IMCOLOGY EXCELLENCE FOR RISK ASSISSMEVT

   a nonprofit corporation dedicjted to
 the best use ot toxicily data lor risk \jlues
     EDUCATION
           Health Risk
Assessment/Characterization
               of the
Drinking Water Disinfection
      Byproduct Bromate
        RISK
      ESTIMATES
       This Document was Prepared by:

   Toxicology Excellence for Risk Assessment
           4303 Hamilton Avenue
           Cincinnati, OH 45223
        PEER
       REVIEW
        ITER
      DATABASE
      METHODS
    DEVELOPS
           Under the Direction of:

     Health and Ecological Criteria Division
       Office of Science and Technology
             Office of Water
     U.S. Environmental Protection Agency
          Washington, DC 20460
                                 Under Purchase Order No.
                                    8W-0766-NTLX

                                   September 30, 199a

-------
Health Risk
Assessment/Characterization of the
Drinking Water Disinfection Byproduct
Bromate
This Document was Prepared by:
Toxicology Excellence for Risk Assessment
4303 Hamilton Avenue
Cincinnati, OH 45223
Under the Direction of:
Health and Ecological Criteria Division
Office of Science and Technology
Office of Water
U.S. Environmental Protection Agency
Washington, DC 20460
Under Purchase Order No.
8W-0766-NTLX
September 30, 1998

-------
Foreword
The purpose of this document is to provide scientific support and rationale for the hazard
identification and dose-response information pertaining to chronic oral exposure to bromate It is
not intended to be a comprehensive treatise on the chemical or toxicology of bromate. Matters
considered in this nsk charactenzation include knowledge gaps, uncertainties, quality of data and
scientific controversies. This characterization is presented 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
nsk assessment process.
An earlier draft of this document underwent external peer review by three independent
experts and experts within EPA. The charge to external peer reviewers and their comments are
presented in Appendix C. Reviewers’ comments were considered in preparing the final version
of this document.
The quantitative dose-response analysis in this document was prepared by the National
Center for Environmental Assessment, Office of Research and Development, U.S
Environmental Protection Agency.
Bromate Hazard Characterization 2 September 30, 1998

-------
Table of Contents
FOREWORD 2
TABLE OF CONTENTS 3
LIST OF TABLES 4
1 0 [ NTRODUCTION 5
2.0 HAZARD ASSESSMENT 6
2.1 Studies in Humans 6
2.2 Studies in Ammals 7
2 2.1 Toxicokinetics. 7
2.2.2. Cancer Bioassays. 7
2.3. Mode of Carcinogenic Action 11
2 4 Cancer Hazard Characterization 15
3 0 DOSE-RESPONSE ANALYSIS 17
3.1 Choice of Critical Study Strengths and Weaknesses 18
3 2. Method of Analysis 19
3.3 Dose-response Charactenzatton 23
40 RISK CHARACTERIZATION 24
5 0 REFERENCES 27
APPENDIX A 30
APPENDIX B 32
APPENDIX C 38
APPENDIX D 44
Bromate Hazard Characterization 3 September 30, 1998

-------
List of Tables
Table 2-i. Tumor Incidence in Male Rats from U S. EPA Study (DeAngelo et al., 1998)
9
Table 2-2. Summary of Tumor Incidence in Male Rats (Kurokawa et al., 1986a)
10
Table 2-3. Tumor Incidence for Male and Female Rats (Kurokawa et al., 1986b)
11
Table 3-1. Parameter estimates for one-stage Weibull time-to-tumor model
20
Table 3-2. Human unit cancer risk estimates for bromate ion.
21
Brornate Hazard Characterization 4 September 30, 1998

-------
1.0 Introduction
Treatment of typical raw water with ozone results in the formation of a very Iar e
number of by-products Most are aliphatic aldehydes, ketones, alcohols or carboxylic acids,
although some aromatic by-products also occur. Reaction of ozone with inorganic substituents
in water can also produce other by-products. For example, during the ozonation process, ozone
oxidizes bromide (Bf) to bromate (BrO 3 ). A survey of bromate in drinking water revealed that
bromate concentrations in water from plants not using ozone were less than the detection limit of
10 g/L (U.S. EPA, 1994). In contrast, median bromate concentrations in ozonated dnnking
water, based on a survey of water utilities nationwide, are estimated to be about 5 p g/L. The
highest concentration of bromate reported was 90 g/L, which occurred when ozone was used in
combination with hydrogen peroxide to treat water. Based on a consideration of reported by-
product levels in ozonated water and the availability of toxicological data, the ozonation by-
product, bromate, has been selected for evaluation in this document
The characterization of bromate will focus on the potential for carcinogenicity following
U.S. EPA’s proposed cancer guidelines (April 23, 1996 61 F.R. 17960). Although no studies in
humans have evaluated the potential for bromate carcinogenicity, the carcinogenicity of bromate
has been evaluated in three long-term studies in rats and mice of both sexes by the oral route of
exposure. Very little information was found concerning bromate’s mode of action. This
information was limited to a few studies on mutagenicity, oxidative DNA adducts, and
metabolism.
U S. EPA prepared a critena document on bromate in 1994 (U.S. EPA, 1994) which
provides a thorough review of the complete database on bromate and evaluates the
toxicokinetics, health effects, and mechanism of action for bromate. The reader is referred to this
document for detailed information on the earlier studies on bromate. The purpose of this
document is to describe new studies, in particular the cancer bioassay by DeAngelo et al. (1998),
published after EPA’s 1994 critena document and to summarize key lines of evidence related to
bromate’s potential human carcinogemcity. The new data will be synthesized and integrated into
an overall characterization of hazard and dose-response.
Bromate Hazard Characterization 5 September 30, 1998

-------
2.0 Hazard Assessment
Limited data are available on the effects of chronic exposure to bromate in humans,
however, bromate toxicity following chronic exposure has been evaluated in rodents. In
addition, the mutagenic potential of bromate has been evaluated both in vitro and in vivo
genotoxicity assays Kidney tumors, thyroid tumors, and peritoneal mesothelioma have been
consistently observed in the rodent bioassays. The weight of evidence shows that bromate is
genotoxic in both in vivo and in vitro systems.
2.1 Studies in Humans
No epidemiological studies are available to demonstrate the carcinogenic potential of
bromate in humans A number of cases of acute bromate intoxication have been reported in
humans following accidental or suicidal ingestion of permanent hair wave neutralizing solutions
(U.S. EPA, 1994). These products usually contain either 2% potassium bromate or 10% sodium
bromate. The most common acute symptoms are severe gastrointestinal irritation (vomiting,
pain, diarrhea) and central nervous system (CNS) depression (lethargy, hypotension,
hypotonicity, loss of reflexes). Anemia from intravascular hemolysis may also occur These
effects ai:e usually reversible. Later sequelae (usually within several days) include marked renal
injury (oliguria, anuna, acidosis, elevated blood urea nitrogen) and hearing loss. Death from
renal failure may ensue if medical intervention is not successful. If support is successful, renal
function generally returns after 5—10 days. Hearing loss is usually irreversible. Estimated doses
in these cases range from about 20 to 1,000mg Br0 3 /kg body weight (U S. EPA, 1994). These
human case reports of short-term exposure are insufficient for denying quantitative estimates of
toxicity, including a cancer potency estimate, after long-term exposure
Bromate Hazard Characterization 6 September 30, 1998

-------
2.2 Studies in Animals
2.2.1. Toxicokinetics .
U S EPA (1994) provides a complete discussion of the absorption, distribution,
metabolism, and excretion of bromate; the reader is referred to the cntena document for detailed
information In summary, bromate is rapidly absorbed from the gastrointestinal tract, at least in
part unchanged It is distnbuted throughout the body appearing in plasma and urine unchanged
and in other tissues as bromide. Bromate is reduced to bromide in several body tissues, probably
by glutathione or other sulthydryl-containing compounds. Most bromate is excreted in the unne,
either as bromate or bromide, but some may leave the body in the feces. Bromine has been
detected in adipose tissue of mice following long-term treatment with bromate.
2.2.2. Cancer Bioassays .
The carcinogenicity of bromate has been evaluated by the oral route of exposure in both
sexes of F344 rats and B6C3F I mice. Exposure to bromate has caused renal tumors in male and
female rats after dnnking water exposure (Kurokawa et al., 1986a, 1986b; Kurokawa et al., 1987;
Kurata et al., 1992; DeAngelo et a!., 1998). In addition, exposure to bromate has caused thyroid
follicular cell tumors in both male and female rats and peritoneal mesotheliomas in male rats
(Kurokawa et al., 1986a, 1986b; DeAngelo et al, 1998). Studies on the time course of tumor
development indicate that the minimum treatment period for the induction of tumors is 13 weeks
(Kurokawa et al, 1987). Kurokawa et a!. (1986a, 1986b) was descnbed in detail in U.S EPA
(1994) and is only summanzed here. The DeAngelo et a!. (1998) study is discussed below.
In a recent study, U.S. EPA (DeAngelo et al., 1998) administered potassium bromate to
male F344 rats or male B6C3F I mice (78/group) in drmking water at concentrations of 0, 0.02,
0.1, 0.2 and 0.4 g/L or 0, 0.08, 0.4, and 0.8 g/L, respectively, for 100 weeks. Time-weighted
mean daily doses were calculated by the authors from mean daily water consumption and the
measured concentrations of potassium bromate. Bromate doses for the rat were 0, 1.1, 6.1, 12.9,
and 28 7 mg BrO 3 ikg-day. For rats, 6 animals/group were included for interim sacrifices which
occurred at 12, 26, 52, and 77 weeks. Parameters evaluated included survival, body weights,
organ weights, serum chemistry, and histopathology.
Bromate Hazard Characterization 7 September 30, 1998

-------
In male rats. survival in the 28 7 mg Br0 3 /kg-day dose group was dec eased compared
with controls beginning by approximately week 79 (Wolf, 1998a, see Appendix D for survival
curves); this decrease was statistically significant by study termination. In the 12.9 mg Br0 3
1kg-day dose group, survival was decreased compared with controls beginning by approximately
week 88 (Wolf, 1998a, see Appendix D for survival curves); this decrease was also significant by
study termination. Male rats in the 28.7 mg BrO 3 ikg-day dose group also had a statistically
significant decrease (18%) in the final mean body weight compared with controls. The decrease
in survival and body weight were attributed to an excessive mesothelioma burden (Wolf, 1998a).
The effects on survival and body weight in rats indicates that the maximum tolerated dose
(MTD) was reached ifl this study.
Tumor incidence for the interim sacrifices is presented in Appendix A; tumor incidence
for the terminal sacnfice is presented in Table 2-1. Statistically significant, dose-dependent
increased tumor incidence was observed in the kidney (adenomas and carcinomas combined, and
carcinomas alone), thyroid (adenomas and carcinomas combined, and carcinomas alone), and
tunica vaginalis testis (mesotheliomas). Using data from the National Toxicology Program
historical controls database (NTP, 1998), the historical control rates for these tumor types in
male F344 rats are 0.6% for kidney renal tubule adenomas and carcinomas, 2.1% for thyroid
follicular cell adenomas and carcinomas, and 1.5% for mesothehomas. The earliest renal tumors
and mesotheliomas in DeAngelo et a!. (1998) were observed at 52 weeks, the thyroid tumors
were first seen at 26 weeks (see Appendix A).
Results of the U.S. EPA study (DeAngelo Ct al., 1998) in male B6C3FI mice indicate
that mice may be less sensitive to the effects of bromate exposure than rats. Time-weighted
mean daily doses were calculated by the authors from mean daily water consumption and the
measured concentrations of potassium bromate. Bromate doses for the mouse were 0, 6.9, 32.5,
and 59.6 mg BrO 3 ikg-day. For mice, 7 animals/group were included for intenm sacrifice,
which occurred at 14, 31, 53, and 78 weeks. Bromate in drinking water had no effect on the
survival or body weight of male mice. Water consumption was decreased by 17% in the 59.6 mg
Br0 3 /kg-day dose group, this decrease was statistically significantly different from controls.
The only type of tumor reported for male mice was kidney tumors; however, the incidence of
Brorna e Hazard Characterization 8 September 30, 1998

-------
TumorType
Control
1.1mg
BrO 1 ikg-day
6.1 mg
BrO 1 ikg-day
12.9mg
BrO /kg-day
28.7mg
BrO 1 fkg-day
Kidney, adenomas
and carcinomas
combined
1/45
(2%)
1/43
(2%)
6/47
(13%)
3/39
8%
12/32**tt
(38%)
Kidney, carcinomas
alone
0/45
0/43
2/47
(4%)
1/39
(3%)
4/32’t
(13%)
Thyroid, adenomas
andcarcinomas
combined
0/36
4/39
(10%)
1/43
(2%)
4/35*
(11%)
14/30**tt
(47%)
Thyroid, carcinomas
alone
0/36
2/39
(5%)
0/43
2/35
(6%)
6/30*tt
(20%)
Mesothelioma
0/47
4/49
(8%)
5/49*
(10%)
10/47 1*
(21%)
27/43 1 1 tt
(63%)
* Statistically significant when compared with control, p <0 05
Statistically significant when compared with control, p <0 002
t Statistically significant trend with dose, p <0.05
tt Statistically significant trend with dose, p <0 002
Two additional key cancer bioassays (Kurokawa et al., 1986a, 1986b) are summarized
below for the purposes of this cancer hazard charactenzation for bromate. These studies as well
as additional studies on the time course for tumor development, the cumulative dose and
minimum treatment period needed for tumor induction are descnbed in greater detail in the
cntena document (U.S. EPA 1994). Kurokawa et a!. (1986a) treated groups of 20 to 24 male
F344 rats with water containing potassium bromate at 0, 15, 30, 60, 125, 250 or 500 mg(L for
104 weeks The average doses in Kurokawa et al. (1986a) for male rats was 0, 0 7, 1.3, 2.5, 5.6,
12.3, or 33 mg BrO 3 ikg-day. Compared with controls, the males in the high-dose group had
decreased body weight gain and decreased survival beginning approximately week 70. Survival
and body weight gain was comparable with controls for all remaining dose groups. Tumor
incidence fron this study is summarized in Table 2-2 Statistically significantly increased
incidence was observed for dysplastic foci at the 1.3 mg BrO 3 ikg-day dose and above, for kidney
adenoma and carcinoma combined was not dose-dependent. Incidence for the intenm sacrifices
is presented in Appendix A. Tumor incidence at termin’! sacrifice for combined kidney tumors
in male mice was 0/40, 5/38 (p
-------
Tumor Type Control


0 7
mg BrO,
1kg-day
1 3
mg BrO ,
1kg-day
2 5
mg BrO ,
1kg-day
5 6
mg BrO ,
1kg-day I
12 3
mg BrO ,
1kg-day
33
mg BrO , [ kg-
day
Dysplasuc
0/19
1/19
5/20*
6/24*
12124**
19120**
19/20**
foci’
(5%)
(25%)
(25%)
(50%)
(95%)
(95%)
Kidney,
0/19
0/19
0/20
1/24
5/24*
5/20*
9/20**
adenoma and
(4%)
(2 1%)
(25%)
(45%, 3
carcinoma
carcmomasb)
combined
Thyroid,
0/16
0/19
0/20
1/24
0/24
3/20
7/19*
adenoma and
(4%)
(15%)
(37%)
carcinoma
combined
Mesothelioma
_______________
0/19
0/19
3/20
(15%
4/24
(17%)
2/24
(8%)
3/20
(15%)
15/20*
(75%)
Considered by the authors to be a preneoplastic lesion
Incidence of carcinomas alone. not statistically significant
* Statistically sigmficant when compared with control, p < 0.05
** Statistically significant when compared with control, p <0 001
Kurokawa et at. (1986b) studied the carcinogemc potential of potassium bromate in both
male and female F344 rats and female B6C3F1 mice. Potassium bromate was administered in
drinking water. Time-weighted mean doses of potassium bromate were estimated by the authors
based on measured water consumption and body weight. The average bromate doses for rats
were 0, 9 6 and 21.3 mg Br0 3 /kg-day in males and 0, 9.6 and 19.6 mg Br0 3 /kg-day in females.
The average bromate doses for mice were 0, 43.5 and 91.6 mg Br0 3 fkg-day. Male rats in the
high dose group had a marked decrease in body weight gain and a decrease in survival beginning
approximately week 70 compared with controls. The authors do not describe the cause of the
decreased survival and body weight. For the low dose groups in male rats and all dose groups in
female rats and mice, survival and body weight gain were comparable to controls.
In Kurokawa et al. (I 986b), treatment-related, statistically significant tumors observed
in rats included renal cell adenomas and carcinomas and peritoneal mesotheliomas (in males
only). The tumor incidence for rats are shown in Table 2-3. The authors note that “high
incidence” of tumors was observed in the thyroid; however this incidence was not statistically
tumors at the 5 6 mg BrO kg-day dose group and above, and for the thyroid tumors and
mesothetiomas at the high dose group only
Table 2-2. Summary of Tumor Incidence in Male Rats (Kurokawa et al, 1986a)
Bromate Hazard Characterization
10
September 30, 1998

-------
significant En male rats. the earliest renal tumor was observed at 14 weeks and the earliest
mesothelioma was observed at 72 weeks En female rats, the earliest renal tumor was observed at
85 weeks. In female mice, no significant differences in tumor incidence between exposed and
control animals were apparent after 78 weeks of dosing, based on histological examination of
tissues at week 104. The authors concludeJ that potassium bromate was carcinogenic in rats of
both sexes, but not in mice.
Table 2-3 Tumor Incidence a for Male and Female Rats (Kurokawa et al., I 986b)
Tumor Type
Control
9 6 mg Br0 3 1kg-day
(males) mg BrO 1kg-day
Male Rats
46/52**
Kidney , adenomas and
carcuiomas combined
3/53
(6%)
32/53’
(60%)
(88%)
Kidney, carcinomas alone
3/53
(6%)
24/53**
(45%)
44/52**
(85%)
28/46**
Pentoneum,
mesotheliomas
6/53
(11%)
17/52*
(33%)
(61%)
Female Rats
39/49**
Kidney, adenomas and
carcinomas combined
0/47
28150**
(56%)
(80%)
flKjdney, carcinomas alone
L______________________
0/47
21150’
(42%)
36/49**
(73%)
a Incidence reported for the “effective number of rats”, which is defined by the authors as the number of rats
surviving longer than the time at which the earliest tumor of each type was observed.
* Statistically sigmficant when compared with control, p < 0 01
Statistically significant when compared with control, p <0001
The evidence presented by the key carcinogenicity (Kurokawa et al., 1986a, 1986b;
DeAngelo et al., 1998) studies demonstrates that bromate is carcinogenic by the oral route of
exposure in both sexes of rats, but not in mice.
2.3. Mode of Carcinogenic Action
The weight of evidence demonstrates that bromate is genotoxic in in vivo and in vitro
systems. Bromate was mutagenic in Salmonella typhimurium (strain TAIOO) with metabolic
activation (Ishidate et a!., 1984). Bromate caused chromosomal aberrations in Chinese hamster
fibroblasts (Ishidate et al., 1984) and rat bone marrow cells (Fujie et al., 1988). Oral exposure to
Bromate Hazard Characterization
11
September 30, 1998

-------
brornare resulted in a dose-dependent increase in micronuclei in both rats and mice (Fujie et a!
1988. E- [ ayashi et al., 1988, Hayashi et al , 1989; Sai et al 1992a). The reader is directed to the
critena document (U S. EPA, 1994) for additional Lnformatlon regarding the mutagenicity of
bromate
A series of studies from one laboratory indicates that bromate may exert its kidney effects
by a mechanism that involves the formation of oxygen radicals. In vivo exposure to bromate
results in the formation of the damaged-DNA product 8-hydroxydeoxyguanosine (8-OH-dG) in
the kidney, but not liver, of rats (Kasai et aL, 1987; Sai et a!, 1991; Sai et aL, 1992a, 1992b,
1 992c). Formation of 8-OH-dG has been associated with mutagenicity caused by agents which
form oxygen radicals (Kasai et al., 1987). Time-course studies indicate that formation of 8-OH-
dG correlates with an increase of lipid peroxidation in the kidney and an increase in relative
kidney weight (Sai et al, 1991). Pretreatment with glutathione (GSH), cysteine, and Vitamin C
inhibited the formation of 8-OH-dG and lipid peroxidation, and it prevented the increased
relative kidney weight associated with bromate treatment. Concurrent treatment with diethyl
maleate, an agent which depletes GSH, increased the formation of 8-OH-dG and lipid
peroxidatiori, and it caused and even greater increase in relative kidney weight compared to the
animals treated with bromate alone (Sai et al., 1992c). The criteria document (U.S. EPA, 1994)
provides additional information about the studies described in this section.
Recent studies (published sin e the cntena document) have been conducted to better
understand how bromate causes its DNA-damaging effects. Sai et. a!. (1994) investigated the
role of oxidatrve damage in potassium bromate (KBrO 3 )-induced carcmogenesis in the renal
proximal tubules (RPT) and nuclear fractions. RPT were isolated from male F344 rats arid
incubated with KBrO 3 (0—10 mM) for 8 hours in 95% air/5% CO 2 . Renal nuclear fractions were
isolated via centrifugation from kidney homogenates and resuspended in KBrO 3 (10 imM) for 2
hours at 37°C. DNA and 8-hydroxydeoxyguanosine (8-OH-dG) were isolated frDm nuclear
fractions of RPT and analyzed. The release of lactate dehydrogeriase (LDH) from RPT was also
determined, At 0.5, 2 and 5 mM K.Br0 3 , a significant increase in the release of LDH and a
significant decrease in protein-SH content in RPT (75%, 68% and 43% of control values) was
seen in a time- and concentration-dependent manner. 8-OH-dG levels in RPT and the ratio of
Bromate Hazard Characterization 12 September 30, 1998

-------
15-pero\idized arachidonic acid to the total isomers (17-, 18- and 19- peroxidized arachidonic
acid), an indicator of lipid peroxidation, were also increased at 2 and 5 mM KBrO 3 8-OH-dG
levels in renal nuclei were also increased approximately 2-fold following incubation with
autooxidized methyl linolenate, a lipid-peroxidizing system. The authors suggested that the
potassium bromate-induced carcinogenesis may be due to lipid peroxidation and the subsequent
DNA damage sustained in RPT, the target site for renal carcinogenesis
In another study, Umemura et al. (1995) investigated the role of 8-OH-dG and
cumulating replicating fractions (CRFs) in potassium bromate-induced carcinogenesis in the rat
kidney and liver Female F344 rats received gavage doses of 100, 200 or 400 mg/kg of KBrO 3 .
Additional female rats were administered 0.05% N-ethyl-N- hydroxyethylnitrosamine (EHEN)
orally for the first two weeks as an initiator with subsequent administration of KBrO 3 at a dose of
500 ppm in drinking water for 30 weeks. Unlike the liver, renal 8-OH-dG levels were
significantly increased at 200 mg/kg (0.83 ± 0.15/10 dG, P<0.01) and 400 mg/kg (1.54 ±
0.15/l0 dO, P<0.01) compared to controls (0.37 ± 0.iO/i0 5 dG). Likewise, renal CRFs in the
proximal convoluted tubular cells were significantly increased at 200 mg/kg (3.67 ± 0.99%,
P<0.05) and 400 mg/kg (12.40 ± 3.04%, P<0.05) compared to controls (2.02 ± 0.66%), but
unchanged in the liver. A significantly higher number of atypical tubules, atypical hyperplasia
and renal cell tumors were seen in animals treated with KBrO 3 after EHEN initiation than in
animals given distilled water after EHEN initiation. However, no significant effect was observed
on liver tumongenesis. The authors concluded that oxidative stress is associated with tumor
promotion in female rats but hypothesized that a 2 -globu1in mediated cell proliferation in male
rats was cooperating with oxidative mechanisms and accounted for the sex differences observed
in bromate kidney toxicity.
Ballmaier and Epe (1995) observed that in cell-free systems or in in vitro mammalian cell
cultures, the reduction of potassium bromate by glutathione actually generates a short-lived
reactive intermediate that induces 8-hydroxyguariine. The damaged DNA was not associated
with cytotoxicity in the cell cultures. The results obtained by Ballmaier and Epe (1995) are in
contrast to earlier in vivo studies in which potassium bromate and GSH were administered
together. The authors conclude that these differences are due to the fact that, in vivo, reduction
Bromate Hazard Characterizauon 13 September 30, 1998

-------
of potassium bron-iate to inactive bromide occurs before bromate reaches the target tissue The
reactive intermediate responsible for the DNA damage is thought to be the bromine radical or
bromine oxides, consistent with a finding that molecular bromine gives nse to the same DNA
damage profile as potassium bromate.
In a recent study, Chipman et al (1998) proposed a dual role for GSH in the genotoxicity
of potassium bromate Consistent with the findings of Ballmaier and Epe (1995), Chipman et al
(1998) foundthat incubation of isolated calf thymus with both potassium bromate ( 15 tM) and
GSH (500 riM) produced elevated 8-OH-dO, while incubation with potassium bromate alone did
not produce any DNA damage. These data suggest a direct, activating role for GSH in vitro
However, data from in vivo systems suggest that GSH has a protective effect. Chipman et al.
(1998) found that 8-OH-dG was not elevated in either total DNA or mitochondnal DNA from rat
kidney perfused in situ with 5 mM potassium bromate for up to 1 hour. A single i.p dose of 100
mg/kg potassium bromate caused a significant increase in lipid peroxides, 8-OH-dG, and
oxidized GSH. Pretreatment with diethyl maleate to deplete OSH enhanced the toxicity of
potassium bromate, as evidenced by clinical signs of toxicity including weight loss reflecting
dehydration. In contrast, a single i.p. dose of 20 mg/kg potassium bromate had no effect on
either toxicity or oxidative stress. The authors conclude that a sustained exposure at a high, toxic
dose of bromate which induces lipid peroxidation is required for DNA oxidation, and that this
study contnbutes to the evidence that the dose-response relationship in renal carcinogenesis is
non-linear.
Although some data are available which suggest oxidative stress as a possible
mechanism for the formation of kidney tumors, no studies are available which directly evaluate
whether antioxidants will prevent the formation of kidney tumors. In addition, no relationship
has been shown between 8-OH-dO and any adverse cellular outcome. No information is
available to suggest possible mecharnsms for the formation of thyroid tumors or mesotheliomas
Although carcinogens that produce thyroid tumors may operate via mutagenic mechanisms, most
chemicals that induce thyroid follicular cell tumors in animals seem to operate by interfenng
with the thyroid-pituitary feedback mechanism (Hill et al., 1989). This mechanism involves a
decrease in circulating levels of the thyroid hormones 13 and T 4 and increased secretion of
thyroid stimulating hormone (TSH) by the pituitary, which results in prolonged stimulation of
Bromate Hazard Characterization 14 September 30, 1998

-------
the th roid gland The progressive morphological changes in the thyroid in response to
prolonged stimulation include hypertrophy, hyperplasia, nodular hyperplasia, and finally,
neoplasia However, no data are available regarding bromate’s effects on thyroid hormone or
TSH levels that would indicate that bromate is affecting the thyroid-pituitary feedback
mechanism In addition, no data are available to suggest a possible mechanism for formation of
pentoneal mesotheliomas.
At present, the available data on bromate are too limited to provide a description of mode
of action In chronic animal bioassays, exposure to bromate consistently results in kidney
adenomas and carcinomas and thyroid follicular cell adenomas and carcinomas in both male and
female rats. Pentoneal mesotheliomas are also consistently observed in male rats. Bromate is
mutagenic in bacterial systems and causes chromosomal aberrations in a variety of cell systems.
No studies are available which fully evaluates the profile of DNA adducts. These studies are
needed to determine if bromate has the potential to interact directly with DNA. It is generally
accepted that mutagenicity can play a role in the carcinogenic process. The other mechanisms by
which bromate may cause the tumors observed in multiple target organs is uncertain at this time.
2.4 Cancer Hazard Characterization
Bromate should be evaluated as a likely human carcinogen by the oral route of exposure.
Data are insufficient to draw conclusions regarding the inhalation route. Although no
epidemiological studies or studies of long-term human exposure to bromate are available,
bromate is carcinogeruc to male and female rats following exposure in dnnking water. Given the
limited data on possible mechanism of carcinogenic action for bromate, it is a reasonable default
assumption that production of tumors in rodents occurs by a mode of action that is relevant to
humans.
Three key studies (Kurokawa et al., I 986a, 1 986b; DeAngelo et al., 1998) demonstrate
the carcinogenicity of bromate in animals. All studies were well conducted using an appropriate
route of exposure and adequate numbers of animals. In the studies, the maximum tolerated dose
was reached as evidenced by effects on survival and body weight in the high dose groups There
Bromate Hazard Charactenzat’on 15 September 30, 19?8

-------
is some concern that the high dose in each of these studies (28 7 mg Br0 1 1kg-day - DeAngelo et
at, 1998, 19 6 mg Br0 3 /kg-day - Kurokawa et at, l986h 33 mg Br0 3 1kg-day - Kurokawa et
al., 1986a) exceeded the MTD for male rats However in all three studies, the decrease in
survival began to appear relatively late in the study. approximately week 70 in the Kurokawa et
al (1986a, 1986b) studies and approximately week 79 in the DeAngelo et at (1998) study. Two
studies reported the time of first tumor observation: in Kurokawa et at (1986b), the first tumor of
any type was observed at 14 weeks and in DeAngelo et at. (1998), the first tumor of any type was
observed at 26 weeks. Therefore, the mate rats in these studies were surviving long enough to
have developed tumors . In addition, in the DeAngelo et at. (1998) study, the decreased survival
and body weight gain appeared to be caused by the heavy mesothelioma burden of the animals
(Wolf, 1998a); the cause of decreased survival and body weight gain in the Kurokawa et al.
(1 986b) study is not apparent. The decreased survival in the high doses in these studies does not
compromise these studies for use in nsk assessment.
Several aspects of these bioassay studies support the conclusion that bromate has the
potential to be a human carcinogen. First, tumors were observed at multiple sites, including
kidney (adenomas and carcinomas), thyroid (folticular cell adenomas and carcinomas), and
pentoneum (mesotheliomas). In DeAngelo et at. (1998) the mesotheliomas arose from the tunica
vaginalis testis and spread throughout the peritoneat cavity on the serosal surfaces of many
organs Kurokawa et at. (1986a, 1986b) do not specify the ongin of the pentoneal
mesothetiomas observed. While mate rats had tumors at all three sites, only kidney tumors were
observed in female rats However, the kidney tumors in female rats developed in the absence of
the significant toxicity observed in the male rats.
Second, there was a clear dose-response relationship both in tumor incidence and in
severity/progression of tumors. Kurokawa (1986a) observed statistically significantly increased
incidence of renal dysplastic foci, a preneoplastic lesion, at doses of 1.3 mg Br0 3 ikg-day and
above; statistically significantly increased incidence of renal adenomas at doses of 5 6 mg Br0 3
1kg-day and aboye; and increased incidence of renal carcinoma at the lugh dose of 33 mg Br0 3
1kg-day. Kurokawa (1986a, 1986b) observed dose-response relationships for two other tumor
types, and U S EPA (DeAngeto et at., 1998) observed does-response relationships for alt three
Bromate Hazard Charactenzation 16 September 30, 1998

-------
tumor t’vpes in rats The consistent observation of an uncommon tumor, mesotheliomas, also
lends support to the human cancer potential of bromate
The evidence is too limited to give high confidence in a conclusion about any mode of
carcinogenic action Oxidative stress may play a role in the formation of kidney tumors, but the
evidence is insufficient to establish lipid perioxidation and free radical production as the key
events responsible for the induction of kidney tumors. In addition, no data are currently
available to suggest any mechanism for the production of thyroid and peritoneal tumors by
bromate Bromate is mutagenic in both rats and mice in viva and in vitro assays; albeit the
testing has been limited to the Ames assay and in vitro cytogenetics and bone marrow assays.
Therefore, in the absence of a biologically-based model, the assumption of low dose linearity is
considered to be a reasonable public health protective approach at this time for estimating the
potential risk for bromate because of the limited data on its mode of action and because of some
evidence of mutagenicity.
3.0 Dose-response Analysis
In 1994, U.s EPA (1994) presented a dose-response assessment of bromate following the
1986 Carcinogen Risk Assessment Guidelines (U.S. EPA, 1987). That assessment used the
male rat kidney tumor incidence from Kurokawa et al. (1986b). and the linearized multistage
(LMS) model to estimate an oral slope factor for bromate ion of 0.65 per (mg/kg-day); the
drinking water unit risk for the bromate ion was 1.8E-5 per (ugfL). The purpose of the current
dose-response analysis is to compare the cancer potency estimated using the more recent U S.
EPA (DeAngelo et al., 1998) bioassay data for the male rat tumors to that determined in 1994
using the Kurokawa et al. (1 986b) male rat tumor data. As in the 1994 assessment, a low dose
linear extrapolation approach is taken because of a lack of understanding of bromate’s mode of
action ant the presence of positive mutagenicity results. The LED/ED approach recommended
by the 1996 Proposed Guidelines for Carcinogen Risk Assessment (U.S. EPA, 1996) is provided
for comparison. In addition, a quantal analysis of the DeAngelo et al. (1998) data using the LMS
model is provided in Appendix B for comparison with the 1994 results based on the Kurokawa et
al. (1986b) data.
Bromate Hazard Characterization 17 September 30, 1998

-------
3.1 Choice of Critical Study: Strengths and Weaknesses
The rodent bioassay studies (Kurokawa et a!, 1986a, 1986b, DeAngelo et a!., 1998)
clearly indicate that bromate induces tumors at multiple sites in rats However, the tumor
incidences among the three studies are different, and the nature of the dose-response is not well
defined. Kurokawa et al (1986b) observed higher incidences of both kidney tumors and
peritoneal mesotheliomas at both dose groups than Kurokawa et al. (1986a) or DeAngelo et al.
(1998) However, Kurokawa et al. (1986b) did not observe the statistically significant increase in
thyroid tumors that was observed in the other two studies. The tumor mcidences at all three sites
were similar for both Kurokawa et al (1986a) and DeAngelo et al. (1998); however, the
statistical significance of the tumor incidence vaned between the studies. DeAngelo et a! (1998)
observed a statistically significant increase of mesothehoma at the 6.1 mg Br0 3 /kg-day doses
and higher, statistically significant increase of thyroid tumors at the 12.9 mg Br0 3 /kg-day doses
and higher, and statistically significant increase of kidney tumors only at the highest dose
Conversely, Kurokawa et al. (1986a) observed statistically significant increase in mesotheliomas
and thyroid tumors only at the highest dose tested, but observed statistically significant increase
of kidney tumors at the 5.6 mg BrO 3 ikg-day doses and higher. Because the DeAngelo et a!
(1998) study used lower doses than the Kurokawa et al. (1 986b) study and used more animals per
group than the Kurokawa et al. (1986a) bioassay, DeAngelo et a!. (1998) was chosen as the
preferred data set for quantif ’ing bromate cancer risk.
The strengths of DeAngelo et a!. (1998) include dose-dependent results at tumor sites
consistent with the Kurokawa et a!. (1986a, 1986b) studies, adequate numbers of animals, and
lower doses than Kurokawa et al, (1986b). The data are considered adequate for dose-response
modeling; moreover, the availability of the individual animal data make it possible to account
for early mortality and include the interim lull results. Although DeAngelo et al. (1998) only
evaluated male rats, the Kurokawa et al. (1986b) found no difference in the response of male and
female rats to kidney tumors. Therefore, it is reasonable to use male rat data and assume that it is
valid for females. There is some concern that decreased survival in the two highest dose groups
compromised the quality of the study. As discussed in section 2.4, the excessive tumor burden
Bromate Hazard Characterization 18 September 30, 1998

-------
appears to be the cause of early mortality and decreased body weight gain in this study (Wolf,
1998a) Thyroid tumors were first seen at week 26 and kidney tumors and testicular
mesotheltomas were first seen at week 52 while survival was comparable to controls until
approximately week 79. Therefore, the rats survived well past the time of first tumor observation
and the study is not compromised for quantifying cancer nsk. Use of the Weibull time-to-tumor
model will account for any effects that early mortality may have had on tumor response.
3.2. Method of Analysis
Time-to-tumor analyses of data from the U.S. EPA study (DeAngelo et a!., 1998) were
performed to account for early mortality in the highest dose group. The analyses were conducted
using the individual male rat data, including the 12-, 26-, 52-, and 77-week intenm kill data, for
each site demonstrating an increased cancer incidence. Bemgn and malignant tumors were
combined for the sites, i.e., testicular mesotheliomas; kidney tubular adenomas and carcinomas;
and thyroid follicular adenomas and carcinomas. Tumors from each of the sites was modeled
separately and then the individual risks combined to represent the total cancer risk.
The general model used for the time-to-tumor analyses was the multistage Weibull
model, which has the form
P(d,t) = I - exp [ -(q 0 ÷ q 1 d + q 2 d 2 + ... + q dk)*(t -
where P(d,t) represents the probability of a tumor by age t (in bioassay weeks) for dose d (rat
dose of bromate in mg/kg-day), and parameters z 1, to?O, and q O for i=O, I, ..., k, where k =
the number of dose groups - 1. The parameter t 0 represents the time between when a potentially
fatal tumor becomes observable and when it causes death (see below). The analyses were
conducted using the computer software TOX-RISK Version 3.5 (Crump et al., ICF Kaiser
International, Ruston, LA), which is based on Weibull models taken from Krewski et al. (1983).
Parameters are estimated using the method of maximum likelihood.
Bromate Hazard Charactenzatlofl 19 September 30, 1998

-------
Specific n-stage Weibull models were selected for the individual tumor types based on
the values of the log likelihoods according to the strategy used by NIOSH (1991). If twice the
difference in log likelihoods was less than a chi-square with degrees of freedom equal to the
difference in the number of stages included in the models being compared, then the models were
considered comparable arid the most parsimonious model (i.e., the lowest stage model) was
selected
hi the time-to-tumor analysis, tumor types are categorized as either fatal or incidental
Fatal tumors are those tumors thought to act rapidly to cause the animal to die, while incidental
tumors are thought not to have caused the death of the animal, or at least not rapidly. Each of the
three tumor types observed in the U S. EPA study was considered incidental (Wolf, 1998b).
Thus, t 0 was set equal to 0.
Parameter estimates for the time-to-tumor analyses for each tumor type are presented in
Table 3-1. For each tumor type, a one-stage model was the preferred model.
Table 3-1. Parameter estimates for one-stage Weibull time-to-tumor model
Tumor
Z
Testicular mesothelioma 0 0 3.94 X i0 9
3 44
Kidney tubular adenomas and carcinomas 3 78 X iO 3 26 X 1O
2.28
Thyroid follicular adenomas and carcinomas 3.95 X 1O I 2.63 X i0
1.28
Incremental lifetime unit extra cancer risks for humans were estimated by TOX-RISK
based on a linearized low-dose extrapolation of the Weibull time-to-tumor models for the rat
tumor sites. Extra risk over the background tumor rate is defined as
(P(d) - P(O)) / (1 - P(0)).
Human equivalent doses of bromate ion were calculated using a 2/3-power surface area
adjustment (e.g , human dose in mg/kg-day = (rat dose m mgfkg-day)(rat body weight/human
body weight)L 3 ) The resulting cancer potency estimates are presented in Tables 3-2.
Bromate Hazard Characterization 20 September 30, 1998

-------
Con ersions to the 34-po er-based estimates,as described in the 1996 Proposed Guidelines for
Carcinogen Risk Assessment (U S. EPA, 1996), will be made at the end of this section
Table 3-2. Human unit cancer nsk estimates for bromate ion(extra risk; computed for a risk
of 10.6), based on the rat tumor sites (DeAngelo et at., 1998), using a one-stage Weibull
time-to-tumor model and 2/3 power surface area scaling.
Tumor
Q*a
((rng1kg-day) ’)
MLEb
((mgikg-day) ’)
0.1/LEDi 0 c
((mgfkg-day) )
Mesothelioma
0.83
0.42
0.79
Kidney
Thyroid
0.28
0.15
0.12
0.075
0.26
0.14
a upper confidence limit on cancer potency
b maximum likelihood estimate of cancer potency
unit cancer risk estU te based on drawing a straigifl jinc from the 95% lower conf dence 1mm qfl the
dose resulting in a IU/o extra cancer risk to the ongin (U dose, ii extra risk), as described [ Or the linear
extrapolation default in EPA’s 1996 Proposed Guidelines for Carcinogen Risk Assessment (U S EPA,
1996) LED 10 modeled using the Weibull time-to-tumor model
The testicular mesotheliomas yield the highest upper bound unit cancer potency estimate
(q 1 *), 0.83 per mg Br0 3 1kg-day.
While the time-to-tumor modeling does help account for decreased survival times in the
rats, considering the tumor sites individually does not convey the total amount of risk potentially
ansing from multiple sites. To get some indication of the total unit risk from multiple tumor
sites, assuming the tumors at these different sites anse independently, the maximum likelihood
estimates (MLE) of the unit potency from the Weibull time-to-tumor models were summed
across tumor sites and an estimate of the 95% upper bound on the sum was calculated. The
TOX_RISK software provides MLEs and 95% UCLs for extra risk at vanous exposure levels,
allowing for the calculation of urut potency estimates at those exposure levels. By summing the
MLE across the three tumor sites, it is not assumed that these tumors are caused by a similar
mechanism.
Bromate l-Iazard Characterization
21
September 30, 1998

-------
The potency estimates were summed using a Monte Carlo analysis and the software
Crystal Ball Version 4 0 (Decisioneenng, Denver CO) Normal distnbutions were assumed for
the potency estimate at a human dose of I ngtkg-day for each tumor site, with the distnbution
mean equal to the MLE of potency and the standard deviation, (, calculated according to the
formula
95% UCL = MLE + 1 .645o.
A distnbution of the sum of the potency estimates was then generated by simulating the sum of
estimates picked from the distributions for each tumor site (according to probabilities prescribed
by those distnbutions) 10,000 times. This procedure yielded a mean value for the total unit nsk
of 0 63 per mg/kg-day continuous lifetime bromate ion exposure, equal to the sum of the MLEs
of unit potency (0.62). The 95% upper bound for the total unit risk was 1.07 per mg BrO 3 ikg-
day. In comparison, summing the q 1 *s across the three tumor sites yielded 1 .26 per mg BrO 3 fkg-
day
The summation analyses were repeated for potency estimates calculated at a human dose
of 0.01 mg/kg-day for comparison with the estimates calculated at I ng/kg-day. The results were
nearly identical (the 95% UCL for total unit nsk was 1.06 per mg BrO 3 ikg-day). Thus, the total
unit potency estimates are effectively linear up to 0 01 mg/kg-day continuous lifetime bromate
ion exposure. A sensitivity analysis based on the contribution to variance reported that the
variability associated with the risk estimate for the testicular mesotheliomas was contributing
over 85% of the variance of the sum In addition, the simulation analysis revealed that the
assumption of normal distributions on the risk estimates is violated because some of the
simulated estimates and sums were negative. Thus, if the potency estimates were constrained to
be non-negative, the true distributions would be skewed to the right rather than symmetrical, and
the 95% UCL on the sum would likely be higher than that predicted under the assumption of
normal distributions, although not as high as the sum of the upper bounds.
Bromate Hazard Charactenzation 22 September 30, 1998

-------
In summar . based on a time-to-tumor analysis of the EPA data (DeAngeto et at. 1998)
for testicular mesothetiomas, kidney tubular tumors, and thyroid follicular tumors in the mate
F344 rat and a 2/3-power surface area scaling factor, the best estimate of an upper bound
incremental lifetime human unit extra cancer risk is 1 07 per mg BrO 3 Jkg-day continuous
lifetime exposure to bromate ion in dnnking water. Using the proposed 3/4-power surface area
adjustment (U S EPA, 1996), the unit potency estimate for potassium bromate would be about
0 70 per mg BrO 3 ikg-day (conversion factor of 0 65 (1 e., ((70/0.4)U4/(70/0.4) 3
3.3 Dose-response Characterization
The hazard characterization of bromate suggests that the dose-response assessment
should apply a linear extrapolation from data in the observable range to the low dose region
because of the lack of understanding of bromate’s mode of action and the positive mutagenicity
data. A time-to-tumor analysis of all 3 male rat tumor sites (modeled independently) from the
U.S. EPA study (DeAngelo et al., 1998) and 2/3 power scaling provides a unit potency estimate
of about 1 07 mg BrO 3 ikg-day. This is slightly higher than U.S. EPA’s earlier estimate of 0 65
per mg/kg-day bromate ion, which was based on a quantal analysis ofjust the kidney tumors
from the Kurokawa et al. (1986b) study using 2/3-power scaling (U S. EPA, 1994). Using the
3/4-power scaling as described in the 1996 Proposed Guidelines for Carcinogen Risk Assessment
(U.S. EPA, 1996), a unit potency of 0.70 per mg/kg-day for bromate ion is estimated based on
the time-to-tumor analysis of the male rat tumor data in DeAngelo et al. (1998).
The potency estimate of 0.70 per mg/kg-day is considered a plausible upper bound
estimate on human extra unit cancer risk from contmuous lifetime exposure to bromate ion in
drinking water, based on linear extrapolation of the cancer risks observed in the most sensitive
species examined, the rat. This potency estimate corresponds to a drinking water unit risk
estimate of 2.0 x I 0 per gfL bromate ion. Assuming a daily water consumption of 2 L/day for
a 70 kg adult, lifetime cancer risks of 10 , 10 , and 10.6 are associated with bromate
concentrations of 5,0 5, and 0.05 .ig(L, respectively.
Bromate Hazard Charactenzalion 23 September 30, 1998

-------
A major source of uncertainty in these estimates is from the interspecles extrapolation ol
risk from rats to humans The limited results in mice (Kurokawa et al , 1986b, DeAngelo et a!
1998) suggest that this species is less sensitive to bromate-induced carcinogenicity than is the rat
The reasons for the interspecies differences are unknown, and it is not known whether humans
are more similar to the rat or the mouse.
A.nother major source of uncertainty in the unit potency estimate is the linear
extrapolation of high-dose risks observed in the rat bioassay to lower doses that would be of
concern from human environmental exposures. A multistage Weibull time-to-tumor model was
used because it can take into account the differences in mortality between the exposure groups in
the rat bioassay, however, it is unknown how well this model is predicting the risks for low
exposure to bromate. While there are also uncertainties pertaining to the specific assumptions
used in conducting the multistage Weibull time-to-tumor analyses and the Monte Carlo analysis
for summing across the tumor sites, these are considered minor when compared to the
uncertainties introduced by the interspectes and high-to-low dose extrapolations.
4.0 Risk Characterization
In 1994, U S. EPA concluded that bromate was a probable human carcinogen (Group B2)
under the 1986 EPA Guidelines for Carcmogen Risk Assessment weight of evidence
classification approach. The new rodent cancer study by DeAngelo et al. (1998) is considered to
provide confirming evidence for the potential human carcinogenicity of bromate. Under the
principles of the 1996 Proposed Guidelines for Carcinogen Risk Assessment, bromate is
considered a likely human carcinogen by the oral route of exposure . This conclusion is
supported by evidence of bromate carcinogenicity in both sexes of rats following exposure in
drinking water, a relevant route of exposure. Tumors were consistently observed at multiple sites
including kidney, thyroid, and pentoneum in several studies. In rats, all three tumor types show
statistically significant dose-response relationships (DeAngelo et al., 1998, and tumors show a
dose-dependent progression in severity from preneoplastic lesions to carcinomas Further, an
uncommon tumor, mesothelioma, is found in all three rodent cancer bioassays
Bromate Hazard Charactenzaflon 24 September 30, 1998

-------
The major limitation of the bromate hazard charactenzation is the lack of data on the
effects of long-term exposure to bromate in humans. The available human data are limited to
case reports of toxicity following acute, accidental ingestion. Therefore, to extrapolate rat tumor
data for bromate to the human situation, it must be assumed that humans will respond like the rat.
Nevertheless, the choice of using the rat tumor data from DeAngelo et al (1998) in the absence
of human data is a reasonable default assumption,
Overall, there is not enough evidence to give high confidence in a conclusion about any
mode of carcinogenic action. Studies are available showing that bromate is weakly mutagenic
and causes chromosomal aberrations (Ishidate et al., 1984; Fujie et al., 1988; Hayashi et al.,
1988; Hayashi et al., 1989; Sal et al., 1992a). The mode of action by which bromate is inducing
mutations and, thus, tumors in the target organs is uncertain. Studies are available showing that
bromate may generate oxygen radicals which increase lipid peroxidation and damage DNA
(Kasat et il., 1987, Sal et aL, 1991; Sai et al., 1992a, 1992b, 1992c; Sal et al., 1994; Umemura et
al., 1995). However, no data are available which link this proposed mechanism with to tumor
induction. Thus, the available evidence is insufficient to establish this mechanism as a key event
in the induction of tumors at the target organs observed. Given the uncertainty about the mode of
action, a science policy decision is made to use a low dose linear extrapolation approach as more
protective of public health. The cancer risk estimation presented for bromate is considered to
be protective of susceptible groups, including children, given that the low dose linear default
approach is used as a conservative approach.
A low dose linear extrapolation based on the U.S. EPA bromate study (DeAngelo et al.,
1998) was conducted using a one-stage Weibull time-to-tumor model. This model was selected
because it can account for the early mortality observed in treated animals as compared with
control animals. Modeling was conducted on the individual tumor types, and cancer potency
estimates were generated for the individual sites and for total risk from all three sites combined.
Incidence of pentoneal mesotheliomas was the most sensitive response; however the total cancer
potency estimate was selected because it accounts for the total cancer risk posed by tumors
arising at multiple sites. It is assumed that these different tumors at different sites arise
Bromate Hazard Characterization 25 September 30, 998

-------
ndependenh1y and that the different tumors are not necessarily induced by similar mechanisms
A source of uncertainty is the Lnterspecles differences bet’ een rat and humans. Studies indicate
that mice are less sensitive to the effects of bromate than rats The reasons for this difference are
unknown and it is also unknown what the relative sensitivity between rats and humans is
Another uncertainty concerns how well the linear extrapolation predicts the low-dose human
risks for bromate
Based on low dose linear extrapolation, a upper bound cancer potency estimate for
bromate ion is 0.70 per mg/kg-day. This potency estimate corresponds to a drinking water unit
risk of 2 x 1 0 per gfL, assuming a daily water consumption of 2 L/day for a 70 kg adult.
Lifetime cancer risks of l0 , and 10.6 are associated with bromate concentrations of 5, 0.5,
and 0.05 g/L, respectively. These estimates are in close agreement with earlier assessments
derived from the Kurokawa et al. (1986b) data that are presented in the criteria document (U.S.
EPA, 1994).
In conclusion, the available data suggest that bromate in drinking water is a likely human
carcinogen A number of uncertainties related to this conclusion are discussed above. The upper
bound estimate of cancer risk for bromate ion is 0 70 per mg BrO 3 ikg-day with a corresponding
drinking water unit risk of 2 x l0 per g/L.
Bromate Hazard Characterization 26 September 30, 1998

-------
5.0 References
Anderson, E L 1983. Quantitative approaches rn use to assess cancer nsk LI S. EPA
Carcinogen Assessment Group Risk Anal. 3: 277-295.
Ballmaier, D and B Epe 1995. Oxidative DNA damage induced by potassium bromate under
cell-free conditions and in mammalian cells. Carcinogenesis. 16: 335-342.
Chipman, J K., J E. Davies, J L. Parsons, J. Nair, G. O’Neill, and J K. Fawell. 1998. DNA
oxidation by potassium bromate; a direct mechanism or linked to lipid peroxidation?
Toxicology In press.
DeAngelo, A.B., M.H. George, S.R. Kilburn, T.M. Moore and D.C. Wolf. 1998.
Carcinogenicity of potassium bromate administered in the drinking water to male B6C3FI mice
and F3441N rats. Toxicol. Path. 26(4): July/August. In press.
Fujie, K, H. Shimazu, M. Matsuda and T. Sugiyama. 1988 Acute cytogenetic effects of
potassium bromate on rat bone marrow cells in vivo. Mutat. Res. 206: 455—458.
Hayashi, M., M. Kishi, T. Sofuni and M. Ishidate Jr. 1988. Micronucleus tests in mice on 39
food additives and eight miscellaneous chemicals. Food Chem. Toxicol. 26: 487—500.
Hayashi, M, S Sutou, H. Shimada, S. Sato, Y.F. Sasaki and A. Wakata, 1989. Difference
between intraperitoneal and oral gavage application in the micronucleus test. The 3rd
collaborative study by CSGMT/JEMS-HMS. Mutat Res. 223: 329—344
Hill, R.N., L.S. Erdreich, 0 E. Paynter, P A. Roberts, S.L. Rosenthal, and CF. Wilkinson. 1989.
Thyroid folhcular cell carcinogeneszs. Fund. Appl. Toxicol. 12: 629-697.
Howe, R B , and Van Landingham, C. 1986. GLOBAL86: A computer program to extrapolate
quanta! animal toxicity data to low doses. EPA Contract No. 68-01-6826. K.S. Crump and
Company, Ruston, LA. Unpublished.
Ishidate, M., T. Sofimi, K. Yoshikawa, M. Hayashi, T. Nohmi, M.. Sawada, and A. Matsuoka.
1984. Pnmary mutagenicity screening of food additives currently used in Japan. Food Chem.
Toxicol. 22: 623—636.
Kasai, H., S. Nishimura, Y. Kurokawa and Hayasha. 1987. Oral administration of the renal
carcinogen, pota.sslum bromate, specifically produces 8-hydroxydeoxyguanosine in rat target
organ DNA. Carcinogenesis 8: 1959—1961.
Bromate Hazard Charactenzaiion 27 September 30, 1998

-------
K.re ski. D Crump. KS Farmer. J et al 1983 A companson of statistical methods for low
dose extrapolation utilizing time-to-tumor data Fundani Appi Toxicol 3 140-160
Kurata, Y , B.A. Diwan, arid J.M. Ward. 1992. Lack of renal tumour-initiating activity of a
single dose of potassium bromate, a genotoxic renal carcinogen in male F344/NCr rats. Food
Chem Toxicol 30(3). 25 1-259
Kurokawa, Y , Y. Matsushima, N. Takamura, T Imazawa, and Y Hayashi. 1987. Relationship
between the duration of treatment and the incidence of renal cell tumors in male F344 rats
administered potassium bromate. Jpn. J. Cancer. Res. 78: 358-364.
Kurokawa, Y , S. Aoki, Y Matsushima, N. Takamura, T Imazawa, and Y. Hayashi. 1986a.
Dose-response studies on the carctnogemcity of potassium bromate in F344 rats after long-term
oral administration. J. Nati Cancer Inst. 77 97 7—982
Kurokawa, Y , S. Takayama, Y. Konishi, Y. Hiasa, S. Asahina, M. Takahashi, A. Maekawa and
Y Hayashi 1986b. Long-term n vivo carcinogemcity tests of potassium bromate, sodium
hypochionte and sodium chlorite conducted in Japan. Environ. Health Perspec. 69: 221—236.
NIOSH 1991 A quantitative assessment of the risk of cancer associated with exposure to 1,3-
butadiene, based on a low dose inhalation study in B6C3F 1 mice. Submitted to the Occupational
Safety and Health Admimstration (OSHA) for Butadiene Docket (#H-041). U.S. Department of
Health and Human Services, Washington, DC.
NTP. 1998. National Toxicology Program Historical Controls Database.
http //ehis niehs.nih.gov
Sai, K, A Takagi, T Umemura, R. Hasegawa and Y. Kurokawa. 1991. Relation of 8-
hydroxydeoxyguanosine formation in rat kidney to lipid peroxidation, glutathione level and
relative organ weight after a single adnumstration of potassium bromate. Jpn. J. Cancer Res.
82(2). 165-169.
Sat, K., M. Hayashi, A. Takagi, R. Hasegawa, T. Sofuni and Y. Kurokawa. 1992a. Effects of
antioxidants on induction of niicronuclei in rat peripheral blood reticulocytes by potassium
bromate Mutat. Res. 269(1): 113-118.
Sai, K., S. Uchiyama, Y. Ohno, R. Hasegawa, and Y. Kurokawa. 1992b. Generation of active
oxygen species n vitro by the interaction of potassium bromate with rat kidney cell.
Carcinogenesis. 13(3): 333-339.
Bromate Hazard Characterization 28 September 30, 1998

-------
Sai. K. T Lmemura. A Takagi, R Hasegawa and Y Kurokawa 1992c The protective role of
glutathione, cysteine and vitamin C against oxidati e DNA damage induced in rat kidney by
potassium bromate Jpn J. Cancer Res 83(1) 45-51
Sai, K, C A. Tyson, D W Thomas, et al. 1994. Oxidative DNA damage induced by
potassium bromate in isolated rat renal proximal tubules and renal nuclei Cancer Lett.
87 1-7
Lrmemura, T., K Sal, A. Takagi, et al. 1995. A possible role for oxidative stress in
potassium bromate (KBrO 3 ) carcinogenesis. Carcinogenesis 16: 593-597.
U S EPA. 1987. Risk Assessment Guidelines of 1986. (EPAJ600J8-87/045).
U S. EPA. 1994. Drinking water criteria document on bromate Office of Water,
Washington, DC.
U S EPA, 1996. Proposed Guidelines for Carcinogen Risk Assessment. Fed. Reg. 6 1(79):
17960-18011.
Wolf, D C. 1998a. Personal communication from Douglas Wolf, National Health and
Environmental Effects Research Laboratory, U.S. EPA to VickiDellarco, Office of Water, U S.
EPA. February 20, 1998.
Wolf, D.C I 998b. Personal communication from Douglas Wolf, National Health arid
Environmental Effects Research Laboratory, U.S. EPA to Jennifer Jinot, National Center for
Environmental Assessment, U.S EPA. January 12, 1998.
Bromate Hazard Characterization 29 September 30, 1998

-------
Appendix A
Interim Sacrifice Data from U.S. EPA bromate study
(DeAngelo et at., 1998)
Bromate Hazard Characterization 30 September 30, 1998

-------
Table A-I. Summary of Intenm Sacnfice Data for Rat.
week control 11 mg Br0 3 6] mg 8r0 3 12.9 mg Br0 3 28.7 mg Br0 3
I. 1kg-day 1kg-day 1kg-day 1kg-day
mesotheliomas
12 0/6
0/6
0/6
I 0/6
I 0/6
26 0/6
0/6
0/6
0/6
0/6
52
0/6
0/6
0/6
1/6
0/6
77
0/6
0/6
0/6
0/6
4/6
kidney tubular adenomas +_carcinomas
12
0/6
0/6
0/6
0/6
0/6
26
0/6
0/6
0/6
0/6
0/6
52
0/6
0/6
0/6
0/6
2/6
77
1 0/6
0/6
0/6
0/5
4/6
thyroidfollicular adenomas + carcinomas
12 0/6 } 0/6
0/6 0/6
016
26 0/6
0/6
1/6 1/6
0/6
52 I 0/6
0/6
0/6 I 0/6
0/6
77
0/6
0/6
0/6 0/5
3/6
Table A-2. Summary of Intenm Sacnfice Data for Mouse (kidney tubular adenomas + carcinomas)
week control 6.9mg BrO /kg-day 32.5 mg 8r0 /kg-day 59.6 mg BrO /kg-day
14 0/7 0/7 0/7 0/7
31
0/7
0/7
0/7
0/7
53
0/7
0/7
[ 0/7
0/6 (week 52)
78
0/7
0/7
0/7
0/5
Bromate Hazard Characterization
31
September 30, 1998

-------
Appendix B
Comparison of Modeling using DeAngelo et al. (1998) and
Kurokawa et al (1986b) Data.
Bromate Hazard Characterization 32 September 30, 1998

-------
B.!. Quantal 4nal ses of the DeAngelo et al. (1998) data.
Quantal analyses using the linearized multistage (LMS) model are provided here
primarily for the purpose of comparison with the cancer potency estimates calculated by EPA in
the criteria document (U S EPA, 1994) based on the 1986 Kurokawa et al. bioassay, for which
individual animal data were not available The estimates derived from the quantal analyses are
considered infenor to those calculated above using time-to-tumor analyses because the effects of
early mortality in the bioassay on lifetime cancer nsk are not fully taken into account. In
addition, data from the interim kills cannot be incorporated. The dose-response data for the
quantal analyses are presented in Table B-I.
Table B-i. Dose-response data for potassium bromate administered to male rats (DeAngelo
et al., 1998)
administered dose
(mg BrO 3 ikg-day)
control
1.1
6.1
12.9
28.7
human equivalent dose
0
.21
1.1
2.25
4.91
(mg BrO 3 ikg day)a
I
testicular
0/47
4/49
5/49
10/47
27/43
mesotheliomas
kidney tubular
1/45
1/43
6/47
3/39
12/3 2
adenomas or-carcinomas
thyroid follicular
0/36
4/39
1/42
4/35
14/30
adenomas or carcinomas
Numberofrats
1/36
6/39
11/42
9/33
26/30
with tumorsb/
Number of rats at nSkc
a human equivalent doses of bromate were calculated using a 2/3-power surface area adjustment
b tumors at any of the three significant sites.
C ra s s viving to the n e of the,first.significant tumor, which was a testicular mesothelioma at week 78 in the
hignest aose group, ana tor wnicn patnatogy aata were available tor all three sites
95% upper-limit incremental lifetime unit cancer risks (extra risk) for humans were
calculated for the individual tumor sites as well as for the number of tumor-bearing animals from
the incidence data in Table B-I using the LMS model. The model has the form
Bromate Hazard Characterization 33 September 30, 1998

-------
P(d) = I - exp [ -(q 0 - - q 1 d q,d 2 -i- ÷ q d”)],
where P(d) represents the lifetime risk (probability) of cancer at human dose d, and parameters q
= 0, for 1=0, 1, ..., k Extra risk over the background tumor rate is defined as (P(d) - P(0)) / (1 -
P(O))
Point estimates of the dose coefficients (q’s), and consequently the extra risk function, at
any dose d, are calculated by maximizing the likelihood function with respect to the tumor
rncidence data. The incremental lifetime unit cancer risk for humans (q 1 *) is defined as the 95%
UCL on the parameter q 1 , which is the linear dose coefficient estimate calculated by the
algorithm. This 95% UCL represents a plausible upper bound for the true risk The 95% UCL
was calculated using the computer program GLOBAL86 (Howe and Van Landingham, 1986).
Both the model and the curve-fitting methodology used are descnbed in detail by Anderson et al.
(1983). The results from the LMS model (GLOBAL86) are presented in Table B-2.
MLEs; q 0 if not specified
b for chi-square goodness of fit
C calculated at dose of 1 x i0 mg/kg-day
d effective dose estimated to result in 10% ex a nsk
e lower 95% limit on ED,,
Table B-2. Results from linearized multistage quanta! model for bromate ion
Tumor
dose
coefficients’
p-values
q 1 *
((mg/kg-
day) ’)
MLE of unit
potencyc
((mg/kg-day) )
ED , 0
(mg/kg-
day)
LED 1 Oe
(mg/kg-
day)
mesothehoma
qO= 1.92 x 10.2
ql=945x10 2
0.17
017
0.094
110 064

kidney
g4=862x10
q02.46 x 102
ql=4.81 x l0
q4=3.40 x 10 ’
0.21
0.099
0.048
2 06
1.07
thyroid
qO=4.48 x 102
any of the 3
q35.02x i0
q07.92 x 10.2
013
0.068
5.Ox io .
2.76
151
.
sites
q1=O.138
g42.05 x iO
0.13
0.25
0.14
0.76
0.43
Bromate Hazard Characterization
34
September 30, 1998

-------
Thus, using the LMS quanta! model, the incremental unit cancer risk estimate (95%
UCL) for humans calculated from the tumor-beanng ani ’ials would be 0 25 per mg Br0 3 1kg-day
for continuous lifetime exposure to bromate. The highest unit potency for an individual tumor
site is 0 17 per mg BrO 3 Ykg-day based on the testicular mesotheliomas. The unit potency based
on the kidney tumors is 0 099 per mg BrO 3 ikg-day.
Under U.S EPA’s proposed cancer nsk assessment guidelines (U.S. EPA, 1996), unit
cancer risk estimates for genotoxic chemicals, such as bromate, would be derived by drawing a
straight line from the LED 10 to the origm (0 dose, 0 risk). Using the LED 10 generated for the
tumor-bearing animals from the LMS model (GLOBAL86) yields a unit cancer risk for
potassium bromate of 0.10/(0.43 mg BrO 3 ikg-day), or 0.23 per mg/kg-day, virtually the same as
the q 1 *
The unit cancer risk estimate (95% UCL) derived above is intended to depict a plausible
upper limit on the risk of developing any bromate-attributable tumor from continuous exposure
in drinking water over a full (70 year) lifetime However, using the quantal incidence data for
total tumor-bearing rats in each dose group does not fully charactenz the cancer potency
reflected by the rat bioassay results. First, the methodology does not take into account the fact
that some of the rats had tumors at multiple significant sites. Second, the methodology ignores
the fact that survival was decreased in the highest dose group, thus it does not reflect the
reduction in rat-weeks at risk during older ages when cancer risks are higher Comparison with
the results of the time-to-tumor analyses conducted above on the same data suggests that the
LMS model may be underestimating the unit potency.
B.2. Comparison with Estimates Derived from the Kurokawa et AL(1986b) Data.
In the criteria document, U.S. EPA (1994) denved a cancer potency estimate (UCL) of
0.65 per mg BrO 3 ikg-day for bromate based on the male rat kidney adenomas and carcinomas,
the most sensitive tumor response from Kurokawa et a!. (1986b). A similar potency estimate
(0.69 per mg BrO 3 ikg-day) is obtained from the Kurokawa et al. (1986b) female rat kidney
Bromate Hazard Characterization 35 September 30, 1998

-------
tumor data Male rats also had significant increases in pentoneal mesotheliomas Sur ival was
decreased in both dose groups for the male rats, but not fr’r females Individual animal data were
not available, and the cancer potency estimate was calculated using the LMS model This
potency estimate of 0.65 per mg BrO 3 ikg-day likely underestimates the cancer potency implied
by the Kurokawa et al. (l986b) results because early mortality was not reflected in the quantal
analysis and because the pentoneal mesotheliomas were not included
The U.S. EPA (DeAngelo et al., 1998) bioassay results generally suggest lower cancer
potencies for the same strain of rats (F344) (See Table B-3). The reasons for the discrepancies
are unknown. In the EPA study, a large number of the animals that died before the end of the
study had incomplete pathology (tissues not available), and if these animals were also more
likely to have tumors, potency estimates based on the incidences for only the animals with
pathology results may underestimate the actual tumor nsk. However, even if one assumes that
all the animals without pathology results had tumors, the incidences would still be lower than
those obtained in the Kurokawa et al. ( 1986b) study.
Bromate Hazard Characterization 36 September 30, 1998

-------
Table B-3. Companson of tumor incidence and modeling results between the Kurokawa et al. (1986b) and U S EPA (DeAngelo ci l.
1998) rat bio assays
Weibull time-to-tumor
model
0 27
0.84
0 I I
007
LMS model
0.70
064
0 10
0.26
0.17
0037
0042
Daily
Human
Kurokawa
Kurokawa
EPA
Kurokawa et al
EPA male
Kurokawa
Kurokawa
1 l’A
intakes .
equivalent
et al
et at. male
male
male peritoneal
testicular
et al
et al male
imile
(mg I3rO 3
dose (mg
female
kidney
kidney
mesotheliornas
mesothelionias
female
thyroid
ihyiotd
1kg-day)
BrO 3 ikg-
day)
kidney
thyroid
0
0
0/47
(0%)
3/53
(6%)
1/45
(2%)
6/53
(11%)
0/47
(0%)
0/52
(0%)
2/53
(4%)
0/36
(0%)
11
.21
1/43
(2%)
4/49
(8%)
4/39
(10%)
6.1
11
.
6/47
(13%)
5/49
(10%)
1/42
96
1.5
28/50
(56%)
1/52
(2%)
9.6
1.8 -
32/53
(60%)
17/52
(33%)
2/53 -
(4%)
12.9
2.2
3139
(8%)
10/47
(21%)
4135
(11%)
19.5
2.8
39/49
(80%)
3/52
(6%)
21.2
3.8
46/52
(88%)
28146
(61%)
7/52
(13%)
28.7
4.9
12/32
(38%)
•
27/43
(63%)
14/30
(47%)
q 1 ((mg BrO 3 ikg-day) )
Bromate Hazard Charactenzation
37
September 30, I )9

-------
Appendix C
Charge to External Peer Reviewers and Reviewers’ Comments
Bromate Hazard Characterization 38 September 30, 1998

-------
Statement of \Vork
TITLE : Peer review of Health Risk Characterization Report on Bromate
BACKGROUND :
The mission of the United States Environmental Protection Agency’s (EPA) Office of
Water (OW) is to protect public health and the environment from adverse effects of
contaminants in media such as ambient water, drinking water, waste water, sewage sludge
and sediments. This procurement relates to the peer review of a health risk assessment on the
ozone disinfection by product, bromate. This risk assessment will be used in support of
EPA’s stage 1. ‘disinfection by product rule which is scheduled to be final in November 1998.
The Safe Drinking Water Act Amendments of 1996 emphasize that “the best peer review
science” be used in carrying out SDWA regulations.
PURPOSE :
A cancer risk assessmentlcharacterization document has been recently prepared that
cites and updates EPA’s 1994 assessment on bromate This 1998 document considers a new
cancer bioassay in rodents and applies the EPA 1996 proposed revisions to its guidelines for
carcinogen risk assessment.
TASK DESCRIPTION :
This purchase will procure a peer review on the 1998 EPA bromate risk cancer
assessment report EPA has attached the 1998 risk assessment document, consisting of
approximately 20-25 pages, on bromate to be reviewed (Attachment 1), as well as supporting
matenals, such as EPA’s 1996 guidelines for carcinogen assessment (Attachment 2), EPA’s
1994 Criteria Document on bromate (Attachment 3), and hard copies of key studies
(Attachment 4). The peer reviewer shall submit written comments that are clear/transparent,
and constructive. They shall comment on whether the document clearly and adequately
discusses
Bromate Hazard Characterization 39 September 30, 1998

-------
- the eight of the evidence -
- the key lines of evidence
- the mode of carcinogenic action understanding
- alternative hypotheses
- uncertainties in the nsk assessment
- scientific basis for the risk assessment dose-response choice (i.e., linear versus
nonlinear default approaches)
The peer reviewers shall indicate where they are in agreement with the report and where they
disagree. If they disagree with any part of the report or find a weakness in the report, they
shall recommend explicit guidance on revising the report. They shall provide comments that
include an overall general summary on the acceptability and adequacy of the risk assessment,
and specific comments as needed (comment I on page X, paragraph X, line X).
Bromate Hazard Characterization 40 September 30, 1998

-------
External Peer Review
Comments on Bromate l-lazard Characterization
General:
I All doses should be shown as adjusted for Br0 3 rather than potassium salt for
consistency, to aid reader in making compansons, and because the bromate ion is the
agent of concern.
We agree All doses in the document text and tables have been adjusted
from doses of the salt to doses of the bromate ion itself.
Introduction:
2. It would be useful to include data on how much bromine is converted to bromate ion
dunng ozonation. Also, include the typical bromate levels that can be found in dnnking
water
This information was added to the Introduction on page 5 of the revised
document.
Hazard Characterization:
3 A summary statement like “The weight of evidence shows that bromate is genotoxic in
both in vivo and in vitro systems” is needed here.
This statement was added to the Hazard Assessment on page 6 of the
revised document.
4 There is no evidence for bromate-induced carcinogenicity in humans from
epidemiological studies, as noted in the report. The data presented for possible
toxicological effects from bromate intake is fairly anecdotal and provides a weaker link to
effects in rodents than is supported in the report Thus, extrapolating tumor data from
rodents to humans would appear to be rather weakly based on sound scientific evidence
As discussed mt eh report, it cannot be established whether humans are more likely to
respond like rats or mice to the effects of bromate. This makes it difficult to provide an
acceptable nsk assessment, since one based on rat data, given the limited toxicological
human data, might well be considered fallacious rather than just conservative. Despite
these arguments, it remains that using the rat tumor data from DeAngelo et al., (1998) is
the most viable of the approaches for conducting a human cancer nsk assessment. The
report needs to reflect these uncertainties a little more explicitly.

-------
The paragraph in the draft document which draws conclusions regarding
the similarity between human and animal responses following short-term
exposure was deleted from section 2 1. The uncertainties associated with a
lack of human data and with extrapolating from rats to humans is
discussed in greater detail in the Risk Characterization section of the
revised document on page 25
5. There may be sufficient data to characterize the dose-response for noncancer
(degeneration necrosis, droplets and regenerative changes in the kidney, urothelial
hyperplasia, and serum chemistry changes). Suggest evaluating this endpoint.
Although the three principal studies of chronic bromate toxicity
(Kurokawa et al., 1986a, 1986b, DeAngelo et al., 1998) seem to support
the qualitative conclusion that kidney toxicity may be the critical effect for
noncancer risk assessment, there appears to be insufficient dose-response
information presented in any of the studies to conduct a quantitative dose
response assessment. Therefore, the primary focus of this document has
remained the cancer characterization and dose-response assessment.
6. Add historical control values for tumors observed in rats to put the numbers into
perspective in terms of their biological significance.
This information was added to Section 2.2.2 of the revised document on
page 8.
7 Isn’t it reasonable to suggest that the formation of oxygen radicals could also result in
noncancer kidney effects?
Yes, we agree that it is reasonable to suggest that the formation of oxygen
radicals could result in noncancer kidney effects. However, since the
focus of the document was determined to be the cancer characterization of
bromate, we chose not to discuss the possible mechanisms of noncancer
toxicity.
8. In evaluating the critical study, the report indicates that “A weakness is the fact that
the study was only conducted in male rats and mice...” However, this may be overstating
the case. Males appear to be as or more sensitive than females for most endpoints.
Therefore, it is reasonable to use male data and to assume that data from males will be
valid for females.
The statement indicated above was deleted from the revised document.
The revised document now discusses the findings of Kurokawa et al
(1 986b) regarding male and female rats and indicates that it is reasonable
to assume male data will be valid for females (page 18).

-------
9 The rodent tumor data certainly indicate that bromate induces tumors in rats although
the n iture of the dose response is not well defined and the incidences are quite different
among the three studies discussed The data for the mouse are even less clear and the
reasons for the differences in outcome among the three studies are not readily explicable
on the basis of the expenmental design. The report needs to be more reflective of the
differences among the vanous studies (incidences of tumor types, effects in mouse and
female rats) rather than presenting a consensus view and proceeding with the risk
assessment on this basis
We agree The discussion of the differences among the various studies has
been expanded in Section 3.1 of the revised document on pages 18-19.
10 The key lines of evidence for the potential for bromate to induce tumors in humans
by drinking water exposure are tumor induction in rats and mutagenicity in rodents.
Other information related to the potential for tumor induction is even more circumstantial
than this. The report needs to place these key pieces of evidence into a clearer framework
of uncertainty given the nature, quality, and consistency of the data.
We agree The Risk Characterization section of the document (Section 4,
pages 24-26) has been revised to present the uncertainties of the
assessment as suggested by this comment.
Mode of Action:
Ii Add a summary statement for genotoxicity in Mode of Action section.
This statement was added to the Mode of Action section, page 11 of the
revised document.
12. Add a statement that in the absence of a nonlinear biologically based model, dose-
response modeling should be linear.
This statement was added to the Cancer Hazard Characterization section,
page 15 of the revised document.
13 The role of oxidative stress in the development of tumors is very speculative. The
report needs to present the high level of uncertainty in the proposed mode of action or
perhaps, more appropriately defer to a conservative linear extrapolation based upon the
available data. The discussion of the pertinent data for deriving a mode of action is
reasonable; the conclusions are too much of a stretch. It has to remain that the data are
too limited for a description of mode of action to be provided.
We agree, The Mode of Action section (pages 11-15) and the Cancer
Hazard Characterization section (page 15-17) have been revised to reflect

-------
the uncertainties associated with determining the mode of action En
addition. the document has been revised to conclude that an assumption of
linear dose response modeling is appropriate yen the uncertainties
regarding mode of action
14 The evidence to support a role for alpha 2 microglobulin in male rat responses is
lacking and this should not be used as part of the mode of action.
We agree This discussion was deleted from the revised document.
15 At this time an alternate hypothesis for mode of action, while plausible in a
theoretical sense, is not appropriate based on available supporting evidence.
We agree. The sections of the draft document discussing alternative mode
of action hypothesis and non linear dose response assessment have been
deleted from the revised document.
16 It is not correct to state that the dose response modeling may underestimate risk,
rather the linear potency may be underestimated. Whether risk itself is underestimated
depends on many factors.
The entire paragraph which contained this statement was extensively
revised. The revised document no longer states that the dose response
modeling will underestimate cancer risk. Rather, the advantages and
disadvantages of using the Weibull time-to-tumor analysis on the available
data sets are discussed in more detail (pages 18-19 of the revised
document).
Risk Characterization:
17. Table 3-3 appears to combine tumors from all sites for analysis. This is not
consistent with current practices where each tumor site is considered to be an independent
event unless it is shown that there is a common mechanism. This is particularly
problematic in this case where the tumors are suspected to have different mechanisms
We agree. The draft document did not clearly explain the modeling
approach used. In the revised document, it is more clearly stated that the
data from each of the sites (kidney, thyroid, peritoneum) was modeled
separately (page 19). The results of the independent modeling of the three
sites is presented in Table 3-2 on page 21 of the revised document. Then,
to determine the total unit risk from multiple tumor sites, the individual
unit risks were summed across tumor sites and an estimate of the 95%

-------
upper bound on the sum was calculated using a N1onte Carlo analysis
(page 21-22 of the revised document)
18. Is there a need to mention that children are not more sensitive than adults to the
carcinogenic potential of bromate 7
Yes. The revised document includes a statement that the nsk estimate is
considered to be protective of susceptible groups, including children (page
26 of the revised document)
19. The report needs to adequately reflect the uncertainty and if appropriate identify how
the uncertainty can be reduced. Areas of uncertainty to be addressed - evidence for
extrapolating from rodents to humans is weak; choice of which data set to model cannot
be readily established; mode of action not established, no reason to depart from a linear
extrapolation
We agree. The Mode of Action section (pages 11-15), the Cancer Hazard
Characterization section (page 15-17), and the Risk Characterization
section (page 24-26) have been revised to reflect the uncertainties
associated with this assessment as suggested by this comment.
20. In the third paragraph of the “nonlinear dose-response analysis” section it is stated
that the mouse is less sensitive than the rat to the carcinogenicity of bromate. While this
is true, I would like to point out that this is not at all unique and a factor of —2 in this case
is well within the intraspecies safety factor of lOX.
The Nonlinear dose response analysis section has been deleted from the
document.
21 fri the nonlinear dose response analysis, there should be some justification that
oxidative damage is a nonlinear mode of action
The Nonlinear dose response analysis section has been deleted from the
document.

-------
Comments on Health Risk Assessment/Characterization
of the Drinking Water Disinfection Byproduct Bromate
Prepared by
A. Julian Preston, PhD
Chemical Industry Institute of Toxicology
6 Davis Drive
Research Triangle Park,
NC 27709
USA
February 14, 1998

-------
Introduction
This peer review of the Health Risk Characteriza on Report for bromate provides a
discussion of the key data and analyses that are utilized for the risk characterization, together with
a discussion of the appropriateness of the mode(s) of action selected for incorporation into a dose
response model. Additional information was obtained from the US EPA 1994 Criteria document on
bromate and the manuscript ‘Bromate carcinogenicity in rodents’ DeAngelo et al., Toxicologic
Pathology 26, in press. In general, the Health Risk Characterization Report for Bromate
representS a sound initial attempt to develop a cancer risk assessment for bromate based on the
rather limited data available.
Weight of the Evidence
There is no evidence for bromate-induced carcinogenicity in humans from epidemiological
studies, as noted in the report. The data presented for possible toxicological effects from bromate
intake is fairly anecdotal and provides a weaker link to effects in rodents than is supported in the
report. Thus, extrapolating tumor data from rodents to humans would appear to be rather weakly
based on sound scientific evidence.
The rodent tumor data certainly indicate that bromate induces tumors (kidney, thyroid and
testicular) in rats, although the nature of the dose response is not well defined, and the incidences
are quite different among the three studies discussed. No explanation can be offered for this latter
observation. The data for the mouse are even less clear. No significant increases in tumor
incidence were noted by Kurokawa et al. (1986). In contrast DeAngelo et al. (1998) did descnbe
an increased renal tumor incidence in male mice, although this was not dose-responsive, and with
the significant increase (p
-------
DeAngelo et al (1998) is the most viable of the approaches for conducting a human cancer risk
assessment. The report needs to reflect these uncertainties a little more explicitly.
The mutagenicity data are relatively sparse, but do allow for the conclusion that bromate
given by gavage or i.p. can induce micronuclei in rodent (rat and mouse) bone marrow cells. In
addition, bromate induces point mutations in Salmonella. Thus, any risk assessment model needs
to be reflective of this, as is the case in the Report.
Key Lines of Evidence
As noted above, the key lines of evidence for the potential for bromate to induce tumors in
humans by drinking water exposure are tumor induction in rats and mutagenicity in rodents. Other
information related to the potential for tumor induction in humans is even more circumstantial than
this. The Report needs to place these key pieces of evidence into a clearer framework of
uncertainty given the nature, quality and consistency of the data.
Mode of Action Understanding
The Report suggests that bromate may be carcinogenic by more than one mechanism.
Based upon fairly limited data, it is suggested that kidney tumors could be formed by oxidative
stress leading to DNA damage (8-hydroxydeoxyguanine) and subsequently to mutations.
Tumors at other sites are proposed to be a consequence of nonoxidative stress modes of action,
given the induction of 8-OHdG in rat renal samples by bromate treatment. While the presentation
does accurately describe the available data by which such a conclusion might be reached, it
needs to be emphasized that the conclusion is speculative. There has been no assessment
of other DNA adducts, no relationship has been shown between 8-OHdG and any adverse
cellular outcome, and no evidence for what might be producing effects in nonrenal tissues absent
oxidative stress. In a similar vein the mutagenicity data in vivo are for bone marrow cells that are
not a target cell type for tumor induction. This does not negate their general utility in developing a
mode of action, but their direct relevance to a cancer nsk assessment model is weak.
The evidence to support a role for d2u globulin enhanced cell proliferation in male ¶at
responses is lacking, and this should not be used as part of the mode of action.
The Report needs to present the high level of uncertainty in the proposed mode of action,
or perhaps more appropriately defer at this time to a conservative, linear extrapolation based

-------
upon the available data. At this point, there seems to be no sound justification for concluding that
the modes of action proposed (page 13, lines 507) are relevant to humans and provide support
for a conclusion that the animal evidence of carcinogenicity is applicablç to humans. The
discussion of the pertinent data for deriving a mode of action is reasonable, the conclusions are
too much of a stretch. It has to remain that the data are too limited for a description of mode of
action to be provided.
Alternative Hypotheses
There seems to be an indication that DNA damage induced by bromat can lead to
mutations if there is adequate cell prol eration, and this might ultimately lead to tumor formation.
However, the various components of the model are not supported by adequate research data. At
this time, an alternate hypothesis, while plausible in a theoretical sense, is not appropriate based
on available supporting evidence.
Uncertainties in the Risk Assessment
The major uncertainties in the risk assessment fall out from the discussion presented
above. The evidence for extrapolating from rodent tumor data to humans is relatively weak, given
the paucity of human data. The choice of rat or mouse data as being the more relevant for human
cancer nsk is undetermined. The choice of which rodent data set to model cannot be readily
established, although the choice in the Report of those from DeAngelo et al. (1998) is as
reasonable as any. The role of mutagenicity in tumor induction is supported by micronucleus data
from rodents and Salmonella mutation data. However, the mode of action whereby mutations
might be induced is very uncertain and considerable additional data are needed to strengthen a
model. Thus, the choice of a risk assessment dose response remains somewhat elusive. There
does not appear, at this time, to be any convincing reason to depart from a linear extrapolation
from the observed range for tumors, with the proviso that the selection of the particular tumor data
has to be more firmly established.
The margin of exposure approach provides some useful guidance, and perhaps would
represent the most viable method of addressing the risk assessment for bromate. However, the
existence of a non linear dose response curve for tumor induction remains elusive.
Conclusion

-------
It is of particular importance for the Report to adequately reflect the uncertainty and, if
appropnate, identify how the uncertainty can be reduced. In addition, there is a need to make the
document more precise (a few examples are given below). The intent of the Report is
appropriate and clearly noted; it is the treatment of the available data that needs some attention.
Examples of Minor Points for Preciseness
Page 4, para 2, line 3. The three long term studies evaluated carcinogenicity not ‘potential’
carcinogenicity - the potential is to humans.
Page 4, para 3, line 2. Need to explain ‘mechanism of bromate’ for what.
Page 4, para 4, line 1. Need to explain ‘Limited data’ for what.
Page 4, para 4, line 4. Non cancer effects need to be identified.
Page 5, para 2, line 4. Need to explain what is being extrapolated.
These are a few examples on the first few pages; it is necessary to read the document for
other areas that need to be more specific.

-------
r u.
Ernest E McConnell. U V M M S (Path)
OlIke 1 *Jephone U)2S I thtn Line
91 cm I .aijicthne Fst
Raleigh N( 27( I
12 February I9O
To ViLky lklarco
Subj I l RA Risk Assessment on Uromaic
l 1 ci voul Icqucs$ following is my rcvicw of the draft document “HedIth Risk As cs ment/
(iiaiacieii,aiion of Drinking Water Disunfcction Byproduct Bromate” Ii represents my prrsonuil
()j)itliUfl based on the supporting documents and my e penence in these malteis I have attciiiptcd
to follow the format outlined in the sow 1 ask I)escription
()vi’i all the repot adequately addrecses the 1) weight of evidence. 2) key lines of evidL’I% C
imiode of cat cino cnIc action. 4) alternative hypotheses. S) uncertainties in the misk assessm in
arid, 6) the ccieniilic basis for thc risk asscssmcnt dose-response thot e in thorough yet cir e
and lu i il niarinri i e it reads eIl Howevcm, itreic arc scvcral comiTIerItsdsu cstrons that I
hunk need to be acldiccscd which could enhance the document
vei r ching olni rent 1 think ould iniprt i t the document, particulat ly iii let ins ol Its
i eadatulit’ is to rice a ccrncictent approaL h to t(lr expressing exposui c/dose It is n v
understanding thdt the potential health problem is as follows bromine tori rn drinking waler has
the 1 )ot(Iitial ho bc convened to 13r0’ non as a result of ulunatlofl Yet, h fl1’CC il) all iiiC
animal stmndie’ use KRtO’ lot exposurc 1 his presents a problem or the reader Ii i conhising to
CL data when dose is referred to as KRrO 4 in one case and BrO iii thc next I think there iieedc
to lie moie comrci Ieflcy and suggest that all doses be shown as the HiD’ ion since tl r t r li ,ii
art’ eKpo eci to although it is aI o irnponanl ho shuw the dose expic cd a K ft ()‘ I hi
sugeest inn includes the tables
I 0 mu uIiu iu,I, — I think it ‘ ould he useful I 0 include data in the mtroduci i On t hat ‘ lii iw h ow
ntmictt hturnmnc is converTed to RrO’ tri water dii i u mg oi,onation (as a ‘o) as desL rihed in the
l’.l’A (‘nteri i Document Also, include the lcvds of Rr0 3 that can he es pectcd in typical
di inking walem (It is gtvcn later in the text hut would he useful hei e)
1 ( I mu s J,a ,ui,, — I’m not sum c that the stdtcnrCnl ‘ tire; e ai c iflsullicicnt dosc.rcspuii e dat,i iii
htiimmns mcl uiimalc i cr adequately charactcri7c the noncaricer eulects ot’ hi ornate” is
wai ranted (see later comment)

-------
2 0 J/uzczrtl ( hc,iacre,,:atioi, - summary statement like ‘The eiglit of the evidence shows that
Br0 1 is gcnotoxic in both in vitro and in i’ivo 5)StemS’ is needed here
2 2 1 Nui,cai, tr J ffec1% — As noted abo e there may be suflicient data to charactcrizt’ (he dose
response for noncancer endpoints Why couldn’t one use the degcneration, necrosis.
droplets and regenerative changes in the kidney, urothelial hyperplasia arid scrum
chemisti j changes (especially BUN) for this? See tables 2, 6, 8 and especially 10 in the
DeAnge o paper 1 suggest a table similar to Table 3-5 to see how this sorts Out.
I would lso like to point out that both the Kurokawa and EPA studies appear to be well
designed and conducted studies, in particular the pathology seems to be well
characrčized Therefore, these data can be viewed as valid
2 2 2 (‘arci1?Og4 .ni i1y - This is where we start getting into the problem of comparing the dose
of KBrO’ to Br0 3 and how this is translated into human exposure This needs to be
reflecte4 in Table 2-1
I would include the “historical rates” of these tumors in this strain of rat’. to put the
.numbers into perspective in terms of Pteir biological significance
2 4 M4e ojAciwn Need a summar ’ statement for genotoxicity
2 4 Isn’t it rcasonable to suggest that the iirrnation of o.xygeri radicals could also tesult in
Cy toiox ciiy, e noncancer kidne> effer_is’
3 1 (‘Iioicc ofCriiica/Siudy - The last sentence A v eakness is ihe fact the study was only
conductçd in male rats and mice, “may he overstating the case Males appear to be
as or more sensitive, than females for most endpoints Therefore. 1 think that it is
reasonable to use male data and and assume that it will bc valid for females
3 2 2 (juanwlAiialy.ccs - Table 3-3 appears to combine tumors from all sites for analysis (last
line) 1 think this is not consistent with current practices where each tumor site is
considei c1 to be an independent e ent unless it is shov n that there is a common
mechanism This is particularly problematic in this case where the tumors are suspected to
have different mechanisms
The inlei pretatson of Table 3-4 may need to be rethought in light of the historical con
incidence of the key tumor types Using NTP data, the historical rates for male rat (this
strain) kIdney tumors (adenomas and carcinomas combined) as reported by the National
Toxicology Program isO 6% (range = 0-6%), thyroid 1 2°io (0-7%). and mesothelioma
2 7% (0.10%)
2

-------
I able i. i the key t hle in my opinion and I think would benefit by c 1e’. ’ Ii1 tile dose
in terms of RrO
I able .S I would like to otler the t llowing interpretation of the data in this isbie i
light of lhc above historical data and the control data in the studies under consideration
!cli ! yjujppr - I think the data suggest that (he LUEL is 7 9 mg/kg based on the
incidence in these tumors in the EPA study (6i47)
McioJh hijrn — I think the data suggest that the LOAL is 16 9 mg/kg based on an
incidence of 10/47 in the EPA study The incidence of S/49 at 7 9 mg/kg and 4/49 at 1 c
mg/kg are within the historical controls (although oii the high side) and are less than thc
Kurokawa study (6/53)
Thyry c tuj .q — think the data suggest that ihc LOAL is also 16 9 mg/kg thr this
tumor, again based on the EPA study I have discounted the 4/39 at 15 mg/kg because of
Lhc lack of any effect at 79mg/kg in this study and at 12 S m /kg Kurokawa) lii fact
one could make an argument thdt the LOAL is dctualiy closet to 27 7 mg/kg Furthei
support for this is the fact thdt there wasn’t any dose-related increase in foiliculai ccli
hyperplasia as one would expect linally. it needs to be remembered that the diagnosis of
thyroid fohlicular adenomas is fairly arhurary. c g, based on the sire of the Ie ion (tIle izc
ofthiiee contiguous follicics, and has little relevance to the biological potential of the
lesion
I dont know what this does to the RA calculations but this is how I interpret the d iia
; ,\ / € ‘ / 1 _I / ponc Anil i in the fluid paragraph of thi; section but it is stated
that the mouse is less sensii e than the rat to the carunogenicii ofhromatc Vhile ibis is
true. I LIuld like to point ou that this is not it )) unique and a factor of 2iri ihi case is
well within the intraspccics safet factor of lox
I 0 1?, h ( ( I C fcr,:cmflw, Ptt e 2 S para m aph 2 It is stated t hut Modeling w s conduL c cl
on the individual tumor types and on the com ’ined incidence of tumors at alt ‘ mte Lven
though ii doesn’t make any difference in this case. I don’t think ii is appropriate to
combinc tumors from various sites, particiiIail hen ii is felt thai the mechanisms of
induction is thotight to be different as in this case I don’t think ills needed and h
doing this suggests that it is a proper procedure
1 would like to stress that the TAR t document. as currently written is iii fairly good shape Most
of what I have pointed out represents fine.tunmng”
Fumiatly does there need to he any mention of special concein for children’ in this document, ii
fui no oilier icacon than you thought about it Ihere appear tube data that suggest that cluldmen
are 1101 more sensitive than adults to the carcinogenic potential of bromate in the watel cmilcrma
document
3

-------
1t you h e qu $lions concerrnn this report pkac fc J f.rcc o gI c rue a cal)
Sincerely.

-------
Comments on the TERA Draft Risk Characterization Document for Bromate
Prepared by: Cohn Park, Dow Chemical Company
Generally I thought the document was well done as a summary document.
It gives sufficient, but not overwhelming detail and the
conclusions/interpretations are clear. Following are a number of
relatively minor suggestions for improving the flow and clarity.
1. Page 9, sec. 2.3. line 6
There appears to be a typo. 0.25 ppm should probably be 25 ppm.
2. P.O, sec. 3 ,flne 2
extra right parenthesis after NTP
3. P.13. lIne 5
delete “In”
4. P.13. first full paragraph, first sentence.
Insert “in the absence of a nonlinear biologically based model”
between
“modeling bromate” and “Since”
5. P.13, se3.O
unit abbreviation should be mg/L not mi t .
8. P.14,Une5.
Replace
“Therefore modeling based upon this data may underestimate risk.”
“Therefore the linear potency may be underestimated”
Whether risk itself is underestimated depends upon many factors
Including animal and man relative sensitivities, accuracy of the linear
model, etc.. Not correcting for early mortality is likely a small issue
relative to bigger assumptions.
7. P.22. sec 3.3, first sentence
There should be some justification that oxidative damage is a nonlinear
mode of action
8. P. 25, second full paragraph
Reword as follows:
Based on linear extrapolation, a cancer potency estimate for bromate
ion is 0 70 per mglkg-day. This potency estimate corresponds to a
drinking water unit risk of 2 x 10-5 assuming a daily water consumption
of 21/day for a 70 kg adult. Lifetime cancer risks of “, I a.
change the punctuatIon.
This paragraph raises a philosophical question which I have wondered
about. In reporting results such as this, does EPA Intend to report
risks of 10-6 .10-5. and 10-4
or
LED1O/1 00000. LEDIO/10000 and LEDIO/1 000?
It has never been clear to me how EPA was going to report these levels
One purpose of the BMD methodology was to get away from the perception

-------
of reporting nsk, therefore the second interpretation would appear to
be more reasonable
9. The document states in the introductIon that there Is Insufficient
data on noncancer effects to tharacterize the dose-response. However,
the 1998 DeAngelo study used 5 doses, Including control, for male rats
and 4 doses for male mice. in these studies, body weights organ
weights, serum chemistry and histopathology were collected. It Is not
clear what parameters were measured in the Kurokawa studies.
Additionally a reproducth,e/deveiopmental study was conducted In rats
over an almost idendcal dose range, albeit for a different stain.
Why is their lnsuffldent data to evaluate noncancer effects?

-------
Appendix D
Personal communication from Douglas Wolf,
National Health and Environmental Effects Research Laboratory,
U.S. EPA
to VickiDellarco, Office of Water, U.S. EPA.
February 20, 1998.
Bromate Hazard Charactenzanon 44) March 13, 1998

-------
Subject One more question -Reply
Date Fri, 20 Feb 1998 10 0356 -0500
From DOUG WOLF 
cc
DEANGELO ANTHONY@EPAMAIL EPA GOV,HAUCI-IMAN FRED@EPAMAIL
EPA GOV, NESNOW STEPI- [ EN@EPAMA [ L EPA GOV
If you look at the Excel tables that Mike George sent up they tell the
week of death for each animal You will see that except for a few
random animals in each dose group, most of the animals in all but the
high dose group survived well past 80 weeks Animals did not really
begin to come off the study in the high dose group till after 70
weeks on study The high dose group was terminated at 94 weeks
because 60% of them had significantly extensive mesothelioma burdens
which caused the terminal weight loss (cachexia of cancer)
The earliest mesothelioma was seen in the interim time point after 52
weeks of treatment This was also the earliest time point for renal
tumors The thyroid tumors were first seen after 26 weeks of
treatment Time to tumor like this may have more to due with the
particular biology and latency of a tumor type rather than the
potency of a chemical carcinogen
As far as which mathematical model to use, that is out of my area of
expertise, but I would strongly suggest the one that most closely
shadows the biology of the response endpoint of interest
>>> VICKI DELLARCO - 2/20/98 9 39 AM>>>
Doug, did the mortality for bromate occur later toward the terminal
sacnfice or earlier in the study 9
A.lso, when did the mesotheliomas occur 9 Early in sacr ifice 7
Bromate Hazard Characterizauon 41 March 13, 1998

-------
February, 1998
Dr. Douglas Wolf, NHEERL/ORD/EPA
Dr. Vicki Dellarco, OST1OWIEPA
Survival Curves from the DeAngelo et. aL, 1998 Study
1
0.8
0.6
0.4
0.2
0
0 10 20 30 40 50 60 70 80 90 100
C
4 . ’
C
(-)
I-
a
FROM:
TO:
RE;
weeks of treatment

-------
February, 1998
FROM: Dr. Douglas Wolf, NHEERLJ ORDI EPA
TO: Dr. Vicki Dellarco, OST/OWIEPA
RE: Survival Curves from the DeAngelo et. al, 1998 Study
60%
40 50 60 70 80 90 100
120%
lOG
80%
—
0)
>
I-
(I)
U,
E
( 0
0
C
C,
0
a
40%
20%
0%
0 10 20 30
weeks of KBrO3 exposure

-------
Bromate Hazard Characterization 41 September 30, 1998

-------
Bromate Hazard Cbaractenzatiofl 42 September 30, 1998

-------
Bromate Hazard Characterization 43 September 30, 1998

-------
Appendix D
Personal communication from Douglas Wolf,
National Health and Environmental Effects Research
Laboratory, U.S. EPA
to Vicki Dellarco, Office of Water, U.S. EPA.
February 20, 1998.
Bromate Hazard Characterization 44 September 30, 1998

-------
Subject One more question -Reply
Date Fri, 20 Feb 1998 10.03 56 -0500
From DOUG WOLF 
CC DEANGELO ANTHONY@EPAMAIL EPA GOV,HAUCHMAN.FRED@EPAMAIL.
EPA GUy, NESNOW STEPHEN@EPAMAIL EPA GOV
If you look at the Excel tables that Mike George sent up they tell the week of death for each
animal. You will see that except for a few random animals in each dose group, most of the
animals in all but the high dose group survived well past 80 weeks Animals did not really
begin to come off the study in the high dose group till after 70 weeks on study The high
dose group was terminated at 94 weeks because 60% of them had sigruficantly extensive
mesothelioma burdens which caused the terminal weight loss (cachexia of cancer). The
earliest mesothelioma was seen in the interim time point after 52 weeks of treatment. This
was also the earliest time point for renal tumors. The thyroid tumors were first seen after 26
weeks of treatment. Time to tumor like this may have more to due with the particular
biology and latency of a tumor type rather than the potency of a chemical carcinogen. As far
as which mathematical model to use, that is out of my area of expertise, but I would strongly
suggest the one that most closely shadows the biology of the response endpoint of interest.
VICKI DELLARCO - 2/20/98 939
Doug, did the mortality for bromate occur later toward the terminal sacnfice or earlier in the
study 9
Also, when did the mesotheliomas occur? Early in sacrifice?
Brornate Hazard Characterization 45 September 30, 1998

-------
Insert Sur i aI Cur ’e for rats
Bromate Hazard Charactenzation 46 September 30, 1998

-------
Insert Sur i aI Curve for mice
Bromate Hazard Characterization 47 September 30, 1998

-------
EJBD
ARCHIVE
EPA
815—
B-
98-
007

-------