Umt*0SwtM
Environment*! Protection .
A«*ncv , November 1984
Research and
Development
METHODOLOGY AND GUIDELINES
FOR RANKING CHEMICALS BASED
ON CHRONIC TOXICITY DATA
Prepared for
OFFICE OF EMERGENCY AND
REMEDIAL RESPONSE
Prepared by
Environmental Criteria and
Assessment Office
Cincinnati OH 45268
-------
DISCLAIMER
Mention of trade names or c owner da 1 products does not constitute
endorsement or recommendation for use.
-------
TABLE OF CONTENTS
Page
1. INTRODUCTION ..................... '
2. CHRONIC TOXICITY RANKIN6 SYSTE* .................
3. APPLICATION OF THE KETHODOlOGY ..................
4. S0« CONCEPTUAL CONSIDERATIONS. ....••• ..........
5. MODIFICATION AND APPLICATIONS ..................
• «•***
6. REFERENCES ......................
111
-------
LIST OF ABBREVIATIONS
ADI Acceptable dally Intake
CERCLA Comprehensive Environmental Response, Compensation and
Liability Act of 1980-
CS Composite score
CWA Clean Hater Act
FEL Frank effect level
ICRP International Commission on Radlologlc Protection
LDso Dose fatal to 50* of test ail"*1?
LOAEL Lowest-observed-adverse-effect level
MED Minimum effective dose
NOAEL No-observed-adverse-effect level
NOEL No-observed-effect level
OERR Office of Emergency and Remedial Response
OSWER Office of Solid Haste and Emergency Response
RQ Reportable quantity
RVj Dose rating value
RVe Effect rating value
1v
-------
PREFACE
During the past three years the Environmental Criteria and Assessment
Office. Cincinnati (ECAO-C1n) of the U.S. EPA has developed a method to rank
chemicals based on chronic toxldty data. This ranking system reflects two
primary attributes of every chemical: the minimum effective dose and the
type of effect elicited at that dose. Although based on observed toxldty
data, H 1s not considered to be a health risk assessment.
The purpose for developing this chronic toxldty ranking system was to
provide the U.S. EPA with the technical background required to adjust the
RQs of hazardous substances designated 1n Section 101(14) of the Comprehen-
sive Environmental Response, Compensation and Liability Act of 1980 (CERCLA
or "Superfund"). This ranking system has undergone a limited peer review
and a public review. It may'have applications to other areas of Interest to
the U.S. EPA and other regulatory agencies where ranking of chemicals based
on chronic toxldty Is desired.
-------
1. INTRODUCTION
The purpose for developing this chronic toxldty ranking system was to
provide the U.S. EPA with the technical background required to adjust the
RQs of hazardous substances deslgnaled In Section 101(14) of the Comprehen-
sive Environmental Response, Compensation and Liability Act of 1980 (CERCLA,
or "Superfund"). Section 103 of CERCLA requires Immediate notification from
any person 1n charge of a vessel or an offshore or an onshore facility who
releases an amount of a hazardous substance equal to or greater than Us RQ.
Under Section 102(b), the RQ of any hazardous substance defined 1n Section
101(14) of CERCLA 1s 1 pound unless a different RQ has been established pur-
suant to Section 311(b) of the Federal Hater Pollution Control Act. This 1s
the statutory RQ for each hazardous substance unless and until the Admin-
istrator of the U.S. EPA promulgates regulations establishing that quantity
of any hazardous substance the release of which shall be reported pursuant
to Section 103 of CERCLA. CERCLA also permits the U.S. EPA to establish a
single RQ for each hazardous substance regardless of the environmental
medium Into which the substance 1s released.
The strategy of the Office of Emergency and Remedial Response (OERR) of
OSUER 1s to adjust the Section 101(14) statutory RQs using six primary
criteria (1gn1tab1l1ty, reactivity, cardnogenldty, aquatic toxldty,
acute mammalian toxldty (oral, dermal. Inhalation) and chronic toxldty)
and three other factors (blodegradatlon, hydrolysis and photolysis).
(Chronic toxldty, for the purposes of this report, 1s defined as toxldty
due to repeated or continuous exposure from a single release or multiple
releases of a designated hazardous substance.) For each criterion, a five-
tiered rating scale Is set up corresponding with RQ values of 1, 10. 100,
1000 and 5000 pounds (X. A. B. C, 0). Since this f1ve-t1ered system was
0721B -1- 03/11/85
-------
successfully used 1n the CUA and the regulated community 1s familiar with
H, CERCLA uses this methodology. Unlike the CUA. CERCLA addresses all
media, not only water. Therefore, while the strategy selected to adjust the
RQs of the designated hazardous sJbstances pursuant to Section 101(14) of
CERCLA was based on the precedence established 1n assigning RQs pursuant to
the CUA, Instead of using aquatic toxldty as the sole criterion for estpb-
«
Ushlng RQs, health and welfare effects for other media were also used to
adjust the RQs. Each hazardous substance 1s evaluated according to the pri-
mary criteria and an RQ value Is determined for each applicable criterion.
The "primary criteria" RQ for each hazardous substance 1s the lowest value
of all the applicable criteria.
, 03/11/85
0721B •*'
-------
2. CHRONIC TOXICITY RANKING SYSTEM
The chronic toxUUy ranking system reflects two primary attributes of
each chemical:
1. The MED levels for a given" effect following chronic exposures
(mg/day for 70 kg man) by alternative environmental media (air.
water).
2. Type of effect at the HED (e.g., liver necrosis, teratogenl-
dty. etc.).
The dose rating for a given chemical 1s based upon the HEO transformed
to values ranging from 1-10 using the graph given 1n Figure 2-1. Substances
having an effect at a low dose (I.e., those that are more highly toxic) will
be given a high rating on this graph, while those requiring a high dose
(less toxic) will be given a low rating. Similarly, the rating for the
observed effect at a given dose for a given chemical will range from 1-10
depending on severity using Table 2-1. with 10 being the most severe. These
values must be assigned on a chemlcal-by-chemlca1 basis.
A final composite score (CS) 1s determined by multiplying the dose rat-
Ing by the effect rating. The possible range of CSs Is thus 1-100. Using
this scheme, only those compounds eliciting what are judged to be the most
severe effects at low levels of exposure would be assigned high CSs; com-
pounds that elicit minimal effects at high doses would be assigned low CSs.
The following text gives step-by-step details for this procedure:
s
1. Identify subchronlc or chronic NOAELs, LOAELs or FELs* based on
animal or human data from the available literature. Note the
dose/exposure and the effect.
2. Convert all NOAELs, LOAELs and FELs to units of mg/kg/day.
Inhalation, dietary or drinking water exposure data will be
converted to units of mg/kg/day doses based on the methods
outlined previously (U.S. EPA. 1980).
3. If the NOAEL, LOAEL or FEL 1s based on subchronlc exposure, a
corresponding chronic value will be estimated by dividing the
subchronlc value by 10 or less. This 1s supported experi-
mentally by Ue11 and NcColllster (1963) and HcNamara (1976).
07218 -3- 03/11/85
-------
MATIMO VALIItl FOR DOSES
i
It
t
•-
>
2
P
J-
I
1
t
• It IF IN MED < -3
• -I.BtofMCD • •• IF-J < tafMtD < S
MVj • 1 IPfefMfD > J
-3-1-1 t I ?
bfl NUMAM MED linffM*!
I
S
I
4
o
OD
\
O
oo
OS
FIGURE 2-1
Rating Values for Doses used (o Rank Chronic ToxUtty
-------
TABU Z-1
Rating Values for HOAEls. LOAEls and fCls used to Rank Chronic Toxlclty
Mimo tmcT
1 f NfVMf INDUCTION Ofl OTHER BIOCHEMICAL CHANGE WITH MO PATHOLOGIC CHANGES AMD NO
CHANGE IN ORGAN WEIGHTS.
t
BUT NO OTHER APPARENT EFFECTS.
WCKWSANWfWMTl.
• IllVtHtiett CltiULAH CHAMOtit CtOUOY HittitHO.KYOHOftCCMAIIOt.OH MTTYCHAIHm,
f NTCflOftt.ONMfTAHAilAiriTNNOA^AMEI^OfCtltMemOfOiraAMnMCTIOM. AMV
NEUROPATHY WITHOUT AIT A fit NT BEHAVIQUAL. SENSORY. ON fHYSIOlOOIC CHANGES.
f Nf CROStf. ATHOWV. MWCMTflOHIV. ON M^TAM. AST A WITH A Of ft CT AW. I Of CUE «f NT Of
ORGAN FUNCTION*. ANY NEUROPATHY WITH A MEASURABLE CHANGE IN BEHAVIORAL.
SENSORY. OR fHVSIOLOGIC ACTIVITY.
• NECROSIS. ATROPHY. HVFff RTMOmV. ON Kit TAW. ASIA WITM Df nWtTVf ONOAN UVSff UNCTION.
ANY NEUROPATHY WITH GROSS CHANOIS IN BEHAVIOR. SCNSOBV.ON MOTOR PEREORMANCE.
ANY DECREASE IN REPRODUCTIVE CAPACITY. ANY EVIDENCE O? FETOTOKICITY.
•
f -.PRONOUNCED PATHOLOGIC CHANGES WITH WVtRE ORGAN OVTOJNCTfON . ANY NEUROPATHY
'WITH LOSS or BEHAVIORAL OR MOTOR CONTROL ON LOSS OP si NSORV ABILITY. REPROOUCTTVI
DYSFUNCTION. ANY TERATOOENIC EFf-ECT WITH MATERNAL 1ONICITY.
ft DC ATM ON PRONOUNCED LIFE SHORTENING. ANY TERATOOENIC f ffECT WITHOUT PONS Of
g MATERNAL TOKICITV.
o
CD
-------
4. The «Ds based on animal data -ni.
.*. Put
KO in units of »g/day for a 70 kg man.*
5 Assign a dose rating value (IV,) to the dose associated with
the KO as described In Figure 2-1.
6 Assign an effect rating value (RVe) to the effect associated
with the "ED »s described In Table 2-1.
7. Calculate the CS as:
CS
,. „ „. than on, K» C
be selected D» the fellowln*
. If adequate cnronlc dit. are available. «1sr«g.rd
based on subchronlc data.
. If .ore than one HCD remains, select the BED uhlch 1s based
on the 'best' data.
for a given route will be used.
.." -Hh the hlgheS, CS.
»s detailed 1. the f.IMm .«"-. «.1.» «U«1f1«.«". «he^ has r«.lt
,d ,„ a reasonable 41.tr1k.t1. .f CSs for the compounds thus far
animal dose (mg/day) x ( an1mal weight
.It the animal dose 1s In.-gAg/day the^uman .dose In^g/da, equals:
x 70 kg.
animal dose (mgAg/day) x
11/02/84
07218
-------
10 The RQs are then assigned based on the following relationship
to CS:
Composite Score *Q
81-100 I
41-80 10
21-40 100
6-20 1000
1-5 5000
As detailed 1n the following section, this classification scheme has result-
ed 1n a reasonable distribution of RQs for the compounds thus far analyzed.
0721B
-7-
09/19/84
-------
3. APPLICATION OF THE «THOOOl06Y
To date. -260 compounds have been assessed using the methodology
described in Chapter 2. In the final evaluation. Insufficient data were
available to derive CSs on 81 cot-pounds. CSs were derived by analogy for 47
conpounds. and CSs were directly derived from experimental data on the 116
remaining compounds. Histograms were constructed from preliminary evalua-
tions showing the distribution of RVds. RVes and maximum CSs. and are
given 1n Figures 3-1. 3-2 and 3-3. respectively. Although the final evalua-
tion has changed several totals, the discussions concerning these histograms
and. more Importantly, the conclusions reached are the same.
A reasonably symmetric distribution of CSs -as obtained (see Figure
3-3). Histograms of the RVds and RVes were generated to better under-
stand the distribution of CSs and to suggest ways in which the methodology
could be modified to account for the available data. As Indicated In Figure
3-1. the distribution of RVds 1s skewed, markedly toward the lower values.
While this partially reflects the distribution of doses In all the available
data, many of the doses given an RV, of 1 were well In excess of 1000
mg/day. suggesting that U may be desirable to expand the RVd scale, at
least at the lower limit. In the distribution of RVes. the predominance
of RVes of 10. which indicate mortality or teratogenlc effects without
signs If maternal toxldt'y. appears to reflect the nature of the data on the
105 compounds and may not be of substantial use In modifying the effects
rating scheme. All RVfts derived for the "lOS compounds, not Just those
associated with the maximum CS. are being reanalyzed to better evaluate this
assertion.
. 09/19/84
07218 ~B"
-------
10-
FIGURE 3-1
Distribution of RVds In Raxlmum Composite Scores for 105 Compounds
0721B
-9-
09/19/84
-------
i
i !•«••• 7. ••«
FIGURE 3-2
Distribution of RV s 1n Haxlmun Composite Scores for 105 Compounds
0721B
-10-
09/19/84
-------
1-
u.16 21-2S
10-20
11-3S 41-45 >K>
S8-40 «f-K
FIGURE 3-3
ComposUe Scores for 105 Compounds
0721B
-11-
09/19/8*
-------
In addition to examining the distribution patterns of CSs, RV.s and
RV s. 1t was also determined If any correlation existed between reported-
acute and calculated-chronic toxlclty values and if any structural correla-
tions could be found vs. the chronic toxlclty values. These determinations
were deemed Important since acute mammalian toxlclty 1s a primary criterion
In establishing toxlclty rating schemes, and the Inclusion of chronic toxl-
clty might possibly be redundant. This latter concern Is possibly justified
given the relationships between acute and chronic toxUUIes noted by
HeNamara (1976). Well and HcColllster (1963) and Ue1l et al. (1969). Conse-
quently, for the 105 compounds on which maximum CSs were directly derived,
the LOcQ 1n experimental mammals was Identified from the Registry of Toxic
Effects of Chemical Substances (NIOSH, 1982) and converted to approximate
human LDc.s using the cubed root of the body weight ratios as specified In
point 4, Chapter 2. Preliminary results of these analyses are given 1n
Figures 3-4. 3-5 and 3-6 for aromatic*.^aliphatic* and Inorganics, respec-
tively. Although an attempt 1s being made to more critically review the
acute toxlclty data, and statistical analyses will be conducted, no correla-
tions between maximum CSs and minimum LD,Qs are apparent. That 1s to say,
the chronic toxlclty of a series of chemicals cannot necessarily be pre-
dicted from acute toxldtles. This Is a well established principle In the
j
field of toxicology. Therefore, the use of chronic toxlclty as a ranking
tool Is not redundant.
Although work on structural correlations Is In a very preliminary stage,
Figure 3-7 Indicates no apparent correlation between maximum CSs and Upo-
phllldty for 21 substituted benzenes. In this figure, log of the octanol/
water partition coefficient given by Hansch and Leo (1981) was used as the
Index of I1poph1l1c1ty. Analyses of other structural parameters are
currently being conducted.
0721B
-12-
09/19/84
-------
» i.O-i
4.0-1
J.I-I
2.0 H
i.o H
o.sH
e •
A
A
e CHicmo
• MITKO
A OTHER
~T i i
10 20 ao
MAXIMUM COMPOSITE SCORE
AROMATICS
10
FIGURE 3-4
Composite Scores vs. H1n1«i» LD5(,s for 28 Aromatic Compounds
07218
-13-
09/19/84
-------
I
f 0-,
4.0-
....
3.0-
2.8-
f
I
2.0-
o o
-
1.0-
O.i-
• CMLORD
o OTHER
10 20 30
COMPOSITE SCORE
ALIPHATIC*
T
40
n
•0
FIGURE 3-5
Composite Scores vs. MlnlM- Utf for 33 Aliphatic Compounds
0721B
-14-
09/19/84
-------
I
4.0i
I.S-
3.0-
2.5-
2.0-
1.5-
1.0-
0.5-
I
10
T
20
1
30
I
40
I
K
COMPOSITE SCORE
INORGANICS
FIBURE 3-6
Composite Scores vs. (Un^mum LDq_s for. 25 Inorganic Compounds
0721B
-15-
09/19/84
-------
i
n-
n-
t-
i
i
I
•
MGURE 3-7
ComposUe Scores vs. Log P for ?1-Substituted Ben/enes
-------
4. SOME CONCEPTUAL CONSIDERATIONS
The development of the quant Hat We rating scheme for effects (see Table
2-1} and the combination of this scheme with a quantitative estimate of dose
(see Figure 2-1) represent a potentially useful Innovation 1n the analysis
and assessment of chronic toxlclty data. The effects rating scheme 1s based
on a combination of biochemical, hlstologU. physiologic and gross effects
arranged 1n Increasing order of severity. While the rating values assigned
to each type of effect are essentially arbitrary, rating values of 1 to 3 or
4 have been generally regarded as NOAELs, 5-7 as LOAELs and 7-10 as FELs.
In an attempt to allow for the necessary scientific judgment, the descrip-
tion of the effects Is Intentionally not too detailed. Although some addi-
tional modifications and expansions of the scheme may be desirable, the
application of the scheme to date suggests that It would be a mistake to
attempt to catalogue every possible specific effect which chemicals can
Induce and to associate each effect with a quantitative value. Also, the
^ ^
ranking scheme Is not organ specific. While the ICRP (1977. 1979) has
addressed the problem of organ sensitivity to specific radlonuc.Hdes. and
their approach nay be useful In predicting the most sensitive organ for cer-
tain radlonucUdes, the effects rating scheme presented 1n Table 2-1 Is
Intended to be applicable across target sites.
Nonetheless, the variation In organ sensitivities 1s an Important factor
In defining the BED associated with a given dose level for a particular
toxicant. This 1s Illustrated In Figure 4-1, hypothetical data 1n which
NOELs. NOAELs. LOAELs and FELs are plotted on a RVd vs. RV£ diagram. At
low doses (I.e., high RV s) only NOELs are observed. As the dose In-
creases (I.e.. the RV^ decreases) effect ratings become Increasingly high.
07218 -17- 09/19/84
-------
ECIESAMD
tOAIL *
NOAEL
V 1-
MOEL
FIGURE 4-1
Hypothetical MOELs. HOAELs. LOAELS and FELs
0721B
.18-
09/19/84
-------
However, at a gl«n «f'«t level, ~Utpl. points can b« e.pected a, the
„„« increases -1th the point furthest to the right representing the »«t
,,r,s1t1« organ or species or 1r.cr.as1n, Incidence of the tfftel. The slope
„, the Hn, oravn to the right of the effect I...1 P«'nts an. l.l.r,Ktl.g
the .-axis to the left of the .OILS represents the 1.1— '«'"» " *»«
S,«r1t, of effect .Hh Increasing dose and «» be »«-d the apparent
,e«r1t, slope. Ho«..r. as Illustrated In ,**. M. the apparent sever-
uy .Lp. «y »e a composite of .wtrll, slopes for specific organs and/or
spedes.
,a 09/19/84
07218 -19'
-------
TftUE CUKVES rOM
PEMENTOHC
Oft SPECIES
MED CURVE
FIGURE 4-2
Apparent BED Curve as a Composite of True Curve for Different
Organs or Species
0721B
-20-
09/19/84
-------
5. MOIFICATION AND APPLICATIONS
The basic methodology outlined 1n Chapter 2 of thU report was developed
In response to a specific need. Having applied the methodology to -260
compounds, explored some of the conceptual considerations Inherent 1n the
methodology, and received preliminary comments from scientists both within
and outside of the Agency, the following modifications have been considered:
1. Derive the CS as the SUB rather than the product of RVd and
RVe.
2. HodUy the dose rating scheme to cover a wider range of doses.
As Indicated 1n Figure 3-1 and discussed 1n Chapter 3 of this
report, several studies from which maximum CSs were derived
Involved dose levels well 1n excess of 1000 mg/day. yet were
still assigned the minimum RVd of 1. The revision currently
under consideration would cover 10 log dose units from 10*
mg/day yielding an RVd of 1 to 10'- mg/day yielding an
RVd of 10.
3. Review and utilize data on pharmacok1net1cs and other relevant
data to obtain more reasonable estimates of absorption from
oral and Inhalation routes. Th5 -current approach assumes 50%
absorption from Inhalation exposures and 100X absorption from
oral exposures. These assumptions were made rather than review-
Ing absorption data In an effort to minimize the cost of this
project. However, these assumptions may be misleading particu-
larly for metals.
4. Modify the effects rating scheme giving more specific guidance
and/or altering some of the criteria.
In addition, the application of the methodology has suggested that the
RV vs. RV plot may be useful to other Agency Offices 1n the estimation
e ' d
of NOELs or NOAELs- used 1n the derivation of ADIs. Kushner et al. (1983)
have recently reviewed the use of AOIs 1n Superfund Implementation, and
Stara and coworkers (Stara et al.. 1980, 1981; Oourson and Stara. 1983) have
reviewed more general problems with the estimation and application of ADIs
In the regulatory process. Two problems Involving the use and derivation of
AOIs may be at least partially alleviated by using the RVd vs. RVg plot.
0721B
.21- 09/19/84
-------
First, the Agency frequently has been criticized for not using all of the
available data 1n estimating the NOAEL from which ADIs are derived. Using
the RV vs. RV, plot along with an appropriate statistical method for
estimating the apparent severity tlope and the ^-Intercept (I.e., maximum
NOEL), all of the available subchronlc and chronic data could be used.
Second, examples have been encountered where no suitable NOAELs or LOAELs
are available from which an ADI could be derived. In such cases, data on
FELs and NOELs could be used to estimate either the maximum NOEL or a suit-
able NOAEL (e.g., dose associated with an RVg of 1 or 2). Again, this
would be dependent on the development of an appropriate mathematic model to
estimate the severity slope and x-1ntercept.
In addition to these Immediate applications, U may be desirable to con-
sider expanding the RV, and RV plot to Include axes for duration of
exposure, species and Incidence of response. This, however, would be a very
complex undertaking that should await further development and applications
of the basic RVft vs. RVd plot.
0721B -22- 03/11/85
-------
6. REFERENCES
Dourson, H.L. and J.F. Stara. 1983. Regulatory history and experimental
support of uncertainty (safety)' factors. Reg. T.«1c.1. Pharmacol.
3: 224-238.
Hansch, C. and A.J. Leo. 1981. Pomona College Medicinal Chemistry Project.
Seaver Chemistry Laboratory. Claremont, CA.
ICRP (international Collision on Radlologlc Protection). 1977. Recommen-
dations of the international Contusion on radlologlc protection. ICRP
Pub.l. 26. Pergamon Press. Elmsford, NY. p. 50.
ICRP (International Comlsslon on Radlologlc Protection), 1979. -Radio-'
nucllde release Into the environment: Assessment of doses to man. ICRP
Publ. 29. Pergamon Press. Elmsford. NY. p. 76.
Ku.hn.r; L.H.. «.C. Wards and V. Pong. 1983. The potential use of the ADI
in Superfund Implementation, mtre Corporation. McLean. VA. p. 68.
Samara. I.P. 1976. Concepts 1n health evaluation of commercial and
industrial chemicals. In: New Concepts 1n Safety Evaluation: Advances in
Modern Toxicology. 1(1): 61-115.
NIOSH (National Institute for Occupational Safety and Health,. 1982.
Registry of Toxic Effects of Chemical Substances. U.S. DHE«. Cincinnati. OH.
,, 03/11/85
0721B •"'
-------
Stara, J.F., D. Kello and P.R. Durkln. 1)80. Human health hazards associ-
ated with chemical contamination of aquatic environment. Environ. Health
Perspect. 34: 145-188.
Stara, J.F., H.I. Dour son and C.T. DeRosa. 1981. Water quality criteria:
Methodology and applications. In: Conference Proceedings: Environmental
Risk Assessment: How New Regulations will Affect the Utility Industry.
Electric Power Research Institute. Palo Alto, CA.
U.S. EPA. 1980. Guidelines and methodology used In the preparation of
health effects assessment chapters of the consent decree water quality
criteria. Federal Register. 45: 79318-79379.
Hell, C.S. and D.O. HcColHster. 1963. Safety evaluation of chemicals.
Relationship between short- and long-term feeding studies 1n designing an
effective toxUUy test. Agrlc. Food Chen. 11: 486-491.
Well. C.S.. H.O. Uoodslde, J.R. Bernard and C.P. Carpenter. 1969. Rela-
•
tlonshlps between slngle-peroral. one-week, and ninety-day rat feeding
studies. Toxlcol. Appl. Pharmacol. 14: 426-431.
07218
-24-
09/19/84
------- |