v>EPA
EPA/63 5/R-08/006F
www.epa.gov/iris
TOXICOLOGICAL REVIEW
OF
ETHYLENE GLYCOL
MONOBUTYL ETHER (EGBE)
(CAS No. 111-76-2)
In Support of Summary Information on the
Integrated Risk Information System (IRIS)
March 2010
U.S. Environmental Protection Agency
Washington, DC

-------
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
11

-------
CONTENTS—TOXICOLOGICAL REVIEW OF ETHYLENE GLYCOL MONOBUTYL
ETHER (CAS No. 111-76-2)
LIST OF TABLES	VI
LIST OF FIGURES	VIII
LIST 01 ABBREVIATIONS AM) ACRONYMS	IX
FOREWORD	XI
AUTHORS, CONTRIBUTORS, AND REVIEWERS	XII
1.	INTRODUCTION	1
2.	CHEMICAL AM) PHYSICAL INFORMATION	3
3.	TOXICOKINETICS	4
3.1.	ABSORPTION AM) DISTRIBUTION	4
3.2.	METABOLISM AM) ELIMINATION	5
4.	HAZARD IDENTIFICATION	15
4.1.	STUDIES IN HUMANS—EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS	15
4.2.	SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIO AS SAYS IN
ANIMALS—ORAL AM) INHALATION	18
4.2.1.	Subchronic Studies	18
4.2.1.1.	Oral	18
4.2.1.2.	Inhalation	24
4.2.2.	Chronic Studies and Cancer Bioassays	28
4.2.2.1. Inhalation	28
4.3.	REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION	33
4.4.	OTHER STUDIES	37
4.4.1.	Acute and Short-Term Exposure Studies	37
4.4.2.	Dermal Exposure Studies	41
4.4.3.	Ocular Exposure Studies	42
4.4.4.	Genotoxicity	43
4.4.5.	Immunotoxicity	46
4.4.6.	Other In Vitro Studies	48
4.5.	SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS AND
MODE OF ACTION: ORAL AND INHALATION	50
4.6.	EVALUATION 01 CARCINOGENICITY	55
4.6.1.	Summary of Overall Weight of Evidence	55
4.6.2.	Synthesis of Human, Animal, and Other Supporting Evidence	56
4.6.3.	Mode-of-Action Information	58
4.6.3.1. Hypothesized MOA for Liver Tumor Development in Male Mice	58
4.6.3.1.1.	Temporal association and species specificity	60
4.6.3.1.2.	Dose-response relationships	61
4.6.3.1.3.	Biological plausibility and coherence of the database	62
4.6.3.1.4.	Relevance of the hypothesized MOA to humans	64
in

-------
4.6.3.1.5. Other possible MOAs for liver tumor development in male mice	65
4.6.3.2.	Hypothesized MOA for Forestomach Tumor Development in Female Mice	67
4.6.3.2.1.	Temporal association	69
4.6.3.2.2.	Dose-response relationships	69
4.6.3.2.3.	Biological plausibility and coherence of the database	70
4.6.3.2.4.	Relevance of the hypothesized MOA to humans	71
4.6.3.2.5.	Other possible MOAs for forestomach tumor development in female mice.71
4.6.3.3.	Conclusions about the Hypothesized MOAs	72
4.7. SUSCEPTIBLE POPULATIONS	73
4.7.1.	Possible Childhood Susceptibility	75
4.7.2.	Possible Gender Differences	76
5. DOSE-RESPONSE ASSESSMENTS	78
5.1.	INHALATION REFERENCE CONCENTRATION (RFC)	78
5.1.1.	Choice of Principal Study and Critical Effect—with Rationale and Justification	78
5.1.2.	Methods of Analysis—Including Models (PBPK, BMD, etc.)	82
5.1.2.1.	BMD Approach Applied to Hemosiderin Staining Data	83
5.1.2.2.	Selection of the POD	86
5.1.3.	RfC Derivation—Including Application of Uncertainty Factors	87
5.1.4.	RfC Comparison Information	89
5.1.5.	Previous Inhalation Assessment	90
5.2.	ORAL REFERENCE DOSE (RID)	90
5.2.1.	Choice of Principal Study and Critical Effect—with Rationale and Justification	90
5.2.2.	Methods of Analysis—Including Models (PBPK, BMD, etc.)	93
5.2.2.1.	BMD Approach Applied to Hemosiderin Endpoint	93
5.2.2.2.	Route-to-Route Extrapolation from Inhalation Data	93
5.2.2.3.	Selection of the POD	93
5.2.3.	RfD Derivation—Including Application of UFs	94
5.2.4.	RfD Comparison Information	95
5.2.5.	Previous Oral Assessment	96
5.3.	UNCERTAINTIES IN THE DERIVATION OF THE INHALATION REFERENCE
CONCENTRATION (RfC) AND ORAL REFERENCE DOSE (RfD)	97
5.3.1.	Choice of Endpoint	98
5.3.2.	Choice of Dose Metric	98
5.3.3.	Use of BMC Approach	99
5.3.4.	Choice of Model for BMCL Derivations	99
5.3.5.	Choice of Animal to Human Extrapolation Method	99
5.3.6.	Route-to-Route Extrapolation	99
5.3.7.	Statistical Uncertainty at the POD	100
5.3.8.	Choice of Bioassay	100
5.3.9.	Choice of Species/Gender	100
5.3.10.	Human Relevance of Noncancer Responses Observed in Mice	101
5.3.11.	Human Population Variability	101
5.4.	CANCER ASSESSMENT	101
5.4.1.	Quantification for Oral and Inhalation Cancer Risk	103
5.4.2.	Uncertainties in Cancer Risk Assessment	103
5.4.2.1.	Choice of Low-Dose Extrapolation Method	104
5.4.2.2.	Human Relevance of Cancer Responses Observed in Mice	105
5.5.	POTENTIAL IMPACT OF SELECT UNCERTAINTIES ON THE RFC	105
iv

-------
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND
DOSE RESPONSE	108
6.1.	HUMAN HAZARD POTENTIAL	108
6.2.	DOSE RESPONSE	110
6.2.1.	Noncancer—Inhalation	110
6.2.2.	Noncancer—Oral	Ill
6.2.3.	Cancer—Oral and Inhalation	112
7. REFERENCES	113
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC COMMENTS
AND DISPOSITION	A-l
APPENDIXB. PBPKMODELS (CORLEYET AI..| 1994; 1997:20051)	B-l
APPENDIX C. RFD AND RFC DERIVATION OPTIONS	C-l
v

-------
LIST OF TABLES
Table 2-1. Physical and chemical properties of EGBE	3
Table 3-1. Summary of species-specific toxicokinetic parameters	8
Table 4-1. Hematology and hemosiderin data from the 13-week drinking water exposure to
EGBE in F344 rats	21
Table 4-2. Incidence11 and severity of selected histopathological changes from the 13-week
drinking water exposure to EGBE in F344 rats and mice	23
Table 4-3. Hematology and hemosiderin data from a 14-week inhalation study of EGBE in F344
rats	26
Table 4-4. Hematology and hemosiderin data from a 14-week inhalation study of EGBE in
B6C3Fi mice	27
Table 4-5. Selected female and male rat and mouse nonneoplastic effects from the 2-year
chronic EGBE inhalation study	29
Table 4-6. Comparison of female and male rat and mouse Hct (manual) values from 3- and 12-
month inhalation exposures to EGBE	31
Table 4-7. Summary of genotoxicity studies on EGBE, BAL, and BAA	44
Table 4-8. Incidence of liver hemangiosarcomas and hepatocellular carcinomas in studies of
NTP chemicals that caused increased hemosiderin in Kupffer cells in male mice ...62
Table 5-1. Results of candidate studies	79
Table 5-2. Female and male rat and mouse liver hemosiderin staining incidence and RBC counts
from subchronic and chronic EGBE inhalation studies	80
Table 5-3. Summary of PBPK models	83
Table 5-4. AUC BAA blood concentrations measured at 12 months in both genders of B6C3Fi
mice and F344 rats	84
Table 5-5. Comparison of BMC/BMCL values for male and female rat liver hemosiderin
staining data from inhalation chronic study using measured blood AUC (12 months)
of the EGBE metabolite BAA as a common dose metric	85
Table 5-6. Comparison of BMC/BMCL values for male and female mouse liver hemosiderin
staining data from inhalation chronic study using measured blood AUC (12 months)
of the EGBE metabolite BAA as a common dose metric	86
Table 5-7. Subchronic 91-day drinking water studies in rats and mice	91
Table 5-8. Summary of uncertainty in the EGBE noncancer and cancer risk assessments	97
Table 5-9. Illustrative potency estimates for tumors in mice, using a linear analysis approach 104
Table B-l. Selected parameters used in the PBPK model for EGBE developed by Corley et al.
(1997, 041984; 2005, 100100)	B-6
Table C-l. Summary of PBPK models	C-2
Table C-2. Model estimates of BAA blood levels in female rats following inhalation
exposures	C-2
vi

-------
Table C-3. Estimated Cmax for BAA in blood for humans continuously exposed to varying
concentrations of EGBE	C-3
Table C-4. Comparison of BMC/BMCL values for female rat RBC count data from a 14-week
subchronic inhalation study3, using modeled blood Cmax (3 months) of the EGBE
metabolite BAA as a common dose metric	C-5
Table C-5. AUC BAA blood concentrations measured at 12 months in both genders of B6C3Fi
mice and F344 rats	C-12
Table C-6. Comparison of BMC/BMCL values for male and female rat liver hemosiderin
staining data from inhalation chronic study using measured blood AUC (12 months)
of the EGBE metabolite BAA as a common dose metric	C-13
Table C-7. Comparison of BMC/BMCL values for male and female mouse liver hemosiderin
staining data from inhalation chronic study using measured blood AUC (12 months)
of the EGBE metabolite BAA as a common dose metric	C-14
Table C-8. PBPK model estimates of BAA Cmax blood levels and incidence of forestomach
epithelial hyperplasia in female mice	C-22
Table C-9. BMDS model estimates of Cmax BMDio and BMDLio values for forestomach
epithelial hyperplasia in female mice	C-23
Table C-10. Female mouse Cmax values for various time points of the NTP (2000, 196293) study
estimated by the Lee et al. (1998, 041983) model	C-23
Table C-l 1. Estimated Cmax for BAA in blood for humans continuously exposed to varying
drinking water concentrations of EGBE	C-24
Table C-12. Estimated Cmax for BAA in blood for humans continuously exposed to varying
concentrations of EGBE	C-24
Table C-13. Modeled estimates of BAA in human blood exposed to EGBE in water	C-28
Table C-14. Model estimates of BAA blood levels in female rats following oral exposures ...C-29
Table C-15. Comparison of female rat RBC count and MCV BMD/BMDL values from an oral
subchronic study using modeled blood Cmax (3 months) of the EGBE metabolite
BAA as a common dose metric	C-30
vii

-------
LIST OF FIGURES
Figure 3-1. Proposed metabolic scheme of EGBE in rats and humans	6
Figure 4-1. Simulated concentrations of EGBE, BAL, and BAA in liver tissues of female mice
exposed via inhalation for 6 hours to 250 ppm EGBE	66
Figure 5-1. PODs for selected endpoints with corresponding applied UFs and derived RfC	89
Figure 5-2. PODs for selected endpoints with corresponding applied UFs and derived RfD	96
Figure 5-3. Potential impact of select uncertainties on the RfC for EGBE	106
Figure B-l. PBPK model of Corley et al. (1994, 041977)	B-2
Figure B-2. PBPK model of Corley et al. (2005, 100100)	B-5
Figure C-l. Hill Model run of female rat RBC count versus Cmax BAA from a 14-week
inhalation study	C-6
Figure C-2. Multistage Model run of hemosiderin deposition in male rats versus AUC BAA at
12 months from a 2-year inhalation study	C-15
Figure C-3. Log-Logistic Model run of hemosiderin deposition in female rats versus AUC BAA
at 12 months from a 2-year inhalation study	C-18
Figure C-4. Log-Logistic Model run of forestomach epithelial hyperplasia in female mice versus
Cmax BAA from a 2-year inhalation study	C-25
Figure C- 5. Hill Model run of female rat RBC count versus Cmax BAA from a 3-month oral
study	C-31

-------
LIST OF ABBREVIATIONS AND ACRONYMS
ADH
alcohol dehydrogenase
AIC
Akaike information criterion
ALDH
aldehyde dehydrogenase
AUC
area under the curve
BAA
2-butoxyacetic acid
BAL
butoxyacetaldehyde
BMC
benchmark concentration
BMCL
benchmark concentration, 95% lower bound
BMD
benchmark dose
BMDL
benchmark dose, 95% lower bound
BMDS
benchmark dose software
BMR
benchmark response
CASRN
Chemical Abstracts Service Registry Number
CHO
Chinese hamster ovary
CHR
contact hypersensitivity response
CI
confidence interval
Cls
clearance rate
Cmax
maximum concentration
con-A
concanavalin-A
COP
cardiac output
DNA
deoxyribonucleic acid
EA
ethyl aery late
ECETOC
European centre for ecotoxicology and toxicology of chemicals
EG
ethylene glycol
EGBE
ethylene glycol monobutyl ether
EGEE
ethylene glycol ethyl ether
EGME
ethylene glycol methyl ether
GD
gestational day
GFR
glomerular filtration rate
G6PD
glucose-6-phosphate dehydrogenase
GSH
glutathione
Hb
hemoglobin
Hct
hematocrit
HEC
human equivalent concentration
HED
human equivalent dose
HH
hereditary hemochromatosis
Hp
haptoglobin
i.p.
intraperitoneal
i.v.
intravenous
IRIS
Integrated Risk Information System
KLH
keyhole limpet hemocyanin
LOAEL
lowest-observed-adverse-effect level
MAA
2-methoxyacetic acid
MCH
mean corpuscular hemoglobin
MCHC
mean corpuscular hemoglobin concentration
MCV
mean cell volume
IX

-------
ME
2-methoxyethanol
MN
micronuclei
MOA
mode of action
NK
natural killer
NOAEL
no-observed-adverse-effect level
NTP
National Toxicology Program
NZW
New Zealand white
8-OHdG
8-hydroxydeoxyguanosine
OR
osmotic resistance
OXA
oxazolone
PBPK
physiologically based pharmacokinetic
PFC
plaque-forming cell
POD
point of departure
RBC
red blood cell
RfC
reference concentration
RfD
reference dose
ROS
reactive oxygen species
s.c.
subcutaneous
SCE
sister chromatid exchange
SD
standard deviation
tVz
half-life
TNFa
tumor necrosis factor-alpha
TNP-LPS
trinitrophenyl-lipopolysaccharide
UF
uncertainty factor
U.S. EPA
U.S. Environmental Protection Agency
vd
volume of distribution
WBC
white blood cell
X

-------
FOREWORD
The purpose of this Toxicological Review is to provide scientific support and rationale
for the hazard and dose-response assessment in IRIS pertaining to chronic exposure to ethylene
glycol monobutyl ether (EGBE). It is not intended to be a comprehensive treatise on the
chemical or toxicological nature of EGBE.
The intent of Section 6, Major Conclusions in the Characterization of Hazard and Dose
Response, is to present the major conclusions reached in the derivation of the reference dose,
reference concentration and cancer assessment, where applicable, and to characterize the overall
confidence in the quantitative and qualitative aspects of hazard and dose response by addressing
the quality of data and related uncertainties. The discussion is intended to convey the limitations
of the assessment and to aid and guide the risk assessor in the ensuing steps of the risk
assessment process.
For other general information about this assessment or other questions relating to IRIS,
the reader is referred to EPA's IRIS Hotline at (202) 566-1676 (phone), (202) 566-1749 (fax), or
hotline.iris@epa.gov (email address).
XI

-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHEMICAL MANAGERS/AUTHORS
Paul Reinhart, Ph.D., DABT
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Jeffrey Gift, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Angela Howard, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
REVIEWERS
This document was provided for review to EPA scientists, interagency reviewers from
other federal agencies and White House offices, and the public, and peer reviewed by
independent scientists external to EPA. A summary and EPA's disposition of the comments
received from the independent external peer reviewers and from the public is included in
Appendix A.
INTERNAL EPA REVIEWERS
Jane Caldwell, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
IIa Cote, Ph.D., DABT
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Lynn Flowers, Ph.D., DABT
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Gary Foureman, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Xll

-------
Jennifer Jinot, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Connie Meacham, M.S.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Reeder Sams II, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
John Vandenberg, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Debra Walsh, M.S.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
John Whalan
National Center for Environmental Assessment
U.S. Environmental Protection Agency
EXTERNAL PEER REVIEWERS
David Jollow, Ph.D.
Professor Emeritus
Medical University of South Carolina
Michael Pereira, Ph.D.
College of Medicine and Public Health
Ohio State University
Andrew Salmon, Ph.D.
Senior Toxicologist
Office of Environmental Health Hazard Assessment
California Environmental Protection Agency
Fletcher Hahn, DVM, Ph.D.
Scientist Emeritus
Lovelace Respiratory Research Institute
Rochelle Tyl, Ph.D.
Senior Fellow
Life Sciences and Toxicology
RTI International
Xlll

-------
D. Alan Warren, M.PH., Ph.D.
Academic Program Director
Environmental Health Sciences
University of South Carolina Beaufort
Gregory Travlos, DVM, DACVP
Veterinary Medical Officer
National Institute of Environmental Health Sciences
xiv

-------
1. INTRODUCTION
This document presents background information and justification for the Integrated Risk
Information System (IRIS) Summary of the hazard and dose-response assessment of ethylene
glycol monobutyl ether (EGBE). IRIS Summaries may include oral reference dose (RfD) and
inhalation reference concentration (RfC) values for chronic and other exposure durations, and a
carcinogenicity assessment.
The RfD and RfC, if derived, provide quantitative information for use in risk assessments
for health effects known or assumed to be produced through a nonlinear (presumed threshold)
mode of action (MO A). The RfD (expressed in units of mg/kg-day) is defined as an estimate
(with uncertainty spanning perhaps an order of magnitude) of a daily exposure to the human
population (including sensitive subgroups) that is likely to be without an appreciable risk of
deleterious effects during a lifetime. The inhalation RfC (expressed in units of mg/m3) is
analogous to the oral RfD, but provides a continuous inhalation exposure estimate. The
inhalation RfC considers toxic effects for both the respiratory system (portal of entry) and for
effects peripheral to the respiratory system (extrarespiratory or systemic effects). Reference
values are generally derived for chronic exposures (up to a lifetime), but may also be derived for
acute (< 24 hours), short-term (>24 hours up to 30 days), and subchronic (>30 days up to 10% of
lifetime) exposure durations, all of which are derived based on an assumption of continuous
exposure throughout the duration specified. Unless specified otherwise, the RfD and RfC are
derived for chronic exposure duration.
The carcinogenicity assessment provides information on the carcinogenic hazard
potential of the substance in question and quantitative estimates of risk from oral and inhalation
exposure may be derived. The information includes a weight-of-evidence judgment of the
likelihood that the agent is a human carcinogen and the conditions under which the carcinogenic
effects may be expressed. Quantitative risk estimates may be derived from the application of a
low-dose extrapolation procedure. If derived, the oral slope factor is a plausible upper bound on
the estimate of risk per mg/kg-day of oral exposure. Similarly, a plausible inhalation unit risk is
an upper bound on the estimate of risk per (j,g/m3 air breathed.
Development of these hazard identification and dose-response assessments for EGBE has
followed the general guidelines for risk assessment as set forth by the National Research Council
(NRC, 1983, 194806). U.S. Environmental Protection Agency (U.S. EPA) Guidelines and Risk
Assessment Forum Technical Panel Reports that may have been used in the development of this
assessment include the following: Guidelines for the Health Risk Assessment of Chemical
Note: Hyperlinks to the reference citations throughout this document will take you to the NCEA HERO database (Health and
Environmental Research Online) at http://epa.gov/hero. HERO is a database of scientific literature used by U.S. EPA in the process of
developing science assessments such as the Integrated Science Assessments CISA1 and the Integrated Risk Information System (IRIS).
l

-------
Mixtures (U.S. EPA, 1986, 001468). Guidelines for Mutagenicity Risk Assessment (U.S. EPA,
1986, 001466). Recommendations for and Documentation of Biological Values for Use in Risk
Assessment (U.S. EPA, 1988, 064560). Guidelines for Developmental Toxicity Risk Assessment
(U.S. EPA, 1991, 008567) Interim Policy for Particle Size and Limit Concentration Issues in
Inhalation Toxicity (U.S. EPA, 1994, 076133). Methods for Derivation of Inhalation Reference
Concentrations and Application of Inhalation Dosimetry (U.S. EPA, 1994, 006488). Use of the
Benchmark Dose Approach in Health Risk Assessment (U.S. EPA, 1995, 005992). Guidelines for
Reproductive Toxicity Risk Assessment (U.S. EPA, 1996, 030019). Guidelines for Neurotoxicity
Risk Assessment (U.S. EPA, 1998, 030021). Science Policy Council Handbook. Risk
Characterization (U.S. EPA, 2000, 052149). Benchmark Dose Technical Guidance Document
(U.S. EPA, 2000, 052150). Supplementary Guidance for Conducting Health Risk Assessment of
Chemical Mixtures (U.S. EPA, 2000, 004421). A Review of the Reference Dose and Reference
Concentration Processes (U.S. EPA, 2002, 088824). Guidelines for Carcinogen Risk Assessment
(U.S. EPA, 2005, 086237), Supplemental Guidance for Assessing Susceptibility from Early-Life
Exposure to Carcinogens (U.S. EPA, 2005, 088823). Science Policy Council Handbook: Peer
Review (U.S. EPA, 2006, 194566). and A Framework for Assessing Health Risks of
Environmental Exposures to Children (U.S. EPA, 2006, 194567).
The literature search strategy employed for this compound was based on the Chemical
Abstracts Service Registry Number (CASRN) and at least one common name. Any pertinent
scientific information submitted by the public to the IRIS Submission Desk was also considered
in the development of this document. The relevant literature was reviewed through December
2008.
2

-------
2. CHEMICAL AND PHYSICAL INFORMATION
EGBE is also known as 2-butoxyethanol. EGBE is widely used as a solvent in various
applications, such as in surface coatings, spray lacquer, quick-dry lacquers, enamels, varnishes,
varnish removers, latex paint, metal cleaners, and in commercially available cleaning products.
EGBE has been estimated to range in concentration from 1 to 30% (volume/volume) in industrial
and commercial products. The average concentration of EGBE in household products in 1977
was 2.6%. EGBE is a high production volume chemical with an estimated 390 million pounds
produced in the United States in 1992 (NTP, 2000, 196293). Some relevant physical and
chemical properties of EGBE are shown in Table 2-1.
Table 2-1. Physical and chemical properties of EGBE
CASRN
111-76-2
Empirical formula
c4h9-o-ch2ch2-oh
Molecular weight
118.2
Vapor pressure
0.88 mm Hg at 25°C (about 1,200 ppm)
Water solubility
Miscible
Log Kow
0.81
Henry's law constant
2.08 x 10"7-2.08 x 10"8 atm-m3/mole (25°C)
Flash point
62°C (closed cup); 70°C (open cup)
Conversion factor
1 ppm = 4.83 mg/m3; 1 mg/m3 = 0.207 ppm
Source: HSDB (2005, 5975371
EGBE exists as a colorless liquid at ambient temperature and pressure. Its evaporation
rate relative to butyl acetate is 0.08; thus, it is considered a "slow evaporator." It is miscible in
water and partitions about equally between phases of octanol and water. Considering the
magnitude of the octanol: water partition coefficient (~ 7 :1), it is unlikely that EGBE
bioaccumulates. Based on the magnitude of the Henry's law constant, partitioning of EGBE
between water and air greatly favors the water phase.
Note: Hyperlinks to the reference citations throughout this document will take you to the NCEA HERO database (Health and
Environmental Research Online) at http://epa.gov/hero. HERO is a database of scientific literature used by U.S. EPA in the process of
developing science assessments such as the Integrated Science Assessments (ISA) and the Integrated Risk Information System (IRIS).
3

-------
3. TOXICOKINETICS
3.1. ABSORPTION AND DISTRIBUTION
EGBE is absorbed and rapidly distributed in humans following inhalation, ingestion, or
dermal exposure. Kumagai et al. (1999, 100171) examined 10 polar organic solvents, including
EGBE, during short-term inhalation by humans. Four healthy male research subjects inhaled
25 ppm EGBE for 10 minutes; the mean uptake was 79.7% in the last 5 minutes of EGBE
respiration.
Johanson and Boman (1991, 006759) attempted to define the relative importance of the
skin to the total absorption of EGBE vapors by humans by comparing mouth-only and body-only
exposures. Four research subjects were exposed to 50 ppm EGBE mouth-only for 2 hours,
followed by 1 hour of no exposure, then 2 hours of 50 ppm body-only exposure (i.e., exposed in
a chamber while breathing fresh air via a respirator). Blood samples were collected periodically
for analysis of EGBE under the assumption that the finger-prick blood samples represented
mixed arterial blood. Since the areas under the curve (AUCs) for the concentration of EGBE in
the subjects' blood samples following body-only exposures were three- to fourfold greater than
following mouth-only exposure, the authors concluded that the skin accounted for approximately
75% of the total uptake of EGBE in a whole-body exposure.
Corley et al. (1994, 041977) suggested that Johanson and Boman's (1991, 006759)
conclusion of greater absorption of EGBE vapor through the skin than from the respiratory tract
was inconsistent with the physiological differences (relative surface area, blood perfusion,
barrier thickness) favoring absorption of vapors through the lungs. They reanalyzed the kinetic
data of Johanson and Boman, assuming that the finger-prick blood samples represented venous
blood draining the skin prior to mixing systemically. These revised calculations resulted in
dermal uptake contributing no more than 22% of the total uptake of EGBE in a whole-body
exposure at average temperature and humidity (skin permeability coefficient of 3 cm/hour),
assuming no clothing to hinder absorption.
To provide experimental validation of the skin's role in the uptake of EGBE vapors,
Corley et al. (1997, 041984) conducted a study in which human research subjects exposed one
arm to 50 ppm [13C]-EGBE for 2 hours. Catheters installed in the antecubital vein of the
unexposed arm served as the primary site for blood collection, which was analyzed for both
EGBE and 2-butoxyacetic acid (BAA). Finger-prick blood samples were collected from the
exposed arm at the end of the 2-hour exposure. If Johanson and Boman's (1991, 006759)
Note: Hyperlinks to the reference citations throughout this document will take you to the NCEA HERO database (Health and
Environmental Research Online) at http://epa.gov/hero. HERO is a database of scientific literature used by U.S. EPA in the process of
developing science assessments such as the Integrated Science Assessments (ISA) and the Integrated Risk Information System (IRIS).
4

-------
assumption that finger-prick blood samples represented systemic arterial blood was correct, then
the concentrations of EGBE and BAA in the finger-prick blood samples taken from the exposed
arm at the end of the 2-hour exposure should have been comparable to the corresponding
catheter sample taken from the unexposed arm. This was not the case, since the concentration of
EGBE averaged nearly 1,500-fold higher in the finger-prick blood samples than in the samples
collected from the unexposed arm, confirming the potential for portal of entry to have a major
effect when using the finger-prick sampling technique.
3.2. METABOLISM AND ELIMINATION
The metabolism of EGBE has been studied extensively in rodents, particularly in rats,
and the large body of literature on this subject has been thoroughly reviewed
(Commonwealth of Australia, 1996, 594448; ECETOC, 1994, 594447). Proposed pathways for
the metabolism of EGBE in rats and humans are presented in Figure 3-1. The principal products
from metabolic processes in rats or humans are BAA (all species) and the glutamine or glycine
conjugate of BAA (humans). Other potential metabolic products, such as the glucuronide
conjugate of EGBE, ethylene glycol (EG), butoxyacetaldehyde (BAL), and CO2 are minor
metabolites or are transitory in nature (e.g., BAL) and do not accumulate in blood, tissues, or
excreta.
5

-------
CO,
t
carboligase oxidase dehydrogenase
CH:i CH2 CH2CH2OH + HOCH2CH2OH
(Butanol)	(Ethylene Glycol)
CH3 CH2 CH2CH2 OCH2CH2 O-Gluc
(EGBE - Glucuronide)
(Rats Only?)
(Rats Only?)
dealkylase
CH3CH2CH2CH2OCH2CH2O-SQ3H
(EGBE - Sulfate)
(Rats Only?)
CH:, CH2 CH2 CH2 OCH2CH2 OH
(EGBE)
(Rats and Human)
1
alcohol dehydrogenase
CH:,CH2CH2CH2OCH2CHO
(BAL)
CH3CH2CH2CH20CH2C02GIU
(BAA - Glutamine)
(Human Only)
aldehyde dehydrogenase
CH3CH2CH2CH2 0CH2C02 -Gly
(BAA - Glycine)
(Human Only)
CH:1CH2CH2CH20CH2C02H
(BAA)
1
C02
dealkyl carboligase
Sources: Adapted from Corley et al. (1997, 041984) and Medinsky et al. (1990, 041986).
Figure 3-1. Proposed metabolic scheme of EGBE in rats and humans.
The two main oxidative pathways of EGBE metabolism observed in rats are via alcohol
dehydrogenase (ADH) and O-dealkylation by a cytochrome P450 dealkylase (CYP 2E1).
Because BAA is excreted in the urine of both rats and humans following EGBE exposure, it has
been suggested that the production of BAA through the formation of BAL by ADH would be
applicable to both rats and humans (Corley et al., 1997, 041984; Corley et al., 2005, 100100;
Medinsky et al., 1990, 041986). However, the other proposed metabolic pathways of EGBE
may only be applicable to rats, since the metabolites of these pathways (i.e., EG, EGBE
glucuronide, and EGBE sulfate) have been observed in the urine of rats (Bartnik et al., 1987,
100086; Ghanayem et al., 1987, 041645). but not in humans (Corley et al., 1997, 041984). In
addition, Corley et al. (1997, 041984) confirmed an observation of Rettenmeier et al. (1993,
042198) that approximately two-thirds of the BAA formed by humans is conjugated with
6

-------
glutamine and, to a lesser extent, glycine. The BAA-glutamine and BAA-glycine conjugation
pathways have not been detected in the rat.
Carpenter et al. (1956, 066464) first identified BAA as the metabolite responsible for the
hemolytic toxicity of EGBE by incubating BAA with whole blood from a variety of species.
Blood from rats, mice, and rabbits was more rapidly hemolyzed than blood from humans,
monkeys, dogs, or guinea pigs when incubated in vitro at 37.5°C with a saline solution of 0.1%
of the sodium salt of BAA. These results correlated well with osmotic fragility studies using
blood from these same species following in vivo inhalation exposures to EGBE. In contrast, a
much higher concentration (2.5%) of EGBE was required to produce a similar degree of
hemolysis in vitro. Subsequent investigations have shown that hemolytic blood concentrations
of BAA can be produced following oral or dermal administration or inhalation of EGBE.
The uptake and metabolism of EGBE is essentially linear following a 6-hour inhalation
exposure of up to 438 ppm, a concentration that causes mortality in animals (Sabourin et al.,
1992, 006763). BAA is the primary metabolite in rats following drinking water (Medinsky et al.,
1990, 041986) and inhalation (Dill et al., 1998, 041981) exposures. EGBE is eliminated
primarily as BAA in urine. Lesser amounts of the glucuronide and sulfate conjugates of EGBE
have been observed in the urine of rats (Bartnik et al., 1987, 100086; Ghanayem et al., 1987,
041645) but not humans (Corley et al., 1997, 041984). No significant differences in the urinary
levels of BAA were found following administration of equivalent doses of EGBE dermally or in
drinking water (Medinsky et al., 1990, 041986; Sabourin et al., 1993, 597595; Shyr et al., 1993,
006766). Corley et al. (1997, 041984) reported that the elimination kinetics of EGBE and BAA
appear to be independent of the route of exposure. Elimination of EGBE and BAA following
repeated inhalation exposure appears to be dependent on species, gender, age, time of exposure,
and exposure concentration (Dill et al., 1998, 041981; NTP, 2000, 196293). In rodents,
dose-dependent clearances of EGBE and BAA have been observed (Corley et al., 1994, 041977;
Ghanayem et al., 1990, 042016). A summary of species-specific toxicokinetic parameters is
shown in Table 3-1 followed by a brief summary of key individual studies.
7

-------
Table 3-1. Summary of species-specific toxicokinetic parameters
EGBE toxicokinetics
tin in blood (hr)
Species
Gender
Route
Mean
Reference
Human
Male
Inhalation
0.65
Johanson (1986. 006758)
Human
Male
Dermal vapor
0.53-0.6
Johanson and Boman (1991. 006759)
Human
Male
Dermal vapor
0.66
Corlevetal. (1997. 041984)
Human
Male/female
Inhalation
0.93
Jones and Cocker (2003. 056771)
F344 rat
Male
i.v.
0.11-0.17
Ghanavem et al. (1990. 042016)
F344 rat
Male
Inhalation
0.13-0.69
Dill et al. (1998. 041981)
F344 rat
Female
Inhalation
0.12-0.50
Dill et al. (1998. 041981)
B6C3F, mouse
Male
Inhalation
0.05-0.16
Dill et al. (1998. 041981)
B6C3F, mouse
Female
Inhalation
0.06-0.14
Dill et al. (1998. 041981)
B6C3Fi mouse
Female
i.p.
0.16
Poet et al. (2003. 100190)
B6C3Fi mouse
Female
Gavage
0.35
Poet et al. (2003. 100190)
Clearance (mL/min/kg body weight)
Species
Gender
Route
Mean
Reference
Human
Male
Inhalation
16.2
Johanson (1986. 006758)
Guinea pig
Female
i.v.
128
Johanson (1986. 006758)
F344 rat
Male
i.v.
5.9-13.3
Ghanavem et al. (1990. 042016)
Sprague-Dawley rat
Male
Inhalation
2.2-2.3
Johanson (1994. 069322)
BAA toxicokinetics
tin in blood (hr)
Species
Gender
Route
Mean
Reference
Human
Male
Inhalation
4.3
Johanson and Johnsson (1991. 100158)
Human
Male
Dermal vapor
3.27
Corlevetal. (1997. 041984)
F344 rat
Male
i.v.
1.5-3.2
Ghanavem et al. (1990. 042016)
F344 rat
Male
Inhalation
0.55-4.96
Dill et al. (1998. 041981)
F344 rat
Female
Inhalation
0.79-6.6
Dill et al. (1998. 041981)
B6C3Fi mouse
Male
Inhalation
0.36-4.0
Dill et al. (1998. 041981)
B6C3Fi mouse
Female
Inhalation
0.38—4.5
Dill et al. (1998. 041981)
B6C3Fi mouse
Female
i.p.
1.05-1.42
Poet et al. (2003. 100190)
B6C3Fi mouse
Female
Gavage
1.55-2.11
Poet et al. (2003. 100190)
Clearance (mL/min/kg body weight)
Species
Gender
Route
Mean
Reference
Sprague-Dawley rat
Male
Inhalation
0.49-0.58
Johanson (1994. 069322)
i.p. = intraperitoneal; i.v. = intravenous
Percutaneous absorption of EGBE in rats is rapid and produces measured blood levels of
BAA sufficient to produce hemolysis (Bartnik et al., 1987, 100086). Metabolism, disposition,
and pharmacokinetic studies in male F344 rats conducted by Corley et al. (1994, 041977)
produced hemolytic blood concentrations of BAA (0.5 mM) following a single oral dose of 126
mg/kg of [14C]-labeled EGBE. Using their physiologically based pharmacokinetic (PBPK)
8

-------
model, they predicted that such hemolytic blood concentrations would also be produced in rats
following a single 6-hour EGBE inhalation exposure >200 ppm. A report that evaluated the
National Toxicology Program (NTP, 2000, 196293) inhalation bioassay suggests that BAA
blood concentrations in female rats, which achieved higher blood concentrations than males,
exceeded 0.5 mM (approximately 67 jag BAA/g blood), following exposure to 62.5 ppm EGBE
for both 1-day and 12-month exposure durations (Dill et al., 1998, 041981).
The metabolic basis for the hematotoxicity of EGBE was studied in male F344 rats by
using pyrazole and cyanamide as metabolic inhibitors of ADH and aldehyde dehydrogenase
(ALDH), respectively (Ghanayem et al., 1987, 041608). Male F344 rats, 9-13 weeks old, were
pretreated with pyrazole or cyanamide followed by administration of 500 mg/kg EGBE by
gavage. The use of pyrazole protected rats from EGBE-induced hematotoxicity and resulted in a
10-fold lower ratio of BAA to conjugated EGBE excreted in urine. Cyanamide treatment
significantly reduced the hematotoxic response in a manner similar to that of pyrazole; it also
resulted in a high mortality rate in rats given cyanamide and EGBE. This effect was possibly
due to the increased levels of unmetabolized BAL; the effect was not observed in animals treated
with cyanamide or EGBE alone. Pyrazole completely blocked the increase in spleen weight to
body weight ratios seen in EGBE-treated animals. Gavage administration of either BAL or BAA
at equivalent molar doses to 125 mg/kg EGBE produced identical increased spleen to body
weight ratios and identical increases in free hemoglobin (Hb) levels in plasma. Pretreatment of
rats with cyanamide prior to administration of BAL provided significant protection against BAL-
induced hematotoxicity. These studies confirm the central role of BAA in the hematotoxic
response elicited in rats.
The effects of age, dose, and metabolic inhibitors on the toxicokinetics of EGBE were
studied in male F344 rats (Ghanayem et al., 1990, 042016). Rats aged 3-4 and 12-13 months
were administered a single gavage dose of 31.2, 62.5, or 125 mg/kg EGBE. Pretreatments
included pyrazole, cyanamide, or probenecid, an inhibitor of renal anion transport.
Toxicokinetic parameters for EGBE, including AUC, maximum plasma concentration (Cmax),
and clearance rate (Cls), were dose dependent; AUC and Cmax increased and Cls decreased as
dose levels increased. Other measured parameters were unaffected by dose. Age had no effect
on half-life (t/2), volume of distribution (Vd), or Cls of EGBE, but Cmax and AUC increased with
age. As expected from previous studies, inhibition of EGBE metabolism by either pyrazole or
cyanamide resulted in significantly increased ty2 and AUC, as well as decreased Cls. BAA
toxicokinetics were also altered by dose and age, as well as by administration of metabolic
inhibitors. Statistically significant, slight increases in Cmax, AUC, and tu were seen at higher
doses; these results were more pronounced in older rats. Probenecid pretreatment at EGBE dose
levels of 31.2 and 62.5 mg/kg produced no changes in the measured toxicokinetic parameters for
EGBE. Pretreatment produced two- to threefold increases in AUC, and two- to sixfold increases
in ty2 for BAA. These results indicate that renal organic acid transport is vital to renal elimination
9

-------
of BAA. The increased Cmax, AUC, and tu in older versus younger rats may be due to
differences in relative contributions of the two primary metabolic pathways previously discussed,
or to compromised renal clearance.
Dill et al. (1998, 041981) described the toxicokinetics of EGBE and BAA in male and
female F344 rats and B6C3Fi mice as part of the 2-year EGBE inhalation toxicity and
carcinogenicity study conducted by the NTP (2000, 196293). Blood samples were collected
from a satellite group of animals postexposure (i.e., after the daily 6-hour exposure) after 1 day,
2 weeks, and 3,6, 12, and 18 months of exposure to target EGBE concentrations of 0, 31.2 (rats
only), 62.5, 125, or 250 ppm (mice only) by whole-body inhalation; the samples were assayed
for EGBE and BAA. Postexposure time points varied from 10 to 720 minutes following 1 day,
2 weeks, and 3 and 6 months and varied from 10 to 2,880 minutes following 12 months.
Postexposure 16-hour urine samples were collected after 2 weeks and 3, 6, 12, and 18 months of
exposure and assayed for BAA. In addition, a separate set of aged mice were kept in the control
chamber and exposed to EGBE for 3 weeks when they were approximately 19 months old.
Postexposure blood samples were collected after 1 day and 3 weeks of exposure; 16-hour urine
samples were collected after 2 weeks of exposure. Overall, mice eliminated both EGBE and
BAA from blood faster than rats: for example after the 1-day exposure, ty2 for rats (males and
females, over three concentrations) averaged 8.6 minutes for EGBE, while the ty2 for mice (males
and females, over three concentrations) averaged 4 minutes for EGBE. In contrast, the rate of
BAA elimination from blood decreased as the exposure concentration increased. As exposure
continued, the rates of elimination for both EGBE and BAA decreased in both species, resulting
in longer residence times in the blood. At 1 day postexposure, t/2 in male rats was 9.4 minutes,
and at 18 months postexposure, ty2 was 15.8 minutes. Female rats were significantly less
efficient in clearing BAA from their blood than males, possibly as a result of reduced renal
clearance in female rats. The aged mice were observed to eliminate BAA from blood >10 times
slower than young mice after 1 day of exposure, but this difference was less obvious after 3
weeks of exposure. These findings provide evidence that the elimination kinetics of EGBE and
BAA, following repeated inhalation exposure to EGBE, appear to be dependent on various
factors, including species, gender, age, time of exposure, and exposure concentration.
Green et al. (2002, 041610) explored reasons that female mice develop marked
hyperkeratosis in the forestomach when given oral doses (1/day for 10 days) of either EGBE or
BAA. Irritation from the carboxylic acid BAA is hypothesized to cause cell damage followed by
cell proliferation and eventually the observed hyperkeratosis. Their studies examined the
activity and localization of ADH and ALDH (the principal enzymes involved in the metabolism
of EGBE in the stomach tissues of mice and rats) the localization of these enzymes in a human
stomach sample, and whole body autoradiography of mice exposed to radiolabeled EGBE (see
Section 4.4.1). Tissue homogenates were prepared from female B6C3Fi mice (n = 30) and rats
(n = 10; gender and species not specified) and centrifuged at 41,000 x g3 with the supernatants
10

-------
used to examine the metabolism of EGBE by ADH and ALDH. The stomachs were separated
into fore and glandular sections and used to measure the metabolism of EGBE to BAL and BAL
to BAA by ADH and ALDH, respectively. A marked species difference in ALDH activity was
observed between rats and mice:
Rats: Km = 0.29 mM
Vmax = 1.627 nmol/minute per mg protein, forestomach, Km = 0.73 mM
Vmax = 2.170 nmol/minute per mg protein, glandular stomach
Mice: Km= 46.59 mM
Vmax = 17.094 nmol/minute per mg protein, forestomach, Km = 87.01 mM
Vmax = 13.986 nmol/minute per mg protein, glandular stomach
Km values were over two orders of magnitude greater in mice compared to rats. Based
upon the Km and Vmax values reported, while the mouse ALDH enzyme has a lower affinity than
the rat enzyme for EGBE, the mouse enzyme has a much greater capacity to metabolize EGBE
to the intermediate without becoming saturated. The fact that EGBE is held in the forestomach,
along with the information that rates for the ADH enzyme were of the same order of magnitude
for rats and mice, suggests that mice are capable of generating more BAA in the forestomach
than rats for the same dose and exposure duration.
Green et al. (2002, 041610) also examined the distribution of ALDH and ADH in rat,
mouse, and human stomach tissue sample from a single individual, using histochemical staining.
The stratified squamous epithelium of the forestomach of both rats and mice contained the
highest staining intensity for ALDH and ADH. These enzymes were found throughout the
mucosa in the human stomach tissue sample; the highest concentration was found in the mucus-
producing cells at the surface. Data indicates that the distribution of these enzymes in humans is
more closely comparable to that found in the rodent glandular stomach than in the rodent
forestomach. This finding, combined with the difference in ALDH and ADH activity between
mouse and rat forestomach, suggests that humans are at much lower risk for the tissue irritation
seen in the mouse forestomach.
Deisinger and Boatman (2004, 594446) determined the extent of the in vivo formation of
BAL and BAA from EGBE and their elimination kinetics from blood, liver, and forestomach of
mice. Male and female B6C3Fi mice (4/gender/time point) were administered oral doses of
600 mg/kg EGBE dissolved in distilled water. At 5, 15, 45, and 90 minutes following the dose,
blood, liver, and forestomach tissues, along with forestomach contents, were collected and
processed to determine EGBE, BAL, and BAA concentrations in the samples. High EGBE
concentrations were measured at all time points; maximum concentrations occurred 5 minutes
after dosing, with a mean of 123 mM in females and 129 mM in males. EGBE levels in blood
11

-------
and liver were also at maximum concentrations at 5 minutes postdosing, but at levels that were
roughly 50-fold lower than in the forestomach. BAA concentrations in all organs were
substantial in the 5-minute samples, and concentrations continued to increase until leveling off in
the 45- and 90-minute samples. Concentrations of BAA measured in the forestomach were
lower than concentrations in blood and liver tissues. Furthermore, BAA was found to be
associated with forestomach tissues, rather than forestomach contents. BAL levels were highest
in the initial samples, 5 minutes postdose, and then declined. Levels of BAL measured in the
forestomach were 10-fold to 100-fold lower than the parent compound or carboxylic acid
metabolite. No differences between male and female mice were apparent in the parent
compound or BAA organ concentrations at comparable time points following dosing, but the
BAL concentrations were up to twofold greater at some time points in the liver and forestomach
of female mice compared to male mice.
Using rate constants derived from mouse stomach fractions (Green et al., 2002, 041610)
and making several assumptions about the use of these enzyme activity data, Corley et al. (2005,
100100) estimated that 250 ppm EGBE would result in peak Cmax values of 7 [jM EGBE, 0.5 |iM
BAL, and 3,250 |iM BAA in liver tissue of male mice at the end of a 6-hour inhalation exposure.
The model includes the metabolism of EGBE to BAL via ADH, and the subsequent metabolism
of BAL to BAA via ALDH in both the liver and forestomach. The model predicts that the
concentrations of BAL in gastrointestinal tract tissues of male and female mice at 5 minutes
postdosing, the time of maximal concentration, would be 18 and 33 |iM, respectively, following
gavage exposure to 600 mg/kg EGBE. This compares well with the levels of BAL actually
observed in forestomach tissue of male and female mice at 5 minutes postdosing: 19 and 33 |iM,
respectively, following gavage exposure to EGBE at 600 mg/kg (Deisinger and Boatman, 2004,
594446).
For humans, the elimination kinetics of EGBE and BAA appear to be independent of the
route of exposure. For whole-body exposures under exercise conditions, the elimination ty2 for
EGBE and BAA were 0.66 and 4 hours, respectively (Johanson, 1986, 006758; Johanson and
Johnsson, 1991, 100158). For dermal exposure to neat liquids, the ty2 for elimination of EGBE
and BAA were 1.3 and 3.1 hours, respectively (Johanson et al., 1988, 100156). For dermal
exposure to vapors, the elimination ty2 for EGBE was 0.53-0.6 hours.
Haufroid et al. (1997, 042040) conducted a study on 31 male workers exposed to low
levels of EGBE in a beverage package production plant. The average airborne EGBE exposure
concentration was 2.91 ± 1.30 mg/m3 (0.59 ± 0.27 ppm). Postshift urine samples showed an
average BAA concentration of 10.4 mg/g creatinine. One exposed individual who exhibited a
very low urinary BAA excretion was found to possess a genetic polymorphism for CYP 2E1 that
produced increased oxidative activity. However, the researchers did not measure BAA
conjugated to glutamine, an alternative pathway for BAA excretion in humans. Further
12

-------
investigations on the influence of genetic polymorphisms for CYP 2E1 on urinary BAA
excretion rate are needed before conclusions can be drawn.
Johanson and Johnsson (1991, 100158) analyzed venous blood samples collected at 0, 2,
4, and 6 hours postexposure from five healthy, male research subjects exposed to 20 ppm EGBE
via inhalation for 2 hours during light physical exercise on a bicycle ergometer. Blood samples
were analyzed for BAA concentrations. An average peak blood concentration of 45 [jM BAA
was reached 2-4 hours after exposure. The range of concentration was from 36 to 60 |iM, The
average ty2 for elimination of BAA from blood was 4.3 hours, with a range from 1.7 to 9.6 hours,
suggesting that blood levels of BAA would probably not increase following prolonged
occupational exposures to concentrations of EGBE vapor at or below existing occupational
exposure limits of 20-25 ppm. Thus, blood levels would not reach those shown to cause adverse
effects in vitro. The average renal clearance of BAA was 23-39 mL/minute, which was only
about one-third of the glomerular filtration rate (GFR). The authors suggested that the low
clearance of BAA relative to the GFR could have been related to the binding of BAA to proteins
in blood or to a low efficiency in renal tubular secretion. The low pKa of 3.5 estimated by the
researchers indicates that tubular reabsorption is unlikely, because more than 99% of the BAA in
normal human urine (pH ~6) is ionized. The Vd averaged 15 L (range 6.5-25 L) based on whole
blood measurements, and was approximately equal to the volume of extracellular water (13-
16 L), a further indication of binding of BAA to blood proteins.
Laitinen (1998, 100173) reported BAA levels in a study of eight silkscreen printers
(gender not specified) exposed to a mixture of EGBE and 2-butoxyethylacetate. Daily mean 8-
hour air concentrations ranged from 0.1 to 0.6 ppm during a 5-day period. Urine samples from
these workers contained 75 mg BAA/g creatinine immediately after the work shift, and 58 mg/g
creatinine the following morning, 14-16 hours postshift. Laitinen et al. (1998, 100174) reported
similar postshift urinary levels of 60 mg BAA/g creatinine in another group of 37 male and 15
female silkscreen workers exposed to 5 ppm EGBE and its acetate for one 8-hour workday.
Several PBPK models have been developed for EGBE, each sequentially building upon
the advances from the previous model. The first model was developed by Johanson (1986,
006758) to describe the kinetics of EGBE in the blood of human volunteers exposed for 2 hours
to 20 ppm EGBE in air while exercising. Shyr et al. (1993, 006766) published a model to
describe the pharmacokinetics of EGBE in male F344 rats based upon the drinking water
exposure data of Medinsky et al. (1990, 041986) and the inhalation and dermal data of Sabourin
et al. (1992, 006763; 1993, 597595). Corley et al. (1994, 041977) then extended the Johanson
(1986, 006758) model to describe the kinetics of EGBE, as well as the major metabolite, BAA,
in rats and humans and later validated the human dermal exposure model in Corley et al. (1997,
041984). Lee et al. (1998, 041983) followed with a model that included young and old, male
and female rats and mice to describe the kinetics of EGBE and BAA in the NTP 2-year
inhalation bioassay (data in Dill et al., 1998, 041981). Based upon the data of Dill et al. (1998,
13

-------
041981). there were species, gender, age, and exposure concentration-dependent differences in
the kinetics of BAA. Lee et al. (1998, 041983) made several assumptions about the plasma
protein binding of BAA, the metabolism of EGBE to BAA and other metabolites, and the renal
clearance of BAA (all initially described by Corley et al. (1994, 041977) and Shyr et al. (1993,
006766) in male rats) to describe the kinetic data in female rats and mice as a function of age and
chronic exposure to EGBE. Corley et al. (2005, 100100) replaced the assumptions used by Lee
et al. (1998, 041983) with experimental data. This model, along with the Lee et al. (1998,
041983) rat and mouse model and Corley et al. (1997, 041984) human model, is used in this
current review to calculate the internal dose of EGBE (Cmax of BAA in blood) used in the
development of the RfC and RfD. This model is described in more detail in Appendix B.
14

-------
4. HAZARD IDENTIFICATION
4.1. STUDIES IN HUMANS—EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS
Carpenter et al. (1956, 066464) conducted three controlled inhalation studies. In the first
study, a group of two men and six rats were exposed simultaneously for 4 hours to an EGBE
concentration of 113 ppm in a 1,250 cubic foot room. Effects observed in humans included nasal
and ocular irritation, a metallic taste, and belching. Erythrocyte osmotic fragility did not change
for the men, yet rose appreciably for the rats. In a second study, a group of two men, one
woman, and three rats were exposed to 195 ppm EGBE for two 4-hour periods, separated by a
30-minute recess, in a 6.5 cubic foot room. There was no change in the subjects' blood pressure,
erythrocyte fragility, or pulse rate. They experienced nose and throat irritation, followed by
ocular irritation and disturbed taste; one subject reported a headache. In the rats, an increase in
erythrocyte fragility values was noted. In the third study, two men and two women were
exposed for 8 hours to a 100 ppm EGBE concentration. No changes in blood pressure,
erythrocyte fragility, or pulse rate were observed. Again, nasal and throat irritation followed by
ocular irritation and a disturbing metallic taste were experienced. Two subjects reported
headaches.
There are a number of case reports of acute ingestion of EGBE, consisting primarily of
accidental or intentional ingestion. Bauer et al. (1992, 100087) reported the effects of acute
ingestion of 500 mL of window cleaner containing 9.1% EGBE and 2.5% ethanol by a 53-year-
old alcoholic male. He was comatose with metabolic acidosis, shock and noncardiogenic
pulmonary edema when brought to a hospital, approximately 10 hours after ingestion. He had
increased heart rate, decreased blood pressure, and transient polyuria and hypoxemia.
Hypochromic anemia was evident with an Hb concentration of 9.1 g/100 mL, a hematocrit (Hct)
of 25%), and thrombocytopenia. The patient recovered and was discharged after 15 days.
Gijsenbergh et al. (1989, 100134) reported that a 23-year-old woman weighing 64 kg
ingested approximately 25-30 g of EGBE (-400-500 mg/kg) and ethanol (-4:1 ratio) as a
window cleaner in an apparent suicide attempt. She was comatose when admitted to the
hospital, exhibiting dilated pupils, obstructive respiration, and metabolic acidosis, including
depression of blood Hb concentration and hematuria. The presence of EGBE in the blood and
dialysis fluid was confirmed. Treatment consisted of supportive therapy, forced diuresis,
bicarbonate administration, and hemodialysis. Her Hb concentration fell from 11.9 g Hb/100 mL
upon admission to 8.9 g Hb/100 mL. She was discharged after 8 days.
Note: Hyperlinks to the reference citations throughout this document will take you to the NCEA HERO database (Health and
Environmental Research Online) at http://epa.gov/hero. HERO is a database of scientific literature used by U.S. EPA in the process of
developing science assessments such as the Integrated Science Assessments (ISA) and the Integrated Risk Information System (IRIS).
15

-------
Gualtieri et al. (1995, 594443; 2003, 100140) reported a case of a suicide attempt with an
industrial-strength window cleaner. The 18-year-old male weighed 71 kg; he consumed between
360 and 480 mL of a concentrated glass cleaner that contained 22% EGBE, a dose equivalent to
1,131-1,509 mg/kg. He was admitted to the hospital with no abnormalities other than epigastric
discomfort within 3 hours postingestion. Approximately 10 hours postadmission, the patient was
noticeably lethargic, weak, and hyperventilating, symptoms consistent with the onset of
metabolic acidosis. BAA was measured; the highest serum concentration found was
4.86 mmol/L, collected approximately 16 hours postingestion. The patient was transferred to a
tertiary care hospital where hemodialysis was initiated at approximately 24 hours postingestion.
Ethanol therapy was started 30 minutes later. Treatment also consisted of intravenous (i.v.)
doses of 100 mg thiamine and 50 mg folic acid every 12 hours and 50 mg pyridoxine every
6 hours. Following 4 hours of dialysis, the patient was alert and remained hemodynamically
stable. Ten days after discharge, the patient was readmitted following a second ingestion of
480 mL of the same cleaner, an EGBE dose equivalent to 1,509 mg/kg. Treatment included
ethanol therapy and hemodialysis, and was initiated within a few hours of ingestion to control the
metabolic acidosis. Due to this early treatment, ethanol therapy had an impact on the disposition
of EGBE and BAA. As with the first episode, metabolic acidosis was manifest. This high-dose
oral ingestion was nearly 1.1-1.5 g EGBE/kg body weight. The highest serum BAA
concentration was 2.07 mmol/L, collected 22 hours postingestion. No evidence of hemolysis or
renal abnormalities was detected.
A 50-year-old woman ingested approximately 250-500 mL of a window cleaner
containing 12% EGBE, representing -30-60 mL, in an apparent suicide attempt (Rambourg-
Schepens et al., 1988, 100191). She was diagnosed with metabolic acidosis, hypokalemia, a rise
in serum creatinine level, and a marked increase in urinary excretion of oxalate crystals.
Moderate hemoglobinuria appeared on the third day postexposure, and a progressive
erythropenia was noted. In the absence of more complete hematologic details from this and
other similar case studies, it is not possible to determine whether these effects were due to
hemolysis or other factors related to the profound blood chemistry changes observed. The
clinical status improved gradually and the patient was discharged on the 10th day.
Burkhart and Donovan (1998, 056375) summarized the case of a 19-year-old male who
ingested 20-30 ounces (or -590-885 mL) of a product that contained 25-35% EGBE (an
exposure equivalent to -177-265 mL, estimated at >3,000 mg/kg) along with 15-25% propylene
glycol, 5-10%) monoethanolamine, and l-3%> potassium hydroxide. On his arrival at the hospital
3.5 hours after ingestion, the patient was deeply comatose with severe hypotension. Hematuria
developed on the second day, with no evidence of renal or hepatic toxicity; however, pulmonary
toxicity consisting of severe aspiration pneumonia was present. The patient had a significant
recovery, despite severe neurologic deficits that were slow to resolve.
16

-------
Osterhoudt (2002, 100186) reported on a 16-month-old girl who ingested an unknown
amount of cleaning solution containing EGBE (10-30%), monoethanolamine (5-10%),
alkoxylated linear alcohols (1—5%), ethylenediaminetetraacetic acid (1—5%), and potassium
hydroxide (1—5%). Metabolic acidosis was manifest, and a single dose (15 mg/kg) of the ALDH
inhibitor fomepizole was administered. Within 2 hours, the metabolic acidosis was completely
resolved, and there was no evidence of alkaline mucosal injury, hepatic or renal dysfunction, or
hemolysis.
Dean and Krenzelok (1991, 597279) reported that 24 children, aged 7 months to 9 years,
were observed subsequent to oral ingestion of at least 5 mL of glass window cleaner containing
EGBE in the 0.5-9.9% range. Two children drank more than 15 mL, and were treated by gastric
lavage. No symptoms of EGBE poisoning, such as metabolic acidosis, and no hemolysis were
observed in any of the children.
Raymond et al. (1998, 100193) reported on seven clerical workers who were evaluated 8
months after they entered a file room where the supervisor believed that EGBE had been applied
overnight to strip the floor. Exact details of the product used were unknown, but based on
containers found and exposure symptoms of noted intense eye and respiratory irritation, marked
dyspnea, nausea, and faintness, the authors suggested that they were exposed to EGBE
concentrations of 200-300 ppm. Of major concern were skin spots—cherry angiomas—that
appeared between 4 and 22 weeks after exposure in six of the seven workers. All workers
continued to experience recurrent eye and tracheobronchial irritation; four had a dry cough.
Workplace air sampling conducted by a certified industrial hygienist 1 week after the floor
stripping found no detectable EGBE, although traces (0.1-0.2 ppm) of formaldehyde were
identified. Five years after the exposure, four of the workers who could be contacted reported
that they continued to have outbreaks of new cherry angiomas. It should be noted that no other
studies linking EGBE exposure to outbreaks of cherry angiomas are available in the literature.
The authors included the observation that, since this report, they had seen three patients who they
believe were also exposed to EGBE vapor in an unrelated incident, and who did not develop any
skin spots. Cherry angiomas are the most common cutaneous vascular lesion; they are benign
and formed by a proliferation of dilated venules. The spots occur more frequently with
increasing age but can appear in younger individuals. There are reports in the literature of cherry
angiomas appearing following individual exposure to other chemicals, such as bromides (Cohen
et al., 2001, 100096). glutaraldehyde (Raymond et al., 1998, 100193). and sulfur mustard gas
(Firooz et al., 1999, 1001 15).
A cross section of 31 male workers, aged 22-45, employed for 1-6 years, who were
exposed to low levels of EGBE in a beverage packing production plant were monitored by
Haufroid et al. (1997, 042040). The effect of external EGBE exposure and internal BAA levels
on erythrocyte lineage were investigated by studying red blood cell (RBC) count, Hb, Hct, mean
cell volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin
17

-------
concentration (MCHC), haptoglobin (Hp), reticulocyte count, and osmotic resistance (OR), a
measure of osmotic fragility. Also studied were serum glutamic-oxaloacetic and glutamic-
pyruvic transaminases and renal creatinine and urinary retinol binding protein parameters. The
average airborne concentration of EGBE was 2.91 mg/m3, or 0.6 ppm (standard deviation [SD]
of ±1.30 mg/m3 or 0.27 ppm). In addition, there was coexposure to methyl ethyl ketone. Single
determinations of BAA in postshift urine samples were used to assess exposure to low levels of
EGBE. No differences were observed for RBC counts, Hb, MCV, MCH, Hp, reticulocyte count,
or between exposed and control workers. The only statistically significant change observed in
exposed workers when compared with a matched control group (n = 21) was a 3.3% decrease in
Hct (p = 0.03) and a 2.1% increase in MCHC (p = 0.02). The implications of these small
erythroid effects are unclear. Both values are within their corresponding normal clinical ranges
and, given that no statistically significant changes were observed in other erythroid parameters,
they do not appear to be related to the more severe effects observed in laboratory animals.
Furthermore, no correlation was found between any of the nine erythroid parameters measured
and the parameters of internal exposure. No significant differences were observed in hepatic and
renal biomarkers.
Several human studies investigated the dermal absorption of EGBE. Jakasa et al. (2004,
100151) dermally exposed six male research subjects, ages 22-55, to 50%, 90%, or neat EGBE
for 4 hours on the forearm over an area of 40 cm2. The dermal absorption of EGBE from
aqueous solutions was markedly higher than from neat EGBE. In Jones et al. (2003, 056771).
four research subjects were exposed via inhalation of 50 ppm EGBE for 2 hours on nine separate
occasions, with each occasion separated by 3 weeks, at varying temperatures and humidity
levels. Results show that "baseline" dermal contribution to total body absorption of EGBE vapor
in appropriately dressed workers was, on average, 11%. Higher temperature (30°C, mean 14%,
p = 0.03) and greater humidity (65% relative humidity, mean 13%,p = 0.1) both increased
dermal absorption. The wearing of whole-body overalls did not attenuate absorption (mean
10%). By combining several factors together in the industrial scenario, dermal absorption of
vapors was reported to be as high as 39% of the total absorbed dose.
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIOASSAYS IN
ANIMALS—ORAL AND INHALATION
4.2.1. Subchronic Studies
4.2.1.1. Oral
A number of subchronic studies by the gavage route of exposure have been conducted.
Krasavage (1986, 066474) conducted a toxicity study using groups of 10 COBS CD (Sprague-
Dawley) BR adult male rats treated by gavage with 222, 443, or 885 mg/kg-day undiluted EGBE
5 days/week for 6 weeks. Endpoints evaluated throughout the study included body weight, food
18

-------
consumption, clinical signs, and survival. Hematology and serum clinical chemistry parameters
were determined after the last treatment. Dose-related changes were observed in the RBC counts
of all treatment groups, including statistically significant decreases in RBC count and Hb
concentration and a statistically significant increase in MCH. Statistically significant
hematological changes occurring at 443 and 885 mg/kg-day were increased MCV and decreased
MCHC. The increased MCV at higher doses is likely due to both an increase in MCV and in the
number of larger reticulocytes in the circulation following the erythropoietic response (NTP,
2000, 196293). Based on decreased RBC count and trends in Hb and other hematological
endpoints, the lowest-observed-adverse-effect level (LOAEL) was determined to be 222 mg/kg-
day, the lowest dose tested. A no-observed-adverse-effect level (NOAEL) was not identified.
Nagano et al. (1979, 100179) performed a toxicity study in male JCL/ICR mice
(five/group) using gavage doses of 0, 357, 714, or 1,430 mg/kg-day EGBE 5 days/week for
5 weeks. Parameters evaluated at the end of the study were hematology (RBC and white blood
cell [WBC] counts, MCV, and Hb), absolute and relative weights of testes, and testicular
histology. Mean RBC counts were significantly lower than the control values in the 357 and
714 mg/kg-day groups, but WBC counts were not affected. All animals in the 1,430 mg/kg-day
group died before examinations were performed; mortality was not observed in the lower dose
groups, and no differences in testes weights or histology were found. The LOAEL for this study,
based on the reduced RBC count, was 357 mg/kg-day. A NOAEL was not determined.
Siesky et al. (2002, 042204) investigated whether subchronic exposure to EGBE in F344
male rats, 8-10 weeks old, and B6C3Fi male mice, 8-10 weeks old, produced an increase in
oxidative damage and deoxyribonucleic acid (DNA) synthesis in endothelial cells and
hepatocytes in the mouse liver, the putative cancer target cells. Mice (60/group) were treated via
gavage with doses of 0, 225, 450, and 900 mg/kg-day and rats (20/group) with 0, 225, and
450 mg/kg-day for 90 days. A dose-related increase in hemolysis was observed in both rats and
mice. An increase in the percentage of iron-stained Kupffer cells was observed following 450
and 900 mg/kg in mice and 225 and 450 mg/kg in rats. An increase in oxidative damage, as
measured by 8-hydroxydeoxyguanosine (8-OHdG) levels, was seen in mouse livers at 450 and
900 mg/kg-day after 7 or 90 days, while no increase was seen in rat livers at any dose or time
point examined. Vitamin E levels were reduced by all doses of EGBE in the mouse and rat liver
(statistically significant at 7 and 90 days in both mice and rats); however, the basal level of
vitamin E was -2.5-fold higher in rat than in mouse liver. The LOAEL for this study was
450 mg/kg-day in mice and 225 mg/kg-day in rats, based on the percentage of iron-stained
Kupffer cells. The NOAEL was 225 mg/kg-day in mice, and a NOAEL was not determined in
rats.
NTP (1993, 042063) performed a 13-week toxicity study in F344 rats and B6C3Fi mice
where groups of 10 animals/gender/species received EGBE in drinking water at doses of 0, 750,
1,500, 3,000, 4,500, and 6,000 ppm in rats and 0, 750, 1,500, 3,000, 4,500, and 6,000 ppm in
19

-------
mice. The corresponding doses in mg/kg-day, based on measured drinking water consumption
were: 0, 69, 129, 281, 367, or 452 mg/kg-day in male rats; 0, 82, 151, 304, 363, or 470 mg/kg-
day in female rats; 0, 118, 223, 553, 676, or 694 mg/kg-day in male mice; and 0, 185, 370, 676,
861, or 1,306 mg/kg-day in female mice. Due to a dose-related decrease in water consumption
in the 2-week studies, the test chemical was administered at a constant concentration (ppm) in
the 13-week studies rather than on a mg/kg body weight basis. Hematology was performed on
rats but not on mice. Complete histological exams were performed on all control animals and all
animals in the highest dose group. Vaginal cytology and sperm indices were evaluated in rats
and mice from the control and three highest dose groups. Hematologic changes in both genders
persisting until or developing by 13 weeks included dose-related indications of mild to moderate
anemia. Portions of the hematologic results from the NTP 13-week rat drinking-water study are
presented in Table 4-1. The various results shown in this table are indicative of the various
degrees of hemolysis caused by exposure to increasing concentrations of EGBE. Overall, the
dose-related increase in MCV is indicative of erythrocyte swelling that would be expected to
precede cell lysis and an increase in the number of reticulocytes. Deficits in RBCs as a result of
lysis manifest through dose-related decreases in the measures of both RBC count and Hb
concentration. Hct would be expected to decrease but did not. The increases noted both in
reticulocytes (young RBCs) and, at higher doses, in nucleated erythrocytes (immature and
prematurely released blood cells) are homeostatic responses that would be anticipated to occur as
the lysed blood cells are being replaced. More specifically, male rats evaluated at 13 weeks
showed significantly reduced RBC counts at > 281 mg/kg-day (3,000 ppm) and reduced Hb
concentration, reduced platelets, and increased bone marrow cellularity at > 367 mg/kg-day
(4,500 ppm). These data also suggest that female rats are more sensitive to the effects from
EGBE, since several statistically significant effects occurred at the 750 ppm concentration, the
lowest level tested in females; males did not show statistically significant effects until two dose
levels higher (3,000 ppm). In addition, the degree to which these various measures were affected
was somewhat greater in females than males (indicated as percent control in the tables),
particularly at the three highest exposure concentrations.
20

-------
Table 4-1. Hematology and hemosiderin data from the 13-week drinking
water exposure to EGBE in F344 rats
Endpoint"
Control
750 ppm
1,500 ppm
3,000 ppm
4,500 ppm
6,000 ppm
N
Males
Females
8
10
10
10
10
10
10
10
10
10
9
9
Hct (%)
Males
Females
44.8 ±0.8
44.8 ±0.6
45.0 ±0.6 (100)
43.2 ±0.8 (96)
44.7	± 0.4 (99)
42.8	±0.7 (95)
44.1 ±0.7 (98)
43.6 ±0.7 (97)
42.3	± 0.6b (94)
44.4	± 0.7 (99)
43.4 ±0.4 (97)
46.1 ±0.7 (103)
Hb (g/dL)
Males
Females
15.0 ±0.2
14.9 ±0.2
15.2 ±0.1 (101)
14.4 ± 0.2b (97)
14.9 ±0.1 (99)
13.9 ±0.2 (93)
14.6 ±0.1 (97)
14.2 ±0.2 (95)
14.0 ± 0.1° (93)
14.0 ± 0.2° (94)
13.7 ± 0.2° (91)
13.4 ± 0.2° (90)
Erythrocytes
(106/|iL)
Males
Females
8.64 ±0.15
8.15 ±0.09
8.74 ±0.10 (101)
7.59 ± 0.15° (93)
8.54 ±0.09
(99)
7.09 ± 0.14°
(87)
8.11 ±0.12b
(94)
7.00 ± 0.12°
(86)
7.48 ± 0.12°
(86)
6.80 ± 0.11°
(83)
7.18 ± 0.12°
(83)
6.58 ± 0.14°
(81)
Reticulocytes
(106/|iL)
Males
Females
0.14 ±0.03
0.12 ±0.02
0.24 ± 0.06
0.17 ±0.03
0.15 ±0.02
0.19 ±0.03
0.18 ±0.02
0.28 ± 0.03°
0.22 ±0.05
0.28 ± 0.05°
0.46 ± 0.07°
0.27 ± 0.05°
Nucleated
erythrocytes
(103/|iL)
Males
Females
0.00 ± 0.00
0.01 ±0.01
0.00 ± 0.00
0.03 ±0.02
0.01 ±0.01
0.02 ±0.01
0.01 ±0.01
0.05 ±0.02
0.00 ± 0.00
0.10 ± 0.03b
0.04 ± 0.02°
0.16 ± 0.04°
MCV (fL)
Males
Females
52.0 ±0.4
54.8 ±0.3
51.5 ±0.3 (99)
57.0 ±0.4C (104)
52.3 ±0.4
(100)
60.5 ± 0.4°
(110)
54.4 ± 0.3°
(105)
62.4 ± 0.6°
(114)
56.7 ± 0.5°
(109)
65.3 ±0.6C
(119)
60.6 ± l.lc
(116)
70.1 ± 0.9°
(128)
MCH (pg)
Males
Females
17.4 ±0.2
18.3 ±0.2
17.4 ±0.1
18.9 ±0.2
17.5 ±0.2
19.7 ±0.2
18.0 ± 0.2b
20.2 ± 0.3°
18.7 ± 0.3°
20.6 ± 0.2°
19.1 ± 0.3°
20.4 ± 0.1°
Hemosiderin
(incidence)
Males
Females
0/10
0/10
0/10
0/10
0/10
2/10
0/10
10/10
0/10
10/10
7/10
10/10
aValues listed are mean ± standard error at various EGBE concentrations in ppm for the 13-week time point.
Percent of control values in parentheses.
Statistically significant difference, p < 0.05.
Statistically significant difference,/) <0.01.
Source: NTP (1993, 042063).
Statistically significant hematologic effects in female rats at week 13 included reduced
RBC counts and Hb concentrations at > 82 mg/kg-day and increased reticulocytes, decreased
platelets, and increased bone marrow cellularity at approximately 304 mg/kg-day, all being
21

-------
indicative of hemolysis. There were no histopathological changes in the testes or epididymis at
>129 mg/kg-day.
Table 4-2 shows that liver effects, including cytoplasmic alterations, hepatocellular
degeneration, and pigmentation were observed in the mid- and high-dose groups (129, 281, 367,
and 452 mg/kg-day for males and 151, 304, 363, and 470 mg/kg-day for females; statistics not
reported). As with the hematologic effects, these effects appeared to be more severe in females
than in males. Cytoplasmic alterations of liver hepatocytes, consisting of hepatocytes staining
more eosinophilic and lacking the basophilic granularity of the cytoplasm present in hepatocytes
from control animals, were observed in the lowest-dose groups tested (69 mg/kg-day for males
and 82 mg/kg-day for females). The lack of cytoplasmic granularity or "ground-glass"
appearance of the hepatocytes suggests that this response was not due to enzyme induction
(Greaves, 2000, 089176). The cytoplasmic alterations were judged to increase in severity in both
genders, but especially in females, with the severity in the two highest dose groups being judged
as "moderate." Liver pigmentation, colored brown to green and staining strongly positive for
iron (indicative of hemosiderin accumulation), was noted in the cytoplasm of Kupffer cells in
both genders of rats. In females, liver pigmentation was noted in 0/10 controls and 0/10 at
82 mg/kg-day, 2/10 with a severity grade of 1 (minimal) at 151 mg/kg-day, and 10/10 in the
three highest dose levels; the severities increased from a numerical grade of 1.2 in the 304
mg/kg-day group to 1.9 the upper two dose groups. In males, the hemosiderin pigmentation was
noted in animals exposed to the highest dose only (452 mg/kg-day) at an incidence of 7/10 and a
severity rating of 1 (minimal). No hepatic pigmentation was reported in the mice exposed for
13 weeks. The hematological (decreased RBC count and Hb) and hepatic changes were dose-
related; 69-82 mg/kg-day was considered a LOAEL. A NOAEL was not identified.
22

-------
Table 4-2. Incidence3 and severity of selected histopathological changes from
the 13-week drinking water exposure to EGBE in F344 rats and mice

Control
750 ppm
1,500 ppm
3,000 ppm
4,500 ppm
6,000 ppm
Rat
N
10
10
10
10
10
10
Liver cytoplasmic alterations (NR)
Males
0
4(1.0)
8 (1.0)
7(1.1)
10 (2.0)
10(1.8)
Females
0
5 (1.4)
9 (2.0)
10 (2.2)
10 (3.0)
10 (3.0)
Hepatocellular degeneration (NR)
Males
0
0
0
8(1.0)
8(1.0)
8(1.0)
Females
0
0
0
10(1.3)
10(1.3)
10(1.1)
Kupffer cell pigmentation (NR)
Males
0
0
0
0
0
7 (1.0)
Females
0
0
2(1.0)
10(1.2)
10(1.9)
10(1.9)
Mouse
N
10
10
10
10
10
10
Necropsy body weight (g)b
Females
(only)
31.1 ± 0.7
31.8 ± 0.8
30.9 ± 1.5
28.0 ± 0.7°
28.4 ± 0.5°
27.8 ± 0.9d
Relative kidney weight (right) (mg organ wt/g body wt)b
Females
(only)
6.33 ±0.10
7.69 ± 0.14d
8.06 ± 0.29d
7.47 ± 0.19d
7.55 ± 0.18d
8.21 ± 0.26d
incidences represent the number of animals with lesions. Average severity (in parentheses) is based on the number
of animals with lesions; 1 = minimal, 2 = mild, 3 = moderate, 4 = marked.
bMean ± standard error.
Statistically significant difference,/) < 0.05.
Statistically significant difference, p < 0.01.
NR = Statistics not reported
Source: NTP (1993, 042063}.
Female mice showed statistically significant reductions in body weight gain starting at
3,000 ppm and statistically significant increases in relative kidney weight at all doses. Changes
at the higher doses followed the reductions in body weight at those dose levels, while the
increases at lower doses (750 and 1,500 ppm) were due to increased absolute kidney weights at
those doses. Body weight reduction followed the decreased water consumption data. No
histopathologic changes were noted at any dose level, even though relative kidney weights
showed a statistically significant increase at 750 and 1,500 ppm in the absence of reduction in
body weight gain.
Keith et al. (1996, 041625) administered EGBE at 120 mg/kg-day for 120 days by gavage
to transgenic FVB/N mice (25 mice/gender/group) carrying the v-Ha-ras oncogene and observed
the animals for an additional 120 days. EGBE did not induce an increase in tumors at any site.
23

-------
4.2.1.2. Inhalation
Wistar rats (23 animals/group, gender not specified) were exposed to 0, 135, or 320 ppm
EGBE 7 hours/day, 5 days/week for 5 weeks (Werner et al., 1943, 597282). Hematologic
endpoints—RBC, WBC, differential, and reticulocyte counts and Hb concentration—were
evaluated. Exposure to 320 ppm EGBE resulted in an increased percentage of circulating
immature granulocytes, a decreased Hb concentration and RBC count, and an increased
reticulocyte count. These hematologic changes were not severe; they were reversed 3 weeks
after discontinuing exposure. No effect on WBC count was observed. In another study, Werner
et al. (1943, 100219) exposed groups of two dogs of unspecified strain to subchronic inhalation
doses of 0 or 415 ppm EGBE 7 hours/day, 5 days/week for 12 weeks. Necropsies were
performed 5 weeks postexposure; hematologic parameters were examined before, during, and
after the exposure. No statistical analysis was presented. The authors concluded that exposure
of dogs to EGBE vapors resulted in decreased Hb concentration and RBC count with increased
hypochromia, polychromatophilia, and microcytosis. These hematologic effects were not severe
and they were reversed 5 weeks after the end of exposure.
Carpenter et al. (1956, 066464) studied the hemolytic effects of EGBE vapor inhalation
in rats, mice, dogs, and monkeys, in addition to humans. An unspecified strain of rats
(15/gender/group) was exposed via inhalation to 54, 107, 203, 314, or 432 ppm EGBE 7
hours/day, 5 days/week for 6 weeks. Erythrocyte osmotic fragility was observed in rats
immediately after a single 7-hour exposure to > 107 ppm. Osmotic fragility in females exceeded
that for males. In almost all cases, these high fragility values returned to normal after the rats
rested overnight. In the same study, the authors exposed groups of 10 male C3H mice to 100,
200, or 400 ppm EGBE 7 hours/day for 30, 60, or 90 days. An increase in erythrocyte osmotic
fragility occurred at all concentrations and was consistent throughout the exposures. In all
instances, erythrocyte osmotic fragility was normal after a 17-hour rest period. The LOAELs for
these rat and mouse studies were 54 and 100 ppm, respectively. NOAELs were not reported.
The authors reported slight increases in erythrocyte osmotic fragility for a male and a female dog
(basenji hybrids) exposed to 200 ppm EGBE for 31 days (7 hours/day). RBC counts and Hb
concentrations were slightly decreased in the female. Erythrocyte permeability, as determined
by radio-iodine uptake, was increased in both genders, but was not statistically different when
compared with control values. A female dog succumbed after 8 days of inhalation exposure to
385 ppm of EGBE (7 hours/day). Symptoms included loss of weight, transitory increases in
erythrocyte osmotic fragility, nasal and ocular infection, weakness, apathy, anorexia, and
increased WBC count. Necropsy of this animal revealed severe congestion and hemorrhage of
the lungs and congestion of the liver and both kidneys. In addition, a severe subcapsular
hemorrhage was found in one adrenal gland. A male dog survived after 28 days of inhalation
exposure to 385 ppm of EGBE for 7 hours/day. Effects in the male were similar to the female,
but developed more slowly. At autopsy, congestion of the kidneys was not observed. In studies
24

-------
on male and female monkeys, occasional rises in erythrocyte osmotic fragility were reported that
were more frequent in the female than in the male following 90-day inhalation exposure to 100
ppm of EGBE.
A 90-day subchronic inhalation study was performed using F344 rats (16/gender/group)
exposed to EGBE for 6 hours/day, 5 days/week at concentrations of 0, 5, 25, and 77 ppm (Dodd
et al., 1983, 066465). After 31 completed exposures (6 weeks), the 77 ppm female rats had
slight but statistically significant decreases in RBC counts (13% below control value) and Hb
concentrations, accompanied by an 11% increase above the control value in MCH. The 77 ppm
males exhibited slight (5%) but statistically significant decreases in RBC counts and Hb
concentration that were accompanied by increases in MCH. At the end of the 90-day study
(66 exposures), the hematologic effects seen in the 77 ppm exposed animals had either lessened
or returned to the ranges of control values and were no longer statistically significant. The
NOAEL was determined to be 25 ppm, and the LOAEL was 77 ppm.
In the subchronic portion of the inhalation NTP (2000, 196293) study, F344 rats and
B6C3Fi mice (10/gender) were exposed to EGBE concentrations of 0, 31, 62.5, 125, 250, and
500 ppm (0, 150, 302, 604, 1,208, and 2,416 mg/m3) 6 hours/day, 5 days/week for 14 weeks.
Hematologic and hemosiderin staining results are presented in Tables 4-3 and 4-4. These results
are indicative of the various degrees of hemolysis caused by exposure to increasing
concentrations of EGBE. Both rat genders exhibited clinical signs at the three highest doses,
consistent with the hemolytic effects of EGBE, including: (1) deficits in RBCs as a result of
lysis manifestation through the dose-related decrease in Hct—a finding consistent with decreases
noted for both RBC count and Hb concentrations; and (2) increases in both reticulocytes and
nucleated erythrocytes at higher doses—homeostatic responses that would be anticipated to
occur as the lysed blood cells are being replaced. Female rats may be somewhat more sensitive;
several statistically significant effects occur at the 31 ppm level in females, as opposed to a
single parameter for males. In addition, the degree to which these various measures are affected
is somewhat greater in females than males (indicated as percent control) particularly at the three
highest concentrations. Hematologic evaluation showed mild-to-moderate regenerative anemia
at all concentrations in females and at the three highest concentrations in males. Exposure-
related trends were noted for reticulocyte count, RBC count, MCV, Hb concentration, and Hct.
Liver-to-body-weight ratios increased significantly in males at the two highest concentrations
and in females at the highest concentration. Histopathologic effects at concentrations >62.5 ppm
for male rats and >31 ppm for females consisted of excessive splenic congestion in the form of
extramedullary hematopoiesis, hemosiderin accumulation in Kupffer cells, liver necrosis,
centrilobular hepatocellular degeneration, renal tubular degeneration, intracytoplasmic Hb and
hemosiderin deposition, and bone marrow hyperplasia. In addition, five moribund female rats
were sacrificed from the highest concentrations and one from the 250 ppm group. The LOAEL
25

-------
for hematological alterations was 31 ppm for female rats and 62.5 ppm for male rats. The 31
ppm exposure level was considered a NOAEL for male rats.
Table 4-3. Hematology and hemosiderin data from a 14-week inhalation
study of EGBE in F344 rats
Endpoint"
Control
31 ppm
(150 mg/m3)
62.5 ppm
(302 mg/m3)
125 ppm
(604 mg/m3)
250 ppm
(1,208 mg/m3)
500 ppm
(2,416 mg/m3)
Hct (%)
Males
Females
46.8 ±0.5
48.5 ±0.5
45.8 ±0.6 (98)
46.0 ± 0.5°(95)
47.0 ±0.4(100)
45.2 ± 0.5°(93)
44.5 ± 0.5° (95)
42.9 ± 0.4° (88)
41.1 ± 0.3° (88)
40.0 ± 0.3° (82)
37.3 ± 0.4° (80)
36.2 ± 0.6° (75)
Hb (g/dL)
Males
Females
15.5	±0.1
15.6	±0.1
14.8 ±0.3 (95)
15.0 ± 0.1°(96)
15.4 ±0.1 (99)
14.6 ± 0.1°(94)
14.5	± 0.2° (94)
13.6	± 0.1° (87)
13.1 ± 0.1° (85)
12.5 ± 0.1° (80)
11.7 ±0.1° (75)
10.5 ± 0.3° (67)
Erythrocytes
(106/|iL)
Males
Females
9.05 ±0.08
8.48 ±0.05
8.71 ± 0.14b(96)
8.08 ± 0.07° (95)
8.91 ±0.06(98)
7.70 ± 0.08°(91)
8.01 ± 0.08°(89)
6.91 ± 0.05°(81)
7.10 ± 0.07° (78)
6.07 ± 0.04° (72)
5.97 ± 0.05°(66)
4.77 ± 0.15° (56)
Reticulocytes
(106/|iL)
Males
Females
0.16 ±0.02
0.13 ±0.02
0.17 ±0.03
0.10 ±0.01
0.15 ±0.02
0.16 ±0.02
0.30 ± 0.04°
0.26 ± 0.04b
0.48 ± 0.06°
0.34 ± 0.04°
0.68 ± 0.07°
0.40 ±0.11°
Nucleated
erythrocytes
(103/|iL)
Males
Females
0.04 ± 0.02
0.04 ± 0.02
0.05 ±0.01
0.05 ± 0.02
0.04 ±0.03
0.12 ± 0.03b
0.11 ±0.03
0.18 ±0.07
0.17 ± 0.04°
0.61 ± 0.24°
0.20 ± 0.06°
0.73 ± 0.27°
MCV (fL)
Males
Females
50.4 ±0.3
55.1 ±0.3
50.2	±0.2 (100)
55.3	±0.2 (100)
50.7 ±0.2 (100)
56.4 ±0.2 (102)
53.1 ± 0.2°(105)
58.7 ± 0.2°(107)
53.8 ±0.3C (107)
61.6 ±0.2C (112)
58.5 ± 0.3°(117)
66.8 ±0.9C (121)
MCH (pg)
Males
Females
17.1 ±0.1
18.4 ±0.1
17.0 ±0.1
18.6 ±0.2
17.3 ±0.1
19.0 ± 0.0°
18.1 ±0.1°
19.6 ± 0.1°
18.4 ± 0.1°
20.6 ± 0.1°
19.5 ± 0.1°
22.0 ± 0.1°
Hemosiderin
(incidence)
Males
Females
0/10
0/10
0/10
0/10
0/10
10/10
7/10
10/10
10/10
9/9
10/10
5/5
aValues listed are mean ± standard error (percent of control).
Statistically significant difference,/) < 0.05.
Statistically significant difference, p < 0.01.
Source: NTP (2000, 196293}.
26

-------
Table 4-4. Hematology and hemosiderin data from a 14-week inhalation
study of EGBE in B6C3Fi mice
Endpoint"
Control
31 ppm
(150 mg/m3)
62.5 ppm
(302 mg/m3)
125 ppm
(604 mg/m3)
250 ppm
(1,208 mg/m3)
500 ppm
(2,416 mg/m3)
Hct (%)
Males
Females
47.3 ± 1.0
46.2 ±0.3
48.3 ±0.4(102)
45.9 ±0.3(99)
47.6 ±0.5(101)
45.8 ±0.3(99)
46.6 ± 0.4(99)
45.1 ± 0.2°(98)
44.2	± 0.4°(93)
42.3	± 0.4°(92)
36.3 ± 1.4°(77)
37.8 ± 1.0°(82)
Hb (g/dL)
Males
Females
15.7 ±0.4
15.7 ±0.1
16.0 ±0.1(102)
15.4 ± 0.1b(98)
15.9 ±0.1(101)
15.4 ± 0.1b(98)
15.4 ± 0.1°(98)
14.8 ± 0.1°(94)
14.4 ± 0.1°(92)
13.7 ± 0.1°(87)
11.4 ± 0.4°(73)
11.6 ± 0.1°(74)
Erythrocytes
(106/|iL)
Males
Females
9.71	±0.22
9.72	±0.05
10.04 ±0.08(103)
9.55 ± 0.06b(98)
9.77 ±0.1(101)
9.51 ± 0.06b(98)
9.47 ± 0.06b(98)
9.18 ± 0.05°(94)
8.90 ± 0.07°(92)
8.57 ± 0.06°(88)
7.21±0.23°(74)
7.35±0.07°(76)
Reticulocytes
(106/|iL)
Males
Females
0.21 ±0.03
0.18 ±0.02
0.22 ±0.03
0.21 ±0.03
0.21 ±0.02
0.19 ±0.02
0.32 ± 0.03b
0.29 ± 0.02°
0.45 ± 0.04°
0.47 ± 0.04°
0.79 ± 0.20°
1.17 ± 0.28°
MCV (fL)
Males
Females
49.1 ±0.4
48.3 ±0.3
48.5 ±0.3(99)
48.8 ±0.2(101)
49.0 ±0.4 (100)
48.8 ±0.2 (101)
49.7 ±0.4(101)
49.5 ±0.5 (102)
49.8 ±0.4(101)
49.0 ±0.3(101)
48.3± 0.9(98)
48.8± 1.0(101)
MCH (pg)
Males
Females
16.2 ±0.1
16.1 ±0.1
16.0 ±0.1(99)
16.0 ±0.1(99)
16.2 ±0.1(100)
16.2 ±0.1(101)
16.2 ±0.0(100)
16.1 ±0.1(100)
16.2 ±0.1(100)
16.0 ±0.0(99)
15.8 ±0.2(98)
15.8 ±0.1(98)
Hemosiderin
(incidence)
Males
Females
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
10/10
6/6
6/6
"Values listed are mean ± standard error (percent of control).
Statistically significant difference, p < 0.05.
Statistically significant difference,/) <0.01.
Source: NTP (2000, 196293}.
The mice exposed via the inhalation route exhibited clinical signs consistent with the
hemolytic effects of EGBE at the two highest concentrations for both genders (NTP, 2000,
196293). Hematologic evaluation indicated a moderate regenerative anemia (marked by
decreased RBC counts, increased reticulocyte counts, and increased MCV) with an increase in
platelets at the three highest concentrations in both genders. Histopathological effects consisted
of excessive extramedullary splenic hematopoiesis, renal tubular degeneration, hemosiderin
deposition in the spleen and kidney and accumulation in Kupffer cells, and testicular
degeneration. Forestomach necrosis, ulceration, inflammation, and epithelial hyperplasia were
observed at concentrations >31 ppm for females and >62.5 ppm for males. In addition, four
females and four males either died or were sacrificed moribund at the highest concentration. The
NOAEL for male and female mice was 31 ppm and the LOAEL was 62.5 ppm, based on
histopathological changes in the forestomach.
27

-------
4.2.2. Chronic Studies and Cancer Bioassays
4.2.2.1. Inhalation
NTP (2000, 196293) also completed a two-species, 2-year inhalation study on EGBE in
both genders of rats and mice. In this chronic study, animals were exposed to EGBE 6
hours/day, 5 days/week at concentrations of 0, 31, 62.5, and 125 ppm (0, 150, 302, and 604
mg/m3) for groups of 50 F344/N rats and 0, 62.5, 125, and 250 ppm (0, 302, 604, and 1,208
mg/m3) for groups of 50 B6C3Fi mice. The researchers stated that the highest exposure was
selected to produce a 10-15% depression in hematologic indices. They reported that no effect on
survival was observed in rats, but survival was statistically significantly decreased in male mice
exposed to 125 or 250 ppm, compared with chamber controls (54, 52, and 78% respectively
(NTP, 2000, 196293)). Although statistics were not reported for mean body weights, the rats
exposed to 31 and 62.5 ppm had similar mean body weights to the control rats. Mean body
weights of the exposed mice were generally less than those of controls, with females
experiencing greater and earlier reductions. From week 17 to the end of the study, the mean
body weights of 125 ppm female rats were generally less than those of controls. Nonneoplastic
effects in rats included hyaline degeneration of the olfactory epithelium in males (13/48, 21/49,
23/49, 40/50) and females (13/50, 18/48, 28/50, 40/49) and Kupffer cell pigmentation in the
livers of males (23/50, 30/50, 34/50, 42/50) and females (15/50, 19/50, 36/50, 47/50)
(Table 4-5). The severity of the olfactory lesion was not affected by exposure. The Kupffer cell
pigmentation is a result of hemosiderin accumulation and is a recognized secondary effect of the
hemolytic activity of EGBE (NTP, 2000, 196293).
28

-------
Table 4-5. Selected female and male rat and mouse nonneoplastic effects
from the 2-year chronic EGBE inhalation study

Control
31 ppm
62.5 ppm
125 ppm
250 ppm
Rat





Kupffer cell pigmentation, hemosiderin in the liver
Male
23/50
30/50
34/503
42/503
NT
Female
15/50
19/50
36/503
47/503
NT
Hyaline degeneration of the olfactory epithelium
Male
13/48
21/493
23/49a
40/503
NT
Female
13/50
18/48
28/503
40/493
NT
Mouse





Kupffer cell pigmentation, hemosiderin in the liver
Male
0/50
NT
0/50
8/4 9b
30/49b
Female
0/50
NT
5/50a
25/49b
44/50b
Hematopoietic cell proliferation in the spleen
Male
12/50
NT
11/50
26/48b
42/50b
Female
24/50
NT
29/50
32/49
35/503
Hemosiderin in the spleen
Male
0/50
NT
6/50a
45/48b
44/49b
Female
39/50
NT
44/50
46/49b
48/50b
Forestomach ulcers
Male
1/50
NT
2/50
9/4 9b
3/48
Female
1/50
NT
7/50a
13/49b
22/50b
Forestomach epithelial hyperplasia
Male
1/50
NT
7/50a
16/49b
21/48b
Female
6/50
NT
27/50b
42/49b
44/50b
Hyaline degeneration of the olfactory epithelium
Females (only)
6/50
NT
14/50
11/49
12/50
Bone marrow hyperplasia
Males (only)
0/50
NT
1/50
9/49a
5/50a
"Statistically significant difference,/) < 0.05.
Statistically significant difference, p < 0.01.
NT = not tested
Source: NTP (2000, 196293).
Nonneoplastic, statistically significant effects in mice included forestomach ulcers and
epithelial hyperplasia, hematopoietic cell proliferation and hemosiderin pigmentation in the
spleen, Kupffer cell pigmentation in the livers, and bone marrow hyperplasia (males only).
Hyaline degeneration of the olfactory epithelium (females only) was increased relative to
chamber controls but was not statistically significant. As in the rats, the Kupffer cell
pigmentation was considered a secondary effect of the hemolytic activity of EGBE. Bone
marrow hyperplasia, hematopoietic cell proliferation, and hemosiderin pigmentation in the
spleen were also attributed to the primary hemolytic effect; it was followed by regenerative
29

-------
hyperplasia of the hematopoietic tissue. The forestomach lesions did not appear to be related to
the hemolytic effect of EGBE. Incidences of ulcer were significantly increased in all exposed
female groups, as well as males exposed to 125 ppm. Incidences of epithelial hyperplasia,
usually focal, were significantly increased in all exposed groups of males and females. The
hyperplasia was often associated with ulceration, particularly in the females, and consisted of
thickness of the stratified squamous epithelium and sometimes the keratinized layer of the
forestomach. Ulceration consisted of a defect in the forestomach wall that penetrated the full
thickness of the epithelium and frequently contained accumulations of inflammatory cells and
debris.
Using the same exposure levels described above, additional groups of rats (27/gender/
exposure group) and mice (30/gender/exposure group) in the 2-year study were examined at 3, 6,
and 12 months (8-10 animals/time point) for hematologic effects (NTP, 2000, 196293). Nine
male and nine female rats were exposed to 31 ppm EGBE, specifically to evaluate hematology at
3 months and to receive a total evaluation at 6 months. Animals were continuously exposed, as
described above, until their sacrifice at 3, 6, or 12 months. As in the 14-week study, inhalation
of EGBE by both species resulted in the development of exposure-related hemolytic effects,
inducing a responsive anemia. In rats, the anemia was persistent and did not progress or
ameliorate in severity from 3 months to the final blood collection at 12 months. Statistically
significant (p < 0.05) decreases in automated and manual Hct values, Hb concentrations, and
RBC counts occurred at 3, 6, and 12 months in the 125 and 250 ppm female mice and the 250
ppm male mice. Statistically significant decreases in these same endpoints were also observed in
62.5 ppm females at 6 months and in 125 ppm males at 6 and 12 months (decreases in Hct were
observed only at 3 and 6 months). MCV was increased in female mice at the highest duration
(12 months) and exposure (250 ppm) levels. Reticulocyte counts were increased significantly in
the 125 ppm females at 3 and 6 months and in the 125 ppm males at 6 months of exposure.
Table 4-6 shows the responses available for a representative measure of the hematologic effects
from EGBE exposure. Hct levels for male and female rats and mice measured after 3 months or
12 months are presented.
30

-------
Table 4-6. Comparison of female and male rat and mouse Hct (manual)
values from 3- and 12-month inhalation exposures to EGBE

Control
31.2 ppm
62.5 ppm
125 ppm
250 ppm
Female rats3
3 mos
12 mos
46.5 ±0.5
45.4 ±0.2
46.1 ±0.5 (95)
43.3 ± 0.5° (93)
45.3 ±0.3 (100)
42.2	± 0.5° (91)
42.3	±0.4C (93)
-
Male rats3
3 mos
12 mos
44.9 ±0.2
47.8 ±0.4
46.9 ±0.5 (104)
44.8	±0.4 (100)
45.9	± 0.8b (96)
42.9 ± 0.5b (95)
42.9 ± 1.2° (90)
-
Female mice3
3 mos
12 mos
49.3 ±0.5
46.9 ±0.4
-
48.9 ±0.4 (99)
46.3 ± 0.4 (99)
46.2 ± 0.5° (94)
43.8 ± 0.4° (93)
43.7	± 0.5° (89)
41.8	± 0.3° (89)
Male mice3
3 mos
12 mos
47.5 ±0.3
47.9 ±0.4
-
47.3 ±0.5 (100)
48.7 ± 1.9(102)
46.0 ± 0.4b (97)
46.4 ± 1.0 (97)
43.7 ± 0.2° (92)
42.1 ± 0.4° (88)
aThese results are from a serial sacrifice conducted as a part of the 2-year chronic inhalation study. Values listed
are mean ± standard error (percent of control).
Statistically significant difference, p < 0.05.
Statistically significant difference,/) <0.01.
- = data were not available
Source: NTP (2000, 196293).
In vitro studies by Ghanayem (1989, 042014) have shown that the hemolysis caused by
the EGBE metabolite BAA is preceded by erythrocyte swelling. If the observed increase in
MCV is in response to cell swelling, it could be a preliminary indicator of the hemolytic effect.
Other researchers, however, have attributed the increased MCV at all exposures and the
increased MCH at higher exposures to the erythropoietic response subsequent to hemolysis and
the corresponding increase in the number of larger reticulocytes in circulation (cited in NTP,
2000, 196293). Reticulocyte counts were significantly increased in female rats at 62.5 ppm (6
and 12 months) and in male rats at 125 ppm (3 and 6 months). Since a statistically significant
increase in reticulocyte count was not observed at any time point in males or females exposed to
31 ppm or in males exposed to 62.5 ppm, it appears that reticulocyte count alone cannot account
for the increase in MCV. The observed increases in MCV may be a combined result of both
erythrocyte swelling prior to, and an increased number of reticulocytes subsequent to, hemolysis;
the former would be more influential at lower exposure levels, and the latter would have more
relative impact at higher levels.
Similar effects indicating anemia were also observed in mice, where females were more
sensitive. As in rats, the anemia response was observed at slightly higher doses, but was
persistent and did not progress or ameliorate in severity from 3 months to the final blood
collection at 12 months. Table 4-6 shows the manual Hct values in male and female rats and
mice at 3 months and 12 months. Statistically significant (p < 0.05) decreases in automated and
31

-------
manual Hct values and Hb and RBC counts occurred at 3, 6, and 12 months in the 125 ppm
females and the 250 ppm males and females. Statistically significant decreases in these
endpoints were also observed in 62.5 ppm females exposed for 6 months and in 125 ppm males
exposed for 6 and 12 months (decreases in Hct were observed only at 3 and 6 months). No
changes were observed in the MCV of mice, except for an increase in female mice at the highest
duration (12 months) and exposure (250 ppm) levels. Reticulocyte counts were increased
significantly in 125 ppm females at 3 and 6 months and in 125 ppm males at 6 months of
exposure.
At the end of the 2-year chronic bioassay (NTP, 2000, 196293). no significant neoplastic
effects were observed in male or female rats. In female rats, the combined incidence of benign
and malignant pheochromocytoma of the adrenal medulla was 3/50, 4/50, 1/49, and 8/49. The
incidence in the high-dose group (16%) does not represent a statistically significant increase over
the chamber control group (6%), but did exceed the historical control range (6.4 ± 3.5%; range
2-13%) for this effect.
Male mice exposed to 125 and 250 ppm EGBE had a low survival rate. A high rate of
hepatocellular carcinomas was found in these exposure groups (10/50 [control], 11/50, 16/49,
21/49); the increase at the high-exposure level was statistically significant (p < 0.01). When
hepatocellular adenomas and carcinomas are combined, no significant increase was observed in
any exposure group. However, the incidence of hemangiosarcomas in males exposed to
250 ppm (8%>) was significantly increased (p = 0.046) relative to chamber controls (0/50, 1/50,
2/49, 4/49) and exceeded the range of historical controls (14/968; 1.5 ± 1.5%; range 0-4%). No
organisms consistent with Helicobacter hepaticus were found in the 14 mice evaluated (NTP,
2000, 196293). The researchers concluded from this that H. hepaticus was not a factor in the
development of liver neoplasms. No significant increases in benign or malignant hepatocellular
tumors or hemangiosarcomas were noted in the female mice, and the incidence of hepatocellular
adenomas actually decreased significantly (p < 0.05) in relation to the control chamber group
(16/50, 8/50, 7/49, 8/49).
Forestomach squamous cell papillomas and carcinomas, combined, were significantly
increased (trend test = 0.003) in female mice relative to the chamber controls (0/50, 1/50, 2/50,
6/50). The incidence of these tumor types (12%) at the highest exposure level was statistically
significant and exceeded the range for the occurrence of these tumors in historical controls
(0.9 ± 1.1%; range 0-3%). The first incidence of these tumors appeared in the group exposed to
250 ppm at 582 days, compared with 731 days at 62.5 and 125 ppm. This indicates a decreased
latency period in the highest exposure group. While the incidence of these types of forestomach
tumors was not significantly increased over controls in male mice (1/50, 1/50, 2/50, 2/50), the
incidence of squamous cell papillomas (4%) in the two highest exposure groups exceeded the
range for historical controls (0.5 ± 0.9%; range 0-2%). The increased incidences of forestomach
neoplasms in males, as in females, occurred in groups with ulceration and hyperplasia.
32

-------
Section 4.6 has a discussion of the cancer data from this study. With respect to the
noncancer findings, a NOAEL could not be determined, and a LOAEL of 62.5 ppm was
determined in mice for hemosiderin deposition. In rats, a NOAEL could not be determined, and
a LOAEL of 31 ppm was determined for hemosiderin deposition.
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION
Due to the known reproductive toxicity, such as effects to male testes and sperm, of two
other glycol ethers, ethylene glycol methyl ether (EGME; 2-methoxyethanol) and ethylene glycol
ethyl ether (EGEE; 2-ethoxyethanol), the reproductive toxicity of EGBE was studied in a variety
of oral studies (Exon et al., 1991, 100112; Foster et al., 1987, 100116; Grant et al., 1985,
006770; Heindel et al., 1990, 042042; Nagano et al., 1979, 100179; Nagano et al., 1984, 100180;
NTP, 1993, 042063) and inhalation studies (Dodd et al., 1983, 066465; NTP, 2000, 196293)
using rats, mice, and rabbits. Several developmental studies have addressed EGBE toxicity from
conception to sexual maturity, including toxicity to the embryo and fetus, following oral (Sleet et
al., 1989, 042089; Wier et al., 1987, 042123). inhalation (Nelson et al., 1984, 031878; Tyl et al.,
1984, 100209). and dermal (Hardin et al., 1984, 042030) exposures in rats, mice, and rabbits. In
many instances, LOAELs and NOAELs were reported for both parental and developmental
effects; therefore, the developmental studies can also be used to assess systemic toxicity as well
as developmental toxicity.
EGBE did not cause significant effects in the reproductive organs, including testes, in any
study. In a two-generation reproductive toxicity study, fertility was reduced in mice only at very
high, maternally toxic doses (>1,000 mg/kg). Maternal toxicity, related to the hematologic
effects of EGBE, and relatively minor developmental effects have been reported in
developmental studies and are discussed below. No teratogenic effects were noted in any of the
studies. It can be concluded from these studies that EGBE is not significantly toxic to the
reproductive organs of adult males or females, or to the developing fetuses of laboratory animals.
As discussed in Section 4.2, Nagano et al. (1979, 100179) performed a toxicity study in
male JCL/ICR mice (five/group), using gavage doses of 0, 357, 714, or 1,430 mg/kg-day EGBE
5 days/week for 5 weeks. A LOAEL of 357 mg/kg-day based on reduced RBC count was
identified, but no changes in testes weight or histology were observed. In another study, Nagano
et al. (1984, 100180) used the same dosing regimen up to 2,000 mg/kg-day to test EGBE and
other glycol ethers. Testicular atrophy was observed for EGEE and EGME, but not for EGBE.
Grant et al. (1985, 006770) exposed male F344 rats (six/group) to gavage doses of 0, 500,
or 1,000 mg/kg-day EGBE and EGME for 4 days. Severe testicular atrophy was observed in rats
fed 500 mg/kg-day EGME, but no significant effect was noted in rats fed up to 1,000 mg/kg-day
EGBE.
Krasavage (1986, 066474) conducted a toxicity study using groups of 10 COBS
CD(Sprague-Dawley)BR adult male rats treated by gavage with 222, 443, or 885 mg/kg-day
33

-------
undiluted EGBE 5 days/week for 6 weeks. They found no effects on testicular weight and no
histopathological lesions in the testes, seminal vesicles, epididymides, or prostate gland at any
exposure level.
Foster et al. (1987, 100116) fed Alpk/AP (Wistar-derived) male rats single gavage doses
of 0, 174, 434, or 868 mg/kg BAA. Occasional significant decreases in the weight of the
prostate gland and seminal vesicles were observed, but the decreases were not time- nor dose-
related. No treatment-related lesions were noted following histologic examination of the testes,
epididymides, or prostate. BAA did not produce any changes in testicular cell populations when
introduced in vitro at 5 mM. Simultaneous testing of the acids of EGME and EGEE resulted in
significant spermatocyte cell loss and damage in vivo and in vitro.
Subchronic reproductive studies were conducted using male and female Swiss CD-I
mice. Heindel et al. (1990, 042042) exposed them to EGBE in drinking water at doses of 0, 700,
1,300, and 2,000 mg/kg-day for 7 days premating; subsequently, they exposed the mice for 98
days while cohoused as breeding pairs. The higher two dose levels resulted in mortality: 13/20
died during the study in the 2,000 mg/kg-day group and 6/20 died in the 1,300 mg/kg-day dose
group, as compared with 1 each in the 700 mg/kg-day group and the control group. Statistically
significant toxic effects seen in the 1,300 and 2,000 mg/kg-day dose groups with adult mice
included decreased body weight gain, increased kidney and liver weights, and dose-related
decreases in water consumption. Statistically significant developmental effects observed in the
1,300 and 2,000 mg/kg-day dose groups included decreased pup weight and fewer and smaller
litters produced per pair. A significant reduction (5%) in live pup weight was also observed in
the 700 mg/kg-day dose group. No statistically significant effect on fertility was observed in the
700 mg/kg-day dose group.
At the completion of the 98-day continuous breeding phase, F0 breeding pairs were
separated and housed individually, while exposure to EGBE continued. When the last litter was
weaned, a 1-week crossover mating trial was performed to determine effects by gender.
F0 males and females from the 1,300 mg/kg-day dose group were mated with male and female
control animals. The exposed mice had significantly lower body weights and increased relative
kidney weights, but reproductive organ weights, sperm motility and morphology, and estrous
cycle length and frequency did not differ from control mice. In the only histopathological
examination carried out on treated females, no kidney lesions were observed. The proportion of
successful copulation was equivalent in all groups, and no developmental effects were observed
in any offspring. However, the number of fertile females was significantly reduced in the group
where treated females were mated with control males, suggesting that fertility effects were
primarily due to effects on the female mice.
A final phase of this study assessed the fertility and reproductive effects of EGBE in first-
generation (Fl) pups. There were insufficient numbers of offspring to assess the two highest
dose groups, and no statistically significant effect on fertility was noted when offspring of the
34

-------
low, 700 mg/kg-day dose group were mated. Thus, the researchers concluded that the 700 and
1,300 mg/kg-day dose levels are considered to be NOAEL and LOAEL values, respectively, for
both maternal and reproductive effects. A minimal LOAEL for developmental effects was
700 mg/kg-day, where a very slight decrease in pup weight was observed.
A study by Exon et al. (1991, 100112). also discussed in Section 4.4.5, looked at
reproductive parameters in male rats. Groups of six Sprague-Dawley rats were exposed to
EGBE in drinking water at doses of 0, 180, or 506 mg/kg-day (males) for 21 days. While
testicular atrophy and necrosis and a reduced number of spermatogenic cells were observed in
males exposed to EGME, no statistically significant effects on fertility parameters were seen in
males at any of the tested doses of EGBE.
NTP (1993, 042063) evaluated the effects of EGBE on the reproductive systems of male
and female B6C3Fi mice (five/gender/group) following 2-week drinking water exposure to doses
of 93, 148, 210, 370, or 627 mg/kg-day for males and 150, 237, 406, 673, or 1,364 mg/kg-day
for females. No deaths were reported, and there were no effects on body weight. Thymus
weights were decreased in the highest male dose group. There were no treatment-related gross
lesions in any of the reproductive organs and histopathological examinations were not
performed. NTP (1993, 042063) also exposed male and female F344 rats (five/gender/group) to
EGBE for 2 weeks in drinking water. Male rats received doses of 73, 108, 174, 242, or
346 mg/kg-day, and females received 77, 102, 152, 203, or 265 mg/kg-day. No treatment-related
deaths occurred during the study, and no changes in body weight were observed in male rats that
could be related to treatment. However, female rats had lower weight gain in the highest dose
group. Water consumption was lowest in the highest dose group in both genders, and no
treatment-related gross lesions of reproductive organs were reported.
Dodd et al. (1983, 066465) (also discussed in Section 4.2) performed a 90-day subchronic
inhalation study on F344 rats. Male and female rats (16/gender/group) were exposed to EGBE
for 6 hours/day, 5 days/week at concentrations of 0, 5, 25, and 77 ppm. They reported no
changes in testicular weight or in the pathology of the epididymides or testes of male rats at any
exposure level; reproductive organs of the female rats were not examined histologically.
NTP (2000, 196293) performed chronic and subchronic inhalation studies of EGBE in
F344 rats and B6C3Fi mice in which reproductive organs were examined. In the subchronic
portion of the NTP (2000, 196293) studies, rats and mice (10/gender/group) were exposed to
concentrations of 0, 31, 62.5, 125, 250, and 500 ppm of EGBE 6 hours/day, 5 days/week for
14 weeks. Testicular degeneration was reported in 2/4 mice from the 500 ppm group that died or
were killed moribund. No other effects were noted in the reproductive organs of rats or mice.
Exposure concentrations were 0, 31, 62.5, and 125 ppm for groups of 50 F344/N rats and 0, 62.5,
125, and 250 ppm for groups of 50 B6C3Fi mice. No effects were noted in the reproductive
organs of either species, but the researchers reported that survival was significantly decreased in
male mice at 125 and 250 ppm (54.0 and 53.1%, respectively).
35

-------
Prenatal and postnatal developmental toxicity tests were conducted in CD-I mice by
Wier et al. (1987, 042123). Animals received 0, 350, 650, 1,000, 1,500, or 2,000 mg/kg-day via
gavage on gestational days (GDs) 8-14. Maternal toxicity included mortality of 3/6 animals in
the 1,500 mg/kg-day group and 6/6 in the 2,000 mg/kg-day group. Treatment-related clinical
observations were lethargy, abnormal breathing, and green or red vaginal discharge (the latter at
> 1,500 mg/kg-day). Based on clinical signs in the prenatal study, the LOAEL for maternal
effects was 650 mg/kg-day. The LOAEL for developmental toxicity was determined to be
1,000 mg/kg-day based on a statistically significant increase in the number of resorptions and a
reduced number of viable fetuses. The corresponding NOAEL for prenatal effects was
350 mg/kg-day. In the postnatal study, reproductive effects were evaluated in offspring of
CD-I mice administered EGBE via gavage at 0, 650, or 1,000 mg/kg-day on GDs 8-14.
Maternal body weight was lowered at 1,000 mg/kg-day. Survival and body weight gain of
offspring were unaffected by treatment. No statistically significant developmental effects were
observed. In a simultaneous study with EGEE, developmental toxicity was noted at doses below
maternal toxicity levels.
Developmental toxicity was investigated following the administration of EGBE in
distilled water by gavage to groups of 28-35 pregnant F344 rats at doses of 0, 30, 100, or
200 mg/kg-day on GDs 9-11, or doses of 0, 30, 100, or 300 mg/kg-day on GDs 11-13 (Sleet et
al., 1989, 042089). GDs 9-13 were the most critical periods of fetal cardiovascular
development. Food and water measurements, body and organ weights, clinical signs,
hematologic analyses of dams, amount of corpora lutea, uterine contents, and number of dead
and live fetuses were monitored. Maternal effects of EGBE given in either dosing sequence
included marked, dose-related reductions in body weight and/or weight gain, increases in kidney
and spleen weights, severe hematotoxicity as evidenced by a decrease in HCT, Hb, and RBC
counts, and an increase in reticulocytes at doses >100 mg/kg-day. No indications of
developmental toxicity were observed at the two lower doses. Viability of embryos was reduced
by EGBE treatment at the 200 mg/kg-day dose (GDs 9-11) but not at 300 mg/kg-day (GDs 9-
13). A decreased platelet count was noted in the fetuses at 300 mg/kg-day (GDs 9-13).
Cardiovascular or other types of malformations were not found at any dose. The LOAEL for
maternal toxicity was 100 mg/kg-day based on signs of hematotoxicity, with a NOAEL
established at 30 mg/kg-day. The LOAEL for developmental toxicity was 200 mg/kg-day based
on decreased viability of embryos, with a NOAEL for this endpoint at 100 mg/kg-day.
Sprague-Dawley rats (15/group) were exposed to 0, 150, or 200 ppm EGBE via
inhalation for 7 hours/day on GDs 7-15 (Nelson et al., 1984, 031878). Rats exposed to 200 ppm
showed some evidence of hematuria on the first day of exposure; no significant effects were
noted thereafter, or at any time in offspring. The LOAEL was 200 ppm for slight maternal
toxicity; a NOAEL was identified at 150 ppm. The NOAEL for developmental toxicity was
36

-------
200 ppm. Simultaneous testing revealed that 50 ppm exposures to EGME were toxic at all levels
of embryonic and fetal development.
Pregnant F344 rats (36/group) and New Zealand white (NZW) rabbits (24/group) were
exposed to 0, 25, 50, 100, or 200 ppm EGBE via inhalation for 6 hours/day on GDs 6-15 for rats
or GDs 6-18 for rabbits (Tyl et al., 1984, 100209). Fetuses were weighed and evaluated for
viability, body weight, and morphological development, including external, visceral, and skeletal
malformations. In rats, fetotoxicity was observed at 100 and 200 ppm in the form of retarded
skeletal ossification of vertebral arches or centra, sternebrae, or phalanges. Maternal toxicity
was also evident at 100 and 200 ppm by increased incidence of hematuria, reduced RBC count,
decreased weight gain, and reduced food consumption. For both maternal and developmental
toxicity in the rat, the NOAEL and LOAEL values were 50 and 100 ppm, respectively. In
rabbits, fetal skeletal ossification of sternebrae and rudimentary ribs was delayed at 200 ppm.
Maternal toxicity was also evident at 200 ppm as an increased incidence of clinical signs,
reduced gravid uterine weight, and decreased weight gain. For both maternal and developmental
effects in the rabbit, the NOAEL and LOAEL values were 100 and 200 ppm, respectively.
Reproductive toxicity tests were performed on female Sprague-Dawley rats (number not
specified) via dermal administration during GDs 6-15, 4 times/day at 1,800 and 5,400 mg/kg-
day (Hardin et al., 1984, 042030). In the highest dose group, 10/11 rats died between days 3 and
7 of treatment. Signs associated with treatment included red-stained urine, ataxia, inactivity,
rough coats, and necrosis of the tail tip. At the lower dose, body weight was slightly reduced, yet
there was no evidence of embryo- or fetotoxicity or gross malformations or variations.
4.4. OTHER STUDIES
4.4.1. Acute and Short-Term Exposure Studies
Ghanayem et al. (1987, 066470) conducted acute toxicity studies designed to assess the
effect of age on toxicity by comparing effects in treated young (4-5 weeks old) and adult rats (9-
13 weeks, 5-6 months, and/or 16 months old). The researchers exposed male F344 rats
(six/group) using single gavage doses of EGBE in water (99% purity) to concentrations of 0, 32,
63, 125, 250, 250, or 500 mg/kg-day. Evaluations included total RBC and WBC counts, urine
Hb concentration, organ weights, and histology of the liver, spleen, bladder, kidney, and testes.
Focal necrosis of the liver was observed in adult rats exposed at a dose of 250 or 500 mg/kg.
Hematologic effects were found to be dose- and age-dependent, with older rats being more
sensitive than younger rats. Significant decreases in RBC counts, Hct, and Hb and increases in
free plasma Hb occurred at 125 mg/kg-day in both young and adult rats, with the younger rats
exhibiting significantly less pronounced responses. The incidence of hemoglobinuria was also
dose- and age-dependent. Concentrations of free Hb in urine were also significantly higher in
older rats than in younger rats at all doses. These researchers suggested that the metabolic basis
37

-------
of the age-dependent toxicity of EGBE may be due to a reduced ability in older rats to
metabolize the toxic metabolite BAA to CO2 and a diminished ability to excrete BAA in the
urine. Based on increased Hb concentrations in the urine and associated hemolytic effects at
higher doses, the LOAEL for this study was determined to be 32 mg/kg-day for both young and
adult rats. A NOAEL was not identified.
Ghanayem and Sullivan (1993, 041609) performed acute oral toxicity studies in male
F344 rats (N not specified), using single gavage doses of 250 mg/kg-day EGBE in tap water.
MCV and Hct values increased immediately after treatment and decreased with time following
exposure. Hemolysis and decreases in Hb concentrations and RBC counts were reported.
Grant et al. (1985, 006770) gavaged groups of 24 male F344 rats with EGBE (purity
99.9%) in water at doses of 0, 500, or 1,000 mg/kg-day for 4 days. Six rats per dose were
examined at 1, 4, 8, and 22 days after the last dose. The animals were evaluated for changes in
body weight, hematology, organ weight, and histology. Hematology evaluations showed marked
dose-related effects on circulating RBCs and WBCs. Changes at 500 and 1,000 mg/kg-day on
postdosing day 1 included significant dose-related decreases in Hb concentration and total WBC
and lymphocyte counts and increases in MCV, reticulocyte counts, and MCH. Hct was also
reduced at 1,000 mg/kg-day. Most of the RBC changes subsequently returned to normal,
although MCV and MCH remained increased at day 22. Body weight gain was sufficiently
reduced throughout the posttreatment period at 1,000 mg/kg-day. Changes in relative organ
weights were evident on posttreatment day 1, including increased liver and spleen weights at 500
and 1,000 mg/kg-day and increased kidney and reduced thymus weights at 1,000 mg/kg-day.
These changes returned to normal by posttreatment day 22, except for liver and spleen weights at
1,000 mg/kg-day, which increased somewhat (~5 and -20%, respectively). The authors
determined that EGBE appears to be relatively inactive as a bone marrow toxicant due to the
observed proliferative response and the lack of hemorrhage at any time in the bone marrow of
EGBE treated animals. Based on hemolytic anemia with associated reticulocytosis and increased
hematopoiesis, a LOAEL was established at 500 mg/kg-day, the lowest dose tested. A NOAEL
was not identified.
Ghanayem et al. (1992, 100128) administered EGBE to male F344 rats (six/group) via
gavage for 12 days at dose levels of 0 and 125 mg/kg-day. These investigators identified effects
of EGBE exposure similar to those identified above. Significant hemolysis occurred, becoming
more pronounced up to the third day of dosing. Gradual recovery was observed up to day 12.
MCV, ATP concentration, reticulocyte counts, and relative spleen-to-body weight ratios
increased up to the sixth day of dosing and declined thereafter. Liver-to-body-weight ratios were
slightly lowered on days 3 and 6 and slightly increased on day 12.
Several studies investigated EGBE-induced effects on specific organs and cells. Four
male and four female F344 rats were exposed to two, three, or four daily doses of EGBE at
250 mg/kg-day. Ezov et al. (2002, 100113) investigated hemolytic anemia and disseminated
38

-------
thrombosis in rats by investigating the organs for hemolysis and histopathologic evidence of
disseminated thrombosis. Significant morphological changes in erythrocytes were noted in both
genders of rats, while disseminated thrombosis and infarction were seen mainly in females and
consisted of tissue necrosis in the brain, liver, bones, eyes, lungs, and heart. Renal tubular
necrosis associated with Hb casts was seen in both genders. Koshkaryev et al. (2003, 100168)
measured changes in adherence, aggregability, and deformability of RBCs. EGBE exposure did
not affect RBC aggregability, and its effect on deformability was inconclusive; however, the
exposure enhanced RBC adherence to endothelial cells, with adherence highest at day 2 (the first
day examined), after which it decreased sharply with time. Shabat et al. (2004, 100200) studied
bone marrow injury and reported extensive vascular thrombosis resulting in necrosis of bone
marrow cells, bone-lining cells, and cortical and trabecular osteocytes. The authors concluded
that, in EGBE-treated rats, interactions of several factors may generate a thrombotic crisis, such
as the release of procoagulant factors from destroyed erythrocytes; they further concluded that
disturbed blood flow may result from alterations in the rheology of erythrocytes, including self-
aggregation, deformation, and adherence to the endothelium of the blood vessel wall. Redlich et
al. (2004, 100194) investigated the dental effects from EGBE-induced hemolysis and
thrombosis. Odontoblastic necrosis in the dental pulp of incisors and molars and muscle-cell
damage in the tongue were observed; the most severe changes occurred in females. These
effects were probably the result of ischemic events in the blood vessels supplying these tissues,
rather than a direct cytotoxic effect of EGBE.
Corley et al. (1999, 042003) conducted a series of studies in B6C3Fi mice investigating
aspects of EGBE toxicity, including the occurrence of forestomach lesions in both oral and
inhalation exposure routes, the dose-response of forestomach irritation, and the occurrence of
forestomach lesions as a consequence of systemic-only exposure. To determine the extent that
activities during inhalation exposures (e.g., grooming) could contribute to observed forestomach
lesions, groups of female mice were exposed for 6 hours to target concentrations of 250 ppm
EGBE via whole-body (n = 20) and nose-only (n = 20) exposures and concentrations on the fur
that were available for oral consumption via grooming measured. For whole-body exposures,
9.2 ± 2.9 mg/kg was available compared with 7.5 ± 2.3 mg/kg for the nose-only exposures.
Little difference was detected in the internal dose of EGBE from whole-body versus nose-only
exposures, as measured by the analysis of EGBE and in BAA detected in blood and urine
postexposure. To inform the dose response of toxicity in the forestomach tissues in mice, neat
EGBE was administered to male and female mice (five/gender/dose) via gavage (no vehicle) for
1 week at doses of 100, 400, or 800 mg/kg-day. The dose in the 100 mg/kg-day group was
increased to 1,200 mg/kg-day after 2 days. Severe hemolysis and mortality were seen, and the
2-week study was terminated after only four doses. Forestomach lesions consisting of focal
areas of irritation and epithelial hyperplasia were seen at all exposure levels in this study. Next,
the researchers administered saline solutions of EGBE to groups of three mice by either
39

-------
intraperitoneal (i.p.) or subcutaneous (s.c.) injection at 400 or 600 mg/kg-day for 3 consecutive
days or 0 and 400 mg/kg-day for 5 consecutive days. Focal irritation in the forestomach, similar
to that seen in gavage and inhalation studies, was seen in the three mice administered EGBE by
i.p. injection at 600 mg/kg for 3 days, while 1/3 mice at 400 mg/kg i.p. and at 400 and 600 mg/kg
s.c. for 3 days also had forestomach lesions, minimal effects. At 400 mg/kg (5-day study),
1/6 mice (i.p.) and 2/6 mice (s.c.) also had minimal lesions. From these results, it can be
concluded that: (1) the contribution to forestomach exposure from grooming activities, etc.,
during whole-body inhalation exposures is incidental; (2) the exposure of forestomach lesions to
EGBE is similar from gavage and inhalation exposures; and (3) forestomach tissues show a
similar irritative response whether EGBE exposure is systemic or portal of entry.
In another series of studies, male and female B6C3Fi mice (16/gender/group) were
exposed by gavage to 0, 400, 800, or 1,200 mg/kg-day of neat EGBE for 2 days (Poet et al.,
2003). A high level of mortality was seen; the dose was reduced by half and the dosing
discontinued when survival did not improve after two additional doses. Lesions, including
epithelial hyperplasia and inflammation of the forestomach, were seen at the higher dose levels
in both males and females, and minimal-to-mild forestomach epithelial hyperplasias were seen in
both genders of the lower-dose groups. In a study similar to that performed by Corley et al.
(1999, 042003). female mice were exposed, either by whole-body or nose-only inhalation, to a
single 6-hour exposure of 250 ppm EGBE; the concentrations on the fur available for oral
consumption via grooming were measured. An average of 205 ± 69 |ig of EGBE was detected
on the fur of the mice exposed whole-body, while an average of 170 ± 52 |ig was detected on the
fur of the mice exposed nose-only (Poet et al., 2003, 100190).
Green et al. (2002, 041610) conducted a series of experiments to examine the effects and
distribution of EGBE in vivo. First, female B6C3Fi mice (five/group) were given a single, daily
gavage dose of either EGBE or BAA (0, 50, 150, or 500 mg/kg) for 10 days. Eighteen hours
after the last dose, animals were sacrificed to look at cell proliferation in the forestomach and
glandular stomach. The only dose-dependent finding was a thickening of the keratinized layers
or hyperkeratosis of the forestomach (statistics not reported). A NOAEL of 150 mg/kg EGBE
and 50 mg/kg BAA was observed. No effects were seen in the glandular stomachs at any dose
levels. In the next set of studies, female B6C3Fi mice (n = 12) were exposed whole-body to
250 ppm 2- butoxy[l-14C]ethanol (specific activity 0.365 mCi/mmol) for 6 hours. Following the
exposure, animals were given free access to food and water; they were terminated (four per time
point) at 5 minutes and 24 and 48 hours after exposure, and whole body autoradiography was
performed. Female B6C3Fi mice (n = 12) were given a single i.v. injection of 10 mg/kg
2-butoxy[l-14C]ethanol (850 (j,Ci/kg). The animals were then given free access to food and water
and terminated (four per time point) at 4, 24, and 48 hours after dosing, and whole body
autoradiograms were prepared. These studies showed that whether EGBE was delivered by
inhalation or by i.v., radiolabeled EGBE was found in the buccal cavity, esophagus, and
40

-------
stomach. This suggests that EGBE somehow enters the stomach via the buccal cavity and
esophagus following inhalation exposure.
Administration of a 2,000 mg/kg oral dose of EGBE to guinea pigs caused complete
mortality of females and 60% mortality of males (Shepard, 1994, 594430; Shepard, 1994,
594428). but a dose of 1,000 mg/kg caused only a 20% mortality of either gender. Clinical signs
and gross necropsy indicated toxicity was due to irritation of the stomach. There was no
evidence of hemolytic toxicity.
Gingell et al. (1998, 100135) performed acute oral and inhalation toxicity studies in the
guinea pig. A dose of 2,000 mg/kg EGBE was administered by gavage to five males and five
females. After excessive mortality (3/5 males, 5/5 females) was observed at this dose, reduced
doses of 500 and 1,000 mg/kg were administered. No animals died at 500 mg/kg, and 1/5 males
and 1/5 females died at 1,000 mg/kg. The acute oral median lethal dose for both genders was
1,414 mg/kg (95%) confidence interval [CI] = 1,020-1,961 mg/kg). Clinical signs in the guinea
pigs included slight-to-severe weakness, salivation and staining of face or abdomen hair, and
respiratory difficulties in a few males. No evidence of RBC toxicity or hemolysis was observed.
In the inhalation study, male and female guinea pigs were exposed for 1 hour to 633 ppm and
691 ppm, respectively. No mortality, clinical signs of toxicity, or exposure-related pathological
signs were noted. Thus, the median lethal concentration for a 1-hour exposure for guinea pigs
was >633 ppm in males and >691 ppm in females.
4.4.2. Dermal Exposure Studies
EGBE appears to be readily absorbed after contact with animal skin. Rats and rabbits
exhibit varying degrees of hematotoxicity following dermal application of EGBE (Allen, 1993,
594452; Allen, 1993, 594455; Allen, 1993, 594459; Allen, 1993, 594460; Bartnik et al., 1987,
100086; Tyler, 1984, 006769). Bartnik et al. (1987, 100086) performed acute dermal toxicity
tests using Wistar rats (six/gender/group). A single application of 200, 260, 320, 375, or
500 mg/kg EGBE was placed on the dorsal shaved skin of rats and covered with a glass capsule.
Hemolytic and/or hemoglobinuria effects were observed at 500 mg/kg EGBE within 6 hours of
application. No effects were observed at 200 mg/kg.
Repeated occluded application of EGBE either neat or as a dilute aqueous solution to
NZW rabbits (five/gender/group) at 18, 90, 180, or 360 mg/kg (6 hours/day, nine consecutive
applications) produced hemoglobinuria in males at 360 mg/kg and in females at 180 or
360 mg/kg (Tyler, 1984, 006769). Only female rabbits showed decreased RBC counts,
decreased Hb and MCHC, and increased MCH concentrations at the highest treatment level.
Recovery was noted after 14 days. In a separate 13-week study, occluded dermal administration
of EGBE to NZW rabbits (10/gender/group) at exposure levels of 10, 50, or 150 mg/kg for
6 hours/day, 5 days/week produced no observable hematological effects (Tyler, 1984, 006769).
41

-------
Occlusion or semi-occlusion of the site of EGBE administration was also a determining
factor. For example, some studies have shown no clinical signs of hematotoxicity in Sprague-
Dawley rats (five/gender/group) administered EGBE dermally at 2,000 mg/kg (24-hour
exposure) either semi-occluded or occluded (Allen, 1993, 594452; Allen, 1993, 594455).
However, clinical signs of systemic toxicity were noted following the occluded exposure. In
similar studies in NZW rabbits (five/gender/group), red-stained urine was reported at semi-
occluded doses of 2,000 mg/kg EGBE, along with other clinical signs of systemic toxicity
(Allen, 1993, 594459; Allen, 1993, 594460). Similar effects occurred at occluded doses of 500,
707, and 1,000 mg/kg in this species; deaths occurred at the 500 and 1,000 mg/kg exposures.
Thus, hematotoxicity varied from nonexistent to severe. In guinea pigs, dermal administration of
EGBE at 2,000 mg/kg produced no deaths, clinical signs of toxicity, or treatment-related signs of
organ toxicity (Gingell et al., 1998, 100135; Shepard, 1994, 594428).
In an assessment of immune parameters, female BALB/c mice (five/group) were
topically exposed to EGBE at 100, 500, 1,000, and 1,500 mg/kg-day for 4 consecutive days
(Singh et al., 2001, 100204). A statistically significant increase in spleen-to-body-weight ratio,
and a 29% increase in splenic cellularity was observed at 1,500 mg/kg. Splenic proliferative
responses to the T-cell mitogen, concanavalin-A (con-A), were significantly decreased by 32% at
500 mg/kg-day and 35% at 1,000 mg/kg-day. Allogeneic antigen-driven lymphoproliferative
responses in the mixed lymphocyte response were significantly reduced by 55% at 500 mg/kg-
day and 56%) at 1,000 mg/kg-day. However, natural killer (NK) cell activity, cytotoxic T-
lymphocyte activity, and the T-dependent plaque-forming cell (PFC) response were not
significantly affected by EGBE exposure. A dose of 100 mg/kg-day was a NOAEL.
Singh et al. (2002, 100203) exposed female BALB/c mice (five/group) via gavage to 50,
150, or 400 mg/kg EGBE, or topically on the ear to 0.25, 1.0, 4.0, or 16.0 mg EGBE. The
researchers measured the oxazolone (OXA)-induced contact hypersensitivity response (CHR).
Mice that received the gavage doses of EGBE for 10 consecutive days did not exhibit a
significantly altered OXA-induced CHR as measured by ear swelling 24 hours postchallenge. In
contrast, topical exposure to EGBE significantly suppressed the OXA-induced CHR at a dose of
4.0 mg EGBE/ear, but not at any other dose.
The studies indicate that while the dermal route of exposure can be expected to contribute
to overall exposure, the concentrations at which effects occur in animals are higher than those
found following oral and inhalation exposure. In humans, toxicokinetic studies have shown that
dermal absorption of EGBE vapors do contribute to the total body burden, showing the
importance of the dermal route of exposure.
4.4.3. Ocular Exposure Studies
EGBE has been found to be an irritant when instilled in rabbit (Jacobs and Martens,
1989, 100150; Kennah et al., 1989, 054156). Kennah et al. (1989, 054156) performed the Draize
42

-------
eye irritation test in rabbits. The percent EGBE concentration and corresponding scores by the
Texaco single-digit toxicity classification system were 100%—66, 70%—49, 30%—39, 20%—
2, and 10%—1. In an assessment that measured corneal thickness, the highest concentration was
classified as severely irritating, the 70% concentration was moderately irritating, and the others
were mildly irritating. Jacobs and Marten (1989, 100150) conducted ocular tests on NZW
rabbits (n = 6) to determine the effects of EGBE (100 [xL, 99% pure) on eye irritation. The
undiluted chemical was dropped onto the lower lid of one eye; the other eye served as a control.
The eyes were examined and graded for ocular reactions at 4, 24, 48, 72, 96, and 168 hours
postinstillation. The authors determined that EGBE should be classified as an irritant based on
the mean erythema scores and percent corneal thickening.
4.4.4. Genotoxicity
Although weakly genotoxic responses have been obtained in two laboratories (Elias et al.,
1996, 042011; Hoflack et al., 1995, 100147). EGBE is not expected to be mutagenic or
clastogenic based on the available data (summarized in Table 4-7). The NTP reported negative
responses for mutagenicity when EGBE was tested in Salmonella typhimurium strains TA97,
TA98, TA100, TA1535, and TA1537 at up to 10 mg/plate with and without metabolic activation
(Zeiger et al., 1992, 095748). However, Hoflack et al. (1995, 100147) reported that at 38
[j,mol/plate (4.5 mg/plate), EGBE induced a weak mutagenic response in salmonella tester strain
TA97a in the absence of S9 mix (Hoflack et al., 1995, 100147). The work of Hoflack and
colleagues was repeated by Gollapudi et al. (1996, 100137). and EGBE was found to be negative
in these tester strains when evaluated at 0.5, 1.0, 2.5, 5.0, 8.5, and 10 mg/plate in the presence
and absence of Aroclor-induced rat liver S9 mix. Thus, the weak positive result reported in
salmonella TA97a by Hoflack et al. (1995, 100147) is unconfirmed. A plausible explanation put
forth by Gollapudi et al. (1996, 100137) is that, given the sensitivity of the Ames test, perhaps
the weak positive result reported by Hoflack et al. (1995, 100147) is attributed to an impurity in
their test material.
43

-------
Table 4-7. Summary of genotoxicity studies on EGBE, BAL, and BAA
Type of test, test species
Dose3
Result
Reference
In vitro tests: EGBE
Reverse mutation, S. typhimurium,
TA97, TA98, TA100, TA1535,
TA1537
10 mg/plate
Negative (w/ and w/o
metabolic activation)
Zeigeretal. (1992,
095748)
(work done for NTP)
Reverse mutation, S. typhimurium his-
TA 98, TA 100, TA 102
115 (imol/plate
(14.0 mg/plate)
Negative (w/ and w/o
metabolic activation)
Hoflack et al. (1995,
100147)
Reverse mutation, S. typhimurium his-
TA97a
38 |imol/platc
(4.5 mg/plate)
Weakly positive (w/o
metabolic activation)
Hoflack et al. (1995,
100147)
Reverse mutation, S. typhimurium his-
TA97a, TA 100; Escherichia coli
WP2uvrA
10 mg/plate
Negative (w/ and w/o
metabolic activation)
Gollapudi et al. (1996,
100137)
Sister chromatid exchanges (SCEs),
micronuclei (MN) and aneuploidy (AP)
in V79 cells
10-100 mM (SCE)
8.46 mM (MN)
16.8 mM (AP)
Weakly positive (w/o
metabolic activation)
Elias et al. (1996,
04201 l)b
Potentiation of clastogenicity induced
by methyl methanesulfonate
8.5 mM
Positive (w/o metabolic
activation)
Elias et al. (1996,
04201 l)b
Chromosomal aberrations, V79 cells
and human lymphocytes
Not available
Negative (w/o metabolic
activation)
Elias et al. (1996,
04201 l)b
Gene mutation, Chinese hamster ovary
cells
38.1 mM°
(4.5 mg/mL)
Negative (w/o metabolic
activation)
Chiewchanwit and Au
(1995. 041999)
DNA damage, SVEC4-10 mouse
endothelial cells
10 mM
Negative
Klaunig and
Kamendulis (2005,
100165)
In vitro tests: BAL
Reverse mutation, S. typhimurium his-
TA 97a, TA 98, TA 100 and TA 102
43 (imol/plate
(5.0 mg/plate)
Negative (w/ and w/o
metabolic activation)
Hoflack et al. (1995,
100147)
Chromosomal aberrations, V79 cells
and human lymphocytes
0.1-1 mM;
cytotoxicity not
reported
Positive (w/o metabolic
activation)
Elias et al. (1996,
04201 l)b
DNA damage, SVEC4-10 mouse
endothelial cells
1 mM
Negative
Klaunig and
Kamendulis (2005,
100165)
44

-------
Type of test, test species
Dose"
Result
Reference
In vitro tests: BAA
Reverse mutation, S. typhimurium his-
TA 97a, TA 98, TA 100 and TA 102
8 (imol/plate
(0.9 mg/plate)
Negative (w/ and w/o
metabolic activation); dose
limited by toxicity
Hoflack et al. (1995,
100147)
SCEs and CAs, V79 cells
0.8 mM
Negative (w/o metabolic
activation)
Elias et al. (1996,
04201 l)b
Aneuploidy, V79 cells
0.38 mM
Weakly positive (w/o
metabolic activation)
Elias et al. (1995,
100147)b
MN assay, V79 cells
10 mM
Positive (w/o metabolic
activation
Elias et al. (1995,
100147)b
DNA damage, SVEC4-10 mouse
endothelial cells
10 mM
Negative
Klaunig and
Kamendulis (2005,
100165)
In vivo tests: EGBE
MNs, bone marrow erythrocytes of
male mice or rats
550 mg/kg-day, mice
450 mg/kg-day, rats
Negative
Negative
NTP (1996. 042064)
DNA adducts FVB/N mice
Sprague-Dawley rats
120 mg/kg-day, mice
and rats
No changes in DNA
methylation
Keith etal. (1996,
041625)
aDoses are either the lowest effective dose or the highest ineffective dose.
bAll in vitro assays were performed without the addition of an exogenous metabolic activation system.
The authors found that this dose was cytotoxic.
Elias et al. (1996, 042011) reported that EGBE did not induce chromosomal aberrations
in Chinese hamster V79 fibroblast cells but that EGBE, at treatment concentrations of > 8.5 mM,
weakly induced sister chromatid exchanges (SCEs) and micronuclei and potentiated the
clastogenicity induced by methyl methanesulfonate. Elias et al. (1996, 042011) also reported
that EGBE weakly induced aneuploidy (numerical chromosomal anomalies) in V79 cells;
however, this response was found at very high concentrations (16.8 mM EGBE).
When tested at doses nearing toxicity, EGBE and its metabolite BAL were not mutagenic
in an in vitro gene mutation assay using Chinese hamster ovary (CHO) cells (CHO-AS52)
(Chiewchanwit and Au, 1995, 041999). In contrast, Elias et al. (1996, 042011) reported that
both EGBE and BAL weakly induced gene mutations in Chinese hamster V79 cells only at high
treatment concentrations (> 7.5 mg/mL). It should be noted that Chiewchanwit and Au (1995,
041999) reported high cytotoxicity at 38.1 mM EGBE (4.5 mg/mL). The gene mutation data
presented by Elias et al. (1996, 042011) is in graphic form only with mean values and no SDs
presented. The presence or absence of cytotoxicity was not reported. BAL was also tested for
induction of DNA damage in the mouse endothelial cell line, SVEC4-10, using the comet assay.
BAL failed to produce a statistically significant increase in DNA strand breaks at any of the
concentrations or time points examined (Klaunig and Kamendulis, 2004, 594442; Klaunig and
Kamendulis, 2005, 100165; Reed et al., 2003, 594436). Other lines of evidence indicate that
direct interaction of BAL with the DNA molecules does not play a significant role in the
carcinogenic activity of EGBE. First, BAL causes cytotoxicity at levels associated with
45

-------
chromosome effects, and cytotoxicity itself can have effects that result in chromosome damage,
such as reduction in the repair of SCEs. Second, acetaldehyde is recognized as "weakly
mutagenic" and structural comparisons of the aldehyde metabolites of glycol ethers shows that
longer-chain aldehydes such as BAL are less mutagenic (Chiewchanwit and Au, 1995, 041999).
Third, if BAL were a stable mutagenic metabolite in any of the in vitro assays exposed to EGBE,
one would expect them to give positive results; however, the results were generally negative.
Elias et al. (1996, 042011) suggested that the V79 cells possess neither ALDH nor ADH. The
relevance of these studies, or of any systems that lack these enzymes, is of limited value in
elucidating the MOA of toxicity in biological systems that possess these enzymes. BAA has
been found negative for reverse mutations in S. typhimurium his" with and without metabolic
activation (Hoflack et al., 1995, 100147). Concentrations of up to 8 [j,mol/plate were tested, and
dose was limited by toxicity. BAA (up to 10 mM) was also found negative for induction of
DNA damage in SVEC4-10 mouse endothelial cells (Klaunig and Kamendulis, 2005, 100165)
and in an SCE assay in V79 cells (Elias et al., 1996, 042011). BAA was weakly positive for
aneuploidy in V79 cells at 0.38 mM and positive for micronuclei induction in the same cell line
at 10 mM, as reported by Elias et al. (1996, 042011). As noted above, the data means are
presented in graphic form without SDs and cannot be critically evaluated; no cytotoxicity data
are reported.
EGBE did not increase the incidence of micronuclei in the bone marrow cells of male
mice or rats (NTP, 1996, 042064). Animals were given three i.p. injections of EGBE 24 hours
apart and sacrificed 24 hours after the last injection; rats were dosed at 0, 7, 14, 28, 56, 112.5,
225, or 450 mg/kg and mice were dosed at 0, 17, 34, 69, 137.5, 275, or 550 mg/kg (NTP, 1996,
042064). There was high mortality (2/5 mice survived) in mice injected with 1,000 mg/kg doses
of EGBE. Keith et al. (1996, 041625) treated Sprague-Dawley rats and transgenic FVB/N mice
carrying the v-Ha-ras oncogene with a single oral dose of 120 mg/kg EGBE; there was no
increase in DNA adducts in the brain, liver, kidney, testes, or spleen of the rats, and no changes
in DNA methylation patterns in either species.
In conclusion, EGBE has been tested in conventional genotoxicity tests for its potential to
induce gene mutations in systems and cytogenetic damage both in vitro and in vivo. Available
data do not support a mutagenic or clastogenic mechanism for EGBE. Two laboratories (Elias et
al., 1996, 042011; Hoflack et al., 1995, 100147) reported weak genotoxicity responses in vitro at
toxic doses. These results, however, are questionable given limited published information.
Elliott and Ashby (1997, 100111) reviewed the results of the available genotoxicity studies on
EGBE and concluded that the data indicate that EGBE has no significant genotoxic activity.
4.4.5. Immunotoxicity
Based on the results of the Exon et al. (1991, 100112) study, it appears that the immune
system is not a sensitive target of EGBE. Groups of six Sprague-Dawley rats were exposed to
46

-------
EGBE in drinking water at doses of 0, 180, or 506 mg/kg-day (males) or 0, 204, or 444 mg/kg-
day (females) for 21 days. All rats were injected s.c. with heat-aggregated aqueous keyhole
limpet hemocyanin (KLH), a T-cell dependent antigen, on days 7 and 13 following the start of
dosing. Endpoints evaluated on day 21 included body weight, absolute and relative organ
weights (spleen, thymus, liver, kidney, testis), and histology of thymus, liver, kidney, and testis.
Splenic histology was not assessed because this tissue was used as a source of cells for immune
function assays. Immune function assays included NK cell cytotoxicity, antibody response to a
T-cell dependent antigen as determined by measurement of KLH-specific serum
immunoglobulin G levels, delayed-type hypersensitivity reaction, interleukin-2 and interferon
production, and spleen cell counts. Terminal body weights were somewhat lower than controls
in all exposed groups, and the decreases were statistically significant in all groups except the
180 mg/kg-day males. No dose-related changes in organ weights or histology were observed.
NK cell cytotoxic responses were significantly enhanced in males at 180 mg/kg-day and females
at 204 mg/kg-day. At the highest dose tested in males and females, the NK cell cytotoxic
response was increased above control, but this finding was not statistically significant.
A decreased NK cell cytotoxic response is an indication of compromised nonspecific immune
system integrity. Given that this study showed an increased response and no dose-response
relationship, these findings are not considered to be an indication of an adverse response. No
significant alterations in other immune parameters were noted.
Smialowicz et al. (1992, 004389) reported on results of experiments that examined the
primary antibody response to a T-cell independent antigen (trinitrophenyl-lipopolysaccharide
[TNP-LPS]) measured with a PFC assay to determine the immunotoxic potential of a variety of
glycol ethers. This author had previously reported that the shorter-chain glycol ether 2-
methoxyethanol (ME) and its principal metabolite 2-methoxyacetic acid (MAA) suppressed the
antibody response to TNP-LPS as measured by the PFC assay in F344 rats but not CD-I mice
(Smialowicz et al., 1992, 100205). Having established the sensitivity of F344 rats to suppression
of the antibody response to the T-cell independent antigen TNP-LPS by ME and MAA, the
authors examined other glycol ethers, including 2-butoxyethanol, in the same dose range, for
their ability to suppress the antibody response to TNP-LPS by using the same PFC assay. Male
F344 rats were immunized with a single i.v. injection of 0.5 mL of 40 |ig/mL TNP-LPS, then
dosed (six/dose group) by gavage with 50-400 mg/kg-day of various glycol ethers, including
EGBE (0, 50, 100, 200, 400 mg/kg-day) for 2 days. All rats exposed to 400 mg/kg-day EGBE
died, and the 200 mg/kg-day EGBE dose resulted in one dead and one moribund rat. This
finding was not unexpected, as the hematotoxicity of EGBE in older rats has been reported in the
literature (Carpenter et al., 1956, 066464; Ghanayem et al., 1987, 066470; Tyler, 1984, 006769).
EGBE did not suppress the primary antibody response to TNP-LPS in the PFC assay.
In an assessment of immune parameters, female BALB/c mice (five/group) were
topically exposed to EGBE at 100, 500, 1,000, and 1,500 mg/kg-day for 4 consecutive days
47

-------
(Singh et al., 2001, 100204). A statistically significant increase in spleen-to-body-weight ratio
was observed at 1,500 mg/kg, and splenic cellularity was increased by 29% at this dose. Splenic
proliferative responses to the T-cell mitogen, con-A, were significantly decreased by 32% at
500 mg/kg-day and 35% at 1,000 mg/kg-day. Allogeneic antigen-driven lymphoproliferative
responses in the mixed lymphocyte response were significantly reduced by 55% at 500 mg/kg-
day and 56%) at 1,000 mg/kg-day. However, NK cell activity, cytotoxic T-lymphocyte activity,
and the T-dependent PFC response were not significantly affected by EGBE exposure. For those
immune parameters measured, 100 mg/kg-day was aNOAEL.
Singh et al. (2002, 100203) exposed female BALB/c mice (five/group) via gavage to 50,
150, or 400 mg/kg EGBE or topically to 0.25, 1.0, 4.0, or 16.0 mg EGBE on the ear and
measured the OXA-induced CHR. Mice that received the gavage doses of EGBE for 10
consecutive days did not exhibit a significantly altered OXA-induced CHR as measured by ear
swelling 24 hours postchallenge. In contrast, topical exposure to EGBE significantly suppressed
the OXA-induced CHR at a dose of 4.0 mg EGBE/ear, but not at any other dose.
4.4.6. Other In Vitro Studies
Ghanayem (1989, 042014) compared the metabolic and cellular basis of EGBE-induced
hemolysis of rat and human erythrocytes in vitro. EGBE is not metabolized when incubated with
blood from male F344 rats and causes no hemolysis or metabolic alterations at concentrations of
up to 10 mM. A concentration of 20 mM EGBE was required to produce significant hemolysis
of rat blood. This may be due to a nonspecific effect occurring at a concentration that is not
physiologically relevant. In contrast, incubation of rat blood with BAL or BAA at
concentrations of 0.5, 1.0, or 2.0 mM caused a time- and concentration-dependent increase in
cell swelling (i.e., increased Hct) followed by hemolysis. This response was more pronounced
for BAA, with nearly complete hemolysis observed after a 4-hour incubation at 2.0 mM. BAL
produced only slight hemolysis under the same conditions. The addition of ADH (with nicotine
adenine dinucleotide cofactor) to rat blood followed by BAL produced a potentiation of the
hemolytic effects. Addition of cyanamide, an ADH inhibitor, significantly decreased the effects
with or without added ADH. Both BAL and BAA caused a time- and concentration-dependent
decrease in blood ATP concentrations, although this effect may be secondary to the swelling and
lysis observed. Addition of exogenous ATP failed to reverse the hemolytic effects. Neither
EGBE nor its metabolites, BAL and BAA, caused any detectable changes in the concentrations
of glutathione (GSH) or glucose-6-phosphate dehydrogenase (G6PD) in rat erythrocytes. Blood
from male and female human volunteers was unaffected by 4-hour incubations with BAA at
concentrations of up to 4.0 mM. At 8 mM, slight but significant hemolysis of human blood was
observed: blood from female volunteers showed a slightly greater sensitivity. These studies
show that the erythrocyte membrane is the likely target for BAA, and that humans of both
genders are relatively insensitive to the hemolytic effects of BAA, as compared with rats.
48

-------
Udden (2000, 042110) compared effects on RBC morphology in male F344 rats
(five/group) dosed with 125 or 250 mg/kg EGBE via gavage with the effects of incubation of rat
erythrocytes in vitro with BAA (1 or 2 mM). In vivo exposure resulted in stomatoacytosis and
spherocytosis in blood smears and cup-shaped cells and spherocytosis in fixed samples. In vitro
incubation resulted in erythrocytes with cup shapes and spherocytosis in the fixed samples.
Since in vivo and in vitro exposure caused similar changes in RBC morphology in rats, this study
provides additional evidence for the role of BAA in causing the hemolytic effects from EGBE
exposure in rats.
Udden (2002, 042 111) performed in vitro comparisons of sub-hemolytic and hemolytic
effects in rat and human RBCs in response to exposure of high BAA, using samples from
hospitalized adults, hospitalized children, and well adults. Erythrocyte parameters measured
included deformability, size distribution, density, MCV, count, osmotic fragility, and hemolysis.
Exposure for 4 hours resulted in loss of deformability, although at concentrations 150-fold
different; the first significant alterations noted in cells exposed was at 0.05 mM BAA in rats and
at 7.5 mM BAA in humans. Significant increases in MCV after a 4-hour exposure to BAA were
noted at 0.05 mM in rats and at 10.0 mM in humans, a 200-fold difference. A similar differential
in concentration range was noted in osmotic fragility. Testing of erythrocytes for changes in
MCV and percent hemolysis from in vitro exposure to either 0 or 10 mM BAA showed no
differences among the well adults or the hospitalized children. The percent hemolysis seen in the
average response of the hospitalized adults to 10 mM BAA was significantly increased, from
0.54 to 0.69%. Significant increases in MCV were noted for all three of these groups, including
the well adults. The mechanisms underlying erythrocyte damage and the resistance of human
cells to these effects remain unknown. It should be noted, however, that although the sub-
hemolytic responses were similar for human and rat erythrocytes, other aspects of the
erythrocytes, including morphology and cell density, were altered in the rats but not in human
cells. These differences between species suggest that the mechanisms underlying the observed
prehemolytic effects may be different between rats and humans.
The possibility exists that certain human subpopulations, including the aged and those
predisposed to hemolytic disorders, might be at increased risk from EGBE exposure. Udden
(1994, 042109; 1996, 042209) investigated this possibility using blood from the elderly (mean
age 71.9; range 64-79 years; five men and four women), from seven patients with sickle cell
disease, and from four subjects with hereditary spherocytosis, three of whom were studied
postsplenectomy and one studied presplenectomy. Using a sensitive assay for erythrocyte
deformability (Udden, 1994, 042109; Udden and Patton, 1994, 056374). it was shown that blood
from all of these potentially sensitive groups was unaffected by incubations of up to 4 hours with
2 mM BAA.
Udden and Patton (2005, 100213) examined the role of osmolarity and cation
composition of the cell suspension buffers in the mechanism of BAA-induced hemolysis of rat
49

-------
RBCs. Adding sucrose to the cell suspension media or replacing external sodium with potassium
protected rat RBCs from BAA-induced hemolysis. The authors also observed that adding as
little as 0.05 mM CaCl2 to the buffer delayed the time course of the hemolytic response, while
adding MgCh had no effect. Use of the calcium-activated potassium channel inhibitor
charybdotoxin blocked the protective effect of calcium. From this, the authors suggest that BAA
causes sodium and calcium to enter the cell. While calcium initially has a protective effect via a
loss of potassium through the calcium-activated potassium channel, compensating for the
osmotic effect of increased cell sodium, calcium may subsequently have other deleterious effects
through activation of proteases and other calcium-activated processes.
Corthals et al. (2006, 100102) tested EGBE, BAL, and BAA in the comet assay to
determine their ability to induce DNA damage in SVEC4-10 mouse endothelial cells. EGBE (up
to 10 mM), BAL (up to 1 mM), and BAA (up to 10 mM) did not produce significant increases in
DNA damage relative to controls at any of the time points examined (2, 4, and 24 hours). The
researchers next tested the effect of hemolyzed mouse RBC lysate and ferrous sulfate in the same
system and found that the hemolyzed RBCs produced a statistically significant increase in DNA
damage at the highest concentration tested (10 x 106 hemolyzed RBCs) for 4 hours. No other
time points were significant. Ferrous sulfate produced statistically significant increases in DNA
damage at the highest time point and the lowest concentration tested (24 hours, 0.1 |iM) and at
all time points (2, 4, and 24 hours) in the mid- and high doses tested (0.5 [iM and 1.0 (j,M). The
next experiment examined the ability of EGBE, BAA, ferrous sulfate, and hemolyzed RBCs to
stimulate tumor necrosis factor-alpha (TNFa) release from cultured mouse macrophages
(RAW 264.7 cells). Hemolyzed RBCs (10 x 106 cells) resulted in a statistically significant
increase (p < 0.05) in TNFa release following a 4-hour treatment. Treatment with EGBE, BAA,
or ferrous sulfate did not result in increased TNFa release. Finally, the authors report that
macrophages activated with hemolyzed RBCs (10 x 106) for 4 hours were able to increase DNA
synthesis in mouse endothelial cells through co-culturing for 24 hours. These macrophages were
not, however, able to increase endothelial cell DNA damage (after 4- or 24-hour treatment) as
measured by the comet assay. The authors did find that LPS activated macrophages, after a
4-hour treatment, produced statistically significant increases (p < 0.05) in endothelial cell DNA
damage, as measured by the comet assay.
4.5. SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS AND
MODE OF ACTION: ORAL AND INHALATION
Hemolytic anemia is the primary response elicited in sensitive species following
inhalation, oral, or dermal administration of EGBE. The carboxylic acid metabolite of EGBE,
BAA, has been shown to be the causative agent in this hemolysis. The mechanisms underlying
the hemolytic events are unknown. In vitro tests have shown that BAA produces a
concentration- and time-dependent swelling of rat erythrocytes, and changes the normal
50

-------
erythrocyte morphology from the typical discocyte form to a spherocyte form prior to lysis. One
potential mechanism that could explain the RBC effects of BAA is the activation of membrane
phospholipid scrambleases. The outer leaflet of the RBC membrane normally has a net positive
charge due to the abundance of phosphatidylcholine while the cytoplasmic side contains
phosphatidylserine. Activation of phospholipid scrambleases results in the translocation of
phosphatidylserine to the outer leaf while shuffling phosphatidylcholine to the cytoplasmic side
of the membrane. The externalization of the phosphatidylserine changes the outer membrane net
charge to negative. The perturbations in membrane homeostasis may result in a loss of
membrane integrity leading to a loss of deformability of the erythrocyte. The fate of the
damaged red cells, whether by direct lysis in the circulation or splenic sequestration, is unknown.
It is likely that under low EGBE exposure conditions, splenic sequestration will predominate but
at higher exposure conditions, splenic spillage and/or frank intravascular lysis will occur. Heme
transport mechanisms would then be overloaded and iron containing fragments will accumulate
in the phagocytic cells of the liver (Kupffer cells) as hemosiderin. It is interesting to note that
phosphatidylserine externalization has been associated with impending apoptosis in nucleated
cells. Older erythrocytes are more sensitive to the hemolytic effects of BAA than are younger
cells or newly formed reticulocytes. Hemolysis can be induced in vivo following administration
of EGBE or in vitro following addition of BAA to either whole blood or washed erythrocytes.
Additionally, it has been reported that EGBE exposure to rats caused disseminated
thrombosis and infarction (Ezov et al., 2002, 1001 13; Nyska et al., 1999, 042747; Nyska et al.,
1999, 042746; Ramot et al., 2007, 100192; Yoshizawa et al., 2005, 100224). The mechanism by
which this occurs is presently unknown but several possibilities exist. The changes in
erythrocyte morphology and decrease in deformability could result in intravascular occlusion.
Secondly, the lysis of the erythrocyte would result in release of procoagulants. Finally, the
thrombotic response could be the result of the appearance of anionic phospholipids, particularly
phosphatidylserine, on the cell surface activating the prothrombinase complex resulting in the
formation of thrombin (Bevers et al., 1982, 100088; Connor et al., 1989, 100097). Nevertheless,
thrombosis only occurs at higher EGBE exposure levels and is not the most sensitive endpoint.
The primary response of hemolysis is indicated via dose-related clinical observations of
decreases in Hct, Hb concentration, and RBC count in the blood of laboratory animals exposed
to EGBE. The hemolysis-related events of macrocytosis and increased MCV were also observed
in the rat, considered to be a sensitive species, and are attributed at least partly to the increased
number of larger reticulocytes in the circulation following the erythropoietic compensatory
response (NTP, 2000, 196293). These alterations were persistent throughout the chronic animal
exposures but do not appear to progress with extended exposure (from 3 months to 1 year for
rats). These changes persist despite functioning, compensatory, homeostatic mechanisms.
Liver effects were noted in the NTP (2000, 196293) reports of subchronic and chronic
inhalation studies in rats and mice and in the subchronic drinking water study in rats. These
51

-------
included statistically significant increases in the iron-staining hepatic pigmentation attributed to
hemosiderin, the storage protein for insoluble iron, believed to be derived from the Hb released
during hemolysis. Nyska et al. (2004, 056379) examined the possible association between
chemically induced hemosiderin deposition and hemangiosarcomas in the liver of mice from 130
NTP bioassays, and found a highly significant association between liver hemangiosarcoma and
apparent Kupffer cell pigmentation (see Roberts et al. (2007, 100196) for an overview of the role
of Kupffer cells in hepatotoxicity and carcinogenicity). The cause for the hemosiderin
deposition in all cases was the erythrocyte hemolytic effect of the compounds. However, in an
NTP (1993, 042063) subchronic drinking water study, hepatocellular cytoplasmic changes were
observed in male rats at an exposure level (750 ppm) below the level at which hematological
changes were recorded (1,500 ppm). This finding raises the possibility of a direct, primary
hepatic toxicity due to either EGBE or an EGBE metabolite. Similar liver effects observed in
female rats at the 750 ppm exposure level were accompanied by hematological effects.
In the same NTP (1993, 042063) report, no liver lesions were reported in mice exposed to
drinking water containing up to 6,000 ppm EGBE. The lesions reported in rats consisted of
cytoplasmic alterations, hepatocellular degeneration, and pigmentation. The cytoplasmic
alterations, the only lesion observed at the 750 ppm exposure level (corresponding to a
consumption of roughly 55 mg/kg-day EGBE for adult male rats), were described as hepatocytes
staining more eosinophilic and lacking the amphophilic-to-basophilic granularity of the
cytoplasm present in hepatocytes from control animals. Greaves (2000, 089176) suggested that
the lack of cytoplasmic granularity or ground-glass appearance of the hepatocytes is an
indication that the response does not involve enzyme induction. The hepatocellular degeneration
and pigmentation observed at the higher exposure levels in both genders was centrilobular,
which is consistent with the Kupffer cell pigmentation and hemosiderin deposition reported in
the NTP (2000, 196293) inhalation studies. This information, along with the observation that all
other rat and mouse oral and inhalation studies of EGBE report hemolysis at or below exposure
levels that result in liver effects, suggests that at least these cytoplasmic hepatocellular changes
in male rats reported in the NTP (1993, 042063) drinking water study may reflect adaptation to a
subclinical level of hemolysis. However, focal necrosis of the liver observed in male rats
following gavage administration of 250 and 500 mg/kg EGBE (Ghanayem et al., 1987, 041608)
was judged to be inconsistent with typical anoxic centrilobular necrosis associated with anemia
(Edmonson and Peters, 1985, 042168). The effects observed in the Ghanayem et al. (1987,
041608) study may be associated with the high bolus exposures employed.
The liver alterations documented throughout these studies suggest a defined progression
of pathological events with increasing doses of EGBE with increasing levels of hemolysis. In
particular, hepatic hemosiderin deposition in the liver is a dose-related sequela of the hemolytic
activity caused by EGBE exposure. This deposition was noted to follow a dose-response
relationship as well as to increase in severity in the chronic rat and mouse NTP studies; it shows
52

-------
a statistically significant increase relative to the chamber controls. Although some hemosiderin
deposition in the spleen and liver may be expected with increasing age, the extent of hemosiderin
deposits noted by NTP in the livers of EGBE exposed animals is not normal. For these reasons,
hemosiderin deposition in the liver has been considered a pathological finding (Muller et al.,
2006, 597291). The following issues relate to the relevance of these hemolytic and hepatic
effects to humans and to the MO A of EGBE.
The weight of evidence from a variety of studies in animals and humans suggests that
certain species are more susceptible to the hemolytic effects of EGBE. It appears that humans
are less sensitive to the hemolytic effects of EGBE than rats and mice. On one end of the
sensitivity range is the guinea pig, which displays no hemolytic effects from EGBE at exposure
levels as high as an oral dose of 1,000 mg/kg or a dermal dose of 2,000 mg/kg. The rat,
conversely, displays increased osmotic fragility of erythrocytes at single inhalation exposures
below 100 ppm and single oral exposures below 100 mg/kg EGBE. No hemolysis was observed
in controlled laboratory acute inhalation exposures of human volunteers at up to 195 ppm EGBE.
Some reversible hemolytic effects have been described in addition to more debilitating effects in
humans who consumed single oral doses of 400-1,500 mg/kg EGBE in cleaning formulations
(see Section 4.1). Effects in humans from chronic exposure to EGBE have not been studied.
With respect to gender sensitivity, it has been consistently noted (Carpenter et al., 1956,
066464; Dodd et al., 1983, 066465; Ezov et al., 2002, 1001 13; NTP, 2000, 196293) that female
rats are more sensitive to EGBE-induced hemolysis than males. This gender difference is
consistent with toxicokinetic data for male and female rats reported by the NTP (2000, 196293)
2-year study. Female rats eliminated BAA, the toxic metabolite of EGBE, more slowly from the
blood, resulting in a larger AUC for the blood concentration of BAA versus time. This may be a
result of the reduced renal excretion observed in female versus male rats. NTP (2000, 196293)
also reported that, like female rats, female mice tended to have greater blood concentrations of
BAA at any given time than males. This may explain the slight increase in incidence and
severity of the anemic response found in female mice, as compared to males. However, unlike
female rats, female mice excrete slightly more BAA than male mice; no significant difference
between female and male mice has been noted in the overall rate of elimination or the ty2 of
BAA.
Some studies (Ghanayem et al., 1987, 066470; Ghanayem et al., 1990, 042016) were
designed to assess the effect of age on the toxicokinetics and hemolytic effects in young and
adult rats treated with single EGBE gavage exposures. Both blood retention (Ghanayem et al.,
1990, 042016) and hematologic effects (Ghanayem et al., 1987, 066470) were found to be dose-
and age-dependent; older rats retained more of the EGBE metabolite BAA in their blood and
were more sensitive than younger rats. The increased blood retention of BAA (as measured by
increased Cmax, AUC, and tu) in older rats versus younger rats may be due to metabolic
differences or compromised renal clearance. The researchers suggested that the pharmacokinetic
53

-------
basis of the age-dependent toxicity of EGBE may be due to a reduced ability of older rats to
metabolize the toxic metabolite BAA to CO2, and a diminished ability to excrete BAA in the
urine. No comparable studies exist for an analysis of liver effects.
While older rats appear to be more severely affected by acute doses of EGBE, continuous
exposures appear to impart a certain level of tolerance to rats and mice over time. Apparent
tolerance to EGBE-induced hemolysis in rats and mice has been seen in subchronic (Grant et al.,
1985, 006770; Krasavage, 1986, 066474) and chronic (NTP, 2000, 196293) studies. Ghanayem
et al. (1990, 042016; 1992, 100128) investigated this effect in the male F344 rat. Daily gavage
administration of EGBE at 125 mg/kg for 12 days resulted initially in hemolytic anemia, which
was more pronounced following the third day, but the animals recovered gradually to near
pretreatment levels by day 12. Additionally, rats treated for 3 days at 125 mg/kg followed by a
7-day recovery period were significantly less sensitive to subsequent treatment with EGBE at
either 125 or 250 mg/kg, as were rats that were bled and subsequently treated. Ghanayem et al.
(1990, 042016; 1992, 100128) proposed that the tolerance to hemolysis following repetitive
dosing is not due to changes in EGBE metabolism, but to the replacement of older and more
susceptible erythrocytes with younger, less susceptible cells. However, chronic studies in rats
and mice (NTP, 2000, 196293) have shown that any increased resistance imparted by these
immature erythrocytes diminishes with age. Rats and mice chronically exposed to EGBE
experienced anemia that persisted with no apparent progression or amelioration of severity for 9
months, up to the final blood collection at 12 months of age. There may be a balance in these
rodents between the release of reticulocytes to the circulation and the aging process, so that the
level of susceptible cells and severity of anemia remains relatively constant.
A number of secondary effects resulting from the hemolytic toxicity of EGBE have been
reported in studies with rats, mice, and rabbits. In the rat, the organs generally affected include
most prominently the liver (see discussions above) but also the kidneys, spleen, bone marrow,
and, to a lesser extent, the thymus (Exon et al., 1991, 1001 12; Grant et al., 1985, 006770; NTP,
1993, 042063; Shabat et al., 2004, 100200). Typically, increased liver and kidney weights are
observed with corresponding decreases in body weights at doses that produce a hematotoxic
response. Accompanying this is hepatocellular degeneration, hemosiderin deposition in the liver,
and congested spleens. Renal damage is often reported, accompanied by hemosiderin
accumulation, renal tubular degeneration, and intracytoplasmic Hb. Often these effects are more
pronounced in females. Hematopoiesis in bone marrow and spleen, increased cellularity of bone
marrow, and splenic congestion are all secondary to the hematotoxicity of EGBE and develop as
a compensatory response to hemolysis. In addition, intact erythrocytes have been observed
histopathologically in spleens from EGBE-treated rats, but not in spleens from control animals.
This suggests an increased rate of removal of damaged erythrocytes in EGBE-treated rats
(Ghanayem et al., 1987, 066470). Mild lymphopenia and neutrophilia were observed at
hemolytic doses of EGBE (Ghanayem et al., 1987, 066470) and were reported to be consistent
54

-------
with a "stress" leukogram produced by the release of endogenous corticosteroids (Wintrobe et
al., 1981, 100222). Neutrophilia, commonly associated with acute hemolysis or hemorrhage
(Wintrobe, 1981, 594424). was also observed.
In the NTP (2000, 196293) study, the incidence of hyaline degeneration of the olfactory
epithelium was significantly increased in all exposed groups of male rats and in the 62.5 and
125 ppm groups of females. Hyaline degeneration is the accumulation of intracytoplasmic
globules of highly eosinophilic material in the epithelial cells. While not unique to EGBE,
hyaline degeneration has been shown to occur in rodents exposed to other gases and vapors like
dimethylamine (Buckley et al., 1985, 062668) and pyridine (Nikula et al., 1995, 100182). It also
has been shown to develop in unexposed aged animals (St Clair and Morgan, 1992, 594426).
In conclusion, humans are significantly less sensitive to the hemolytic toxicity of EGBE
than are typical laboratory species such as mice, rats, or rabbits, although human erythrocytes do
appear capable of responding similarly to the causative EGBE metabolites, albeit at much higher
exposures. This marked species difference in sensitivity has been demonstrated in several
laboratory studies and through the use of in vitro studies using either whole blood or washed
erythrocytes. Based on the results of in vitro testing, blood concentrations of the hemolytically
active metabolite BAA must reach levels in human blood in excess of 7.5 mM for prehemolytic
changes to occur. Comparable effects in rat blood occur at in vitro concentrations approximately
150-fold lower. In addition, blood from potentially sensitive individuals, including the elderly or
those with congenital hemolytic disorders, does not show an increased hemolytic response when
incubated with up to 2 mM BAA for 4 hours. Based on simulations from PBPK modeling,
6-hour whole-body exposure of humans to saturated atmospheres of EGBE will result in
maximum blood concentrations of BAA below those needed to produce hemolysis (Corley et al.,
2005, 100100).
Most of the liver alterations documented throughout the EGBE exposure database are
related directly to hemolysis. Prominent among these alterations is hemosiderin deposition, a
pathological finding whose occurrence is related to hemolysis (Muller et al., 2006, 597291).
Humans also experience hemosiderin deposition in the liver, principally in hepatocytes, as a
consequence of excessive hemolysis, such as with thalassemia, a hereditary form of hemolytic
anemia (Iancu et al., 1976, 100149). Hemosiderin deposition in the liver is a pathological
response that follows the precursor hematologic effects, which, as a group, do not appear to
progress with changes in duration of exposure from subchronic to chronic.
4.6. EVALUATION OF CARCINOGENICITY
4.6.1. Summary of Overall Weight of Evidence
Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005, 086237). EGBE
is deemed "not likely to be carcinogenic to humans" at environmental concentrations below or
55

-------
equivalent to the RfD and RfC, based on laboratory animal evidence, mode-of-action
information, and limited human study information. The available data indicate that carcinogenic
effects from EGBE are not likely to occur in humans in the absence of the critical noncancer
effects, including hepatic hemosiderin staining and irritant effects at the portal of entry, and are
not likely to be carcinogenic to humans exposed to levels at or below the RfC and RfD values
derived in this assessment. Carpenter et al. (1956, 066464) (see Section 4.1) reported that no
changes in erythrocyte osmotic fragility were found in human subjects exposed to up to 195 ppm
(942 mg/m3; -600 times the RfC) for two 4-hour periods separated by a 30-minute break. At
oral doses of 400-500 mg/kg with a one-time bolus dose (see Section 4.1), hematuria has been
noted in two human case reports. This dose is 3,000-3,500 times the RfD and would need to be
sustained for a significant period of time to produce hemosiderin deposition. This is unlikely to
occur because the primary response of humans to high oral doses of EGBE, as shown in the case
studies in Section 4.1, is metabolic acidosis, which, if not treated, can lead to shock and
eventually death. No information is available on the carcinogenic effects of EGBE via the oral
or inhalation route in humans. A 2-year inhalation bioassay with mice and rats (NTP, 2000,
196293) reported tumors of the liver in male mice, forestomach tumors in female mice, and
tumors of the adrenal medulla in female rats. Nonneoplastic effects in rats included hyaline
degeneration of the olfactory epithelium and Kupffer cell pigmentation. Nonneoplastic effects in
mice included forestomach ulcers and epithelial hyperplasia, hematopoietic cell proliferation,
Kupffer cell pigmentation, hyaline degeneration of the olfactory epithelium (females only), and
bone marrow hyperplasia (males only).
EGBE has been tested in conventional genotoxicity tests for its potential to induce gene
mutations in vitro and cytogeneticity in both in vitro and in vivo, and the available data do not
support a mutagenic or clastogenic mechanism for EGBE. Two laboratories (Elias et al., 1996,
042011; Hoflack et al., 1995, 100147) did report weak genotoxicity responses in vitro at high
treatment concentrations, but results were not replicated in five other labs reporting negative
results.
The hypothesized MOA for the tumors observed following EGBE treatment involves
exposure to high doses for prolonged periods of time. This MOA is described in the sections that
follow. The weight of evidence indicates that EGBE is not likely to be carcinogenic to humans
at expected environmental concentrations.
4.6.2. Synthesis of Human, Animal, and Other Supporting Evidence
NTP (2000, 196293) conducted a 2-year inhalation study on EGBE in both genders of
F344/N rats and B6C3Fi mice. Rats (50/gender/group) were exposed to concentrations of 0, 31,
62.5, and 125 ppm (0, 150, 302, and 604 mg/m3) and mice (50/gender/group) were exposed to
concentrations of 0, 62.5, 125, and 250 ppm (0, 302, 604, and 1,208 mg/m3). The NTP report
56

-------
stated that the highest exposure was selected to produce a 10-15% depression in hematologic
indices, and survival was significantly decreased in male mice at 125 and 250 ppm (54.0 and
53.1%, respectively). While the NTP researchers report that no effect on survival was observed
in rats, the female rats appeared to show a trend toward decreased survival that may have been
attributable to the hematological effects. Mean body weights of rats exposed to 31 and 62.5 ppm
were similar to those of control animals. Mean body weights of the exposed mice were generally
less than those of controls, with females experiencing greater and earlier reductions. From
week 17 to the end of the study, the mean body weights of 125 ppm female rats were generally
less than those of controls.
At the end of the 2-year chronic bioassay (NTP, 2000, 196293). neoplastic effects were
observed in female rats and in male and female mice. In female rats, the combined incidence of
benign and/or malignant pheochromocytoma of the adrenal medulla was 3/50, 4/50, 1/49, and
8/49. The incidence in the high-dose group (16%) did not represent a statistically significant
increase over the chamber control group (6%), but it exceeded the historical control (6.4 ± 3.5%;
range 2-13%) for this effect.
The low survival rate in male mice exposed to 125 and 250 ppm EGBE may have been
due to carcinogenic effects in the liver. A high rate of hepatocellular carcinomas was found in
these exposure groups (10/50 [control], 11/50, 16/50, 21/50); the increase at the high-exposure
level was statistically significant (p < 0.01). However, when hepatocellular adenomas and
carcinomas were combined, no significant increase was observed in any exposure group. The
incidence of hemangiosarcomas in males exposed to 250 ppm (8%) was also significantly
increased (p = 0.046) relative to chamber controls (0/50, 1/50, 2/49, 4/49) and exceeded the
range of historical controls (14/968; 1.5 ± 1.5%; range 0-4%). No significant increases in
benign or malignant hepatocellular tumors or hemangiosarcomas were noted in the female mice,
and the incidence of hepatocellular adenomas actually decreased significantly (p < 0.05) in
relation to the control chamber group (16/50, 8/50, 7/49, 8/49). It should be noted that in light of
the high survival rate of the exposed female mice relative to controls (29/50, 31/50, 33/50,
36/50), the high exposure of 250 ppm may not have provided the maximum tolerated dose.
Forestomach squamous cell papillomas and carcinomas, combined, were significantly
increased (trend test = 0.003) in female mice relative to the chamber control group (0/50, 1/50,
2/50, 6/50). The incidence of these tumor types (12%) at the highest exposure level was also
statistically significant and exceeded the range for the occurrence of these tumors in historical
controls (0.9 ± 1.1%; range 0-3%). The first incidence of these tumors appeared in the group
exposed to 250 ppm at 582 days, as compared to 731 days at 62.5 and 125 ppm, indicating a
decreased latency period in the highest exposure group. While the incidence of these types of
forestomach tumors was not significantly increased over controls in male mice (1/50, 1/50, 2/50,
2/50), the incidence of squamous cell papillomas (4%) in the two highest exposure groups
exceeded the range for historical controls (0.5 ± 0.9%; range 0-2%). The increased incidence of
57

-------
forestomach neoplasms in males, as in females, occurred in groups with ulceration and
hyperplasia.
The NTP (2000, 196293) study concluded that there was no evidence showing
carcinogenic activity in male F344/N rats, and equivocal evidence of carcinogenic activity in
female F344/N rats based on increased combined incidences of benign (mainly) and malignant
pheochromocytoma of the adrenal medulla. The researchers reported some evidence of
carcinogenic activity in male B6C3Fi mice based on increased incidences of hemangiosarcoma
of the liver and an increase in the incidence of hepatocellular carcinoma, as well as some
evidence of carcinogenic activity in female B6C3Fi mice based on increased incidence of
forestomach squamous cell papilloma (mainly) or carcinoma.
With respect to the pheochromocytomas reported in female rats, while the data showed a
positive trend (p = 0.044) and the high-dose tumor frequencies (16%) were above the upper
range of historical controls (13%), the tumor incidence data were not statistically significant.
Further, the NTP (2000, 196293) report noted that pheochromocytomas can be difficult to
distinguish from nonneoplastic adrenal medullary hyperplasia. The presence of mild-to-
moderate compression of the adjacent tissue is a primary criterion used to distinguish
pheochromocytomas from medullary hyperplasia; most tumors observed were small and not
substantially larger than the more severe grades of adrenal medullary hyperplasia. Interpretation
of these tumors should be done cautiously. Given the marginal dose response, lack of tumor
evidence in any other organ system of the rats, and reported difficulties in distinguishing
pheochromocytomas from nonneoplastic adrenal medullary hyperplasia, this tumor type was not
given significant weight in the qualitative or quantitative assessment of EGBE cancer potential.
4.6.3. Mode-of-Action Information
4.6.3.1. Hypothesized MOA for Liver Tumor Development in Male Mice
The hypothesized MOA for EGBE-induced liver tumors in male mice is believed to
involve iron accumulation and subsequent oxidative stress due to the hemolytic effects of a
metabolite of EGBE. Male mice developed hepatocellular carcinomas and hemangiosarcomas in
the liver that appeared to be exposure-related. The incidence of hemangiosarcomas was
statistically significant and increased over both concurrent and historical control groups. The
hepatocellular carcinomas were within the range of historical controls for male mice, but are also
considered in this discussion because the dose-response trend is significant and because a similar
MOA has been suggested for this tumor (Klaunig and Kamendulis, 2005, 100165). The
incidences in the high dose group of these two types of tumors were only slightly higher than the
upper end of the range for historical controls. Furthermore, these two tumor types were not
increased in other organs (e.g., bone, bone marrow) and were not noted in either rats or female
mice. The hypothesized MOA involves the hemolysis of RBCs, the accumulation of
hemosiderin, and subsequent oxidative stress that leads to neoplasia in the two cell types
58

-------
believed to be the origin for these two tumors: hepatocytes for hepatocellular carcinoma and
endothelial cells for hemangiosarcomas. Only the male mice developed these tumor types, and
this is hypothesized to be due to their susceptibility to oxidative stress. Mice are known to be
more susceptible to oxidative stress than rats because of their lower antioxidant capacity
(Bachowski et al., 1997, 042161; Siesky et al., 2002, 042204). The available studies suggest that
iron accumulation from the hemolytic effects of EGBE produces liver oxidative damage that is
more severe in mice and increased DNA synthesis in both endothelial cells and hepatocytes that
may be unique to mice (Corthals et al., 2006, 100102; Siesky et al., 2002, 042204). From this
and reported differences in antioxidant capacity2 and background rates of these tumors3 between
male and female mice, it can be hypothesized that events leading to oxidative stress could
contribute to the development of hemangiosarcomas and hepatocellular carcinomas in male
mice. A series of events that may lead to the formation of liver tumors follows:
(1)	EGBE is metabolized to BAL, which is subsequently oxidized to BAA;
(2)	BAA causes RBC swelling, triggering sequestration in the spleen by resident
macrophages. When the capacity of these macrophages becomes overwhelmed, the
damaged RBCs make their way into the liver;
(3)	Excess Hb from damaged RBCs is taken up by phagocytic (Kupffer) cells of the liver and
stored as hemosiderin;
(4)	Oxidative damage and increased synthesis of endothelial and hepatocyte DNA are
initiated by one or more of the following events:
(a)	Generation of reactive oxygen species (ROS) from Hb-derived iron within
Kupffer cells and perhaps from within hepatocytes and sinusoidal endothelial cells;
and
(b)	Activation of Kupffer cells to produce cytokines/growth factors that suppress
apoptosis and promote cell proliferation.
(5)	ROS results in oxidative DNA damage to hepatocytes and endothelial cells;
(6)	ROS modulates hepatocyte and endothelial cell gene expression;
(7)	ROS stimulates hepatocyte and endothelial cell proliferation;
(8)	ROS promotes initiation of hepatocyte and endothelial cells; and
(9)	ROS promotes neoplasm formation.
2While the reason for the sex difference in liver tumor susceptibility between male and female mice is not clear, it has been shown that
estrogens can be protective through their antioxidant capacities and through their modulation of the activities of other antioxidants (Nyska
et al., 2004, 056379V
3NTP has observed liver hemangiosarcomas in 105/4183 (2.51%) male versus just 35/4177 (0.84%) female historical controls (Klaunig
and Kamendulis, 2005, 100165: NTP, 2000, 196293). In addition, other chemicals reported by NTP to cause both early onset hemosiderin
buildup and liver tumors have also exhibited this male specificity (U.S. EPA, 2005, 594432).
59

-------
The first two steps, the metabolism of EGBE to BAA and the association of BAA with
hemolytic effects, have been established in many studies, both in vitro and in vivo, with
sensitivities ranging from extreme sensitivity as in rats, mice, rabbits and dogs, to moderate to
extreme insensitivity seen in monkeys, guinea pigs, and humans (see Section 3). The third step
has been seen in both genders of rats and mice exposed to EGBE in multiple studies that
observed hemosiderin within Kupffer cells and hepatocytes after RBC breakdown (Ghanayem
and Sullivan, 1993, 041609; Ghanayem et al., 1987, 041645; Ghanayem et al., 1987, 041608;
Kamendulis et al., 1999, 042182; Krasavage, 1986, 066474; NTP, 2000, 196293; Siesky et al.,
2002,	042204). A number of studies (Kamendulis et al., 1999, 042182; Park et al., 2002,
030770; Siesky et al., 2002, 042204) provide support for step 4a by showing that in vivo
exposure to EGBE increases 8-OHdG levels (an indicator of oxidative damage) in mice, but not
rats, and decreases vitamin E levels in rats and mice. Using rat and mouse hepatocytes, Park et
al. (2002, 030770) showed that FeSC>4 produced dose-related changes in these same indicators in
mouse hepatocytes, but not in rat hepatocytes, and that treatment with EGBE or BAA did not
produce changes in these oxidative stress parameters. Additionally, Nyska et al. (2004, 056379)
analyzed 130 2-year carcinogenicity studies of B6C3Fi mice from NTP bioassays and concluded
that a significantly increased risk of inducing hepatic hemangiosarcomas in male B6C3Fi mice
exists in studies with compounds that caused increased tissue burdens of ROS. Klaunig and
Kamendulis (2005, 100165) and Corthals et al. (2006, 100102) provided support for step 4b, by
showing that the activation of Kupffer cells, either through RBC hemolytic components and/or
iron accumulation in the Kupffer cells, results in the production of cytokines such as TNFa. The
comet assay has been used to assess DNA damage to endothelial cells from ROS (step 5)
(Klaunig and Kamendulis, 2004, 594442; Klaunig and Kamendulis, 2005, 100165; Reed et al.,
2003,	594436). While step 6 has not been shown directly for endothelial cells or hepatocytes
exposed to EGBE, induction of oxidative damage has been shown to modify gene expression in
mammalian cells. In addition, ROS production can stimulate cell proliferation and the inhibition
of apoptosis (Klaunig and Kamendulis, 2005, 100165; Nyska et al., 2004, 056379). Siesky et al.
(2002, 042204) observed increased DNA synthesis in endothelial cells and hepatocytes in vivo in
mice, but not in rats, at doses that produced hemangiosarcomas in the mouse liver (NTP, 2000,
196293) (step 7). Steps 8 and 9 are consistent with the lack of direct genotoxicity of EGBE (see
Section 4.4.4) and the high rate of spontaneous endothelial neoplasms in the male mouse liver
relative to the rat (Klaunig and Kamendulis, 2005, 100165). The observation of decreased
antioxidant capacity and increased 8-OHdG levels in male mice also lends support to the
proposed steps of initiation and promotion of neoplasms by ROS (Klaunig et al., 1998, 024780).
4.6.3.1.1. Temporal association and species specificity. Key steps in the proposed MOA (i.e.,
hemolysis, hepatic hemosiderin buildup, and oxidative damage) have all been observed in
60

-------
subchronic or shorter-duration rat and mouse studies of EGBE (Kamendulis et al., 1999, 042182;
NTP, 2000, 196293; Siesky et al., 2002, 042204) well in advance of tumor formation.
Differences between rats and mice with respect to these responses may help to explain the
relative sensitivity of male mice to the formation of liver tumors following EGBE exposure. In
mice, Siesky et al. (2002, 042204) observed a dose-dependent increase in levels of liver
oxidative stress indicators at exposure days 7 and 90, increased endothelial cell DNA synthesis at
exposure days 7 and 14, and increased hepatocyte DNA synthesis at 90 days. No increase in
ROS or the DNA synthesis of either cell type was observed in rats at any time point.
4.6.3.1.2. Dose-response relationships. Five chemicals have been determined by the NTP to
cause hemosiderin buildup in the livers of mice. As shown in Table 4-8, all three studies that
reported hemosiderin buildup in Kupffer cells of male mice within 13 weeks of exposure also
showed an increased incidence of hemangiosarcomas and hepatocellular carcinomas following
chronic exposure. The dose responses for endpoints describing possible precursor effects,
splenic hematopoietic cell proliferation, and liver hemosiderin accumulation appear to be dose-
related and coincident to the formation of tumors. Dose-responses for several hemolytic effects
were also observed in rats exposed to EGBE, but liver tumors were not increased in rats at any
dose. However, the high dose used in the EGBE rat study was only half that of the high dose
used in the mouse study, and the 2-year duration of these studies represents a smaller fraction of
the rat lifespan, leaving the possibility that similar responses could have been observed in rats if
higher and longer EGBE exposures had been administered.
61

-------
Table 4-8. Incidence of liver hemangiosarcomas and hepatocellular
carcinomas in studies of NTP chemicals that caused increased hemosiderin
in Kupffer cells in male mice
Chemical (NTP TR#)
Hemosiderin
SCa
Hemangiosarcoma
Hepatocarcinoma
Type
2-Butoxyethanol
(EGBE) (TR-484)
0/50, 0/50, 8/49°,
30/49°
Yes
0/50, 1/50, 2/49, 4/49b
10/50, 11/50, 16/49,
21/49°
I
p-Chloroaniline
hydrochloride (TR-351)
0/50, 0/49, 0/50, 50/50d
Yes
2/50, 2/49, 1/50, 6/50
3/50, 7/49, 11/50b, 17/50°
G
p-Nitroaniline
(TR-418)
1/50, 1/50, 8/50b,
50/50°
Yes
0/50, 1/50, 2/50, 4/50
10/50, 12/50, 13/50, 6/50
G
C.I. Pigment Red 3
(TR-407)d
0/50, 5/50, 30/50, 41/50
No
0/50, 1/50, 1/50, 0/50
5/50, 10/50, 8/50, 4/50
F
o-Nitroanisole
(TR-416)
0/50, 0/50, 3/50, 16/50°
No
2/50, 2/50, 1/50, 0/50
7/50, 12/50, 11/50, 7/50
F
aChemicals that caused hemosiderin accumulation in Kupffer cells following subchronic (SC) exposure are
identified with a "yes" in this column.
Statistically significant difference, p < 0.05.
Statistically significant difference,/) <0.01.
Statistics not reported.
F = feed; G = gavage; I = inhalation
Of the five chemicals listed in Table 4-8, liver hemangiosarcomas were observed with
only three chemicals that induced hemosiderin buildup by week 13, but not with the other two
chemicals for which hemosiderin buildup was not observed until the end of the 2-year study
(Gift, 2005, 100133; U.S. EPA, 2005, 594432). Two of the three chemicals that induced early
liver hemosiderin accumulation and increased hemangiosarcoma incidence, EGBE and p-
chloroaniline hydrochloride, also induced an increase in hepatocellular carcinomas. Early
buildup of hemosiderin combined with early increases in endothelial cell and hepatocyte DNA
synthesis would result in a longer exposure of cells to oxidative damage via iron-generated
radicals (step 4). This would be consistent with a mechanism involving a continuing cycle of
damage and repair and accumulation of DNA mutations (steps 5 and 6). In addition to an earlier
onset of hemosiderin buildup, mice also show evidence of a more sustained hemolytic response
to EGBE than rats.4
4.6.3.1.3. Biological plausibility and coherence of the database. Oxidative damage plays an
important role in the pathogenesis of several diseases, including cancer and cardiovascular
disease (Djordjevic, 2004, 100106; Klaunig et al., 1998, 024780; Lesgards et al., 2002, 042183).
In support of the proposed hypothesis, increased ROS are known to accompany the release of
4Mice experienced an increase in liver and splenic hematopoietic cell proliferation throughout the 2-year NTP (2000, 196293) study, while
rats tended to compensate for the effects of EGBE after a few months. This increased tolerance in rats is evidenced by a lack of induction
of splenic hematopoiesis at the end of the 2-year NTP (2000, 196293) study.
62

-------
large amounts of iron from hemolysis (Ziouzenkova et al., 1999, 024904). If EGBE causes
oxidative stress via hemolysis, then the production of protein and DNA damage would be
expected, including the production of 8-OHdG, accompanied by a decrease in antioxidant levels,
such as Vitamin E (Houglum et al., 1997, 024777; Wang et al., 1995, 024797; Yamaguchi et al.,
1996, 024876). These effects were verified by both Siesky et al. (2002, 042204) and
Kamendulis et al. (1999, 042182). who measured a dose-dependent increase in levels of ROS
indicators, including 8-OHdG and malondialdehyde, and a decrease in vitamin E levels in the
livers of mice, but not in rats, after acute and subchronic exposure to EGBE. The fact that mice
appear more susceptible than rats to ROS from EGBE exposure is consistent with the proposed
MO A, and is a reasonable explanation for why hemangiosarcomas are not observed in rats
following chronic EGBE exposure, despite the fact that rats are at least as sensitive as mice to the
hemolytic effects of EGBE.
Iron, which is known to accumulate in cells of rodent livers following EGBE exposure,
can produce hydroxyl radicals in combination with oxidative by-products via the Fenton reaction
(Kamendulis et al., 1999, 042182). The damaging effects of iron overload to liver sinusoidal
cells have been shown in rats following a single i.p. injection of 200 mg iron/kg (Junge et al.,
2001, 042180). In addition, endothelial cells appear to be relatively sensitive to oxidative stress
(DeLeve, 1998, 100104; Spolarics, 1999, 024792). Liver hemangiosarcomas develop from the
endothelial cell component of the vascular sinusoidal cells of the liver (Frith and Ward, 1980,
024775).
In vivo studies have indicated that pretreatment of rats with an ALDH inhibitor, pyrazole,
prior to a single 125 mg EGBE/kg gavage exposure protected against hemolysis Ghanayem et al.
(1987, 041608). presumably by blocking the production of both BAL and BAA. Pretreatment of
rats with an ADH inhibitor, cyanamide, prior to a single 125 mg EGBE/kg gavage exposure,
reduced hemolytic responses, but increased RBC swelling, increased mortality, decreased BAA
formation and excretion in the urine, and increased the urinary excretion of EGBE conjugates
with glucuronide and sulfate (Ghanayem et al., 1987, 041608). This hematotoxicity in the
presence of cyanamide may be due to BAL, but residual BAA may also be a factor. EGBE +
cyanamide decreased BAA concentrations in rats; however, some BAA was formed and the
BAA ty2 was increased (Ghanayem et al., 1990, 042016). and, when Ghanayem et al. (1987,
041608) administered a gavage dose of 125 mg BAL/kg + cyanamide to rats, the researchers
observed almost no hemolytic activity. Furthermore, gavage administration to rats of 125 mg
EGBE/kg and the molar equivalent of BAL and BAA resulted in no significant difference
between the hemolytic effects of the three chemicals between 2 and 24 hours postexposure
(Ghanayem et al., 1987, 041608). These data suggest that EGBE's hemolytic activity (without
coexposures) is due to BAA and that the metabolism of EGBE and BAL to BAA takes place
rapidly and completely.
63

-------
4.6.3.1.4. Relevance of the hypothesized MOA to humans. The occurrence of liver tumors in
mice exposed to EGBE is hypothesized to occur through an MOA that requires first a dosage of
EGBE that is high enough to cause sustained hemolysis of RBCs and, second, leads to sufficient
buildup of hemosiderin in the Kupffer cells of the liver to produce ROS and subsequent
neoplasm formation.
Several studies have examined the susceptibility of RBCs to BAA-induced hemolysis and
have found a range in sensitivity from the sensitive (rats, mice, rabbits, and dogs) to the less
sensitive (monkeys, guinea pigs, and humans). Human volunteers experienced no hemolysis
from controlled laboratory acute inhalation exposures (up to 195 ppm), a dosage which caused
significant erythrocyte fragility in rats (Carpenter et al., 1956, 066464). Only mild hemolytic
effects have been observed in humans acutely exposed to oral doses of EGBE (400-1,500
mg/kg), doses that have been shown to cause marked hemolytic effects in rats (Ghanayem et al.,
1987, 066470; Grant et al., 1985, 006770). RBCs from populations that were potentially more
sensitive to hemolysis in general—the elderly, and individuals with sickle cell anemia and
hereditary spherocytosis—were tested in vitro and not found to exhibit hemolysis after exposure
to concentrations 40-fold higher (the highest tested in the study) than those shown to induce
hemolysis in rat RBCs (Udden, 1994, 042109; Udden and Patton, 1994, 056374). In an in vitro
study of RBCs from hospitalized children and adults, concentrations of up to 150-fold higher
than those used in rat studies did not produce hemolysis (Udden, 2002, 042 111). The resistance
of human RBCs to the initial event of hemolysis makes it unlikely that they would experience the
subsequent effects of increased hemosiderin deposition through this pathway, and consequently,
humans would not be at increased risk of tumor development through this MOA.
64

-------
4.6.3.1.5. Other possible MOAs for liver tumor development in male mice. Although certain
key events in EGBE's MOA for the development of liver tumors in male mice are fairly well-
described and plausible, some alternative considerations (also supported by scientific literature)
may be involved. ROS can potentially be derived from two sources: iron overloading in the
liver (through Fenton and Haber-Weiss reactions) and Kupffer cell activation. Via either source,
oxygen radicals can induce oxidative damage to DNA and lipids as documented in the liver
following EGBE treatment (Siesky et al., 2002, 042204). The activation of Kupffer cells,
through phagocytosis of RBC hemolytic components or iron in the Kupffer cell, results in the
production of cytokines, possibly including vascular endothelial growth factor that may elicit a
growth response on endothelial cells (Corthals et al., 2006, 100102). In addition to the
production of oxidative DNA damage, ROS, whether derived from Kupffer cell activation or
other biological processes, can alter gene expression (e.g., MAP kinase/AP-1 and NFkB),
resulting in stimulation of cell proliferation and/or inhibition of apoptosis (Klaunig and
Kamendulis, 2004, 594442).
Another recognized mechanism for the development of chemically-induced liver
hemangiosarcomas involves direct interaction with DNA. This MOA is recognized for vinyl
chloride and thorotrast, two agents that are known to induce hemangiosarcomas in humans. The
EGBE metabolite BAL is considered to have the greatest potential to interact with DNA, since it
has been shown to cause in vitro SCE at concentrations ranging from 0.2 to 1 mM (Elliott and
Ashby, 1997, 100 111). However, high ADH activity in the liver, as in the forestomach, is
expected to result in very short residence time and low Cmax liver tissue concentrations of BAL.
Corley et al. (2005, 100098) extended their 1994 model (Appendix B) to include the metabolism
of EGBE to BAL via ALDH and the subsequent metabolism of BAL to BAA via ADH in both
the liver and forestomach. As shown in Figure 4-1, using rate constants derived from mouse
stomach fractions (Green et al., 2002, 041610) and making several assumptions about the use of
these enzyme activity data, Corley et al. (2005, 100100) estimated that 250 ppm EGBE (the
highest concentration used in the NTP (2000, 196293) study) would result in peak Cmax values of
7 [iM EGBE, 0.5 [xM BAL, and 3,250 [xM BAA in liver tissue of male mice at the end of a 6-
hour exposure period.
65

-------
BE
BAL
BAA
Cmax (uM)
7
0.5
3,250
BE & Metabolites in Liver
50 i
3,500
3,000
2,500
2 2.000
o 1,500
c
0
o
Time (h)
BE = EGBE
Source: Adapted from Corley et al. (2005, 100098).
1
-BE
• BAL
-BAA
6	12
Time (h)
18
Figure 4-1. Simulated concentrations of EGBE, BAL, and BAA in liver
tissues of female mice exposed via inhalation for 6 hours to 250 ppm EGBE.
Thus, the Corley et al. (2005, 100098) PBPK model suggests that the high cytotoxic
concentrations of BAL that showed some evidence of clastogenicity may not be relevant to the
target organ where lower concentrations of BAL would exist in the presence of metabolizing
enzymes. A recent gavage study performed by Deisinger and Boatman (2004, 594446) provided
support for the Corley et al. (2005, 100098) model and the predicted low levels of the BAL
metabolite in liver tissue.5 In addition, as discussed in Section 4.4.4, evidence from in vivo and
in vitro genotoxicity assays does not suggest that BAL would have any significant genotoxicity
in vivo.6 Furthermore, the MO A for hemangiosarcoma induction by genotoxins such as vinyl
chloride and thorotrast involves the initiation of hepatocellular and sinusoidal cell hyperplasia
and sinusoidal compression, leading to the development of fibrous septa, generally in the
periportal area, out of which eventually develop multiple areas of angiosarcomas (Foster, 2000,
042172). EGBE exposure does not generate this same pattern of effects prior to the development
of cancer in mice.
5The Corley et al. (2005, 100098) model predicts that the concentrations of BAL in liver tissues of male and female mice would be 17 and
29 jaM, respectively, following oral gavage exposure to 600 mg/kg EGBE. The levels of BAL actually observed in the liver tissue of male
and female mice, following oral gavage exposure to EGBE at 600 mg/kg, 3 and 4 jaM, respectively, were even lower than the predicted
values (Deisinger and Boatman, 2004, 594446).
6The exposure concentrations that would be necessary to cause these effects in humans, if attainable at all, are likely to be much higher
than the RfC and RfD.
66

-------
4.6.3.2. Hypothesized MO A for Forestomach Tumor Development in Female Mice
A significant increase over controls (experimental and historical) of papillomas and one
carcinoma of the forestomach (6/50; 12%) was reported by NTP (2000, 196293) in female mice
exposed for 2 years to 250 ppm EGBE by inhalation. An increase was also seen in squamous
cell papillomas in male mice, although this did not reach the level of statistical significance.
Significant increases in forestomach papillomas and carcinomas were not observed in rats of
either gender. The study also showed statistically significant, dose-dependent increases in
hyperplasia for both male and female mice, and for ulceration in female mice. Male mice
showed significant increases in ulceration at the 125 ppm exposure. The process of irritation,
hyperplasia, and ulceration is thought to be a part of the cell injury and regeneration process
elicited by irritant chemicals such as EGBE, and the study authors hypothesized that the
neoplasia may occur due to an increase in the cell replication associated with regeneration.
A hypothesized series of events that may lead to the formation of forestomach tumors is
as follows:
(1)	Deposition of EGBE/BAA in the stomach and forestomach via consumption or
reingestion of EGBE laden mucus, salivary excretions, and fur material;
(2)	Retention of EGBE/BAA in food particles of the forestomach long after being cleared
from other organs;
(3)	Metabolism of EGBE to BAL, which is rapidly metabolized to BAA systemically and in
the forestomach;
(4)	Irritation of target cells by BAA leading to hyperplasia and ulceration;
(5)	Continued injury by BAA and degeneration leading to high cell proliferation and
turnover; and
(6)	High levels of cell proliferation and turnover leading to clonal growth of spontaneously
initiated forestomach cells.
There are a number of studies that have demonstrated steps 1 and 2 (i.e., the deposition
and retention of EGBE and BAA in the forestomach). Studies have shown this occurs following
whole-body exposure (Green et al., 2002, 041610; Poet et al., 2002, 594439), nose-only
inhalation (Poet et al., 2002, 594439). i.v. exposure (Bennette, 2001, 041810; Green et al., 2002,
041610; Poet et al., 2002, 594439). i.p. exposure (Corley et al., 1999, 042003; Poet et al., 2002,
594439). s.c. exposure (Corley et al., 1999, 042003). and gavage exposures (Ghanayem et al.,
1987, 041645; Ghanayem et al., 1987, 041608; Green et al., 2002, 041610; Poet et al., 2002,
594439). It is of note that following i.v. and inhalation exposures in mice, EGBE metabolites
rapidly accumulate in salivary secretions and are swallowed (Bennette, 2001, 041810; Green et
al., 2002, 041610). leading to the collection and retention of the chemical(s) in the forestomach.
67

-------
The same process is likely to occur through other systemic exposures. The metabolism of EGBE
to BAA (step 3) has been shown in both in vitro and in vivo tests with rats, mice, rabbits, guinea
pigs, dogs, monkeys, and humans, (see Section 3) and is further supported by the EGBE PBPK
model developed by Corley et al. (2005, 100098). Step 4, the irritation of target cells, has been
seen in both genders of B6C3Fi mice, (Green et al., 2002, 041610; NTP, 2000, 196293; Poet et
al., 2002, 594439), with irritation and a compensatory proliferative response reported following
exposure to EGBE. Female mice were shown to have more extensive and severe forestomach
lesions than male mice and were observed in the NTP (2000, 196293) study to have statistically
significant increases in forestomach tumors. This suggests the importance of the continued
damage and high cell proliferation (step 5) that is associated with tumor formation. Green et al.
(2002, 041610) found that the number of cells in S-phase (an indication of cell turnover)
increased in a dose-dependent fashion after exposure to EGBE and BAA, even though none of
the changes were statistically significant due to the high turnover for the control group. Step 6,
high levels of cell proliferation and turnover, leads to clonal growth of spontaneously initiated
cells and is supported by the continuum of effects observed in the mice (Green et al., 2002,
041610; NTP, 2000, 196293) and the effects seen with other irritant compounds (Kroes and
Wester, 1986, 041626).
Green et al. (2002, 041610) also provided relevant information regarding step 3 through
examination of the activity and localization of ALDH and ADH in the stomach tissues of mice,
rats, and a human sample. Whole body autoradiography of mice that had been exposed to
radiolabeled EGBE was also performed. Histochemical staining of stomach tissues from the
rodent species showed the dehydrogenase enzymes to be heavily concentrated in the stratified
squamous epithelium of the forestomach of both rats and mice, whereas their distribution in the
glandular stomach was more diffuse. By comparison, histochemical analysis of a human
stomach tissue sample showed both enzymes to be present and evenly distributed throughout the
epithelial cells of the entire stomach mucosa. A marked species difference in ALDH activity in
the forestomach was observed between rats (Km = 0.29 mM; Vmax = 1.627 nmol/minute per mg
protein) and mice (Km = 46.59 mM; Vmax = 17.094 nmol/minute per mg protein) with Km values
up to one order of magnitude greater in mice compared to rats. These differences indicate that
mice forestomach tissues would have the capacity to metabolize appreciably larger amounts of
EGBE to BAL, and subsequently to BAA, than would the rat forestomach. Whole body
autoradiography of mice exposed to EGBE demonstrated selective accumulation in the
forestomach, which would provide substrate for these enzymes. Collectively, these data
demonstrate several points regarding the observed forestomach toxicity in mice following
exposure to EGBE, including: (1) the accumulation of EGBE in the target tissue, the
forestomach, of mice; (2) a high degree of localization in the forestomach (as compared to the
glandular stomach) tissues of both rats and mice of the enzymes that metabolize EGBE to the
corresponding carboxylic acid; and (3) kinetic differences in these enzymes consistent with mice
68

-------
being capable of metabolizing more EGBE to BAA than rats. The data also suggest that human
stomach tissues do not have a high localization of the EGBE metabolizing enzymes; the diffuse
distribution of these enzymes in the human stomach sample examined is more similar to the
distribution seen in the glandular portions of the rodent species examined. These observations
suggest that human stomach tissues would be less capable of accumulating and localizing BAA
than rat tissues and, thus, would be less likely to be exposed to the irritating effects of BAA.
4.6.3.2.1.	Temporal association. All of the steps in the proposed MOA have been observed to
occur in female mice prior to tumor formation. NTP (2000, 196293) reported that female mice
experienced epithelial hyperplasia (1/10, 5/10, 9/10, and 10/10) after just 13 weeks of exposure
at the same exposure levels used in the chronic study, 0, 62.5, 125, and 250 ppm. The reported
incidence of a forestomach papilloma or carcinoma in female mice was 731, 731, and 582 days
in the 62.5, 125, and 250 ppm exposure groups, respectively. This is consistent with the findings
of Ghanayem et al. (1986, 011728; 1993, 597328; 1994, 042760). who investigated the temporal
relationship between the induction of this type of forestomach lesion by another nonmutagenic
irritant, ethyl acrylate (EA), and the development of squamous cell papillomas and carcinomas.
These researchers observed cell proliferation/hyperplasia in the forestomach of all rats that
received EA by gavage (200 mg/kg, 5 days/week) for 6 or 12 months. All these potentially
precancerous forestomach lesions regressed in animals treated with EA for 6 months and allowed
2 or 15 months of recovery, and no forestomach neoplasms were observed. Although EA, an
unsaturated aldehyde, is not a metabolite of EGBE, it is an analog of BAL and a much more
potent carcinogen than either EGBE or BAL (Gold et al., 1993, 056382).
For EGBE, the high incidence of forestomach hyperplasia, the relatively lower incidence
of papillomas, and the late occurrence of a single carcinoma in the high 250 ppm exposure group
are suggestive of a temporal relationship and tumor progression following EGBE exposure to
female mice. Male mice may show the beginnings of tumorigenic effects as the incidence of
papillomas increases, but such findings have not been statistically significant compared to
concurrent or historical controls. No hyperplasia and no tumors were observed in inhalation
studies of rats (NTP, 2000, 196293) or in drinking water studies of mice (NTP, 1993, 042063).
supporting the need for these steps prior to tumor formation.
4.6.3.2.2.	Dose-response relationships. The incidences of epithelial hyperplasia (6/50, 27/50,
42/49, 44/50) and ulceration (1/50, 7/50, 13/49, 22/50) in EGBE-exposed female mice were
dose-related and significantly increased over both concurrent and historical controls at lower
dose levels than the forestomach tumors. The hyperplasia was often associated with ulceration,
particularly in the female mice. Forestomach tumors observed by NTP (2000, 196293)
(incidence 0/50, 1/50, 2/50, 6/50) increased over control animals only at exposure levels above
those that caused significant hyperplasia. The increased incidence of the forestomach neoplasms
69

-------
occurred in groups with ulceration and hyperplasia, suggesting a dose-dependent relationship
between the nonneoplastic and the neoplastic lesions.
4.6.3.2.3. Biological plausibility and coherence of the database. Both mutagenic and
nonmutagenic chemicals have been shown to induce forestomach tumors in rodents (Ghanayem,
1986, 011728; Ghanayem et al., 1993, 597328; 1994, 042760; Kroes and Wester, 1986, 041626;
NTP, 2000, 196293). Some nonmutagenic substances that cause such tumors appear to require
long-term contact with the forestomach epithelium, leading to irritation, cell proliferation, and
neoplasia. The overstimulation of repair processes and enhancement of growth-promoting
factors are believed to be involved (Harrison et al., 1991, 041622). Promotion and other
activities associated with the stimulation of cell proliferation have been observed for many of
these compounds (Clayson et al., 1991, 006215; Ghanayem et al., 1994, 042760). High
concentrations of EGBE and its BAA metabolite sequestered in the forestomach are assumed to
cause chronic irritation and the more serious damage observed in the forestomach lining of
female mice. The incidence of ulcers was significantly increased in all exposed groups of
females. NTP (2000, 196293) suggests that EGBE exposure-induced irritation caused
inflammatory and hyperplastic effects in the forestomach and that the neoplasias (papillomas and
one carcinoma) were associated with a continuation of the injury/degeneration process.
Other substances that induce forestomach hyperplasia in male and female mice following
inhalation exposure include acetonitrile, 1,3-butadiene, and chloroprene (U.S. EPA, 2005,
594432). Both propionic and butyric acid have been shown to induce proliferative responses in
forestomach epithelium after only 7 days, and long-term propionic acid exposure has produced
papillomas in the rat forestomach (Kroes and Wester, 1986, 041626). Since high levels of EGBE
and BAA have been observed in the stomachs of mice following i.v., i.p., gavage, and inhalation
exposures, it is apparent that the chemical partitions to the forestomach via multiple routes,
including grooming of fur, systemic blood circulation, ingestion of salivary excretions and
respiratory tract mucus, and possibly repartitioning of the stomach contents (Green et al., 2002,
041610; Poet et al., 2003, 100190). Because the forestomach functions as a storage organ, there
is a reduced requirement for vascularization. The planar capillary network within the epithelial
layers of the rodent forestomach contrasts strongly with the thick mucosal network of capillaries
in the glandular stomach of rodents (Browning et al., 1983, 041615). The cells of the
forestomach epithelium, especially the more superficial squamous cells, are separated from
capillaries by substantial diffusion distances (Browning et al., 1983, 041615; Bueld and Netter,
1993, 041617). In addition, the glandular stomach contains a complex mucosal protection and
buffering system necessary to withstand the high acidity of the digestion process. As a result,
irritant substances that concentrate in the forestomach may produce hyperplasia in the
forestomach, but not in the glandular stomach or other gastrointestinal tissue (Kroes and Wester,
1986, 041626).
70

-------
4.6.3.2.4.	Relevance of the hypothesized MOA to humans. While this proposed MOA is
thought to be of qualitative relevance to humans, the EGBE exposure concentrations that would
be necessary to cause hyperplastic effects and tumors in humans, if attainable at all, are likely to
be much higher than the concentrations necessary to cause forestomach effects in mice for
several reasons:
(1)	The mouse forestomach serves a food storage function and the slow rate of emptying
provides a sink for EGBE where it is metabolized to BAA and remains in contact with
squamous epithelium long after EGBE has been cleared from the rest of the body. While
the human esophagus is histologically similar to the murine forestomach, the contact time
with food and other ingested substances is short, because this organ does not have a
storage function. Thus, the risk for esophageal tumors is low (Kroes and Wester, 1986,
041626). The human stomach also has a faster rate of emptying than the rodent
forestomach, and is further protected from irritant compounds by a mucous layer that is
not present in the rodent forestomach;
(2)	The localization of the enzymes needed for acid production in the human stomach tissue
is not the same as it is in the target, the rodent forestomach (Green et al., 2002, 041610);
(3)	A benchmark dose (BMD) analysis (see Appendix C) indicates that the exposure
concentrations necessary to cause hyperplastic effects in humans would be much higher
than the existing RfD and RfC for EGBE.
4.6.3.2.5.	Other possible MO As for forestomach tumor development in female mice.
Though the evidence favors the hypothesis that BAA is the principal toxic metabolite of EGBE,
roles for BAL (Ghanayem et al., 1987, 041608) and butyric acid (Harrison et al., 1991) have
been suggested. It is not likely that butyric acid plays a significant role in the toxicity of EGBE,
particularly at environmentally relevant concentrations. High concentrations of butyric acid have
caused ulceration and other preneoplastic lesions in mice (Harrison, 1992, 041621). However,
low concentrations of butyric acid do not appear to be harmful, since it naturally occurs in the
diet through the fermentation of fiber and starch and as a significant portion (up to 10 mol%) of
total bovine milk fatty acid (Smith and German, 1995, 042103).
Another possible alternative MOA could exist if EGBE or one of its metabolites were to
have the capability of damaging a cell through direct interaction with its DNA. As has been
discussed in Section 4.4.4, there is very limited evidence that EGBE or BAA is genotoxic. BAL,
a short-lived metabolite of EGBE, has been found to be clastogenic in in vitro assays without
enzyme activation at concentrations ranging from 0.2 to 1 mM (Elliott and Ashby, 1997,
100111). However, as has been discussed, in vivo and in vitro genotoxicity assays do not
suggest that BAL would have any significant genotoxicity in vivo. In addition, chemicals for
which mutagenesis/genotoxic effects play a significant role generally induce more tumors at
71

-------
earlier time points than near the end of the conducted bioassays due to their ability to both
initiate and promote tumor pathogenesis. The mutagenic compound ethylene dibromide, for
instance, was reported to induce forestomach tumors in all dose groups 168-280 days from the
start of exposure (U.S. EPA, 2004, 594429). EGBE is consistent with other forestomach
carcinogens that are not mutagenic, such as EA, in that observed tumors generally did not
progress to carcinoma and were not observed until well into the study (i.e., after long periods of
forestomach cell damage and repair). The first reported incidence of forestomach papilloma or
carcinoma in female mice was 731, 731, and 582 days in the 62.5, 125, and 250 ppm EGBE
exposure groups, respectively (NTP, 2000, 196293).
It does not appear that EGBE, BAL, or BAA preferentially binds to stomach tissue
macromolecules (Green et al., 2002, 041610; Poet et al., 2003, 100190). Poet et al. (2003,
100190) found that high levels of EGBE concentrate in the food content of the forestomach
following i.p. exposure (Poet et al., 2003, 100190). indicating that the observed sequestering of
EGBE in the forestomach is related to its retention in the food that remains there, not to
preferential binding to proteins within forestomach tissue.
4.6.3.3. Conclusions about the HypothesizedMOAs
Inhalation exposure of B6C3Fi mice to EGBE gave rise to tumors in the liver and
forestomach. The liver tumors, hemangiosarcomas, and hepatocellular carcinomas occurred in
males only and were significantly elevated over controls with a positive trend test for the
hemangiosarcomas. The forestomach tumors occurred in females only, had a positive trend test
result, and were significantly increased over controls only at the highest dose.
The MOAs that have been developed for these tumors reflect the evidence for the
nonmutagenic nature of EGBE and its metabolites. For the liver tumors, the hypothesized key
steps of the MOA are metabolism of EGBE to BAA, hemolysis of RBCs with release of Hb and
hepatic hemosiderin accumulation, followed by oxidative stress, modulation of gene expression,
cell proliferation, promotion, and neoplasm, leading to the formation of liver tumors. For the
forestomach tumors, the hypothesized steps are metabolism to BAA, followed by tissue irritation
and subsequent cytotoxicity, compensatory proliferation, and the induction of forestomach
tumors. No other viable MOAs have been identified that adequately explain the existing
laboratory animal and human observations.
Both of these MOAs have some degree of qualitative significance for humans since the
principal biological components supporting them are all present and the processes can occur in
humans. Collectively, however, the evidence presented in this assessment for these MOAs
suggests that both MOAs have only limited quantitative significance to humans, principally due
to kinetic/dynamic differences from the rodents. In the case of the liver tumors, in vitro data
suggest there is a 40- to 150-fold difference in the dose that produces hemolytic changes in the
RBCs of humans as compared to rodents. This difference is supported by the Carpenter et al.
72

-------
(1956, 066464) study in which no changes in erythrocyte fragility were measured in humans at
the highest tested concentration, 195 ppm, but increased erythrocyte fragility was measured in
coexposed rats. Further, PBPK model simulations performed by Corley et al. (2005, 100100)
predict that given the vapor pressure of EGBE, the maximum blood level of BAA that can be
obtained from inhalation exposure would be lower than the predicted concentrations from bolus
exposures that have not resulted in hemolytic effects, and lower than concentrations that have
been shown to produce an effect on human RBCs in vitro (Udden, 2002, 042 111). In the case of
the forestomach tumors, the primary difference between mice and humans is in the degree of
kinetics in the metabolizing enzymes involved in the production and clearance of BAA. Thus,
the hypothesized key events in the MOAs for the animal tumors (liver and forestomach) are not
likely to occur in humans, especially at low doses.
Based on the preceding analysis, EGBE is deemed not likely to be carcinogenic to
humans at the calculated RfC and RfD values presented in this document when examining it on
its physical-chemical properties, toxicokinetic and dynamic factors, and MOA information.
4.7. SUSCEPTIBLE POPULATIONS
The hemolytic effect of EGBE is presumed to be caused by its primary metabolite, BAA,
interacting with the RBC membrane. Potentially susceptible subpopulations would include
individuals with enhanced metabolism or decreased excretion of BAA. As discussed in
Section 4.7.1, older rats have reduced ability to metabolize the toxic metabolite BAA to CO2 and
a diminished ability to excrete BAA in the urine (Ghanayem et al., 1987, 066470; Ghanayem et
al., 1990, 042016). However, the relevance of this finding to the possible susceptibility of
elderly humans is uncertain; as discussed in Section 3, humans may have conjugation pathways
for the excretion of BAA, such as BAA-glutamine and BAA-glycine, which are not present in
rats.
It would also be expected that individuals whose RBC membranes are more susceptible
to the lysis caused by BAA would be more sensitive to effects from EGBE exposure. However,
RBCs from normal, aged, sickle-cell anemia, and hereditary spherocytosis patients were no more
sensitive to the hemolytic effects of BAA than RBCs from healthy volunteers when tested in
vitro (Udden, 1994, 042109). As work in this area continues, further information on the
metabolic or structural differences that result in the lower sensitivity of human RBCs compared
to rat RBCs may eventually identify characteristics in humans that may indicate increased
susceptibility. For instance, it is unknown if a genetic predisposition to hemolytic anemia from
other causes, such as G6PD deficiency, would lead to increased susceptibility to EGBE-induced
hemolysis. G6PD deficiency appears in approximately 400 variants, thus describing a
genetically heterogeneous disorder. The label has been applied to all types of moderate to severe
enzyme deficiency with intermittent, induced hemolytic episodes with chronic hemolytic anemia.
To date, complete deficiency of this enzyme has not been identified. The clinical result is the
73

-------
reduced ability to produce nicotinamide adenine dinucleotide phosphate, an enzyme required for
reactions of various biosynthetic pathways, as well as for the stability of catalase and the
maintenance of GSH levels. Catalase and glutathione peroxidase are the primary enzymes in the
detoxification of hydrogen peroxide. Thus, cells are dependent on G6PD for this pathway;
without it, they are vulnerable to oxidative damage. RBCs are sensitive to this loss of enzyme:
they rely on this system for their antioxidant defenses. Other human risk factors for anemia
include ingestion of certain therapeutic drugs, infections, family history, diet, and systemic
illnesses (Berliner et al., 1995, 001232).
Individuals with hereditary hemochromatosis (HH) represent a population potentially
susceptible to increased release of iron from any source. It is reported that 5/1,000 persons of
northern European descent are homozygous for the gene or genes that cause hemochromatosis,
although it is unknown what proportion of this population will go on to develop the HH
phenotype (Pietrangelo, 2004, 100189). Individuals with this disorder are not able to reduce
their absorption of iron in response to increasing iron levels in the body. Iron stores in the body
continue to increase. The iron stores normally start out as ferritin, then eventually become
aggregates of a breakdown product of ferritin called hemosiderin. HH is a condition
characterized by excessive iron deposition in the form of hemosiderin, found in the liver, heart,
skin, joints, pancreas, and other endocrine organs. It is unknown whether individuals with this
condition would be susceptible to the effects of EGBE exposure. There is no indication in the
literature, however, that RBCs in individuals with HH are more fragile, and it is therefore
unlikely that HH would increase the risk of hemolysis or additional hemosiderin deposition from
EGBE exposure. Studies have shown differences in the localization of iron between HH patients
and rodents. In HH patients, iron appears to accumulate preferentially in the parenchymal cells
of the liver, early on as ferritin and later as hemosiderin, due to increased iron absorption from
the duodenum; late in the disease, iron storage is seen in Kupffer cells and reticular endothelial
cells of the bone marrow (Knutson and Wessling-Resnick, 2003, 094393; Valberg et al., 1975,
100217). In contrast, EGBE-induced toxicity in mice and rats results in initial and preferential
accumulation of hemosiderin in Kupffer cells by phagocytosing senescent RBCs (NTP, 2000,
196293). While it is clear that macrophages and other cells can in fact contain hemosiderin, the
relative level compared to hepatocytes is much less; staining in these cells is typically seen in
late stages of the disease (Kwittken and Tartow, 1966, 597334). The human course of
developing hepatocellular carcinomas as a consequence of HH is also quite different compared
to the development in the mouse model of hemangiosarcoma. Human cases of HH-induced
hepatocellular carcinoma are typically observed in the presence of cirrhosis of the liver, another
long-term process that reflects the chronic nature of the disease progression in humans (Harrison
and Bacon, 2005, 100144). Hemangiosarcomas, the tumor type of concern in the male mice,
have not been associated with HH in the literature.
74

-------
4.7.1. Possible Childhood Susceptibility
A number of factors may differentially affect children's responses to toxicants. The only
human toxicity information available on the toxicity of EGBE to children is from the case study
by Dean and Krenzelok (1991, 597279). They observed 24 children, aged 7 months to 9 years,
after oral ingestion of at least 5 mL of glass window cleaner containing EGBE in the 0.5-9.9%
range (potentially 25-1,500 mg EGBE exposures). Two children who had taken >15 mL did
well after gastric emptying or lavage and observation in the hospital. The remainder were
watched at home after receiving diluting oral fluids. No symptoms of EGBE poisoning or
hemolysis were observed. While the effects reported in adult poisonings have been more severe
than those reported in these children, the adults tended to consume larger volumes and different
concentrations of EGBE, making it extremely difficult to assess the correlation of toxic effects
with age sensitivity.
As discussed above, there are numerous risk factors for anemia that might predispose an
individual to, or compound the effects of, EGBE-induced hemolysis. It is generally recognized,
however, that children do not share the same risk factors for anemia as adults for a number of
reasons, including: (1) a higher rate of RBC turnover; (2) lower incidence of neoplastic disease
in childhood as either a direct or indirect cause of anemia (<7,000 of the 1,000,000 new cases of
cancer each year in the U.S. occur in individuals <15 years of age); (3) the fact that iron
deficiency is almost always secondary to nutritional factors in children; (4) the relative rarity of
alcoholism and its related liver disease; (5) a much lower incidence of anemia associated with
thyroid disease; and (6) a rarity of cardiovascular disease other than congenital heart diseases,
with the result that valve replacement, malignant hypertension, and the use of certain drugs are
not usually a factor (Berliner et al., 1995, 001232; Hord and Lukens, 1999, 594445).
Anemia in children is usually associated with an abnormality of the hematopoietic system
(Berliner et al., 1995, 001232; Hord and Lukens, 1999, 594445). Studies of the osmotic fragility
and deformability of RBCs exposed to BAA, the toxic metabolite of EGBE (Udden, 1994,
042109). suggest that certain patients with abnormal hematopoietic systems, such as sickle-cell
anemia or hereditary spherocytosis, are not more sensitive to the hemolytic effects of EGBE than
normal adults. Other studies suggest that the RBCs of children may be pharmacodynamically
less sensitive to hemolysis than those of adults. RBCs of neonates and children up to 6 months
of age differ from normal adult RBCs in that they are larger and have higher levels of Hb F
versus adult Hb A (Lewis, 1970, 594441). Frei et al. (1963, 100118) showed that the larger calf
erythrocytes containing Hb F were osmotically more resistant than smaller, adult erythrocytes
containing Hb A and suggested that, as fetal erythrocytes are replaced by postnatal erythrocytes,
the total population of RBCs becomes more susceptible to lysis.
The effect of age on EGBE-induced hematotoxicity was studied in male F344 rats by
Ghanayem et al. (1987, 066470; 1990, 042016). These studies also demonstrated the time course
for the onset and resolution of the hematological and histopathologic changes accompanying
75

-------
hemolysis. Adult male F344 rats were significantly more sensitive to the hemolytic effects of
EGBE than were young (4-5 week) male rats following administration of a single gavage dose
of EGBE. Concurrent metabolism studies found increased blood retention of EGBE metabolite
BAA (as measured by increased Cmax, AUC, and tu) in young rats and that these rats eliminated a
significantly greater proportion of the administered EGBE dose as exhaled CO2 or as urinary
metabolites. The rats also excreted a greater proportion of the EGBE conjugates, glucuronide
and sulfate, in the urine. These researchers suggested that a reduced ability of older rats to
metabolize the toxic metabolite BAA to CO2 along with a diminished ability to excrete BAA in
the urine may explain the age-dependent toxicity of EGBE.
NTP (2000, 196293) also found that young mice eliminated BAA 10 times faster than
aged mice following a 1-day exposure to EGBE. This difference was not as apparent after 3
weeks of exposure, suggesting that factors other than age may be involved (Dill et al., 1998,
0419811
Available in vitro information suggests that children are no more and are possibly less
sensitive to the hemolytic effects of BAA than adults. Udden (2002, 042 111) compared the in
vitro responses of erythrocytes (percent hemolysis and MCV alterations) obtained from
hospitalized adults (n = 29-40) and hospitalized children (n = 25-46) to 0 or 10 mM BAA for 4
hours. BAA (10 mM) produced comparable significant increases in MCV in both adults (from
87.1 to 89.2 [j.m3; 2%) and children (92.8-95.2 [j,m3; 3%). In the case of hemolysis in response to
BAA, the response was noted as being significantly increased for hospitalized adults (0.54-
0.69%; 28%) but not for hospitalized children (0.68-0.75%; 10%).
Relatively minor developmental effects due to maternal toxicity related to hematologic
effects of EGBE exposure were found in studies using rats, mice, and rabbits dosed orally, by
inhalation, or dermally (Hardin et al., 1984, 042030; Heindel et al., 1990, 042042; NTP, 1993,
042063; Sleet et al., 1989, 042089; Tyl et al., 1984, 100209; Wier et al., 1987, 042123). No
teratogenicity was noted in any studies. It can be concluded that EGBE is not significantly toxic
to developing fetuses of laboratory animals.
4.7.2. Possible Gender Differences
Gender differences have been noted in a number of animal and human studies: females
were more susceptible to effects from EGBE exposure. In the NTP (1993, 042063) 2-week
drinking water studies with EGBE, the absolute and relative thymus weights in female F344 rats
at the highest exposure level (265 mg/kg-day) were slightly reduced. In the 13-week studies,
male rats in the highest three dose groups and females in all dose groups suffered mild (males) to
moderate (females) anemia.
Gender differences have also been noted in some studies that observed the hemotoxic
effects of dermal administration of EGBE. Repeated application of EGBE either neat or as a
dilute aqueous solution (occluded) to male or female NZW rabbits at exposure levels of 18, 90,
76

-------
180, or 360 mg/kg (6 hours/day, nine applications) produced hemoglobinuria in males at
360 mg/kg and in females at 180 or 360 mg/kg (Tyler, 1984, 006769). Only female rabbits
showed decreased RBC counts, Hb concentrations, and MCHC along with increased MCH at the
highest treatment level. Recovery was noted following a 14-day observation period.
A number of secondary effects resulting from the hemolytic toxicity of EGBE, such as
effects on the rat liver, kidneys, spleen, bone marrow, and, to a lesser extent, the thymus, were
more pronounced in females. In the NTP drinking water studies (1993, 042063). liver lesions in
females, but not males, included cytoplasmic alterations, hepatocellular degeneration, and
pigmentation. These effects were most pronounced in the three highest dose groups.
Carpenter et al. (1956, 066464) reported female rats to be more sensitive than males to
the hemolytic effects of EGBE. In dogs, slight increases in erythrocyte osmotic fragility in both
the male and female basenji hybrids were seen, but RBC counts and Hb concentrations were
slightly decreased in the female. Overall, the effects were seen in both genders, but appeared to
develop more slowly in the males. In monkeys, occasional rises in erythrocyte osmotic fragility
were recorded during the exposure period more frequently in the female than in the male.
In the process of studying and comparing the metabolic and cellular basis of EGBE-
induced hemolysis of rat erythrocytes in vitro with human erythrocytes, Ghanayem (1989,
042014) observed that the blood from male and female human volunteers was unaffected by 4-
hour incubations with BAA at concentrations of up to 4.0 mM. At 8 mM, only slight but
significant hemolysis of human blood was observed, with blood from females showing a slightly
greater sensitivity.
The NTP 2-year inhalation bioassay (Dill et al., 1998, 041981; NTP, 2000, 196293) also
reported evidence of gender specificity in mice and rats, particularly with respect to the
elimination of BAA in rats. Female rats eliminated BAA more slowly from the blood, as
indicated by a smaller elimination rate constant, longer elimination ty2, and larger AUC. In
addition, the Cmax of BAA was greater for females at each concentration and time point. It has
been suggested because a smaller amount of BAA was excreted in the urine of female rats, that
higher blood concentrations of BAA accumulated in the females (Dill et al., 1998, 041981).
Mouse data from the NTP (2000, 196293) study also suggest a slightly increased hematologic
effect among female mice; however, while female mice tended to have higher blood
concentrations of BAA, they excreted more BAA in urine than male mice.
77

-------
5. DOSE-RESPONSE ASSESSMENTS
5.1. INHALATION REFERENCE CONCENTRATION (RFC)
In general, the RfC is an estimate, with uncertainty spanning perhaps an order of
magnitude, of a daily exposure to the human population—including susceptible subgroups—that
is likely to be without an appreciable risk of adverse health effects over a lifetime. It is derived
from a lower confidence limit on the BMD, a NOAEL, a LOAEL, or another suitable point of
departure (POD), with uncertainty/variability factors applied to reflect limitations of the data
used. The RfC is expressed in terms of mg/m3 of exposure to an agent and is derived by a
methodology similar to the RfD. Ideally, studies with the greatest duration of exposure and
conducted via the inhalation route of exposure give the most confidence for derivation of an RfC.
5.1.1. Choice of Principal Study and Critical Effect—with Rationale and Justification
Only one human occupational exposure study to low levels of EGBE is available, which
did not observe changes outside the normal clinical ranges in hepatic, renal, or hematologic
parameters (Haufroid et al., 1997, 042040) (Section 4.1). The animal studies considered for
selection as principal studies include the 14-week and 2-year inhalation studies by NTP (2000,
196293) in rats and mice, the developmental toxicity study by Tyl et al. (1984, 100209) in rats
and rabbits, the developmental toxicity study by Nelson et al. (1984, 031878) in rats, and the
subchronic study by Dodd et al. (1983, 066465) in rats. The NTP (2000, 196293) study was
selected as the principal study because it was conducted in two species and provides data for
different durations and for more dose groups than the other studies. The developmental toxicity
studies identified effects at doses higher than the doses associated with the critical effects
identified in the NTP (2000, 196293) study and were not used for quantitative purposes. While
the subchronic study by Dodd et al. (1983, 066465) was well-conducted, the NTP (2000,
196293) study contained more dose groups, more animals per group, and a longer duration of
exposure. Thus, Dodd et al. (1983, 066465) was not used for quantitative purposes. Two
endpoints from the NTP (2000, 196293) study—the hemolytic endpoint from the 14-week
inhalation study and the hemosiderin deposition endpoint from the 2-year inhalation study—
were used for the critical effect. The hemolytic endpoints in the 1999 EGBE Toxicological
Review were used to derive the reference values (see Section 5.1.5), but were not used to derive
the values in this updated assessment. New MOA information published since the 1999 EGBE
Toxicological Review is included in this document, and this information supports the
Note: Hyperlinks to the reference citations throughout this document will take you to the NCEA HERO database (Health and
Environmental Research Online) at http://epa.gov/hero. HERO is a database of scientific literature used by U.S. EPA in the process of
developing science assessments such as the Integrated Science Assessments (ISA) and the Integrated Risk Information System (IRIS).
78

-------
hemosiderin deposition endpoint as an important key event in the proposed MOA. A
comparison of the NOAELs and LOAELs for the candidate studies are summarized in Table 5-1.
Table 5-1. Results of candidate studies
Reference
Species
(strain)
Gender
Number/
dose group
Duration/effect
Effect levels (ppm)
NOAEL
LOAEL
NTP (2000,
196293s)
Rat (F344)
M
9-10
14 wk, hematologic
31
62.5

50
2 yr, hematologic,
hemosiderin (liver)
-
31
F
9-10
14 wk, hematologic
-
31
50
2 yr, hematologic,
hemosiderin (liver)
-
31
NTP (2000,
196293)
Mouse
(B6C3F0
M
50
2 yr, histopathology of the
forestomach
-
62.5
50
2 yr, hematologic,
hemosiderin (liver)
62.5
125
F
50
2 yr, histopathology of the
forestomach
-
62.5
50
2 yr, hematologic,
hemosiderin (liver)
-
62.5
Tyl etal. (1984,
100209)
Rat (F344)
F
36
GD 6-15, hematologic
50
100
Nelson et al.
(1984. 031878)
Rat (Sprague-
Dawley)
F
15
GD 7-15, hematologic
150
200
Dodd et al.
(1983. 066465s)
Rat (F344)
M, F
16
13 wk, hematologic
25
77
The primary effects of EGBE exposure were hematological effects and were observed in
both species and genders tested. Female rats (NTP, 2000, 196293) appeared to be most sensitive
among animals studied. A mild-to-moderate regenerative anemia was observed in females
exposed to all concentrations, with a LOAEL of 31 ppm identified for hematological effects in
male and female rats and no NOAEL. Exposure-related trends were noted for reticulocyte count,
RBC count, MCV, Hb concentration, and Hct. The hematological endpoints were considered for
the derivation of the RfC; however, they presented a number of difficulties. It was not clear
which of the hematological endpoints (changes in RBC count, reticulocyte count, MCV, Hb
concentration, and Hct) observed in EGBE-exposed animals should be used to derive an RfC. In
the case of BMD analysis, the proper benchmark response (BMR) level for the BMD derivation
was uncertain. In addition, while these hematologic effects were observed in both the subchronic
and chronic studies and persisted with exposure duration, they did not progress in severity in the
subchronic-to-chronic study (see Tables 4-3, 4-6, and 5-2). Further, better model fits were
obtained from the BMD analysis of the subchronic study, which used two more exposure
79

-------
concentrations than the chronic study. For these reasons, the hematologic responses from the
14-week subchronic study were chosen for use in the BMD analyses of this endpoint (see
Appendix C). Selection of the most appropriate hematologic endpoints for use in the BMD
analysis also required consideration of EGBE's MOA for hemolysis.
Table 5-2. Female and male rat and mouse liver hemosiderin staining
incidence and RBC counts from subchronic and chronic EGBE inhalation
studies
Effect/time
Control
31 ppm
62.5 ppm
125 ppm
250 ppm
500 ppm
Female rat
Hemosiderin
14 wk
2 yr
0/10
15/50
0/10
19/50
10/10°
36/50b
10/10°
47/50°
9/9°
NT
5/5°
NT
RBC count3
14 wk
1 yr
8.48 ±0.05
7.81 ±0.05
8.08 ± 0.07° (95)
NT
7.70 ± 0.08° (91)
7.42 ± 0.06° (95)
6.91 ±0.05° (82)
6.75 ± 0.05° (86)
6.07 ± 0.04° (72)
NT
4.77 ±0.15° (56)
NT
Male rat
Hemosiderin
14 wk
2 yr
0/10
23/50
0/10
30/50
0/10
34/50b
7/10°
42/50b
10/10°
NT
10/10°
NT
RBC count3
14 wk
1 yr
9.05 ±0.08
8.88 ±0.08
8.71 ±0.14b (96)
NT
8.91 ±0.06 (94)
8.39 ± 0.15° (94)
8.01 ±0.08° (89)
7.43 ± 0.20° (84)
7.10 ±0.07° (78)
NT
5.97 ± 0.05° (66)
NT
Female mouse
Hemosiderin
14 wk
2 yr
0/10
0/50
0/10
NT
0/10
5/50°
0/10
25/493
10/103
44/503
6/63
NT
RBC count3
14 wk
1 yr
9.72 ±0.05
9.32 ±0.09
9.55 ± 0.06b (98)
NT
9.51 ± 0.06b (98)
9.14 ±0.08 (98)
9.18 ±0.05° (94)
8.50 ±0.12° (91)
8.57 ± 0.06° (88)
8.08 ± 0.09° (87)
7.35 ± 0.07° (76)
NT
Male mouse
Hemosiderin
14 wk
2 yr
0/10
0/50
0/10
NT
0/10
0/50
0/10
8/493
0/10
30/493
6/63
NT
RBC count3
14 wk
1 yr
9.71 ±0.22
9.58 ±0.07
10.04 ±0.08
(103)
NT
9.77 ±0.10 (101)
9.73 ±0.49 (102)
9.47 ± 0.06b (98)
9.36 ± 0.32b (98)
8.90 ± 0.07° (92)
8.33 ±0.10° (87)
7.21 ±0.23° (74)
NT
aMean ± standard error with RBC counts expressed aslO'/|iL; percent of control in parentheses.
Statistically significant difference, p < 0.05.
Statistically significant difference, p < 0.01.
NT = not tested
Source: NTP (2000,196293).
The suggested MOA of EGBE hemolysis is based on data indicating that BAA, an
oxidative metabolite of EGBE and the first hypothesized event in the MOA, is likely to be the
80

-------
causative agent in hemolysis (Carpenter et al., 1956, 066464; Ghanayem et al., 1987, 041608;
Ghanayem et al., 1990, 042016). The second event in the MO A is erythrocyte swelling and cell
lysis, which is believed to be preceded by an increase in the osmotic fragility and a loss of
deformability of the erythrocyte (Ghanayem, 1989, 042014; Udden, 1994, 042109; Udden, 1996,
042209; Udden and Patton, 1994, 056374). This results in decreased values for RBC count, Hb,
and Hct and in response, an increase in the production of immature RBCs (reticulocytes) by the
bone marrow.
Although changes in reticulocyte and nucleated erythrocyte counts sometimes represent
the largest measurable differences between exposed animals and unexposed control animals, this
parameter is highly variable and does not always exhibit a dose-dependent trend (NTP, 1993,
042063; 2000, 196293). While these endpoints can be indirect markers of RBC lysis, they are
governed by multiple feedback control processes that can be both very sensitive and variable.
Therefore, a change in reticulocyte or nucleated erythrocyte count is not considered a suitable
endpoint for deriving the RfC or RfD.
Until more is known about the molecular interaction between BAA and specific cellular
molecules, it is reasonable to assume that changes in MCV and RBC count are measurements of
precursor events in response to both oral and inhalation EGBE exposure. Therefore, dose-
response information on MCV and RBC count are key endpoints used in the BMD analyses and
were considered for derivation of the RfC and RfD for EGBE.
While the toxicokinetic data suggest that MCV should theoretically be the earlier
indicator of hemolytic effects from EGBE exposure, recent studies suggest that the relationship
between the rate of MCV increase and RBC count decrease may not be consistent across
exposure protocols. In the gavage studies of Ghanayem et al. (1987, 066470) and the inhalation
studies of NTP (2000, 196293). Hct, a measure of RBC volume relative to blood volume, tended
to decrease along with RBC count and Hb at all exposure levels for which a hematologic effect
was observed. However, Hct did not change as RBC count and Hb decreased following drinking
water exposures (NTP, 1993, 042063). Thus, the loss of erythrocytes in the drinking water
studies (reduced RBC count) may have been offset by a concurrent increase in the size of the
individual cells (increased MCV). This was not the case in the gavage and inhalation studies.
For these reasons, greater weight is given to reduced RBC count, as opposed to increased MCV.
While the hemolytic effects appeared to be among the earliest effects from EGBE
exposure, the hemosiderin deposition endpoint was selected as the critical effect. This effect was
found to occur in both species and genders of animals tested, with rats being the more sensitive
species; the effect also occurred in the 14-week subchronic NTP inhalation study. The suggested
MO A of EGBE-induced liver effects is based on the observation that the hemolytic effects led to
compensatory erythropoiesis and significant increases in blood degradation products, including
an increased accumulation of hemosiderin in the liver Kupffer cells of EGBE-exposed animals.
The hemosiderin accumulation seen in the Kupffer cells was found to increase in severity with
81

-------
increasing dose and exposure duration (Table 5-2), unlike the hemolytic endpoints, such as
decreased Hct, which did not progress from 3 to 12 months (Table 4-6). Thus, hemosiderin
deposition in Kupffer cells in the rat liver is believed to be a sequela to the hematologic effects.
Because of the progression of this effect with chronic exposure, hemosiderin is deemed to be the
most sensitive effect. A NOAEL was not identified, while a LOAEL of 31 ppm was identified in
both male and female rats.
The 2-year chronic inhalation study by the NTP (2000, 196293) observed forestomach
ulcers in female mice at all exposure levels, but this effect has not been observed in any other
species, including mice exposed orally to EGBE (NTP, 1993, 042063). Though the incidence of
this lesion increased with exposure, severity of the lesion did not increase with increasing dose.
While this effect was not considered a critical effect for the derivation of an RfC, Appendix C
contains the BMD analysis for this endpoint for comparison purposes.
5.1.2. Methods of Analysis—Including Models (PBPK, BMD, etc.)
PODs for the RfC derivation in terms of the human equivalent concentrations (HECs)
have been calculated via the application of PBPK modeling and the use of internal dose metrics
published by Dill et al. (1998, 041981) to obtain NOAEL and benchmark concentration, 95%
lower bound (BMCL) estimates. Details of the various POD derivation approaches that were
considered are provided in Appendix C. The selected approach is described in Section 5.1.2.1
below.
The PBPK models developed for EGBE are briefly summarized in Table 5-3. Shyr et al.
(1993, 006766) and Johanson (1986, 100159) do not address BAA distribution and are only
parameterized for humans and rats, respectively. In the 1999 EGBE ToxicologicalReview, the
model described by Lee et al. (1998, 041983) was considered to be the most appropriate model
for the estimation of rat and mouse internal doses following inhalation exposure. Since the 1999
Toxicological Review (U.S. EPA, 1999, 597365). Corley et al. (2005, 100100) published a
revision to the Lee et al. (1998, 041983) model for rats and mice where several assumptions used
by Lee et al. (1998, 041983) were replaced with measured values (e.g., protein binding, partition
coefficients, metabolism rate constants for multiple pathways, and renal clearance) as a function
of species, gender, and age. As is described in Appendix C, that model was used to estimate the
Cmax of BAA in rat blood for the derivation of reference values from hematologic endpoints, and
the human PBPK model of Corley et al. (1994, 041977; 1997, 041984) was used to obtain
estimates of human inhalation exposure concentrations associated with the rat BAA Cmax
estimates. In the analysis of the hemosiderin effect in rats and mice described in Section 5.1.2.1
below, the human PBPK model of Corley et al. (1994, 041977; 1997, 041984) was used to obtain
estimates of human inhalation exposure concentrations associated with the BMDs derived from
82

-------
rat BAA AUC levels reported by Dill et al. (1998, 041981).1 Established U.S. EPA (2006,
194568) methods and procedures were used to review, select and apply these chosen PBPK
models.2
Table 5-3. Summary of PBPK models
Model
Species
Routes of exposure
Comments
Johanson (1986. 006758s)
Human
Inhalation
BAA not addressed
Shvretal. a993. 006766)
Rat
Inhalation, oral, dermal
BAA excretion
Corlev et al. (1994. 041977;
1997. 041984s)
Rat and human
Inhalation, oral, dermal
BAA distribution and excretion; male
rats only
Lee et al. (1998. 041983)
Rat and mouse
Inhalation
BAA distribution and excretion; males
and females
Corlev etal. C2005. 100100)
Rat and mouse
Inhalation, oral, dermal,
i.p., i.v.
Age-dependent BAA distribution,
metabolism and excretion, males and
females
Franks etal. (2006. 100117)
Human
Inhalation and dermal
Extended Corlev et al. < 1997. 041984)
model to include bladder compartment
for human biomonitoring studies
5.1.2.1. BMD Approach Applied to Hemosiderin Staining Data
For the purposes of deriving an RfC for EGBE, hemosiderin staining data were evaluated
in male and female rats from the 2-year chronic study by NTP (2000, 196293). The current
BMD technical guidelines (U.S. EPA, 2000, 052150) suggest the use of 10% extra risk as a
BMR level for quantal data, as this is at or near the limit of sensitivity in most cancer bioassays
and in some noncancer bioassays as well. Because the hemosiderin staining endpoint was
observed in control animals and a 10% increase in incidence was within the observable range of
the data, 10% extra risk was considered an appropriate BMR and a BMCLio an appropriate POD
for derivation of the RfC (1995, 005992; 2000, 052150). All BMD assessments in this review
were performed using U.S. EPA benchmark dose software (BMDS) version 1.4.1. Graphical
figures and text output files for selected benchmark concentration (BMC) analyses are provided
in Appendix C.
The AUC was selected as the appropriate dose metric due to the nature of the endpoint,
hemosiderin deposition. This endpoint increased in severity with increased duration (subchronic
to chronic) and is believed to be the result of the cumulative exposure to EGBE as opposed to a
peak event. Table 5-4 reports AUC BAA blood concentrations measured at 12 months3
published by Dill et al. (1998, 041981) in both genders of B6C3Fi mice and F344 rats exposed to
the same concentrations used in the NTP (2000, 196293) chronic studies of these test animals.
lrThe basic components of the Corley model are summarized in Appendix B.
2EPA notes that a review of the PBPK models was conducted prior to their use in the 1999 EGBE toxicological review.
3Dill et al. (1998, 041981) also reported 18 month data, but due to the smaller number of animals and higher variability in this data, the 12
month data were used for the purposes of this analysis.
83

-------
Table 5-4. AUC BAA blood concentrations measured at 12 months in both
genders of B6C3Fi mice and F344 rats
Rats


AUCbaa (jtmol-hr/L)a
Exposure concentration (ppm)
Gender
n
Mean
SE
31.2
Male
7
358.3
16.6
Female
5
638.8
18.7
62.5
Male
6
973.0
86.2
Female
9
1,128.9
50.9
125
Male
9
2,225.6
71.1
Female
12
3,461.8
154.8
Mice


AUCbaa (jtmol-hr/L)a
Exposure concentration (ppm)
Gender
n
Mean
SE
62.5
Male
10
1,206.6
205.6
Female
12
1,863.6
112.4
125
Male
9
2,819.8
685.1
Female
6
5,451.6
508.9
250
Male
10
17,951.5
1,770.4
Female
11
18,297.1
609.7
aAuthors reported AUC values in terms of (ig/min and g, which were converted to units consistent with the PBPK
model of (imol-hour/L by dividing by 60 minute/hour and 132.16 g/mol and multiplying by 1,060 g/L.
Source: Dill et al. (1998, 0419811.
The fit statistics and BMC information derived from the dichotomous models available in
the BMD software as applied to the male and female rat hemosiderin staining data versus AUC
BAA are shown in Table 5-5. All models were fit using restrictions and option settings suggested
in the U.S. EPABMD technical guidance document (U.S. EPA, 2000, 052150). The best model
fit to these data, as determined by visual inspection, examination of low dose model fit
(i.e., scaled residual for the dose group closest to the BMD), and comparison of overall fit
(i.e., Akaike information criterion [AIC] values), was obtained using a multistage model (1st
degree) for the male response data and a Log-Logistic model for the female response data. The
male rat BMCio was 196 [j,mol-hour/L and the BMCLio was determined to be 133 [j,mol-hour/L,
using the 95% lower confidence limit of the dose-response curve expressed in terms of the AUC
for BAA in blood. The BMC io and BMCLio values for the female rat were determined to be 425
and 244 [j.mol-hour/L, respectively. Assuming continuous exposure (24 hour/day), the Corley
et al. (1997, 041984) PBPK model was used to back-calculate HECs of 3.4 ppm (16 mg/m3)
from the male rat data and 4.9 ppm (24 mg/m3) from the female rat data.
84

-------
Table 5-5. Comparison of BMC/BMCL values for male and female rat liver
hemosiderin staining data from inhalation chronic study using measured
blood AUC (12 months) of the EGBE metabolite BAA as a common dose
metric
Model
BMCio (jtmol-hr/L)
BMCLio (nmol-hr/L)
/7-Value
AICa
Scaled residualb
Male rats
Multistage-lst degree0
196.252
133.141
0.8680
247.234
0.441
Gamma0
196.253
133.141
0.8680
247.234
0.441
Logistic
259.296
192.773
0.7692
247.476
0.526
Log-logistic
166.376
69.3279
0.5623
249.283
0.313
Probit
271.525
205.882
0.7450
247.54
0.517
Log-probit
368.336
241.992
0.6309
247.876
0.765
Weibull0
196.253
133.141
0.8680
247.234
0.441
Female rats
Multistage-lst degree
122.166
214.555
0.0698
218.868
-1.945
Gamma
316.635
134.02
0.0554
219.229
-1.238
Logistic
273.693
221.689
0.0993
218.188
-1.294
Log-logisticc
424.527
243.69
0.1533
217.526
-0.896
Probit
291.017
241.206
0.0683
218.985
-1.260
Log-probit
427.728
248.683
0.1238
217.884
-0.965
Weibull
266.515
130.801
0.0454
219.58
-1.377
AIC = -2L + 2P, where L is the log-likelihood at the maximum likelihood estimates for the parameters and P is the
number of model degrees of freedom (usually the number of parameters estimated).
b%2 residual (measure of how model predicted response deviates from the actual data) for the dose group closest to
the BMC scaled by an estimate of its SD. Provides a comparative measure of model fit near the BMC. Residuals
that exceed 2 in absolute value should cause one to question model fit in this region.
°Model choice based on adequate p-valuc (>0.1), visual inspection, low AIC, and low (absolute) scaled residual.
The Multistage (1st degree) is referred to as the chosen model for male rats, though equivalent fit was obtained by
the restricted Gamma and Weibull models.
Likewise, the fit statistics and BMC information for male and female mouse hemosiderin
staining data versus AUC BAA are shown in Table 5-6. All models were fit using restrictions
and option settings suggested in the U.S. EPA BMD technical guidance document (U.S. EPA,
2000, 052150). The best model fit to these data, as determined by visual inspection, examination
of low dose model fit (i.e., scaled residual for the dose group closest to the BMD), and
comparison of overall fit (i.e., AIC values), was obtained using a log-probit model for both the
male and female response data. The male mouse BMCio was 3,077 [j,mol-hour/L and the
BMCLio was determined to be 2,448 [j,mol-hour/L using the 95% lower confidence limit of the
dose-response curve expressed in terms of the AUC for BAA in blood. The BMCio and BMCLio
values for the female mouse were determined to be 1,735 and 1,322 [j,mol-hour/L, respectively.
Assuming continuous exposure (24 hour/day), the Corley et al. (1997, 041984) PBPK model
described in Appendix B was used to back-calculate HECs of 36 ppm (174 mg/m3) from the
male mouse data and 20 ppm (97 mg/m3) from the female mouse data.
85

-------
Table 5-6. Comparison of BMC/BMCL values for male and female mouse
liver hemosiderin staining data from inhalation chronic study using
measured blood AUC (12 months) of the EGBE metabolite BAA as a
common dose metric
Model
BMC10
(nmol-hr/L)
BMCL10
(nmol-hr/L)
/7-Value
AICa
Scaled residualb
Male mice
Multistage-1st
degree
2,100.07
1,613.9
0.3067
117.571
-1.766
Gamma
2,725.35
1,702.27
0.1452
118.559
1.358
Logistic
6,605.45
5,333.72
0.0022
127.326
2.789
Log-logistic
2,616.51
1,628.48
0.1882
118.02
1.193
Probit
5,917.06
4,825.09
0.0031
126.405
2.734
Log-probitc
3,076.8
2,448.3
0.1290
116.614
1.946
Weibull
2,689.76
1,687.09
0.1445
118.712
-1.448
Female mice
Multistage-1st
degree
946.491
769.879
0.3680
142.669
-1.583
Gamma
1,402.92
818.367
0.3420
143.288
-0.817
Logistic
2,897.15
2,341.03
0.0002
162.338
-0.942
Log-logistic
1,705.75
1,121.43
0.8223
141.501
-0.343
Probit
2,860.03
2,364.52
0.0002
161.681
-0.829
Log-probitc
1,734.53
1,322.06
0.8237
141.498
-0.315
Weibull
1,282.82
804.234
0.2958
143.631
-0.988
AIC = -2L + 2P, where L is the log-likelihood at the maximum likelihood estimates for the parameters and P is the
number of model degrees of freedom (usually the number of parameters estimated).
b%2 residual (measure of how model predicted response deviates from the actual data) for the dose group closest to
the BMC scaled by an estimate of its SD. Provides a comparative measure of model fit near the BMC. Residuals
that exceed 2 in absolute value should cause one to question model fit in this region.
°Model choice based on adequate p-valuc (>0.1), visual inspection, low AIC, and low (absolute) scaled residual.
The Log-probit model provides a slightly better fit than other models for both genders.
5.1.2.2. Selection of the POD
Consideration of the available data has led to the selection of the 2-year inhalation study
(NTP, 2000, 196293) and increased hemosiderin staining in the liver of male F344 rats as the
principal study and critical effect for deriving the chronic RfC for EGBE. This is a high-quality
study and, when coupled with information on the MO A, U.S. EPA concluded that this is a
precursor to an adverse effect and is appropriate for use in deriving the RfC. ABMCLio of 133
[j,mol-hour/L for hemosiderin staining in liver of male rats chronically exposed to EGBE (NTP,
2000, 196293) was used as the POD to calculate the RfC. A human PBPK model (Corley et al.,
1997, 041984) was used to back-calculate to a HEC of 16 mg/m3 (3.4 ppm) for the BMCLrec-
86

-------
5.1.3. RfC Derivation—Including Application of Uncertainty Factors
Uncertainty factors (UFs) are applied to account for recognized uncertainties in
extrapolation from experimental conditions to the assumed human scenario (e.g., chronic
exposure over a lifetime when subchronic studies are used for deriving a toxicity value). The
1/2
application of UFs may include the use of a partial UF of 10 (3.2) based on the assumption that
the actual values for the UFs are log-normally distributed. Application of these factors in the
assessments is such that, when a single partial UF is applied, the factor is rounded to 3—for
instance, the total factor for individual UFs of 3 and 10 would be 30 (3 x 10). When two partial
UFs are evoked, however, they are not rounded, such that a UF of 3, 3, and 10 would result in a
total uncertainty factor of 100 (actually 101/2 x 101/2 x 101) (U.S. EPA, 2002, 0888241 UFs
applied for this RfC assessment and the justification for their use are as follows.
A factor of 10 was selected to account for the uncertainty associated with the variability
of the human response (UFH) to the effects of EGBE. Potentially susceptible subpopulations
include individuals with enhanced metabolism or decreased excretion of BAA and individuals
whose RBC membranes are more susceptible to the lysis caused by BAA, the precursor step to
developing hemosiderin staining in the liver. Human in vitro studies suggest that the elderly and
patients with fragile RBCs would not be more sensitive to the hemolytic effects of EGBE than
normal adults. Laboratory animal studies suggest that older animals are more sensitive than
neonates and that females are more sensitive than males. While developmental studies do not
reveal increased susceptibility in infants, none of the developmental studies examined fetal or
infant blood for signs of effects from prenatal exposure to EGBE. Additionally, human
responses to EGBE have not been observed under a broad range of exposure conditions (e.g.,
repeated or long-term exposures) and potentially sensitive subjects (e.g., individuals predisposed
to hemolytic anemia or infants).
A factor of 1 was selected to account for the uncertainty associated with interspecies
variability resulting from toxicodynamic and toxicokinetic differences between animals and
humans (UFA). Traditionally, these components (toxicodynamic and toxicokinetic) are
individually represented by partial UFs of 3 for a total UF of 10 in the absence of chemical-
specific information; thus, application of a full UF of 10 would depend on two areas of
uncertainty (i.e., toxicokinetic and toxicodynamic uncertainties). In this assessment, the
toxicokinetic uncertainty is addressed by the determination of an HEC, using a combination of
measured internal blood levels in the test animals and PBPK modeling. Thus, a value of 1 was
selected for the toxicokinetic portion of the UFA. Regarding toxicodynamics, in vivo (Carpenter
et al., 1956, 066464) and in vitro (Ghanayem and Sullivan, 1993, 041609; Udden, 2002, 042 111;
Udden and Patton, 1994, 056374) studies indicate that humans may be significantly less sensitive
than rats to the hematological effects of EGBE. For this reason, a value of 1 was selected for the
toxicodynamic portion of the UFA.
87

-------
A factor to account for extrapolation from subchronic to chronic exposure (UFS) was not
needed because the RfC was derived from a chronic inhalation study.
A factor to account for the extrapolation from a LOAEL to a NOAEL (UFL) was not
applied because the current approach is to address this extrapolation as one of the considerations
in selecting a benchmark response (BMR) for BMD modeling. In this case, EPA concluded a
10% increase in hemosiderin staining, indicating a precursor to an adverse effect, is appropriate
for use in deriving the RfC.
A factor of 1 was selected to account for deficiencies in the database (UFD). Studies that
are available include chronic and subchronic studies for two species (rats and mice), and several
reproductive and developmental studies, including a two-generation reproductive toxicity study.
There are also limited human studies available following short-term inhalation exposure.
A total UF of 10 (10 for UFH, 1 for UFA, and 1 for UFD) was used in the derivation of the
RfC. The combined PBPK and BMC modeling method using hemosiderin as an endpoint was
used to derive the RfC. In addition, MO A information was used to inform the choice of the
critical effect. The RfC for EGBE based on hemosiderin deposition in the liver was calculated as
follows:
RfC = BMCLhec - UF
= 16 mg/m3 10
= 1.6 mg/m3
Thus, the RfC is 1.6 mg/m3.
88

-------
5.1.4. RfC Comparison Information
For comparison purposes, Figure 5-1 presents the POD, applied UFs, and derived
reference values, including the RfC, for the effect endpoints discussed. BMC modeling was
done using U.S. EPA BMDS version 1.4.1 (U.S. EPA, 2000, 052150). and results are provided in
Appendix C. This comparison is intended to provide information on alternative endpoints
associated with EGBE exposure. The selected RfC value is circled; BMC analyses are provided
in Appendix C.
~ PODh
~ UF^
ufa
m
UF,
RfC
1000
100 -
10 -
0.1
NTP (2000); 2 yr
male and female
rat study;
hematological
effects; LOAEL
NTP (2000);
3 mo female rat
study; RBC
effects;
BMCLos(HEC)
NTP (2000); 2 yr
female rat study;
hemosiderin
staining;
BMCL10(HEC)
NTP (2000); 2 yr
male rat study;
hemosiderin
staining;
BMCL10(HEC)
NTP (2000); 2 yr
female mouse
study;
hemosiderin
staining;
BMCL-io(HEC)
NTP (2000); 2 yr
male mouse
study;
hemosiderin
staining;
BMCL10(HEC)
NTP (2000); 2 yr
female mouse
study;
forestomach
hyperplasia;
BMCL10(HEC)
Figure 5-1. PODs for selected endpoints with corresponding applied UFs
and derived RfC.
Figure 5-1 shows PODs and comparison reference values (including the RfC) that could
be derived from the various endpoints to allow a comparison with the chosen critical effect and
the resultant RfC for the critical effect. Hemolytic effects and effects related to hemolysis (i.e.,
hemosiderin deposition) are the most sensitive endpoints for identification of a NOAEL or a
BMCL in the subchronic and chronic studies available; these endpoints have been considered as
89

-------
the critical endpoint for derivation of the RfC. The BMCL05 for RBC count changes in female
rats was 133 [xM, using Cmax at 3 months, and was converted to an inhalation HEC (BMCLrec)
of 225 mg/m3 using the U.S. EPA model. Though adequate model fit per U.S. EPABMD
technical guidance (U.S. EPA, 2000, 052150) could not be obtained for the NTP (2000, 196293)
3-month male rat RBC response, BMCs derived for hemolytic endpoints in male rats of this
study were approximately twofold higher than for female rats (data not shown). For the
hemosiderin endpoint, both male and female data from the chronic study were considered. The
BMCL 10 for hemosiderin stcunin.^ in mcile fcits Wcis 133 pJVI~houx/L usin^ the A_T_JC for BA..A in
arterial blood at 12 months and was converted to a BMCLrec of 16 mg/m3 using the Corley et al.
(1994, 041977; 1997, 041984) human PBPK model. The BMCLio for hemosiderin staining in
female rats was 244 [xM-hour/L using the AUC for BAA in arterial blood at 12 months and was
converted to a BMCLrec of 24 mg/m3 using the Corley et al. (1994, 041977; 1997, 041984)
human PBPK model.
5.1.5. Previous Inhalation Assessment
The previous IRIS assessment for EGBE was entered into the database on December 31,
1999; it contains an inhalation RfC of 13 mg/m3. The RfC was based on the BMCos(hec) of
380 mg/m3 for changes in RBC count in female F344 rats following a 14-week inhalation
exposure (NTP, 1998, 594421). A total UF of 30 was used to account for human variability and
extrapolation from an adverse effect level.
5.2. ORAL REFERENCE DOSE (RfD)
In general, the RfD is an estimate, with uncertainty spanning perhaps an order of
magnitude, of a daily exposure to the human population—including susceptible subgroups—that
is likely to be without an appreciable risk of adverse health effects over a lifetime. It is derived
from a statistical benchmark dose, 95% lower bound (BMDL), a NOAEL, a LOAEL, or another
suitable POD, with uncertainty/variability factors applied to reflect limitations of the data used.
The RfD is expressed in terms of mg/kg-day of exposure to an agent and is derived by a similar
methodology to the RfC. Ideally, studies with the greatest duration of exposure and conducted
via the oral route of exposure give the most confidence for derivation of an RfD. The database
of oral studies for EGBE is more limited than the database of inhalation studies. For this reason,
a PBPK model for EGBE has been applied to the inhalation data for derivation of an RfD.
5.2.1. Choice of Principal Study and Critical Effect—with Rationale and Justification
No studies have been reported in which humans have been exposed subchronically or
chronically to EGBE by the oral route of exposure, and thus would be suitable for derivation of
an oral RfD. No chronic oral laboratory animal studies are currently available for EGBE. The
90

-------
results of the only two available subchronic 91-day drinking water studies in rats and mice (NTP,
1993, 042063) are summarized in Table 5-7.
Table 5-7. Subchronic 91-day drinking water studies in rats and mice
Reference
Species
(strain)
Gender
Animals/dose
Effect
Effect levels
(mg/kg-d)
NOAEL
LOAEL
NTP (1993,
042063s)
Rat
(F344)
M
10
Hepatocellular changes
-
54.9a
F
10
Hematological
-
58.6a
NTP (1993,
042063s)
Mouse
(B6C3F0
M
10
Body weight
223
553b
F
10
Body weight
370
676b
aDoses were calculated using water consumption rates and body weights measured during the last week of exposure
and, therefore, differ slightly from those reported by the authors and presented in Section 4.2.
bThe LOAEL in mice was based on reduced body weight and body weight gain.
Based on a comparison of NOAELs and LOAELs for hematological and liver effects, rats
are clearly more sensitive to the noncancer effects of EGBE than mice. As discussed in
Section 4.2, hematologic and hepatocellular changes were noted in both rat genders. In females,
both hematologic and hepatocellular changes were noted at the low-dose level (58.6 mg/kg-day,
using water consumption rates and body weights measured during the last week of exposure).
Only hepatocellular cytoplasmic changes were observed in low-dose male rats (54.9 mg/kg-day,
using water consumption rates and body weights measured during the last week of exposure).
In the female rat study (NTP, 1993, 042063). groups of 10 F344 rats were exposed to 0,
82, 151, 304, 363, and 470 mg/kg-day EGBE via drinking water for 13 weeks. Body and organ
weights were measured, and clinical, hematological, gross, and histopathologic examinations
were conducted. Decreases in body weight were observed in female rats exposed to the two
highest dose levels. Hematologic changes were observed at all dose levels after 13 weeks and
were indicative of mild-to-moderate anemia. These changes included reduced RBC count, Hb,
and Hct, as well as increased reticulocyte count and MCV. Liver hemosiderin pigmentation was
noted in the cytoplasm of Kupffer cells in both genders. In females it was noted in 0/10 controls
in 82 mg/kg-day treated animals, 2/10 with a severity grade of 1 (minimal) at 151 mg/kg-day,
and 10/10 in the three highest dose levels, with the severities noted as increasing from a
numerical grade of 1.2 in the 304 mg/kg-day group to 1.9 in both of the upper two dose groups.
In males the pigmentation was noted in animals exposed to the highest dose only (452 mg/kg-
day) at an incidence of 7/10 and a severity rating of 1 (minimal). No hepatic pigmentation was
reported in the mice exposed for 13 weeks.
Hematological endpoints indicative of hemolysis do not progress with increasing duration
of inhalation exposure, whereas the incidence of hemosiderin pigmentation did progress
considerably with chronic exposure (Table 5-2). A primary consideration with respect to the
91

-------
NTP (1993, 042063) study was that it did not provide a sufficient duration of exposure to EGBE
to allow for maximal accumulation of hemosiderin deposition. In contrast, hemosiderin
accumulation was one of the most sensitive effects observed in the NTP (2000, 196293) chronic
inhalation study of EGBE. Furthermore, uncertainty regarding the mechanism of action of
EGBE does not allow for a biologically-informed determination regarding which hematological
endpoints—changes in RBC count, reticulocyte count, MCV, Hct, and Hb—observed in EGBE-
exposed animals should be used to derive an RfC and, in the case of BMD analysis, what a
proper BMR level should be for the BMD derivation. Finally, hematologic effects signified by
changes in RBC count, reticulocyte count, MCV, HCT, and Hb are considered precursor effects
to the pathological finding of hemosiderin deposition (Section 4.5). For these reasons and
because the hemolytic effects of EGBE appear to be consistent between oral and inhalation
routes of exposure, an RfD has been derived via the application of a PBPK model to perform a
route-to-route extrapolation (Section 5.2.2) from the BMD analysis (Section 5.1.2) of
hemosiderin pigmentation observed in the NTP (2000, 196293) chronic inhalation study of
EGBE. BMD/NOAEL analyses of hematologic endpoints and hemosiderin pigmentation
observed in the oral NTP (1993, 042063) subchronic study are provided below for comparison
purposes.
Another issue that needs to be addressed with respect to the NTP (1993, 042063) study is
the lack of reported forestomach lesions at even the highest drinking water doses administered in
this study relative to the considerable incidence of forestomach hyperplasia and ulceration
observed at all exposure levels in the NTP (2000, 196293) subchronic and chronic inhalation
studies of mice. This is difficult to explain, considering the lowest dose in the subchronic
drinking water study is predicted to result in similar, if not higher, Cmax blood levels of the
EGBE metabolite BAA (the presumed irritant) compared to the lowest exposure concentration in
the NTP (2000, 196293) subchronic inhalation study. There is no clear explanation. It has been
suggested that oral nonbolus dosing of EGBE does not result in high enough local concentrations
of EGBE and BAA (Poet et al., 2003, 100190). Studies with other forestomach carcinogens that
are not mutagenic have demonstrated that forestomach effects are dependent not only on the dose
but also on the chemical concentration in the dosing solution (Ghanayem et al., 1985, 042768).
and other effects of EGBE appear to be highly dependent on the concentration attained
(Ghanayem et al., 2000, 042748; Ghanayem et al., 2001, 042755; Long et al., 2000, 042753;
Nyska et al., 1999, 042747; Nyska et al., 1999, 042746). In addition, first-pass liver metabolism
of orally administered EGBE may affect the extent to which EGBE reaches the forestomach via
the route that has been proposed following i.v. injection: distribution to salivary glands followed
by the swallowing of EGBE-laden saliva (Green et al., 2002, 041610; Poet et al., 2003, 100190).
In any case, since forestomach irritation was not reported in rats or mice in the NTP (1993,
042063) drinking water study, this is not considered a sensitive endpoint, and route-to-route
92

-------
extrapolation of this endpoint from inhalation data is not considered appropriate for use in the
RfD derivation.
5.2.2. Methods of Analysis—Including Models (PBPK, BMD, etc.)
PODs for the RfD derivation in terms of the human equivalent doses (HEDs) have been
calculated via the application of PBPK modeling to NOAEL and BMDL estimates. Details of
the various POD derivation approaches that were considered are provided in Appendix C. The
selected approach is described below.
5.2.2.1.	BMD Approach Applied to Hemosiderin Endpoint
Due to the limited oral database, U.S. EPA concluded that a route-to-route extrapolation
will increase the confidence in the determination of the RfD POD. Inhalation studies considered
for derivation of the RfC are used to supplement the oral database using the route-to-route
extrapolation as described below.
5.2.2.2.	Route-to-Route Extrapolation from Inhalation Data
A route-to-route extrapolation was performed on the data used in the derivation of the
RfC from the NTP (2000, 196293) chronic inhalation study because of the lack of a chronic oral
study for EGBE. As with the animal to human extrapolation used in the development of the
RfC, the dose metric used for interspecies (rat to human) and route-to-route (inhalation to oral)
extrapolation was the AUC of BAA at 12 months in arterial blood. This dose metric was used
for dose-response modeling of chronic inhalation data (Section 5.1.2) to derive the POD of 133
[j.mol-hour/L, expressed as a BMDL. The BMDL was then back-calculated using the human
PBPK model (Corley et al., 1994, 041977; 1997, 041984) to obtain an equivalent human oral
drinking water dose (BMDLhed) of 1.4 mg/kg-day. As for the alternative HED estimations
described in Section C.2 of Appendix C, a simplifying assumption was used that the entire dose
of drinking water EGBE was consumed over a 12-hour period each day.
5.2.2.3.	Selection of the POD
The BMCL chosen for the RfC is used to determine the POD for the RfD. This value is
based on a more comprehensive chronic data set and is below the range of estimates from
available oral data of shorter duration of exposure. Hemosiderin deposition in male rat liver is
the critical effect chosen for derivation of the RfC. New MOA information (see Section 4.6.3.1)
supports the hemosiderin deposition endpoint as an important key event in the proposed MOA.
The BMCL for the RfC (AUC of 133 [xM-hour/L BAA in arterial blood at 12 months) is
converted using the Corley et al. (1994, 041977; 1997, 041984) model to an oral HED
(BMDLhed) of 1.4 mg/kg-day. This extrapolated oral value is consistent with and slightly lower
than the LOAELred of 18 mg/kg-day and the BMDLhed of 6.8 mg/kg-day estimated from the
subchronic oral (NTP, 1993, 042063) study (see Appendix C).
93

-------
5.2.3. RfD Derivation—Including Application of UFs
UFs are applied to account for recognized uncertainties in extrapolation from
experimental conditions to the assumed human scenario (e.g., chronic exposure over a lifetime
when subchronic studies are used for deriving a toxicity value). The application of UFs may
1/2
include the use of a partial UF of 10 (3.2) based on the assumption that the actual values for
the UFs are log-normally distributed. Application of these factors in the assessment is such that,
when a single partial UF is applied, the factor is rounded to 3 (e.g., the total factor for individual
UFs of 3 and 10 would be 30 [3 x 10]). When two partial UFs are evoked, however, they are not
rounded, such that a UF of 3, 3, and 10 would result in a total uncertainty of 100 (actually
101/2 x 101/2 x 101). UFs applied for this RfD assessment and the justification for their use
follow.
A factor of 10 was selected to account for the uncertainty associated with the variability
of the human response (UFH) to the effects of EGBE. Potentially susceptible subpopulations
include individuals with enhanced metabolism or decreased excretion of BAA and individuals
whose RBC membranes are more susceptible to the lysis caused by BAA, the precursor step to
developing hemosiderin staining in the liver. Human in vitro studies suggest that the elderly and
patients with fragile RBCs would not be more sensitive to the hemolytic effects of EGBE than
normal adults. Laboratory animal studies suggest that older animals are more sensitive than
neonates and that females are more sensitive than males. While developmental studies do not
reveal increased susceptibility in infants, none of the developmental studies examined fetal or
infant blood for signs of effects from prenatal exposure to EGBE. Additionally, human
responses to EGBE have not been observed under a broad range of exposure conditions (e.g.,
repeated or long-term exposures) and potentially sensitive subjects (e.g., individuals predisposed
to hemolytic anemia or infants).
A factor of 1 was selected to account for the uncertainty associated with interspecies
variability resulting from toxicodynamic and toxicokinetic differences between animals and
humans (UFA). Traditionally, these components (toxicodynamic and toxicokinetic) are
individually represented by partial UFs of 3 for a total UF of 10 in the absence of chemical-
specific information; thus, application of a full UF of 10 would depend on two areas of
uncertainty (i.e., toxicokinetic and toxicodynamic uncertainties). In this assessment, the
toxicokinetic uncertainty is addressed by the determination of an HEC, using a combination of
measured internal blood levels in the test animals and PBPK modeling. Thus, a value of 1 was
selected for the toxicokinetic portion of the UFA. Regarding toxicodynamics, in vivo (Carpenter
et al., 1956, 066464) and in vitro (Ghanayem and Sullivan, 1993, 041609; Udden, 2002, 042 111;
Udden and Patton, 1994, 056374) studies indicate that humans may be significantly less sensitive
than rats to the hematological effects of EGBE. For this reason, a value of 1 was selected for the
toxicodynamic portion of the UFA.
94

-------
A factor to account for extrapolation from subchronic to chronic exposure (UFS) was not
needed because the RfD was derived from a chronic inhalation study.
A factor to account for the extrapolation from a LOAEL to a NOAEL (UFL) was not
applied because the current approach is to address this extrapolation as one of the considerations
in selecting a benchmark response (BMR) for BMD modeling. In this case, EPA concluded a
10% increase in hemosiderin staining, indicating a precursor to an adverse effect, is appropriate
for use in deriving the RfD.
A factor of 1 was selected to account for deficiencies in the database (UFD). While no
chronic oral studies or adequate human data are available for EGBE, PBPK models allow for
deriving a BMDL from the chronic inhalation study using measured internal dose metrics and
then extrapolating it back to an equivalent human oral dose. The database for inhalation exposure
includes chronic and subchronic studies in two species (rats and mice), and several reproductive
and developmental studies, including a two-generation reproductive toxicity study.
A total UF of 10 (10 for UFH, 1 for UFA, and 1 for UFD) was used in the derivation of the
RfD. The RfD for EGBE based on hemosiderin deposition in the liver was calculated as follows:
RfD = BMDLhed - UF
= 1.4 mg/kg-day -MO
= 0.1 mg/kg-day
Thus, the RfD is 0.1 mg/kg-day.
5.2.4. RfD Comparison Information
For comparison purposes, Figure 5-2 presents the POD, applied UFs, and derived
reference values, including the RfD, for the effect endpoints discussed. BMC modeling was
done using U.S. EPA BMDS version 1.4.1 (U.S. EPA, 2000, 052150). and results are provided in
Appendix C. This comparison is intended to provide information on alternative endpoints
associated with EGBE exposure. The selected RfD value is circled; BMD analyses are provided
in Appendix C.
95

-------
100
a>
a>
o
cc
CD
O
c
o
O
0.1
0.01
~ PODhed I I UFh
(mg/kg-d) 1	1
UF,
10 r
NTP 1993; 91 d
oral study;
hematological
changes;
LOAEL(HED)
NTP 2000; 2 yr
Inhalation study;
hemosiderin
staining;
BMDLio(HED)
RfD
(mg/kg-d)
NTP 1993; 91 d
oral study;
RBC changes;
BMDL05(HED)
Figure 5-2. PODs for selected endpoints with corresponding applied UFs
and derived RfD.
Figure 5-2 shows PODs and comparison reference values (including the RfD) that could
be derived from the various endpoints to allow a comparison with the chosen critical effect and
the resultant RfD for the critical effect. Hematological effects and effects related to hemolysis
(i.e., hemosiderin deposition) are the most sensitive endpoints for identification of a NOAEL or a
BMDL in the subchronic and chronic studies available; these endpoints have been considered as
the critical endpoint for derivation of the RfD.
5.2.5. Previous Oral Assessment
The previous IRIS assessment for EGBE was entered into the database on December 30,
1999 and contains an oral RfD of 0.5 mg/kg-day. The RfD was based on the BMD05(hed) of
5.1 mg/kg-day for changes in MCV in female F344 rats following a 91-day drinking water
exposure (NTP, 1993, 042063). A total UF of 10 was used to account for human variability.
This assessment was conducted prior to the adoption of the current BMD technical guidance
document (U.S. EPA, 2000, 052150).
96

-------
5.3. UNCERTAINTIES IN THE DERIVATION OF THE INHALATION REFERENCE
CONCENTRATION (RfC) AND ORAL REFERENCE DOSE (RfD)
The following is a more extensive discussion of uncertainties associated with the RfC and
RfD for EGBE beyond the quantitative discussion in Sections 5.1.2, 5.1.3, 5.2.2, and 5.2.3. A
summary of these uncertainties, along with uncertainties specific to the Section 5.4 cancer
analysis, is presented in Table 5-8.
Table 5-8. Summary of uncertainty in the EGBE noncancer and cancer risk
assessments
Consideration
Potential impact
Decision
Justification
Choice of low-
dose
extrapolation
method
A linear low-dose
extrapolation would likely
drive up the risk
estimation when
combined with site-
specific exposure data.
Nonlinear approach; RfC
and RfD are considered
protective of the key
events leading to cancer.
Proposed key events in the two MO As
(forestomach irritation and hemolysis
leading to increased liver hemosiderin) are
not likely to occur in humans at the RfC or
RfD.
Choice of
endpoint
Use of forestomach
endpoint could increase
RfC by up to 18-fold (see
Section 5.3).
RfC is based on the most
sensitive endpoint,
increased liver
hemosiderin.
Chosen endpoint is considered most
relevant due to forestomach toxicokinetic
and exposure issues.
Choice of dose
metric
Alternatives could
increase or decrease
RfC/RfD (e.g., use of Cmax
BAA increases RfC by
two to threefold).
AUC for BAA in arterial
blood.
Evidence for a causal association between
the proposed key steps of BAA and
hemolysis leading to increased liver
hemosiderin and increased liver tumors.
AUC chosen because hemosiderin
increased with cumulative exposure to
EGBE/BAA.
POD derivation
method
RfC/RfD threefold lower
than for NOAEL.
BMD method used.
Advantages include capacity to account for
sample size and to provide confidence
bounds on dose.
Choice of model
for BMCL
derivation
Alternative models could
increase RfC up to
threefold (see
Section 5.3).
Multistage (1st degree)
model chosen.
The best-fitting model was chosen based
on U.S. EPA (2000. 052150) BMD
technical guidance.
Choice of animal
to human
extrapolation
method
Alternatives could
increase or decrease
RfC/RfD (e.g., default
would increase RfC by
twofold) (see Section 5.3).
A PBPK model was used
to extrapolate animal to
human concentrations.
Use of a PBPK model reduces uncertainty
associated with the animal to human
extrapolation. AUC blood levels of BAA
are an appropriate dose metric, and a peer-
reviewed and verified PBPK model exists
that estimates this metric.
Statistical
uncertainty at
POD
POD would be -40%
higher if BMD were used.
BMDL used per U.S.
EPA BMD guidance
(U.S. EPA, 2000,
052150).
Limited size of bioassay results in
sampling variability; lower bound is 95%
CI on administered exposure.
Choice of
bioassay
Alternatives could
increase or decrease
RfC/RfD.
NTP T2000. 196293)
study.
Alternative bioassays were inadequate.
97

-------
Consideration
Potential impact
Decision
Justification
Choice of species/
gender
RfC would be increased if
based on another
species/gender.
RfC is based on the most
sensitive endpoint,
increased liver
hemosiderin, in the most
sensitive species and
gender, male rats.
Male mice are the only species/gender to
experience an increase in liver tumors
from EGBE exposure. If one relies on the
MO A, however, liver hemosiderin increase
in male rats is the appropriate key event
and species to utilize for the RfC. Female
mice exhibit forestomach papilloma but at
higher doses.
Human relevance
of rodent tumor
data
If MO As for tumors from
EGBE were deemed not
relevant, cancer descriptor
would be "suggestive of
human carcinogenic
potential."
MO As for liver and
forestomach tumors are
used.
Although EGBE has some evidence for
liver tumors in male mice and forestomach
tumors in female mice, the MO As describe
rationales as to why humans are unlikely to
experience appreciable risk at
environmental concentrations.
Human
population
variability
Low-dose toxicity would
increase to an unknown
extent.
10-fold UF applied to
derive the RID and RfC
values.
10-fold UF is applied principally because
of limited data on human variability or
potential susceptible subpopulations.
5.3.1.	Choice of Endpoint
Comparison RfC values were calculated (see Section 5.1.4) and are intended to provide
information on alternative health effects associated with EGBE exposure. The comparison RfCs
ranged from 1 to 23 mg/m3, depending on whether irritation (forestomach), hematologic effects,
or hemosiderin deposition data were used to derive the POD, with the latter endpoint
representing the lower end of the RfC range.
5.3.2.	Choice of Dose Metric
The AUC for BAA in arterial blood was selected as the appropriate measure of dose due
to evidence for a causal association between BAA and hemolysis, between hemolysis and the
accumulation of hemosiderin in the liver, and between hemosiderin accumulation in the liver and
increased incidence of liver hemangiosarcoma. AUC is considered to be a more appropriate
response measure because hemosiderin pigmentation increases in incidence and severity with
increased duration (subchronic to chronic) and is believed to be the result of the cumulative
exposure to EGBE/BAA as opposed to a peak exposure. The corresponding aldehyde of BAA,
BAL, was also considered as a choice for an internal dose measure. As discussed in Section 4.6,
BAL is the EGBE metabolite considered to have the greatest potential to interact directly with
DNA. However, high ADH activity in the liver and forestomach is expected to result in very
short residence time and in very low tissue concentrations of BAL; this scenario has been
demonstrated in simulations using the Corley et al. (2005, 100098) PBPK model. Also, the
Corley et al. (2005, 100098) PBPK model along with the gavage study of Deisinger and
Boatman (2004, 594446) suggest that the conditions of in vitro assays showing BAL to be
clastogenic (e.g., no metabolic activation; high cytotoxic concentrations of BAL) are considered
to be of little relevance to the expected target organ (liver) environment (e.g., high metabolic
98

-------
activity; low concentrations of BAL). Use of an alternate measure of internal dose, for instance,
a parent compound or Cmax, would be more difficult to justify based on available empirical
information. However, for comparative purposes, a BMD analysis was done on the critical
endpoint to determine the impact that choosing Cmax of BAA in blood rather than AUC of BAA
in blood would have had on the BMCLi0(hec) derivation. If Cmax had been used as the dose
measure, the BMCLi0(hec) value would have been 109 mg/m3, approximately 6.8-fold higher
than the 16 mg/m3 BMCLi0(hec) value derived using AUC as the dose measure.
5.3.3.	Use of BMC Approach
Utilization of the BMC approach has advantages over other approaches to dose-response
analysis, such as the NOAEL/LOAEL approach. These advantages include the capacity of the
BMC approach to accommodate study sample size and reflect this in providing confidence
bounds to the lower limit on dose. As shown in Figures 5-1 and 5-2, use of the BMC approach
on the incidence of chronic hemosiderin deposition resulted in RfC and RfD values lower than
RfC and RfD values that would have been derived via the NOAEL/LOAEL approach.
5.3.4.	Choice of Model for BMCL Derivations
BMCLio estimates used in the derivation of the RfC, which formed the basis for both the
noncancer and cancer quantitative assessments, spanned a twofold range (60-124 |imol-hour/L)
for female rats and threefold range (40-126 (j,mol-hour/L) for male rats. All of the models fit
reasonably well (p-values above 0.1). Hence, this range of results can be considered a reflection,
in part, of model uncertainty.
5.3.5.	Choice of Animal to Human Extrapolation Method
A PBPK model (Corley et al., 1997, 041984) was used to extrapolate animal to human
concentration. An AUC blood level of BAA associated with a 10% increase in male mice with
hemosiderin pigmentation of 69.6 [j,mol-hour/L was estimated using the mouse PBPK model; the
human PBPK model was used to convert back to a human equivalent exposure concentration, or
a BMCLio(hec), of 12 mg/m3. If no PBPK models were available, the BMCLio(hec) would have
been derived by dividing the BMCLio for external exposure concentration of 75 mg/m3 by the
threefold pharmacokinetic portion of the animal to human default adjustment factor (U.S. EPA,
1994, 006488). resulting in a BMCLio(hec) of 25 mg/m3. This default value would have been
twofold higher than the value derived using the PBPK model.
5.3.6.	Route-to-Route Extrapolation
To estimate an oral dose POD for chronic hemosiderin deposition, a route-to-route
extrapolation was performed on the inhalation exposure POD used to derive the RfC using a
PBPK model and assuming, as discussed above, that the metric most closely associated with the
effect seen is the AUC measure of BAA in blood. One way of characterizing the uncertainty
99

-------
associated with this approach is to compare dose levels (BMCL values) using this metric for
hemolytic effects—assumed to be associated with chronic hemosiderin deposition—derived
from (1) an existing oral subchronic NTP (1993, 042063) study; or (2) from a model estimating
this metric from an existing inhalation subchronic NTP (1998, 594421; 2000, 196293) study.
This analysis was performed (data not shown) and the values derived using the second procedure
were slightly but consistently lower than those derived using the former, suggesting that
estimates using the procedure employed for route-to-route extrapolation would tend to
overestimate the toxicity value and result in lower RfD estimates.
5.3.7.	Statistical Uncertainty at the POD
Parameter uncertainty can be assessed through CIs. Each description of parameter
uncertainty assumes that the underlying model and associated assumptions are valid. For the
linear multistage model applied to the male mouse hemosiderin data, there is a reasonably small
degree of uncertainty at the 10% excess incidence level (the POD for derivation of the RfC),
with the 95%, single-sided lower confidence limit (BMCL) being about 30% below the
maximum likelihood estimate of the BMC.
5.3.8.	Choice of Bioassay
The NTP (2000, 196293) inhalation study was used for development of the RfC and RfD.
This was a well-designed study, conducted in both genders in two species with an adequate
number of animals per dose group and with examination of appropriate toxicological endpoints
in both genders of rats and mice. Alternative comparable bioassays were unavailable.
5.3.9.	Choice of Species/Gender
The RfC was based on the incidence of liver hemosiderin pigmentation in male rats, the
species and gender most sensitive to this effect (NTP, 2000, 196293). This event also occurs in
female rats and in mice, and is thought to be a precursor to the observed increase in liver tumors
in male mice. The cancer assessment was based on the more sensitive hemosiderin response in
the male rat, not the male mouse, because there is no evidence to suggest that the proposed MOA
for tumor formation would not be relevant to rats, and because the lack of an observed tumor
response in rats may have been due to the fact that rats were exposed to lower doses for a shorter
portion of their average lifespan (see further discussion in Sections 4.6 and 4.7.2).
If the RfC had been based on increased incidence of liver hemosiderin in another
species/gender, such as male mice, a higher RfC value would have been derived. Similarly, the
RfC would also have been higher had it been based on forestomach irritation, an effect less
sensitive than hemosiderin deposition but considered to be a precursor event leading to the
increase in incidence of forestomach squamous cell papillomas and one high-dose carcinoma
observed by NTP (2000, 196293) in female mice.
100

-------
5.3.10.	Human Relevance of Noncancer Responses Observed in Mice
The effects of hemosiderin deposition and forestomach irritation may both have
qualitative relevance to humans. However, for reasons discussed in Section 4.5, the exposure
concentrations that would be necessary to cause these effects in humans, if attainable at all, are
likely to be much higher than the RfC/RfD and well above concentrations necessary to cause
these effects in mice.
5.3.11.	Human Population Variability
The extent of interindividual variability associated with EGBE metabolism is not well
characterized in humans. As is discussed in Section 4.7, the hemolytic effect of EGBE is
presumed to be caused by the interaction of its primary metabolite, BAA, with the RBC
membrane. Potentially susceptible subpopulations or life stages would include individuals with
enhanced metabolism or decreased excretion of BAA. In addition, individuals whose RBC
membranes are more susceptible to the lysis caused by BAA could be more sensitive to EGBE.
However, RBCs from normal, aged, sickle-cell anemia, and hereditary spherocytosis patients
were no more sensitive to the hemolytic effects of BAA than RBCs from healthy volunteers
(Udden, 1994, 042109). As is discussed further in Section 4.7, some potentially susceptible
subpopulations or life stages have not been tested, and, when combined with the lack of
understanding about EGBE's mechanism of hemolytic action, this represents a considerable
source of uncertainty and forms the principal basis for the 10-fold UF applied to derive the RfD
and RfC values.
5.4. CANCER ASSESSMENT
In accordance with the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005,
086237). the method used to characterize and quantify cancer risk from a chemical depends on
what is known about the MOA of carcinogenicity and the shape of the cancer dose-response
curve for that chemical. An assumption of linearity is appropriate when evidence supports an
MOA of gene mutation due to DNA reactivity or an MOA that is anticipated to be linear. The
linear approach is used as a default option if the MOA of carcinogenicity is not known. A
nonlinear approach "can be used for cases with sufficient data to ascertain the MOA and to
conclude that it is not linear at low doses..." (U.S. EPA, 2005, 086237). Alternatively, the MOA
may theoretically have a threshold; that is, the carcinogenicity may be a secondary effect of
toxicity that is itself a threshold phenomenon.
In the case of EGBE, the MOA of carcinogenicity for hepatic hemangiosarcoma,
hepatocellular adenoma and carcinoma, and forestomach tumor formation in animals is
reasonably well understood. An RfC and RfD approach has been used for EGBE because
"When adequate data on MOA provide sufficient evidence to support a nonlinear MOA for the
general population and/or any subpopulations of concern, a different approach—a RfD/RfC that
101

-------
assumes nonlinearity—is used" (U.S. EPA, 2005, 086237). It is recognized, however, that while
this approach fits this case, other nonlinear approaches may be appropriate in other settings.
Available data indicate that EGBE is not likely to be mutagenic and that it is not expected to
produce rodent tumors via a mutagenic MOA. Rather, there is evidence that carcinogenic
responses observed in animals are associated with erythrocyte hemolysis (leading to oxidative
damage, increased hepatocyte and endothelial cell proliferation, and initiation for the liver
tumors) and with the formation of BAA, an acidic metabolite, that leads to regenerative
hyperplasia in response to irritation for the forestomach tumors. Because cytolethality occurs
only at exposure levels above some critical dose, a nonlinear (threshold) approach is considered
to be the most appropriate method for characterizing the cancer risk from EGBE.
As discussed in Sections 4.1 and 4.6, there are currently no human studies addressing the
potential carcinogenicity of EGBE. A 2-year inhalation bioassay in mice and rats (NTP, 2000,
196293) reported some evidence of carcinogenic activity in male B6C3Fi mice based on
increased incidences of hemangiosarcoma of the liver and an increase in the incidence of
hepatocellular carcinoma, as well as some evidence of carcinogenic activity in female B6C3Fi
mice based on increased incidences of forestomach squamous cell papilloma or carcinoma
(mainly papilloma). The study also reported no evidence of carcinogenic activity in male
F344/N rats and equivocal evidence of carcinogenic activity in female F344/N rats based on
increased combined incidences of benign and malignant pheochromocytoma (mainly benign) of
the adrenal medulla. The hypothesized MOA for the induction of hepatic hemangiosarcomas
and hepatocellular carcinomas reported in male mice exposed to EGBE may comprise a
sequence of events that are dose-dependent. First, EGBE is metabolized to the carboxylic acid,
which then hemolyzes RBCs. This event leads to the release of excess iron that in turn could
result in iron-induced formation of ROS and subsequent oxidative damage to target tissues
within the liver. Induction of cell proliferation and neoplasm formation follows. This sequence
of events is considered necessary for the formation of the observed neoplasms. Strategies
intended to control or omit any of these key events, including the initial hemolytic event, would
interrupt the process and prevent formation of neoplasms. For these reasons, formation of
neoplasms in humans would likely be prevented by establishing levels of EGBE exposure in the
dose range that does not result in the initial hemolytic events and subsequent ROS-mediated
cellular damage. Thus, for the assessment of human cancer risk associated with the formation of
hemangiosarcomas and hepatocellular adenomas and carcinomas in animals, the RfD and RfC
derived in Sections 5.1 and 5.2 should be considered protective, as the carcinogenicity may be a
secondary effect of toxicity that is itself a threshold phenomenon.
Available data suggest that the MOA for the induction of forestomach tumors reported in
female mice exposed to EGBE is dependent on the initial formation of the irritating metabolite
BAA. This acidic metabolite produces chronic cytotoxicity, which results in compensatory
epithelial cell regeneration. Chronic cell proliferation in preneoplastic cells is in turn associated
102

-------
with the formation of forestomach papillomas and carcinomas. In addition, the available data
indicate that BAA may be preferentially formed in the forestomach due to the presence of levels
of ADH that are higher than those found in the human stomach or esophagus. BMD modeling of
the dose response for epithelial hyperplasia in female mice forestomachs was performed as a
comparison to reference values derived for precursor effects in the liver as discussed above and
is presented in Appendix C. The analysis in Appendix C shows that, had hyperplastic effects in
female mice been used as a POD, the resultant RfD and RfC values would have been much
higher than the RfD and RfC values derived in Sections 5.1 and 5.2 using hemosiderin
deposition as the critical effect for EGBE exposure. Thus, the analysis indicates that the RfD
and RfC based on liver effects would also be protective of forestomach toxicity. The exposure
concentrations that would be necessary to cause these effects in humans, if attainable at all, are
likely to be much higher than the RfC and RfD.
Female rats reported a marginally significant trend for pheochromocytomas, and the high
dose frequencies reported in the rats were only slightly different from the upper range of
historical controls (see Section 4.6). In addition, the histopathologic data indicated that there
was difficulty distinguishing pheochromocytomas from nonneoplastic adrenal medullary
hyperplasia. Thus, these lesions are interpreted with caution as tumors. Given the marginal dose
response, lack of tumor evidence in any other organ system of the rats, and reported difficulties
in distinguishing pheochromocytomas from nonneoplastic adrenal medullary hyperplasia, this
tumor was not given significant weight in the qualitative or quantitative assessment of EGBE
cancer potential.
5.4.1.	Quantification for Oral and Inhalation Cancer Risk
Following the U.S. EPA (2005, 088823) Guidelines for Carcinogen Risk Assessment, a
nonlinear approach to dose-response assessment is taken for agents, such as EGBE, for which the
most plausible mode of action at low doses is consistent with nonlinearity. The RfC of 1.6
mg/m3 and RfD of 0.1 mg/kg-day derived in Sections 5.1 and 5.2 represent the outcome of
nonlinear assessments based on hemolytic effects (i.e., hemosiderin deposition)associated with
both oral and inhalation exposures to EGBE. Doses (or concentrations) of EGBE below the RfC
(or RfD) would not be expected to produce hemolytic effects (i.e., hemosiderin deposition) and
therefore are not expected to produce any increase in cancer risk.
5.4.2.	Uncertainties in Cancer Risk Assessment
The cancer assessment of EGBE is based largely on the premise that key events in the
MO As proposed for mice (forestomach irritation and hemolysis leading to increased hemosiderin
deposition) are not likely to occur in humans at concentrations at or below the RfC and RfD
values. Uncertainties in the RfC and RfD derivations are addressed in part in Sections 5.1.2,
5.1.3.	5.2.2, 5.2.3, and 5.3. This section will discuss additional uncertainties relative to the
103

-------
human relevance of cancer responses observed in mice and the extrapolation method employed
for the estimation of low-dose cancer risk. All of these uncertainties are summarized in
Table 5-8.
5.4.2.1. Choice of Low-Dose Extrapolation Method
The MOA is a consideration in estimating risks. A linear low-dose extrapolation
approach was not considered to be optimal for the estimation of human carcinogenic risk
associated with EGBE exposure because of the physicochemical properties of EGBE,
toxicokinetic limitations, specific MO As articulated earlier, and limitations in data to
parameterize appropriate models. It should be noted that the demonstration that a chemical is not
mutagenic is insufficient, alone, to postulate a nonlinear dose response. Key events in the MO As
proposed for mice (forestomach irritation and hemolysis leading to increased hemosiderin
deposition) are not likely to occur in humans at concentrations at or below the RfC and RfD
values. This assumes that the proposed MO As in mice are reasonably correct. If, for instance,
hemosiderin accumulation in the mouse liver is not the result of increased hemolysis but of
EGBE interaction with other cell types, a more direct, linear MOA for the observed increase in
male mouse hemangiosarcomas might be hypothesized. In order to illustrate the predicted
cancer risk under such a scenario, the cancer risks associated with the tumor types that were
increased following EGBE exposure were calculated using the default approach of low-dose
linear extrapolation outlined in the cancer guidelines (U.S. EPA, 2005, 086237). The results of
this linear analysis are presented in Table 5-9.
Table 5-9. Illustrative potency estimates for tumors in mice, using a linear
analysis approach

BMCLiohec
(mg/m3)a
Inhalation unit risk
0. 1/BMCLiohec
(per mg/m3)
Hepatocellular carcinoma (males)
208
4.8 x 10-4
Hemangiosarcomas (males)
575
1.7 x 10-4
Papilloma or carcinoma of the forestomach (females)
544
1.8 x 10-4
"BMDLiohec values were calculated using AUC as the dose metric.
For illustrative purposes, an estimate of the increased cancer risk at the RfC (1.6 mg/m3)
was calculated, using the inhalation unit risk for hepatocellular tumors in male mice, that was
derived using a default linear low-dose extrapolation approach. The estimated increased cancer
risk at the RfC in this comparison exercise would be 1.6 mg/m3 x 4.8 x 10"4 = 7.6 x 10"4 This
value is only for illustrative purposes and indicates the differences in the two approaches. It
should not be misconstrued as an estimate of the cancer risk at the RfC since the default linear
approach is not recommended.
104

-------
5.4.2.2. Human Relevance of Cancer Responses Observed in Mice
The hypothesized MO As for EGBE-induced liver tumors observed in male mice and
EGBE-induced forestomach tumors observed in female mice may both have qualitative
relevance to humans. However, for reasons discussed in Section 4.6, the exposure
concentrations that would be necessary to cause these effects in humans, if attainable at all, are
likely to be much higher than the RfC and RfD.
5.5. POTENTIAL IMPACT OF SELECT UNCERTAINTIES ON THE RFC
In this assessment, the RfC forms the basis for the RfD. The range of possible results
associated with some of the areas of uncertainty in the RfC derivation can be estimated (see
Sections 5.3 and 5.4). Figure 5-3 graphically illustrates the change in the RfC that would result
had particular choices, other than those presented in this assessment, been made (see summary in
Table 5-8). These specific areas were presented in this illustration because there are data
available that could be used to quantify their contribution to the uncertainty in the noncancer and
cancer assessments. The "Cancer approach" illustrates the dose that would represent a 10"6
increased cancer risk if a default linear low-dose extrapolation approach were used. This dose is
expressed as X-fold change in the RfC.
105

-------
Areas of Uncertainty - X-fold Change from Alternative Choices
a)
at
£
cs
£
O
O
£
2
$
X
20
15
10
5
0
-5
-10
-15
-50
-55
18
18
6.8
n ft n
56
1	Nonlinear—* Linear
2	Hemosiderin —~ Forestomach
SAUC^C^
4	Low BMDL —~ NOAEL
5	Low —~ High BMDL
6	PBPK —> Default
7	BMDL —> BMD
8	Rat—»Mouse
Cancer
Approach1
Endpoint2
Dose
Metric3
POD
Method4
BMD
Model5
HEC Benchmark
Derivation6 POD7
Species
Figure 5-3. Potential impact of select uncertainties on the RfC for EGBE.
The area of uncertainty that has the potential to have the greatest impact on the EGBE
assessment is that which is associated with the approach used to extrapolate human cancer risk
from the available rodent studies; that is, if human cancer risk had been estimated from a linear
extrapolation of responses observed in rodent studies. Figure 5-3 provides a graphic of the areas
of uncertainty, described in Table 5-8, for which there is quantitative information and impact on
the RfC can be estimated. The "Cancer Approach" value in Figure 5-3 represents the difference
in the dose that was selected for the RfC and the dose that would represent a 10"5 increased
cancer risk if a default linear low-dose extrapolation approach were to be used. In this instance,
the comparison indicated that the dose associated with a 10"6 cancer risk, estimated via the linear
dose-response method using the inhalation unit risk for hepatocellular tumors in male mice as
indicated in Table 5-9, is 56-fold lower than the RfC. Note that the default linear approach is not
recommended.
These discussions and illustrative tools should not be viewed as a comprehensive analysis
of all possible uncertainty considerations; rather, they exemplify what are believed to be
important areas of uncertainties for which data are available. This characterization is presented
106

-------
in an effort to make apparent the limitations of the assessment and to aid and guide the risk
assessor in the ensuing steps of the risk assessment process.
107

-------
6. MAJOR CONCLUSIONS IN Till CHARACTERIZATION OF HAZARD AND
DOSE RESPONSE
6.1. HUMAN HAZARD POTENTIAL
EGBE is a clear, miscible solvent used in formulating cleaning products and protective
coatings. It is metabolized primarily to the carboxylic acid, BAA, the proximate toxicant, in
both humans and animals. BAA and its conjugates are readily excreted in the urine.
Hemosiderin deposition in the liver of male rats, as a consequence of hemolysis, is
identified as the critical endpoint of concern in toxicological studies on EGBE. Toxicologically,
this effect increases with both exposure concentration and duration, is not compensated for (as
with hemolysis), and is considered to be a pathological finding. Mechanistically, this effect is at
the interface between noncancer and cancer effects from exposure to EGBE.
Observations regarding the potential relevance of EGBE toxicity to humans include the
insensitivity of human RBCs to the hemolytic effects of EGBE and its metabolite, BAA. While
it is established that humans can experience hemosiderin deposition in the liver as a consequence
of excessive hemolysis (Iancu et al., 1976, 100149). the relative insensitivity of human blood to
the effects of EGBE have been demonstrated in numerous in vitro studies through the use of
either whole blood or washed erythrocytes (e.g., Ghanayem and Sullivan, 1993, 041609; Udden,
2002, 042 111). Humans appear significantly less sensitive to the hemolytic toxicity of EGBE
than are typical laboratory species, such as mice, rats, or rabbits, with reports from analyses of
isolated RBCs demonstrating that human RBCs are 40- to 150-fold less sensitive than rat RBCs
(Udden, 2002, 042 111). These observations are inclusive of human RBCs from individuals with
hereditary spherocytosis and sickle cell anemia, conditions characterized by RBC sensitivity to
hemolysis. Available in vivo information with human exposure supports this species disparity in
sensitivity to the hemolytic effects of EGBE. Male rats in one study (NTP, 1993, 042063)
experienced mild liver effects at a drinking-water dose lower than that which caused observable
hemolytic effects. Available human toxicity data show that the primary effects after acute oral
ingestion of large doses of EGBE (most often combined with other solvents) are reversible
metabolic acidosis from the production of BAA and some hematological changes. Occupational
exposure to low levels of EGBE were not reported to cause changes outside the normal clinical
ranges in hepatic, renal, or hematologic parameters (Haufroid et al., 1997, 042040).
Due to the known reproductive toxicity (i.e., toxicity to male testes and sperm) of two
other glycol ethers, EGME and EGEE, the reproductive toxicity of EGBE has been studied in a
Note: Hyperlinks to the reference citations throughout this document will take you to the NCEA HERO database (Health and
Environmental Research Online) at http://epa.gov/hero. HERO is a database of scientific literature used by U.S. EPA in the process of
developing science assessments such as the Integrated Science Assessments (ISA) and the Integrated Risk Information System (IRIS).
108

-------
variety of well-conducted oral studies (Exon et al., 1991, 100112; Foster et al., 1987, 100116;
Grant et al., 1985, 006770; Heindel et al., 1990, 042042; Nagano et al., 1979, 100179; Nagano et
al., 1984, 100180; NTP, 1993, 042063) and inhalation studies (Dodd et al., 1983, 066465; Doe,
1984, 006754; Nachreiner, 1994, 081825; NTP, 2000, 196293) using rats, mice, and rabbits. In
addition, several developmental studies have addressed EGBE's toxicity from conception to
sexual maturity, including toxicity to the embryo and fetus, following oral exposures (Sleet et al.,
1989, 042089; Wier et al., 1987, 042123). inhalation exposures (Nelson et al., 1984, 031878; Tyl
et al., 1984, 100209). and dermal exposures (Hardin et al., 1984, 042030) to rats, mice, and
rabbits. EGBE did not cause significant effects in any reproductive organ, including testes, in
any study. In a two-generation reproductive toxicity study, fertility was reduced in mice only at
very high maternally toxic doses (>1,000 mg/kg). Maternal toxicity related to the hematologic
effects of EGBE and relatively minor developmental effects have been reported in
developmental studies. No teratogenic effects were noted in any of the studies. It can be
concluded from these studies that EGBE is not significantly toxic to the reproductive organs
(male or female) of either parents or to the developing fetuses of laboratory animals.
No reliable human epidemiological studies are available that address the potential
carcinogenicity of EGBE. The NTP (2000, 196293) performed a 2-year inhalation bioassay with
rats and mice and found no evidence of carcinogenic activity in male F344/N rats and equivocal
evidence of carcinogenic activity in female F344/N rats based on increased combined incidences
of benign and malignant pheochromocytoma (mainly benign) of the adrenal medulla. The
researchers reported some evidence of carcinogenic activity in male B6C3Fi mice based on an
increased incidence of hemangiosarcoma of the liver and an increase in the incidence of
hepatocellular carcinoma that may have been exposure related. They also reported some
evidence of carcinogenic activity in female B6C3Fi mice based on an increased incidence of
forestomach squamous cell papilloma or carcinoma (mainly papilloma). Based on its physical-
chemical properties, toxicokinetic and dynamic factors, and MOA information, under existing
U.S. EPA guidelines (U.S. EPA, 2005, 086237). EGBE is judged not likely to be carcinogenic to
humans at expected environmental concentrations (see Section 4.6). The MO As presented for
the animal tumors indicate that both high doses and sustained periods of exposure are necessary
for the carcinogenic response. The available human exposure/response information indicates
that these conditions are unlikely to occur because the primary response of humans to high oral
doses of EGBE, as shown in the case studies in Section 4.1 is metabolic acidosis, which, if not
treated, can lead to shock and eventually death. Further, based on simulations from PBPK
modeling, the maximum blood concentrations of BAA that could be produced in humans
following exposure to a saturated atmosphere of EGBE would be below those needed to produce
hemolysis (Corley et al., 2005, 100100). Evidence from the only human inhalation exposure
study available showed that, while nasal and ocular irritation were reported in research subjects
exposed to up to 195 ppm, no changes in erythrocyte fragility were observed (Carpenter et al.,
109

-------
1956, 066464). For a more complete discussion of the carcinogenic potential of EGBE, see
Section 4.6.
6.2. DOSE RESPONSE
6.2.1. Noncancer—Inhalation
Studies have not been reported in which humans were exposed sub chronically or
chronically via inhalation to EGBE. After consideration of the available animal inhalation
studies with EGBE, the NTP (2000, 196293) study was selected as the principal study because it
was conducted in two species and provides data for different durations and for multiple dose
groups compared to other available studies.
For derivation of the RfC, the most sensitive endpoint for dose-response assessment is the
effect of inhalation exposure on hemosiderin staining in Kupffer cells, as reported in the study by
NTP (2000, 196293). The concentration of BAA in the blood was used as an internal dose
metric for EGBE exposure. The human PBPK model of Corley et al. (1994, 041977; 1997,
041984) was used to obtain estimates of human inhalation exposure concentrations associated
with the BMCs derived from rat BAA AUC levels reported by Dill et al. (1998, 041981). The
RfC is based on the human equivalent BMCLio of 16 mg/m3, which was back-calculated from
rat data using the BMD and PBPK approach.
The use of uncertainty factors used to derive the RfC are as follows: a factor of 10 was
selected to account for the uncertainty associated with the variability of the human response
(UFh) to the effects of EGBE, a factor of 1 was selected to account for the uncertainty associated
with interspecies variability resulting from toxicodynamic and toxicokinetic differences between
animals and humans (UFA), and a factor of 1 was selected to account for deficiencies in the
database. A total UF of 10 (10 for UFH, 1 for UFA, and 1 for UFD) was used in the derivation of
the RfC. Thus, the RfC is 16 mg/m3 -h 10 = 1.6 mg/m3.
The overall confidence in the RfC is medium to high. A high degree of confidence is
placed in the RfC because it was derived from internal dose measures (PBPK method and
combined PBPK/BMC method) which account for pharmacokinetic differences between rats and
humans (Corley et al., 1997, 041984; Corley et al., 2005, 100100; Lee et al., 1998, 041983) and
it is based on actual measurements of internal blood concentrations in test animals of interest
(Dill et al., 1998, 041981). There is a high level of confidence in the NTP (2000, 196293) study
because it was a chronic study, it employed both male and female rats and mice, it had a wide
range of exposure levels, and animals were observed twice daily. There is medium-to-high
confidence. While the database lacks long-term human studies, the available short-term human
controlled studies and case reports, as well as laboratory animal and in vitro studies provide
evidence to suggest that, with respect to the hemolytic effects of EGBE, long-term human
exposures would be no more adverse than long-term rat exposures. Additionally, the selection of
110

-------
medium-high confidence is supported because the potential for effects in humans from repeated,
long-term exposures has not been investigated.
6.2.2. Noncancer—Oral
Studies have not been reported in which humans have been exposed sub chronically or
chronically to EGBE by the oral route of exposure, and thus would be suitable for derivation of
an oral RfD. No chronic oral laboratory animal studies are currently available for EGBE.
Hematological effects leading to RBC lysis and organ accumulation of iron in the form of
hemosiderin accumulation appear to be the most sensitive of the effects caused by EGBE in
laboratory animals. Based on the limited oral database and because the critical endpoint,
hemosiderin pigmentation, was more pronounced in the chronic inhalation study (NTP, 2000,
196293). EPA concluded that the use of a route to route extrapolation from the NTP, 2000
(196293) study would increase the confidence in the derived RfD (see Section 5.2.1). An RfD
has been derived via the application of a PBPK model using hemosiderin pigmentation observed
in the NTP (2000, 196293) chronic inhalation study of EGBE. As with the animal to human
extrapolation used in the development of the RfC, the dose metric used for interspecies (rat to
human) and route-to-route (inhalation to oral) extrapolation was the AUC of BAA at 12 months
in arterial blood. This dose metric was used for dose-response modeling of chronic inhalation
data (Section 5.1.2) to derive the POD of 133 [j,mol-hour/L, expressed as a BMDLio. The
BMDLio was then back-calculated using the human PBPK model (Corley et al., 1994, 041977;
Corley et al., 1997, 041984) to obtain an equivalent human oral drinking water dose (BMDLred)
of 1.4 mg/kg-day.
The use of uncertainty factors used to derive the RfD are as follows: a factor of 10 was
selected to account for the uncertainty associated with the variability of the human response
(UFh) to the effects of EGBE, a factor of 1 was selected to account for the uncertainty associated
with interspecies variability resulting from toxicodynamic and toxicokinetic differences between
animals and humans (UFA), and a factor of 1 was selected to account for deficiencies in the
database. A total UF of 10 (10 for UFH, 1 for UFA, and 1 for UFD) was used in the derivation of
the RfD. Thus, the RfD is 1.4 mg/kg-day + 10 = 0.1 mg/kg-day.
The overall confidence in the RfD is medium to high. The RfD has been calculated for
EGBE using a route-to-route extrapolation from the inhalation PBPK/BMC method used to
derive the RfC. High confidence is placed in the RfD since pharmacokinetic differences
between rats and humans were accounted for using a validated PBPK model (Corley et al., 1994,
041977; Corley et al., 1997, 041984). There is high confidence in the NTP (2000, 196293) study
because it was a chronic study, it employed both male and female rats and mice, it had a wide
range of exposure levels, and animals were observed twice daily. There is medium-to-high
confidence in the database, because data are available for a variety of animal species, including
humans. While the database lacks long-term human studies, the available short-term human
111

-------
controlled studies, case reports, laboratory animal, and in vitro studies provide evidence to
suggest that, with respect to the hemolytic effects of EGBE, long-term human exposures would
be no more adverse than long-term rat exposures. Additionally, the selection of medium-high
confidence is supported because the potential for effects in humans from repeated, long-term
exposures has not been investigated.
6.2.3. Cancer—Oral and Inhalation
Information regarding the reported liver and forestomach tumors observed in laboratory
animals exposed to EGBE indicates that the MO As underlying these lesions are nonmutagenic
and include intermediate processes that have nonlinear dose-response characteristics. Control or
omission of these intermediate events would likely be sufficient to prevent the occurrence of
such tumors in humans, including potentially sensitive subpopulations such as children.
Application of nonlinear quantitative assessment methods indicate that the noncancer
RfD (0.1 mg/kg-day) and RfC (1.6 mg/m3) values developed for EGBE are considered protective
of these key events and would serve to prevent the occurrence of carcinogenic effects in
humans.1 The exposure concentrations that would be necessary to cause these effects in humans,
if attainable at all, are likely to be much higher than the RfC and RfD.
The estimation of uncertainty in this analysis is based on the alternative approaches for
estimating the dose response that are discussed and shown in Section 5.1.4 and Appendix C.
These alternatives include using the NOAEL approach combined with the measure of the
internal dose estimated with the PBPK model and the BMC approach combined with the same
PBPK model. Sections 5.1.3 and 5.2.3 summarize the uncertainty associated with each of these
approaches for the RfC and RfD, respectively. Other uncertainties associated with the noncancer
and cancer assessments presented in this Toxicological Review are summarized in Table 5-8 and
in Sections 5.4.1 and 5.5.
1 These analyses are consistent with the nonlinear assessment approach described in the 2005 cancer guidelines (U.S. EPA, 2005a)
112

-------
7. REFERENCES
Allen DJ (1993). Ethylene glycol monobutyl ether: Acute dermal toxicity (limit test) in the rat. Project No.
13/540. Safepharm Laboratories, Ltd for Mitsubishi Petrochemical Co., Ltd.. Derby, UK and Tokyo,
Japan. Project No. 13/540. 594452
Allen DJ (1993). Ethylene glycol monobutyl ether: Acute dermal toxicity (limit test) in the rat. Project No.
13/542. Safepharm Laboratories, Ltd for Mitsubishi Petrochemical Co., Ltd.. Derby, UK and Tokyo,
Japan. Project No. 13/542. 594455
Allen DJ (1993). Ethylene glycol monobutyl ether: Acute dermal toxicity test in the rabbit. Project No. 13/605.
Safepharm Laboratories, Ltd. for Mitsubishi Petrochemical Co., Ltd.. Derby, UK and Tokyo, Japan.
Project No. 13/605. 594459
Allen DJ (1993). Ethylene glycol monobutyl ether: acute dermal toxicity test in the rabbit. Project No. 13/606.
Safepharm Laboratories, Ltd. for Mitsubishi Petrochemical Co., Ltd.. Derby, UK and Tokyo, Japan.
Project No. 13/606. 594460
Bachowski S; Kolaja KL; Xu Y; Ketcham CA; Stevenson DE; Walborg EF Jr; Klaunig JE (1997). Role of
oxidative stress in the mechanism of dieldrin's hepatotoxicity. Ann Clin Lab Sci, 27: 196-209. 042161
Bartnik FG; Reddy AK; Klecak G; Zimmermann V; Hostynek JJ; Kunstler K (1987). Percutaneous absorption,
metabolism, and hemolytic activity of n-butoxyethanol. Fundam Appl Toxicol, 8: 59-70. 100086
Bauer P; Weber M; Mur JM; Protois JC; Bollaert PE; Condi A; Larcan A; Lambert H (1992). Transient non-
cardiogenic pulmonary edema following massive ingestion of ethylene glycol butyl ether. Intensive
Care Med, 18: 250-251. 100087
Bennette D (2001). The distribution of radioactivity in the female B6C3fl mouse following a single
intravenous injection of 2-butoxy[l-14c]ethanol. Central Toxicology Laboratory. Alderley Park,
Macclesfield, Cheshire, UK. CTL/R/1446. 041810
Berliner N; Duffy TP; Abelson HT (1995). Approach to the adult and child with anemia. In Hoffman R
(Ed.),Hematology: basic principles and practice (pp. 468-483). New York, NY and Edinburgh, UK:
Churchill Livingstone. 001232
Bevers EM; Comfurius P; van Rijn JL; Hemker HC; Zwaal RF (1982). Generation of prothrombin-converting
activity and the exposure of phosphatidylserine at the outer surface of platelets. Eur J Biochem, 122:
429-436. 100088
Browning J; Gannon BJ; O'Brien P (1983). The microvasculature and gastric luminal pH of the forestomach of
the rat: a comparison with the glandular stomach. Int J Microcirc Clin Exp, 2: 109-118. 041615
Buckley LA; Morgan KT; Swenberg JA; James RA; Hamm TE Jr; Barrow CS (1985). The toxicity of
dimethylamine in F-344 rats and B6C3F1 mice following a 1-year inhalation exposure. Fundam Appl
Toxicol, 5: 341-352. 062668
Bueld JE; Netter KJ (1993). Factors affecting the distribution of ingested propionic acid in the rat forestomach.
Food Chem Toxicol, 31: 169-176. 041617
Burkhart KK; Donovan JW (1998). Hemodialysis following butoxyethanol ingestion. Clin Toxicol, 36: 723-
725. 056375
Carpenter CP; Pozzani UC; Weil CS; Nair JH III; Keck GA; Smyth HF Jr (1956). The toxicity of butyl
cellosolve solvent. AMAArch Ind Health, 14: 114-131. 066464
Chiewchanwit T; Au WW (1995). Mutagenicity and cytotoxicity of 2-butoxyethanol and its metabolite, 2-
butoxyacetaldehyde, in Chinese hamster ovary (CHO-AS52) cells. Mutat Res, 334: 341-346. 041999
Note: Hyperlinks to the reference citations throughout this document will take you to the NCEA HERO database (Health and
Environmental Research Online) at http://epa.gov/hero. HERO is a database of scientific literature used by U.S. EPA in the process of
developing science assessments such as the Integrated Science Assessments (ISA) and the Integrated Risk Information System (IRIS).
113

-------
Clayson DB; Iverson F; Nera EA; Lok E (1991). Early indicators of potential neoplasia produced in the rat
forestomach by nongenotoxic agents: The importance of induced cellular proliferation. Mutat Res, 248:
321-331. 006215
Cohen AD; Cagnano E; Vardy DA (2001). Cherry angiomas associated with exposure to bromides.
Dermatology, 202: 52-53. 100096
Commonwealth of Australia (1996). National Industrial Chemicals Notification and Assessment Scheme
(NICNAS)-priority existing chemical no. 6-2-butoxyethanol in cleaning products. Australian
Government Publishing Service. Canberra, Australia. 594448
Connor J; Bucana C; Fidler IJ; Schroit AJ (1989). Differentiation-dependent expression of phosphatidylserine
in mammalian plasma membranes: quantitative assessment of outer-leaflet lipid by prothrombinase
complex formation. PNAS, 86: 3184-3188. 100097
Corley RA; Bartels MJ; Carney EW; Weitz KK; Soelberg JJ; Gies RA; Thrall KD (2005). Development of a
physiologically based pharmacokinetic model for ethylene glycol and its metabolite, glycolic acid, in
rats and humans. Toxicol Sci, 85: 476-490. 100098
Corley RA; Bormett GA; Ghanayem BI (1994). Physiologically-based pharmacolinetics of 2-butoxyethanol
and its major metabolite 2-butoxyacetic acid, in rats and humans. Toxicol Appl Pharmacol, 129: 61-79.
041977
Corley RA; Grant DM; Farris E; Weitz KK; Soelberg JJ; Thrall KD; Poet TS (2005). Determination of age and
gender differences in biochemical processes affecting the disposition of 2-butoxyethanol and its
metabolites in mice and rats to improve PBPK modeling. Toxicol Lett, 156: 127-161. 100100
Corley RA; Markham DA; Banks C; Delorme P; Masterman A; Houle JM (1997). Physiologically based
pharmacokinetics and the dermal absorption of 2-butoxyethanol vapor by humans. Fundam Appl
Toxicol, 39: 120-130. 041984
Corley RA; Weitz KK; Mast TJ; Miller RA; Thrall BD (1999). Short-term studies to evaluate the dosimetry
and modes of action of EGBE in B6C3F1 mice [final report]. Battelle Memorial Institute. Richland,
WA. Battelle Project No. 29753. 042003
Corthals SM; Kamendulis LM; Klaunig JE (2006). Mechanisms of 2-butoxyethanol-induced
hemangiosarcomas. Toxicol Sci, 92: 378-386. 100102
Dean BS; Krenzelok EP (1991). Critical evaluation of pediatric ethylene glycol monobutyl ether poisonings.
Vet Hum Toxicol, 33: 362. 597279
Deisinger PJ; Boatman RJ (2004). In vivo metabolism and kinetics of ethylene glycol monobutyl ether and its
metabolites, 2 butoxyacetaldehyde (BAL) and 2 butoxyacetic acid (BAA), as measured in blood, liver
and forestomach homogenates from mice. Eastman Kodak Company for The American Chemistry
Council. Rochester, NY and Arlington, VA. 594446
DeLeve LD (1998). Glutathione defense in non-parenchymal cells. Semin Liver Dis, 18: 403-413. 100104
Dill JA; Lee KM; Bates DJ; Anderson DJ; Johnson RE; Chou BJ; Burka LT; Roycroft JH (1998).
Toxicokinetics of inhaled 2-butoxyethanol and its major metabolite, 2-butoxyacetic acid, in F344 rats
and B6C3F1 mice. Toxicol Appl Pharmacol, 153: 227-242. 041981
Djordjevic VB (2004). Free radicals in cell biology. Int Rev Cytol, 237: 57-89. 100106
Dodd DE; Snellings WM; Maronpot RR; Ballantyne B (1983). Ethylene glycol monobutyl ether: acute, 9-day,
and 90-day vapor inhalation studies in Fischer 344 rats. Toxicol Appl Pharmacol, 68: 405-414. 066465
Doe JE (1984). Further studies on the toxicology of the glycol ethers with emphasis on rapid screening and
hazard assessment. Environ Health Perspect, 57: 199-206. 006754
ECETOC (1994). Butoxyethanol criteria document. Special Report No. 7. European Centre for Ecotoxicology
and Toxicology of Chemicals. Brussels, Belgium. 594447
Edmonson HA; Peters RL (1985). Liver. In JM Kissane (Ed.),Anderson's pathology (pp. 1096-1213). St. Louis,
MO: C.V. Mosby Company. 042168
Elias Z; Daniere MC; Marande AM; Poirot O; Terzetti F; Schneider O (1996). Genotoxic and/or epigenetic
effects of some glycol ethers: Results of different short-term tests. Occup Hyg, 2: 187-212. 042011
114

-------
Elliott BM; Ashby J (1997). Review of the genotoxicity of 2-butoxyethanol. Mutat Res, 387: 89-96. 100111
Exon JH; Mather GG; Bussiere JL; Olson DP; Talcott PA (1991). Effects of subchronic exposure of rats to 2-
methoxyethanol or 2-butoxyethanol: thymic atrophy and immunotoxicity. Fundam Appl Toxicol, 16:
830-840. 100112
Ezov N; Levin-Harrus T; Mittelman M; Redlich M; Shabat S; Ward SM; Peddada S; Nyska M; Yedgar S;
Nyska A (2002). A chemically induced rat model of hemolysis with disseminated thrombosis.
Cardiovasc Toxicol, 2: 181-194. 100113
Firooz A; Komeili A; Dowlati Y (1999). Eruptive melanocytic nevi and cherry angiomas secondary to
exposure to sulfur mustard gas. J Am Acad Dermatol, 40: 646-647. 100115
Foster JR (2000). 2-Butoxyethanol (ethylene glycol butyl ether, EGBE): the proposed mode of action in the
induction of hemangiosarcomas in male mouse liver and its significance for man. Central Toxicology
Laboratory. Cheshire, United Kingdom. Ctl/r/1465. 042172
Foster PM; Lloyd SC; Blackburn DM (1987). Comparison of the in vivo and in vitro testicular effects
produced by methoxy-, ethoxy- and N-butoxy acetic acids in the rat. Toxicology, 43: 17-30. 100116
Franks SJ; Spendiff MK; Cocker J; Loizou GD (2006). Physiologically based pharmacokinetic modelling of
human exposure to 2-butoxyethanol. Toxicol Lett, 162: 164-173. 100117
Frei YF; Perk K; Danon D (1963). Correlation between osmotic resistance and fetal hemoglobin in bovine
erythrocytes. Exp Cell Res, 30: 561-568. 100118
Frith CH; Ward JM (1980). A morphologic classification of proliferative and neoplastic hepatic lesions in
mice. J Environ Pathol Toxicol Oncol, 3: 329-351. 024775
Ghanayem BI (1989). Metabolic and cellular basis of 2-butoxyethanol-induced hemolytic anemia in rats and
assessment of human risk in vitro. Biochem Pharmacol, 38: 1679-1684. 042014
Ghanayem BI Maronpot RR Matthews HB (1986). Association of chemically induced forestomach cell
proliferation and carcinogenesis. Cancer Lett, 32: 271-278. 011728
Ghanayem BI; Blair PC; Thompson MB; Maronpot RR; Matthews HB (1987). Effect of age on the toxicity
and metabolism of ethylene glycol monobutyl ether (2-butoxyethanol) in rats. Toxicol Appl Pharmacol,
91:222-234. 066470
Ghanayem BI; Burka LT; Matthews HB (1987). Metabolic basis of ethylene glycol monobutyl ether (2-
butoxyethanol) toxicity: role of alcohol and aldehyde dehydrogenases. J Pharmacol Exp Ther, 242: 222-
231.041608
Ghanayem BI; Burka LT; Sanders JM; Matthews HB (1987). Metabolism and disposition of ethylene glycol
monobutyl ether (2-butoxyethanol) in rats. Drug Metab Dispos, 15: 478-484. 041645
Ghanayem BI; Long PH; Ward SM; Chanas B; Nyska M; Nyska A (2001). Hemolytic anemia, thrombosis, and
infarction in male and female F344 rats following gavage exposure to 2-butoxyethanol. Exp Toxicol
Pathol, 53: 97-105. 042755
Ghanayem BI; Maronpot RR; Matthews HB (1985). Ethyl acrylate-induced gastric toxicity II Structure-
toxicity relationships and mechanism. Toxicol Appl Pharmacol, 80: 336-344. 042768
Ghanayem BI; Sanchez IM; Maronpot RR; Elwell MR; Matthews HB (1993). Relationship between the time
of sustained ethyl acrylate forestomach cell proliferation and carcinogenicity. Environ Health Perspect,
101 (Suppl. 5): 277-279. 597328
Ghanayem BI; Sanchez IM; Matthews HB (1992). Development of tolerance to 2-butoxyethanol-induced
hemolytic anemia and studies to elucidate the underlying mechanisms. Toxicol Appl Pharmacol, 112:
198-206. 100128
Ghanayem BI; Sanchez IM; Matthews HB; Elwell MR (1994). Demonstration of a temporal relationship
between ethyl acrylate-induced forestomach cell proliferation and carcinogenicity. Toxicol Pathol, 22:
497-509. 042760
Ghanayem BI; Sanders JM; Clark A-M; Bailer J; Matthews HB (1990). Effects of dose, age, inhibition of
metabolism and elimination on the toxicokinetics of 2-butoxyethanol and its metabolites. J Pharmacol
Exp Ther, 253: 136-143. 042016
115

-------
Ghanayem BI; Sullivan CA (1993). Assessment of the haemolytic activity of 2-butoxyethanol and its major
metabolite, butoxyacetic acid, in various mammals including humans. Hum Exp Toxicol, 12: 305-311.
041609
Ghanayem BI; Ward SM; Chanas B; Nyska A (2000). Comparison of the acute hematotoxicity of 2-
butoxyethanol in male and female F344 rats. Hum Exp Toxicol, 19: 185-192. 042748
Gift JS (2005). U.S. EPA's IRIS assessment of 2-butoxyethanol: the relationship of noncancer to cancer effects.
Toxicol Lett, 156: 163-178. 100133
Gijsenbergh FP; Jenco M; Veulemans H; Groeseneken D; Verberckmoes R; Delooz HH (1989). Acute
butylglycol intoxication: a case report. Hum Toxicol, 8: 243-245. 100134
Gingell R; Boatman RJ; Lewis S (1998). Acute toxicity of ethylene glycol mono-n-butyl ether in the guinea
pig. Food Chem Toxicol, 36: 825-829. 100135
Gold LS; Manley NB; Slone TH; Farfinkel GB; Rohrbach L; Ames BN (1993). The fifth plot of the
carcinogenic potency database: results of animal bioassays published in the general literature through
1988 and by the National Toxicology Program through 1989. Environ Health Perspect, 100: 65-135.
056382
Gollapudi BB; Barber ED; Lawlor TE; Lewis SA (1996). Re-examination of the mutagenicity of ethylene
glycol monobutyl ether to Salmonella tester strain TA97a. Mutat Res, 370: 61-64. 100137
Grant D; Sulsh S; Jones HB; Gangolli SD; Butler WH (1985). Acute toxicity and recovery in the hemopoietic
system of rats after treatment with ethylene glycol monomethyl and monobutyl ethers. Toxicol Appl
Pharmacol, 77: 187-200. 006770
Greaves P (2000). Hepatocellular hypertrophy and hyperplasia. In Histopathology of preclinical toxicity
studies: interpretation and relevance in drug safety evaluation (pp. 445-448). New York, NY: Elsevier.
089176
Green T; Toghill A; Lee R; Moore R; Foster J (2002). The development of forestomach tumors in the mouse
following exposure to 2-butoxyethanol by inhalation Studies on the mode of action and relevance to
humans. Toxicology, 180: 257-273. 041610
Gualtieri JF; DeBoer L; Harris CR; Corley R (2003). Repeated ingestion of 2-butoxyethanol: case report and
literature review. J Toxicol Clin Toxicol, 41: 57-62. 100140
Gualtieri JF; Harris CR; Roy R; Corley RA; Manderfield C (1995). Multiple 2-butoxyethanol intoxications in
the same patient:Clinical findings, pharmacokinetics, and therapy. J Toxicol Clin Toxicol, 33: 550-551.
594443
Hardin BD; Goad PT; Burg JR (1984). Developmental toxicity of four glycol ethers applied cutaneously to
rats. Environ Health Perspect, 57: 69-74. 042030
Harrison PTC (1992). Propionic acid and the phenomenon of rodent forestomach tumorigenesis: a review.
Food Chem Toxicol, 30: 333-340. 041621
Harrison PTC; Grasso P; Badescu V (1991). Early changes in the forestomach of rats, mice and hamsters
exposed to dietary propionic and butyric acid. Food Chem Toxicol, 29: 367-371. 041622
Harrison SA; Bacon BR (2005). Relation of hemochromatosis with hepatocellular carcinoma: epidemiology,
natural history, pathophysiology, screening, treatment, and prevention. Med Clin North Am, 89: 391-
409. 100144
Haufroid V; Thirion F; Mertens P; Buchet J-P; Lison D (1997). Biological monitoring of workers exposed to
low levels of 2-butoxyethanol. Int Arch Occup Environ Health, 70: 232-236. 042040
Heindel JJ; Gulati DK; Russell VS; Reel JR; Lawton AD; Lamb JC IV (1990). Assessment of ethylene glycol
monobutyl and monophenyl ether reproductive toxicity using a continuous breeding protocol in Swiss
CD-I Mice. Fundam Appl Toxicol, 15: 683-696. 042042
Hoflack JC; Lambolez L; Elias Z; Vasseur P (1995). Mutagenicity of ethylene glycol ethers and of their
metabolites in Salmonella typhimurium his. Mutat Res, 341: 281-287. 100147
Hord JD; Lukens JN (1999). Anemia unique to infants and young children. In Wintrobe's clinical hematology
(pp. 1518-1537). Baltimore, MD: Williams & Wilkins. 594445
116

-------
Houglum K; Ramm GA; Crawford DHG; Witztum JL; Powell LW; Chojkier M (1997). Excess iron induces
hepatic oxidative stress and transforming growth factor "beta"l in genetic hemochromatosis.
Hepatology, 26: 605-610. 024777
HSDB (2005). Ethylene glycol mono-n-butyl ether(CASRN: 111-76-2). Retrieved 22-MAR-10, from
http://toxnet.nlm.nih.gOv/cgi-bin/sis/search/f7./temp/~lwRTkd: 1. 597537
Iancu TC; Neustein HB; Landing BH (1976). The liver in thalassaemia major: ultrastructural observations. In
Iron metabolism (pp. 293-316). The Netherlands: Ciba Foundation Symposium. 100149
IARC (2006). IARC monographs on the evaluation of carcinogenic risks to humans: Preamble. International
Agency for Research on Cancer, World Health Organization. Lyon, France
.http://monographs.iarc.fr/ENG/Preamble/CurrentPreamble.pdf. 093206
Jacobs GA; Martens MA (1989). An objective method for the evaluation of eye irritation in vivo. Food Chem
Toxicol, 27: 255-258. 100150
Jakasa I; Mohammadi N; Kruse J; Kezic S (2004). Percutaneous absorption of neat and aqueous solutions of 2-
butoxyethanol in volunteers. Int Arch Occup Environ Health, 77: 79-84. 100151
Jepson GW; Hoover DK; Black RK; McCafferty JD; Mahle DA; Gearhart JM (1994). A partition coefficient
determination method for nonvolatile chemicals in biological tissues. Fundam Appl Toxicol, 22: 519-
524. 006570
Johanson G (1986). Physiologically based pharmacokinetic modeling of inhaled 2-butoxyethanol in man.
Toxicol Lett, 34: 23-31. 006758
Johanson G (1994). Inhalation toxicokinetics of butoxyethanol and its metabolite butoxyacetic acid in the male
Sprague-Dawley rat. Arch Toxicol, 68: 588-594. 069322
Johanson G; Boman A (1991). Percutaneous absorption of 2-butoxyethanol vapour in human subjects. Occup
Environ Med, 48: 788-792. 006759
Johanson G; Boman A; Dynesius B (1988). Percutaneous absorption of 2-butoxyethanol in man. Scand J Work
Environ Health, 14: 101-109. 100156
Johanson G; Dynesius B (1988). Liquid/air partition coefficients of six commonly used glycol ethers. Br J Ind
Med, 45: 561-564. 100157
Johanson G; Johnsson S (1991). Gas chromatographic determination of butoxyacetic acid in human blood after
exposure to 2-butoxyethanol. Arch Toxicol, 65: 433-435. 100158
Johanson G; Wallen M; Byfalt Nordqvist M (1986). Elimination kinetics of 2-butoxyethanol in the perfused rat
liver—dose dependence and effect of ethanol. Toxicol Appl Pharmacol, 83: 315-320. 100159
Jones K; Cocker J (2003). A human exposure study to investigate biological monitoring methods for 2-
butoxyethanol. Biomarkers, 8: 360-370. 056771
Jones K; Cocker J; Dodd L; Fraser I (2003). Factors affecting the extent of dermal absorption of solvent
vapours: a human volunteer study. Ann Occup Hyg, 47: 145. 100161
Junge B; Carrion Y; Bosco C; Galleano M; Puntarulo S; Tapia G; Videla L (2001). Effects of iron overload and
lindane intoxication in relation to oxidative stress, Kupffer cell function, and liver injury in the rat.
Toxicol Appl Pharmacol, 170: 23-28. 042180
Kamendulis LM; Park JJ; Klaunig JE (1999). Potential mechanisms of rodent liver toxicity by 2-
butoxyethanol: oxidative stress studies. Indiana University School of Medicine, for The Ethylene Glycol
Ethers Panel; Chemical Manufacturers Association. Indianapolis, IN and Arlington, VA. 042182
Keith G; Coulais C; Edorh A; Bottin MC; Rihn B (1996). Ethylene glycol monobutyl ether has neither
epigenetic nor genotoxic effects in acute treated rats and in subchronic treated v-HA-ras transgenic
mice. Occup Hyg, 2: 237-249. 041625
Kennah HE II; Hignet S; Laux PE; Dorko JD; Barrow CS (1989). An objective procedure for quantitating eye
irritation based upon changes of corneal thickness. Fundam Appl Toxicol, 12: 258-268. 054156
Klaunig JE; Kamendulis LM (2004). Effect of 2-butoxyacetaldehyde on the induction of DNA damage (comet)
in rodent endothelial cells. American Chemistry Council. Arlington, VA. 594442
117

-------
Klaunig JE; Kamendulis LM (2005). Mode of action of butoxyethanol-induced mouse liver hemangiosarcomas
and hepatocellular carcinomas. Toxicol Lett, 156: 107-115. 100165
Klaunig JE; Xu Y; Isenberg JS; Bachowski S; Kolaja KL; Jiang J; Stevenson DE; Walborg EF Jr (1998). The
role of oxidative stress in chemical carcinogenesis. Environ Health Perspect, 106: 289-295. 024780
Knutson M; Wessling-Resnick M (2003). Iron metabolism in the reticuloendothelial system. Crit Rev Biochem
Mol Biol, 38: 61-88. 094393
Koshkaryev A; Barshtein G; Nyska A; Ezov N; Levin-Harrus T; Shabat S; Nyska M; Redlich M; Tsipis F;
Yedgar S (2003). 2-Butoxyethanol enhances the adherence of red blood cells. Arch Toxicol, 77: 465-
469. 100168
Krasavage WJ (1986). Subchronic oral toxicity of ethylene glycol monobutyl ether in male rats. Fundam Appl
Toxicol, 6: 349-355. 066474
Kroes R; Wester PW (1986). Forestomach carcinogens: possible mechanisms of action. Food Chem Toxicol,
24: 1083-1089. 041626
Kumagai S; Oda H; Matsunaga I; Kosaka H; Akasaka S (1999). Uptake of 10 polar organic solvents during
short-term respiration. Toxicol Sci, 48: 255-263. 100171
Kwittken J; Tartow LR (1966). Hemochromatosis and kupffer-cell sarcoma with unusual localization of iron. J
Pathol Bacteriol, 92: 571-573. 597334
Laitinen J (1998). Correspondence between occupational exposure limit and biological action level values for
alkoxyethanols and their acetates. Int Arch Occup Environ Health, 71: 117-124. 100173
Laitinen J; Liesivuori J; Savolainen H (1998). Urinary NAG and GAG as biomarkers of renal effects in
exposure to 2-alkoxyalcohols and their acetates. J Occup Environ Med, 40: 595-600. 100174
Lee KM; Dill JA; Chou BJ; Roycroft JH (1998). Physiologically based pharmacokinetic model for chronic
inhalation of 2-butoxyethanol. Toxicol Appl Pharmacol, 153: 211-226. 041983
Lesgards J-F; Durand P; Lassarre M; Stacker P; Lesgards G; Lanteaume A; Prost M; Lehucher-Michel M-P
(2002). Assessment of lifestyle effects on the overall antioxidant capacity of health subjects. Environ
Health Perspect, 110: 479-486. 042183
Lewis AE (1970). Principles of hematology. New York, NY: Appleton Century Crofts. 594441
Long PH; Maronpot RR; Ghanayem BI; Roycroft JH; Nyska A (2000). Dental pulp infarction in female rats
following inhalation exposure to 2-butoxyethanol. Toxicol Pathol, 28: 246-252. 042753
Medinsky MA; Singh G; Bechtold WE; Bond JA; Sabourin PJ; Birnbaum LS; Henderson RF (1990).
Disposition of three glycol ethers administered in drinking water to male F344/N rats. Toxicol Appl
Pharmacol, 102: 443-455. 041986
Muller A; Jacobsen H; Healy E; McMickan S; Istace F; Blaude MN, Howden P; Fleig H; Schulte A; EU
Working Group on Haemolytic Anaemia (2006). Hazard classification of chemicals inducing
haemolytic anaemia: An EU regulatory perspective . Regul Toxicol Pharmacol, 45: 229-241. 597291
Nachreiner DJ (1994). Ethylene glycol butyl ether: acute vapor inhalation toxicity study in guinea pigs. Union
Carbide Corporation for Chemical Manufacturers Association. Export, PA and Washington, DC.
94N1392. 081825
Nagano K; Nakayama E; Koyano M; Oobayashi H; Adachi H; Yamada T (1979). [Testicular atrophy of mice
induced by ethylene glycol mono alkyl ethers (author's transl)]. Sangyo Igaku, 21: 29-35. 100179
Nagano K; Nakayama E; Oobayashi H; Nishizawa T; Okuda H; Yamazaki K (1984). Experimental studies on
toxicity of ethylene glycol alkyl ethers in Japan. Environ Health Perspect, 57: 75-84. 100180
Nelson BK; Setzer JV; Brightwell WS; Mathinos PR; Kuczuk MH; Weaver TE; Goad PT (1984). Comparative
inhalation teratogenicity of four glycol ether solvents and an amino derivative in rats. Environ Health
Perspect, 57: 261-271. 031878
Nikula KJ; Novak RF; Chang IY; Dahl AR; Kracko DA; Zangar RC; Kim SG; Lewis JL (1995). Induction of
nasal carboxylesterase in F344 rats following inhalation exposure to pyridine. Drug Metab Dispos, 23:
529-535. 100182
118

-------
NRC (1983). Risk assessment in the federal government: Managing the process. National Research Council,
National Academy Press. Washington, DC. 194806
NRC (2008). Science and Decisions: Advancing Risk Assessment. Washington, DC: National Research
Council. 180073
NTP (1993). NTP technical report on toxicity studies of ethylene glycol ethers: 2-methoxyethanol, 2-
ethoxyethanol, 2-butoxyethanol (CAS Nos 109-86-4, 110-80-5, 111-76-2) administered in drinking
water to F344/N rats and B6C3F1 mice. National Toxicology Program. Research Triangle Park, NC. 26;
NIH Publication 93-3349. 042063
NTP (1996). Toxicology and carcinogenesis studies of acetonitrile (CAS No 75-05-8) in F344/N rats and
B6C3F1 mice (inhalation studies). National Toxicology Program. Research Triangle Park,
NC.http://ehp.niehs.nih.gov/ntp/docs/400 4xx doc.html. 042064
NTP (1998). NTP technical report on the toxicology and carcinogenesis studies of 2-butoxyethanol (CAS No.
111-76-2) in F344/N rats and B6C3F1 mice (inhalation studies). National Toxicology Program.
Research Triangle Park, NC. 594421
NTP (2000). NTP technical report on the toxicology and carcinogenesis studies of 2 butoxyethanol (CAS No.
111-76-2) in F344/N rats and B6C3F1 mice (inhalation studies). National Toxicology Program.
Research Triangle Park, NC. NTP TR 484. http://ntp.niehs.nih.gov/?objectid=070AC403-B110-CA79-
3A23AF79DE7B752A. 196293
Nyska A; Haseman JK; Kohen R; Maronpot RR (2004). Association of liver hemangiosarcoma and secondary
iron overload in B6C3F1 mice—the National Toxicology Program experience. Toxicol Pathol, 32: 222-
228. 056379
Nyska A; Maronpot RR; Ghanayem BI (1999). Ocular thrombosis and retinal degeneration induced in female
F344 rats by 2-butoxyethanol. Hum Exp Toxicol, 18: 577-582. 042747
Nyska A; Maronpot RR; Long PH; Roycroft JH; Hailey JR; Travlos GS; Ghanayem BI (1999). Disseminated
thrombosis and bone infarction in female rats following inhalation exposure to 2-butoxyethanol. Toxicol
Pathol, 27: 287-294. 042746
Osterhoudt KC (2002). Fomepizole therapy for pediatric butoxyethanol intoxication. J Toxicol Clin Toxicol,
40: 929-930. 100186
Park J; Kamendulis LM; Klaunig JE (2002). Effects of 2-butoxyethanol on hepatic oxidative damage. Toxicol
Lett, 126: 19-29. 030770
Pauluhn J (2003). Issues of dosimetry in inhalation toxicity. Toxicol Lett, 140-141: 229-238. 100188
Pietrangelo A (2004). Hereditary hemochromatosis~a new look at an old disease. N Engl J Med, 350: 2383.
100189
Poet TS; Soelberg JJ; Curry TL; Studniski KG; Corley RA (2002). In vivo kinetic studies with 2-
butoxyethanol. Part 1. Target tissue dosimetry. Battelle Memorial Institute for American Chemistry
Council. Columbus, OH and Arlington, VA. 4097f4. 594439
Poet TS; Soelberg JJ; Weitz KK; Mast TJ; Miller RA; Thrall BD; Corley RA (2003). Mode of action and
pharmacokinetic studies of 2-butoxyethanol in the mouse with an emphasis on forestomach dosimetry.
Toxicol Sci, 71: 176-189. 100190
Rambourg-Schepens MO; Buffet M; Bertault R; Jaussaud M; Journe B; Fay R; Lamiable D (1988). Severe
ethylene glycol butyl ether poisoning. Kinetics and metabolic pattern. Hum Toxicol, 7: 187-189. 100191
Ramot Y; Lewis DA; Ortel TL; Streicker M; Moser G; Elmore S; Ward SM; Peddada S; Nyska A (2007). Age
and dose sensitivities in the 2-butoxyethanol F344 rat model of hemolytic anemia and disseminated
thrombosis. Exp Toxicol Pathol, 58: 311-322. 100192
Raymond LW; Williford LS; Burke WA (1998). Eruptive cherry angiomas and irritant symptoms after one
acute exposure to the glycol ether solvent 2-butoxyethanol. J Occup Environ Med, 40: 1059-1064.
100193
Redlich M; Maly A; Aframian D; Shabat S; Ezov N; Levin-Harrus T; Nyska M; Nyska A (2004).
Histopathologic changes in dental and oral soft tissues in 2-butoxyethanol-induced hemolysis and
thrombosis in rats*. J Oral Pathol Med, 33: 424-429. 100194
119

-------
Reed JM; Kamendulis LM; Klaunig JE (2003). Examination of DNA damage in endothelial cells following
treatment with 2 butoxyethanol using the single cell gel electrophoresis (Comet) assay. Toxicologist, 72:
206. 594436
Rettenmeier AW; Hennigs R; Wodarz R (1993). Determination of butoxyacetic acid and N-butoxyacetyl-
glutamine in urine of lacquerers exposed to 2-butoxyethanol. Int Arch Occup Environ Health, 65: S151-
S153.042198
Roberts RA; Ganey PE; Ju C; Kamendulis LM; Rusyn I; Klaunig JE (2007). Role of the Kupffcr cell in
mediating hepatic toxicity and carcinogenesis. Toxicol Sci, 96: 2-15. 100196
Russel FG; Wouterse AC; van Ginneken CA (1987). Physiologically based pharmacokinetic model for the
renal clearance of phenolsulfonphthalein and the interaction with probenecid and salicyluric acid in the
dog. J Pharmacokinet Biopharm, 15: 349-368. 100197
Sabourin PJ; Medinsky MA; Birnbaum LS; Griffith WC; Henderson RF (1992). Effect of exposure
concentration on the disposition of inhaled butoxyethanol by F344 rats. Toxicol Appl Pharmacol, 114:
232-238. 006763
Sabourin PJ; Medinsky MA; Thurmond F; Birnbaum LS; Henderson RF (1993). Erratum to: Effect of dose on
the disposition of methoxyethanol, ethoxyethanol, adn butoxyethanol administered dermally to male
F344/N rats. Fundamental and Applied Toxicology 19:124-132. Fundam Appl Toxicol, 20: 508-510.
597595
Shabat S; Nyska A; Long PH; Goelman G; Abramovitch R; Ezov N; Levin-Harrus T; Peddada S; Redlich M;
Yedgar S; Nyska M (2004). Osteonecrosis in a chemically induced rat model of human hemolytic
disorders associated with thrombosis—a new model for avascular necrosis of bone. Calcif Tissue Int, 74:
220-228. 100200
Shepard KP (1994). Ethylene glycol monobutyl ether: Acute dermal toxicity study in the guinea pig. Eastman
Kodak Company for Chemical Manufacturers Association. Rochester, NY and Arlington, VA. 594428
Shepard KP (1994). Ethylene glycol monobutyl ether: Acute oral toxicity study in the guinea pig. Eastman
Kodak Company for Chemical Manufacturers Association. Rochester, NY and Arlington, VA. 594430
Shyr LJ; Sabourin PJ; Medinsky MA; Birnbaum LS; Henderson RF (1993). Physiologically based modeling of
2-butoxyethanol disposition in rats following different routes of exposure. Environ Res, 63: 202-218.
006766
Siesky AM; Kamendulis LM; Klaunig JE (2002). Hepatic effects of 2-butoxyethanol in rodents. Toxicol Sci,
70:252-260. 042204
Singh P; Morris B; Zhao S; Blaylock BL (2002). Suppression of the contact hypersensitivity response
following topical exposure to 2-butoxyethanol in female BALB/c mice. Int J Toxicol, 21: 107-114.
100203
Singh P; Zhao S; Blaylock BL (2001). Topical exposure to 2-butoxyethanol alters immune responses in female
BALB/c mice. Int J Toxicol, 20: 383-390. 100204
Sleet RB; Price CJ; Marr MC (1989). Teratologic evaluation of ethylene glycol monobutyl ether (Cas no 111-
76-2) administered to Fischer-344 rats on either gestational days 9 through 11 or days 11 through 13
(Final report). National Institute of Environmental Health Sciences. Research Triangle Park, NC. NTP
CTER 86 103. 042089
Smialowicz RJ; Riddle MM; Williams WC; Copeland CB; Luebke RW; Andrews DL (1992). Differences
between rats and mice in the immunosuppressive activity of 2-methoxyethanol and 2-methoxyacetic
acid. Toxicology, 74: 57-67. 100205
Smialowicz RJ; Williams WC; Riddle MM; Andrews DL; Luebke RW; Copeland CB (1992). Comparative
immunosuppression of various glycol ethers orally administered to Fischer 344 rats. Fundam Appl
Toxicol, 18: 621-627. 004389
Smith JG; German JB (1995). Molecular and genetic effects of dietary derived butyric acid. , 49: 87-90.
042103
Spolarics Z (1999). A carbohydrate-rich diet stimulates glucose-6-phosphate dehydrogenase expression in rat
hepatic sinusoidal endothelial cells. JNutr, 129: 105-108. 024792
120

-------
St Clair MBG; Morgan KT (1992). Changes in the upper respiratory tract. In Mohr U; Dungworth DL; Cappen
CC (Ed.),Pathobiology of the aging rat (pp. 111-127). Washington, DC: ILSI Press. 594426
Tyl RW; Millicovsky G; Dodd DE; Pritts IM; France KA; Fisher LC (1984). Teratologic evaluation of ethylene
glycol monobutyl ether in Fischer 344 rats and New Zealand white rabbits following inhalation
exposure. Environ Health Perspect, 57: 47-68. 100209
Tyler TR (1984). Acute and subchronic toxicity of ethylene glycol monobutyl ether. Environ Health Perspect,
57: 185-191. 006769
U.S. EPA (1986). Guidelines for mutagenicity risk assessment. Risk Assessment Forum, U.S. Enviornmental
Protection Agency. Washington, DC. EPA/630/R-98/003. 001466
U.S. EPA (1986). Guidelines for the health risk assessment of chemical mixtures. Risk Assessment Forum,
U.S. Environmental Protection Agency. Washington, DC. EPA/630/R-98/002. 001468
U.S. EPA (1988). Recommendations for and documentation of biological values for use in risk assessment.
Environmental Criteria and Assessment Office, Office of Health and Environmental Assessment, Office
of Research and Development, U.S. Environmental Protection Agency. Cincinnati, OH. EPA/600/6-
87/008. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=34855. 064560
U.S. EPA (1991). Guidelines for developmental toxicity risk assessment. U.S. Environmental Protection
Agency, Risk Assessment Forum. Washington, DC. EPA/600/FR-91/001.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=23162. 008567
U.S. EPA (1994). Interim policy for particle size and limit concentration issues in inhalation toxicity studies.
U.S. Environmental Protection Agency, Health Effects Division. Washington,
DC.http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=l86068. 076133
U.S. EPA (1994). Methods for derivation of inhalation reference concentrations and application of inhalation
dosimetry. Environmental Criteria and Assessment Office, Office of Health and Environmental
Assessment, Office of Research and Development, U.S. Environmental Protection Agency. Research
Triangle Park, NC. EPA/600/8-90/066F . http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=71993.
006488
U.S. EPA (1995). The use of the benchmark dose approach in health risk assessment. Risk Assessment Forum,
U.S. Environmental Protection Agency. Washington, DC. EPA/630/R-94/007.
http://cfpub.epa.gov/ncea/raf/recordisplay.cfm?deid=42601. 005992
U.S. EPA (1996). Guidelines for reproductive toxicity risk assessment. Risk Assessment Forum, U.S.
Environmental Protection Agency. Washington, DC. EPA/630/R-96/009.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=2838. 030019
U.S. EPA (1998). Guidelines for neurotoxicity risk assessment. National Center for Environmental
Assessment; Office of Research and Development; U.S. Environmental Protection Agency. Washington,
DC. EPA/630/R-95/001Fa. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=12479. 030021
U.S. EPA (1999). Toxicological Review of ethylene glycol monobutyl ethter (EGBE) in support of summary
information on the Integrated Risk Information System (IRIS). U.S. Environmental Protection Agency.
Washington, DC.http://www.epa.gov/ncea/iris/toxreviews/0500tr.pdf. 597365
U.S. EPA (2000). Benchmark dose technical guidance document [external review draft]. Risk Assessment
Forum, U.S. Environmental Protection Agency. Washington, DC. EPA/630/R-00/001.
http://www.epa.gov/raf/publications/benchmark-dose-doc-draft.htm . 052150
U.S. EPA (2000). Science policy council handbook: Risk characterization. Office of Science Policy, Office of
Research and Development, U.S. Environmental Protection Agency. Washington, D.C.. EPA 100-B-00-
002. www.epa.gov/spc/pdfs/rchandbk.pdf. 052149
U.S. EPA (2000). Supplementary guidance for conducting health risk assessment of chemical mixtures. Risk
Assessment Forum, U.S. Environmental Protection Agency. Washington, DC. EPA/630/R-00/002.
004421
U.S. EPA (2002). A review of the reference dose and reference concentration processes. Risk Assessment
Forum, U.S. Environmental Protection Agency. Washington, DC. EPA/630/P-02/0002F.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=55365. 088824
121

-------
U.S. EPA (2004). Toxicological review of 1,2-dibromoethane (CASRN 106-93-4) in support of summary
information on the Integrated Risk Information System (IRIS). U.S. Environmental Protection Agency.
Washington, D.C..http://www.epa.gov/iris/subst/0361.htm. 594429
U.S. EPA (2005). An evaluation of the human carcinogenic potential of ethylene glycol butyl ether, Final
Report. National Center for Enviornmental Assessment, Office of Research and Development, U.S.
Environmental Protection Agency. Washington, D.C.. EPA 600/R 04/123. 594432
U.S. EPA (2005). Guidelines for carcinogen risk assessment, Final Report. Risk Assessment Forum, U.S.
Environmental Protection Agency. Washington, DC. EPA/630/P-03/001F.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=l 16283. 086237
U.S. EPA (2005). Supplemental guidance for assessing susceptibility from early-life exposure to carcinogens.
Risk Assessment Forum, U.S. Environmental Protection Agency. Washington, DC. EPA/630/R-03/003F.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=l 16283. 088823
U.S. EPA (2006). A framework for assessing health risk of environmental exposures to children (final).
National Center for Environmental Assessment, Office of Research and Development, U.S.
Environmental Protection Agency. Washington, DC. EPA/600/R-05/093F.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=158363. 194567
U.S. EPA (2006). Approaches for the application of physiologically based pharmacokinetic (PBPK) models
and supporting data in risk assessment (final report). National Center for Environmental Assessment,
Office of Research and Development, U.S. Environmental Protection Agency. Washington, DC.
EPA/600/R-05/043F. 194568
U.S. EPA (2006). Science Policy Council Handbook: Peer Review. Office of Science Policy, Office of
Research and Development, U.S. Environmental Protection Agency. Washington, DC. EPA/100/B-
06/002. http://www.epa.gov/peerreview/pdfs/peer_review_handbook_2006.pdf. 194566
U.S. EPA (2008). Toxicological Review of Ethylene Glycol Mono-Butyl Ether (EGBE) (External Review
Draft). U.S. Environmental Protection Agency. Washington, DC. EPA/635/R-08/006A.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=189631. 597575
Udden M; Patton C (2005). Butoxyacetic acid-induced hemolysis of rat red blood cells: effect of external
osmolarity and cations. Toxicol Lett, 156: 81-93. 100213
Udden MM (1994). Hemolysis and deformability of erythrocytes exposed to butoxyacetic acid, a metabolite of
2-butoxyethanol: II Resistance in red blood cells from humans with potential susceptibility. J Appl
Toxicol, 14: 97-102. 042109
Udden MM (1996). Effects of butoxyacetic acid on human red cells. Occup Hyg, 2: 283-290. 042209
Udden MM (2000). Rat erythrocyte morphological changes after gavage dosing with 2-butoxyethanol: a
comparison with the in vitro effects of butoxyacetic acid on rat and human erythrocytes. J Appl Toxicol,
20: 381-387. 042110
Udden MM (2002). In vitro sub-hemolytic effects of butoxyacetic acid on human and rat erythrocytes. Toxicol
Sci, 69: 258-264. 042111
Udden MM; Patton CS (1994). Hemolysis and deformability of erythrocytes exposed to butoxyacetic acid, a
metabolite of 2-butoxyethanol: I. Sensitivity in rats and resistance in normal humans. J Appl Toxicol,
14:91-96. 056374
Valberg L; Simon J; Manley P; Corbett W; Ludwig J (1975). Distribution of storage iron as body stores expand
in patients with hemochromatosis. J Lab Clin Med, 86: 479. 100217
Wang YJ; Ho YS; Lo MJ; Lin JK (1995). Oxidative modification of DNA bases in rat liver and lung during
chemical carcinogenesis and aging. Chem Biol Interact, 94: 135-145. 024797
Werner H; Mitchell J; Miller J; Von Oettingen W (1943). Effects of repeated exposure of dogs to monoalkyl
ethylene glycol ether vapors. J Ind Hyg Toxicol, 25: 409-414. 100219
Werner HW; Nawrocki CZ; Mitchell JL; Miller J; Oettingen WV (1943). Effects of repeated exposure of rats
to vapors of monoalkyl ethers of ethylene glycol. J Ind Hyg Toxicol, 25: 374-379. 597282
122

-------
Wier PJ; Lewis SC; Traul KA (1987). A comparison of developmental toxicity evident at term to postnatal
growth and survival using ethylene glycol monoethyl ether, ethylene glycol monobutyl ether, and
ethanol. Teratog Carcinog Mutagen, 7: 55-64. 042123
Wintrobe M; Lee G; Boggs D (1981). Variations of leukocytes in disease. In Wintobe MM (Ed.),Clinical
hematology (pp. 1284-1323). Philadelphia, PA: Lea& Febiger. 100222
Wintrobe MM (1981). The normocytic, normochromic anemia. In Wintrobe MM (Ed.),Clinical hematology
(pp. 677-697). Philadelphia, PA: Lea & Febiger. 594424
Yamaguchi R; Hirano T; Asami S; Chung M-H; Sugita A; Kasai H (1996). Increased 8-hydroxyguanine levels
in DNA and its repair activity in rat kidney after administration of a renal carcinogen, ferric
nitrilotriacetate. Carcinogenesis, 17: 2419-2422. 024876
Yoshizawa K; Kissling G; Johnson J; Clayton N; Flagler N; Nyska A (2005). Chemical-Induced Atrial
Thrombosis in NTP Rodent Studies. Toxicol Pathol, 33: 517-532. 100224
Zeiger E; Anderson B; Haworth S; Lawlor T; Mortelmans K (1992). Salmonella mutagenicity tests: V Results
from the testing of 311 chemicals. Environ Mol Mutagen, 19: 2-141. 095748
Ziouzenkova O; Asatryan L; Sevanian A (1999). Oxidative stress resulting from hemolysis and formation of
catalytically active hemoglobin: protective strategies. Int J Clin Pharmacol Ther, 37: 125-132. 024904
123

-------
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC
COMMENTS AND DISPOSITION
The Toxicol ogical Review of Ethylene Glycol Monobutyl Ether (U.S. EPA, 2008,
597575) has undergone formal external peer review performed by scientists in accordance with
U.S. EPA guidance on peer review (U.S. EPA, 2006, 194566). The external peer reviewers were
tasked with providing written answers to general questions on the overall assessment and on
chemical-specific questions in areas of scientific controversy or uncertainty. A summary of
significant comments made by the external reviewers and U.S. EPA's responses to these
comments arranged by charge question follow. In many cases the comments of the individual
reviewers have been synthesized and paraphrased for development of Appendix A. Public
comments that were received are summarized and addressed following the peer-reviewers
comments and disposition.
On April 10, 2008, U.S. EPA introduced revisions to the IRIS process for developing
chemical assessments. As part of the revised process, the disposition of peer reviewer and public
comments, as found in this Appendix, and the revised IRIS Toxicological Review was provided
to the external peer review panel members for their comment on May 15, 2009. Any additional
peer review panel comments received as part of this second review and U.S. EPA's responses are
included at the end of this Appendix.
A.l. GENERAL CHARGE QUESTIONS:
1. Is the Toxicological Review logical and clear? Has EPA accurately, clearly and
objectively represented and synthesized the scientific evidence for noncancer and cancer
hazard?
Comments:
In general, reviewers found the Toxicological Review to be logical and clear. In addition,
they found that the review accurately, clearly, and objectively represented and synthesized the
scientific evidence for noncancer and cancer hazard.
More specifically, several reviewers provided suggestions to improve the clarity of the
document and one reviewer questioned the quality of analytical procedures used in the critical
study.
Note: Hyperlinks to the reference citations throughout this document will take you to the NCEA HERO database (Health and
Environmental Research Online) at http://epa.gov/hero. HERO is a database of scientific literature used by U.S. EPA in the process of
developing science assessments such as the Integrated Science Assessments (ISA) and the Integrated Risk Information System (IRIS).
A-l

-------
Two reviewers stated that the PBPK modeling of EGBE and BAA was not well
illustrated or adequately described in the document and one reviewer felt that Figure 4-2 was not
very informative.
One reviewer commented that the low renal clearance of BAA relative to GFR needs to
be discussed in more depth. The same reviewer posed several questions: is there data on plasma
albumin binding of BAA and does it vary between rodents and humans? If the binding is less
than -95%, it is unlikely to be the explanation of the restricted renal elimination. The alternate
explanation of active reuptake by a carrier mechanism is amenable to study: is there relevant
information in the literature? Resolution of this question is deemed crucial in understanding the
factors that determine both the Cmax and AUC of BAA in the various species/sexes of rodents as
well as the determination of HECs.
One reviewer commented that the toxicokinetic discussion (Section 3.2) is, perhaps of
necessity, somewhat diffuse and suggested a table incorporating some of the relevant kinetic
information.
Response:
The suggestions for document improvement were incorporated into the Toxicological
Review wherever possible and appropriate. Additional text has been added in Section 4.5 to
clarify the effects of EGBE on RBCs and the possible pathophysiological mechanisms involved.
Comments concerning the quality of the critical study were considered, but do not negate the
positive findings nor preclude the use of study dose-response data.
A brief summary of the existing PBPK models and rationale for use has been included in
Section 3.2 as well as in Appendix B. We have eliminated Figure 4-2 as PBPK simulations for
peak (Cmax) concentrations of EGBE, BAL, and BAA in the liver of male mice following 6-hour
inhalation exposures to 250 ppm (the highest concentration used in the NTP bioassay) were
adequately described in the text.
The PBPK modeling paper by Corley et al. (1994, 041977) describes the basis for
inclusion of plasma protein binding and renal clearance of BAA for male F344 rats and humans.
The parameters for protein binding were assumed to be similar to phenolsulfophthalein and renal
clearance optimized from in vivo pharmacokinetic studies that indicated that the elimination of
BAA in urine by male F344 rats was saturable, leading to nonlinear increases in blood BAA Cmax
and AUC values at high dose levels. Parameters developed from rats were successfully scaled to
humans as demonstrated by Franks et al. (2006, 100117). Jones et al. (2003, 100161). and Corley
et al. (1994, 041977: 1997, 041984). for 'low-dose' inhalation and dermal exposures, and
Gualtieri et al. (1995, 594443: 2003, 100140) for 'high dose' oral exposures (suicide attempt).
In their 1998 publication, Lee et al. (1998, 041983) updated the model parameters using the in
vivo toxicokinetic data for male and female, young and old, rats and mice. As identified by the
reviewer, these processes are experimentally accessible. Thus, Corley et al. (2005, 100100)
A-2

-------
conducted a series of studies to determine the species, gender, and age-dependent tissue:blood
partitition coefficients for BAA, plasma protein binding of BAA, metabolism of EGBE and
BAA, and renal active transport of BAA. These parameters were used to update the PBPK
model of Lee et al. (1998, 041983). The toxicokinetic summary (Section 3.2) has been revised to
include additional information from Corley et al. (2005, 100100) who determined plasma protein
binding, partition coefficients, and renal elimination of BAA in young vs. aged male and female
rats and mice.
A table of relevant toxicokinetic information has been included in Section 3.2 to
summarize the information presented in the text.
2. Please identify any additional studies that should be considered in the assessment of the
noncancer and cancer health effects of EGBE.
Comments:
Most reviewers did not identify any additional studies that should be considered in the
assessment of the health effects. However, a few reviewers did provide suggestions for
supporting references and expanded discussions that are listed below.
a.	References on the subjects of thrombosis and infarction.
b.	References on the subjects of hematology instrument methodology differences and iron
overload.
c.	References on the subject of the role of Kupffer cells in hepatotoxicity and
carcinogenicity.
d.	While no specific references were supplied, one reviewer suggested an increased
discussion on the MO A of EGBE on the RBC and its fate.
e.	An increased discussion on the subject of olfactory hyaline membrane degeneration.
Response:
a. The subject of thrombosis and infarction is briefly discussed in Section 4.4.1. Additional
text has been added in Section 4.5 of the Toxicological Review including the suggested
reference as well as others. The added references are as follows:
Bevers, EM; Comfurius, P; van Rijn, JL; et al. (1982) Generation of prothrombin-
converting activity and the exposure of phosphatidylserine at the outer surface of
platelets. Eur J Biochem 122(2):429-436. (1982, 100088)
Connor, J; Bucana, C; Fidler, IJ; et al. (1989) Differentiation-dependent expression of
phosphatidylserine in mammalian plasma membranes: quantitative assessment of
outer-leaflet lipid by prothrombinase complex formation. Proc Natl Acad Sci USA.
86(9):3184-3188. (1989. 100097)
Ezov, N; Levin-Harrus, T; Mittelman, M; et al. (2002) A chemically induced rat
model of hemolysis with disseminated thrombosis. Cardiovasc Toxicol 2:181-194.
(2002, 1001 13)
A-3

-------
Nyska, A; Maronpot, RR; Long, PH; et al. (1999b) Disseminated thrombosis and
bone infarction in female rats following inhalation exposure to 2-butoxyethanol.
Toxicol Pathol 27(3):287-294.(1999, 042746)
Ramot, Y; Lewis, DA; Ortel, TL; et al. (2007) Age and dose sensitivities in the
2-butoxyethanol F344 rat model of hemolytic anemia and disseminated thrombosis.
Exp Toxicol Pathol 58(5):311-322. (2007, 100192)
Yoshizawa, K; Kissling, GE; Johnson, JA; et al. (2005) Chemical-induced atrial
thrombosis in NTP rodent studies. Toxicol Pathol 33(5):517-532. (2005, 100224)
b.	All of the references cited in the Toxicological Review pertaining to MCV used the
impedance type analyzers (the more sensitive of the two types commented on).
Therefore, no changes were made to the document. The references suggested on iron
overload were reviewed. It was determined that the information in the papers would not
add significantly to the outcome of the health assessment.
c.	Additional text and the suggested reference have been added to the Toxicological Review
on the role of Kupffer cells in hepatotoxicity and carcinogenicity in Section 4.5. The
following reference was added:
Roberts, RA; Ganey, PE; Ju, C; et al. (2007) Role of the Kupffer cell in mediating
hepatic toxicity and carcinogenesis. Toxicol Sci 96(1):2-15. (2007, 100196)
d.	Additional text has been added to the Toxicological Review discussing the MOA of
EGBE on the RBC and its fate in Section 4.5.
e.	Additional text and references have been added to the Toxicological Review discussing
the effect of olfactory hyaline membrane degradation and the U.S. EPA's rationale for its
exclusion as a critical effect. The references added are as follows:
Buckley, LA; Morgan, KT; Swenberg, JA; et al. (1985) The toxicity of
dimethylamine in F344 rats and B6C3Fi mice following a 1-year inhalation
exposure. Fundam Appl Toxicol 5(2):341-352. (1985, 062668)
Nikula, KJ; Novak, RF; Chang, IY; et al. (1995) Induction of nasal carboxylesterase
in F344 rats following inhalation exposure to pyridine. Drug Metab Dispos
23(5):529—535. (1995, 100182)
St. Clair, MBG; Morgan, KT. (1992) Changes in the upper respiratory tract. In Mohr,
U; Dungworth, DL; Capen, CC; eds. Pathobiology of the Aging Rat, Vol. 1, ILSI
Press; Washington, DC: pp. 111-127. (1992, 594426)
3. Please discuss research that you think would be likely to increase confidence in the
database for future assessments of EGBE.
Comments:
All reviewers provided research suggestions that may increase confidence in the database
for future EGBE assessments. The following is a brief summary of these comments and
suggestions:
• Measure iron levels in the liver;
A-4

-------
•	Do dose/response for oxidant-induced liver cell initiation and promotion;
•	Define the age of erythrocyte susceptibility to BAA;
•	Definition of the MOA of the toxic insult to the erythrocyte with definition of dose
metrics;
•	Determine the loss of sidedness in the phospholipid composition of the outer leaflet of the
red cell membrane to the premature loss of red cells in the circulation and the possibility
that this may induce microvascular thrombosis and infarction;
•	Determine the role of the spleen vs. the liver in the removal of damaged red cells and
effects on Hp and hemopexin levels and saturation during chronic exposure to EGBE;
•	Determine the time and dose/response relationships for hemosiderin deposition in the
liver as a whole and in specific cell types;
•	Determine the MOA and dose/response relationships for hemosiderin in
hemangiosarcoma development;
•	Experiments designed to determine the fate of BAA damaged RBCs and the role of
spleen and liver macrophages;
•	Measure the phosphatidylserine levels on RBCs following exposure to BAA;
•	Experiments designed to better define the cellular effects of BAA exposure and their role
in hemosiderin deposition and ROS toxicity;
•	Determination of the dose and temporal relationships between EGBE-induced
hemosiderin and its induction of hemolysis;
•	Experiments designed to define the nature and dose response of effects seen in humans
following moderate exposures;
•	Experiments designed to characterize the species differences in RBC membrane
physiology;
•	Long term human inhalation and oral studies of EGBE;
•	Experiments designed to address the mechanism(s) by which BAA induces RBC
hemolysis in various species;
•	Experiments designed to investigate the role of ROS in modulation of endothelial and
hepatic cell gene expression;
•	In vivo studies designed to quantify the internal threshold doses that must be met to
progress from one key precursor event in the mechanistic sequence to another; and
A-5

-------
• Experiments designed to determine the relative susceptibility of endothelial cells and
hepatocytes to oxidative damage.
Response:
A number of research suggestions were received, and U.S. EPA agrees further research
may enhance future risk assessments of EGBE. No additional changes to the Toxicological
Review are indicated at this time.
4. Please comment on the identification and characterization of sources of uncertainty in
Sections 5 and 6 of the assessment document. Please comment on whether the key sources
of uncertainty have been adequately discussed. Have the choices and assumptions made in
the discussion of uncertainty been transparently and objectively described? Has the impact
of the uncertainty on the assessment been transparently and objectively described?
Comments:
All of the reviewers found Sections 5 and 6 to be well written and complete. A number
of reviewers had comments on the selection of the factors that account for uncertainty (UFs).
Additional suggestions included the following: (1) an alternative endpoint of olfactory
epithelial degeneration; (2) Sections 5.1.3 and 5.2.3 be expanded for clarity; and (3) further
analyses (Hazard Quotients and Hazard Indices) be performed for Section 6.
Response:
Specific comments on the values of the UFs are addressed in the response to charge
question A5.
The potential use of olfactory epithelial degeneration as an endpoint is discussed in
Section 5. Additional references have been added that support the position that olfactory
epithelial degeneration is not an adverse effect but an adaptive response.
Substantial improvements were made to Section 5 that address the suggestion for
improved clarity. Supporting text for alternative derivations has been moved to Appendix C to
improve the readability of the document.
Computing Hazard Quotients and Hazard Indices would require an exposure analysis,
and while an portent component of risk assessment, is beyond the scope of this human health risk
assessment.
A-6

-------
A.2. CHEMICAL-SPECIFIC CHARGE QUESTIONS:
A.2.1. (A) Inhalation reference concentration (RfC) for EGBE
Al. The 2-year inhalation study by the National Toxicology Program (NTP, 2000,196293)
was selected as the basis for the chronic inhalation RfC. Please comment on whether the
selection of this study as the principal study has been scientifically justified. Has this study
been transparently and objectively described in the document? Please identify and provide
the rationale for any other studies that should be selected as the principal study.
Comments:
All reviewers agreed with the selection of the NTP (2000, 196293). No additional studies
were identified for use as the principal study. A number of editorial comments were provided to
improve the document. One reviewer commented that since humans are substantially less
sensitive than rodents to the EGBE-induced hemolysis, they could be prone to other effects
which are masked in the rodent study. He concludes with the statement that there does not seem
to be any adequate alternative to the approach taken.
Response:
Based on the favorable comments from the reviewers, no substantial changes were made
to the Toxicological Review with respect to the selection of the principal study for the RfC.
Suggested editorial comments have been reviewed and revisions were made to the Toxicological
Review where appropriate.
A2. The incidence of hemosiderin staining in the liver of male rats was selected as the
critical effect because it is considered by EPA to be a precursor to an adverse effect. Please
comment on whether the selection of this critical effect has been scientifically justified. Are
the criteria and rationale for this selection transparently and objectively described in the
document? Please provide a detailed discussion. Please identify and provide the rationale
for any other endpoints that should be considered in the selection of the critical effect.
Comments:
Five of the seven reviewers agreed with the conclusion that hemosiderin staining in the
liver of male rats was the most appropriate choice for the critical effect. However, two of those
five reviewers communicated reservations regarding this conclusion and discussed the
deficiencies of its use. It was suggested that both hemosiderin and Hct, as parameters for POD
purposes, be discussed and illustrated in the review.
Two reviewers thought that alternate endpoints were more appropriate, such as RBC
counts, Hct levels, MCV, or hyaline degeneration.
A-7

-------
Response:
RfC and RfD derivations from all endpoints were considered and are now illustrated in
Appendix C as well as summarized in Sections 5.1.4 and 5.2.4 of the assessment. The dose-
response assessments used in the derivation of the RfC and RfD are reviewed in Sections 5.1.2
and 5.1.4, respectively. Based on the most recent information on the MOA for EGBE,
hemosiderin deposition was selected as the critical effect. While the U.S. EPA recognizes that
other hematological endpoints are biologically relevant, hemosiderin provides the best overall
choice based on pathological considerations, dose-response relationships, and model fit. The
support and rationale for the choice of the critical effect can be found in Sections 5.1, 5.2, 6.1,
and 6.2.
A3. Benchmark dose (BMD) modeling was applied to incidence data for hemosiderin
staining in male rat liver to derive the point of departure (POD) for the RfC. Please
provide comments with regard to whether BMD modeling is the best approach for
determining the POD. Has the BMD modeling been appropriately conducted and
objectively and transparently described? Has the benchmark response (BMR) selected for
use in deriving the POD (i.e., 10% extra risk of hemosiderin staining in the liver) been
scientifically justified, and transparently and objectively described? Please identify and
provide the rationale for any alternative approaches for the determination of the POD and
discuss whether such approaches are preferred to EPA's approach.
Comments:
All reviewers agreed that BMD modeling was the best approach for determining the
POD. However, alternatives were suggested for the choice of species and gender. One reviewer
commented that male mice should have been used since they have shown correlation between the
critical effect and the adverse effect (tumor). Two other reviewers commented that female rats
should have been used since they have been shown to be more sensitive to the hemolytic effects
and were used in the previous IRIS assessment. Lastly, one reviewer commented on the choice
of 10% response level. This reviewer felt that a 5% response level is scientifically supported
because it approximates the NOAEL.
Response:
Based on the majority of comments agreeing with the choice of gender and endpoint to
model, no substantial changes to the ToxicologicalReview are indicated. Rationale for the
selection of the species and gender are discussed in Section 5.3.9. Briefly, the male rat was
chosen based on the NTP (2000, 196293) report showing male rats were the most sensitive
species and gender with respect to the critical effect (hemosiderin deposition). In addition, the
A-8

-------
modeling of the data for male rats provided a much better fit than the modeling for the female rat
data. As for the selection of the response rate, utilizing U.S. EPA guidance on BMD modeling
of qualitative data 10% was chosen because it was within the range of experimental responses
and appropriate for the power of the study. Additional justification has been added regarding the
fact that this tumor response may not have been observed in the rats because the rats were
exposed to lower doses and for a shorter duration of their average lifespan than the mice.
A4. PBPK modeling was used to extrapolate the POD from rats to humans. Please
comment on whether the PBPK modeling for interspecies extrapolation is scientifically
justified, and transparently and objectively described in the document. Does the model
properly represent the toxicokinetics of the species under consideration? Was the model
applied properly? Are the model assumptions, parameter values, and selection of dose
metrics clearly presented and scientifically supported?
Comments:
All reviewers that provided comments agreed that the PBPK modeling was the
appropriate method for extrapolating the POD from rats to humans. Two reviewers commented
that a comparison between the 1999 assessment and this one would be helpful.
Response:
Based on the comments, no change in the ToxicologicalReview is indicated.
Comparisons to the previous assessment are made throughout the document in the appropriate
sections.
A5. Please comment on the selection of all of the uncertainty factors applied to the POD
for the derivation of the chronic RfC. For instance, are they scientifically justified, and
transparently and objectively described in the document? An UF of 10 for extrapolation
from animals to humans (UFA) is generally applied when data are not available to inform
potential pharmacokinetic (PK-UF) and pharmacodynamic (PD-UF) differences. In this
assessment, an UFA of 1 was applied.
•	A PBPK model was used to inform pharmacokinetic differences and a PK-UF of 1 was
selected. Please comment on whether this selection is scientifically justified. Is the
rationale transparently and objectively described? Please comment on whether there are
sufficient scientific data and support for the use of this PBPK model to estimate
interspecies toxicokinetic differences and to replace the default interspecies factor for
1/2
toxicokinetic differences (i.e., 10 ).
•	Evidence from human and animal in vitro and in vivo studies were used to inform
pharmacodynamic differences and a PD-UF of 1 was selected. Please comment on
A-9

-------
whether this selection is scientifically justified. Is the rationale transparently and
objectively described? Please comment on whether a higher value for the PD-UF should
be used (e.g., to account for the limited information available on the potential for effects
in human cell types other than red blood cells) or alternatively, should a lower (i.e.,
fractional) PD-UF be used (e.g., to account for the 40-150-fold difference in the
concentrations that cause prehemolytic effects in human red blood cells (RBCs),
including RBCs from potential susceptible populations such as the elderly, and patients
suffering from anemia and RBC disorders that weaken the cellular membrane such as
hereditary spherocytosis).
Please identify and provide the rationale for any alternative approaches for the selection
of the uncertainty factors.
Comments:
Several reviewers commented that the selection of factors to account for uncertainty
(UFs) was not supported by the available science.
a.	Two reviewers concluded that the UFA should be <1.
b.	Two reviewers commented that the UFH should be <10.
c.	One reviewer concluded that the UFD was inconsistent with the confidence level.
d.	Most reviewers agreed that the selection of the UF values should be more clearly
justified.
Response:
The current UFa value in the ToxicologicalReview is 1. The UFa accounts for a
reduction from 3 for the toxicokinetic differences between animals and humans through the use
of a PBPK model for extrapolation of doses. The toxicodynamic portion, likewise was reduced
from 3 to 1. The toxicological effect in question being the deposition of hemosiderin is
identified as a key event in the MOA for the development of liver hepatomas in mice. To
implement a further reduction in the UFA (i.e., fractional) would logically indicate that a
preponderance of data are available to describe this key event (toxicodynamics) in both animals
and humans to the extent of describing why humans are less sensitive to the hemolytic effects
leading to hemosiderin deposition. This is not the case, although in vitro data (Ghanayem, 1989,
042014; Udden, 2000, 042110; Udden, 2002, 042111; Udden andPatton, 2005, 100213) do
suggest humans are less sensitive than rodents to the hemolytic effects of EGBE. Likewise, the
few human studies (Carpenter et al., 1956, 066464; Haufroid et al., 1997, 042040) indicate the
same finding. However, these studies (Carpenter et al., 1956, 066464; Ghanayem, 1989,
042014; Haufroid et al., 1997, 042040; Udden, 2000, 042110; Udden, 2002, 042 111; Udden and
A-10

-------
Patton, 2005, 100213) do not characterize hemosiderin deposition, the key event for the MOA In
addition, little is known of the long-term or repeated exposure responses in humans to EGBE.
a.	There is very limited information on the sensitivity of various human subpopulations to
the hemolytic effects of EGBE. In addition, the long-term and repeated exposure effects
in potentially sensitive populations remain to be determined. Therefore, the default UFH
of 10 is appropriate. This subject is discussed in Section 5.2.3.
b.	Response to this comment can be found in question A6 below.
c.	Additional text has been added to Sections 5.1.3 and 5.2.3 for improved justification of
the chosen UFs; the sections were reformatted for improved clarity.
A6. Please comment specifically on the database uncertainty factor of 1 applied in the RfC
derivation. Are the criteria and rationale for the selection of the database uncertainty
factor transparently and objectively described in the document? Please comment on the
body of information regarding the hemato and hepatic toxicity of EGBE and the use of the
toxicokinetic data in the determination of the database uncertainty factor. Please comment
on whether the selection of the database uncertainty factor for the RfC has been
scientifically justified. Has this selection been transparently and objectively described in
the document?
Comments:
The majority of reviewers commented that the UFD value was appropriate. One reviewer
commented that the value was inconsistent with the U.S. EPA's confidence level of medium to
high.
Response:
Based on the reviewers comments, no change in the document is indicated. The value
assigned for the database UF is based on the completeness of the database in terms of
toxicological studies assessing the range of likely potential effects including reproductive and
developmental effects as well as information from more than one species.
A.2.2. (B) Oral reference dose (RfD) for EGBE
Bl. A conclusion was reached that the available oral toxicity data are inadequate to
support derivation of a chronic oral RfD value. Is the rationale for not developing an RfD
from the available database of oral studies transparently and objectively described? If
other oral studies are identified that would be suitable for the derivation of the RfD, please
identify and provide the rationale for their use.
A-ll

-------
Comments:
All reviewers commented that the 91-day drinking water study should be considered in
the RfD development and that the justification for not using it to at least derive comparative RfD
estimates was inadequate.
Response:
Currently, there are no chronic oral studies of EGBE and very limited sub-chronic
studies. The larger database of inhalation studies, including 2 year studies that identify chronic
endpoints (e.g. hemosiderin deposition), supports the use of a route-to-route extrapolation using
the chronic inhalation study and PBPK modeling to produce an RfD estimate. The 91-day
drinking water study has been included in Chapter 5 for comparison purposes and Section 5.2.1
was amended to include a more concise rationale for the exclusion of the oral study in
determining the RfD. Alternative endpoints including the endpoint selected for the previous
assessment are presented graphically and discussed in a new section (5.2.4).
B2. A route-to-route extrapolation was performed to derive the chronic RfD, using the
chronic inhalation study and PBPK modeling. The Human Equivalent Concentration
(HEC) was based on a continuous oral exposure to EGBE in drinking water that would
yield the same AUC for the metabolite BAA (in the arterial blood over three months) as
that estimated for the rat following an external inhalation exposure to EGBE at the level of
the proposed POD (i.e., the BMCLio). Please comment on whether the PBPK model is
adequate for use to conduct a route-to-route extrapolation for EGBE to derive an RfD in
the absence of adequate oral animal or human dose-response data to derive the RfD
directly. Was the extrapolation correctly performed and objectively and transparently
documented?
Comments:
Five of seven reviewers agreed with the extrapolation. Two reviewers abstained from
commenting citing their lack of expertise in this area. Additionally, it was noted by two
reviewers that while the extrapolation was correctly performed, it would not be necessary if the
drinking water study were used.
Response:
Based on the reviewer's comments, no change in the ToxicologicalReview is indicated.
A-12

-------
B3. Please comment specifically on the database uncertainty factor of 1 applied in the RfD
derivation. Are the criteria and rationale for the selection of the database uncertainty
factor transparently and objectively described in the document? Measured internal doses
in rats and a human PBPK model were used to perform a route-to-route extrapolation to
derive the RfD. Please comment on the use of the PBPK model and the inhalation database
in the determination of the database uncertainty factor for the RfD. Please comment on
whether the selection of the database uncertainty factor for the RfD has been scientifically
justified. Has this selection been transparently and objectively described in the document?
Comments:
There was general agreement among the reviewers with the selection of the database UF
for the RfD. One reviewer commented that the explanation of the rationale for the UF selection
should be improved.
Response:
Based on the reviewer's comments, the explanation of the rationale for the UF selection
has been revised for improved transparency in the ToxicologicalReview where indicated.
A.2.3. (C) Carcinogenicity of EGBE
CI. Under the EPA's 2005 Guidelines for Carcinogen Risk Assessment
http://www.eDa.gov/cancerguidelines/. the Agency concluded that EGBE is not likely to be
carcinogenic to humans at expected exposure concentrations. Please comment on the
scientific justification for the cancer weight of evidence characterization and describe the
basis for your view. Has the scientific justification for the weight of evidence descriptor
been sufficiently, transparently and objectively described?
Comments:
Four of seven reviewers indicated that they agreed with the Agency's conclusion
regarding the cancer descriptor for EGBE. One reviewer abstained from commenting while
another reviewer only commented on the wording of the cancer descriptor. One reviewer
commented that the cancer descriptor was carefully explained. Two reviewers suggested the
phrase "at expected exposure concentrations" should either be deleted or better defined with
exposure concentrations encountered. These reviewers provided a possible revision as follows:
"at the calculated RfC and RfD values presented in this document."
A-13

-------
Response:
Because all reviewers that commented agreed with the Agency conclusion that EGBE is
not likely to be carcinogenic to humans at expected exposure concentrations, qualitatively this
conclusion was not revised. As suggested by two of the reviewers on the External Peer Review
Panel, the specific text to the weight of evidence descriptor was revised to account for their
suggested language. Lastly, all reviewers that commented, responded favorably concerning the
scientific justification for the cancer weight of evidence descriptor, thus, no revisions were
indicated for the document.
C2. EPA has proposed a mode of action (MOA) for male mouse liver cancer involving
metabolism, hemolysis of RBCs, hemosiderin deposition in the liver, oxidative damage and
proliferation leading to tumor induction as key events best supported by the data. Please
provide detailed comments on whether this analysis regarding the MOA for liver cancer is
scientifically sound, and transparently and objectively described in the Toxicological
Review. Considerations include the scientific support regarding the plausibility for the
hypothesized MOA and the characterization of uncertainty regarding this MOA.
Comments:
In general, the reviewers agreed with the conclusions of the analysis regarding the MOA
for liver cancer, although several reviewers provided additional scientific considerations
regarding the MOA. Most notably, remarks regarding hemosiderin deposition include: (1) Why
does hemosiderin deposition not occur in the spleen?; (2) Tumor induction from initiated cells is
speculation; (3) If male mice represent the most sensitive species for tumor formation due to
exposure to EGBE, why is a more marked dose response for hemosiderin deposition observed in
female mice?; and (4) Two reviewers commented that hemosiderin deposition was a biomarker
of exposure, not effect.
Response:
Responses to this charge question will be addressed in the order in which they were
presented in the comment overview (above): (1) One reviewer inquired as to why hemosiderin
deposition does not occur in the spleen. NTP (2000, 196293) reported hemosiderin deposition
does occur in the spleen for which these data are presented in Table 4-5; (2) Tumor induction
from initiated cells is not speculated within the Toxicological Review. With respect to the
hypothesized MOA for forestomach tumors, the MOA is the same as that proposed for
hepatocellular tumors observed in male mice (2000, 196293). Within this MOA, the occurrence
of initiated cells is not specifically addressed but could occur from two of the key events
including oxidative stress and cytotoxicity. Although data do not currently exist for EGBE, the
possibilities also include epigenetic pathways that could lead to the formation of initiated cells.
A-14

-------
A key point regarding the hypothesized MOA is that the key events represent obligatory (rate
limiting) steps or biomarkers of such events leading to the formation of tumors, but do not
represent a complete description of the biologic mechanisms that lead to tumor development;
(3) Hemosiderin deposition observed was indeed greater in female compared to male mice (NTP,
2000, 196293; see table 4-5). NTP has observed liver hemangiosarcomas in 105/4,183 (2.51%)
male versus just 35/4,177 (0.84%) female historical controls (Klaunig and Kamendulis, 2005,
100165; NTP, 2000, 196293). In addition, other chemicals reported by NTP to cause both early
onset hemosiderin buildup and liver tumors have also exhibited this male specificity (U.S. EPA,
2005, 594432). While the reason for the sex difference in liver tumor susceptibility between
male and female mice is not clear, it has been shown that estrogens can be protective through
their antioxidant capacities and through their modulation of the activities of other antioxidants
(Nyska et al., 2004, 056379; Section 4.6.3). Given the significant mortality of the male mice in
the study, the doses received by the female mice may not have represented a sufficient dose to
induce tumors (i.e., maximum tolerated dose); and (4) Hemosiderin deposition is a key event in
the MOA for the formation of liver tumors in male mice. As stated previously, a key event
represents an obligatory (rate limiting) step or a biomarker of such an event. This would not
preclude the use of hemosiderin deposition as a biomarker of such an event, as long as a dose-
response and temporal relationship are observed to indicate it as an obligatory step in the MOA.
C3. EPA has proposed a MOA for female mouse forestomach tumors involving
metabolism, irritation and regenerative proliferation leading to tumor induction as key
events best supported by the data. Please provide detailed comments on whether this
analysis regarding the MOA for forestomach tumors is scientifically sound, and
transparently and objectively described in the ToxicologicalReview. Considerations
include the scientific support regarding the plausibility for the hypothesized MOA and the
characterization of uncertainty regarding this MOA.
Comments:
Five of seven reviewers generally agreed with the conclusions of the analysis regarding
the MOA for forestomach tumors. One reviewer abstained from commenting while another
focused their comments on editorial aspects stating that the discussion on the MOA was too
speculative and that steps 5 and 6 should be deleted. A few reviewers provided comments to
refine the supporting text and conclusions for the MOA.
Response:
The overall conclusions regarding the MOA for forestomach tumors were not modified.
Editorial and clarification changes to refine the supporting text were made to the document.
A-15

-------
While step 5 of the proposed sequence of events leading to forestomach tumors seems
redundant, it is necessary to illustrate that the development of tumors is not a single exposure
scenario but requires a repeated cycle of injury and repair. Regarding step 6, EGBE-specific
evidence of ROS modulation in endothelial cells and hepatocytes is a biologically plausible step
within the hypothesized MO A, even though it is not definitively supported. It is important to
note that key events within the MOA may have different levels of supporting scientific evidence.
C4. EPA has not proposed a MOA for the female rat pheochromocytomas of the adrenal
medulla. NTP rated the female rat pheochromocytomas as providing equivocal evidence of
carcinogenic activity and the pathology report expressed concern as to whether the
observed tumors met the criteria used to diagnose pheochromocytomas. For these reasons,
this tumor was not given significant weight in the qualitative or quantitative assessment of
EGBE cancer potential. Please provide detailed comments on whether this analysis
regarding the female rat pheochromocytomas is scientifically sound, and transparently and
objectively described in the Toxicological Review. Please comment on whether and the
extent to which the female rat pheochromocytomas are adequate to support alternative
analyses of qualitative and quantitative cancer risks to humans and discuss approaches to
consider if such analyses are warranted.
Comments:
Six of seven reviewers generally agreed with the conclusions of the analysis regarding
female rat pheochromocytomas. No dissenting scientific opinions were submitted that would
warrant alternative analysis. One reviewer did suggest a possible extension of the NTP
conclusion that pheochromocytomas represented equivocal findings. This reviewer also
suggested the possibility of including a linear low-dose extrapolation for this tumor type as an
academic exercise to explore the range of possibilities but did not recommend its use for
determining risk. One reviewer abstained from commenting.
Response:
As stated in the charge question, U.S. EPA has not proposed a MOA for the female rat
pheochromocytomas of the adrenal medulla. NTP concluded that the data for the female rat
pheochromocytomas provided equivocal evidence of carcinogenic activity and the pathology
report expressed concern as to whether the observed tumors met the criteria used to diagnose
pheochromocytomas. For these reasons, this tumor was not given significant weight in the
qualitative or quantitative assessment of EGBE cancer potential.
Thus, cancer quantification using a linear low-dose extrapolation for pheochromocytomas
was not included in the Toxicological Review.
A-16

-------
C5. Please comment on the choice of the nonlinear threshold approach for the quantitative
assessment of the carcinogenic potential of EGBE. Please comment on whether this
approach is scientifically sound, and transparently and objectively described. Please
comment on whether the example calculations using linear low-dose extrapolation for
cancer as discussed in Section 5.4.1 represent useful characterizations of the potential
quantitative uncertainty associated with exposure to EGBE. Please comment on whether
the linear analysis should be presented as an alternative to the threshold approach
considering the Agency conclusion that EGBE is not likely to be carcinogenic to humans at
expected exposure concentrations.
Comments:
Six of seven reviewers that commented supported the choice of the nonlinear threshold
approach. Most reviewers found the presentation of the linear low-dose extrapolation to be
informative for comparative purposes, but two reviewers felt it should not be presented as an
alternative to the threshold approach.
Response:
Based on the majority of comments, no change in the ToxicologicalReview is indicated.
A.3. PUBLIC COMMENTS
Comment:
The use of hemosiderin staining as the critical effect for RfC and RfD development is not
biologically appropriate and is inconsistent with the available dose-response data.
Response:
The use of hemosiderin staining as the critical effect for RfC and RfD development has
been peer reviewed by a panel of experts and found to be appropriately used. While some
increase in hemosiderin accumulation can be expected with advancing age, excessive
accumulation beyond that contributed by age is considered pathological (Muller et al., 2006,
597291). While the hemolytic effects appeared to be among the earliest effects from EGBE
exposure, the hemosiderin deposition endpoint was selected as the critical effect. This effect was
found to occur in both species and genders of animals tested, with rats being the more sensitive
species; the effect also occurred in the 14-week subchronic NTP inhalation study. The suggested
MO A of EGBE-induced liver effects is based on the observation that the hemolytic effects led to
compensatory erythropoiesis and significant increases in blood degradation products, including
an increased accumulation of hemosiderin in the liver Kupffer cells of EGBE-exposed animals.
The hemosiderin accumulation seen in the Kupffer cells was found to increase in severity with
A-17

-------
increasing dose and exposure duration (Table 5-2), unlike the hemolytic endpoints, such as
decreased Hct, which did not progress from 3 to 12 months (Table 4-6). Thus, hemosiderin
deposition in Kupffer cells in the rat liver is believed to be a sequela to the hematologic effects.
Because of the progression of this effect with chronic exposure, hemosiderin is deemed to be the
most sensitive effect (Section 5.1.1). Additional support for the use of hemosiderin, as opposed
to the various hematological parameters, is the location of the effect. Kupffer cell accumulation
of hemosiderin in the liver is consistent with the pathological findings of liver
hemangiosarcomas and hepatocellular carcinomas found in EGBE-exposed male mice.
Comment:
The use of male rat data as the POD is inappropriate because the available data
convincingly demonstrate that female rats are more sensitive to the hemolytic effects of EGBE.
Response:
U.S. EPA agrees that female rats are more sensitive to the hemolytic effects of EGBE.
However, hemolysis is one step in a complex process leading to the critical endpoint,
hemosiderin accumulation. Male rats have been shown to be more sensitive to hemosiderin
accumulation than females (NTP, 2000, 196293). The fit statistics and BMC information
derived from the dichotomous models available in the BMD software as applied to the male and
female rat hemosiderin staining data versus AUC BAA are shown in Table 5-5. All models were
fit using restrictions and option settings suggested in the U.S. EPA BMD technical guidance
document (U.S. EPA, 2000, 052150). The best model fit to these data, as determined by visual
inspection, examination of low dose model fit (i.e., scaled residual for the dose group closest to
the BMD), and comparison of overall fit (i.e., AIC values) was obtained using a multistage
model (1st degree) for the male response data and a Log-Logistic model for the female response
data. The male rat BMCio was 196 [j,mol-hour/L and the BMCLio was determined to be 133
[j,mol-hour/L, using the 95% lower confidence limit of the dose-response curve expressed in
terms of the AUC for BAA in blood. The BMCio and BMCL io values for the female rat were
determined to be 425 and 244 [j.mol-hour/L, respectively. Assuming continuous exposure (24
hours/day), the Corley et al. (1997, 041984) PBPK model was used to back-calculate HECs of
3.4 ppm (16 mg/m3) from the male rat data and 4.9 ppm (24 mg/m3) from the female rat data
(Section 5.1.2.1). If you combine the males and females for the BMC analysis, the result is
<10% of the original BMCLio for males alone. This indicates that there is an inconsistency with
at least one of the data sets in the female study. Based on all the information, the use of male
rats in the assessment is warranted.
Comment:
A-18

-------
The choice of dose metric, and the application of PBPK modeling in the derivation of
candidate RfCs based on hematological data should be re-examined. The selection of an internal
dose metric to be used in the dose-response assessment using the incidence of hemosiderin
staining is not adequately justified.
The Lee et al. (1998, 041983) model has been incorrectly applied in the derivation of the
Cmax BAA values reported for female rats in Table 5-4.
Similar discrepancies (current draft vs. 1999 document) are evident in the BMD
modeling based on hematological endpoints.
Response:
The AUC for BAA in arterial blood was selected as the appropriate measure of dose due
to evidence for a causal association between BAA and hemolysis, between hemolysis and the
accumulation of hemosiderin in the liver, and between hemosiderin accumulation in the liver and
increased incidence of liver hemangiosarcoma. AUC is considered to be a more appropriate
response measure because hemosiderin pigmentation increases in incidence and severity with
increased duration (subchronic to chronic) and is believed to be the result of the cumulative
exposure to EGBE/BAA as opposed to a peak exposure. However, for comparative purposes, a
BMD analysis was done on the critical endpoint to determine the impact that choosing Cmax of
BAA in blood rather than AUC of BAA in blood would have had on the BMCLi0(hec) derivation.
If Cmax had been used as the dose measure, the BMCLi0(hec) value would have been 39 mg/m3,
approximately 2.4-fold higher than the 16 mg/m3 BMCLio(hec) value derived using AUC as the
dose measure (Section 5.3.2). The choice of hemosiderin as the critical effect warrants the use of
AUC as the dose metric.
The reviewer is correct; the arterial blood BAA Cmax estimates for female rats reported in
Table 5-4 were in error. Table 5-4 was revised using the most recent PBPK model of Corley et
al. (2005, 100100). which replaced several of the assumptions utilized by Lee et al. (1998,
041983) with measured values for protein binding, partition coefficients, metabolism, and renal
clearance. For comparison, the Lee et al. (1998, 041983) simulations along with the simulations
using the Corley et al. (2005, 100100) model are as follows:
Exposure	Female rat
concentration	body weight 	BAA in arterial blood (uM)	
(ppm)	(g) Lee et al. (1998, 041983) Corley et al. (2005, 100100)
31.25
216
285
167
61.5
211
603
408
125
214
1,243
1,091
250
210
1,959
2,752
500
201
4,227
6,483
A-19

-------
As the reviewer pointed out, these revised Cmax estimates are now consistent with the
peak blood concentrations of BAA observed following 3 months of exposure to the three
concentrations used in the NTP chronic inhalation bioassay (31.25, 61.5, and 125 ppm) reported
by Dill et al. (1998, 041981). Table 5-3 and Appendix A were also revised to include a
description of the Corley et al. (2005, 100100) PBPK model.
The BMD calculations (Table 5-5) were revised using the Corley et al. (2005, 100100)
estimates of arterial blood BAA Cmax values for female rats as described above. The HEC
calculations were again based upon the Corley et al. (1997, 041984) model.
Comment:
The use of inhalation data to derive the RfD is inappropriate. The current IRIS Review
of EGBE (U.S. EPA, 1999, 597365) http://www.epa.gov/ncea/iris/toxreviews/05Q0tr.pdf selects
a 91-day drinking water study in rats as the principal study for the derivation of the RfD.
Because information available from chronic inhalation studies with EGBE indicates that the
primary effect, hematological changes, does not become more severe with prolonged exposures,
the 91-day drinking water study in rats remains the appropriate POD for deriving the RfD. Use
of this study would preclude the need for a route-to-route extrapolation from chronic inhalation
data, with the substantial errors that can be associated with this type of analysis (Pauluhn, 2003,
100188).
In evaluating the results of the route-to-route extrapolation, a comparison to what is
observed in the available oral studies is critical to determine whether the modeling results are
realistic or whether, instead, the modeling results contribute additional uncertainty to the derived
RfD. The route-to-route extrapolation in the draft IRIS Review is based on the observation of
male rat liver hemosiderin staining from the NTP (2000, 196293) inhalation study. With that
data set as the POD, PBPK modeling is used to derive the proposed new RfD of 0.14 mg/kg-day.
In comparing this RfD to the results from the 13-week oral study in rats (NTP, 1993, 042063). a
significant increase in hemosiderin staining was not observed in male rats until doses of 452
mg/kg-day were achieved or in female rats until a dose of 281 mg/kg-day was achieved
(Table 4-1). In addition, significant changes in hematological endpoints were observed at lower
doses (69 mg/kg-day). The doses associated with significant increases in hemosiderin staining
are orders of magnitude above the POD of 1.4 mg/kg-day, suggesting that the use of the route-to-
route endpoint is not appropriate. Additional analyses are needed to determine why a difference
in response by route of exposure would be observed.
While route-to-route extrapolation is a valuable tool for chemicals for which no adequate
study is available for a selected route of exposure, in the case of EGBE, an adequate oral study is
available and should be used to derive the RfD.
A-20

-------
Response:
Hematological endpoints indicative of hemolysis (e.g., RBC count) do not progress with
increasing duration of inhalation exposure, whereas the incidence of hemosiderin pigmentation
did progress considerably with chronic exposure (Table 5-2). A primary consideration with
respect to the NTP (1993, 042063) study was that it did not provide a sufficient duration of
exposure to EGBE to allow for maximal accumulation of hemosiderin deposition. In contrast,
hemosiderin accumulation was one of the most sensitive effects observed in the NTP (2000,
196293) chronic inhalation study of EGBE. Furthermore, uncertainty regarding the mechanism
of action of EGBE does not allow for a biologically-informed determination regarding which
hematological endpoints—changes in RBC count, reticulocyte count, MCV, Hct, and Hb—
observed in EGBE-exposed animals should be used to derive an RfC and, in the case of BMD
analysis, what a proper BMR level should be for the BMD derivation. Finally, hematologic
effects signified by changes in RBC count, reticulocyte count, MCV, HCT, and Hb are
considered precursor effects to the pathological finding of hemosiderin deposition (Section 4.5).
For these reasons and because the hemolytic effects of EGBE appear to be consistent between
oral and inhalation routes of exposure, an RfD has been derived via the application of a PBPK
model to perform a route-to-route extrapolation (Section 5.2.2) from the BMD analysis (Section
5.1.2) of hemosiderin pigmentation observed in the NTP (2000, 196293) chronic inhalation study
of EGBE. BMD/NOAEL analyses of hematologic endpoints and hemosiderin pigmentation
observed in the oral NTP (1993, 042063) subchronic study are provided below for comparison
purposes.
The 91-day drinking water study has been included in Chapter 5 for comparison
purposes. Alternative endpoints including the endpoint selected for the previous assessment are
presented graphically and discussed in a new section (5.2.4). Inhalation studies considered for
derivation of the RfC were used to supplement the oral database using the route-to-route
extrapolation (Section 5.2.2.1).
Comment:
The intrahuman and interspecies UFs applied in the derivation of the RfC and RfD are
greatly overprotective and should be reevaluated. The draft IRIS Review repeatedly (e.g., pages
50, 59, and 109) mentions that humans are much less sensitive to the toxic effects of EGBE than
are the rodents that provide the basis for derivation of the RfC and RfD. These differences in
sensitivity are not simply due to pharmacokinetic differences (which are addressed in the PBPK
modeling), but differences in inherent sensitivity (pharmacodynamics) as illustrated by the
marked differences in sensitivity of human and rodent blood cells in vitro to EGBE and BAA.
For example, the work of Udden (2002, 042 111) demonstrates about a 150-fold greater
sensitivity of rat blood cells than human blood cells to the effects of BAA on RBC deformability,
osmotic fragility, and hemolysis. Even potentially hypersensitive subgroups of the human
A-21

-------
population (the elderly, and patients with sickle cell disease or hereditary spherocytosis) show
similar resistance to these effects of BAA (Udden, 2002, 042 111).
Given the consistent, substantial difference in sensitivity between human and rats and the
data that is the basis for the RfD and RfC, there is no scientific justification for using a
pharmacodynamic UF as large as 1. The weight of the evidence supports the use of a fractional
value of perhaps 0.01, or even less.
Furthermore, the default UF of 10 for human variability is not needed. This factor is
typically applied to account for variations in human sensitivity or to be protective of sensitive
subpopulations. In this case, however, typical hypersensitive subgroups, such as the elderly, also
show resistance to the hematological effects of BAA, as do individuals with disease conditions
(patients with sickle cell disease or hereditary spherocytosis) that might be expected to make
them more sensitive (Udden, 2002, 042 111). While animal studies suggest that older animals are
more sensitive than neonates, and females are more sensitive than males, these have been shown
to be a reflection of differences in pharmacokinetics, not pharmacodynamics or sensitivity
(Corley et al., 2005, 100100). Based on these findings, the proposed 10-fold intraspecies UF is
clearly excessive. For further information, see the response for charge question A5.
Response:
The current evidence indicates that the human RBC response to EGBE exposure is
considerably less than that of rodents. However, this information comes from a relatively small
number of studies. The current UFa value in the ToxicologicalReview is 1 and considers the
limited number of studies available as well as the unknown effect of EGBE on the cellular events
leading to hemolysis in human populations. In addition, little is known of the long-term or
repeated exposure responses in humans to EGBE.
The UF for human variation has been assigned a value of 10. Some studies have shown
that in vitro exposure of RBCs from proposed sensitive populations, such as the elderly and those
with iron handling diseases, produces similar effects as in the general population. However, the
database is small and represents only acute or short-term exposures. The effects of longer or
repeated exposures to the general population as well as the proposed sensitive populations
remain to be determined. Therefore, the use of the default value for UFH is justified.
Comment:
There are several problems with BMD modeling. While benchmark modeling is the most
scientifically appropriate approach for determining the POD using the available noncancer data
for EGBE, there are several problems in the implementation of the procedure. The additional
documentation of the modeling results provided in Appendix B of the draft has several
deficiencies. The output provided in Appendix B for the hemosiderin modeling does not use the
AUC doses so is not an example of the output used to derive the RFC. Specifically, the
A-22

-------
multistage model output labeled "BMD Method for RFC: Hemosiderin deposition in male rats
versus AUC BAA, 2-year inhalation study (NTP, 2000, 196293)" in Appendix B does not use
the doses indicated in Table 5-6 as the AUC doses. The BMD and BMDL in this output are not
the values reported in Table 5-7. The same is also true of the log-logistic output labeled "BMD
Method for RfC: Hemosiderin deposition in female rats versus AUC BAA, 2-year inhalation
study, (NTP, 2000, 196293)" in Appendix B. The BMD and BMDLs reported in these outputs
are not given anywhere else in the document. In addition, in Table 5-7, the female rat multistage
(1-stage) output has the BMDL in the BMD column and an incorrect number in the BMDL
column (source unknown).
Response:
The reviewer is correct that the outputs of the BMD modeling do not match the cited
source studies. Older BMD outputs were inadvertently placed in the document appendix while
the correct values incorporated into the text. The BMD calculations have been revised and the
corrected outputs placed in the appendix.
A.4. ADDITIONAL EXTERNAL PEER REVIEW PANEL COMMENTS - SECOND
REVIEW IN RESPONSE TO REVISIONS AS INDICATED ABOVE
A.4.1. General Charge Questions
Comment:
One reviewer questioned the replacement of the term "adverse" with the terms
"significant" and "biologically significant" throughout the text.
Response:
The term "adverse" was generally replaced throughout the text to provide a more
meaningful description. Many of the instances where "biologically significant" was used in the
document has been replaced with the words "statistically significant."
A.4.2. Chemical-Specific Charge Questions
(A) Inhalation reference concentration (RfC) for EGBE
A2.
Comment:
One reviewer commented that hyaline membrane degeneration could be regarded as a
precursor lesion and, while not suggesting that this is a suitable critical endpoint in isolation, it
might be a useful endpoint for quantitative comparison with other endpoints.
Response:
A-23

-------
The potential use of hyaline membrane degeneration as endpoints was considered and
discussed in chapter 5 of the Toxicological Review. Based on the available literature, U.S. EPA
maintains the position that olfactory hyaline membrane degeneration is not suitable as a critical
endpoint. Based on the available literature for EGBE, U.S. EPA concludes that hyaline
membrane degeneration represents an adaptive response and not an adverse effect. Any
quantitative comparison between hyaline membrane degeneration and hemosiderin deposition
would therefore not be appropriate for comparative purposes.
Comment:
One reviewer felt the discussion on hemolytic anemia in Section 4.5 still required
revisions and provided alternative text. Specifically, the discussion on osmotic fragility and its
relationship to EGBE-induced hemolytic anemia required clarification.
Response:
The alternative text suggested by the reviewer has been incorporated into the discussion
on hemolytic anemia in Section 4.5 and the Toxicological Review was reviewed for other
occurrences of the phrase "osmotic fragility". The alternative text clearly discusses the available
information and limitation of what is known about the mechanism of EGBE-induced hemolytic
anemia.
A3.
Comment:
One reviewer questioned the selection of male rats as the most sensitive gender when the
data demonstrate that females are more sensitive to the hematological effects of EGBE and
display an increased incidence of hemosiderin deposition.
Response:
It is agreed that female rats are more sensitive to the hemolytic effects of EGBE.
However, hemolysis is one step in a complex process leading to the critical endpoint,,
hemosiderin accumulation. While the data from the subchronic study show females as more
sensitive, it is the 2-year chronic study that is the focus of our analysis. In that study, male rats
were shown to be more sensitive to hemosiderin accumulation than females (NTP, 2000,
196293). In addition, the cumulative blood concentration of the active metabolite BAA was
chosen as the internal dose metric, as opposed to the exposure concentration of EBGE. The fit
statistics and BMC information derived from the dichotomous models available in the BMD
software as applied to the male and female rat hemosiderin staining data versus AUC BAA are
shown in Table 5-5. All models were fit using restrictions and option settings suggested in the
U.S. EPA BMD technical guidance document (U.S. EPA, 2000, 052150). The best model fit to
A-24

-------
these data, as determined by visual inspection, examination of low dose model fit (i.e., scaled
residual for the dose group closest to the BMD), and comparison of overall fit (i.e., AIC values)
was obtained using a multistage model (1st degree) for the male response data and a Log-
Logistic model for the female response data. The male rat BMCio was 196 |j,mol-hour/L and the
BMCLio was determined to be 133 |j,mol-hour/L, using the 95% lower confidence limit of the
dose-response curve expressed in terms of the AUC for BAA in blood. The BMCio and BMCLio
values for the female rat were determined to be 425 and 244 |j,mol-hour/L, respectively.
Assuming continuous exposure (24 hours/day), the Corley et al. (1994, 041977) PBPK model
was used to back-calculate concentrations for the HECs of 3.4 ppm (16 mg/m3) from the male rat
data and 4.9 ppm (24 mg/m3) from the female rat data (Section 5.1.2.1). If the data for males and
females are combined for the BMC analysis, the result is <10% of the original BMCLio for
males alone. This indicates that there is a inconsistency with at least one of the data sets in the
female study. Based on all the information, the use of male rats in the assessment is warranted.
A5.
Comment:
One reviewer continued to question the use of a UFA of 1 when in their view the
scientific data support a UFA <1.
Response:
The basis for selection of a UFA of 1 is presented in the EPA response to the comment on
charge question A5. A new rationale has not been provided by the reviewer to alter the U.S.
EPA's selection of this value. The current UFa value in the Toxicological Review is 1. The UFa
accounts for a reduction from 3 for the toxicokinetic differences between animals and humans
through the use of a PBPK model for extrapolation of doses. Likewise, the toxicodynamic
portion was reduced from 3 to 1. The toxicological effect in question being the deposition of
hemosiderin is a key event of the MOA for the development of liver hepatomas in mice. To
implement a further reduction in the UFA (i.e., fractional) would logically indicate that a
preponderance of data are available to fully describe this key event (toxicodynamics) in both
animals and humans to the extent of describing why humans are less sensitive to the hemolytic
effects leading to hemosiderin deposition. This is not the case, although in vitro data
(Ghanayem, 1989, 042014; Udden, 2000, 042110; Udden, 2002, 042 111; Udden and Patton,
2005, 100213) do suggest humans are less sensitive than rodents to the hemolytic effects of
EGBE. Likewise, the few human studies (Carpenter et al., 1956, 066464; Haufroid et al., 1997,
042040) indicate the same finding. However, these studies (Carpenter et al., 1956, 066464;
Ghanayem, 1989, 042014; Haufroid et al., 1997, 042040; Udden, 2000, 042110; Udden, 2002,
042 111; Udden and Patton, 2005, 100213) do not characterize hemosiderin deposition, the key
event for the MOA. The current UFa value in the Toxicological Review is 1 and considers the
A-25

-------
limited number of studies available as well as the unknown effect of EGBE on the cellular events
leading to hemolysis in human populations. In addition, little is known of the long-term or
repeated exposure responses in humans to EGBE.
A6.
Comment:
One reviewer commented that our interpretation of his original comment on the UFd in
charge question A6 was inaccurate. The reviewer stated that he did not think the value of the
UFd should be higher but that it was inconsistent with the Agency's confidence statement.
Response:
The inaccuracies have been corrected in the reviewer responses for charge question A6.
The assigning of a value of 1 for the UFD reflects that the database contains a composite of the
basic studies needed for an overall toxicological assessment. However, U.S. EPA's conclusion
regarding the overall confidence for the derivation of a human health toxicity value (i.e., RfD,
RfC) considers the quality, strength, and adequacy of the principal study as well as the database.
The selection of a confidence level of medium - high is appropriate and also considers the
potential for effects in humans from repeated, long-term exposures has not been investigated.
A-26

-------
APPENDIX B. PBPK MODELS (CORLEY ET AL.[1994;1997;2005])
Corley et al. (1994, 041977) developed PBPK models for rats and humans with the
primary objective of describing the concentration of BAA in the target tissue (blood) of rats and
humans for use in risk assessment (Figure B-l). The models incorporate allometrically scalable
physiological and biochemical parameters (e.g., blood flows, tissue volumes, and metabolic
capacity) in place of the standard values for a 70-kg human. These parameters normalize
standard values to the actual body weights of the subjects in several human kinetic studies. The
physiology of humans under exercise conditions initially used by Johanson (1986, 006758) to
describe the kinetics of EGBE following inhalation exposures while exercising was maintained
in the model. The rat was included to expand the database for model validation and to assist in
interspecies comparisons of target tissue doses (BAA in blood). The formation of BAA from
EGBE was assumed to occur only in the liver and was simulated in a second model linked via
the formation of BAA.
Note: Hyperlinks to the reference citations throughout this document will take you to the NCEA HERO database (Health and
Environmental Research Online) at http://epa.gov/hero. HERO is a database of scientific literature used by U.S. EPA in the process of
developing science assessments such as the Integrated Science Assessments (ISA) and the Integrated Risk Information System (IRIS).
B-l

-------
Model for 2-Butoxyethanol
Model for 2-Butoxyacetic Acid
Inhalation
Exhalation
| iv infusion |
Lungs and
Arterial Blood
Lungs and
Arterial Blood
Rapidly Perfused
Organs
Rapidly Perfused
Organs
Slowly Perfused
Organs
Slowly Perfused
Organs
Fat
Fat
Skin
Skin
Muscle
Muscle
Liver
Liver
i k
i i
Metabolism
to BAA
Gastro-
intestinal
Tract
Gastrointestinal
Tract
Other
metabolites
Kidney
Drinking water
Urine
(Butoxyacetic acid)
Source: Modified from Corley et al. (1994, 041977).
Figure B-l. PBPK model of Corley et al. (1994, 041977).
The Corley et al. (1994, 041977) model included additional routes of exposure such as
gavage, drinking water, i.v. infusion, and dermal (liquids and vapor) to facilitate comparisons to
several published toxicokinetic studies utilizing these routes of administration. The formation of
BAA was assumed to occur only in the liver, using the rat liver perfusion data of Johanson et al.
(1986, 006758) scaled to the human. A second model was linked to the EGBE model
specifically to track the disposition of BAA following its formation in the liver. The kidney was
added to the BAA model because it is the organ of elimination for BAA. All other metabolic
routes for EGBE (formation of EG and glucuronide conjugate) were combined, as they were
B-2

-------
used only to account for the total disposition of EGBE in the rat metabolism studies and not for
cross-species extrapolations.
Contrary to observations in rats, Corley et al. (1997, 041984) found no evidence of
metabolites in urine that would indicate that humans form conjugates of EGBE or EG. Thus,
these pathways, which were lumped together in the model of Corley et al. (1994, 041977) to
simulate rat kinetic data, were eliminated for human simulations. The human blood:air partition
coefficient of 7,965, from Johanson and Dynesius (1988, 100157). was also used in the Corley et
al. (1994, 041977) model. In addition, the partition coefficients for both EGBE and BAA were
measured in human blood, rat blood, and rat tissues by using a modification of the Jepson et al.
(1994, 006570) technique for ultrafiltration. Human tissue:blood partition coefficients were
assumed to be equal to those of the rat. The skin:air partition coefficient, used to calculate the
dermal uptake of vapors, was assumed to be the same as the blood:air partition coefficient. With
the exception of the lung:blood partition coefficient for EGBE (11.3), the tissue:blood partition
coefficients ranged from 0.64 to 4.33 for EGBE and from 0.77 to 1.58 for BAA. Protein binding
of BAA in blood and saturable elimination of BAA by the kidneys were necessary components
to describe the BAA kinetic data in rats and humans as discussed above. Since no direct
measurements of protein binding were available, these parameters were arbitrarily set to the
molar equivalent values reported for phenolsulfonphthalein as described by Russel et al. (1987,
100197). Constants for the saturable elimination of BAA by the kidneys were then estimated by
optimization from the data of Ghanayem et al. (1990, 042016). where rats were administered
EGBE i.v. and the concentrations of BAA in blood were determined, following three different
dose levels. These parameters were then held constant (protein binding) or scaled by (body
weight)0'74 x (renal elimination) for all simulations.
Corley et al. (1997, 041984) published the results from a human dermal exposure study
and updated their 1994 human model. This study tested the hypothesis put forward by Johanson
and Boman (1991, 006759) that dermal absorption can be more significant than inhalation
absorptions by humans exposed to EGBE vapors. The results from this study verified the
predictions from the Corley et al. (1994, 041977) model that blood samples taken via finger-
prick sampling methods are confounded by locally high concentrations of EGBE on the surface
or within the skin compartment and do not represent system blood concentrations. This study,
and PBPK simulations have more recently been verified by Franks et al. (2006, 1001 17) and
Jones et al. (2003, 100161).
Lee et al. (1998, 041983) published an upgrade to the rat PBPK model of Corley et al.
(1994, 041977) and included female rats, male and female rats, and the effects of aging to
simulate the results from the kinetic studies conducted as part of the NTP 2-year bioassay (Dill et
al., 1998, 041981). In their model, Lee et al. (1998, 041983) adjusted the rates of metabolism of
EGBE to BAA in female rats, plasma protein binding of BAA, and the rates of renal active
transport of BAA into urine to describe the kinetics of EGBE and BAA following short-term and
B-3

-------
long-term exposures. This model was used in the 1999 EGBE Toxicological Review to calculate
the blood BAA Cmax values in female rats, which were used as the basis for RfC determination.
Corley et al. (2005, 100100) continued the work of Lee et al. (1998, 041983) by
experimentally measuring the blood and tissue partition coefficients for BAA, plasma protein
binding of BAA, liver metabolism of EGBE, and renal clearance of BAA in young and old, male
and female F344 rats and B6C3Fi mice to replace assumptions used in the rat and mouse PBPK
model of Lee et al. (1998, 041983). This revised PBPK model (Figure B-2) was used as the
basis for calculating the Cmax for BAA in the blood of female rats in the subchronic NTP
inhalation study as the basis for the current RfC. The HEC was then calculated from BMD
analysis of internal dose-response for hemolysis using the Corley et al. (1997, 041984) human
PBPK model. A summary of the female rat and human model parameters are shown in Table
B-l.
The model was based upon the PBPK model of Lee et al. (1998, 041983) with the
exception that (a) the spleen was combined with the rapidly perfused tissue group; (b) the i.p.
injection route was added to the EGBE submodel; (c) salivary glands were added; (d) the skin
compartment was separated into exposed and unexposed skin; and (e) metabolism of BAA to
conjugates or CO2 were included.
B-4

-------
Butoxyethanol Butoxyacetic Acid


Lungs & U
Arterial Blood
Rapidly Perf
Organs
Slowty Perf
Organs
Dermal
Liver
Liver
Gl Trad
BAA
Kidney
Kidney
Fat
Fat
Lungs &
Arterial Blood
Slowly Perf
Organs
'Exposed Skin
Skin
Salivary Gland
Skin
Gavage be giuc eg
Urine BAA-Conj & C02
Source: Corley et al. (2005, 100100)
Figure B-2. PBPK model of Corley et al. (2005, 100100).
B-5

-------
Table B-l. Selected parameters used in the PBPK model for EGBE
developed by Corley et al. (1997, 041984: 2005, 100100)
Parameter
Human
Young female rat
Old female rat
Weights
Body weight (kg)
70
0.20
0.34
Liver (%BW)
3.14
3.22
2.52
Kidney (EGBE model)
N/A
0.69
0.694
(BAA model) (%BW)
0.44
0.69
0.694
Rapidly perfused (EGBE model)
3.71
4.39
4.39
(BAA model) (%BW)
3.27
4.39
4.39
Slowly perfused (% BW)
9.4
24.6
14.3
Flows
Alveolar ventilation (L/hr)
347.9
5.47
8.10
Cardiac output (COP) (L/hr)
347.9
5.47
8.10
Liver (% COP)
25.0
18.3
18.3
Kidney (% COP)
25.0
14.1
14.1
Rapidly perfused (EGBE model) (% COP)
50.0
23.3
23.3
(BAA model)
25.0
23.3
23.3
Slowly perfused (% COP)
2
2
2
Partition coefficients
Blood/air
7,965
7,965
7,965
Liver/blood
1.46
1.48
1.48
(BAA model)
1.30
0.66
0.66
Kidney/blood (EGBE model)
1.83
1.83
1.83
(BAA model)
1.07
0.87
0.87
Rapidly perfused/blood
1.46
1.47
1.47
(BAA model)
1.30
0.66
0.66
Slowly perfused/blood
0.64
0.65
0.65
(BAA model)
1.31
0.54
0.54
B-6

-------
Parameter
Human
Young female rat
Old female rat
Metabolic constants
EGBE to BAA



VmaxlC (mg/hr/kg BW)
375
213
189
Kml (mg/L)
26.9
20.1
20.1
EGBE to others (human only)



Vmax2C (mg/hr/kg BW)
5


Km2 (mg/L)
0.5


EGBE to EG (rat only)



Vmax2C (mg/hr/kg BW)

3.3
3.3
Km2 (mg/L)

2.7
2.7
EGBE to glucuronide conjugate (rat only)



Vmax3C (mg/hr/kg BW)

30
30
Km3 (mg/L)

55.7
55.7
BAA to C02 (rat only)



Vmax4C (mg/hr/kg BW)

2.6
2.6
Km4 (mg/L)

31.8
31.8
Plasma protein binding of BAA
P (binding sites; mg/L)
164
n/a
n/a
Kd (dissociation constant; mg/L)
46
n/a
n/a
Bind (unitless fraction)
n/a
0.298
0.433
Renal elimination of BAA
VmaxEC (mg/hr/kg BW)
20.4
4.0
10.8
KmEC (mg/L)
21.9
40.0
40.0
B-7

-------
APPENDIX C. RfD AND RfC DERIVATION OPTIONS
C.l. DERIVATION OF THE RfC
C.l.l. Potential RfC Derivations Based on Hematologic Data
The lowest NOAEL/LOAEL observed in any subchronic or chronic study of EGBE is the
31 ppm LOAEL for hematologic effects observed in the NTP study (2000, 196293) in rats. In
order to estimate a corresponding human equivalent exposure, an internal dose metric associated
with this exposure level is estimated and a PBPK model is used to estimate the human exposure
level that would result in that internal dose.
Initially, it was important to determine what estimate of internal dose (i.e., dose metric)
could serve as the most appropriate metric for the effects under consideration. PBPK models of
Lee et al. (1998, 041983) and Corley et al. (1994, 041977; 1997, 041984) are capable of
calculating several measures of dose for both EGBE and BAA, including:
CmaxC The peak concentration of EGBE or BAA in the blood during the exposure
period;
AUCC Represents the cumulative product of concentration and time for EGBE and BAA
in the blood.
Dill et al. (1998, 041981) published measured AUC, but not Cmax, blood concentrations
of EGBE and its principal metabolite BAA at various exposure durations in both genders of
B6C3Fi mice and F344 rats exposed to the same concentrations used in the NTP (2000, 196293)
chronic studies. Cmax values would need to be derived from a PBPK model. Two pieces of
information were used to select Cmax for BAA in the blood as the more appropriate dose metric
for the main hemolytic endpoint associated with this LOAEL. First, as discussed in Section 4.5,
there is convincing evidence to indicate that an oxidative metabolite, BAA, is the causative agent
for EGBE-induced hemolysis (Carpenter et al., 1956, 066464; Ghanayem et al., 1987, 041608;
Ghanayem et al., 1990, 042016). With this in mind, dose metrics for BAA in blood appear to be
more appropriate than those for EGBE in blood, since they are more closely linked
mechanistically to the toxic response. Second, EGBE-induced hemolysis appears to be
dependent upon the dose rate. Ghanayem et al. (1987, 066470) found that gavage doses to F344
male rats of 125 mg/kg EGBE resulted in hemolytic effects including reduced RBC count, Hb,
and Hct, as well as kidney pathology (Hb casts and intracytoplasmic Hb). However, it should be
Note: Hyperlinks to the reference citations throughout this document will take you to the NCEA HERO database (Health and
Environmental Research Online) at http://epa.gov/hero. HERO is a database of scientific literature used by U.S. EPA in the process of
developing science assessments such as the Integrated Science Assessments (ISA) and the Integrated Risk Information System (IRIS).
c-i

-------
noted that hemolytic effects were not reported at a similar acute drinking water dose of 140
mg/kg (Medinsky et al., 1990, 041986). While a drop in RBC count and Hb (9 and 7%,
respectively) was noted in F344 male rats after 1 week of drinking water exposure to 129 mg/kg-
day EGBE, dose-related kidney pathology was not observed in these rats, even after 13 weeks of
drinking water exposure up to 452 mg/kg-day EGBE (NTP, 1993, 042063). Consistent with the
hypothesis that exposure concentration plays a larger role than exposure duration for EGBE-
induced hemolytic effects, hematological endpoints indicative of hemolysis do not progress with
increased inhalation duration (Section 5.1.1). Corley et al. (1994, 041977) also suggested that
Cmax is a more appropriate dose metric for the hemolytic effects of EGBE than AUC.
The PBPK models developed for EGBE are briefly summarized in Table C-l. Shyr et al.
(1993, 006766) and Johanson (1986, 006758) do not address BAA distribution, and are only
parameterized for humans and rats, respectively. In the 1999 EGBE ToxicologicalReview
(U.S. EPA, 1999, 597365). the model described by Lee et al. (1998, 041983) was determined to
be the most appropriate model for the estimation of rat and mouse internal doses following
inhalation exposure. Since the 1999 Toxicological Review (U.S. EPA, 1999, 597365). Corley et
al. (2005, 100100) published a revision to the Lee et al. (1998, 041983) model for rats and mice
where several assumptions used by Lee et al. (1998, 041983) were replaced with measured
values (e.g., protein binding, partition coefficients, metabolism rate constants for multiple
pathways, and renal clearance) as a function of species, gender, and age. That model is used
here to estimate the Cmax of BAA in blood following inhalation exposure to female rats, the more
sensitive gender. For transparency, the results from both PBPK models are presented in Table
C-2 below. However, only the values derived from the Corley et al. (2005, 100100) model were
utilized in the derivation of the LOAEL and BMD in Sections C. 1.1.1 and C. 1.1.2. The human
PBPK model of Corley et al. (1994, 041977; 1997, 041984) was then used to obtain estimates of
human inhalation exposure concentrations associated with the female rat BAA blood
concentrations.1 Established U.S. EPA (2006, 194568) methods and procedures were used to
review, select, and apply these chosen PBPK models.2
lrThe basic components of the Corley model are summarized in Appendix B.
2EPA notes that the review of the PBPK models was conducted prior to their use in the 1999 EGBE toxicological review.
C-2

-------
Table C-l. Summary of PBPK models
Model
Species
Routes of exposure
Comments
Johanson (1986. 006758s)
Human
Inhalation
BAA not addressed
Shyretal. (1993,
0067661
Rat
Inhalation, oral, dermal
BAA excretion
Corley et al. (1994,
041977; 1997. 0419841
Rat and human
Inhalation, oral,
dermal, i.v.
BAA distribution and excretion; male rats
only
Lee et al. (1998. 0419831
Rat and mouse
Inhalation
BAA distribution and excretion; males and
females
Corley et al. (2005,
1001001
Rat and mouse
Inhalation, oral,
dermal, i.p., i.v.
Age-dependent BAA distribution,
metabolism, and excretion, males and females
Franks et al. (2006,
1001171
Human
Inhalation and dermal
Extended Corlev et al. (1997. 0419841 model
to include bladder compartment for human
biomonitoring studies
Table C-2. Model estimates of BAA blood levels in female rats following
inhalation exposures
Exposure
concentration (ppm)
Female rat body weight (g)
BAA in arterial blood Cmax in
female rats (juIVI)
Lee et al. (1998.0419831
BAA in arterial blood Cmax
in female rats (jiM)
Corlev et al. (2005.1001001
31
216
285
167
61.5
211
603
408
125
214
1,243
1,091
250
210
1,959
2,752
500
201
4,227
6,483
Source: Lee et al. (1998, 0419831. Model results used in the 1999 EGBE Toxicological Review are included for
comparison to the updated model of Corley et al. (2005, 1001001.
C.l.1.1. NOAEL/LOAEL Method Applied to Hematologic Data
A five-step procedure was used to calculate the LOAEL HEC:
Step 1: Calculate the internal dose surrogate (Cmax for BAA in blood) corresponding to
female rat LOAEL (Corley et al., 2005, 100100) by using the actual experimental
exposure regimen (6 hours/day, 5 days/week) in model simulations.
Female rat LOAEL = 31 ppm
Cmax BAA = 167 [iM
Step 2: Verify that steady state was achieved (e.g., no change in BAA Cmax as a result of
prolonging the exposure regimen).
There were no changes in the Cmax of BAA in blood during any 24-hour simulation
period, using a 6 hour/day, 5 day/week exposure regimen at the female rat LOAEL,
indicating that steady state was achieved.
C-2

-------
Step 3: Simulate the internal dose surrogate by calculating the Cmax for BAA in blood for
humans continuously exposed (24 hours/day, 7 days/week) to varying concentrations
of EGBE.
Table C-3. Estimated Cmax for BAA in blood for humans
continuously exposed to varying concentrations of EGBE
Concentration of EGBE
Cmax BAA in blood
in air (ppm)
(HM)
1
2.6
5
13.0
10
26.1
20
52.9
50
137.1
100
295.0
200a
732.3
"Steady-state concentrations are not achieved within 5 days of continuous exposure to EGBE
concentrations >200 ppm.
Source: Corley et al. (1994, 041977: 1997, 0419841
Step 4: Calculate the LOAELrec of EGBE for continuous human exposure in air that
resulted in the same internal dose (Cmax of BAA) in blood calculated for the animal
study in step 1.
Female rat Cmax BAA =167 [jM
HEC continuous exposure = 56 ppm (calculated by regression of internal dose
versus the concentration of EGBE in air from step 3).
Step 5: Convert the EGBE exposure units from ppm to mg/m3
LOAELrec (mg/m3) = conversion factor x LOAELrec (ppm)
= 4.84 (mg/m3) ^ (ppm) x 56 ppm
= 271 mg/m3
C.l.1.2. BMC Method Applied to Hematologic Data
For the purposes of deriving an RfC for EGBE from hemolytic endpoints, both MCV and
RBC count response data were evaluated in female rats from the 14-week subchronic NTP
(2000, 196293) study (see Section 5.1.1). BMCs derived for these same hemolytic endpoints in
male rats of this study were approximately twofold higher than for female rats (data not shown).
The current BMD technical guidelines (U.S. EPA, 2000, 052150) suggest the use of 1 SD
from the control mean for the BMR level for continuous data in the absence of additional
information. Because the hemolytic endpoints are continuous measurements that have a
relatively small historical variance in rats, and because low-dose responses for these endpoints
C-3

-------
were generally near or within 5% of the control mean, the BMCL05 was considered to be a more
appropriate POD for derivation of the RfC (U.S. EPA, 1995, 005992; U.S. EPA, 2000, 052150).
The steepest concentration-response curves (and the lowest BMCL05 estimate) were obtained for
decreased RBC count in female rats, and a 5% change was found to be statistically significant.
Higher levels of response (e.g., > 10% reductions) are in the exposure range where other more
severe responses related to anemia occur, such as MCV increases and increased reticulocyte
counts. Lower levels of response, for instance, 1 SD from the control mean, an approximate 2%
reduction for these data, are relatively distant from the observable data and other responses
related to anemia. Cmax for BAA in arterial blood of rats was determined by using the PBPK
model of Corley et al. (2005, 100100). Dermal exposure to EGBE vapor was not considered in
the predicted blood levels because the estimated relative contribution of the skin to the total
uptake of unclothed humans exposed to 25 ppm EGBE for 8 hours ranged from only 4.6 to
27.5%, depending on temperature, humidity, and exercise level (Corley et al., 1997, 041984).
Thus, dermal uptake is predicted to contribute <10%, even if 50% of an individual's skin is
exposed. The results of this modeling effort are summarized in Table C-2.
All BMD analyses were performed using models in U.S. EPA BMDS, version 1.4.1c
(U.S. EPA, 2000, 052150). The fit statistics and BMC information derived from the continuous
models available in the BMD software as applied to the female rat RBC count data versus Cmax
BAA are shown in Table C-4. The best model fit to these data, from visual inspection and
comparison of AIC values, was obtained using the Hill model. The BMCL05 was determined to
be 133 [xM using the 95% lower confidence limit of the dose-response curve expressed in terms
of the Cmax for BAA in blood. The graphic and textual output from the Hill model run is
displayed after Table C-4. BMCLSd values are provided for comparative purposes. The Corley
et al. (1997, 041984) PBPK model was used to back-calculate a HEC of 46.5 ppm (225 mg/m3)
from the 133 [xM BMCL05, assuming continuous exposure (24 hours/day).
C-4

-------
Table C-4. Comparison of BMC/BMCL values for female rat RBC count
data from a 14-week subchronic inhalation study3, using modeled blood Cmax
(3 months) of the EGBE metabolite BAA as a common dose metric
Model
BMC 05 (jtM)
BMC L0s
(jiM)
bmcsd
(jiM)
bmclsd
(jiM)
/7-Value
AICa
Scaled
residualb
Polynomial (2°)
495.468
460.317
357.566
301.089
<0.0001
-66.057235
-1.51
Power0
119.313
82.782
29.4231
16.5251
0.022
-106.196640
1.29
Hillcd
189.394
133.005
72.9703
38.8183
0.621
-111.611016
-0.287
aTo obtain adequate model fits, the high dose group data was dropped from the analysis.
bAIC = -2L + 2P, where L is the log-likelihood at the maximum likelihood estimates for the parameters and P is the
number of model degrees of freedom (usually the number of parameters estimated).
°%2 residual (measure of how model predicted response deviates from the actual data) for the dose group closest to
the BMC scaled by an estimate of its SD. Provides a comparative measure of model fit near the BMC. Residuals
that exceed 2.0 in absolute value should cause questioning of the model fit in this region.
dPower and Hill models were run with power terms unrestricted to obtain adequate fit; estimates of the power terms
were 0.56 and 0.97 for the Power and Hill model, respectively.
"Model choice based on adequate p-valuc (>0.1), visual inspection, low AIC, and low (absolute) scaled residual.
C-5

-------
Hill Model with 0.95 Confidence Level
Hill
8.5
8
7.5
7
6.5
6
BMDL
BMD
0
500
1000
1500
2000
2500
dose
16:26 02/27 2009
Source: NTP (2000, 196293)
Figure C-l. Hill Model run of female rat RBC count versus Cmax BAA from
a 14-week inhalation study.
Hill Model. (Version: 2.14; Date: 06/26/2008)
Input Data File: C:\Usepa\BMDS21Beta\Data\EGBE\femalerat\2Hilrbchil.(d)
Gnuplot Plotting File: C:\Usepa\BMDS21Beta\Data\EGBE\femalerat\2Hilrbchil.plt
Fri Feb 27 16:26:00 2009
BMDS Model Run
The form of the response function is:
Y[dose] = intercept + v*doseAn/(kAn + doseAn)
Dependent variable = RBC
Independent variable = Cmax
C-6

-------
rho is set to 0
Power parameter 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 = 0.0357308
rho = 0 Specified
intercept = 8.48
v= -2.41
n = 1.34368
k = 775.437
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been specified by the
and do not appear in the correlation matrix )
alpha intercept
v
n
k
alpha 1 9.2e-008 -5.9e-008 -8.5e-008 2.5e-008
intercept 9.2e-008 1 -0.36 -0.47 0.23
v -5.9e-008 -0.36
1 0.95 -0.99
n -8.5e-008 -0.47 0.95
1 -0.92
k 2.5e-008 0.23 -0.99 -0.92
1
Parameter Estimates
C-7

-------
95.0% Wald Confidence Interval
Variable	Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha	0.0323151	0.00646302 0.0196478 0.0449824
intercept	8.47617	0.0566598 8.36512	8.58722
v	-3.86075	0.809864 -5.44805 -2.27344
n	0.971125	0.157912 0.661623	1.28063
k	1634.54	734.333 195.274	3073.81
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
10
8.48
8.48
0.158
0.18
0.0674
167
10
8.08
8.1
0.221
0.18
-0.287
408
10
7.7
7.68
0.253
0.18
0.353
1091
10
6.91
6.92
0.158
0.18
-0.175
2752
10
6.07
6.07
0.126
0.18
0.0415
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
C-8

-------
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model	Log(likelihood)	#Param's AIC
A1	60.927568	6 -109.855137
A2	64.091535	10 -108.183070
A3	60.927568	6 -109.855137
fitted 60.805508 5 -111.611016
R -18.641036 2 41.282072
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.)
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
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1
Test 2
Test 3
Test 4
165.465	8	<.0001
6.32793	4	0.176
6.32793	4	0.176
0.244121	1	0.6212
C-9

-------
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 = 0.05
Risk Type = Relative risk
Confidence level = 0.95
BMD = 189.394
BMDL = 133.005
C-10

-------
C.1.2. BMC Method Applied to Hemosiderin Data
For the purposes of deriving an RfC for EGBE, hemosiderin staining data were evaluated
in male and female rats from the 2-year chronic study by NTP (2000, 196293). The current
BMD technical guidelines (U.S. EPA, 2000, 052150) suggest the use of 10% extra risk as a
BMR level for quantal data as this is at or near the limit of sensitivity in most cancer bioassays
and in some noncancer bioassays as well. Because the hemosiderin staining endpoint was
observed in control animals and a 10% increase in incidence was within the observable range of
the data, 10% extra risk was considered an appropriate BMR and a BMCLio an appropriate POD
for derivation of the RfC (U.S. EPA, 1995, 005992; U.S. EPA, 2000, 052150). All BMD
assessments in this review were performed using U.S. EPA BMDS version 1.4.1c.
The AUC was selected as the appropriate dose metric due to the nature of the endpoint,
hemosiderin deposition. This endpoint increased in severity with increased duration (subchronic
to chronic) and is believed to be the result of the cumulative exposure to EGBE as opposed to a
peak event. Table C-5 reports AUC BAA blood concentrations measured at 12 months3
published by Dill et al. (1998, 041981) in both genders of B6C3Fi mice and F344 rats exposed to
the same concentrations used in the NTP (2000, 196293) chronic studies of these test animals.
3Dill et al. (1998, 041981) also reported 18 month data, but due to the smaller number of animals and higher variability in this data the 12
month data were used for the purposes of this analysis.
C-ll

-------
Table C-5. AUC BAA blood concentrations measured at 12 months in both
genders of B6C3Fi mice and F344 rats
Exposure concentration (ppm)
Gender

AUCbaa (jimol-hr/L)a
n
Mean
SE
Rats
31.2
Male
7
358.3
16.6

Female
5
638.8
18.7
62.5
Male
6
973.0
86.2

Female
9
1,128.9
50.9
125
Male
9
2,225.6
71.1

Female
12
3,461.8
154.8
Mice
62.5
Male
10
1,206.6
205.6

Female
12
1,863.6
112.4
125
Male
9
2,819.8
685.1

Female
6
5,451.6
508.9
250
Male
10
17,951.5
1,770.4

Female
11
18,297.1
609.7
"Authors reported AUC values in terms of (ig-min/g, which were converted to units consistent with the PBPK model
of (imol-hr/L by dividing by 60 min/h and 132.16 g/mol and multiplying by 1,060 g/L.
Source: Dill et al. (1998, 0419811.
The fit statistics and BMC information derived from the dichotomous models available in
the BMD software as applied to the male and female rat hemosiderin staining data versus AUC
BAA are shown in Table C-6. All models were fit using restrictions and option settings
suggested in the U.S. EPA BMD technical guidance document (U.S. EPA, 2000, 052150). The
best model fit to these data, as determined by visual inspection, examination of low dose model
fit (i.e., scaled residual for the dose group closest to the BMD), and comparison of overall fit
(i.e., AIC values) was obtained using a multistage model (1st degree) for the male response data
and a Log-Logistic model for the female response data. The male rat BMCio was 196 [j,mol-
hour/L and the BMCLio was determined to be 133 [j,mol-hour/L, using the 95% lower confidence
limit of the dose-response curve expressed in terms of the AUC for BAA in blood. The BMCio
and BMCLio values for the female rat were determined to be 425 and 244 [j,mol-hour/L,
respectively. The graphic and textual output from the model runs that resulted in these male and
female rat BMCio and BMCLio estimates are displayed below, after Table C-7. Assuming
continuous exposure (24 hour/day), the Corley et al. (1997, 041984) PBPK model was used to
back-calculate HECs of 3.4 ppm (16 mg/m3) from the male rat data and 4.9 ppm (24 mg/m3)
from the female rat data.
C-12

-------
Table C-6. Comparison of BMC/BMCL values for male and female rat liver
hemosiderin staining data from inhalation chronic study using measured
blood AUC (12 months) of the EGBE metabolite BAA as a common dose
metric
Model
BMCio (jimol-hr/L)
BMCL10 (nmol-hr/L)
/7-Value
AICa
Scaled residualb
Male rats
Multistage-lst degree0
196.252
133.141
0.8680
247.234
0.441
Gamma0
196.253
133.141
0.8680
247.234
0.441
Logistic
259.296
192.773
0.7692
247.476
0.526
Log-Logistic
166.376
69.3279
0.5623
249.283
0.313
Probit
271.525
205.882
0.7450
247.54
0.517
Log-Probit
368.336
241.992
0.6309
247.876
0.765
Weibull0
196.253
133.141
0.8680
247.234
0.441
Female rats
Multistage-lst degree
122.166
214.555
0.0698
218.868
-1.945
Gamma
316.635
134.02
0.0554
219.229
-1.238
Logistic
273.693
221.689
0.0993
218.188
-1.294
Log-Logisticc
424.527
243.69
0.1533
217.526
-0.896
Probit
291.017
241.206
0.0683
218.985
-1.260
Log-Probit
427.728
248.683
0.1238
217.884
-0.965
Weibull
266.515
130.801
0.0454
219.58
-1.377
aAIC = -2L + 2P, where L is the log-likelihood at the maximum likelihood estimates for the parameters and P is the
number of model degrees of freedom (usually the number of parameters estimated)
b%2 residual (measure of how model predicted response deviates from the actual data) for the dose group closest to
the BMC scaled by an estimate of its SD. Provides a comparative measure of model fit near the BMC. Residuals
that exceed 2.0 in absolute value should cause one to question model fit in this region.
°Model choice based on adequate p-valuc (>0.1), visual inspection, low AIC, and low (absolute) scaled residual.
The Multistage (1st degree) is referred to as the chosen model for male rats, though equivalent fit was obtained by
the restricted Gamma and Weibull models.
C-13

-------
Table C-7. Comparison of BMC/BMCL values for male and female mouse
liver hemosiderin staining data from inhalation chronic study using
measured blood AUC (12 months) of the EGBE metabolite BAA as a
common dose metric
Model
BMC10
(nmol-hr/L)
BMCL10
(nmol-hr/L)
/7-Value
AICa
Scaled residualb
Male mice
Multistage-1st
degree
2,100.07
1,613.9
0.3067
117.571
-1.766
Gamma
2,725.35
1,702.27
0.1452
118.559
1.358
Logistic
6,605.45
5,333.72
0.0022
127.326
2.789
Log-logistic
2,616.51
1,628.48
0.1882
118.02
1.193
Probit
5,917.06
4,825.09
0.0031
126.405
2.734
Log-probitc
3,076.8
2,448.3
0.1290
116.614
1.946
Weibull
2,689.76
1,687.09
0.1445
118.712
-1.448
Female mice
Multistage-1st
degree
946.491
769.879
0.3680
142.669
-1.583
Gamma
1,402.92
818.367
0.3420
143.288
-0.817
Logistic
2,897.15
2,341.03
0.0002
162.338
-0.942
Log-logistic
1,705.75
1,121.43
0.8223
141.501
-0.343
Probit
2,860.03
2,364.52
0.0002
161.681
-0.829
Log-probitc
1,734.53
1,322.06
0.8237
141.498
-0.315
Weibull
1,282.82
804.234
0.2958
143.631
-0.988
aAIC = -2L + 2P, where L is the log-likelihood at the maximum likelihood estimates for the parameters and P is the
number of model degrees of freedom (usually the number of parameters estimated).
b%2 residual (measure of how model predicted response deviates from the actual data) for the dose group closest to
the BMC scaled by an estimate of its SD. Provides a comparative measure of model fit near the BMC. Residuals
that exceed 2.0 in absolute value should cause one to question model fit in this region.
°Model choice based on adequate p-valuc (>0.1), visual inspection, low AIC, and low (absolute) scaled residual.
The Log-probit model provides a slightly better fit than other models for both genders.
Likewise, the fit statistics and BMC information for male and female mouse hemosiderin
staining data versus AUC BAA are shown in Table C-7. All models were fit using restrictions
and option settings suggested in the U.S. EPA BMD technical guidance document (U.S. EPA,
2000, 052150). The best model fit to these data, as determined by visual inspection, examination
of low dose model fit (i.e., scaled residual for the dose group closest to the BMD), and
comparison of overall fit (i.e., AIC values) was obtained using a log-probit model for both the
male and female response data. The male mouse BMCio was 3,077 [j,mol-hour/L and the
BMCLio was determined to be 2,448 [j,mol-hour/L using the 95% lower confidence limit of the
dose-response curve expressed in terms of the AUC for BAA in blood. The BMCio and BMCLio
values for the female mouse were determined to be 1,735 and 1,322 [j,mol-hour/L, respectively.
C-14

-------
Assuming continuous exposure (24 hour/day), the Corley et al. (1997, 041984) PBPK model was
used to back-calculate HECs of 36 ppm (174 mg/m3) from the male mouse data and 20 ppm
(97 mg/m3) from the female mouse data.
Multistage Model with 0.95 Confidence Level
Multistage
0.9
0.8
0.7
0.6
0.5
0.4
0.3
BMDL
BMD
0
500
1000
1500
2000
dose
16:50 02/27 2009
Source: NTP (2000, 196293)
Figure C-2. Multistage Model run of hemosiderin deposition in male rats
versus AUC BAA at 12 months from a 2-year inhalation study.
Multistage Model. (Version: 2.8; Date: 02/20/2007)
Input Data File: U:\BMDS\DATA\EGBE\RAT-HEMOSID-12MTH-DILL-
AUC\MALE-MULT 1. (d)
Gnuplot Plotting File: U:\BMDS\DATA\EGBE\RAT-HEMOSID-12MTH-DILL-
AUC\MALE-MULT1 pit
Mon Mar 24 14:56:58 2008
C-15

-------
BMD Model Run
The form of the probability function is:
P[response] = background + (l-background)*[l-EXP(
-betal*doseAl)]
The parameter betas are restricted to be positive
Dependent variable = HemoM
Independent variable = AUC-M-Dill
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 = 0.481866
Beta(l) = 0.000526977
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background	1 -0.64
Beta(l) -0.64 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.477747 *	*	*
Beta(l) 0.000536865 *	*	*
C-16

-------
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) #Param's Deviance Testd.f. P-value
Full model -121.475 4
Fitted model -121.617 2 0.28443 2 0.8674
Reduced model -130.097 1 17.2453 3 0.0006292
AIC: 247.234
Goodness of Fit
Scaled
Dose Est.Prob. Expected Observed Size Residual
0.0000
0.4777
23.887
23
50
-0.251
358.3000
0.5691
28.457
30
50
0.441
973.0000
0.6902
34.512
34
50
-0.157
2225.6000
0.8419
42.094
42
50
-0.037
ChiA2 = 0.28 d.f. = 2 P-value = 0.8680
Benchmark Dose	Computation
Specified effect =	0.1
Risk Type =	Extra risk
Confidence level =	0.95
BMD =	196.252
BMDL =	133.141
BMDU =	342.382
Taken together, (133.141, 342.382) is a 90 % two-sided confidence
interval for the BMD
C-17

-------
Log-Logistic Model with 0.95 Confidence Level
1
0.9
0.8
o
>2
5	0.6
c
o
o
2	0.5
LL
0.4
0.3
0.2
0	500	1000 1500 2000 2500 3000 3500
dose
17:03 02/27 2009
Source: NTP (2000, 196293)
Figure C-3. Log-Logistic Model run of hemosiderin deposition in female rats
versus AUC BAA at 12 months from a 2-year inhalation study.
Logistic Model. (Version: 2.10; Date: 09/23/2007)
Input Data File: U:\BMDS\DATA\EGBE\RAT-HEMOSID-12MTH-DILL-
AUC\FEMALE-LOGLOG.(d)
Gnuplot Plotting File: U:\BMDS\DATA\EGBE\RAT-HEMOSID-12MTH-DILL-
AUCVFEMALE-LOGLOG.plt
Mon Mar 24 15:20:55 2008
BMD Model Run
The form of the probability function is:
P[response] = background+(l-background)/[l+EXP(-intercept-slope*Log(dose))]
Dependent variable = Hemo F
C-18
Log-Logistic
BMDL BMD

-------
Independent variable = AUC-F-Dill
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 = 0.3
intercept = -17.7354
slope = 2.49241
Asymptotic Correlation Matrix of Parameter Estimates
background intercept slope
background	1 -0.24 0.19
intercept -0.24	1 -1
slope 0.19 -1	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable
background
intercept
slope
Estimate
0.280971
-16.4459
2.35478
Std. Err.
*
*
Lower Conf. Limit Upper Conf. Limit
* *
- Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) #Param's Deviance Testd.f. P-value
Full model -104.742 4
Fitted model -105.763 3 2.04091 1 0.1531
C-19

-------
Reduced model -135.725 1 61.9658 3 <.0001
AIC: 217.526
Goodness of Fit
Scaled
Dose Est.Prob. Expected Observed Size Residual
0.0000
0.2810
14.049
15
50
0.299
638.8000
0.4430
22.149
19
50
-0.896
1128.9000
0.6595
32.974
36
50
0.903
3461.8000
0.9566
47.828
47
50
-0.575
ChiA2 = 2.04 d.f. = 1 P-value = 0.1533
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 424.527
BMDL = 243.69
C-20

-------
C.1.3. BMD Method Applied to Forestomach Lesions in Female Mice
Cmax of blood BAA is considered a reasonable measure of internal dose for the
forestomach lesions reported from EGBE exposure. BAA is believed to be the toxic moiety
responsible for the forestomach effects observed following EGBE exposure, with concentration
(rather than AUC) appearing to be determinative in the development of these effects as well
(Corley et al., 2005, 100098). Other information supportive of the Cmax as an appropriate metric
include the findings of NTP (1993, 042063). where no signs of forestomach irritation were
observed in mice at very high dose levels of 1,400 mg/kg-day in 2-week and 13-week drinking-
water studies conducted by NTP (1993, 042063). It is likely that such oral nonbolus dosing of
EGBE does not result in high enough local concentrations of EGBE and BAA (Poet et al., 2003,
100190) to produce irritation. Consistent with this observation are results with other
forestomach carcinogens that are not mutagenic, demonstrating that forestomach effects are
dependent not only on the dose but also on the chemical concentration in the dosing solution
(Ghanayem et al., 1985, 042768).
The blood level of BAA is a valid surrogate for the local (i.e., forestomach) BAA
concentration as blood levels of BAA follow the severity of the irritant and irritant-associated
hyperplastic responses in the forestomach. Further, this relationship between blood BAA levels
and irritant response holds true for routes of EGBE administration other than oral, (e.g.,
inhalation, i.p., and i.v.) (Corley et al., 1999, 042003; Green et al., 2002, 041610). The basis for
this route-independent response may be related to the Green et al. (2002, 041610) results from
whole-body autoradiography of mice exposed via inhalation, showing appreciable amounts of
EGBE-associated radioactivity being present in salivary glands and ducts.
Plausible evidence exists for considering the incidence of forestomach tumors in female
mice, following chronic inhalation exposure to EGBE, to occur via a nonlinear, nongenotoxic
MO A. EGBE appears to be one of a group of compounds that are not mutagenic but can
indirectly cause forestomach tumors through the sustained cytotoxicity and cell regeneration
brought about by irritant and irritant-associated hyperplastic effects and breakdown of the
forestomach's gastric mucosal barrier. This sequence of events is considered obligate for the
formation of the observed neoplasms. Strategies intended to control or omit any of these key
mode-of-action events, including the initial hyperplastic event, would interrupt the process and
prevent formation of neoplasms. While this MOA may be of qualitative relevance to humans,
the exposure concentrations that would be necessary to cause these hyperplastic effects and
resultant tumors in humans, if attainable, are likely to be much higher than the concentrations
necessary to cause forestomach effects in mice, primarily because humans lack a comparable
organ for storage and long-term retention of EGBE.
Another line of reasoning that may be used to address issues relating to the occurrence of
these irritant/hyperplastic lesions (and potential for progression) in humans is to order the dose-
C-21

-------
response among those effects noted for EGBE, most prominently the hematologic effects that
underlie the hepatic tumors and that are the basis of the oral RfD and inhalation RfC. A BMD
analysis of the irritant/hyperplastic response observed in the NTP (2000, 196293) follows.
The endpoint used in this analysis was epithelial hyperplasia of the female mouse
forestomach, since it was the most sensitive forestomach effect observed in the NTP (2000,
196293) study. Consistent with the 1999 IRIS assessment, four steps were employed to estimate
human equivalent oral and inhalation benchmark exposures from this endpoint: (1) estimate a
BMDLio value using modeled "end-of-the-week" internal dose (Cmax BAA in blood) levels;
(2) verify that steady state was achieved (e.g., no change in BAA Cmax as a result of prolonging
the exposure regimen); (3) simulate the internal dose surrogate (Cmax BAA in blood) for humans
(continuous air exposure; drinking water assumption was that a 70-kg human consumes an
average of 2 L of water during a 12-hour awake cycle); and (4) calculate the HED/concentration
that resulted in the same internal dose (Cmax BAA) simulated for the animal in step 1.
Step 1: Estimation of BMDLio (Cmax) dose.
Cmax for BAA in arterial blood was determined using the PBPK model of Lee et al. (1998,
041983). The model results and incidence data for the endpoint of concern are summarized in
Table C-8.
Table C-8. PBPK model estimates of BAA Cmax blood levels and incidence
of forestomach epithelial hyperplasia in female mice
Air concentration (ppm)
Cmax BAA (jliIVI)
Incidence of forestomach
hyperplasia
0
0
0/50
62.5
529
6/50
125
1,200
27/50
250
2,620
42/49
BMD and BMDLio estimates were derived using the available models in version 1.3.2 of
the U.S. EPABMDS. The estimates for each model, along with statistical goodness-of-fit
information, are provided in Table C-9.
C-22

-------
Table C-9. BMDS model estimates of Cmax BMDi0 and BMDLio values for
forestomach epithelial hyperplasia in female mice
BMDS model
BMD (jiM)
BMDL (jiM)
AICa
(lowest = best fit)
/j-Valuc
(>0.1 = adequate fit)
Gamma
420.56
266.87
151.16
0.5287
Logistic
544.757
444.896
162.191
0.0067
Log-logisticb
462.513
329.04
150.153
0.8717
Multistage (1st degree)
177.442
145.713
156.244
0.0648
Multistage (2nd degree)
338.483
202.437
152.681
0.0976
Multistage (3rd degree)
338.485
197.436
152.681
0.2535
Probit
525.521
430.612
161.304
0.0086
Log-probit
470.876
344.412
150.163
0.8673
Weibull
376.085
238.952
151.855
0.3807
aAIC = -2L + 2P, where L is the log-likelihood at the maximum likelihood estimates for the parameters and P is the
number of model degrees of freedom (usually the number of parameters estimated).
bModel choice based on adequate p-valuc (>0.1), visual inspection, low AIC, and low (absolute) scaled residual.
The Log-logistic model provides a slightly better fit than other models.
Step 2: Verification of steady state.
As can be seen from Table C-10, Cmax levels are relatively constant through 6 months
then increase at and beyond 12 months, presumably due to clearance problems in aging animals.
However, the earlier steady-state levels are appropriate for use in this assessment, because that is
the more conservative approach, and because similar effects were observed during the
subchronic portion of the NTP (2000, 196293) study at the same dose levels, indicating that the
higher internal doses at and beyond 12 months were not required for the effects to appear.
Table C-10. Female mouse Cmax values for various time points of the NTP
(2000, 196293) study estimated by the Lee et al. (1998, 041983) model
Months on study
62.5 ppm
125 ppm
250 ppm
Male
Female
Male
Female
Male
Female
1
403
529
921
1,200
2,080
2,620
3
402
527
925
1,202
2,120
2,652
6
399
523
914
1,184
2,071
2,582
12
484
639
1,079
1,414
2,349
2,951
16
643
849
1,443
1,839
2,798
3,501
18
756
995
1,625
2,102
3,067
3,803
Step 3: Simulation of internal human doses.
The tables below summarize the results of model simulations of the internal dose
surrogate (Cmax BAA in blood) for a 70-kg human who consumes an average of 2 L of drinking
C-23

-------
water during a 12-hour awake cycle (Table C-l 1) or is continuously exposed to air
concentrations (Table C-12) of EGBE.
Table C-ll. Estimated Cmax for BAA in blood for humans continuously
exposed to varying drinking water concentrations of EGBE
EGBE concentration in water
Calculated dose of EGBE from
Cmax BAA
(ppm)
drinking water (mg/kg-d)
in blood (jliIVI)
24
0.7
9
48
1.4
18
94
2.7
36
188
5.4
73
375
10.7
147
750
21.4
299
Sources: Corley et al. (1994, 041977: 1997, 0419841.
Table C-12. Estimated Cmax for BAA in blood for humans continuously
exposed to varying concentrations of EGBE
Concentration of EGBE in air (ppm)
Cmax BAA in blood (juIVI)
1
2.6
5
13.0
10
26.1
20
52.9
50
137.1
100
295.0
200
733.7
Sources: Corley et al. (1994, 041977: 1997, 0419841.
Step 4: Calculate the HED/concentration
The Corley et al. (1994, 041977; 1997, 041984) PBPK model was used to back-calculate
a human equivalent oral dose of 23.6 mg/kg-day from the Cmax BMDLio of 329 |iM estimated in
step 1, assuming that mice and humans receive their entire dose of EGBE from drinking water
over a 12-hour period each day. The Corley et al. (1997, 041984) PBPK model was used to
back-calculate a human equivalent air concentration of 551 mg/m3 (113 ppm) from the Cmax
BMDLio of 320 |iM estimated in step 1, assuming continuous exposure (24 hours/day).
The PODs calculated above are significantly higher than the PODs of 1.4 mg/kg-day and
16 mg/m3 used to derive the RfD and RfC, respectively. Thus, these results indicate that the RfD
and RfC values for EGBE, which were based on hemosiderin accumulation due to hemolytic
C-24

-------
effects in rats, should be adequate for the prevention of gastrointestinal hyperplastic effects as
well.
Log-logistic Model with 0.95 Confidence Level
1
Log-logistic
0.8
0.6
0.4
0.2
0
BMDL
BMD
Concentration (mM)
500
>000
2500
Dose
Source: NTP (2000, 196293)
Figure C-4. Log-Logistic Model run of forestomach epithelial hyperplasia in
female mice versus Cmax BAA from a 2-year inhalation study.
Logistic Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:20 $
Input Data File: F:\BMDS\DATA\EGBE\F_MOUSE_HYP_LOG-LOGIST.(d)
Gnuplot Plotting File: F:\BMDS\DATA\EGBE\F_MOUSE_HYP_LOG-LOGIST.plt
Fri Jul 11 19:53:31 2003
BMDS Model Run
The form of the probability function is:
P[response] = background+(l-background)/[l+EXP(-intercept-slope*Log(dose))]
Dependent variable = F Hyperplasia
C-25

-------
Independent variable = Dose
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 = -16.768
slope = 2.36735
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 1-1
slope -11
Parameter Estimates
Variable Estimate Std. Err.
background 0 NA
intercept -16.7132 2.64108
slope 2.36545 0.372243
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) Deviance Test DF P-value
Full model -72.9391
Fitted model -73.0765 0.274637 2 0.8717
Reduced model -131.841 117.804 3 <0001
AIC: 150.153
C-26

-------
Goodness of Fit
Scaled
Dose Est.Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0 50 0
529.0000 0.1324 6.622 6 50 -0.2596
1200.0000 0.5145 25.725 27 50 0.3608
2620.0000 0.8705 42.653 42 49 -0.2777
Chi-square = 0.27 DF = 2 P-value = 0.8717
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 462.513
BMDL = 329.04
C.2. DERIVATION OF THE RfD
C.2.1. Potential RfD Derivations Based on Hematologic Data
PODs for the RfD derivation in terms of the HEDs have been calculated via the
application of PBPK modeling to NOAEL and BMDL estimates.
Of the available PBPK models (Table C-l), the Corley et al. (1997, 041984; 2005,
100100) model is considered the most complete and appropriate for use in the derivation of the
oral RfD because it has been experimentally validated for the most sensitive species (rats) and
humans, it covers both oral and inhalation routes of exposure, and it addresses both the
distribution and excretion of the toxic metabolite, BAA, following oral EGBE exposure. This
model is summarized in Appendix B. As in the case of the RfC (see Section C. 1.1), Cmax is
considered a more appropriate dose metric than AUC for the hematological effects. The PBPK
model of Corley et al. (1997, 041984; 2005, 100100) was used to obtain estimates of human Cmax
values from the female rat drinking water study data.
The four steps involved in using the Corley et al. (1997, 041984; 2005, 100100) PBPK
model to calculate the HED corresponding to the LOAEL identified in the animal study
(LOAELred) were to: (1) calculate the internal dose surrogate (Cmax BAA in blood)
corresponding to the female rat LOAEL, assuming that the drinking water was consumed only
during a 12-hour awake cycle on a 7 day/week schedule in model simulations; (2) verify that
steady state was achieved (e.g., no change in BAA Cmax as a result of prolonging the exposure
C-27

-------
regimen); (3) simulate the internal dose surrogate (Cmax BAA in blood) for humans consuming
EGBE in drinking water, assuming that a 70-kg human consumes an average of 2 L of water
during a 12-hour awake cycle; and (4) calculate the HED (mg/kg-day) for the amount of EGBE
consumed in 2 L of water that resulted in the same internal dose (Cmax BAA) simulated for the
animal in step 1 as shown below.
C.2.1.1. NOAEL/LOAEL Method and PBPK Model Applied to Hematologic Data
Step 1: Calculate the Cmax for BAA in blood corresponding to female rat LOAEL.
Female rat LOAEL = 59 mg/kg-day (calculated for final week of 13-week study to
correspond with the final hematological determination) Cmax BAA = 255 [jM.
Step 2: Verify steady state.
There were no changes in the Cmax of BAA in blood during any 24-hour simulation
period using a 12 hour/day, 7 day/week drinking water exposure regimen at the female rat
LOAEL, indicating that steady state was achieved.
Step 3: Calculate the Cmax for BAA in blood for humans continuously exposed to varying
concentrations of EGBE.
Table C-13 shows modeled estimates of BAA in blood of humans exposed continuously
to varying concentrations of EGBE in water (Corley et al., 1994, 041977; Corley et al., 1997,
041984). Drinking water volume is 2 L consumed over 12 hours in a day.
Table C-13. Modeled estimates of BAA in human blood exposed to EGBE in
water.
EGBE concentration (ppm) in
water
Calculated dose of EGBE from
drinking water (mg/kg-d)
Cmax BAA in blood (jtM)
24
0.7
9
48
1.4
18
94
2.7
36
188
5.4
73
375
10.7
147
750
21.4
299
Step 4: Calculate the LOAEL HED for a 70-kg human consuming EGBE in 2 L of
drinking water/day that results in the same internal dose of EGBE (Cmax of BAA in blood)
calculated for the animal study in step 1.
Female rat Cmax for BAA in blood at LOAEL = 255 [xM
LOAELred continuous exposure =18 mg/kg-day (calculated by regression of the internal
dose versus the dose of EGBE from step 3).
The internal dose surrogate, Cmax for BAA in blood, is highly dependent on the rate of
water ingestion. Since drinking-water exposures are highly complex and variable, a simplifying
C-28

-------
assumption was used in all simulations that the entire dose of EGBE in drinking water was
consumed over a 12-hour period each day corresponding to the awake cycle for both rats and
humans. This assumption resulted in higher Cmax blood concentrations of BAA in both rats and
humans than would have been calculated using the original Corley et al. (1994, 041977) structure
that assumed that drinking-water uptake occurred over a 24 hour/day dosing period.
C.2.1.2. BMD Method and PBPK Model Applied to Hematologic Data
Although a lower LOAEL was reported in male rats, this value gives no indication of the
relative slopes of the dose-response curves for males and females. Because this is an important
factor for BMD analyses (U.S. EPA, 1995, 005992; U.S. EPA, 2000, 0521501 a comparison of
the MCV and RBC count results for both male and female rats was performed and demonstrated
that female rats are more sensitive to the effects of EGBE than are males. Therefore, dose-
response information on these hematological effects in female rats was selected as the basis for
the oral RfD BMD analyses discussed below.
As was discussed in Section 5.2.1, MCV and RBC count are continuous response
measurements of precursor events associated with EGBE exposure and are considered the most
appropriate hematologic endpoints for use in a BMD analysis. Cmax is considered the more
appropriate dose metric for use in evaluating the chosen hemolytic endpoints. Cmax for BAA in
arterial blood was determined using the PBPK model of Corley et al. (1997, 041984; 2005,
100100). The results of this modeling effort are summarized in Table C-14.
Table C-14. Model estimates of BAA blood levels in female rats following
oral exposures
EGBE concentration
in water (ppm)
Water EGBE intake
(L/d)
Female body
weight (g)
BAA in blood
Dose (mg/kg-d)
Cmax (jlM)
750
0.0147
188
59
255
1,500
0.0155
185
125
914
3,000
0.0125
180
208
2,370
4,500
0.0101
164
277
3,23 la
6,000
0.0101
150
404
5,464a
aSteady-state not reached within 5 days.
Source: Corley et al. (2005, 1001001.
A BMD analysis was performed using U.S. EPA BMDS version 1.4.1
(http://www.epa.gov/ncea/bmds/about.html). As can be seen from the results in Table C-15,
RBC count was the more sensitive of the two hematological endpoints assessed. All models
were fit using restrictions and option settings suggested in the U.S. EPA BMD technical
guidance document (U.S. EPA, 2000, 052150) except for the choice of BMR.
C-29

-------
Table C-15. Comparison of female rat RBC count and MCV BMD/BMDL
values from an oral subchronic study using modeled blood Cmax (3 months)
of the EGBE metabolite BAA as a common dose metric
Model
BMD0S
(jiM)
BMDLos
(jiM)
BMDsd
(jiM)
BMDLsd
(jiM)
/7-Value
AICa
Scaled residualb
RBC count
Polynomial (1°)
1,623.31
1,350.41
1,937.36
1,532.88
<0.0001
-27.563915
-0.85
Power
1,623.31
1,350.41
1,937.36
1,532.88
<0.0001
-27.563915
-0.85
Hillc
180.667
93.9053
166.258
84.5854
0.4084
-47.398016
-0.228
MCV
Polynomial (1°)
1,036.38
934.455
578.088
442.513
<0.0001
146.041856
3.37
Power
1,036.38
934.455
578.088
442.513
<0.0001
146.041856
3.37
Hill
475.072
347.356
156.553
100.308
0.001166
135.552419
0.947
aAIC = -2L + 2P, where L is the log-likelihood at the maximum likelihood estimates for the parameters and P is the
number of model degrees of freedom (number of parameters estimated).
b%2 residual (measure of how model predicted response deviates from the actual data) for the dose group closest to
the BMD scaled by an estimate of its SD provides a comparative measure of model fit near the BMD. Residuals
that exceed 2.0 in absolute value should cause questioning of model fit in this region.
°Model choice based on adequate p-valuc (>0.1), visual inspection, low AIC, and low (absolute) scaled residual.
Source: NTP (1993, 042063}.
For continuous response data, the current BMD technical guidelines (U.S. EPA, 2000,
052150) suggest the use of 1 SD for the BMR level in the absence of additional information,
such as a minimal level of change in the endpoint that is generally considered to be biologically
significant. Because the chosen hemolytic endpoints are continuous measurements that have a
relatively small historical variance in rats, and because low-dose responses for these endpoints
were generally near or within 5% of the control mean, the BMDL05 was considered to be a more
appropriate POD for derivation of the RfD (U.S. EPA, 1995, 005992; U.S. EPA, 2000, 052150).
Higher levels of response (e.g., > 10% reduction) are in the exposure range where other more
severe responses related to anemia occur (e.g., RBC decreases, increased reticulocyte counts).
Lower levels of response (e.g., 1 SD, approximately a 3% reduction for these data) are relatively
distant from the observable data and other responses related to anemia.
Adequate model fit could not be obtained for the MCV data. The best model fit to the
RBC count data (from visual inspection and comparison of AIC values and scaled residuals near
the BMD) was obtained using a Hill model (see Table C-15). A graphical plot and textual
description of the results of the Hill model assessment of RBC count responses in female rats
(NTP, 2000, 196293) versus corresponding PBPK estimates of Cmax for BAA in female rat blood
are provided below.
C-30

-------
The BMD05 was 181 [xM and the BMDL05 was determined to be 94 [xM using the 95%
lower confidence limit of the dose-response curve expressed in terms of the Cmax for BAA in
blood. The Corley et al. (1997, 041984; 2005, 100100) PBPK model was used to back-calculate
a HED (BMDLred) of 6.8 mg/kg-day, assuming that rats and humans receive their entire dose of
EGBE from drinking water over a 12-hour period each day.
Hill Model with 0.95 Confidence Level
5
Hill
8
7.5
7
6.5
3MDL
BMD
1000	2000	3000	4000	5000
dose
08:27 03/04 2009
Source: NTP (1993, 042063)
Figure C- 5. Hill Model run of female rat RBC count versus Cmax BAA from
a 3-month oral study.
Hill Model. (Version: 2.14; Date: 06/26/2008)
Input Data File: C:\Usepa\BMDS21Beta\Data\EGBE\femaleratoral\lHilrbchil.(d)
Gnuplot Plotting File: C:\Usepa\BMDS21Beta\Data\EGBE\femaleratoral\lHilrbchil.plt
Tue Mar 03 15:29:39 2009
BMDS Model Run
C-31

-------
The form of the response function is:
Y[dose] = intercept + v*doseAn/(kAn + doseAn)
Dependent variable = RBC
Independent variable = Cmax
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
Default Initial Parameter Values
alpha = 0.154717
rho = 0 Specified
intercept = 8.15
v= -1.57
n = 0.982521
k = 551.55
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	v	k
alpha 1 2.2e-010	1.9e-009	6.6e-009
intercept 2.2e-010 1	-0.55	-0.54
v 1.9e-009 -0.55	1	-0.29
k 6.6e-009 -0.54	-0.29	1
Parameter Estimates
C-32

-------
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha	0.146124 0.0266785
intercept	8.13659 0.120911
v	-1.57421	0.156637
n	1	NA
k	518.419	227.998
0.0938351
7.89961
-1.88122
71.5517
0.198413
8.37357
-1.26721
965.286
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 N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
10
8.15
8.14
0.28
0.382
0.111
255
10
7.59
7.62
0.48
0.382
-0.228
914
10
7.09
7.13
0.43
0.382
-0.348
2370
10
7
6.84
0.37
0.382
1.28
3231
10
6.8
6.78
0.36
0.382
0.165
5464
10
6.58
6.7
0.41
0.382
-0.983
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model Log(likelihood) #Param's AIC
C-33

-------
A1 29.145609 7 -44.291218
A2 30.744881 12 -37.489762
A3 29.145609 7 -44.291218
fitted 27.699008 4 -47.398016
R -3.574142 2 11.148285
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.)
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 =	0.05
Risk Type = Relative risk
Confidence level = 0.95
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1
Test 2
Test 3
Test 4
68.638	10	<.0001
3.19854 5	0.6694
3.19854 5	0.6694
2.8932	3	0.4084
BMD
180.667
C-34

-------
BMDL = 93.9053
C-35

-------