	*

Beta (1)	0.0029518	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-30.9876	4

Fitted model -31.0692 2	0.163345 2	0.9216

Reduced model -34.3291 1	6.68308 3	0.08272

AIC:	66.1385

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0124 0.920 1.000	74	0.084

1.9460 0.0181 0.905 1.000	50	0.101

5.8440 0.0293 1.408 1.000	48	-0.349

32.0560 0.1016 4.775 5.000	47	0.109

Chi ^2 = 0.15	d.f. = 2	P-value = 0.9274

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	3.4 04 81

BMDL =	1 . 68615

BMDU =	11.3501

Taken together, (1.68615, 11.3501) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00593067

This document is a draft for review purposes only and does not constitute Agency policy.

F-41	DRAFT—DO NOT CITE OR QUOTE

-------
F. 1.14.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

09:59 04/02 2010

National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-42	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.15. National Toxicology Program, 1982: Hematopoietio System: Lymphoma or
Leukemia

F. 1.15.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.977

261.445

1.145E+00

6.091E-01



Multistage Cancer,
2-Degree

1

0.869

263.426

1.704E+00

6.102E-01



Multistage Cancer,
3-Degree

1

0.869

263.426

1.704E+00

6.102E-01

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F. 1.15.2. Output for Selected Model: Multistage Cancer, 1 -Degree

National Toxicology Program, 1982: Hematopoietio System: Lymphoma or Leukemia

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\15 mscl IPerc mice f lymphoma.(d)

Gnuplot Plotting File: C:\l\Blood\15 mscl IPerc mice f lymphoma.pit

Fri Apr 02 11:00:07 2010

Table 15 page 64 Hematopoietic System Lymphoma or Leukemia

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.23423

Beta(1) = 0.00892991

This document is a draft for review purposes only and does not constitute Agency policy.

F-43	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background
Beta(1)

1

-0.54

-0.54
1

Variable
Background
Beta(1)

Parameter Estimates

Estimate
0.236159
0. 00877894

Std. Err.

Indicates that this value is not calculated.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table

Log(likelihood)
-128.699
-128.723
-131.412

Param's

4
2
1

Deviance Test d.f.

P-value

0.0465401
5. 42487

0 . 977
0.1432

AIC:

261.445

Dose

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000
1.9460
5.8440
32.0560

0.2362
0.2491
0.2744
0.4235

17.476
12.455
13.169
19.905

18.000

12.000
13.000
20.000

74

50
48
47

0.143

-0.149
-0.055
0. 028

Chi ^2

0. 05

d.f.

P-value

0. 9770

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	1.14 4 82

BMDL =	0.609084

BMDU =	4 .29581

Taken together, (0.609084, 4.29581) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0164181

This document is a draft for review purposes only and does not constitute Agency policy.

F-44	DRAFT—DO NOT CITE OR QUOTE

-------
F. 1.15.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

10:00 04/02 2010

National Toxicology Program, 1982: Hematopoietio System: Lymphoma or Leukemia

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-45	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.16. National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma
F. 1.16.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.340

155.213

1.488E+00

8.265E-01



Multistage Cancer,
2-Degree

2

0.340

155.213

1.488E+00

8.265E-01

final fi=0

Multistage Cancer,
3-Degree

2

0.340

155.213

1.488E+00

8.265E-01

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F. 1.16.2. Output for Selected Model: Multistage Cancer, 1 -Degree

National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\16_mscl_lPerc_mf_LivAdenCarc.(d)

Gnuplot Plotting File: C:\l\Blood\16_mscl_lPerc_mf_LivAdenCarc.plt

~Fri Apr 02 11:04:11 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.080941

Beta (1) = 0.00583089

This document is a draft for review purposes only and does not constitute Agency policy.

F-46	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.57

Beta(1)	-0.57	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0692161	*	*	*

Beta (1)	0.00675636	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-74.5177	4

Fitted model -75.6063 2	2.17736 2	0.3367

Reduced model -79.6703 1	10.3053 3	0.01614

AIC:	155.213

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0692 5.053 3.000	73	-0.947

1.9460 0.0814 4.069 6.000	50	0.999

5.8440 0.1053 5.052 6.000	48	0.446

32.0560 0.2505 11.772 11.000	47	-0.260

Chi ^2 = 2.16	d.f. = 2	P-value = 0.3395

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	1.487 54

BMDL =	0.826482

BMDU =	3.9863

Taken together, (0.826482, 3.9863 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0120995

This document is a draft for review purposes only and does not constitute Agency policy.

F-47	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.16.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	10:04 04/02 2010

3

4	National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-48	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.17. National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma
F. 1.17.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degreea

2

0.179

75.385

3.127E+00

1.380E+00



Multistage Cancer,
2-Degree

2

0.179

75.385

3.127E+00

1.380E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.179

75.385

3.127E+00

1.380E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F. 1.17.2. Output for Selected Model: Multistage Cancer, 1 -Degree
National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\17 mscl IPerc mice f thyroid aden.(d)

Gnuplot Plotting File: C:\l\Blood\17 mscl IPerc mice f thyroid aden.pit

Fri Apr 02 11:04:39 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0202346
Beta (1) = 0.00292833

This document is a draft for review purposes only and does not constitute Agency policy.

F-49	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.58

Beta(1)	-0.58	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0153082	*	*	*

Beta (1)	0.00329742	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-32.0017	4

Fitted model -34.3904 2	4.77738 2	0.09175

Reduced model -37.2405 1	10.4776 3	0.01491

AIC:	72.7807

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.9460
5.8440
32.0560

0.0153
0.0216
0.0341
1141

056

080
603

0

5.24E

0.	000
3. 000

1.	000
5. 000

69

50
47
46

-1.036
1.867
-0.484
-0.115

Chi ^2

d.f.

P-value

0.0904

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	3.04794

BMDL =	1.43569

BMDU =	138 87 6

Taken together, (1.43569, 138876 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00696528

This document is a draft for review purposes only and does not constitute Agency policy.

F-50	DRAFT—DO NOT CITE OR QUOTE

-------
F. 1.17.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

10:04 04/02 2010

National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-51	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.1.18. National Toxicology Program, 1982: Lung: Alveolar/Bronchiolar Adenoma or
Carcinoma

F. 1.18.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degree

2

0.088

168.342

6.499E-01

3.512E-01



Multistage
Cancer, 2-Degreea

2

0.167

166.946

2.528E+00

4.135E-01



Multistage Cancer,
3-Degree

2

0.182

166.799

4.147E+00

4.230E-01



a Best-fitting model, BMDS output presented in this appendix

F. 1.18.2. Output for Selected Model: Multistage Cancer, 2-Degree

National Toxicology Program, 1982: Lung: Alveolar/Bronchiolar Adenoma or Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\18 msc2 IPerc lung aden carc.(d)

Gnuplot Plotting File: C:\l\Blood\18 msc2 IPerc lung aden carc.pit

Fri Apr 02 11:05:09 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal*dose/Nl-beta2*dose/N2 ) ]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0868577
Beta(1) =	0

This document is a draft for review purposes only and does not constitute Agency policy.

F-52	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Beta(2) = 0.00165722

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Beta(l)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Background	Beta(2)

Background	1	-0.46

Beta(2)	-0.46	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf.	Limit Upper Conf. Limit

Background 0.0942466 *	*	*

Beta(1) 0 *	*	*

Beta(2) 0.00157255 *	*	*

Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-79.5959	4

Fitted model	-81.4729	2	3.754	2	0.153

Reduced model	-85.3351	1	11.4782	3	0.009402

AIC:	166.946

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000	0.0942	6.692	10.000	71	1.344

0.7665	0.0951	4.564	2.000	48	-1.262

2.2711	0.1016	4.875	4.000	48	-0.418

11.2437	0.2575	12.877	13.000	50	0.040

Chi ^2 = 3.57	d.f. = 2	P-value = 0.1674

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	2.52806

BMDL =	0.413504

BMDU =	4 .19905

Taken together, (0.413504, 4.19905) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0241835

This document is a draft for review purposes only and does not constitute Agency policy.

F-53	DRAFT—DO NOT CITE OR QUOTE

-------
F. 1.18.3. Figure for Selected Model: Multistage Cancer, 2-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

10:05 04/02 2010

National Toxicology Program, 1982: Lung: Alveolar/Bronchiolar Adenoma or Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-54	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

F.1.19. National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma
F. 1.19.1. Summary Table of BMDS Modeling Results

Degrees ,	BMD BMDL

Model	of 7 ;	AIC , „ ,, (ng/kg-	Notes

„ , Value	(ng/kg-d)

Freedom	** &	d)

Multistage	2	0.928 258.548 2.110E-01 1.378E-

Cancer, 1-Degree	01

a

Multistage	1	0.779 260.475 3.072E-01 1.385E-

Cancer, 2-Degree	01

Multistage	1	0.790 260.468 2.934E-01 1.385E-

Cancer, 3-Degree	01

a Best-fitting model, BMDS output presented in this appendix

F. 1.19.2. Output for Selected Model: Multistage Cancer, 1 -Degree

National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\19 mscl IPerc mice m liver aden carc.(d)

Gnuplot Plotting File: C:\l\Blood\19 mscl IPerc mice m liver aden carc.pit

Fri Apr 02 11:05:36 2010

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.201679

Beta (1) = 0.04864 92

Asymptotic Correlation Matrix of Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

F-55	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

Background
Beta(1)

Background
1

-0.53

Beta(1)

-0.53
1

Parameter Estimates

Variable
Background
Beta(1)

Estimate
0.204258
0. 0476385

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)	# Param'

-127.199	4

-127.274	2

-135.589	1

Deviance Test d.f.

0.149955
16.7801

P-value

0.9278
0.0007843

AIC:

258.548

Dose

Est. Prob.

Goodness of Fit

Expected

Observed

Scaled
Residual

0.0000
0.7665
2.2711
11.2437

0.2043
0.2328
0.2859
0.5343

14.911

11.407
14.007
26.713

15.000

12.000
13.000
27.000

73

49

49

50

0. 026
0.201
-0.318
0. 081

Chi ^2

0.15

d.f.

P-value

0.9283

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	0.210971

BMDL =	0.137771

BMDU =	0.383981

Taken together, (0.137771, 0.383981) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0725843

This document is a draft for review purposes only and does not constitute Agency policy.

F-56	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.19.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	10:05 04/02 2010

3

4	National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-57	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

F.1.20. National Toxicology Program, 2006: Liver: Cholangiocarcinoma
F. 1.20.1. Summary Table of BMDS Modeling Results

Model

Multistage
Cancer, 1-Degree

Multistage
Cancer, 2-Degree

Multistage
Cancer, 3-
Degreea	

Degrees
of

Freedom

5

5
5

i: p-

Value

0.001

0.405
0.993

BMD BMDL

(ng/kg-d) (ng/kg-d)

AIC

138.456	9.481E-01	7.114E-01

119.374	4.263E+00	2.959E+00

113.508	7.574E+00	4.133E+00

Notes

a Best-fitting model, BMDS output presented in this appendix

F. 1.20.2. Output for Selected Model: Multistage Cancer, 3-Degree
National Toxicology Program, 2006: Liver: Cholangiocarcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\20_msc3_lPerc_liv_cho-carc.(d)

Gnuplot Plotting File: C:\l\Blood\2 0 msc3 IPerc liv cho-carc.plt

Fri Apr 02 11:06:03 2010

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/Nl-beta2*dose/N2-beta3*dose/N3)

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) =	0

Beta(2) =	0

Beta(3) = 2.44727e-005

This document is a draft for review purposes only and does not constitute Agency policy.

F-58	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l)	-Beta(2)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(3)

Beta(3)	1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)
Beta(3)

Estimate

0
0
0

2 . 31301e-005

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-55.408
-55.7538
-96.9934

Param's Deviance Test d.f.

0.691671
83.1708

P-value

0.9834
C.0001

AIC:

113.508

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000

0

0000

0

000

0

000

49

0

000

2.5565

0

0004

0

019

0

000

48

-0

136

5.6937

0

0043

0

196

0

000

46

-0

444

9.7882

0

0215

1

073

1

000

50

-0

071

16.5688

0

0 9 9 9

4

893

4

000

49

-0

426

29.6953

0

4543

24

078

25

000

53

0

254

Chi/N2 = 0.47	d.f. = 5	P-value = 0.9933

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	7.57416

BMDL =	4 . 13304

BMDU =	8 . 42557

Taken together, (4.13304, 8.42557) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00241953

This document is a draft for review purposes only and does not constitute Agency policy.

F-59	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.20.3. Figure for Selected Model: Multistage Cancer, 3-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	10:06 04/02 2010

3

4	National Toxicology Program, 2006: Liver: Cholangiocarcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-60	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

F.1.21. National Toxicology Program, 2006: Liver: Hepatocellular adenoma
F. 1.21.1. Summary Table of BMDS Modeling Results

Model

Degrees of x p-

AIC

Freedom
5

Value
0.026

BMD
(ng/kg-d)

BMDL
(ng/kg-d)

0.509

Multistage Cancer,

1-Degree

Multistage Cancer,

2-Degree

Multistage Cancer,

3-Degreea

a Best-fitting model, BMDS output presented in this appendix

Notes

87.024 2.192E+00 1.455E+00

76.982 6.602E+00 4.342E+00

0.933 72.782 1.022E+01 6.527E+00

F. 1.21.2. Output For Selected Model: Multistage Cancer, 3-Degree
National Toxicology Program, 2006: Liver: Hepatocellular adenoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\21_msc3_lPerc_liv_hepat_ad.(d)

Gnuplot Plotting File: C:\l\Blood\21_msc3_lPerc_liv_hepat_ad.plt

~Fri Apr 02 11:06:32 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/Nl-beta2*dose/N2-beta3*dose/N3) ]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) =	0

Beta(2) =	0

Beta(3) = 1.08896e-005

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l)	-Beta(2)

This document is a draft for review purposes only and does not constitute Agency policy.

F-61	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(3)

Beta(3)	1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)
Beta(3)

Estimate

0
0
0

9. 41228e-006

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)
-34.4075
-35.3907
-56.3333

Param's Deviance Test d.f.

1.96648
43.8515

P-value

0.8538
C.0001

72.7815

Goodness of Fit

Scaled

Dose

Est

. Prob.

Expected

Observed

Size

Residual

0.0000

0.

0000

0

000

0

000

49

0

000

2.5565

0.

0002

0

008

0

000

48

-0

087

5.6937

0.

0017

0

080

0

000

46

-0

283

9.7882

0.

0088

0

439

0

000

50

-0

6 6 6

16.5688

0.

0419

2

054

1

000

49

-0

751

29.6953

0.

2184

11

577

13

000

53

0

473

Chi ^2

1. 32

d.f.

P-value

0.9330

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	10.221

BMDL =	6.52683

BMDU =	11.9754

Taken together, (6.52683, 11.9754) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00153214

This document is a draft for review purposes only and does not constitute Agency policy.

F-62	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.21.3. Figure For Selected Model: Multistage Cancer, 3-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	10:06 04/02 2010

3

4	National Toxicology Program, 2006: Liver: Hepatocellular adenoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-63	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

F.1.22. National Toxicology Program, 2006: Oral mucosa: squamous cell carcinoma
F. 1.22.1. Summary Table of BMDS Modeling Results

Model

Multistage
Cancer, 1-
Degreea
Multistage
Cancer, 2-Degree

Multistage
Cancer, 3-Degree

Degrees
of

Freedom

x2 p-

Value

AIC

BMD BMDL

(ng/kg-d) (ng/kg-d)

Notes

0.270 126.963 2.204E+00 1.389E+00

0.538 123.896 7.108E+00 2.158E+00
0.565 123.295 1.103E+01 2.298E+00

a Best-fitting model, BMDS output presented in this appendix

F. 1.22.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 2006: Oral mucosa: squamous cell carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\22_mscl_lPerc_oral_carc.(d)

Gnuplot Plotting File: C:\l\Blood\22_mscl_lPerc_oral_carc.plt

Fri Apr 02 11:07:00 2010

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-006
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) = 0.00629243

Asymptotic Correlation Matrix of Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

F-64	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Background	Beta(1)

Background	1	-0.67

Beta(1)	-0.67	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0139169	*	*	*

Beta(1)	0.00456055	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-57.5353	6

Fitted model	-61.4815	2	7.89233	4	0.0956

Reduced model	-67.7782	1	20.4858	5	0.001013

AIC:	126.963

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0

0139

0

682

1

000

49

0.388

2.5565

0

0253

1

217

2

000

48

0.719

5.6937

0

0392

1

803

1

000

46

-0.610

9.7882

0

0570

2

848

0

000

50

-1.738

16.5688

0

0857

4

198

4

000

49

-0.101

29.6953

0

1388

7

357

10

000

53

1. 050

Chi ^2 = 5.17	d.f. = 4	P-value = 0.2700

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	2.20376

BMDL =	1.38901

BMDU =	4 . 3103

Taken together, (1.38901, 4.3103 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00719939

This document is a draft for review purposes only and does not constitute Agency policy.

F-65	DRAFT—DO NOT CITE OR QUOTE

-------
F. 1.22.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

10:07 04/02 2010

National Toxicology Program, 2006: Oral mucosa: squamous cell carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-66	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

F.1.23. National Toxicology Program, 2006: Pancreas: adenoma or carcinoma
F. 1.23.1. Summary Table of BMDS Modeling Results

Model

Multistage
Cancer, 1-
Degreea
Multistage
Cancer, 2-Degree

Multistage
Cancer, 3-Degree

Degrees
of

Freedom

x2 p-

Value
0.640

0.929
0.986

AIC

BMD BMDL

(ng/kg-d) (ng/kg-d)

Notes

29.373 1.052E+01 4.630E+00

27.061 1.458E+01 7.227E+00
25.972 1.739E+01 9.373E+00

a Best-fitting model, BMDS output presented in this appendix

F. 1.23.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 2006: Pancreas: adenoma or carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\23 mscl IPerc pane ad carc.(d)

Gnuplot Plotting File: C:\l\Blood\23 mscl IPerc pane ad carc.pit

Fri Apr 02 11:07:29 2010

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-006
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) = 0.00191132

Asymptotic Correlation Matrix of Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

F-67	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta (1)	0.000955662	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-11.4096	6

Fitted model	-13.6865	1	4.55375	5	0.4727

Reduced model	-16.7086	1	10.598	5	0.05996

AIC:	29.373

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0

0000

0

000

0

000

48

-0

007

2.5565

0

0024

0

117

0

000

48

-0

343

5.6937

0

0054

0

250

0

000

46

-0

501

9.7882

0

0093

0

466

0

000

50

-0

686

16.5688

0

0157

0

754

0

000

48

-0

875

29.6953

0

0280

1

427

3

000

51

1

336

Chi/N2 = 3.39	d.f. = 5	P-value = 0.6403

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	10.5166

BMDL =	4 . 62967

BMDU =	32.8573

Taken together, (4.62967, 32.8573) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00215998

This document is a draft for review purposes only and does not constitute Agency policy.

F-68	DRAFT—DO NOT CITE OR QUOTE

-------
F. 1.23.3. Figure for Selected Model: Multistage Cancer, 1-Degree

7D

aj
o

0.15

0.1

0.05

0

Multistage Cancer Model with 0.95 Confidence Level

Multistage Cancer
Linear extrapolation

BMDL

BMD

0	5	10	15	20

dose

10:07 04/02 2010

National Toxicology Program, 2006: Pancreas: adenoma or carcinoma

25

30

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-69	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

F.1.24. National Toxicology Program, 2006: Lung: Cystic keratinizing epithelioma

F. 1.24.1. Summary Table of BMDS Modeling Results

Model

Multistage
Cancer, 1-Degree
Multistage
Cancer, 2-Degree

a

Multistage
Cancer, 3-Degree

Degrees
of

Freedom

i: p-

Value

0.062

0.507

0.845

BMD BMDL

(ng/kg-d) (ng/kg-d)

AIC

64.034 3.445E+00 2.084E+00

56.943 8.304E+00 5.245E+00

53.558 1.193E+01 7.765E+00

Notes

a Best-fitting model, BMDS output presented in this appendix

F. 1.24.2. Output for Selected Model: Multistage Cancer, 2-Degree
National Toxicology Program, 2006: Lung: Cystic keratinizing epithelioma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\24_msc2_lPerc_lung_epith.(d)

Gnuplot Plotting File: C:\l\Blood\2 4_msc2_lPerc_lung_epith.plt

Fri Apr 02 11:07:57 2010

The form of the probability function is:

background + (1-background)*[1-EXP(
-betal*dose/Nl-beta2*dose/N2 ) ]

P[response]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-006
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) =	0

Beta(2) = 0.000216412

This document is a draft for review purposes only and does not constitute Agency policy.

F-70	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(2)

Beta(2)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0	*	*	*

Beta(2)	0.000145744	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-23.958
-27 .4714
-40.2069

Param's Deviance Test d.f.

P-value

7.02662
32.4976

0.2187
C.0001

AIC:

56.9427

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000

0

0000

0

000

0

000

49

0

000

2.5565

0

0010

0

046

0

000

48

-0

214

5.6937

0

0047

0

217

0

000

46

-0

467

9.7882

0

0139

0

679

0

000

49

-0

830

16.5688

0

0392

1

922

0

000

49

-1

414

29.6953

0

1206

6

271

9

000

52

1

162

Chi/N2 = 4.30	d.f. = 5	P-value = 0.5067

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =

BMDL =
BMDU =

8.30415
5.24499
11. 2298

Taken together, (5.24499, 11.2298) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00190658

This document is a draft for review purposes only and does not constitute Agency policy.

F-71	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.24.3. Figure for Selected Model: Multistage Cancer, 2-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	10:07 04/02 2010

3

4	National Toxicology Program, 2006: Lung: Cystic keratinizing epithelioma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-72	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

F.1.25. Toth et al., 1979: Liver: Tumors

F. 1.25.1. Summary Table of BMDS Modeling Results

Model

Degrees
of

Freedom

r p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

1

0.293

155.740

3.684E-01

2.096E-01



Multistage Cancer,
2-Degree

1

0.293

155.740

3.684E-01

2.096E-01

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F. 1.25.2. Output for Selected Model: Multistage Cancer, 1-Degree
Toth et al., 1979: Liver: Tumors

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\25 mscl IPerc adr cor lyr.(d)

Gnuplot Plotting File: C:\l\Blood\25 mscl IPerc adr cor lyr.pit

Fri Apr 02 11:08:26 2010

Table 1

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 1
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.234952

Beta (1) = 0.0269892

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

This document is a draft for review purposes only and does not constitute Agency policy.

F-73	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

Background
Beta(1)

1	-0.55

-0.55	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.235297	*	*	*

Beta(1)	0.0272796	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-75.3127	3

Fitted model -75.8702 2	1.11506 1	0.291

Reduced model -79.4897 1	8.35401 2	0.01534

AIC:	155.74

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.2353 8.941 7.000	38	-0.742

0.5732 0.2472 10.875 13.000	44	0.743

14.2123 0.4811 21.167 21.000	44	-0.050

Chi ^2 = 1.11	d.f. = 1	P-value = 0.2931

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	0.368419

BMDL =	0.209642

BMDU =	1.01064

Taken together, (0.209642, 1.01064) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0477004

This document is a draft for review purposes only and does not constitute Agency policy.

F-74	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.25.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	10:08 04/02 2010

3

4	Toth et al., 1979: Liver: Tumors

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-75	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

F.1.26. Delia Porta et al., 1987: Table 4, B6C3 mice, male, hepatocellular carcinoma
F. 1.26.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degree

1

0.036

165.333

9.239E-01

6.933E-01



Multistage
Cancer, 2-Degreea

1

0.525

161.217

7.143E+00

1.170E+00



a Best-fitting model, BMDS output presented in this appendix

F. 1.26.2. Output for Selected Model: Multistage Cancer, 2-Degree
Delia Porta et al., 1987: Table 4, B6C3 mice, male, hepatocellular carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\94_DPorta_l987_Male_Hep_Carc_MultiCanc2_l.(d)

Gnuplot Plotting File: C:\l\Blood\94_DPorta_1987_Male_Hep_Carc_MultiCanc2_l.plt

Fri Apr 02 13:52:21 2010

Table 4, B6C3 mice, Male, Hepatocellular carcinoma

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal*dose/Nl-beta2*dose/N2 ) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 3

Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0865895
Beta(1) =	0

Beta ( 2) = 0.000211877

Asymptotic Correlation Matrix of Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

F-76	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

( *** The model parameter(s) -Beta(l)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Background	Beta(2)

Background	1	-0.64

Beta(2)	-0.64	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf.	Limit Upper Conf. Limit

Background 0.107218 *	*	*

Beta (1) 0 *	*	*

Beta(2) 0.00019698 *	*	*

- Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-78.4036	3

Fitted model -78.6083 2	0.409345 1	0.5223

Reduced model -94.7394 1	32.6717 2	<.0001

AIC:	161.217

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.1072 4.610 5.000	43	0.192

37.9990 0.3282 16.740 15.000	51	-0.519

67.7695 0.6387 31.936 33.000	50	0.313

Chi/N2 = 0.40	d.f. = 1	P-value = 0.5249

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	7.142 98

BMDL =	1.16991

BMDU =	8.58118

Taken together, (1.16991, 8.58118) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0085477

This document is a draft for review purposes only and does not constitute Agency policy.

F-77	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.26.3. Figure for Selected Model: Multistage Cancer, 2-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	12:52 04/02 2010

3

4	Delia Porta et al., 1987: Table 4, B6C3 mice, male, hepatocellular carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-78	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

F.1.27. Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular adenoma
F. 1.27.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degree

1

0.380

99.614

3.599E+00

2.186E+00



Multistage
Cancer, 2-Degreea

1

0.863

98.833

1.449E+01

2.342E+00



a Best-fitting model, BMDS output presented in this appendix

F. 1.27.2. Output for Selected Model: Multistage Cancer, 2-Degree
Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular adenoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\95_DPorta_l987_Female_Hep_Aden_MultiCanc2_l.(d)

Gnuplot Plotting File: C:\l\Blood\95_DPorta_1987_Female_Hep_Aden_MultiCanc2_l.plt

Fri Apr 02 13:52:51 2010

Table 4, B6C3 mice, Female, Hepatocellular adenoma

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal*dose/Nl-beta2*dose/N2 ) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 3

Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0364319
Beta(1) =	0

Beta(2) = 4.92861e-005

Asymptotic Correlation Matrix of Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

F-79	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

( *** The model parameter(s) -Beta(l)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Background	Beta(2)

Background	1	-0.69

Beta(2)	-0.69	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf.	Limit Upper Conf. Limit

Background 0.0392633 *	*	*

Beta(1) 0 *	*	*

Beta(2) 4.78928e-005 *	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-47.4015	3

Fitted model -47.4165 2	0.0299957 1	0.8625

Reduced model -51.6367 1	8.47042 2	0.01448

AIC:	98.8329

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0393 1.924 2.000	49	0.056

37.5865 0.1021 4.289 4.000	42	-0.147

66.9741 0.2250 10.800 11.000	48	0.069

Chi/N2 = 0.03	d.f. = 1	P-value = 0.8634

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	14.4862

BMDL =	2.34 21

BMDU =	22.1663

Taken together, (2.3421	, 22.1663) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope	Factor = 0.00426967

This document is a draft for review purposes only and does not constitute Agency policy.

F-80	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.27.3. Figure for Selected Model: Multistage Cancer, 2-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	12:52 04/02 2010

3

4	Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular adenoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-81	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

F.1.28. Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular carcinoma
F. 1.28.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

1

0.019

115.539

2.302E+00

1.545E+00



Multistage Cancer,
2-Degree

1

0.019

115.539

2.302E+00

1.545E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F. 1.28.2. Output for Selected Model: Multistage Cancer, 1-Degree
Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\l\Blood\96_DPorta_l987_Female_Hep_Carc_MultiCancl_l.(d)

Gnuplot Plotting File: C:\l\Blood\96_DPorta_1987_Female_Hep_Carc_MultiCancl_l.plt

Fri Apr 02 13:53:20 2010

Table 4, B6C3 mice, Female, Hepatocellular carcinoma

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 3

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0787329
Beta(1) = 0.00304814

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

This document is a draft for review purposes only and does not constitute Agency policy.

F-82	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

Background
Beta(1)

1	-0.8

-0.8	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0268873	*	*	*

Beta(1)	0.00436529	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-53.1726	3

Fitted model	-55.7697	2	5.19425	1	0.02266

Reduced model	-60.7146	1	15.084	2	0.0005303

AIC:	115.539

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0269 1.317 1.000	49	-0.280

37.5865 0.1741 7.314 12.000	42	1.907

66.9741 0.2736 13.131 9.000	48	-1.338

Chi/N2 = 5.50	d.f. = 1	P-value = 0.0190

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	2.30233

BMDL =	1.54 47 9

BMDU =	4.37768

Taken together, (1.54479, 4.37768) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00647339

This document is a draft for review purposes only and does not constitute Agency policy.

F-83	DRAFT—DO NOT CITE OR QUOTE

-------
1 F. 1.28.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	12:53 04/02 2010

3

4	Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-84	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

F.2. ADMINISTERED DOSE BMDS RESULTS

F.2.1. Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal
turbinates

F.2.1.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

3

0.928

30.745

1.344E+01

6.515E+00



Multistage Cancer,
2-Degree

3

0.998

29.961

3.490E+01

7.216E+00



Multistage Cancer,
3-Degree

3

1.000

29.885

4.941E+01

7.297E+00



a Best-fitting model, BMDS output presented in this appendix

F.2.1.2. Output for Selected Model: Multistage Cancer, 1-Degree

Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal turbinates

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\l mscl IPerc palate nasal.(d)

Gnuplot Plotting File: C:\Canc\l mscl IPerc palate nasal.pit

Thu Apr 01 12:47:40 2010

Source - Table 4

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

This document is a draft for review purposes only and does not constitute Agency policy.

F-85	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

Background =	0

Beta(1) = 0.000858074

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.00074801	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-13.9385	4

Fitted model -14.3726 1	0.868297 3	0.8331

Reduced model -20.2589 1	12.6409 3	0.005481

AIC:	30.7452

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0000 0.000 0.000	85	0.000

1.0000 0.0007 0.037 0.000	50	-0.193

10.0000 0.0075 0.373 0.000	50	-0.613

100.0000 0.0721 3.604 4.000	50	0.217

Chi/N2 = 0.46	d.f. = 3	P-value = 0.9276

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	13.4361

BMDL =	6.51522

BMDU =	34.82 9

Taken together, (6.51522, 34.829 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00153487

This document is a draft for review purposes only and does not constitute Agency policy.

F-86	DRAFT—DO NOT CITE OR QUOTE

-------
F.2.1.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

11:47 04/01 2010

Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal turbinates

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-87	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.2. Kociba et al., 1978: Stratified squamous cell carcinoma of tongue
F.2.2.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.451

48.368

1.742E+01

7.146E+00



Multistage Cancer,
2-Degree

2

0.451

48.368

1.742E+01

7.146E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.451

48.368

1.742E+01

7.146E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.2.2. Output for Selected Model: Multistage Cancer, 1-Degree
Kociba et al., 1978: Stratified squamous cell carcinoma of tongue

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\2 mscl IPerc tongue.(d)

Gnuplot Plotting File: C:\Canc\2 mscl IPerc tongue.pit

Thu Apr 01 12:48:16 2010

Source - Table 4

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0113883
Beta(1) = 0.000508703

This document is a draft for review purposes only and does not constitute Agency policy.

F-88	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.52

Beta(1)	-0.52	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.00809154	*	*	*

Beta (1)	0.000576915	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-21.1523	4

Fitted model -22.1838 2	2.06309 2	0.3565

Reduced model -24.1972 1	6.08976 3	0.1073

AIC:	48.3677

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0081 0.688 0.000	85	-0.833

1.0000 0.0087 0.433 1.000	50	0.865

10.0000 0.0138 0.690 1.000	50	0.376

100.0000 0.0637 3.185 3.000	50	-0.107

Chi/N2 = 1.59	d.f. = 2	P-value = 0.4506

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	17.4208

BMDL =	7 .14637

BMDU = 3.20359e+006

Taken together, (7.14637, 3.20359e+006) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00139931

This document is a draft for review purposes only and does not constitute Agency policy.

F-89	DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.2.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:48 04/01 2010

3

4	Kociba et al., 1978: Stratified squamous cell carcinoma of tongue

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-90	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.3. Kociba et al., 1978: Adenoma of adrenal cortex
F.2.3.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

3

0.376

53.518

7.587E+00

4.317E+00



Multistage Cancer,
2-Degree

3

0.376

53.518

7.587E+00

4.317E+00

final fi=0

Multistage Cancer,
3-Degree

3

0.376

53.518

7.587E+00

4.317E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.3.2. Output for Selected Model: Multistage Cancer, 1-Degree
Kociba et al., 1978: Adenoma of adrenal cortex

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\3 mscl IPerc adre adenoma.(d)

Gnuplot Plotting File: C:\Canc\3 mscl IPerc adre adenoma.pit

Thu Apr 01 12:48:52 2010

Source - Table 5

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.00927818
Beta(1) = 0.00098105

This document is a draft for review purposes only and does not constitute Agency policy.

F-91	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.00132464	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-24.6514	4

Fitted model	-25.759	1	2.2152	3	0.529

Reduced model	-31.4904	1	13.6781	3	0.003378

AIC:	53.5179

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.0000
10.0000
100.0000

0.0000
0.0013
0.0132
0.1241

0. 000
0. 066
0. 658
6.203

0. 000
0. 000
2 . 000
5. 000

85

50
50
50

0.	000
-0.257

1.	666
-0.516

Chi ^2

3.11

d.f.

P-value

0.3755

Benchmark Dose Computation

Specified effect	=	0.01

Risk Type	=	Extra risk

Confidence level	=	0.95

BMD	=	7.587 22

BMDL	=	4 . 31737

BMDU	=	17.638

Taken together, (4.31737, 17.638 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00231623

This document is a draft for review purposes only and does not constitute Agency policy.

F-92	DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.3.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:48 04/01 2010

3

4	Kociba et al., 1978: Adenoma of adrenal cortex

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-93	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.4. Kociba et al., 1978: Hepatocellular adenoma(s) or carcinoma(s)
F.2.4.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.034

146.199

1.769E+00

1.225E+00



Multistage Cancer,
2-Degree

2

0.034

146.199

1.768E+00

1.225E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.034

146.199

1.768E+00

1.225E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.4.2. Output for Selected Model: Multistage Cancer, 1-Degree
Kociba et al., 1978: Hepatocellular adenoma(s) or carcinoma(s)

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\4 mscl IPerc liver ad carc.(d)

Gnuplot Plotting File: C:\Canc\4 mscl IPerc liver ad carc.pit

Thu Apr~01~12:49:25 2010

Source - Table 1 in Goodman and Sauer 1992

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0591902
Beta(1) = 0.00458516

This document is a draft for review purposes only and does not constitute Agency policy.

F-94	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.47

Beta(1)	-0.47	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0328755	*	*	*

Beta(1)	0.00568299	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-68.2561	4

Fitted model -71.0993 2	5.68634 2	0.05824

Reduced model -89.1983 1	41.8843 3	<.0001

AIC:	146.199

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0329 2.827 2.000	86	-0.500

1.0000 0.0384 1.918 1.000	50	-0.676

10.0000 0.0863 4.315 9.000	50	2.359

100.0000 0.4521 20.346 18.000	45	-0.703

Chi ^2 = 6.77	d.f. = 2	P-value = 0.0339

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	1.7685

BMDL =	1.22517

BMDU =	2.77641

Taken together, (1.22517, 2.77641) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00816214

This document is a draft for review purposes only and does not constitute Agency policy.

F-95	DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.4.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:49 04/01 2010

3

4	Kociba et al., 1978: Hepatocellular adenoma(s) or carcinoma(s)

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-96	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.2.5. Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal
turbinates

F.2.5.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

3

0.928

30.745

1.344E+01

6.515E+00



Multistage Cancer,
2-Degree

3

0.998

29.961

3.490E+01

7.216E+00



Multistage Cancer,
3-Degree

3

1.000

29.885

4.941E+01

7.297E+00



a Best-fitting model, BMDS output presented in this appendix

F.2.5.2. Output for Selected Model: Multistage Cancer, 1-Degree

Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal turbinates

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\5 mscl IPerc nasal.(d)

Gnuplot Plotting File: C:\Canc\5 mscl IPerc nasal.pit

Thu Apr 01 12:49:59 2010

Source - Table 5

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.00343283
Beta(1) = 0.000825276

This document is a draft for review purposes only and does not constitute Agency policy.

F-97	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.000953868	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-18.7562	4

Fitted model -19.0532 1	0.594034 3	0.8978

Reduced model -24.1972 1	10.882 3	0.01238

AIC:	40.1064

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.0000
10.0000
100.0000

0.0000
0.0010
0.0095
0.0910

0. 000
0. 048
0.475
4 .458

0. 000

0.	000

1.	000
4 . 000

50
50
49

0. 000
-0.218
0.766
-0.227

Chi ^2

0.69

d.f.

P-value

0. 8764

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	10.5364

BMDL =	5.46907

BMDU =	25.8 64

Taken together, (5.46907, 25.864 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00182846

This document is a draft for review purposes only and does not constitute Agency policy.

F-98	DRAFT—DO NOT CITE OR QUOTE

-------
F.2.5.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

11:49 04/01 2010

Kociba et al., 1978: Stratified squamous cell carcinoma of hard palate or nasal turbinates

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-99	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.6. Kociba et al., 1978: Keratinizing squamous cell carcinoma of lung
F.2.6.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

3

0.837

43.792

7.311E+00

4.159E+00



Multistage Cancer,
2-Degree

3

0.994

42.346

2.568E+01

4.917E+00



Multistage Cancer,
3-Degree

3

1.000

42.207

4.026E+01

5.022E+00



a Best-fitting model, BMDS output presented in this appendix

F.2.6.2. Output for Selected Model: Multistage Cancer, 1-Degree
Kociba et al., 1978: Keratinizing squamous cell carcinoma of lung

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\6 mscl IPerc kera carc.(d)

Gnuplot Plotting File: C:\Canc\6 mscl IPerc kera carc.pit

Thu Apr 01 12:50:34 2010

Source - Table 5

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) = 0.00158635

This document is a draft for review purposes only and does not constitute Agency policy.

F-100 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta (1)	0.0013747	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-20.0957	4

Fitted model -20.8959 1	1.60041 3	0.6593

Reduced model -31.4904 1	22.7894 3	<.0001

AIC:	43.7918

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.0000
10.0000
100.0000

0.0000
0.0014
0.0137
0.1284

0. 000
0. 069
0. 683
6.294

0. 000
0. 000
0. 000
7 . 000

50
50
49

0. 000
-0.262
-0.832
0.302

Chi ^2

0. 85

d.f.

P-value

0. 8370

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	7.31091

BMDL =	4 . 15929

BMDU =	14 . 6306

Taken together, (4.15929, 14.6306) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00240426

This document is a draft for review purposes only and does not constitute Agency policy.

F-101 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.6.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:50 04/01 2010

3

4	Kociba et al., 1978: Keratinizing squamous cell carcinoma of lung

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-102 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.7. National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma
F.2.7.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.146

76.377

9.761E+00

3.964E+00



Multistage Cancer,
2-Degree

2

0.146

76.377

9.761E+00

3.964E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.146

76.377

9.761E+00

3.964E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.7.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\7 mscl IPerc sub fibro.(d)

Gnuplot Plotting File: C:\Canc\7 mscl IPerc sub fibro.pit

Thu Apr 01 12:51:07 2010

Source - Table 10

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.030595

Beta(1) = 0.000799545

This document is a draft for review purposes only and does not constitute Agency policy.

F-103 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.54

Beta(1)	-0.54	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0231556	*	*	*

Beta (1)	0.00102962	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-33.5998	4

Fitted model -36.1883 2	5.17698 2	0.07513

Reduced model -37.7465 1	8.29346 3	0.04032

AIC:	7 6.3766

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0232 1.737 0.000	75	-1.333

1.4000 0.0246 1.228 2.000	50	0.705

7.1000 0.0303 1.514 3.000	50	1.227

71.0000 0.0920 4.509 4.000	49	-0.252

Chi ^2 = 3.84	d.f. = 2	P-value = 0.1463

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	9.76124

BMDL =	3.96354

BMDU = 1.03301e+006

Taken together, (3.96354, 1.03301e+006) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.002523

This document is a draft for review purposes only and does not constitute Agency policy.

F-104 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.7.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:51 04/01 2010

3

4	National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-105 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.2.8. National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular
Carcinoma

F.2.8.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.398

133.832

2.554E+00

1.600E+00



Multistage Cancer,
2-Degree

2

0.503

133.436

1.334E+01

1.652E+00



Multistage Cancer,
3-Degree

2

0.503

133.436

1.334E+01

1.652E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.8.2. Output for Selected Model: Multistage Cancer, 1-Degree

National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\8 mscl IPerc liver nod.(d)

Gnuplot Plotting File: C:\Canc\8 mscl IPerc liver nod.pit

Thu Apr~01 12:51:41 2010

Source - Table 10

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0383072
Beta(1) = 0.00417257

This document is a draft for review purposes only and does not constitute Agency policy.

F-106 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.47

Beta(1)	-0.47	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0451327	*	*	*

Beta(1)	0.00393556	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-63.9149	4

Fitted model	-64.916	2	2.00214	2	0.3675

Reduced model	-74.0195	1	20.2092	3	0.0001536

AIC:	133.832

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.4000
7.1000
71.0000

0.0451

0.0504
0.0714
0 . 277 9

3.385
2.469
3.572
13.618

5. 000
1. 000
3. 000
14.000

75

49

50
49

0. 898
-0.959
-0.314
0.122

Chi ^2

1.

d.f.

P-value

0.3984

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	2 . 55373

BMDL =	1.5 9 983

BMDU =	4.74206

Taken together, (1.59983, 4.74206) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00625067

This document is a draft for review purposes only and does not constitute Agency policy.

F-107 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.8.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:51 04/01 2010

3

4	National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-108 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.2.9. National Toxicology Program, 1982: Adrenal: Cortical Adenoma, or Carcinoma or
Adenoma, NOS

F.2.9.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.405

203.380

3.672E+00

1.871E+00



Multistage Cancer,
2-Degree

2

0.501

202.885

1.577E+01

1.974E+00



Multistage Cancer,
3-Degree

2

0.513

202.832

2.600E+01

1.986E+00



a Best-fitting model, BMDS output presented in this appendix

F.2.9.2. Output for Selected Model: Multistage Cancer, 1-Degree

National Toxicology Program, 1982: Adrenal: Cortical Adenoma, or Carcinoma or Adenoma,
NOS

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\9 mscl IPerc adre cort ad carc.(d)

Gnuplot Plotting File: C:\Canc\9 mscl IPerc adre cort ad carc.pit

Thu Apr 01 12:53:57 2010

Source - Table 10

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.140663

This document is a draft for review purposes only and does not constitute Agency policy.

F-109 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

Beta(1) = 0.00289845

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background
Beta(1)

1

-0.48

-0.48
1

Parameter Estimates

Variable
Background
Beta(1)

Estimate
0.143284
0.00273674

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table

Log(likelihood)
-98.7282
-99.6898
-102.201

Param's

4
2
1

Deviance Test d.f.

P-value

1.92318
6.94 636

0.3823
0. 07363

AIC:

203.38

Goodness of Fit

Est. Prob.

Expected

Observed

Scaled
Residual

0.

1.
7 .

71.

0000
4000
1000
0000

0.1433
0.1466
0.1598
0.2946

10.460
7 .181
7.829
13.551

11.
9.
5.

14 .

000
000
000
000

73

49
49
46

0.180
0.735
-1.103
0.145

Chi ^2

1. 81

d.f.

P-value

0.4046

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	3.67237

BMDL =	1 . 87133

BMDU =	15.4002

Taken together, (1.87133, 15.4002) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00534381

This document is a draft for review purposes only and does not constitute Agency policy.

F-l 10 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.9.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:53 04/01 2010

3

4	National Toxicology Program, 1982: Adrenal: Cortical Adenoma, or Carcinoma or Adenoma,

5	NOS

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-lll DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.10. National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma
F.2.10.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.661

92.020

7.571E+00

3.488E+00



Multistage Cancer,
2-Degree

2

0.769

91.639

2.257E+01

3.656E+00



Multistage Cancer,
3-Degree

2

0.781

91.601

3.302E+01

3.675E+00



a Best-fitting model, BMDS output presented in this appendix

F.2.10.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\10_mscl_lPerc_thy_ad.(d)

Gnuplot Plotting File: C:\Canc\10_mscl_lPerc_thy_ad.plt

Thu Apr 01 12:54:31 2010

Source - Table 10

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.032089

Beta(1) = 0.00143599

This document is a draft for review purposes only and does not constitute Agency policy.

F-l 12 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.5

Beta(1)	-0.5	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0345958	*

Beta(1)	0.00132742	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-43.5264	4

Fitted model -44.0098 2	0.966786 2	0.6167

Reduced model -46.2299 1	5.40699 3	0.1443

AIC:	92.0196

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0346 2.525 3.000	73	0.304

1.4000 0.0364 1.637 2.000	45	0.289

7.1000 0.0437 2.139 1.000	49	-0.796

71.0000 0.1214 5.707 6.000	47	0.131

Chi ^2 = 0.83	d.f. = 2	P-value = 0.6614

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	7.57131

BMDL =	3.4 8 815

BMDU =	964 541

Taken together, (3.48815, 964541 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00286685

This document is a draft for review purposes only and does not constitute Agency policy.

F-113 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.10.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:54 04/01 2010

3

4	National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-l 14 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.2.11. National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular
Carcinoma

F.2.11.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.398

133.832

2.554E+00

1.600E+00



Multistage Cancer,
2-Degree

2

0.503

133.436

1.334E+01

1.652E+00



Multistage Cancer,
3-Degree

2

0.503

133.436

1.334E+01

1.652E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.11.2. Output for Selected Model: Multistage Cancer, 1 -Degree

National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\ll mscl IPerc liver nod.(d)

Gnuplot Plotting File: C:\Canc\ll mscl IPerc liver nod.pit

Thu Apr 01 12:55:05 2010

Source - Table 9

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) = 0.000900399

This document is a draft for review purposes only and does not constitute Agency policy.

F-l 15 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta(1)	0.000775683	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-11.3484	4

Fitted model	-11.6976	1	0.698469	3	0.8736

Reduced model	-15.9189	1	9.14109	3	0.02747

AIC:	25.3952

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.4000
7.1000
71.0000

0.0000
0.0011
0.0055
0.0536

0. 000
0. 054
0.275
2.679

0. 000
0. 000
0. 000
3. 000

74

50
50
50

0. 000
-0.233
-0.525
0.201

Chi ^2

0.37

d.f.

P-value

0.9462

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	12.9568

BMDL =	5.70369

BMDU =	3 9.987 8

Taken together, (5.70369, 39.9878) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00175325

This document is a draft for review purposes only and does not constitute Agency policy.

F-l 16 DRAFT—DO NOT CITE OR QUOTE

-------
F.2.11.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

11:55 04/01 2010

National Toxicology Program, 1982: Liver: Neoplastic Nodule or Hepatocellular Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-117 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.2.12. National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma or
Carcinoma

F.2.12.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.028

151.224

3.521E+00

1.916E+00



Multistage Cancer,
2-Degree

2

0.028

151.224

3.521E+00

1.916E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.028

151.224

3.521E+00

1.916E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.12.2. Output for Selected Model: Multistage Cancer, 1-Degree

National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma or Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\12_mscl_lPerc_thyroid.(d)

Gnuplot Plotting File: C:\Canc\12 mscl IPerc thyroid.pit

Thu Apr 01 12:55:38 2010

Source - Table 9

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0867382
Beta(1) = 0.00232055

This document is a draft for review purposes only and does not constitute Agency policy.

F-l 18 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.53

Beta(1)	-0.53	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0704713	*	*	*

Beta (1)	0.00285481	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance Table

Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value

Full model	-69.5946	4

Fitted model	-73.6119	2	8.03468	2	0.018

Reduced model	-77.5267	1	15.8643	3	0.001209

AIC:	151.224

Goodness of Fit

Scaled

Est. Prob. Expected Observed	Size	Residual

0.0000
1.4000
7.1000
71.0000

0. 0705
0. 0742
0. 0891
0.2410

4 .863
3.561
4 .456
12.051

1. 000
5. 000
8 . 000
11.000

69

48
50
50

-1.817
0.793
1.759
-0.347

Chi ^2

7 .14

d.f.

P-value

0.0281

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	3.5205

BMDL =	1.91558

BMDU =	9.7 6663

Taken together, (1.91558, 9.76663) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00522034

This document is a draft for review purposes only and does not constitute Agency policy.

F-l 19 DRAFT—DO NOT CITE OR QUOTE

-------
F.2.12.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

11:55 04/01 2010

National Toxicology Program, 1982: Thyroid: Follicular-Cell Adenoma or Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-120 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.13. National Toxicology Program, 1982: Adrenal cortex: Adenoma
F.2.13.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.054

199.672

1.400E+01

3.444E+00



Multistage Cancer,
2-Degree

2

0.054

199.672

1.400E+01

3.444E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.054

199.672

1.400E+01

3.444E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.13.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 1982: Adrenal cortex: Adenoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\13 mscl IPerc adre cort.(d)

Gnuplot Plotting File: C:\Canc\13 mscl IPerc adre cort.pit

Thu Apr~01 12:56:10 2010

Source - Table 9

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.168444

Beta(1) = 0.000395949

This document is a draft for review purposes only and does not constitute Agency policy.

F-121 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.53

Beta(1)	-0.53	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.153096	*	*	*

Beta (1)	0.000718012	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-94.8672	4

Fitted model -97.8359 2	5.93732 2	0.05137

Reduced model -98.0432 1	6.35197 3	0.09569

AIC:	199.672

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0.1531

11.023

6. 000

72

-1.644

1.4000

0.1539

7 . 697

9. 000

50

0.510

7.1000

0.1574

7 .713

12.000

49

1. 682

71.0000

0.1952

9.564

9. 000

49

-0.203

Chi/N2 = 5.83	d.f. = 2	P-value = 0.0541

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	13.9974

BMDL =	3.4 4 43

BMDU did not converge for BMR = 0.010000
BMDU calculation failed
BMDU = Inf

This document is a draft for review purposes only and does not constitute Agency policy.

F-122 DRAFT—DO NOT CITE OR QUOTE

-------
F.2.13.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

T3

aj
o

0.4

0.35

0.3

0.25

0.2

0.15

0.1

0.05

0

Multistage Cancer
Linear extrapolation

BMDL

BMD

0	10	20	30	40	50

dose

11:56 04/01 2010

National Toxicology Program, 1982: Adrenal cortex: Adenoma

60

70

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-123 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.14. National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma
F.2.14.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.146

76.377

9.761E+00

3.964E+00



Multistage Cancer,
2-Degree

2

0.146

76.377

9.761E+00

3.964E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.146

76.377

9.761E+00

3.964E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.14.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\14 mscl IPerc subcu fibro.(d)

Gnuplot Plotting File: C:\Canc\14 mscl IPerc subcu fibro.pit

Thu Apr 01 12:56:41 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0143554
Beta(1) = 0.000341874

This document is a draft for review purposes only and does not constitute Agency policy.

F-124 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.5

Beta(1)	-0.5	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0145028	*

Beta(1)	0.000338561	*

Indicates that this value is not calculated.

Analysis of Deviance Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-30.9876	4

Fitted model -31.0199 2	0.0645971 2	0.9682

Reduced model -34.3291 1	6.68308 3	0.08272

AIC:	6 6.0397

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0145 1.073 1.000	74	-0.071

5.7000 0.0164 0.820 1.000	50	0.200

28.6000 0.0240 1.152 1.000	48	-0.143

286.0000 0.1055 4.956 5.000	47	0.021

Chi/N2 = 0.07	d.f. = 2	P-value = 0.9675

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	2 9.6855

BMDL =	14.3524

BMDU =	100.382

Taken together, (14.3524, 100.382) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.000696747

This document is a draft for review purposes only and does not constitute Agency policy.

F-125 DRAFT—DO NOT CITE OR QUOTE

-------
F.2.14.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

11:56 04/01 2010

National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-126 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.2.15. National Toxicology Program, 1982: Hematopoietio System: Lymphoma or
Leukemia

F.2.15.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.987

261.425

1.034E+01

5.456E+00



Multistage Cancer,
2-Degree

2

0.987

261.425

1.034E+01

5.456E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.987

261.425

1.034E+01

5.456E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.15.2. Output for Selected Model: Multistage Cancer, 1-Degree

National Toxicology Program, 1982: Hematopoietio System: Lymphoma or Leukemia

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\15 mscl IPerc mice f lymphoma.(d)

Gnuplot Plotting File: C:\Canc\15 mscl IPerc mice f lymphoma.pit

Thu Apr~01 12:57:14 2010

Table 15 page 64 Hematopoietic System Lymphoma or Leukemia

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.242959

Beta(1) = 0.000967723

This document is a draft for review purposes only and does not constitute Agency policy.

F-127 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background
Beta(1)

1

-0.48

-0.48
1

Variable
Background
Beta(1)

Parameter Estimates

Estimate
0.242712
0. 000971954

Std. Err.

Indicates that this value is not calculated.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table

Log(likelihood)
-128.699
-128.712
-131.412

Param's

4
2
1

Deviance Test d.f.

P-value

0. 0264819
5. 42487

0.9868
0.1432

AIC:

261.425

Dose

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000
5.7000
28.6000
286.0000

0.2427
0.2469
0.2635
0.4265

17.961

12 . 345
12.647
20.045

18 . 000

12.000
13.000
20.000

74

50
48
47

0. 011
-0.113
0.116
-0.013

Chi ^2

0. 03

d.f.

P-value

0.9868

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	10.3403

BMDL =	5.45599

BMDU =	38.9139

Taken together, (5.45599, 38.9139) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00183285

This document is a draft for review purposes only and does not constitute Agency policy.

F-128 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.15.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:57 04/01 2010

3

4	National Toxicology Program, 1982: Hematopoietio System: Lymphoma or Leukemia

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-129 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.16. National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma
F.2.16.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.244

156.001

1.458E+01

7.829E+00



Multistage Cancer,
2-Degree

2

0.244

156.001

1.458E+01

7.829E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.244

156.001

1.458E+01

7.829E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.16.2. Output for Selected Model: Multistage Cancer, 1-Degree

National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\16 mscl IPerc mice f liv aden carc.(d)

Gnuplot Plotting File: C:\Canc\16 mscl IPerc mice f liv aden carc.pit

Thu Apr~01 12757:47 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0888873
Beta (1) = 0.000616931

This document is a draft for review purposes only and does not constitute Agency policy.

F-130 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.5

Beta(1)	-0.5	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0788077	*

Beta(1)	0.000689385	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-74.5177	4

Fitted model -76.0006 2	2.96597 2	0.227

Reduced model -79.6703 1	10.3053 3	0.01614

AIC:	156.001

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0788 5.753 3.000	73	-1.196

5.7000 0.0824 4.121 6.000	50	0.966

28.6000 0.0968 4.646 6.000	48	0.661

286.0000 0.2436 11.452 11.000	47	-0.153

Chi ^2 = 2.82	d.f. = 2	P-value = 0.2436

Benchmark Dose Computation

Specified effect	=	0.01

Risk Type	=	Extra risk

Confidence level	=	0.95

BMD	=	14.5787

BMDL	=	7 . 82902

BMDU	=	42 . 4536

Taken together, (7.82902, 42.4536) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0012773

This document is a draft for review purposes only and does not constitute Agency policy.

F-131 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.16.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:57 04/01 2010

3

4	National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-132 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.17. National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma
F.2.17.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.146

76.377

9.761E+00

3.964E+00



Multistage Cancer,
2-Degree

2

0.146

76.377

9.761E+00

3.964E+00

final fi=0

Multistage Cancer,
3-Degree

2

0.146

76.377

9.761E+00

3.964E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.17.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\17 mscl IPerc mice f thyroid aden.(d)

Gnuplot Plotting File: C:\Canc\17 mscl IPerc mice f thyroid aden.pit

Thu Apr~01 12:58:20 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.02405

Beta(1) = 0.000315564

This document is a draft for review purposes only and does not constitute Agency policy.

F-133 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background	1	-0.51

Beta(1)	-0.51	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0207192	*	*	*

Beta (1)	0.000331835	*	*	*

* - Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-32.0017	4

Fitted model -34.6122 2	5.22112 2	0.07349

Reduced model -37.2405 1	10.4776 3	0.01491

AIC:	73.2245

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0207 1.430 0.000	69	-1.208

5.7000 0.0226 1.128 3.000	50	1.782

28.6000 0.0300 1.409 1.000	47	-0.350

286.0000 0.1094 5.032 5.000	46	-0.015

Chi ^2 = 4.76	d.f. = 2	P-value = 0.0927

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	30.2871

BMDL =	13.993

BMDU =	130.014

Taken together, (13.993	, 130.014) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope	Factor = 0.000714641

This document is a draft for review purposes only and does not constitute Agency policy.

F-134 DRAFT—DO NOT CITE OR QUOTE

-------
F.2.17.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

11:58 04/01 2010

National Toxicology Program, 1982: Subcutaneous Tissue: Fibrosarcoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-135 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

F.2.18. National Toxicology Program, 1982: Lung: Alveolar/Bronchiolar Adenoma or
Carcinoma

F.2.18.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

x2 p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degree

2

0.138

167.341

3.706E+00

2.026E+00



Multistage
Cancer, 2-Degreea

2

0.181

166.805

1.590E+01

2.139E+00



Multistage Cancer,
3-Degree

2

0.185

166.777

2.618E+01

2.145E+00



a Best-fitting model, BMDS output presented in this appendix

F.2.18.2. Output for Selected Model: Multistage Cancer, 2-Degree

National Toxicology Program, 1982: Lung: Alveolar/Bronchiolar Adenoma or Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\18 msc2 IPerc lung aden carc.(d)

Gnuplot Plotting File: C:\Canc\18 msc2 IPerc lung aden carc.pit

Thu Apr 01 12:58:55 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal*dose/Nl-beta2*dose/N2 ) ]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0889033
Beta(1) =	0

This document is a draft for review purposes only and does not constitute Agency policy.

F-136 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Beta(2) = 4.12413e-005

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Beta(l)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Background	Beta(2)

Background	1	-0.45

Beta(2)	-0.45	1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)

Estimate

0.0953987
0

3.97 322e-005

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

* - Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)
-79.5959
-81.4024
-85.3351

Param's

4
2
1

Deviance Test d.f.

3.61287
11.4782

P-value

0.1642
0.009402

166.805

Dose

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000	0.0954	6.773 10.000	71	1.304

1.4000	0.0955	4.583 2.000	48	-1.268

7.1000	0.0972	4.666 4.000	48	-0.325

71.0000	0.2596	12.979 13.000	50	0.007

Chi ^2 = 3.41 d.f.	= 2	P-value = 0.1814

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =

BMDL =
BMDU =

15.9045

2 .1388
26.2712

Taken together, (2.1388 , 26.2712) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00467551

This document is a draft for review purposes only and does not constitute Agency policy.

F-137 DRAFT—DO NOT CITE OR QUOTE

-------
F.2.18.3. Figure for Selected Model: Multistage Cancer, 2-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

11:58 04/01 2010

National Toxicology Program, 1982: Lung: Alveolar/Bronchiolar Adenoma or Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-138 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.19. National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma
F.2.19.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

2

0.916

258.572

1.338E+00

8.620E-01



Multistage Cancer,
2-Degree

2

0.916

258.572

1.338E+00

8.620E-01

final fi=0

Multistage Cancer,
3-Degree

2

0.916

258.572

1.338E+00

8.620E-01

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.19.2. Output for Selected Model: Multistage Cancer, 1-Degree

National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\19 mscl IPerc mice m liver aden carc.(d)

Gnuplot Plotting File: C:\Canc\19 mscl IPerc mice m liver aden carc.pit

Thu Apr~01 12:59:28 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 4

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.22264

Beta(1) = 0.0074005

This document is a draft for review purposes only and does not constitute Agency policy.

F-139 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background
Beta(1)

1

-0. Ai

-0. 4t

Variable
Background
Beta(1)

Parameter Estimates

Estimate
0.219315
0.00750879

Std. Err.

Indicates that this value is not calculated.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table

Log(likelihood)
-127.199
-127.286
-135.589

258.572

Param's

4
2
1

Deviance Test d.f.

0.174343
16.7801

P-value

0. 9165
0.0007843

Dose

Goodness of Fit

Est. Prob.

Expected

Observed

Size

Scaled
Residual

0.0000
1.4000
7.1000
71.0000

Chi ^2

0.17

0.2193
0.2275
0.2598
0.5419

d.f.

16.010
11.146
12.732
27.096

15.000

12.000
13.000
27.000

P-value

0. 9164

73

49

49

50

-0.286

0.291
0. 087
-0.027

Benchmark Dose Computation

Specified effect
Risk Type
Confidence level

BMD
BMDL
BMDU

0. 01
Extra risk

0. 95
1.33848
0. 861975
2 .4671

Taken together, (0.861975, 2.4671 ) is a 90
interval for the BMD

two-sided confidence

Multistage Cancer Slope Factor

0. 0116013

This document is a draft for review purposes only and does not constitute Agency policy.

F-140 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.19.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	11:59 04/01 2010

3

4	National Toxicology Program, 1982: Liver: Hepatocellular Adenoma or Carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-141 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.20. National Toxicology Program, 2006: Liver: Cholangiocarcinoma
F.2.20.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degree

5

0.024

129.070

1.872E+00

1.404E+00



Multistage Cancer,
2-Degree

5

0.947

114.349

9.440E+00

5.290E+00



Multistage
Cancer, 3-Degreea

4

0.995

115.158

1.310E+01

4.468E+00



a Best-fitting model, BMDS output presented in this appendix

F.2.20.2. Output for Selected Model: Multistage Cancer, 3-Degree
National Toxicology Program, 2006: Liver: Cholangiocarcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\20 msc3 IPerc liv cho-carc.(d)

Gnuplot Plotting File: C:\Canc\20 msc3 IPerc liv cho-carc.pit

Thu Apr 01 13:00:03 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/Nl-beta2*dose/N2-beta3*dose/N3) ]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) = 0.000561481
Beta(2) = 1.74365e-005
Beta(3) = 1.40248e-006

This document is a draft for review purposes only and does not constitute Agency policy.

F-142 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(2)	Beta(3)

Beta(2)	1	-0.99

Beta(3)	-0.99	1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)
Beta(3)

Estimate

0
0

4 . 35 92 7e-0 05
1.14186e-006

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-55.408
-55.5789
-96.9934

Param's Deviance Test d.f.

0.34181
83.1708

P-value

0. 987
0001

AIC:

115.158

Goodness of Fit

Scaled

Dose

Est

. Prob.

Expected

Observed

Size

Residual

0.0000

0.

0000

0. 000

0

000

49

0

000

2.1400

0.

0002

0. 010

0

000

48

-0

101

7.1400

0.

0026

0.121

0

000

46

-0

349

15.7000

0.

0150

0.752

1

000

50

0

288

32.9000

0.

0841

4 121

4

000

49

-0

062

71.4000

0.

4716

24.994

25

000

53

0

002

Chi ^2

0.22

d.f.

P-value

0.9945

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	13.1014

BMDL =	4 . 46755

BMDU =	19.17 83

Taken together, (4.46755, 19.1783) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00223836

This document is a draft for review purposes only and does not constitute Agency policy.

F-143 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.20.3. Figure for Selected Model: Multistage Cancer, 3-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	12:00 04/01 2010

3

4	National Toxicology Program, 2006: Liver: Cholangiocarcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-144 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.21. National Toxicology Program, 2006: Liver: Hepatocellular adenoma
F.2.21.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degree

5

0.131

82.310

4.393E+00

2.915E+00



Multistage Cancer,
2-Degree

5

0.857

73.656

1.475E+01

8.618E+00



Multistage
Cancer, 3-Degreea

5

0.999

71.216

2.379E+01

1.153E+01



a Best-fitting model, BMDS output presented in this appendix

F.2.21.2. Output for Selected Model: Multistage Cancer, 3-Degree
National Toxicology Program, 2006: Liver: Hepatocellular adenoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\21 msc3 IPerc liv hepat ad.(d)

Gnuplot Plotting File: C:\Canc\21 msc3 IPerc liv hepat ad.pit

Thu Apr 01 13:00:36 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/Nl-beta2*dose/N2-beta3*dose/N3) ]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) =	0

Beta(2) =	0

Beta(3) = 7.77141e-007

This document is a draft for review purposes only and does not constitute Agency policy.

F-145 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l)	-Beta(2)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(3)

Beta(3)	1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)
Beta(3)

Estimate

0
0
0

7 . 46408e-007

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)
-34.4075
-34.6078
-56.3333

Param's Deviance Test d.f.

0.40065
43.8515

P-value

0.9953
C.0001

AIC:

71. 2156

Dose

Est. Prob.

Goodness of Fit
Expected Observed

Scaled
Residual

0.0000

0

0000

0

000

0

000

49

0

000

2.1400

0

0000

0

000

0

000

48

-0

019

7.1400

0

0003

0

012

0

000

46

-0

112

15.7000

0

0029

0

144

0

000

50

-0

380

32.9000

0

0262

1

285

1

000

49

-0

255

71.4000

0

2379

12

609

13

000

53

0

126

Chi/N2 = 0.24	d.f. = 5	P-value = 0.9986

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	23.7 904

BMDL =	11.5343

BMDU =	27 . 8755

Taken together, (11.5343, 27.8755) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.000866978

This document is a draft for review purposes only and does not constitute Agency policy.

F-146 DRAFT—DO NOT CITE OR QUOTE

-------
F.2.21.3. Figure for Selected Model: Multistage Cancer, 3-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

12:00 04/01 2010

National Toxicology Program, 2006: Liver: Hepatocellular adenoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-147 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.22. National Toxicology Program, 2006: Oral mucosa: squamous cell carcinoma
F.2.22.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

4

0.386

125.484

4.751E+00

2.956E+00



Multistage Cancer,
2-Degree

4

0.587

123.245

1.635E+01

3.845E+00



Multistage Cancer,
3-Degree

4

0.587

123.245

1.635E+01

3.844E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.22.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 2006: Oral mucosa: squamous cell carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\22 mscl IPerc oral carc.(d)

Gnuplot Plotting File: C:\Canc\22 mscl IPerc oral carc.pit

Thu Apr~01 13:01:11 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.00607545
Beta(1) = 0.00265195

This document is a draft for review purposes only and does not constitute Agency policy.

F-148 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

Background
Beta(1)

1

-0. (

-0. (

Parameter Estimates

Variable
Background
Beta(1)

Estimate

0.0171416
0.00211536

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-57.5353
-60.7418
-67.7782

Param's Deviance Test d.f.

6.41293
20.4858

P-value

0.1704
0.001013

125.484

Goodness of Fit

Scaled

Dose

Est

. Prob.

Expected

Observed

Size

Residual

0.0000

0.

0171

0

840

1

000

49

0

176

2.1400

0.

0216

1

036

2

000

48

0

958

7.1400

0.

0319

1

466

1

000

46

-0

391

15.7000

0.

04 92

2

462

0

000

50

-1

609

32.9000

0.

0832

4

078

4

000

49

-0

040

71.4000

0.

1549

8

211

10

000

53

0

679

Chi ^2

4 .15

d.f.

P-value

0.3855

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	4 .75111

BMDL =	2.9556

BMDU =	9.194 54

Taken together, (2.9556	, 9.19454) is a 90 % two-sided confidence
interval for the BMD

Multistage Cancer Slope	Factor = 0.0033834

This document is a draft for review purposes only and does not constitute Agency policy.

F-149 DRAFT—DO NOT CITE OR QUOTE

-------
F.2.22.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

12:01 04/01 2010

National Toxicology Program, 2006: Oral mucosa: squamous cell carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-l50 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

F.2.23. National Toxicology Program, 2006: Pancreas: adenoma or carcinoma
F.2.23.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

5

0.796

28.316

2.120E+01

9.335E+00



Multistage Cancer,
2-Degree

5

0.977

26.230

3.270E+01

1.389E+01



Multistage Cancer,
3-Degree

5

0.997

25.427

4.057E+01

1.755E+01



a Best-fitting model, BMDS output presented in this appendix

F.2.23.2. Output for Selected Model: Multistage Cancer, 1-Degree
National Toxicology Program, 2006: Pancreas: adenoma or carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\23 mscl IPerc pane ad carc.(d)

Gnuplot Plotting File: C:\Canc\23 mscl IPerc pane ad carc.pit

Thu Apr~01~13:01:43 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) = 0.000817541

Asymptotic Correlation Matrix of Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

F-151 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

( *** The model parameter(s) -Background

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(1)

Beta(1)

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0	*	*	*

Beta (1)	0.000474004	*	*	*

- Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-11.4096	6

Fitted model -13.1581 1	3.49702 5	0.6238

Reduced model -16.7086 1	10.598 5	0.05996

AIC:	28.3163

Goodness of Fit

Scaled

Dose	Est. Prob. Expected Observed	Size	Residual

0.0000

0

0000

0

000

0

000

48

0

000

2.1400

0

0010

0

049

0

000

48

-0

221

7.1400

0

0034

0

155

0

000

46

-0

395

15.7000

0

0074

0

371

0

000

50

-0

611

32.9000

0

0155

0

743

0

000

48

-0

869

71.4000

0

0333

1

697

3

000

51

1

017

Chi ^2 = 2.37	d.f. = 5	P-value = 0.7964

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	21.2031

BMDL =	9.33481

BMDU =	65.4351

Taken together, (9.33481, 65.4351) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00107126

This document is a draft for review purposes only and does not constitute Agency policy.

F-152 DRAFT—DO NOT CITE OR QUOTE

-------
F.2.23.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

12:01 04/01 2010

National Toxicology Program, 2006: Pancreas: adenoma or carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-153 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

F.2.24. National Toxicology Program, 2006: Lung: Cystic keratinizing epithelioma

F.2.24.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degree

5

0.192

60.806

6.922E+00

4.187E+00



Multistage
Cancer, 2-Degreea

5

0.771

54.363

1.858E+01

1.069E+01



Multistage Cancer,
3-Degree

5

0.961

51.847

2.778E+01

1.556E+01



a Best-fitting model, BMDS output presented in this appendix

F.2.24.2. Output for Selected Model: Multistage Cancer, 2-Degree
National Toxicology Program, 2006: Lung: Cystic keratinizing epithelioma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\24_msc2_lPerc_lung_epith.(d)

Gnuplot Plotting File: C:\Canc\24 msc2 IPerc lung epith.plt

Thu Apr~01 13:02:19 2010

0

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal*dose/Nl-beta2*dose/N2 ) ]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 6

Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta(1) =	0

Beta(2) = 3.77591e-005

This document is a draft for review purposes only and does not constitute Agency policy.

F-154 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Beta(2)

Beta(2)	1

Parameter Estimates

Variable
Background
Beta(1)
Beta(2)

Estimate

0
0

2.91011e-005

Std. Err.

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

Indicates that this value is not calculated.

Analysis of Deviance Table

Model
Full model
Fitted model
Reduced model

Log(likelihood)

-23.958
-26.1815
-40.2069

Param's Deviance Test d.f.

4.44693
32.4976

P-value

0. 487
0001

AIC:

54.363

Goodness of Fit

Scaled

Dose

Est

. Prob.

Expected

Observed

Size

Residual

0.0000

0.

0000

0

000

0

000

49

0

000

2.1400

0.

0001

0

006

0

000

48

-0

080

7.1400

0.

0015

0

068

0

000

46

-0

261

15.7000

0.

0071

0

350

0

000

49

-0

594

32.9000

0.

0310

1

519

0

000

49

-1

252

71.4000

0.

1379

7

170

9

000

52

0

736

Chi ^2

2 . 54

d.f.

P-value

0.7708

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	18.5839

BMDL =	10.687 8

BMDU =	25.1324

Taken together, (10.6878, 25.1324) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.000935646

This document is a draft for review purposes only and does not constitute Agency policy.

F-155 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.24.3. Figure for Selected Model: Multistage Cancer, 2-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	12:02 04/01 2010

3

4	National Toxicology Program, 2006: Lung: Cystic keratinizing epithelioma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-l56 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

F.2.25. Toth et al., 1979: Liver: Tumors
F.2.25.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

1

0.254

155.946

2.689E+00

1.522E+00



Multistage Cancer,
2-Degree

1

0.254

155.946

2.689E+00

1.522E+00

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.25.2. Output for Selected Model: Multistage Cancer, 1-Degree
Toth et al., 1979: Liver: Tumors

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Canc\25 mscl IPerc adr cor lyr.(d)

Gnuplot Plotting File: C:\Canc\25 mscl IPerc adr cor lyr.pit

Thu Apr 01_13:10:25 2010

Table 1

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = Mean
Independent variable = Dose

Total number of observations = 3

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.240176

Beta (1) = 0.00374745

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

This document is a draft for review purposes only and does not constitute Agency policy.

F-157 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

Background
Beta(1)

1	-0.53

-0.53	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.2418	*	*	*

Beta(1)	0.00373791	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-75.3127	3

Fitted model -75.9728 2	1.3201 1	0.2506

Reduced model -79.4897 1	8.35401 2	0.01534

AIC:	155.946

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.2418 9.188 7.000	38	-0.829

1.0000 0.2446 10.764 13.000	44	0.784

100.0000 0.4783 21.044 21.000	44	-0.013

Chi ^2 = 1.30	d.f. = 1	P-value = 0.2537

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	2.68876

BMDL =	1.52183

BMDU =	7.54263

Taken together, (1.52183, 7.54263) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00657103

This document is a draft for review purposes only and does not constitute Agency policy.

F-158 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.25.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	12:10 04/01 2010

3

4	Toth et al., 1979: Liver: Tumors

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-l59 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

F.2.26. Delia Porta et al., 1987: Table 4, B6C3 mice, male, hepatocellular carcinoma
F.2.26.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage Cancer,
1-Degree

1

0.073

164.110

9.255E+00

6.946E+00



Multistage
Cancer, 2-Degreea

1

0.899

160.823

7.359E+01

9.825E+00



a Best-fitting model, BMDS output presented in this appendix

F.2.26.2. Output for Selected Model: Multistage Cancer, 2-Degree
Delia Porta et al., 1987: Table 4, B6C3 mice, male, hepatocellular carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\1\94_DPorta_l987_Male_Hep_Carc_MultiCanc2_l.(d)

Gnuplot Plotting File: C:\l\94_DPorta_1987_Male_Hep_Carc_MultiCanc2_l.plt

Fri Apr 02 13:58:02 2010

Table 4, B6C3 mice, Male, Hepatocellular carcinoma

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal*dose/Nl-beta2*dose/N2 ) ]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 3

Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background =	0.110507

Beta(1) =	0

Beta(2) = 1.88069e-006

Asymptotic Correlation Matrix of Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

F-160 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

( *** The model parameter(s) -Beta(l)

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

Background	Beta(2)

Background	1	-0.62

Beta(2)	-0.62	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf.	Limit Upper Conf. Limit

Background 0.114031 *	*	*

Beta(1) 0 *	*	*

Beta(2) 1.8559e-006 *	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-78.4036	3

Fitted model -78.4116 2	0.0160146 1	0.8993

Reduced model -94.7394 1	32.6717 2	<.0001

AIC:	160.823

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.1140 4.903 5.000	43	0.046

357.1429 0.3008 15.340 15.000	51	-0.104

714.2857 0.6563 32.815 33.000	50	0.055

Chi/N2 = 0.02	d.f. = 1	P-value = 0.8994

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	73.5891

BMDL =	9.82517

BMDU =	88.9247

Taken together, (9.82517, 88.9247) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00101779

This document is a draft for review purposes only and does not constitute Agency policy.

F-161 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.26.3. Figure for Selected Model: Multistage Cancer, 2-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	12:58 04/02 2010

3

4	Delia Porta et al., 1987: Table 4, B6C3 mice, male, hepatocellular carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-162 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

F.2.27. Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular adenoma
F.2.27.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

1

0.468

99.355

3.695E+01

2.245E+01



Multistage Cancer,
2-Degree

0

NA

100.803

1.345E+02

2.353E+01



a Best-fitting model, BMDS output presented in this appendix

F.2.27.2. Output for Selected Model: Multistage Cancer, 1-Degree
Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular adenoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\1\95_DPorta_l987_Female_Hep_Aden_MultiCancl_l.(d)

Gnuplot Plotting File: C:\l\95_DPorta_1987_Female_Hep_Aden_MultiCancl_l.plt

Fri Apr 02 13:58:32 2010

Table 4, B6C3 mice, Female, Hepatocellular adenoma

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 3

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0244051
Beta(1) = 0.000306055

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

This document is a draft for review purposes only and does not constitute Agency policy.

F-163 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

Background
Beta(1)

1	-0.72

-0.72	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0369416	*	*	*

Beta(1)	0.000272012	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-47.4015	3

Fitted model -47.6775 2	0.552146 1	0.4574

Reduced model -51.6367 1	8.47042 2	0.01448

AIC:	9 9.3551

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0369 1.810 2.000	49	0.144

357.1429 0.1261 5.296 4.000	42	-0.602

714.2857 0.2070 9.936 11.000	48	0.379

Chi ^2 = 0.53	d.f. = 1	P-value = 0.4677

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	36.94 82

BMDL =	22.4 477

BMDU =	8 6.1826

Taken together, (22.4477, 86.1826) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.000445481

This document is a draft for review purposes only and does not constitute Agency policy.

F-164 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.27.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	12:58 04/02 2010

3

4	Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular adenoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-165 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

F.2.28. Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular carcinoma
F.2.28.1. Summary Table of BMDS Modeling Results

Model

Degrees of
Freedom

xV

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

Multistage
Cancer, 1-Degreea

1

0.010

116.588

2.425E+01

1.605E+01



Multistage Cancer,
2-Degree

1

0.010

116.588

2.425E+01

1.605E+01

final fi=0

a Best-fitting model, BMDS output presented in this appendix

F.2.28.2. Output for Selected Model: Multistage Cancer, 1-Degree
Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular carcinoma

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\1\96_DPorta_l987_Female_Hep_Carc_MultiCancl_l.(d)

Gnuplot Plotting File: C:\l\96_DPorta_1987_Female_Hep_Carc_MultiCancl_l.plt

Fri Apr 02 13:59:01 2010

Table 4, B6C3 mice, Female, Hepatocellular carcinoma

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(
-betal^dose^l)]

The parameter betas are restricted to be positive

Dependent variable = DichEff
Independent variable = Dose

Total number of observations = 3

Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

Background = 0.0903848
Beta(1) = 0.000261828

Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(1)

This document is a draft for review purposes only and does not constitute Agency policy.

F-166 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

Background
Beta(1)

1	-0.8

-0.8	1

Parameter Estimates

95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit

Background	0.0300271	*	*	*

Beta(1)	0.000414523	*	*	*

Indicates that this value is not calculated.

Analysis of Deviance	Table

Model Log(likelihood) # Param's	Deviance Test d.f. P-value

Full model	-53.1726	3

Fitted model -56.2941 2	6.24292 1	0.01247

Reduced model -60.7146 1	15.084 2	0.0005303

AIC:	116.588

Goodness of Fit

Scaled

Dose Est. Prob. Expected Observed	Size	Residual

0.0000 0.0300 1.471 1.000	49	-0.395

357.1429 0.1635 6.867 12.000	42	2.142

714.2857 0.2786 13.373 9.000	48	-1.408

Chi/N2 = 6.72	d.f. = 1	P-value = 0.0095

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	24.2455

BMDL =	16.0512

BMDU =	4 9.7176

Taken together, (16.0512, 49.7176) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.000623007

This document is a draft for review purposes only and does not constitute Agency policy.

F-167 DRAFT—DO NOT CITE OR QUOTE

-------
1 F.2.28.3. Figure for Selected Model: Multistage Cancer, 1-Degree

Multistage Cancer Model with 0.95 Confidence Level

dose

2	12:59 04/02 2010

3

4	Delia Porta et al., 1987: Table 4, B6C3 mice, female, hepatocellular carcinoma

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

F-168 DRAFT—DO NOT CITE OR QUOTE

-------
F.3. REFERENCES

1	Delia Porta G; Dragani TA; Sozzi D; Sozzi G. (1978) Carcinogenic effects of infantile and long-term 2,3,7,8-

2	tetrachlorodibenzo-p-dioxin treatment in the mouse. Tumori 73: 99-107.

3	Goodman, DG; Sauer, RM. (1992) Hepatotoxicity and carcinogenicity in female Sprague-Dawley rats treated with

4	2,3,7,8-tetrachlorordibenzo-p-dioxin (TCDD): a Pathology Working Group reevaluation. Regul Toxicol Pharmacol

5	15:245-252.

6	Kociba, RJ; Keyes, DG; Beyer, JE; et al. (1978) Results of a two-year chronic toxicity and oncogenicity study of

7	2,3,7,8-tetrachlorodibenzo-p-dioxin in rats. Toxicol Appl Pharmacol 46(2):279-303.

8	NTP (National Toxicology Program). (1982) Bioassay of 2,3,7,8-tetrachlorodibenzo-p-dioxin for possible

9	carcinogenicity (gavage study). Tech. Rept. Ser. No. 201. U.S. Department of Health and Human Services, Public

10	Health Service, Research Triangle Park, NC.

11	NTP (National Toxicology Program). (2006) NTP technical report on the toxicology and carcinogenesis studies of

12	2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) (CAS No. 1746-01-6) in female Harlan Sprague-Dawley rats (Gavage

13	Studies). Natl Toxicol ProgramTech Rep 521. Public Health Service, National Institute of Health, U.S. Department

14	of Health and Human Services, Research Triangle Park, NC.

15	Toth, KJ; Sugar, S; Somfai-Relle S; et al. (1978) Carcinogenic bioassay of the herbicide 2,4,5-trichlorphenoxy

16	ethanol (TCPE) with Swiss mice. Prog Biochem Pharmacol 14:82-93.

This document is a draft for review purposes only and does not constitute Agency policy.

F-169 draft—do not cite or quote

-------
[This page intentionally left blank.]

-------
DRAFT

DO NOT CITE OR QUOTE

May 2010
External Review Draft

APPENDIX G

Endpoints Excluded From Reference Dose
Derivation Based on Toxicological Relevance

NOTICE

THIS DOCUMENT IS AN EXTERNAL REVIEW DRAFT. It has not been formally released
by the U.S. Environmental Protection Agency and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its technical accuracy and policy
implications.

National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH

-------
1	APPENDIX G. ENDPOINTS EXCLUDED FROM REFERENCE DOSE DERIVATION

2	BASED ON TOXICOLOGICAL RELEVANCE

3

4

5	The National Academy of Sciences (NAS) committee commented on the low dose model

6	predictions and the need to discuss the biological significance of the noncancer health effects

7	modeled in the 2003 Reassessment. In selecting point of departure (POD) candidates from the

8	animal bioassays for derivation of the reference dose (RfD), U.S. Environmental Protection

9	Agency (EPA) had to consider the toxicological relevance of the identified endpoint(s) from any

10	given study. Often endpoints/effects may be sensitive, but lack general toxicological

11	significance due to not being clearly adverse (defined in the Integrated Risk Information System

12	(IRIS) glossary as a biochemical change, functional impairment, or pathologic lesion that affects

13	the performance of the whole organism, or reduces an organism's ability to respond to an

14	additional environmental challenge), being an adaptive response, or not being clearly linked to

15	downstream functional or pathological alterations. It is standard EPA RfD derivation policy not

16	to base a reference value on endpoints that are not adverse or not obvious precursors to an

17	adverse effect. For select studies, a rationale for lack of toxicological relevance of particular

18	endpoints reported is listed here. These endpoints were not considered for derivation of the RfD.

19	Kitchin and Woods (1979) administered female Sprague-Dawley rats a single gavage

20	dose of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and measured cytochrome P450 levels and

21	benzo(a)pyrene hydroxylase (BPH) activity as a marker of hepatic microsomal cytochrome

22	P448-mediated enzyme activity. They found a statistically significant increase in BPH at doses

23	>2 ng/kg and a significant increase in cytochrome P450 levels at doses >600 ng/kg. Aryl

24	hydrocarbon hydrolase and EROD were both significantly increased 3 months after exposure;

25	however the elevation did not maintain statistical significance at 6 months. No other indicators

26	of hepatic effects were analyzed. CYP induction alone is not considered a significant

27	toxicologically adverse effect given that CYPs are induced as a means of hepatic processing of

28	xenobiotic agents. Additionally, the role of CYP induction in hepatotoxicity and carcinogenicity

29	of TCDD is unknown, and CYP induction is not considered a relevant POD without obvious

30	pathological significance.

31	In multiple studies by Hassoun et al. (1998, 2000, 2002, 2003), various indicators of

32	oxidative stress were measured in hepatic and brain tissue of female B6C3F1 mice and Sprague-

This document is a draft for review purposes only and does not constitute Agency policy.

G-l	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

Dawley rats following 13 or 30 weeks of TCDD gavage dosing (5 days a week). Biomarkers for
oxidative stress included production superoxide anion, lipid peroxidation, and DNA single-strand
breaks. The authors report a statistically significant effect on several oxidative stress markers as
a result of TCDD exposure, the lowest dose producing an effect being 0.32 ng/kg-day (Hassoun
et al., 1998). In this study, all oxidative stress markers were significantly effected, but no other
indicators of brain pathology were assessed. Thus, it is impracticable to link the markers of
oxidative stress to a toxicological outcome in the brain, and this study and its endpoints are not
considered relevant POD candidates.

Burleson et al. (1996) analyzed the effect of a TCDD on viral host resistance following a
single gavage dose of TCDD by measuring mortality mediated by influenza virus challenge in
B6C3F1 female mice. The study authors found that TCDD at >10 ng/kg-day increased
influenza-induced mortality. The experimental design calls for a 30% mortality in untreated
animals (15% was achieved); mortality, itself, is not a direct result of TCDD exposure. None of
the other immunologically-relevant measures were affected by TCDD treatment in this study,
and no other effects were reported. The interpretation of these results with respect to humans is
problematic. Furthermore, the findings were not reproduced by Nohara et al. (2002) using the
same experimental design (see Section 2.4.2). Therefore, this endpoint is not considered relevant
as a POD candidate.

To examine the central nervous system response to TCDD, Kuchiiwa et al (2002)
analyzed the effect of in utero and lactational TCDD exposure on the serotonergic system in the
brainstem of male ddY mice. Female mice were administered TCDD by oral gavage once a
week for 8 weeks prior to pregnancy and, using an immunocytochemical detection method, the
raphe nuclei in the brainstem of male offspring was monitored for serotogergic neurons. TCDD
at 0.7 ng/kg-day caused a 25—50% reduction in the immunostaining of serotonin, however there
were no differences in external morphology, birth or postnatal body weights between
TCDD-exposed and control offspring. The authors suggest that these findings may indicate that
TCDD acts as a neuroteratogen by mediating long-term alterations in neuronal serotonin
synthesis and serotonergic function. However, no other relevant neurotoxicity endpoints were
examined or reported. Thus, reduced serotonin is not an adverse endpoint of toxicological
significance in and of itself, and this study is deemed unsuitable as a POD candidate.

This document is a draft for review purposes only and does not constitute Agency policy.

G-2 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

Mally and Chipman (2002) evaluated the effect of TCDD on gap junctions,
hypothesizing that as a nongenotoxic carcinogen, TCDD may induce tumor formation by
disturbing tissue homeostasis. Female F344 rats were dosed with TCDD by oral gavage for
either 3 consecutive days or 2 days a week for 28 days. Gap junction connexin (Cx) plaque
expression and hepatocyte proliferation was measured. The study authors report a decrease in
Cx32 plaque number and area in the liver of rats exposed to 0.7 ng/kg-day and higher, however
they did not find an associated increase in hepatocyte proliferation. No clinical signs of toxicity
were observed, and histological examination of the liver revealed no abnormalities. In the
absence of additional indicators of hepatotoxicity, a decrease in Cx32 plaque formation is not
clearly linked to TCDD-mediated hepatotoxicity or hepatocarcinogenicty, nor is it considered an
adverse effect. This endpoint is not considered a toxicologically relevant POD.

Vanden Heuvel et al. (1994) analyzed changes in hepatic mRNA following a single
administration of TCDD to female Sprague-Dawley rats by oral gavage. Four days after
treatment, animals were sacrificed and livers were excised. Using reverse transcriptase-
polymerase chain reaction (RT-PCR) on hepatic RNA, they compared levels of "dioxin
responsive" mRNA's (CYP1A1, UDP-glucuronosyltransferase I, plasminogen activator inhibitor
2, and transforming growth factor a) at various doses of TCDD and at control (baseline) levels.
They determined that CYP1 Al elicited the most sensitive response to TCDD, with a statistically
significant increase (3-fold) in mRNA from rat livers exposed to 1 ng/kg-day TCDD. Induction
of CYP1 Al expression is not considered an adverse effect, as the role of CYP1 Al in
TCDD-mediated carcinogenicity is unsettled. Therefore, in the absence of other indicators of
hepatoxicity, increases in liver CYP1 Al cannot be considered toxicologically relevant for a POD
candidate.

Devito et al. (1994) assessed the activity of CYP1A1 and CYP1A2, the amount of
phosphorylation of phosphotyrosyl proteins (pp32, pp34, and pp38), and the levels of estrogen
receptor in the liver, uterus, lung and skin tissue of female B6C3F1 mice administered TCDD for
5 days a week for 13 weeks. The authors hypothesized that these measurements may be
sensitive biomarkers for exposure to TCDD. Body weights were also recorded weekly.

Induction of CY1A1 and CYP1A2, as well as increased phosphorylated forms of pp32, pp34,
and pp38 were sensitive indicators of TCDD exposure, with statistically significant changes seen

at 1.07 ng/kg-day. EROD activity in the ling, skin, and liver was also observed with significant

This document is a draft for review purposes only and does not constitute Agency policy.

G-3 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

increases at this dose. However, the authors did not find a change in rat body or terminal organ
weights, nor did they note any pathology in the animals at this dose level. The role of CYPs and
phosphorylated pp32, pp34, and pp38 in TCDD-mediated toxicity is unknown, and changes in
the activity or function of these proteins are not considered adverse. Therefore, these endpoints
are not considered suitable as PODs.

Because TCDD had been detected in the soil of contaminated locations, determining the
bioavailability of TCDD from ingested soil may be important to the calculation of safe exposure
levels. Lucier et al. (1986) fed adult female Sprague-Dawley rats TCDD contaminated soil or
gave them TCDD in corn oil at various doses and compared the effects of TCDD on biochemical
parameters from liver tissue. They found that equivalent doses of TCDD in corn oil and soil
produced similar increases in hepatic aryl hydrocarbon hydroxylase activity (AHH) and UDP
glucuronyltransferase activity. They determined that AHH was statistically induced 1.8-fold at
15 ng/kg in corn oil and 40 ng/kg in soil. Cytochrome P450 was significantly increased at higher
doses. No clinical signs of acute toxicity or changes in body weight were observed. The
association between AHH activity and TCDD-mediated hepatotoxicity is unknown and no
adverse endpoints were measured. Thus, this endpoint is not suitable as a POD candidate.

Sugita-Konishi et al. (2003) investigated the change in host resistance of mice offspring
lactationally exposed to TCDD. Pregnant C57BL/6NCji mice were administered TCDD via
drinking water from parturition to weaning of the offspring (17 days). One group of offspring
was then infected with Listeria monocytogenes and blood and spleen samples were collected
various time points post infection. Uninfected, TCDD exposed offspring were weighed and their
spleens and thymuses removed for assay of cellular content and protein expression. TCDD
exposure caused a statistically-significant decrease in relative spleen weight and a statistically-
significant increase in thymic CD4+ cells in the high-dose group (11.3 ng/kg-day). Offspring
infected with Listeria following TCDD exposure exhibited a statistically significant increase in
serum tumor necrosis factor alpha (TNF-a) 2 days after infection in both sexes in the low-
(1.14 ng/kg-day) and high-dose groups. The authors conclude that exposure to TCDD disrupted
the host resistance of the offspring at the lowest dose tested, despite the primary immune
parameters being unaffected. Without an obvious association between TCDD and immune
function, however, this endpoint is not suitable for identification of a LOAEL. Thus, the

LOAEL for this study is 11.3 ng/kg-day, and the NOAEL is 1.14 ng/kg-day.

This document is a draft for review purposes only and does not constitute Agency policy.

G-4 DRAFT—DO NOT CITE OR QUOTE

-------
1	Sewall et al. (1993) investigated alterations in the epidermal growth factor receptor

2	(EGFR) pathway in a two-stage initiation promotion model of TCDD hepatic cancer. EGFR

3	signaling has been implicated in the altered cell growth induction by tumor promoters. Female

4	Sprague-Dawley rats were administered TCDD biweekly by oral gavage for 30 weeks following

5	initiation by a single dose of diethylnitrosamine (DEN). A group also received TCDD without

6	prior DEN initiation. Livers were harvested and fixed from sacrificed animals and sections

7	tested for EGFR binding, autophosphorylation, immunolocalization, and hepatic cell

8	proliferation. The authors report a significant dose-dependent decrease in plasma membrane

9	EGFR maximum binding capacity in TCDD-exposed rats beginning at 3.5 ng/kg-day. However,

10	at this same dose, the authors note a statistically significant decrease in cell proliferation (as

11	measured by DNA replication labeling), with increases in proliferation only occurring at higher

12	doses (125 ng/kg-day). No other indicators of hepatic toxicity or tumorigenicity were assessed.

13	The role of EGFR in TCDD-mediated hepatotoxicity and hepatocarcinogenicity is unknown, and

14	as such, this endpoint cannot be unequivocally linked to TCDD-induced hepatic effects nor

15	labeled as adverse. Thus, it is not suitable as a POD candidate.

16

17	G.l. REFERENCES

18	Burleson, GR; Lebrec, H; Yang, YG; et al. (1996) Effect of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) on

19	influenza virus host resistance in mice. Fund Appl Toxicol 29:40-47.

20	Devito, MJ; Ma, X; Babish, JG; et al. (1994) Dose-response relationships in mice following subchronic exposure to

21	2,3,7,8-tetrachlorodibenzo-p-dioxin: CYP1A1, CYP1A2, estrogen receptor, and protein tyrosine phosphoylation.

22	Toxicol Appl Pharmacol 124:82-90.

23	Hassoun, EA; Wilt, SC; DeVito, MJ; et al. (1998) Induction of oxidative stress in brain tissues of mice after

24	subchronic exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Toxicol Sci 42:23-27.

25	Hassoun, EA; Li, F; Abushaban, A; et al. (2000) The relative abilities of TCDD and its congeners to induce

26	oxidative stress in the hepatic and brain tissues of rats after subchronic exposure. Toxicology 145:103-113.

27	Hassoun, EA; Wang, H; Abushaban, A. (2002) Induction of oxidative stress following chronic exposure to TCDD,

28	2,3,4,7,8-pentachlorodibenzofuran, and 2,3 ',4,4',5-pentachlorobiphenyl. J Toxicol Environ Health A 65:825-842.

29	Hassoun, EA; Al-Ghafri, M; Abushaban, A. (2003) The role of antioxidant enzymes in TCDD-induced oxidative

30	stress in various brain regions of rats after subchronic exposure. Free Rad Biol Medicine 35(9): 1028-1036.

31	Kitchin, KT; Woods, JS. (1979) 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) effects on hepatic microsomal

32	cytochrome P-448-mediated enzyme activities. Toxicol Appl Pharmacol 47:537-546.

This document is a draft for review purposes only and does not constitute Agency policy.

G-5 DRAFT—DO NOT CITE OR QUOTE

-------
1	Kuchiiwa, S; Cheng, S-B; Nagatomo, I; et al. (2002) In utero and lactational exposure to 2,3,7,8-tetrachlorodibenso-

2	/j-dioxin decreases serotonin-immunoreactive neurons in raphe nuclei of male mouse offspring. Neurosci Lett

3	317:73-76.

4	Lucier, GW; Rumbaugh, RC; McCoy, Z; et al. (1986) Ingestion of soil contaminated with 2,3,7,8-tetrachloro-

5	dibenzo-p-dioxin (TCDD) alters hepatic enzyme activities in rats. Fund Appl Toxicol 6:364-371.

6	Mally, A; Chipman, JK. (2002) Non-genotoxic carcinogens: early effects on gap junctions, cell proliferation and

7	apoptosis in the rat. Toxicology 180:233-248.

8	Nohara, K; Izumi, H; Tamura, S; et al. (2002) Effect of low-dose 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) on

9	influenza A virus-induced mortality in mice. Toxicology 170:131-138.

10	Sewall, CH; Lucier, GW; Tritscher, AM; et al. (1993) TCDD-mediated changes in hepatic epidermal growth factor

11	receptor may be a critical event in the hepatocarcinogenic action of TCDD. Carcinogenesis 14:1885-1893.

12	Sugita-Konishi, Y; Kobayashi, K; Naito, H; et al. (2003) Effect of lactational exposure to 2,3,7,8-

13	tetrachlorodibenzo-p-dioxin on the susceptibility to Listeria infection. Biosci Biotechnol Biochem 67(l):89-93.

14	U.S. EPA. (2003) Exposure and human health reassessment of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and

15	related compounds [NAS review draft]. Volumes 1-3. National Center for Environmental Assessment, Washington,

16	DC; EPA/600/P-00/001 Cb. Available at: http://www.epa.gov/nceawwwl/pdfs/dioxin/nas-review/.

17	Vanden Heuvel, JP; Clark, GC; Tritscher, A; et al. (1994) Accumulation of polychlorinated dibenzo-p-dioxins and

18	dibenzofurans in liver of control laboratory rats. Fundam Appl Toxicol 23:465-469.

19

This document is a draft for review purposes only and does not constitute Agency policy.

G-6 DRAFT—DO NOT CITE OR QUOTE

-------
[This page intentionally left blank.]

-------
DRAFT

DO NOT CITE OR QUOTE

May 2010
External Review Draft

APPENDIX H
Cancer Precursor Benchmark Dose Modeling

NOTICE

THIS DOCUMENT IS AN EXTERNAL REVIEW DRAFT. It has not been formally released
by the U.S. Environmental Protection Agency and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its technical accuracy and policy
implications.

National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH

-------
CONTENTS—APPENDIX H: Cancer Precursor Benchmark Dose Modeling

APPENDIX H. CANCER PRECURSOR BENCHMARK DOSE MODELING	H-1

II I. BMDS INPUT TABLES	II-l

H. 1.1. Hassoun et al. (2000)	II-l

11.1.2.	Kitchin and Woods (1979)	II-l

H. 1.3. National Toxicology Program (2006), 31 Week Exposure	H-2

H. 1.4. National Toxicology Program (2006), 53 Week Exposure	H-2

H. 1.5. Vanden Heuvel et al. (1994)	H-2

II.2. ALTERNATE DOSE: WHOLE BLOOD BMDS RESULTS	11-3

H.2.1. Hassoun et al., 2000: Cytochrome C Reductase	H-3

H.2.1.1. Summary Table of BMDS Modeling Results	H-3

H.2.1.2. Output for Selected Model: Exponential (M5)	H-3

H.2.1.3. Figure for Selected Model: Exponential (M5)	H-6

H.2.2. Hassoun et al., 2000: DNA Single-Strand Breaks	H-7

H.2.2.1. Summary Table of BMDS Modeling Results	H-7

H.2.2.2. Output for Selected Model: Exponential (M4)	H-7

H.2.2.3. Figure for Selected Model: Exponential (M4)	H-10

H.2.2.4. Output for Additional Model Presented: Power, Unrestricted	H-l 1

H.2.2.5. Figure for Additional Model Presented: Power, Unrestricted	H-l3

11.2.3.	Hassoun et al., 2000: TBARS	11-14

H.2.3.1. Summary Table of BMDS Modeling Results	H-14

H.2.3.2. Output for Selected Model: Hill	H-14

H.2.3.3. Figure for Selected Model: Hill	H-17

H.2.4. Kitchin and Woods, 1979: Bap Hydroxylase Activity	H-18

H.2.4.1. Summary Table of BMDS Modeling Results	H-18

H.2.4.2. Output for Selected Model: Exponential (M5)	H-18

H.2.4.3. Figure for Selected Model: Exponential (M5)	H-21

H.2.5. National Toxicology Program, 2006: Liver EROD 53 Weeks	H-22

H.2.5.1. Summary Table of BMDS Modeling Results	H-22

H.2.5.2. Output for Selected Model: Hill	H-22

H.2.5.3. Figure for Selected Model: Hill	H-25

H.2.6. National Toxicology Program, 2006: Lung Erod 53 Weeks	H-26

H.2.6.1. Summary Table of BMDS Modeling Results	H-26

H.2.6.2. Output for Selected Model: Exponential (M4)	H-26

H.2.6.3. Figure for Selected Model: Exponential (M4)	H-29

H.2.6.4. Output for Additional Model Presented: Power, Unrestricted	H-30

H.2.6.5. Figure for Additional Model Presented: Power, Unrestricted	H-32

H.2.7. National Toxicology Program, 2006: Labeling Index 31 Weeks	H-33

H.2.7.1. Summary Table of BMDS Modeling Results	H-33

H.2.7.2. Output for Selected Model: Polynomial, 5-degree	H-33

H.2.7.3. Figure for Selected Model: Polynomial, 5-degree	H-36

H.2.8. Vanden Heuvel et al., 1994: Hepatic CYP1 Al Mrna Expression	H-37

H.2.8.1. Summary Table of BMDS Modeling Results	H-37

H.2.8.2. Output for Selected Model: Hill	H-37

This document is a draft for review purposes only and does not constitute Agency policy.

H-ii DRAFT—DO NOT CITE OR QUOTE

-------
CONTENTS (continued)

H.2.8.3. Figure for Selected Model: Hill	H-40

11.3. ADMINISTERED DOSE BMDS RESULTS	11-41

H.3.1. Hassoun et al., 2000: Cytochrome C Reductase	H-41

H.3.1.1. Summary Table of BMDS Modeling Results	H-41

H.3.1.2. Output for Selected Model: Exponential (M4)	H-41

H.3.1.3. Figure for Selected Model: Exponential (M4)	H-44

H.3.1.4. Output for Additional Model Presented: Power, Unrestricted	H-44

H.3.1.5. Figure for Additional Model Presented: Power, Unrestricted	H-47

H.3.2. Hassoun et al., 2000: DNA Single-Strand Breaks	H-48

H.3.2.1. Summary Table of BMDS Modeling Results	H-48

H.3.2.2. Output for Selected Model: Hill	H-48

H.3.2.3. Figure for Selected Model: Hill	H-51

H.3.2.4. Output for Additional Model Presented: Hill, Unrestricted	H-51

H.3.2.5. Figure for Additional Model Presented: Hill, Unrestricted	H-54

H.3.3. Hassoun et al., 2000: TBARS	11-55

H.3.3.1. Summary Table of BMDS Modeling Results	H-55

H.3.3.2. Output for Selected Model: Exponential (M4)	H-55

H.3.3.3. Figure for Selected Model: Exponential (M4)	H-58

H.3.3.4. Output for Additional Model Presented: Power, Unrestricted	H-58

H.3.3.5. Figure for Additional Model Presented: Power, Unrestricted	H-61

H.3.4. Kitchin and Woods, 1979: Bap Hydroxylase Activity	H-62

H.3.4.1. Summary Table of BMDS Modeling Results	H-62

H.3.4.2. Output for Selected Model: Exponential (M5)	H-62

H.3.4.3. Figure for Selected Model: Exponential (M5)	H-65

H.3.5. National Toxicology Program, 2006: Liver EROD 53 Weeks	H-66

H.3.5.1. Summary Table of BMDS Modeling Results	H-66

H.3.5.2. Output for Selected Model: Hill	H-66

H.3.5.3. Figure for Selected Model: Hill	H-69

H.3.6. National Toxicology Program, 2006: Lung Erod 53 Weeks	H-70

H.3.6.1. Summary Table of BMDS Modeling Results	H-70

H.3.6.2. Output for Selected Model: Exponential (M4)	H-70

H.3.6.3. Figure for Selected Model: Exponential (M4)	H-73

H.3.6.4. Output for Additional Model Presented: Power, Unrestricted	H-73

H.3.6.5. Figure for Additional Model Presented: Power, Unrestricted	H-76

H.3.7. National Toxicology Program, 2006: Labeling Index 31 Weeks	H-77

H.3.7.1. Summary Table of BMDS Modeling Results	H-77

H.3.7.2. Output for Selected Model: Exponential (M2)	H-77

H.3.7.3. Figure for Selected Model: Exponential (M2)	H-80

H.3.8. Vanden Heuvel et al., 1994: Hepatic CYP1A1 Mrna Expression	H-81

H.3.8.1. Summary Table of BMDS Modeling Results	H-81

H.3.8.2. Output for Selected Model: Hill	H-81

H.3.8.3. Figure for Selected Model: Exponential (M5)	H-84

This document is a draft for review purposes only and does not constitute Agency policy.

H-iii DRAFT—DO NOT CITE OR QUOTE

-------
1	APPENDIX H. CANCER PRECURSOR BENCHMARK DOSE MODELING

2

3

4	H.l. BMDS INPUT TABLES

5	H.l.l. Hassoun et al. (2000)

Endpoint

Administered Dose (ng/kg-day)

0

3

10

22

46

100

Internal Dose (ng/kg blood) 11

0

1.94

4.61

8.15

14.01

25.34

n = 6

n = 6

n = 6

n = 6

n = 6

n = 6

Cytochrome C reductase d

0.15 ±0.07

0.18 ± 0.05b

0.19 ±0.06

0.27 ± 0.06c

0.39 ± 0.06c

0.44 ± 0.11c

DNA single-strand breaks f

7.41 ± 1.54

10.78 ± 1.25 b-c

13.6 ± 1.69c

15.3 ± 1.71c

20.4 ± 2.25c

23.5 ± 1.37c

TBARs e

1.47 ±0.29

1.55 ± 0.54b

2.15 ± 0.36c

2.28 ± 0.25c

2.62 ± 0.52c

2.29 ± 0.49c

aFrom the Emond PBPK model described in 3.3.
bLOEL for selected endpoint.

Statistically significant as compared to control (p < 0.05).
dValues are the mean ± SD. Data obtained from Table 1 in Hassoun et al. 2000.
"Values are the mean ± SD. Data obtained from Table 2 in Hassoun et al. 2000.
Values are the mean ± SD. Data obtained from Table 3 in Hassoun et al. 2000.

6

7

8	H.1.2. Kitchin and Woods (1979)

Endpoint

Administered Dose (ng/kg-day)

0

0.6

2

4

20

60

Internal Dose (ng/kg blood) a

0

0.06

0.20

0.38

1.61

4.15

n = 9

n = 4

n = 4

n = 4

n = 4

n = 4

BaP hydroxylase activity'
(continued on next line)

4.9 ±0.37

4.9 ± 0.59b

6.7 ± 0.70c'd

7.2 ± 0.90 d

8.3 ± 0.13 e

14 ± 2.5e

Endpoint

Administered Dose (ng/kg-day)

200

600

2000

5000

20,000



Internal Dose (ng/kg blood) a

11.59

30.26

90.90

218.02

863.18



n = 4

n = 4

n = 4

n = 4

n = 4



BaP hydroxylase activity'
(continued)

59 ± 3.4e

96 ± 23e

155 ± 8.2e

182 ± 13e

189 ±13e



aFrom the Emond PBPK model described in 3.3.

^NOEL for selected endpoint.

°LOEL for selected endpoint.

Statistically significant as compared to control (p < 0.05).

"Statistically significant as compared to control (p < 0.001).

Values are the mean ± SE. Data obtained from Table 3 in Kitchin and Woods 1979.

This document is a draft for review purposes only and does not constitute Agency policy.

H-1 DRAFT—DO NOT CITE OR QUOTE

-------
1 H.1.3. National Toxicology Program (2006), 31 Week Exposure



Administered Dose (ng/kg-day)



0

2.14

7.14

15.7

32.9

71.4



Internal Dose (ng/kg blood) a



0

2.33

5.32

9.21

15.66

28.13

Endpoint

n = 9

n = 10

n = 10

n = 10

n = 10

n = 10

Labeling Index ,week 31 c

0.33 ±0.0.06

0.85 ±0.21 b

0.96 ± 0.23 b

0.79 ± 0.15 b

1.33 ± 0.36b

3.85 ± 0.97b

aFrom the Emond PBPK model described in 3.3.

Statistically significant as compared to control (p < 0.05).

°Values are the mean ± SE. Data obtained from Table 11 in NTP 2006.

2

3

4	H.1.4. National Toxicology Program (2006), 53 Week Exposure

Endpoint

Administered Dose (ng/kg-day)

0

2.14

7.14

15.7

32.9

71.4

Internal Dose (ng/kg blood) a

0.00

2.46

5.53

9.54

16.18

29.04

n = 8

n = 8

n = 8

n = 8

n = 8

n = 8

Liver EROD, week 53 c

30.22 ± 1.59

569.38 ±
24.62 b

1280.00 ±
95.30b

1551.16 ±
112.36 b

1726.81 ±
107.58b

1871.47 ±
109.14 b

Lung EROD, week 53 c

3.01 ±0.56

27.15 ± 1.87b

42.85 ± 3.94b

36.57 ±4.59b

43.75 ±6.56b

43.71 ±2.24 b

aFrom the Emond PBPK model described in 3.3.

Statistically significant as compared to control (p < 0.01).

°Values are the mean ± SE. Data obtained from Table 12 in NTP 2006.

5

6	H.1.5. Vanden Heuvel et al. (1994)

Endpoint

Administered Dose (ng/kg-day)

0

0.1

1

10

100

1,000

10,000

Internal Dose (ng/kg blood) a

0.00

0.01

0.11

0.88

6.45

48.32

434.50

n = 13

n = 5

n= 12

n = 7

n = 7

n = 11

n = 5

Hepatic CYP1A1
mRNA Expressionc

5.4 ± 1.0

7.2 ±2.5

14.8±4.3b

12.8 ± 1.7b

536± 121b

18000 ±4590

b

36700 ± 9900

b

aFrom the Emond PBPK model described in 3.3.

Statistically significant as compared to control (p < 0.05).

°Values are the mean ± SE. Data obtained from Table 2 in vanden Heuvel 1994.

7

This document is a draft for review purposes only and does not constitute Agency policy.

H-2 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

H.2. ALTERNATE DOSE: WHOLE BLOOD BMDS RESULTS

H.2.1.	Hassoun et al., 2000: Cytochrome C Reductase

I.2.1.1.	Summary Table of BMDS Modeling Results	

Model3

Degrees
of

Freedom

x'/'-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

4

0.016

-143.333

9.274E+00

7.737E+00



exponential (M3)

4

0.016

-143.333

9.274E+00

7.737E+00

power hit bound (d = 1)

exponential (M4)

3

0.339

-150.139

3.364E+00

2.170E+00



exponential

(M5)b

2

0.788

-151.027

5.913E+00

3.102E+00



Hill

2

0.743

-150.910

6.208E+00

3.190E+00



linear

4

0.170

-149.086

5.613E+00

4.429E+00



polynomial, 5-
degree

4

0.170

-149.086

5.613E+00

4.429E+00



power

4

0.170

-149.086

5.613E+00

4.429E+00

power bound hit (power =1)

a Constant variance model selected (p = 0.3871)
b Best-fitting model, BMDS output presented in this appendix

H.2.1.2. Output for Selected Model: Exponential (M5)

Hassoun et al., 2000: Cytochrome C reductase

Exponential Model. (Version: 1.61;

Date: 7/24/2009)

Input Data File: C:\5\Blood\17 Has

2000 CytCLiv ExpCV 1.(d)

Gnuplot Plotting File:





Fri Apr 30 14:14:

TBARs, liver only (Table 2)

The form of the response function by Model:

Model 2:	Y[dose] = a * exp{sign * b * dose}

Model 3:	Y[dose] = a * exp{sign * (b * dosej^d}

Model 4:	Y[dose] = a * [c-(c-l) * exp{-b * dose}]

Model 5:	Y[dose] = a * [c-(c-l) * exp{-(b * dosej^d}]

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.

Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean

This document is a draft for review purposes only and does not constitute Agency policy.

H-3 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values

Variable

Model 5

lnalpha

rho(S)

-5.48625

0

0.1387
0.0225296
6.40231

1

(S)

Specified

Parameter Estimates

Variable

lnalpha
rho

Model 5

-5.47298
0

0.156024
0. 0891513
2.85355
2 .14235

Table of Stats From Input Data

Obs Mean

Obs Std Dev

0 6

0

146

0.06614

1.938 6

0

177

0.05389

4.614 6

0

191

0.05634

8.147 6

0

271

0.05634

14.01 6

0

388

0.06369

25.34 6

0

444

0.1102

Dose

Estimated Values of Interest
Est Mean	Est Std	Scaled Residual

0

0.156

0

0648

-0.3789

1. 938

0.1627

0

0648

0.5416

4 . 614

0.1961

0

0648

-0.1919

8 .147

0.2705

0

0648

0. 01769

14 . 01

0.3874

0

0648

0.02224

25.34

0.4443

0

0648

-0.0107

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

H-4 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	80.75258	7	-147.5052

A2	83.37355	12	-142.7471

A3	80.75258	7	-147.5052

R	55.82002	2	-107.64

5	80.51364	5	-151.0273

Additive constant for all log-likelihoods =	-33.08. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 7a: Does Model 5 fit the data? (A3 vs 5)

Test

Test 1
Test 2
Test 3
Test 7a

Tests of Interest

-2*log(Likelihood Ratio)

55.11
5.242
5.242
0.4779

10

5
5
2

p-value

< 0.0001
0.3871
0.3871
0.7875

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 7a is greater than .1. Model 5 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control

Confidence Level = 0.950000

BMD =	5.91298

BMDL =	3.10234

This document is a draft for review purposes only and does not constitute Agency policy.

H-5 DRAFT—DO NOT CITE OR QUOTE

-------
1 H.2.1.3. Figure for Selected Model: Exponential (M5)

Exponential Model 5 with 0.95 Confidence Level

dose

2	14:14 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-6 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

H.2.2.	Hassoun et al., 2000: DNA Single-Strand Breaks

I.2.2.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

4

<0.0001

111.134

6.551E+00

5.472E+00



exponential (M3)

4

<0.0001

111.134

6.551E+00

5.472E+00

power hit bound (d = 1)

exponential
(M4)b

3

0.231

78.588

1.207E+00

9.165E-01



exponential (M5)

3

0.231

78.588

1.207E+00

9.165E-01

power hit bound (d = 1)

Hill

3

0.230

78.590

1.097E+00

7.966E-01

n lower bound hit (n = 1)

linear

4

<0001

97.616

3.552E+00

2.890E+00



polynomial, 5-
degree

4

<0001

97.616

3.552E+00

2.890E+00



power

4

<0001

97.616

3.552E+00

2.890E+00

power bound hit (power =1)

power,
unrestricted0

3

0.132

79.893

4.522E-01

2.027E-01

unrestricted (power = 0.576)

a Constant variance model selected (p = 0.7521)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

H.2.2.2. Output for Selected Model: Exponential (M4)

Hassoun et al., 2000: DNA single-strand breaks

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\5\Blood\18_Has_20 0 0_SSB_ExpCV_l.(d)

Gnuplot Plotting File:

Fri Apr 30 14:15:16 2010

DNA single-strand breaks, liver only (Table 3)

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

H-7 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

Specified

lnalpha	0.841244

rho(S)	0

a	7.0395

b	0.103521

c	3.50522

d	1

Parameter Estimates
Variable	Model 4

lnalpha	0.960789

rho	0

a	7.7528

b	0.075429

c	3.39665

d	1

Table of Stats From Input Data
Dose	N	Obs Mean	Obs Std Dev

0	6	7.41	1.543

1.938	6	10.78	1.249

4.614	6	13.6	1.69

8.147	6	15.3	1.715

14.01	6	20.4	2.254

25.34	6	23.5	1.372

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	7.753	1.617	-0.5194

1.938	10.28	1.617	0.7575

4.614	13.21	1.617	0.5853

8.147	16.28	1.617	-1.49

14.01	19.87	1.617	0.7958

25.34	23.59	1.617	-0.1293

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

This document is a draft for review purposes only and does not constitute Agency policy.

H-8 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Var{e(ij)} = Sigma/N2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)} = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-33.14239	7	80.28478

A2	-31.81197	12	87.62394

A3	-33.14239	7	80.28478

R	-80.44209	2	164.8842

4	-35.29421	4	78.58842

Additive constant for all log-likelihoods =	-33.08. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

97 .26
2 . 661
2 . 661
4 . 304

10
5
5
3

p-value

< 0.0001
0.7521
0.7521
0.2305

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	1.20684

BMDL =	0.916526

This document is a draft for review purposes only and does not constitute Agency policy.

H-9 DRAFT—DO NOT CITE OR QUOTE

-------
1 H.2.2.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

2	14:15 04/30 2010

3

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-10 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

H.2.2.4. Output for Additional Model Presented: Power, Unrestricted

Hassoun et al., 2000: DNA single-strand breaks

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\5\Blood\18_Has_20 0 0_SSB_PwrCV_U_l.(d)

Gnuplot Plotting File: C:\5\Blood\18_Has_2000_SSB_PwrCV_U_l.plt

Fri Apr 30 14:15:20 2010

DNA single-strand breaks, liver only (Table 3)

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
alpha =	2.7831

rho =	0 Specified

control =	7.41

slope =	2.16848

power =	0.62 004 8

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha	control	slope	power

alpha	1	2.5e-009 -4.6e-009	5.7e-009

control	2.5e-009	1	-0.79	0.66

slope -4.6e-009	-0.79	1	-0.97

power	5.7e-009	0.66	-0.97	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable Estimate Std. Err.	Lower Conf. Limit	Upper Conf. Limit

alpha 2.71022 0.638804	1.45818	3.96225

control 7.26415 0.644159	6.00163	8.52668

slope 2.60017 0.530762	1.55989	3.64044

power 0.575946 0.0589669	0.460373	0.691519

Table of Data and Estimated Values of Interest

This document is a draft for review purposes only and does not constitute Agency policy.

H-l 1 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Dose

Obs Mean

Est Mean

Obs Std Dev Est Std Dev Scaled Res.

0	6	7.41	7.26

1.938	6	10.8	11.1

4.614	6	13.6	13.5

8.147	6	15.3	16

14.01	6	20.4	19.2

25.34	6	23.5	24

1.54	1.65	0.217

1.25	1.65	-0.432

1.69	1.65	0.094

1.71	1.65	-0.993

2.25	1.65	1.85

1.37	1.65	-0.735

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -33.142389	7	80.284779

A2 -31.811970	12	87.623940

A3 -33.142389	7	80.284779

fitted -35.946504	4	79.893008

R -80.442086	2	164.884172

Explanation of Tests

Test 1:

Test 2
Test 3
Test 4
(Note:

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

97 .2602
2.66084
2.66084
5.60823

10
5
5
3

<.0001
0.7521
0.7521
0.1323

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than
model appears to be appropriate here

.1. A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems

This document is a draft for review purposes only and does not constitute Agency policy.

H-12 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

to adequately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 0.452221

BMDL = 0.202688

H.2.2.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

14:15 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-13 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

H.2.3.	Hassoun et al., 2000: TEARS

I.2.3.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

4

0.001

-8.517

1.736E+01

1.223E+01



exponential (M3)

4

0.001

-8.517

1.736E+01

1.223E+01

power hit bound (d = 1)

exponential (M4)

3

0.188

-19.755

2.189E+00

1.151E+00



exponential (M5)

2

0.240

-19.681

3.470E+00

1.525E+00



Hill b

2

0.272

-19.935

3.292E+00

1.737E+00



linear

4

0.002

-9.793

1.444E+01

9.622E+00



polynomial, 5-
degree

4

0.002

-9.793

1.444E+01

9.622E+00



power

4

0.002

-9.793

1.444E+01

9.622E+00

power bound hit (power =1)

a Constant variance model selected (p = 0.3348)
b Best-fitting model, BMDS output presented in this appendix

H.2.3.2. Output for Selected Model: Hill

Hassoun et al., 2000: TBARS

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\5\Blood\19_Has_20 0 0_TBARsLiv_HillCV_l.(d)

Gnuplot Plotting File: C:\5\Blood\19_Has_2000_TBARsLiv_HillCV_l.plt

Fri Apr 30 14:16:02 2010

TBARs, liver only (Table 2)

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter restricted to be greater than 1
A constant variance model is fit

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

H-14 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Default Initial Parameter Values

alpha
rho

intercept

0.178788

0

1.469
1.15
1.2785
5.08547

Specified

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
intercept

alpha
1

2 . 8e-008
-4 . 4e-008
4 . 9e-008
-1. 5e-008

intercept
2 . 8e-008
1

-0. 82
0.48
0.52

v

-4 . 4e-008
-0. 82
1

-0. 61
-0.22

n

4 . 9e-008
0.48
-0. 61
1

0.29

k

-1. 5e-008
0.52
-0.22
0.29
1

Parameter Estimates

Variable
alpha
intercept

Estimate
0.16017
1.46138
0.963033
3.44642
3.63417

Std. Err.

0.0377523
0.152797
0.20228
2.43468
1.02019

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

0.0861764
1.1619
0.566571
-1. 32547
1. 63464

0.234163
1.76086
1. 3595
8 . 21832
5.6337

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 6

1.47

1.46

0.291

0.4

0.0466

938 6

1. 55

1. 56

0.539

0.4

-0.0 6 9 6

614 6

2 .15

2 .13

0.363

0.4

0.12

147 6

2 .28

2 . 37

0.247

0.4

-0.54

. 01 6

2 . 62

2 .42

0.517

0.4

1. 25

. 34 6

2.29

2 .42

0. 487

0.4

-0.803

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

H-15 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	16.269770	7	-18.539539

A2	19.127827	12	-14.255654

A3	16.269770	7	-18.539539

fitted	14.967391	5	-19.934782

R	2.442940	2	-0.885880

Explanation of Tests

Test

1

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adequately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio)

Test df

p-value

Test

Test
Test
Test

33.3698
5.71611
5.71611
2.60476

10

5
5
2

0.000236
0.3348
0.3348
0.2719

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1.
model appears to be appropriate here

A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	3.29185

BMDL =	1.73738

This document is a draft for review purposes only and does not constitute Agency policy.

H-16 DRAFT—DO NOT CITE OR QUOTE

-------
1 H.2.3.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

2	15:22 04/30 2010

3

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-17 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

H.2.4.	Kitchin and Woods, 1979: Bap Hydroxylase Activity

I.2.4.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r/'-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

9

<0.0001

452.100

2.960E+02

1.446E+02



exponential (M3)

9

<0.0001

452.100

2.960E+02

1.446E+02

power hit bound (d = 1)

exponential (M4)

8

0.002

232.110

3.182E-01

2.373E-01



exponential

(M5)b

7

0.015

227.004

9.321E-01

4.900E-01



Hill

8

<0001

479.250

5.340E+00

4.528E+00



linear

9

<0001

291.380

4.552E-01

3.303E-01



polynomial, 8-
degree

6

<0001

468.198

1.012E+03

7.899E-01



power

9

<0001

291.380

4.552E-01

3.303E-01

power bound hit (power =1)

1 Non-constant variance model selected (p = <0.0001)
' Best-fitting model, BMDS output presented in this appendix

H.2.4.2. Output for Selected Model: Exponential (M5)

Kitchin and Woods, 1979: BaP Hydroxylase Activity

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\5\Blood\2 7_Kitchin_l979_Hydrolase_Exp_l.(d)

Gnuplot Plotting File:

Fri Apr 30 14 :17:28 2010

Kitchin 1979, Tbl3, BaP hydrolase activity

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

This document is a draft for review purposes only and does not constitute Agency policy.

H-18 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 11

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 5

lnalpha	-3.27793

rho	1.92227

a	4.655

b	0.0041206

c	42.6316

d	1

Parameter Estimates
Variable	Model 5

lnalpha	-2.64071

rho	1.94 04 6

a	5.46248

b	0.0382278

c	30.9208

d	1.42906

Table of Stats From Input Data

Dose

N

Obs Mean

Obs St

0

9

4 . 9

1.11

0.0645

4

4 . 9

i—1

i—1

CO

0.2023

4

6.7

1 . 4

0.3839

4

7 . 2

i—1

CO

1. 613

4

CO
CO

0.26

4.146

4

14

5

11.59

4

59

CO

30.26

4

96

46

90. 9

4

155

16.4

218

4

182

26

863.2

4

189

26



Estimated

Values

of

Interest

Dose

Est Mean

Est

Std

Scaled Residi

0

5. 462

1.

387

-1.217

0.0645

5. 493

1.

394

-0.8507

0.2023

5. 619

1.

425

1. 516

0.3839

5. 854

1.

483

1. 815

1. 613

8.483

2 .

126

-0.1723

4.146

i—1

G\

CO

4 .

125

-1.358

11.59

49.32

11

.73

1. 65

30.26

121. 2

28

. 06

-1.796

90. 9

168 . 5

38

. 62

-0.6975

218

168 . 9

38

.72

0.6765

863.2

168 . 9

38

.72

1. 038

This document is a draft for review purposes only and does not constitute Agency policy.

H-19 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Other models for which	likelihoods are calculated:

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-158.1306	12	340.2613

A2	-84.80028	22	213.6006

A3	-98.82189	13	223.6438

R	-234.6252	2	473.2504

5	-107.5022	6	227.0044

Additive constant for all log-likelihoods =	-45.03. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 7a: Does Model 5 fit the data? (A3 vs 5)

Test

Test 1
Test 2
Test 3
Test 7a

Tests of Interest
-2*log(Likelihood Ratio)

299.6

146.7
28 . 04
17 . 36

D. F.

20
10
9
7

p-value

<	0.0001

<	0.0001
0.0009381

0.01521

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is less than .1. You may want to
consider a different variance model.

The p-value for Test 7a is less than .1. Model 5 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

This document is a draft for review purposes only and does not constitute Agency policy.

H-20 DRAFT—DO NOT CITE OR QUOTE

-------
BMD =	0.9321

BMDL =	0.490004

7 H.2.4.3. Figure for Selected Model: Exponential (M5)

CD
CO
c
o

Q.

CO
CD

a:

c
m

CD

250

200

150

100

Exponential Model 5 with 0.95 Confidence Level

Exponential

S/IDLBMD

0	100 200 300

400 500
dose

600 700 800 900

14:17 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-21 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

H.2.5.	National Toxicology Program, 2006: Liver EROD 53 Weeks

I.2.5.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

4

<0.0001

648.094

2.011E+01

1.464E+01



exponential (M3)

4

<0.0001

648.094

2.011E+01

1.464E+01

power hit bound (d = 1)

exponential (M4)

3

0.015

521.251

1.430E-02

9.808E-03



exponential (M5)

2

0.354

514.812

7.656E-02

3.202E-02



Hill b

2

0.760

513.286

1.853E-01

9.351E-02



linear

4

<0001

639.841

1.034E+01

6.557E-03



polynomial, 5-
degree

1

<0001

14.000

error

error



power

4

<0001

592.889

2.254E-02

1.527E-02

power bound hit (power = 1)

1 Non-constant variance model selected (p = <.0001)
' Best-fitting model, BMDS output presented in this appendix

H.2.5.2. Output for Selected Model: Hill

National Toxicology Program, 2006: Liver EROD 53 Weeks

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\5\Blood\46_NTP_20 0 6_ERODliv53_Hill_l.(d)

Gnuplot Plotting File: C:\5\Blood\4 6_NTP_2006_ERODliv53_Hill_l.plt

Sun May 02 15:34:21 2010

0

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose

Power parameter restricted to be greater than 1

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

This document is a draft for review purposes only and does not constitute Agency policy.

H-22 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

lalpha
rho

intercept

11. 0197

0

30.215
1841.26
7 . 0105
6.95814

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho

intercept

lalpha
1

-0. 97
-0.18
0. 065
-0.025
0. 046

rho

-0. 97
1

0.17
-0.093
0. 025
-0.048

intercept

-0.18
0.17
1

-0.022
0. 011
0.00084

v

0. 065
-0.093
-0.022
1

-0.73
0. 87

n

-0.025
0. 025
0. 011
-0.73
1

-0. 83

k

0. 046
-0.048
0. 00084
0. 87
-0. 83
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

Estimate
-4 .47504
2.12799
30.2685
1813.88
2 . 02516
3.78554

Std. Err.

0. 923978
0.137849
1.41935
100.554
0.29717
0.349266

Lower Conf. Limit

-6.286
1. 85781
27.4866
1616.8
1.44272
3.101

Upper Conf. Limit

-2.66407
2 .39817
33.0504
2010.96
2 . 6076
4 .47009

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

2 .458
5.533
9.543
16.18
29. 04

30.2

569

1. 28e + 003
1. 55e + 003
1. 73e + 003
1. 87e + 003

30.3

564

1. 27e + 003
1. 6e + 003
1. 75e + 003
1. 82e + 003

4 . 5

6 9.6
270
318
304
309

4 . 02
90.3
214
274
302
313

-0.0377
0.17
0.137
-0.529
-0.248
0.507

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Model A2:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

Mu(i) + e(ij ;

Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi

Var{e(i)}

Mu + e(i)
Sigma/N2

This document is a draft for review purposes only and does not constitute Agency policy.

H-23 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-285.269096	7	584.538193

A2	-249.237836	12	522.475671

A3	-250.368300	8	516.736600

fitted	-250.643212	6	513.286424

R	-338.451300	2	680.902600

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adequately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	178.427	10	<.0001

Test 2	72.0625	5	<.0001

Test 3	2.26093	4	0.6879

Test 4	0.549824	2	0.7596

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.185269

BMDL =	0.0935065

This document is a draft for review purposes only and does not constitute Agency policy.

H-24 DRAFT—DO NOT CITE OR QUOTE

-------
1 H.2.5.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

This document is a draft for review purposes only and does not constitute Agency policy.

H-25 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

H.2.6.	National Toxicology Program, 2006: Lung Erod 53 Weeks

I.2.6.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2p-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

4

<0.0001

314.332

3.281E+01

2.047E+01



exponential (M3)

4

<0.0001

555.061

5.210E+00

8.194E-01

power hit bound (d = 1)

exponential
(M4)b

3

0.302

255.955

9.586E-02

5.907E-02



exponential (M5)

2

0.276

256.882

1.044E+00

6.588E-02



Hill

2

0.275

256.882

1.903E+00

3.469E-01



linear

4

<0001

313.237

2.662E+01

1.251E+01



polynomial, 5-
degree

5

<0001

330.180

error

2.718E+01



power

4

<0001

313.237

2.662E+01

1.251E+01

power bound hit (power =1)

power,
unrestricted0

3

0.032

261.083

1.875E-07

1.875E-07

unrestricted (power =0.18)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

H.2.6.2. Output for Selected Model: Exponential (M4)

National Toxicology Program, 2006: Lung EROD 53 Weeks

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\5\Blood\52_NTP_20 0 6_LungEROD53_Exp_l.(d)

Gnuplot Plotting File:

Fri Apr 30 14:20:27 2010

Tbl 12, Week 53, Lung Microsomes EROD

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

H-26 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values

Variable

lnalpha
rho

Model 4

-0.80064
1.47683
2 .86045
0.134268
16.0581
1

Parameter Estimates
Variable	Model 4

lnalpha	-1.14455

rho	1.63458

a	3.06102

b	0.371249

c	14.1551

d	1

Table of Stats From Input Data

Dose	N	Obs Mean	Obs Std Dev

0	8	3.011	1.584

2.458	8	27.15	5.269

5.533	8	42.85	11.15

9.543	8	36.57	12.99

16.18	8	43.75	18.55

29.04	8	43.71	6.322

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0	3.061	1.408	-0.1005

2.458	27.16	8.383	-0.003073

5.533	38.17	11.07	1.196

9.543	42.16	12.01	-1.318

16.18	43.23	12.26	0.1191

29.04	43.33	12.28	0.08864

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

H-27 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-135.2677	7	284.5353

A2	-115.6885	12	255.3771

A3	-121.1517	8	258.3034

R	-162.0902	2	328.1805

4	-122.9773	5	255.9546

-44.11. This constant added to the

Additive constant for all log-likelihoods =
above values gives the log-likelihood including the term that does not
depend on the model parameters.

Test

1:

Test

2 :

Test

3:

Test

6a

Explanation of Tests

Does response and/or variances differ among Dose levels? (A2 vs. R)

Are Variances Homogeneous? (A2 vs. A1)

Are variances adeguately modeled? (A2 vs. A3)

Does Model 4 fit the data? (A3 vs 4)

Tests of Interest

Test

-2*log(Likelihood Ratio)

D. F.

p-value

Test 1

92 . 8

10

< 0.0001

Test 2

39.16

5

< 0.0001

Test 3

10. 93

4

0.0274

Test 6a

3. 651

3

0.3017

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is less than .1. You may want to
consider a different variance model.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control

Confidence Level = 0.950000

BMD =	0.09586

BMDL = 0.0590734

This document is a draft for review purposes only and does not constitute Agency policy.

H-28 DRAFT—DO NOT CITE OR QUOTE

-------
H.2.6.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

14:20 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-29 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

H.2.6.4. Output for Additional Model Presented: Power, Unrestricted

National Toxicology Program, 2006: Lung EROD 53 Weeks

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\5\Blood\52_NTP_20 0 6_LungEROD53_Pwr_U_l.(d)

Gnuplot Plotting File: C:\5\Blood\52_NTP_2006_LungEROD53_Pwr_U_l.plt

Fri_Apr 30 14:20:33 2010

Tbl 12, Week 53, Lung Microsomes EROD

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
lalpha =	4.76968

rho =	0

control =	3.011

slope =	23.2411

power =	0.187468

Asymptotic Correlation Matrix of Parameter Estimates

lalpha	rho	control	slope	power

lalpha 1	-0.96	-0.49	0.1	-0.045

rho -0.96	1	0.45	-0.13	0.05

control -0.49	0.45	1	-0.14	0.048

slope 0.1	-0.13	-0.14	1	-0.94

power -0.045	0.05	0.048	-0.94	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho
control
slope
power

Estimate

-1. 02668
1. 63033
3. 01543
23.8167
0.179731

Std. Err.

0. 818488
0.24056
0.519355
3.70401
0.0639681

Lower Conf. Limit

-2.63088
1.15884
1. 99751
16.5569
0.054356

Upper Conf. Limit
0.577531
2 .10182
4.03335
31.0764
0.305106

Table of Data and Estimated Values of Interest
Dose	N Obs Mean	Est Mean Obs Std Dev Est Std Dev Scaled Res.

This document is a draft for review purposes only and does not constitute Agency policy.

H-30 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

CO

o

3. 01

3. 02

i—1

en

CO

1 .47

-0.00851

2.458 8

27 .1

31

5.27

9.84

-1 . 11

5.533 8

42 . 8

35. 4

11 . 2

11

1 . 92

9.543 8

3 6.6

38 . 7

13

i—1
i—1

CO

-0.52

16.18 8

43.7

42 . 3

i—1

CO

CJ1

12 . 7

0.323

29.04 8

43.7

46.6

6.32

13.7

-0.605

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-135.267662	7	284.535325

A2	-115.688533	12	255.377067

A3	-121.151707	8	258.303413

fitted	-125.541690	5	261.083380

R	-162.090242	2	328.180484

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

92 .8034
39.1583
10.9263
8 . 77 997

10
5
4
3

<.0001
<.0001
0.0274
0.03236

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is less than .1. You may want to try a different
model

This document is a draft for review purposes only and does not constitute Agency policy.

H-31 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 1.87 4 5e-007

BMDL = 1. 87 4 5e-007

H.2.6.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

14:20 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-32 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

H.2.7.	National Toxicology Program, 2006: Labeling Index 31 Weeks

I.2.7.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r/'-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

4

0.000

46.547

8.660E+00

6.926E+00



exponential (M3)

4

0.000

46.547

8.660E+00

6.926E+00

power hit bound (d = 1)

exponential (M4)

3

<0.0001

50.958

3.151E+00

1.865E+00



exponential (M5)

3

<0.0001

50.958

3.151E+00

1.864E+00

power hit bound (d = 1)

Hill

3

<0001

50.963

3.145E+00

error

n lower bound hit (n = 1)

linear

4

0.000

48.958

3.151E+00

1.865E+00



polynomial, 5-
degree b

3

0.000

46.230

7.607E+00

3.125E+00



power

4

0.000

48.958

3.151E+00

1.865E+00

power bound hit (power =1)

1 Non-constant variance model selected (p = <.0001)
' Best-fitting model, BMDS output presented in this appendix

H.2.7.2. Output for Selected Model: Polynomial, 5-degree

National Toxicology Program, 2006: Labeling Index 31 Weeks

Polynomial Model. (Version: 2.13; Date: 04/08/2008)

Input Data File: C:\5\Blood\3 8_NTP_20 0 6_HepIndex_Poly5_l.(d)

Gnuplot Plotting File: C:\5\Blood\38_NTP_2006_HepIndex_Poly5_l.plt

Fri_Apr 30 14:21:16 2010

Tbl 11, 31wk, Hep Cell Proliferation Labeling Index

The form of the response function is:

Y[dose] = beta_0 + beta_l*dose + beta_2*dose/N2 + ...

Dependent variable = Mean
Independent variable = Dose

The polynomial coefficients are restricted to be positive

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

This document is a draft for review purposes only and does not constitute Agency policy.

H-33 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

lalpha =	0.708431

rho =	0

beta_0 =	0.327

beta_l =	0

beta_2 =	0

beta_3 =	0

beta_4 =	0

beta 5 =	0

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -beta 2 -beta 3 -beta 4

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

lalpha
rho
beta_0
beta_l
beta 5

lalpha
1

-0.086
0. 012
-0.032
0. 043

rho

-0.086
1

-0.0027
-0.011
0. 076

beta_0

0. 012
-0.0027
1

-0.6
0.23

beta_l
-0.032
-0.011
-0.6
1

-0.53

beta 5

0. 043
0. 076
0.23
-0.53
1

Parameter Estimates

Variable
lalpha
rho

beta_0
beta_l
beta_2
beta_3
beta_4
beta 5

Estimate
-0.501559
1.90452
0.500197
0.0525247
8.00068e-025
0
0

1.08658e-007

Std. Err.
0.185039
0.272948
0.102837
0.0192967
NA
NA
NA

. 10451e-008

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-0.864229
1.36955
0.298641
0.0147038

-1. 0987 9e-008

-0.138889
2 .43948
0.701753
0. 0903456

2 . 28305e-007

NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

2 . 331
5.315
9.207
15. 66
28 .13

9
10
10
10
10
10

0.327
0. 852
0. 956
0.792
1. 33
3. 85

0.5

0. 623
0.78
0. 991
1.42
3.89

0.189

0. 651
0 .737
0. 462
1.12
3. 08

0. 402
0. 496
0. 614
0 .772
1.09
2 .84

-1.29

1.46
0. 907
-0.816
-0.266
-0.0523

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

H-34 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-47.234977	7	108.469953

A2	-8.679256	12	41.358512

A3	-8.980651	8	33.961301

fitted	-18.115050	5	46.230101

R	-63.448285	2	130.896571

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.'

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	109.538	10	<.0001

Test 2	77.1114	5	<.0001

Test 3	0.60279	4	0.9628

Test 4	18.2688	3	0.0003871

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	7.6073

BMDL =	3.12526

This document is a draft for review purposes only and does not constitute Agency policy.

H-3 5 DRAFT—DO NOT CITE OR QUOTE

-------
H.2.7.3. Figure for Selected Model: Polynomial, 5-degree

Polynomial Model with 0.95 Confidence Level

dose

14:21 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-36 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

H.2.8.	Vanden Heuvel et al., 1994: Hepatic CYP1A1 Mrna Expression

I.2.8.1.	Summary Table of BMDS Modeling Results	

Model3

Degrees
of

Freedom

r/'-

Value

AIC

BMD

(ng/kg)

BMDL

(ng/kg)

Notes

exponential (M2)

5

<0.0001

1147.626

1.769E+01

1.257E+01



exponential (M3)

4

<0.0001

1149.626

1.769E+01

1.257E+01

power hit bound (d = 1)

exponential (M4)

4

<0.0001

666.337

6.104E-02

2.871E-02



exponential (M5)

3

<0.0001

635.591

1.252E+00

9.089E-01



Hillb

3

<0001

664.418

2.429E-01

1.679E-01



linear

5

<0001

673.777

4.546E-02

2.487E-02



polynomial, 6-
degree

6

<0001

1213.329

error

1.301E+03



power

4

<0001

673.418

6.269E-02

3.196E-02



a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix

H.2.8.2. Output for Selected Model: Hill

Vanden Heuvel et al., 1994: Hepatic CYP1 Al mRNA Expression

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\Usepa\BMDS21\Data\hil_Vanden_mRNA_Setting.(d)

Gnuplot Plotting File: C:\Usepa\BMDS21\Data\hil_Vanden_mRNA_Setting.plt

Tue May 18 05:24:48 2010

BMDS Model Run

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = mRNA mean
Independent variable = blood cone
Power parameter is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))
Total number of dose groups = 7

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

User Inputs Initial Parameter Values

This document is a draft for review purposes only and does not constitute Agency policy.

H-37 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

lalpha
rho

intercept

1

1. 9
6

36000
1

1000

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho

intercept

lalpha
1

-0.89
-0.43
0.27
0. 68
-0.18

rho

-0.89
1

0.31
-0.42
-0.72
0.22

intercept

-0.43
0.31
1

-0.093
0.14
-0. 04

v

0.27
-0.42
-0.093
1

0. 075
0.7

n

0. 68
-0.72
0.14
0. 075
1

-0.52

k

-0.18
0.22
-0. 04
0.7
-0.52
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

Estimate
-0.191631
2.0275
5.416
41657.2
1.29154
97 .8648

Std. Err.
0.711681
0.132551
1.16292
16561.5
0.100513
41.0376

Lower Conf. Limit

-1. 5865
1.76771
3.13672
9197.25
1.09454
17 .4325

Upper Conf. Limit
1.20324
2 .28729
7 .69529
74117.2
1.48854
178 . 297

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

0.0113
0.106
0.8828

6.46
48 . 32
434 . 5

13

c

12

7

7
11
5

5.4

7 . 2
14 . 8
12 . 8
536
1. 8e + 004
3. 67e + 004

5.42

5.76
11. 6
100
1. 21e + 003
1.19e + 004
3.64 e + 0 0 4

3. 61

5.59
14 . 9
4 . 5
320

1.52e+004
2 . 21e + 004

5. 04

5.36
10. 9
97 . 2
1. 22e + 003
1. 24e + 004
3. 82e + 004

-0.0115

0.	602

1.	03
-2 . 38

-1.48
1. 62
0.0199

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Model A2:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

Mu(i) + e(ij ;

Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi

Var{e(i)}

Mu + e(i)
Sigma/N2

This document is a draft for review purposes only and does not constitute Agency policy.

H-3 8 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-572.470944	8	1160.941889

A2	-290.799287	14	609.598575

A3	-293.809342	9	605.618684

fitted	-326.209186	6	664.418372

R	-603.663396	2	1211.326792

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	625.728	12	<.0001

Test 2	563.343	6	<.0001

Test 3	6.02011	5	0.3043

Test 4	64.7997	3	<.0001

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose	Computation

Specified effect =	24

Risk Type =	Point risk

Confidence level =	0.95

BMD =	0.24 9203

BMDL =	0.167897

This document is a draft for review purposes only and does not constitute Agency policy.

H-39 DRAFT—DO NOT CITE OR QUOTE

-------
1 H.2.8.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

2	05:24 05/18 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-40 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

H.3. ADMINISTERED DOSE BMDS RESULTS
H.3.1. Hassoun et al., 2000: Cytochrome C Reductase

Model3

Degrees
of

Freedom

x'/'-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

4

0.002

-139.075

3.939E+01

3.254E+01



exponential (M3)

4

0.002

-139.075

3.939E+01

3.254E+01

power hit bound (d = 1)

exponential
(M4)b

3

0.637

-151.807

9.085E+00

5.886E+00



exponential (M5)

2

0.786

-151.023

1.420E+01

6.537E+00



Hill

2

0.741

-150.905

1.513E+01

6.277E+00



linear

4

0.032

-144.946

2.470E+01

1.933E+01



polynomial, 5-
degree

4

0.032

-144.946

2.470E+01

1.933E+01



power

4

0.032

-144.946

2.470E+01

1.933E+01

power bound hit (power =1)

power,
unrestricted0

3

0.211

-148.989

6.573E+00

1.966E+00

unrestricted (power = 0.574)

a Constant variance model selected (p = 0.3871)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

H.3.1.2. Output for Selected Model: Exponential (M4)

Hassoun et al., 2000: Cytochrome C reductase

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\5\17_Has_20 0 0_CytCLiv_ExpCV_l.(d)

Gnuplot Plotting File:

Fri Apr 30 21:15:20 2010

TBARs, liver only (Table 2)

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

H-41 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

Specified

lnalpha	-5.48625

rho(S)	0

a	0.1387

b	0.027423

c	3.36121

d	1

Parameter Estimates
Variable	Model 4

lnalpha	-5.43908

rho	0

a	0.141259

b	0.0235562

c	3.42165

d	1

Table of Stats From Input Data
Dose	N	Obs Mean	Obs Std Dev

0 6

0

146

0.06614

3 6

0

177

0.05389

10 6

0

191

0.05634

22 6

0

271

0.05634

46 6

0

388

0.06369

100 6

0

444

0.1102

Estimated Values of Interest

Dose	Est Mean	Est Std	Scaled Residual

0

0.1413

0. 06591

0

1762

3

0.1646

0.06591

0

4609

10

0.2131

0. 06591

-0

8196

22

0.2796

0. 06591

-0

3199

46

0.3676

0. 06591

0

7587

100

0. 4509

0. 06591

-0

2564

This document is a draft for review purposes only and does not constitute Agency policy.

H-42 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Other models for which	likelihoods are calculated:

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	80.75258	7	-147.5052

A2	83.37355	12	-142.7471

A3	80.75258	7	-147.5052

R	55.82002	2	-107.64

4	79.90337	4	-151.8067

Additive constant for all log-likelihoods =	-33.08. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

55.11
5.242
5.242
1.698

D. F.

10
5
5
3

p-value

< 0.0001
0.3871
0.3871
0.6373

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

This document is a draft for review purposes only and does not constitute Agency policy.

H-43 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

EMD =

9.0851

E'.MDL =	5.88612

H.3.1.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

21:15 04/30 2010

H.3.1.4. Output for Additional Model Presented: Power, Unrestricted
Hassoun et al., 2000: Cytochrome C reductase

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\5\17_Has_2000_CytCLiv_PwrCV_U_l.(d)

Gnuplot Plotting File: C:\5\17_Has_2000_CytCLiv_PwrCV_U_l.plt

Fri Apr 30 21:15:26 2010

TBARs, liver only (Table 2)

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-44 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

alpha =	0.004972

rho =	0 Specified

control =	0.146

slope = 0.0109242
power =	0.717 914

Asymptotic Correlation Matrix of Parameter Estimates

alpha
control

The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

. 8e-010

control
3. 8e-010
1

slope
-3.8e-009
-0.77

power
4.5e-009
0. 68

slope -3.8e-009	-0.77	1	-0.98

power	4.5e-009	0.68	-0.98	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
control
slope
power

Estimate

0. 00469717
0.135495
0.0232652
0.573772

Std. Err.

0.00110713
0.0246289
0.013381
0.119032

Lower Conf. Limit

0. 00252723
0.0872229
-0.00296103
0.340474

Upper Conf. Limit

0.00686711
0.183766
0.0494915
0.80707

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 6

0

146

0.135

0.0661

0

0685

0.375

3 6

0

177

0.179

0.0539

0

0685

-0.0784

10 6

0

191

0.223

0.0563

0

0685

-1.13

22 6

0

271

0.273

0.0563

0

0685

-0.056

46 6

0

388

0.345

0.0637

0

0685

1. 54

100 6

0

444

0. 462

0.11

0

0685

-0.653

Model Descriptions for likelihoods calculated

Model A1:	Yij	=	Mu(i) + e(ij)

Var{e(ij)}	=	Sigma/'2

Model A2:	Yij	=	Mu(i) + e(ij)

Var{e(ij)}	=	Sigma(i)^2

Model A3:	Yij	=	Mu(i) + e(ij)

Var{e(ij)}	=	Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

H-45 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	80.752584	7	-147.505168

A2	83.373547	12	-142.747094

A3	80.752584	7	-147.505168

fitted	78.494318	4	-148.988637

R	55.820023	2	-107.640047

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	55.107	10	<.0001

Test 2	5.24193	5	0.3871

Test 3	5.24193	5	0.3871

Test 4	4.51653	3	0.2108

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 6.57302

BMDL = 1.96558

This document is a draft for review purposes only and does not constitute Agency policy.

H-46 DRAFT—DO NOT CITE OR QUOTE

-------
H.3.1.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

21:15 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-47 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

H.3.2.	Hassoun et al., 2000: DNA Single-Strand Breaks

I.3.2.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r/'-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

4

<0.0001

120.828

3.006E+01

2.491E+01



exponential (M3)

4

<0.0001

120.828

3.006E+01

2.491E+01

power hit bound (d = 1)

exponential (M4)

3

0.036

82.814

3.734E+00

2.783E+00



exponential (M5)

3

0.036

82.814

3.734E+00

2.783E+00

power hit bound (d = 1)

Hill b

3

0.068

81.407

2.890E+00

2.007E+00

n lower bound hit (n = 1)

linear

4

<0001

111.165

1.807E+01

1.452E+01



polynomial, 5-
degree

4

<0001

111.165

1.807E+01

1.452E+01



power

4

<0001

111.165

1.807E+01

1.452E+01

power bound hit (power =1)

Hill, unrestricted

C

2

0.133

80.318

9.618E-01

2.114E-01

unrestricted (n = 0.613)

a Constant variance model selected (p = 0.7521)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

H.3.2.2. Output for Selected Model: Hill

Hassoun et al., 2000: DNA single-strand breaks

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\5\18_Has_2000_SSB_HillCV_l.(d)

Gnuplot Plotting File: C:\5\18_Has_2000_SSB_HillCV_l.plt

Fri Apr 30 21:16:28 2010

DNA single-strand breaks, liver only (Table 3)

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter restricted to be greater than 1
A constant variance model is fit

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

This document is a draft for review purposes only and does not constitute Agency policy.

H-48 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Relative Function
Parameter

has been set to: le-008
has been set to: le-008

Default Initial Parameter Values

alpha
rho

intercept

2 .7831

0

7 .41
16.09
0.174831
69.2706

Specified

Asymptotic Correlation Matrix of Parameter Estimates

alpha
intercept

The model parameter(s) -rho -n

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

1. le-007
1.9e-007
1.9e-007

intercept
1.le-007
1

0.099
0. 61

v

1.9e-007
0.099
1

0.79

k

1.9e-007
0. 61
0.79
1

Parameter Estimates

Variable
alpha
intercept

k

Estimate

2 . 82659
8 .16404
20.1253
1

31. 702

Std. Err.

0.666233
0.581043
1.69013
NA

8 . 35815

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

1.5208

7.02522
16.8127

15.3203

4 .13238
9.30286
23.4379

48 . 0836

NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0 6

7 .41

CO

i—1

G\

1. 54

1

68

-1.1

3 6

10.8

9. 9

1. 25

1

68

1.28

10 6

13. 6

13

1.69

1

68

0.889

22 6

15.3

16.4

1.71

1

68

-1. 62

46 6

20.4

20.1

2 . 25

1

68

0.469

100 6

23.5

23. 4

1. 37

1

68

0.0802

Model Descriptions for likelihoods calculated

Model A1:	Yij	= Mu(i) + e(iji

Var{e(ij)} = Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij|

Var{e(ij)} = Sigma(i)^2

This document is a draft for review purposes only and does not constitute Agency policy.

H-49 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i)

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model
A1

A2
A3

fitted
R

Log(likelihood)
-33.142389
-31.811970
-33.142389
-36.703273
-80.442086

Param's

7
12
7
4
2

AIC
80.284779
87 . 623940
80.284779
81.406545
164.884172

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio) Test df

p-value

Test 1

Test 2
Test 3
Test 4

97 .2602
2.66084
2.66084
7 . 12177

10
5
5
3

<.0001
0.7521
0.7521
0.06812

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	2 . 88976

BMDL =	2.0066 9

This document is a draft for review purposes only and does not constitute Agency policy.

H-50 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

H.3.2.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

dose

21:16 04/30 2010

H.3.2.4. Output for Additional Model Presented: Hill, Unrestricted

Hassoun et al., 2000: DNA single-strand breaks

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\5\18_Has_2000_SSB_HillCV_U_l.(d)

Gnuplot Plotting File: C:\5\18_Has_2 000_SSB_HillCV_U_l.plt

Fri Apr 30 21:16:30 2010

DNA single-strand breaks, liver only (Table 3)

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose
rho is set to 0

Power parameter is not restricted
A constant variance model is fit

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-51 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Relative Function
Parameter

has been set to: le-008
has been set to: le-008

Default Initial Parameter Values

alpha
rho

intercept

2 .7831

0

7 .41
16.09
0.174831
69.2706

Specified

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
intercept

alpha
1

-2 . 2e-008
-4 . 6e-008
8 . 4e-009
-4 . 3e-008

intercept
-2 . 2e-008
1

-0.33
0.47
-0.29

v

-4 . 6e-008
-0.33
1

-0. 95
1

n

. 4e-009
0.47
-0. 95
1

-0. 96

k

-4 . 3e-008
-0.29
1

-0. 96
1

Parameter Estimates

Variable
alpha
intercept

Estimate
2 .5942
7 . 47627
36.9014
0.612877
148.104

Std. Err.

0. 611459
0. 665055
25.5466
0.190055
303.532

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

1.39576
6.17278
-13.1689
0.240376
-446.809

3.79264

8 . 77 975
86.9718
0.985377
743.016

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0	6	7.41	7.48

3	6	10.8	10.6

10	6	13.6	13.4

22	6	15.3	16.2

46	6	20.4	19.6

100	6	23.5	23.7

1.54	1.61	-0.101

1.25	1.61	0.313

1.69	1.61	0.286

1.71	1.61	-1.41

2.25	1.61	1.24

1.37	1.61	-0.33

Model Descriptions for likelihoods calculated

Model A1:	Yij	=	Mu(i) + e(ij)

Var{e(ij)}	=	Sigma/'2

Model A2:	Yij	=	Mu(i) + e(ij)

Var{e(ij)}	=	Sigma(i)^2

Model A3:	Yij	=	Mu(i) + e(ij)

Var{e(ij)}	=	Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

H-52 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi = Mu + e(i

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1	-33.142389	7	80.284779

A2	-31.811970	12	87.623940

A3	-33.142389	7	80.284779

fitted	-35.159023	5	80.318046

R	-80.442086	2	164.884172

Explanation of Tests

Test

1

Test

2

Test

3

Test

4

(Note:

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.

Tests of Interest

Test

-2*log(Likelihood Ratio)

Test df

p-value

Test 1

Test 2
Test 3
Test 4

97 .2602
2.66084
2.66084
4.03327

10

5
5
2

<.0001
0.7521
0.7521
0.1331

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1.
model appears to be appropriate here

A homogeneous variance

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adeguately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.961789

BMDL =	0.211403

This document is a draft for review purposes only and does not constitute Agency policy.

H-53 DRAFT—DO NOT CITE OR QUOTE

-------
H.3.2.5. Figure for Additional Model Presented: Hill, Unrestricted

Hill Model with 0.95 Confidence Level

dose

21:16 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-54 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

H.3.3.	Hassoun et al., 2000: TEARS

I.3.3.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r/'-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

4

0.000

-6.143

7.977E+01

5.344E+01



exponential (M3)

4

0.000

-6.143

7.977E+01

5.344E+01

power hit bound (d = 1)

exponential
(M4)b

3

0.340

-21.181

4.916E+00

2.300E+00



exponential (M5)

2

0.240

-19.681

6.732E+00

2.470E+00



Hill

2

0.272

-19.932

6.261E+00

2.575E+00



linear

4

0.001

-7.019

6.904E+01

4.373E+01



polynomial, 5-
degree

4

0.001

-7.019

6.904E+01

4.373E+01



power

4

0.001

-7.019

6.904E+01

4.373E+01

power bound hit (power =1)

power,
unrestricted0

3

0.023

-14.993

2.902E+00

6.150E-02

unrestricted (power = 0.263)

a Constant variance model selected (p = 0.3348)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

H.3.3.2. Output for Selected Model: Exponential (M4)

Hassoun et al., 2000: TBARS

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\5\19_Has_2000_TBARsLiv_ExpCV_l.(d)

Gnuplot Plotting File:

Fri Apr 30 21:17:17 2010

TBARs, liver only (Table 2)

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

H-55 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))
rho is set to 0.

A constant variance model is fit.

Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 4

Specified

lnalpha	-1.90388

rho(S)	0

a	1.39555

b	0.0194898

c	1.97051

d	1

Parameter Estimates

Variable

lnalpha
rho
a
b
c
d

Model 4

-1. 81059

0

1.40436
0.0996859
1.74329

1

Table of Stats From Input Data

Obs Mean

Obs Std Dev

0 6

1.469

0

2915

3 6

1.549

0

5389

10 6

2 .15

0

3625

22 6

2 .28

0

2474

46 6

2 . 619

0

5168

100 6

2 . 292

0

4874

Estimated Values of Interest

Dose

Est Mean

Est Std

Scaled Residual

0

1.404

0

4044

0.3915

3

1. 674

0

4044

-0.7582

10

2 . 063

0

4044

0.527

22

2 . 332

0

4044

-0.3134

46

2 .438

0

4044

1.099

100

2.448

0

4044

-0.9458

Other models for which likelihoods are calculated:

This document is a draft for review purposes only and does not constitute Agency policy.

H-56 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)} = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	16.26977	7	-18.53954

A2	19.12783	12	-14.25565

A3	16.26977	7	-18.53954

R	2.44294	2	-0.8858799

4	14.5907	4	-21.18141

Additive constant for all log-likelihoods =	-33.08. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test

1:

Does response

and/or variances differ among Dose levels? (A2 vs. R)

Test

2 :

Are Variances

Homogeneous? (A2 vs. A1)

Test

3:

Are variances

adeguately modeled? (A2 vs. A3)

Test 6a: Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

33.37
5.716
5.716
3.358

D. F.

10
5
5
3

p-value

0.000236
0.3348
0.3348
0.3396

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

BMD =	4.91639

This document is a draft for review purposes only and does not constitute Agency policy.

H-57 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

E'.MDL =	2.2 9 952

H.3.3.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

dose

21:17 04/30 2010

H.3.3.4. Output for Additional Model Presented: Power, Unrestricted
Hassoun et al., 2000: TBARS

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\5\19_Has_2000_TBARsLiv_PwrCV_U_l.(d)

Gnuplot Plotting File: C:\5\19_Has_2 000_TBARsLiv_PwrCV_U_l.plt

Fri Apr 30 21:17:21 2010

TBARs, liver only (Table 2)

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
rho is set to 0
The power is not restricted
A constant variance model is fit

Total number of dose groups = 6

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-58 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
alpha =	0.178788

rho =	0 Specified

control =	1.469

slope = 0.0756538
power =	0.652114

Asymptotic Correlation Matrix of Parameter Estimates

alpha
control

The model parameter(s) -rho

have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )

alpha
1

1. le-008

control
1. le-008
1

slope
-1.le-009
-0.75

power
-1. 5e-008
0.47

slope -1.le-009	-0.75	1	-0.91

power -1.5e-008	0.47	-0.91	1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
alpha
control
slope
power

Estimate
0.194232
1.42104
0.333105
0.262735

Std. Err.

0. 0457809
0.171077
0.166768
0. 0983956

Lower Conf. Limit
0.104503
1.08573
0. 00624603
0. 0698836

Upper Conf. Limit
0.283961
1.75634
0.659963
0. 455587

Table of Data and Estimated Values of Interest

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0
3
10
22
46
100

1.47

1. 55
2 .15
2 .28
2 . 62
2.29

1.42

1. 87
2 . 03
2 .17
2 . 33
2 . 54

0.291

0.539
0.363
0.247
0.517
0. 487

0.441

0.441
0.441
0.441
0.441
0.441

0.267
-1.76
0. 661
0. 603
1. 6
-1. 37

Model Descriptions for likelihoods calculated

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

Model A2:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2
Model A3 uses any fixed variance parameters that
were specified by the user

This document is a draft for review purposes only and does not constitute Agency policy.

H-59 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

Model R:	Yi = Mu + e(i

Var{e(i) } = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood) # Param's	AIC

A1	16.269770	7	-18.539539

A2	19.127827	12	-14.255654

A3	16.269770	7	-18.539539

fitted	11.496634	4	-14.993268

R	2.442940	2	-0.885880

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adeguately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	33.3698	10	0.000236

Test 2	5.71611	5	0.3348

Test 3	5.71611	5	0.3348

Test 4	9.54627	3	0.02284

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 2.90232

BMDL = 0.0614 971

This document is a draft for review purposes only and does not constitute Agency policy.

H-60 DRAFT—DO NOT CITE OR QUOTE

-------
H.3.3.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

dose

21:17 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-61 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

H.3.4.	Kitchin and Woods, 1979: Bap Hydroxylase Activity

I.3.4.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

r/'-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

9

<0.0001

452.693

7.939E+03

3.663E+03



exponential (M3)

9

<0.0001

452.693

7.939E+03

3.663E+03

power hit bound (d = 1)

exponential (M4)

8

0.015

226.600

5.458E+00

4.099E+00



exponential

(M5)b

7

0.019

226.401

1.022E+01

4.807E+00



Hill

8

<0001

504.527

error

error

n upper bound hit (n = 18)

linear

9

<0001

299.732

8.276E+00

5.945E+00



polynomial, 8-
degree

3

<0001

20.000

error

error



power

9

<0001

299.732

8.276E+00

5.945E+00

power bound hit (power =1)

1 Non-constant variance model selected (p = <0.0001)
' Best-fitting model, BMDS output presented in this appendix

H.3.4.2. Output for Selected Model: Exponential (M5)

Kitchin and Woods, 1979: BaP Hydroxylase Activity

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\5\27_Kitchin_l979_Hydrolase_Exp_l.(d)

Gnuplot Plotting File:

Fri Apr 30 21:18:04 2010

Kitchin 1979, Tbl3, BaP hydrolase activity

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

This document is a draft for review purposes only and does not constitute Agency policy.

H-62 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 11

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values

Variable

lnalpha
rho

Model 5

-3.27793
1.92227
4 . 655
0.000177432
42 . 6316
1

Parameter Estimates

Variable

lnalpha
rho

Model 5

-2 . 64304
1.93753
5. 43423
0. 00191658
31.2033
1. 21503

Table of Stats From Input Data

Dose

Obs Mean

Obs Std Dev

0

0.6
2
4
20
60
200
600
2000
5000
2e+004

7

9

9
7

2

3
14
59
96

155
182
189

1.11

1.18
1. 4
1. 8
0.26
5

6.8
46
16.4
26
26

Estimated Values of Interest

Dose

Est Mean

Est Std

Scaled Residual

0

5. 434

1.

375

-1.166

0.6

5.478

1.

386

-0.8347

2

5. 624

1.

421

1. 514

4

5. 875

1.

483

1.787

20

8 . 525

2 .

127

-0.2115

60

i—1

G\

CO
-J

4

. 12

-1.394

200

49.41

11

. 67

1. 643

600

119. 4

27

.43

-1.705

2000

168 . 6

38

. 31

-0.7091

5000

169.6

38

. 53

0.6454

2e+004

169.6

38

. 53

1. 009

This document is a draft for review purposes only and does not constitute Agency policy.

H-63 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Other models for which	likelihoods are calculated:

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-158.1306	12	340.2613

A2	-84.80028	22	213.6006

A3	-98.82189	13	223.6438

R	-234.6252	2	473.2504

5	-107.2005	6	226.4011

Additive constant for all log-likelihoods =	-45.03. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)

Test 2: Are Variances Homogeneous? (A2 vs. A1)

Test 3: Are variances adeguately modeled? (A2 vs. A3)

Test 7a: Does Model 5 fit the data? (A3 vs 5)

Test

Test 1
Test 2
Test 3
Test 7a

Tests of Interest
-2*log(Likelihood Ratio)

299.6

146.7
28 . 04
16.76

D. F.

20
10
9
7

p-value

<	0.0001

<	0.0001
0.0009381

0.01903

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is less than .1. You may want to
consider a different variance model.

The p-value for Test 7a is less than .1. Model 5 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control
Confidence Level = 0.950000

This document is a draft for review purposes only and does not constitute Agency policy.

H-64 DRAFT—DO NOT CITE OR QUOTE

-------
BMD =	10.2235

BMDL =	4 . 80673

H.3.4.3. Figure for Selected Model: Exponential (M5)

Exponential Model 5 with 0.95 Confidence Level

a>

CO
c
o

Q.

CO
CD

a:

c
m

CD

250

200

150

100

50

Exponential

S/IDL

BMD

5000

10000
dose

15000

20000

21:18 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-65 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

H.3.5.	National Toxicology Program, 2006: Liver EROD 53 Weeks

I.3.5.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

4

<0.0001

210.749

4.068E+01

2.856E+01



exponential (M3)

4

<0.0001

210.749

4.068E+01

2.856E+01

power hit bound (d = 1)

exponential (M4)

3

0.071

98.835

1.912E-01

1.384E-01



exponential (M5)

2

0.040

100.232

2.394E-01

1.433E-01



Hill b

2

0.219

96.847

3.823E-01

2.336E-01



linear

4

<0001

203.577

2.076E+01

8.128E+00



polynomial, 5-
degree

4

<0001

203.577

2.076E+01

8.128E+00



power

4

<0001

203.577

2.076E+01

8.128E+00

power bound hit (power = 1)

1 Non-constant variance model selected (p = <.0001)
' Best-fitting model, BMDS output presented in this appendix

H.3.5.2. Output for Selected Model: Hill

National Toxicology Program, 2006: Liver EROD 53 Weeks

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\5\46_NTP_2006_ERODliv53_Hill_l.(d)

Gnuplot Plotting File: C:\5\4 6_NTP_2006_ERODliv53_Hill_l.plt

Sun May 02 15:05:02 2010

0

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = Mean
Independent variable = Dose

Power parameter restricted to be greater than 1

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

This document is a draft for review purposes only and does not constitute Agency policy.

H-66 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

lalpha
rho

intercept

1.59547

0

3. 614
17 .599
1.38542
8.70663

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho

intercept

lalpha
1

-0. 96
-0.16
0. 086
-0.057
0. 041

rho

-0. 96
1

0.14
-0.11
0. 059
-0.045

intercept

-0.16
0.14
1

-0.18
0.13
0. 069

v

0. 086
-0.11
-0.18
1

-0.72
0.84

n

-0.057
0. 059
0.13
-0.72
1

-0.79

k

0. 041
-0.045
0. 069
0.84
-0.79
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

Estimate
-4 .86522
2.26949
3. 62909
17.9802
1.4314
5.58259

Std. Err.
0.741624
0.287245
0.133823
0.989132
0.162447
0.717084

Lower Conf. Limit
-6.31878
1.7065
3.3668
16.0416
1.11301
4 .17713

Upper Conf. Limit

-3.41167
2 . 83248
3. 89138
19.9189
1.74979
6.98805

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

2 .14
7 .14
15.7
32 . 9
71. 4

3. 61

7 . 27
14 . 8
17 . 3
20.6
21. 2

3. 63

7 . 27
14 . 2
18 . 3
20.3
21. 2

0.486
0.557
1. 61
1.59
3. 05
3. 82

0.379

0. 833
1.78
2 . 37
2 . 67
2 . 8

-0.113

0. 0203
0. 911
-1.19
0.304
0.0606

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Model A2:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

Mu(i) + e(ij ;

Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi

Var{e(i)}

Mu + e(i)
Sigma/N2

This document is a draft for review purposes only and does not constitute Agency policy.

H-67 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1 -59.086537	7	132.173073

A2 -37.515858	12	99.031716

A3 -40.906180	8	97.812359

fitted -42.423278	6	96.846556

R	-116.710291	2	237.420582

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adequately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.

Test

2

Test

3

Test

4

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	158.389	10	<.0001

Test 2	43.1414	5	<.0001

Test 3	6.78064	4	0.1479

Test 4	3.0342	2	0.2193

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.382287

BMDL =	0.233611

This document is a draft for review purposes only and does not constitute Agency policy.

H-68 DRAFT—DO NOT CITE OR QUOTE

-------
1 H.3.5.3. Figure for Selected Model: Hill

Hill Model with 0.95 Confidence Level

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-69 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

H.3.6.	National Toxicology Program, 2006: Lung Erod 53 Weeks

I.3.6.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

4

<0.0001

316.324

8.979E+01

5.757E+01



exponential (M3)

4

<0.0001

316.324

8.979E+01

5.757E+01

power hit bound (d = 1)

exponential
(M4)b

3

0.421

255.120

8.746E-02

5.370E-02



exponential (M5)

2

0.276

256.882

6.769E-01

5.491E-02



Hill

2

0.275

256.882

1.454E+00

1.138E-01



linear

4

<0001

315.961

8.550E+01

4.502E+01



polynomial, 5-
degree

4

<0001

315.961

8.550E+01

4.502E+01



power

4

<0001

315.961

8.550E+01

4.502E+01

power bound hit (power =1)

power,
unrestricted0

3

0.037

260.794

2.688E-10

2.688E-10

unrestricted (power = 0.129)

a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix
0 Alternate model, BMDS output also presented in this appendix

H.3.6.2. Output for Selected Model: Exponential (M4)

National Toxicology Program, 2006: Lung EROD 53 Weeks

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\5\52_NTP_2006_LungEROD53_Exp_l.(d)

Gnuplot Plotting File:

Fri Apr 30 21:22:36 2010

Tbl 12, Week 53, Lung Microsomes EROD

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

This document is a draft for review purposes only and does not constitute Agency policy.

H-70 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values

Variable

lnalpha
rho

Model 4

-0.80064
1.47683
2 .86045
0.054659
16.0581
1

Parameter Estimates
Variable	Model 4

lnalpha	-1.15021

rho	1.63127

a	3.06838

b	0.414677

c	13.847

d	1

Table of Stats From Input	Data

Dose	N	Obs Mean	Obs Std Dev

0	8	3.011	1.584

2.14	8	27.15	5.269

7.14	8	42.85	11.15

15.7	8	36.57	12.99

32.9	8	43.75	18.55

71.4	8	43.71	6.322

Estimated Values of	Interest
Dose Est Mean Est Std Scaled Residual

0	3.068	1.404	-0.1156

2.14	26.26	8.088	0.3116

7.14	40.45	11.5	0.5901

15.7	42.43	11.96	-1.386

32.9	42.49	11.98	0.2972

71.4	42.49	11.98	0.2894

Other models for which likelihoods are calculated:

Model A1:	Yij = Mu(i) + e(ij)

Var{e(ij)} = Sigma/'2

This document is a draft for review purposes only and does not constitute Agency policy.

H-71 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

Model A2 :	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i)) * rho)

Model R:	Yij	= Mu + e(i)

Var{e(ij)}	= Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-135.2677	7	284.5353

A2	-115.6885	12	255.3771

A3	-121.1517	8	258.3034

R	-162.0902	2	328.1805

4	-122.5601	5	255.1202

-44.11. This constant added to the

Additive constant for all log-likelihoods =
above values gives the log-likelihood including the term that does not
depend on the model parameters.

Test

1:

Test

2 :

Test

3:

Test

6a

Explanation of Tests

Does response and/or variances differ among Dose levels? (A2 vs. R)

Are Variances Homogeneous? (A2 vs. A1)

Are variances adeguately modeled? (A2 vs. A3)

Does Model 4 fit the data? (A3 vs 4)

Test

Test 1
Test 2
Test 3
Test 6a

Tests of Interest

-2*log(Likelihood Ratio)

92 . 8
39.16
10. 93
2 . 817

10
5
4
3

p-value

0.0001
0.0001
0.0274
0.4207

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is less than .1. You may want to
consider a different variance model.

The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control

Confidence Level = 0.950000

BMD = 0.0874595
BMDL = 0.0537035

This document is a draft for review purposes only and does not constitute Agency policy.

H-72 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

H.3.6.3. Figure for Selected Model: Exponential (M4)

Exponential Model 4 with 0.95 Confidence Level

60

Exponential

50

40

30

20

10

BVIDLBMD

10

20

30

40

50

60

70

21:22 04/30 2010

dose

H.3.6.4. Output for Additional Model Presented: Power, Unrestricted

National Toxicology Program, 2006: Lung EROD 53 Weeks

Power Model. (Version: 2.15; Date: 04/07/2008)

Input Data File: C:\5\52_NTP_2006_LungEROD53_Pwr_U_l.(d)

Gnuplot Plotting File: C:\5\52_NTP_2006_LungEROD53_Pwr_U_l.plt

Fri Apr 30 21:22:40 2010

Tbl 12, Week 53, Lung Microsomes EROD

The form of the response function is:
Y[dose] = control + slope * doseApower

Dependent variable = Mean
Independent variable = Dose
The power is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008

This document is a draft for review purposes only and does not constitute Agency policy.

H-73 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

Parameter

has been set to: le-008

Default Initial Parameter Values

lalpha =	4 .76968

rho =	0

control =	3.011

slope =	24 . 7003

power =	0.132 9 96

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho
control
slope
power

lalpha
1

-0. 96
-0.48
0.11
-0.048

rho

-0. 96
1

0.45
-0.15
0. 053

control

-0.48
0.45
1

-0.15
0. 05

slope
0.11
-0.15
-0.15
1

-0. 92

power

-0.048
0. 053
0. 05
-0. 92
1

Parameter Estimates

Variable
lalpha
rho
control
slope
power

Estimate

-1.03242
1.63031
3.01793
25.144
0.128894

Std. Err.

0. 815871
0.239764
0.518146
3.39289
0.0448391

95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit

-2 . 6315
1.16038
2.00238
18.494
0.041011

0.566654
2 .10024
4.03348
31.7939
0 . 216777

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

2 .14
7 .14
15.7
32 . 9
71. 4

3. 01

27 .1
42 . 8
3 6.6
43.7
43.7

3. 02
30. 8
35. 4
38 . 9
42 . 5
46.6

I.	58

5.27

II.	2
13

18 . 5
6.32

I.47

9.74
10. 9

II.	8
12 . 7
13.7

-0.0133
-1. 05
1. 92
-0.553
0.286
-0.598

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Model A2:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

Mu(i) + e(ij ;

Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi

Var{e(i)}

Mu + e(i)
Sigma/N2

This document is a draft for review purposes only and does not constitute Agency policy.

H-74 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-135.267662	7	284.535325

A2	-115.688533	12	255.377067

A3	-121.151707	8	258.303413

fitted	-125.397022	5	260.794043

R	-162.090242	2	328.180484

Explanation of Tests

Test 1:

Test

2

Test

3

Test

4

Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Are Variances Homogeneous? (A1 vs A2)

Are variances adequately modeled? (A2 vs. A3)

Does the Model for the Mean Fit? (A3 vs. fitted)

When rho=0 the results of Test 3 and Test 2 will be the same.'

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	92.8034	10	<.0001

Test 2	39.1583	5	<.0001

Test 3	10.9263	4	0.0274

Test 4	8.49063	3	0.03689

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is less than .1. You may want to consider a
different variance model

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD = 2 . 68823e-010

BMDL = 2 . 68823e-010

This document is a draft for review purposes only and does not constitute Agency policy.

H-75 DRAFT—DO NOT CITE OR QUOTE

-------
H.3.6.5. Figure for Additional Model Presented: Power, Unrestricted

Power Model with 0.95 Confidence Level

a>

CO
c
o

Q.

CO
CD

a:

c
m

CD

10

20

30

40

50

60

70

dose

21:22 04/30 2010

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-76 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

H.3.7.	National Toxicology Program, 2006: Labeling Index 31 Weeks

I.3.7.1.	Summary Table of BMDS Modeling Results

Model3

Degrees
of

Freedom

x2p-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential
(M2)b

4

0.000

47.304

2.336E+01

1.867E+01



exponential (M3)

4

0.000

47.304

2.336E+01

1.867E+01

power hit bound (d = 1)

exponential (M4)

3

<0.0001

53.331

1.233E+01

7.562E+00



exponential (M5)

2

<0.0001

51.057

3.279E+01

2.055E+01



Hill

3

0.000

49.057

3.277E+01

error

n upper bound hit (n = 18)

linear

4

<0001

51.331

1.233E+01

7.563E+00



polynomial, 5-
degree

3

0.000

48.698

2.510E+01

1.192E+01



power

3

<0001

49.826

3.238E+01

1.723E+01



1 Non-constant variance model selected (p = <0.0001)
' Best-fitting model, BMDS output presented in this appendix

H.3.7.2. Output for Selected Model: Exponential (M2)

National Toxicology Program, 2006: Labeling Index 31 Weeks

Exponential Model. (Version: 1.61; Date: 7/24/2009)

Input Data File: C:\5\38_NTP_2006_HepIndex_Exp_l.(d)

Gnuplot Plotting File:

Fri Apr 30 21:23:28 2010

Tbl 11, 31wk, Hep Cell Proliferation Labeling Index

The form of the response function by Model:

Model 2
Model 3
Model 4
Model 5

Y[dose]
Y[dose]
Y[dose]
Y[dose]

exp{sign *	b * dose}

exp{sign *	(b * dosej^d}

[c-(c-l) *	exp{-b * dose}]

[c-(c-l) *	exp{-(b * dosej^d}

Note: Y[dose] is the median response for exposure = dose;
sign = +1 for increasing trend in data;
sign = -1 for decreasing trend.

Model 2 is nested within Models 3 and 4.
Model 3 is nested within Model 5.

Model 4 is nested within Model 5.

Dependent variable = Mean
Independent variable = Dose

Data are assumed to be distributed: normally
Variance Model: exp(lnalpha +rho *ln(Y[dose]))

This document is a draft for review purposes only and does not constitute Agency policy.

H-77 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 6

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

MLE solution provided: Exact

Initial Parameter Values
Variable	Model 2

lnalpha	-0.674004

rho	2.29189

a	0.576363

b	0.0266174

c	0

d	1

Parameter Estimates

Variable	Model 2

lnalpha	-0.471424

rho	1.902 98

a	0.616539

b	0.0253715

c	0

d	1

Table of Stats From Input Data

Dose N Obs Mean	Obs Std Dev

0	9

2.14	10

7.14	10

15.7	10

32.9	10

71.4	10

Estimated Values of Interest
Dose	Est Mean	Est Std	Scaled Residual

0.327
0. 852
0. 956

0.792

1.	333
3.846

0.189
0.6514
0.7368
0.4617
1.123
3. 08

0	0.6165	0.4986	-1.742

2.14	0.6509	0.5251	1.211

7.14	0.739	0.5924	1.158

15.7	0.9182	0.7284	-0.548

32.9	1.421	1.103	-0.2511

71.4	3.773	2.795	0.08251

Other models for which	likelihoods are calculated:

Model A1:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma/'2

Model A2:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= Sigma(i)^2

Model A3:	Yij	= Mu(i) + e(ij)

Var{e(ij)}	= exp(lalpha + log(mean(i)) * rho)

This document is a draft for review purposes only and does not constitute Agency policy.

H-78 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Model R:	Yij = Mu + e(i)

Var{e(ij)} = Sigma/'2

Likelihoods of Interest

Model	Log(likelihood)	DF	AIC

A1	-47.23498	7	108.47

A2	-8.679256	12	41.35851

A3	-8.980651	8	33.9613

R	-63.44829	2	130.8966

2	-19.65195	4	47.30389

Additive constant for all log-likelihoods =	-54.22. This constant added to the

above values gives the log-likelihood including the term that does not
depend on the model parameters.

Explanation of Tests

Does response and/or variances differ among Dose levels? (A2

Are Variances Homogeneous? (A2 vs. A1)

Are variances adeguately modeled? (A2 vs. A3)

Does Model 2 fit the data? (A3 vs. 2)

Tests of Interest
Test	-2*log(Likelihood Ratio)	D. F.	p-value

Test 1	109.5	10	< 0.0001

Test 2	77.11	5	< 0.0001

Test 3	0.6028	4	0.9628

Test 4	21.34	4	0.0002708

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.

The p-value for Test 2 is less than .1. A non-homogeneous
variance model appears to be appropriate.

The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.

The p-value for Test 4 is less than .1. Model 2 may not adeguately
describe the data; you may want to consider another model.

Benchmark Dose Computations:

Specified Effect = 1.000000

Risk Type = Estimated standard deviations from control

Confidence Level = 0.950000

BMD =	23.3586

BMDL =	18.6683

This document is a draft for review purposes only and does not constitute Agency policy.

H-79 DRAFT—DO NOT CITE OR QUOTE

-------
1 H.3.7.3. Figure for Selected Model: Exponential (M2)

Exponential Model 2 with 0.95 Confidence Level

0	10	20	30	40	50	60	70

dose

2	21:23 04/30 2010

3

t	i	r

Exponential 	

BMDL

BMD

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-80 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

H.3.8.	Vanden Heuvel et al., 1994: Hepatic CYP1A1 Mrna Expression

I.3.8.1.	Summary Table of BMDS Modeling Results	

Model3

Degrees
of

Freedom

r/'-

Value

AIC

BMD

(ng/kg-d)

BMDL

(ng/kg-d)

Notes

exponential (M2)

5

<0.0001

1164.377

4.699E+03

1.729E+03



exponential (M3)

5

<0.0001

1164.377

4.699E+03

1.729E+03

power hit bound (d = 1)

exponential (M4)

4

<0.0001

661.006

4.550E-01

2.643E-01



exponential (M5)

3

<0.0001

635.327

1.516E+01

1.046E+01



Hill b

3

<0001

662.251

8.091E-01

4.844E-01



linear

5

<0001

667.554

4.953E-01

3.093E-01



polynomial, 6-
degree

1

<0001

715.412

5.774E+03

1.204E+01



power

4

<0001

669.441

5.571E-01

3.204E-01



a Non-constant variance model selected (p = <0.0001)
b Best-fitting model, BMDS output presented in this appendix

H.3.8.2. Output for Selected Model: Hill

Vanden Heuvel et al., 1994: Hepatic CYP1 Al mRNA Expression

Hill Model. (Version: 2.14; Date: 06/26/2008)

Input Data File: C:\Usepa\BMDS21\Data\hil_Vanden_mRNA_Setting.(d)

Gnuplot Plotting File: C:\Usepa\BMDS21\Data\hil_Vanden_mRNA_Setting.plt

Wed May 19 14:25:06 2010

BMDS Model Run

The form of the response function is:

Y[dose] = intercept + v*doseAn/(kAn + doseAn)

Dependent variable = mRNA mean

Independent variable = d

Power parameter is not restricted

The variance is to be modeled as Var(i) = exp(lalpha + rho * ln(mean(i)))
Total number of dose groups = 7

Total number of records with missing values = 0
Maximum number of iterations = 250

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values

This document is a draft for review purposes only and does not constitute Agency policy.

H-81 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

lalpha
rho

intercept

18.2064

0

5.4
3 6 6 9 4 . 6
0.720907
18830.3

Asymptotic Correlation Matrix of Parameter Estimates

lalpha
rho

intercept

lalpha
1

-0.89
-0.41
0.37
0.7
-0.2

rho

-0.89
1

0.29
-0.54
-0.75
0.24

intercept

-0.41
0.29
1

-0.11
0.13
-0.034

v

0.37
-0.54
-0.11
1

0.21
0.57

n

0.7
-0.75
0.13
0.21
1

-0.53

k

-0.2

0.24
-0.034
0.57
-0.53
1

Parameter Estimates

95.0% Wald Confidence Interval

Variable
lalpha
rho

intercept

Estimate
-0.28219
2.05171
5.4299
36598.9
1.13992
2012.71

Std. Err.

0.146654
1.14997
13930.2
0.0919476
881.73

Lower Conf. Limit
-1.71928
1.76427
3.17599
9296.23
0. 959705
284 . 554

Upper Conf. Limit
1.1549
2.33915
7.68381
63901.7
1. 32013
3740.87

Table of Data and Estimated Values of Interest

Dose

Obs Mean

Est Mean Obs Std Dev Est Std Dev Scaled Res.

0

0.1

1
10

100
1000
le+004

13

5
12
7
7
11

5.4

7 . 2
14 . 8
12 . 8
536
1. 8e + 004
3. 67e + 004

5.43

5.88
11. 7
91. 8
1.16e + 003
1.14e + 004
3 .15e + 004

3. 61

5.59
14 . 9
4 . 5
320

1.52e+004
2 . 21e + 004

4 . 93

5.35
10.8
89.6
1. 21e + 003
1. 26e + 004
3.58e+004

-0.0219
0.55
0. 991
-2 . 33
-1. 37
1.75
0.323

Model Descriptions for likelihoods calculated

Model A1:	Yij

Var{e(ij ) }

Model A2:	Yij

Var{e(ij ) }

Mu(i) + e(ij ;
Sigma/N2

Mu(i) + e(ij ;

Sigma(i)^2

Model A3:	Yij = Mu(i) + e(ij)

Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user

Model R:	Yi

Var{e(i)}

Mu + e(i)
Sigma/N2

This document is a draft for review purposes only and does not constitute Agency policy.

H-82 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

Likelihoods of Interest

Model	Log(likelihood)	# Param's	AIC

A1	-572.470944	8	1160.941889

A2	-290.799287	14	609.598575

A3	-293.809342	9	605.618684

fitted	-325.125462	6	662.250924

R	-603.663396	2	1211.326792

Explanation of Tests

Test 1: Do responses and/or variances differ among Dose levels?

(A2 vs. R)

Test 2: Are Variances Homogeneous? (A1 vs A2)

Test 3: Are variances adequately modeled? (A2 vs. A3)

Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)

(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)

Tests of Interest

Test	-2*log(Likelihood Ratio)	Test df	p-value

Test 1	625.728	12	<.0001

Test 2	563.343	6	<.0001

Test 3	6.02011	5	0.3043

Test 4	62.6322	3	<.0001

The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data

The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate

The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here

The p-value for Test 4 is less than .1. You may want to try a different
model

Benchmark Dose Computation
Specified effect =	1

Risk Type	=	Estimated standard deviations from the control mean

Confidence level =	0.95

BMD =	0.809125

BMDL =	0.4 84 455

This document is a draft for review purposes only and does not constitute Agency policy.

H-83 DRAFT—DO NOT CITE OR QUOTE

-------
1 H.3.8.3. Figure for Selected Model: Exponential (M5)

Hill Model with 0.95 Confidence Level

dose

2	14:25 05/19 2010

3

4

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

H-84 DRAFT—DO NOT CITE OR QUOTE

-------
[This page intentionally left blank.]

-------
DRAFT

DO NOT CITE OR QUOTE

May 2010
External Review Draft

APPENDIX I

Effect of Background Exposure on
Benchmark-Dose Modeling

NOTICE

THIS DOCUMENT IS AN EXTERNAL REVIEW DRAFT. It has not been formally released
by the U.S. Environmental Protection Agency and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its technical accuracy and policy
implications.

National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH

-------
CONTENTS—APPENDIX I: Effect of Background Exposure on Benchmark-Dose

Modeling

LIST OF FIGURES	I-ii

APPENDIX I. EFFECT OF BACKGROUND EXPOSURE ON BENCHMARK-DOSE

MODELING	1-1

1.1.	NTP, 2006 (CHOLANGIOCARCINOMAS): UNADJUSTED BLOOD
CONCENTRATIONS	I-1

1.2.	NTP, 2006 (CHOLANGIOCARCINOMAS): BACKGROUND DOSE =

MEASURED TCDD CONCENTRATION ONLY	1-4

1.3.	NTP, 2006 (CHOLANGIOCARCINOMAS): BACKGROUND DOSE =

MEASURED TEQ CONCENTRATION (TCDD, PECDF, AND PCB-126)	1-7

1.4.	NTP, 2006 (CHOLANGIOCARCINOMAS): BACKGROUND DOSE = 2x
MEASURED TEQ CONCENTRATION (TCDD, PECDF, AND PCB-126)	I-10

1.5.	NTP, 2006 (CHOLANGIOCARCINOMAS): BACKGROUND DOSE = 10x
MEASURED TCDD CONCENTRATION	1-13

1.6.	REFERENCE	1-15

LIST OF FIGURES

1-1. NTP, 2006: Unadjusted blood concentrations (cholangiocarcinomas)	1-3

1-2. NTP, 2006 (cholangiocarcinomas): Background dose = measured TCDD

concentration only	1-6

1-3. NTP, 2006 (cholangiocarcinomas): Background dose = measured TEQ

concentration (TCDD, PeCDF, and PCB-126)	1-9

1-4. NTP, 2006 (cholangiocarcinomas): Background dose = 2x measured TEQ

concentration (TCDD, PeCDF, and PCB-126)	1-12

1-5. NTP, 2006 (cholangiocarcinomas): Background dose = 10x measured TCDD

concentration	1-15

This document is a draft for review purposes only and does not constitute Agency policy.

I-ii DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

APPENDIX I. EFFECT OF BACKGROUND EXPOSURE ON BENCHMARK-DOSE

MODELING

1.1. NTP, 2006 (CHOLANGIOCARCINOMAS): UNADJUSTED BLOOD
CONCENTRATIONS

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Usepa\BMDS21\Data\msc_NTP_2006_carcin_Setting.(d)

Gnuplot Plotting File: C:\Usepa\BMDS21\Data\msc_NTP_2 00 6_carcin_Setting.plt

Wed Apr 14 12:59:57 2010

BMDS Model Run

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/sl-beta2*dose/s2-beta3* dose^)]

The parameter betas are restricted to be positive

Dependent variable = cholang
Independent variable = bl_nom

Total number of observations = 6

Total number of records with missing values = 0

Total number of parameters in model = 4

Total number of specified parameters = 0

Degree of polynomial = 3

Maximum number of iterations = 25 0

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta (1) =	0

Beta (2) =	0

Beta (3) = 2.44609e-005

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l) -Beta(2)

have been estimated at a boundary point, or have been specified by

the user,

and do not appear in the correlation matrix )

Beta (3)

Beta(3)	1

This document is a draft for review purposes only and does not constitute Agency policy.

1-1 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

Parameter Estimates

95.0% Wald Confidence

Interval

Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf.

Limit

Background

0







Beta (1)

0







Beta (2)

0







Beta (3)

2.30992e-005







* - Indicates that this value is not calculated.

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table
#

Log(likelihood)
-55.408
-55 .7584
-96.9934

Param's
6
1
1

Deviance Test d.f.

0.700706
83.1708

P-value

0.9829
<.0001

AIC:

113.517

Goodness of Fit

Scaled

Dose	Est._Prob. Expected Observed	Size	Residual

0.0000

0.0000

0. 000

0. 000

49

0.000

2.5600

0.0004

0. 019

0. 000

48

-0.136

5.6900

0.0042

0.195

0. 000

46

-0.443

9.7900

0.0214

1. 072

1. 000

50

-0.070

16.6000

0.1003

4.913

4.000

49

-0.434

29.7000

0.4540

24.063

25.000

53

0.259

Chi^2 = 0.48	d.f. = 5	P-value = 0.9930

Benchmark Dose Computation

Specified effect =	0.01

Risk Type =	Extra risk

Confidence level =	0.95

BMD =	7.57754

BMDL =	4.13907

BMDU =	8.42931

Taken together, (4.13907, 8.42931) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.002416

This document is a draft for review purposes only and does not constitute Agency policy.

1-2 DRAFT—DO NOT CITE OR QUOTE

-------
Multistage Cancer Model with 0.95 Confidence Level

dose

12:59 04/14 2010

Figure 1-1. NTP, 2006: Unadjusted blood concentrations
(cholangiocarcinomas).

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

1-3	DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

1.2. NTP, 2006 (CHOLANGIOCARCINOMAS): BACKGROUND DOSE = MEASURED
TCDD CONCENTRATION ONLY

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Usepa\BMDS21\Data\msc_NTP_2006_carcin_Setting.(d)

Gnuplot Plotting File: C:\Usepa\BMDS21\Data\msc_NTP_2 00 6_carcin_Setting.plt

Fri Apr 16 15:47:08 2010

BMDS Model Run

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/sl-beta2*dose/s2-beta3* doseA3)]

The parameter betas are restricted to be positive

Dependent variable = cholang
Independent variable = bl_TCDDadj

Total number of observations = 6

Total number of records with missing values = 0

Total number of parameters in model = 4

Total number of specified parameters = 0

Degree of polynomial = 3

Maximum number of iterations = 25 0

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta (1) =	0

Beta (2) =	0

Beta (3) = 2.43074e-005

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l) -Beta(2)

have been estimated at a boundary point, or have been specified by

the user,

and do not appear in the correlation matrix )

Beta (3)

Beta(3)	1

Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

1-4 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

95.0% Wald Confidence

Interval

Variable Estimate Std. Err.	Lower Conf. Limit Upper Conf.

Limit

Background 0 *	* *

Beta (1) 0 *	* *

Beta (2) 0 *	* *

Beta (3) 2.29144e-005 *	* *

* - Indicates that this value is not calculated.

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table
#

Log(likelihood)
-55.408
-55.771
-96.9934

Param's
6
1
1

Deviance Test d.f.

0.726
83.1708

P-value

0.9815
<.0001

AIC:

113.542

Goodness of Fit

Scaled

Dose	Est._Prob. Expected Observed	Size	Residual

0.0640

0.0000

0. 000

0. 000

49

-0.001

2.6240

0.0004

0. 020

0. 000

48

-0.141

5.7540

0.0044

0.200

0. 000

46

-0.449

9.8540

0.0217

1. 084

1. 000

50

-0.082

16.6640

0.1006

4.930

4.000

49

-0.442

29.7640

0.4535

24.035

25.000

53

0.266

Chi^2 = 0.49	d.f. = 5	P-value = 0.9924

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	7.59785

BMDL =	4.19355

BMDU =	8.4518 8

Taken together, (4.19355, 8.45188) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00238461

This document is a draft for review purposes only and does not constitute Agency policy.

1-5 DRAFT—DO NOT CITE OR QUOTE

-------
Multistage Cancer Model with 0.95 Confidence Level

dose

15:47 04/16 2010

Figure 1-2. NTP, 2006 (cholangiocarcinomas): Background dose = measured
TCDD concentration only.

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

1-6 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

1.3. NTP, 2006 (CHOLANGIOCARCINOMAS): BACKGROUND DOSE = MEASURED
TEQ CONCENTRATION (TCDD, PECDF, AND PCB-126)

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Usepa\BMDS21\Data\msc_NTP_2006_carcin_Setting.(d)

Gnuplot Plotting File: C:\Usepa\BMDS21\Data\msc_NTP_2 00 6_carcin_Setting.plt

Fri Apr 16 15:50:00 2010

BMDS Model Run

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/sl-beta2*dose/s2-beta3* doseA3)]

The parameter betas are restricted to be positive

Dependent variable = cholang
Independent variable = bl_TEQadj

Total number of observations = 6

Total number of records with missing values = 0

Total number of parameters in model = 4

Total number of specified parameters = 0

Degree of polynomial = 3

Maximum number of iterations = 25 0

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta (1) =	0

Beta (2) =	0

Beta(3) = 2.40088e-005

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l) -Beta(2)

have been estimated at a boundary point, or have been specified by

the user,

and do not appear in the correlation matrix )

Beta (3)

Beta(3)	1

Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

1-7 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

95.0% Wald Confidence

Interval

Variable Estimate Std. Err.	Lower Conf. Limit Upper Conf.

Limit

Background 0 *	* *

Beta (1) 0 *	* *

Beta (2) 0 *	* *

Beta (3) 2.25556e-005 *	* *

* - Indicates that this value is not calculated.

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table
#

Log(likelihood)
-55.408
-55.7969
-96.9934

Param's
6
1
1

Deviance Test d.f.

0.777718
83.1708

P-value

0.9784
<.0001

AIC:

113.594

Goodness of Fit

Scaled

Dose	Est._Prob. Expected Observed	Size	Residual

0.1900

0.0000

0. 000

0. 000

49

-0.003

2.7500

0.0005

0. 023

0. 000

48

-0.150

5.8800

0.0046

0.210

0. 000

46

-0.460

9.9800

0.0222

1.109

1. 000

50

-0.104

16.7900

0.1013

4.962

4.000

49

-0.455

29.8900

0.4525

23.981

25.000

53

0.281

Chi^2 = 0.53	d.f. = 5	P-value = 0.9909

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	7.63793

BMDL =	4.29872

BMDU =	8.4 964

Taken together, (4.29872, 8.4964 ) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00232627

This document is a draft for review purposes only and does not constitute Agency policy.

1-8 DRAFT—DO NOT CITE OR QUOTE

-------
Multistage Cancer Model with 0.95 Confidence Level

dose

15:50 04/16 2010

Figure 1-3. NTP, 2006 (cholangiocarcinomas): Background dose = measured
TEQ concentration (TCDD, PeCDF, and PCB-126).

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

1-9 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

1.4. NTP, 2006 (CHOLANGIOCARCINOMAS): BACKGROUND DOSE = 2x
MEASURED TEQ CONCENTRATION (TCDD, PECDF, AND PCB-126)

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Usepa\BMDS21\Data\msc_NTP_2006_carcin_Setting.(d)

Gnuplot Plotting File: C:\Usepa\BMDS21\Data\msc_NTP_2 00 6_carcin_Setting.plt

Fri Apr 16 15:51:30 2010

BMDS Model Run

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/sl-beta2*dose/s2-beta3* doseA3)]

The parameter betas are restricted to be positive

Dependent variable = cholang
Independent variable = bl_TEQ2x

Total number of observations = 6

Total number of records with missing values = 0

Total number of parameters in model = 4

Total number of specified parameters = 0

Degree of polynomial = 3

Maximum number of iterations = 25 0

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta (1) =	0

Beta (2) =	0

Beta(3) = 2.3568e-005

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l) -Beta(2)

have been estimated at a boundary point, or have been specified by

the user,

and do not appear in the correlation matrix )

Beta (3)

Beta(3)	1

Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

I-10 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

95.0% Wald Confidence

Interval

Variable Estimate Std. Err.	Lower Conf. Limit Upper Conf.

Limit

Background 0 *	* *

Beta (1) 0 *	* *

Beta (2) 0 *	* *

Beta (3) 2.20268e-005 *	* *

* - Indicates that this value is not calculated.

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table
#

Log(likelihood)
-55.408
-55.8382
-96.9934

Param's
6
1
1

Deviance Test d.f.

0.860456
83.1708

P-value

0.973
<.0001

AIC:

113.676

Goodness of Fit

Scaled

Dose	Est._Prob. Expected Observed	Size	Residual

0.3800

0.0000

0. 000

0. 000

49

-0.008

2.9400

0.0006

0. 027

0. 000

48

-0.164

6.0700

0.0049

0.226

0. 000

46

-0.477

10.1700

0.0229

1.145

1. 000

50

-0.137

16.9800

0.1022

5.009

4.000

49

-0.476

30.0800

0.4509

23.898

25.000

53

0.304

Chi^2 = 0.59	d.f. = 5	P-value = 0.9884

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	7.6985 6

BMDL =	4.45212

BMDU =	8.5 637 6

Taken together, (4.45212, 8.56376) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor = 0.00224612

This document is a draft for review purposes only and does not constitute Agency policy.

I-11 DRAFT—DO NOT CITE OR QUOTE

-------
Multistage Cancer Model with 0.95 Confidence Level

dose

15:51 04/16 2010

Figure 1-4. NTP, 2006 (cholangiocarcinomas): Background dose = 2x
measured TEQ concentration (TCDD, PeCDF, and PCB-126).

This document is a draft for re\'iew purposes only and does not constitute Agency policy.

1-12 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

1.5. NTP, 2006 (CHOLANGIOCARCINOMAS): BACKGROUND DOSE = 10x
MEASURED TCDD CONCENTRATION

Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)

Input Data File: C:\Usepa\BMDS21\Data\msc_NTP_2006_carcin_Setting.(d)

Gnuplot Plotting File: C:\Usepa\BMDS21\Data\msc_NTP_2 00 6_carcin_Setting.plt

Fri Apr 16 15:55:37 2010

BMDS Model Run

The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(

-betal*dose/sl-beta2*dose/s2-beta3* doseA3)]

The parameter betas are restricted to be positive

Dependent variable = cholang
Independent variable = bl_TEQmax

Total number of observations = 6

Total number of records with missing values = 0

Total number of parameters in model = 4

Total number of specified parameters = 0

Degree of polynomial = 3

Maximum number of iterations = 25 0

Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background =	0

Beta (1) =	0

Beta (2) =	0

Beta(3) = 2.29823e-005

Asymptotic Correlation Matrix of Parameter Estimates

( *** The model parameter(s) -Background -Beta(l) -Beta(2)

have been estimated at a boundary point, or have been specified by

the user,

and do not appear in the correlation matrix )

Beta (3)

Beta(3)	1

Parameter Estimates

This document is a draft for review purposes only and does not constitute Agency policy.

I-13 DRAFT—DO NOT CITE OR QUOTE

-------
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

95.0% Wald Confidence

Interval

Variable Estimate Std. Err.	Lower Conf. Limit Upper Conf.

Limit

Background 0 *	* *

Beta (1) 0 *	* *

Beta (2) 0 *	* *

Beta (3) 2.13264e-005 *	* *

* - Indicates that this value is not calculated.

Model
Full model
Fitted model
Reduced model

Analysis of Deviance Table
#

Log(likelihood)
-55.408
-55.8994
-96.9934

Param's
6
1
1

Deviance Test d.f.

0.982747
83.1708

P-value

0.9639
<.0001

AIC:

113.799

Goodness of Fit

Scaled

Dose	Est._Prob. Expected Observed	Size	Residual

0.6400

0.0000

0. 000

0. 000

49

-0.017

3.2000

0.0007

0. 034

0. 000

48

-0.183

6.3300

0.0054

0.248

0. 000

46

-0.499

10.4300

0.0239

1.195

1. 000

50

-0.181

17.2400

0.1035

5.072

4.000

49

-0.503

30.3400

0.4488

23.785

25.000

53

0.336

Chi^2 = 0.68	d.f. = 5	P-value = 0.9840

Benchmark Dose Computation
Specified effect =	0.01

Risk Type	=	Extra risk

Confidence level =	0.95

BMD =	7.78193

BMDL =	4.65224

BMDU =	8.65 638

Taken together, (4.65224, 8.65638) is a 90	% two-sided confidence

interval for the BMD

Multistage Cancer Slope Factor =	0.0021495

This document is a draft for review purposes only and does not constitute Agency policy.

1-14 DRAFT—DO NOT CITE OR QUOTE

-------
Multistage Cancer Model with 0.95 Confidence Level

dose

1	15:55 04/16 2010

2

3	Figure 1-5. NTP, 2006 (cholangiocarcinomas): Background dose = 10*

4	measured TCDD concentration.

5	1.6. REFERENCE

6	NTP (National Toxicology Program). (2006a) NTP technical report on the toxicology and carcinogenesis studies of

7	2,3,7,8-tetrachlorodibenzo-/?-dioxin (TCDD) (CAS No. 1746-01-6) in female Harlan Sprague-Dawley rats (Gavage

8	Studies). Natl Toxicol ProgramTech Rep 521. Public Health Service, National Institute of Health, U.S. Department

9	of Health and Human Services, Research Triangle Park, NC.

This document is a draft for review purposes only and does not constitute Agency policy.

1-15 DRAFT—DO NOT CITE OR QUOTE

-------