EPA/600/R-05/025
February 2005
Trichloroethylene Issue Paper 4:
Issues in Trichloroethylene Cancer Epidemiology
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20460
-------
DISCLAIMER
This issue paper does not represent and should not be construed to represent any agency
determination or policy. This issue paper has not been externally reviewed. The information is
being provided to assist the National Academy of Sciences in their review of the scientific issues
surrounding trichloroethylene health risks.
-------
CONTENTS
LIST OF TABLES iv
LIST OF ABBREVIATIONS AND ACRONYMS v
PREFACE ! vi
AUTHORS AND CONTRIBUTORS vii
THE TCE ISSUE PAPERS viii
1. INTRODUCTION AND PURPOSE 1
2. EPIDEMIOLOGIC STUDIES ON CANCER EFFECTS AND TCE EXPOSURE 2
2.1. RECENT COHORT STUDIES 3
2.2. CASE-CONTROL STUDIES 6
2.3. COMMUNITY STUDIES 7
3. ISSUES RELATED TO TCE EPIDEMIOLOGIC EVIDENCE 9
3.1. NARRATIVE AND QUANTITATIVE METHODS 11
3.2. SOUNDNESS OF DATABASE 15
3.2.1. Studies of Morbidity or Mortality 15
3.2.2. Exposure Assessment Issues in Epidemiologic Studies 17
3.3. NON-HODGKIN'S LYMPHOMA 20
3.4. CONSIDERATIONS FOR META-ANALYSIS 22
4. SUMMARY 25
REFERENCES 33
m
-------
LIST OF TABLES
1. Epidemiologic studies assessing cancer and trichloroethylene exposure 26
2. Causal inferences as reported in reviews of TCE body of evidence 28
3. Summary of relative risks in analyses using meta-analysis methods 31
IV
-------
IARC
ICD
JEM
JTEM
MOR
NAS
NHL
NTP
OR
PMR
RCC
REAL
SAB
SEER
SES
SIR
TCA
TCE
U-TCA
VHL
LIST OF ABBREVIATIONS AND ACRONYMS
International Agency for Research on Cancer
International Classification of Diseases
Job-exposure matrix
Job-task-exposure matrix
Mortality odds ratio
National Academy of Sciences
Non-Hodgkin's lymphoma
National Toxicology Program
Odds ratio
Proportional mortality ratio
Renal cell carcinoma
Revised European-American Lymphoma (classification)
Science Advisory Board
Surveillance, Epidemiology, and End Results (program)
Socioeconomic status
Standardized incidence ratio
Trichloroacetic acid
Trichloroethylene
Urinary TCA
Von Hippel-Lindau (gene)
-------
PREFACE
Publication of these issue papers is a part of EPA's effort to develop a trichloroethylene
(TCE) human health risk assessment. These issue papers were developed to provide scientific
and technical information to the National Academy of Sciences (NAS) for use in developing
their advice on how to best address the important scientific issues surrounding TCE health risks.
As such, these papers discuss a wide range of perspectives and scientific information (current
through Fall 2004) on some of these important issues, highlighting areas of continuing
uncertainty and data that may be relevant. They are intended to be useful characterizations of
the issues, not a presentation of EPA conclusions on these issues. The papers have undergone
internal review within EPA, but they have not been externally reviewed. The concepts presented
in these papers will eventually be addressed in EPA's revised risk assessment of TCE, after the
advice from the NAS, along with comments from the EPA Science Advisory Board and the
public, as well as recently published scientific literature, have been incorporated.
VI
-------
AUTHORS AND CONTRIBUTORS
V,
Many individuals contributed to the completion of this set of tichloroethylene (TCE)
issue papers. The TCE Risk Assessment Team identified the topics covered by the papers and
prepared them for submission to the National Academy of Sciences. The authors wish to thank
Dr. Peter Preuss, Dr. John Vandenberg, Mr. David Bussard, Mr. Paul White, Dr. Bob Sonawane,
Dr. Hugh Barton, Dr. Aparna Koppikar, Mr. David Bayliss, Dr. William Wood, and Dr. Ila Cote
for their input and comments.
TCE Risk Assessment Team
Jerry Blancato, ORD/NERL
Jane Caldwell*, ORD/NCEA
Chao Chen, ORD/NCEA
Weihsueh Chiu*, ORD/NCEA (TCE Chemical Manager)
Marina Evans, ORD/NHEERL
Jennifer Jinot, ORD/NCEA
Nagu Keshava*, ORD/NCEA
John Lipscomb*, ORD/NCEA
Miles Okino*, ORD/NERL
Fred Power, ORD/NERL
John Schaum, ORD/NCEA
Cheryl Siegel Scott*, ORD/NCEA
* Primary authors of issue papers. Entire team contributed via review and comment.
vu
-------
THE TCE ISSUE PAPERS
BACKGROUND
In August 2001, a draft, Trichloroethylene (TCE) Health Risk Assessment: Synthesis and
Characterization, was released for external review. This draft assessment drew on 16 "state-of-
the-science" papers published as a supplemental issue of Environmental Health Perspectives
(Volume 108, Supplement 2, May 2000). Subsequent to its release, EPA's 2001 draft
assessment underwent a peer review by a panel of independent scientists through EPA's Science
Advisory Board (SAB), which provided a peer review report in December 2002. In addition, the
public submitted more than 800 pages of comments to EPA during a 120-day public comment
period.
There are a number of important issues that EPA will need to examine as it moves
forward in revising the draft TCE assessment. These include issues raised not only in the SAB
peer review and public comments, but also by new scientific literature published since the
release of the state-of-the-science papers and EPA's 2001 draft assessment. Some of this
research is specific to the study of TCE or its metabolites while some of it describes advances in
scientific fields more generally but which have potential relevance to characterizing the human
health risks from TCE.
In February 2004, EPA held a symposium so that authors of some of the TCE-specific
research that had been published since the release of the draft assessment could present their
findings in more detail. This symposium represented only a limited cross section of recently
published research, but was reflective of the breadth of new relevant science that EPA will
consider in revising the assessment (the presentation slides and a transcript of the meeting are
available separately on EPA's website and have already been sent to the NAS).
In 2004, EPA, in cooperation with a number of other federal agencies, initiated a
consultation with the National Academy of Sciences (NAS) to provide advice on scientific issues
related to the health risk assessment of TCE. It was recognized that a review by an NAS panel
of the important scientific issues would be beneficial and informative to clarify the state-of-the-
science as EPA moves forward in completing its health risk assessment. A charge was
developed for the NAS through an Interagency Workgroup led by the White House Office of
Science and Technology Policy.
PURPOSE OF THE TCE ISSUE PAPERS
Although EPA will need to address all of the issues identified in the charge to the NAS
panel in updating its assessment, EPA would like to focus the NAS panel's attention on a subset
Vlll
-------
of issues that EPA believes to be most critical in developing a revised risk assessment, as
summarized in four issue papers developed by EPA staff:
1. Issues in trichloroethylene pharmacokinetics;
2. Interactions of trichloroethylene, its metabolites, and other chemical exposures;
3. Role of peroxisome proliferator-activated receptor agonism and cell signaling in
trichloroethylene toxicity; and
4. Issues in trichloroethylene cancer epidemiology.
Each paper provides an overview of the science issues, a discussion of perspectives on
those issues (including the SAB and public comments), and an outline of some of the recently
published scientific literature. The pharmacokinetics issue paper also summarizes results from a
recent collaboration with the U.S. Air Force on TCE pharmacokinetics, as well as EPA's
planned approach for further refinement of the pharmacokinetic modeling of TCE and its
metabolites. These scientific areas were selected because they are (a) critical to the hazard
and/or dose-response characterization of TCE; (b) scientifically complex and/or controversial;
and (c) areas in which substantial important scientific literature has been recently published. The
input from the NAS on the topics described in the issue papers, as well as other topics put forth
in the charge to the NAS, should help to strengthen EPA's revised TCE assessment.
NEXT STEPS
The advice from the NAS, along with comments already received from the EPA SAB
and the public, as well as recently published scientific literature, will be incorporated into a
revised EPA risk assessment of TCE, strengthening its scientific basis. Because of the
substantial amount of new information and analysis that is expected, the revised draft of the
assessment will undergo further peer review and public comment prior to completion.
IX
-------
1. INTRODUCTION AND PURPOSE
Epidemiologic evidence holds a unique place in an assessment of potential environmental
risk for a number of reasons. Well-conducted epidemiologic studies that show a positive
association between an agent and a disease are accepted as the most convincing evidence about
human risk (NRC, 1983). First, epidemiologic observations are from humans, the target
population for risk assessment practices. Inferences of human risks based on data from
epidemiologic studies can have fewer associated uncertainties than inferences of human risks
based on data from rodent species. Hertz-Picciotto (1995) notes uncertainties associated with
uncontrolled bias or errors in exposure assessment are likely to be less than those stemming from
interspecies extrapolation. Additionally, the patterns and exposure concentrations of
epidemiologic-studied populations are likely to be closer and more similar to the exposure
scenarios of populations for which risks are inferred than exposures of 2-year rodent bioassays.
Furthermore, national and international organizations such as the National Toxicology Program
(NTP) and the International Agency for Research on Cancer (IARC) place greatest weight on
epidemiologic evidence in their overall evaluation of human carcinogenicity (NTP, 2002;
Siemiatycki et al., 2004). Although epidemiologist have become more introspective with
questions on the utility of epidemiologic studies for providing insight on associations of relative
risks of 2.0 or lower (Taubes, 1995, Monson, 1980), epidemiologic evidence continues to inform
the hazard step in risk assessment and is increasingly examined for dose-response inferences.
Epidemiologic data along with toxicological data and data from other biological sciences
provide the foundation for toxicity evaluations of health risk assessment.
The purpose of this issue paper is twofold: (1) to provide the National Academy of
Sciences (NAS) panel with an overview of the epidemiologic evidence on trichloroethylene
(TCE) and the occurrence of cancer, highlighting the new literature published since the U.S.
Environmental Protection Agency's (EPA's) external review draft, Trichloroethylene Health
Risk Assessment: Synthesis and Characterization (U.S. EPA, 2001) and (2) to focus the NAS
panel's advice on approaches to synthesizing the body of epidemiologic evidence as EPA moves
forward in revising the draft TCE assessment. EPA asks the NAS panel to focus on the issues
listed below and discussed in Section 3 as they review the epidemiologic evidence on TCE
exposure. NAS input on how EPA can address the following questions would help to strengthen
EPA's revised assessment:
• What qualitative and/or quantitative approaches can best inform causal inferences of the
TCE body of epidemiologic evidence?
-------
• What are the strengths and limitations of the epidemiologic body of evidence on cancer
occurrence and TCE exposure? Specifically, which studies carry a greater weight in the
hazard evaluation?
- For which site-specific cancers can studies of incidence carry greater weight for hazard
identification than those of mortality?
- What advice can NAS provide EPA on weighting the different exposure assessment
approaches adopted in the TCE epidemiologic studies?
• What advice can NAS provide EPA on factors that may influence interpretation of
epidemiologic observations on non-Hodgkin's lymphoma (NHL)?
• How appropriate and useful is applying meta-analysis methods to synthesize epidemiologic
data on TCE, and what strategies could be employed to implement such an analysis?
This issue paper is not intended to provide a complete survey and synthesis of the
scientific literature on TCE cancer epidemiology. Section 2 identifies the types of epidemiologic
evidence on TCE and carcinogenicity, focusing on more recently published studies. It also
provides brief details of the study designs employed by investigators, the site-specific cancers
examined in these studies, and the information on potential TCE exposure collected for
individual study subjects. The specific methodological issues that arise in analyzing and
interpreting the epidemiologic studies are discussed in Section 3. EPA has received public
comment on the epidemiologic evaluation in their external review draft TCE Health Risk
Assessment (U,S. EPA, 2001) and these comments are also identified and discussed in Section 3.
2. EPIDEMIOLOGIC STUDIES ON CANCER EFFECTS AND TCE EXPOSURE
Three types of epidemiologic study designs are most commonly used to assess cancer
occurrence and TCE exposure: (1) historical or retrospective cohort studies; (2) case-control
studies; and (3) ecologic or community studies. Epidemiologic studies that evaluate TCE
exposure and cancer occurrence are listed in Table 1. They include studies cited in the
Wartenberg et al. (2000) analysis and literature published since 2000, identified from a search of
the National Library of Medicine's PubMed database, (http://www.ncbi.nlm.nih.gov
/entrez/query.fcgi) using the keywords "trichloroethylene epidemiology cancer." The historical
-------
or retrospective cohort study follows a group or cohort1 over time to assess vital status: alive at
end of followup, lost to followup, or cause-specific morbidity or mortality of sick or deceased
individuals. The cohort's health outcomes are compared to those of a reference group, usually
cause-specific morbidity or mortality within the U.S. population or another country's population.
Many of the TCE cohort studies assess mortality, with a few studies of cancer morbidity in
groups of workers who have either documented or presumed exposure to TCE. In some cohort
studies, investigators have assigned TCE exposure to individual study subjects. These cohort
studies have adopted a number of approaches and include the use of biological (exposure)
markers or the job-exposure matrix2 (JEM). Case-control studies are retrospective studies in
which exposure to a suspected agent, e.g., TCE, is determined among the individuals with the
disease (cases) or without the disease (controls)3. Analytical designs such as cohort and case-
control studies are generally relied on for identifying a causal association between human
exposure and adverse health effects and are necessary to infer causality. Ecologic studies are
descriptive epidemiologic studies that examine symptom or disease rates among populations in
relationship to personal characteristics, such as age, gender, race, and temporal or environmental
conditions. Responses, exposure indices, and covariates in an eccological study are at a
community level, and information about individual exposure is lacking. Descriptive studies such
as ecologic studies, alone cannot provide direct evidence of causation, although their results can
provide supporting data concerning a possible causal relationship (NRC, 2005).
2.1. RECENT COHORT STUDIES
More precise exposure assessment methods for qualitative identification of TCE
exposure (versus, for instance, nonspecific solvent exposure) and for developing quantitative
exposure metrics are increasingly included in cohort studies published since 1990. Exposure
assessment in older cohort studies included TCE as a larger category of organic solvents. These
studies were unable to attribute TCE exposure to an individual study subject, thus limiting
inferences about associations between TCE, specifically, and site- or cause-specific cancer. For
epidemiologic data to be useful in determining whether a causal association exists between
health effects and exposure to an agent, adequate characterization of exposure information is
vital. In general, EPA assigns greater weight to studies with more precise and specific exposure
estimates.
1 Cohort is defined as any designated group of individuals followed over a period of time.
* A cross classification of jobs and occupational exposures (Garcia and Checkoway, 2003).
3 A case group is identified with the health endpoint of interest along with the control group without the health
endpoint of interest.
-------
A large body of the TCE cohort studies has described mortality in occupational cohorts,
with cancer incidence assessed as a health endpoint in reports using cancer registries from
Nordic countries, registries with documented high accuracies of recorded cancer incidence. Two
incidence studies on TCE exposure have been published since EPA's draft assessment in 2000.
Hansen and colleagues (2001) reported a study of cancer incidence in 803 Danish workers with
TCE exposure as inferred from biomonitoring of urinary trichloroacetic acid (TCA) (U-TCA) or
from breathing zone measurements. The accompanying papers of Raaschou-Nielsen et al.
(2001,2002) more rully describe exposure assessment methods for this cohort study. Hansen
(2004) reported at an EPA Science Symposium on TCE that the hypothesis of this study was to
examine associations between TCE exposure and liver and biliary tract cancers, kidney cancer,
cervical cancer, and lung and testicular cancers based on previous epidemiologic or bioassay
observations. This cohort of 806 subjects was followed from 1968 to 1996 with 128 cancers
(109 male, 19 female) observed and 123.4 cancers (104.8 male, 18.6 female) expected
standardized incidence ratio (SIR) = 1.04, 95% confidence interval (CI): 0.9,1.2). Expected
numbers of cancer are derived from cancer rates of the Danish populations for the study period.
A second incidence cohort study of 40,049 blue-collar workers further examines the possibility
of an association of elevated risks of certain site-specific cancers with exposure to TCE
(Raaschou-Nielsen et al., 2003). A total of 3,244 incident cancer cases were observed between
1968 and 1997. No information was available on the TCE exposure of individual subjects;
however, Raaschou-Nielsen et al. (2003) attempted to increase the likelihood of a larger
proportion of study subjects with TCE exposures by examining blue-collar workers and
comparing their cancer-incidence experiences with white-collar workers, and by accounting for a
number of factors previously identified in industrial hygiene surveys that increased the potential
for TCE exposure (Raaschou-Nielsen et al., 2001; 2002).
Both studies report statistically significant elevated relative risks for (NHL)4. Hansen et
al. (2001) concluded that their study "identified increased SlRs for non-Hodgkin's lymphoma
and for cancer of the esophagus and cervix." The conclusions of Raaschou-Nielsen et al, (2003)
more strongly associated the increased NHL risk with TCE exposure. Raaschou-Nielsen et al.
(2003) noted "the present results and those of previous studies suggest that occupation exposure
to TCE to past higher levels may be associated with elevated risk for NHL." NHL rates increase
with increasing socioeconomic status (SES) (Raaschou-Nielsen et al., 2003); thus, the finding of
an elevated incidence of NHL in this study of lower SES or "blue-collar" workers who comprise
these two cohorts is noteworthy, given that SES is less likely to produce a false positive finding.
4 Hansen et al. (2001): males, SIR = 3.5 (95% CIil.5, 6.9), 8 cases; females, 0 cases observed, 0.3 cases expected;
Raaschou-Nielsen et al. (2003): cohort, both sexes, SIR = 1.2 (95% CI:1.0, 1.5), 96 cases; subcohort with expected
higher exposure levels, both sexesSiSIR =1.5 (95% CI: 1.2, 2.0), 65 cases.
-------
Both studies also reported statistically significant elevated relative risks for cervical cancer in
women. The excess cervical cancer risks reported in both studies may be more related to social
class given the higher incidence rates seen among females of lower versus higher social class
(Raaschou-Nielsen et al., 2003). Hansen et al. (2001) reported that also noteworthy among
males was the observation of five cases of liver and biliary passage cancer, three more cases than
expected (SIR = 2.6, 95% CI: 0.8, 6.0). Of the five cases, two were cancers of the primary liver
(1.1 expected) and three were cancers of the biliary tract (0.9 expected) (e-mail dated June 13,
2001, J. Hansen to C. Siegel Scott). No cases of liver or biliary tract cancers were observed
among females (0.4 case was expected). Raaschou-Nielsen et al. (2003) reported an excess of
gallbladder-biliary passage cancer in women (SIR = 2.8, 95% CI: 1.1,5.8) but not men, and
primary liver cancer risk in both men and women was close to that expected. Raaschou-Nielsen
et al. (2004) also reported their study indicated an association between TCE exposure and renal
cell carcinoma, which was unlikely to be explained by the lower SES status or by smoking
patterns of the cohort5.
An unexpected finding in Hansen et al. (2001) was an excess of esophageal cancer in
males that was confirmed in the subsequent study of Raaschou-Nielsen et al. (2003)6.
Examination of histological subtypes of esophageal tumors diagnosed from 1980 to 1996 from
men bom during the same period as the TCE study subjects revealed a higher proportion of the
adenocarcinoma among TCE cases (83%) compared with 30% of 2,900 esophageal tumors in the
Danish Cancer Registry reported as adenocarcinoma (Hansen et al., 2001). No major
confounders are known for adenocarcinomas of the esophagus (Hansen, 2004); cigarette
smoking and alcohol consumption are more strongly associated with squamous cell carcinoma
(Crew and Neugut, 2004).
Obtaining incidence rates in populations other than those in Nordic countries is often
difficult, and death certificates are widely used in epidemiologic research owing, in part, to the
ease of obtaining this information from national databases where all deaths are recorded. A
large number of cohort studies have assessed mortality, generally, using job title to infer
exposure. A small number of cohort mortality studies examine aircraft maintenance workers and
assign TCE exposure to subjects. These studies present cause-specific mortality risks for the
TCE subcohorts. Three mortality studies have been published since EPA's draft assessment,
5Raaschou-Nielson et al. (2003): both sexes, cohort, SIR = 1.2 (95% CI: 0.9, 1.5), 76 cases; both sexes, subcohort
with expected higher exposure levels, SIR = 1.4 (95% CI: 1.0, 1.8), 53 cases.
'Hansen et al. (2001): males, SIR = 4.2 (95% CI: 1.5, 9.2), 6 cases; females, 0 observed cases, 0.1 expected cases;
Raaschou-Nielsen et al. (2003): adenocarcinoma, both sexes, cohort, SIR = 1.8 (95% CI: 1.2,217), 23 cases; both
sexes, subcohort with expected higher exposure levels, SIR = 1.7 (95% CI: 0.9,2.9), 13 cases.
-------
those of Chang et al. (2003a, b) and Stern et al, (2001). Chang et al. (2003a) is a proportional
mortality ratio (PMR) analysis of deaths among Taiwanese electronic workers, with a fuller
cohort analysis published as Chang et al. (2003b). The PMR study of Stern et al. (2001)
examines deaths among U.S. construction workers belonging to a union affiliated with the AFL-
CIO. All three investigations lack documentation of the TCE exposure of individual study
subjects, and the studies provide no information to gauge the percentage of the cohort with
potential TCE exposure. Furthermore, the cohort of Chang et al. (2003b) is of a young mean
age, and a full latent period for cancer development has not yet elapsed.
2.2. CASE-CONTROL STUDIES
Studies of a case-control design can inform causal associations, and two studies
published since 2000 are available on TCE exposure and site-specific cancers. A strength of the
case-control design is a better efficiency compared with cohort studies for studying rarer
outcomes such as cancer (Breslow and Day, 1980). Both recent studies are of renal cell
carcinoma (RCC) and can allow a comparison with previous case-control studies and with
findings in cohort studies. Pesche et al. (2004, 2000) examined RCC and occupational TCE
exposure using two exposure assessment approaches, the JEM and the job-task-exposure matrix
(ITEM). This study of 935 RCC cases and 4,298 population controls identified from 5 regions
in eastern and western parts of Germany was adjusted for age, study center, and the number of
pack-years smoking. A strength of the study is that exposure information is from participant
interviews and not from interviews with a relative7. None of the RCC cases in the previous
reports of Henschler et al. (1995) or Vamvakas et al. (1998), cases that arose from a cluster, were
included in the case series of Pesche et al. (2000). Point estimates in males between RCC and
TCE exposure were the same regardless of which exposure assessment was employed
(substantial exposure: JEM, odds ratio (OR) = 1.3; JTEM, OR = 1.3). The ORs were not
statistically significant, and exposure-response relationships were not observed; Pesche et al.
(2000) attributed this finding to nondifferential exposure misclassification that would attenuate
the estimate of risk.
The case-control study of Briining et al. (2003) was designed to examine the hypothesis
of an association between RCC and TCE exposure. This is a study of 134 RCC cases and 401
controls from hospitals or nursing homes in Germany who were matched to cases by sex and
age. None of the kidney cancer cases in Henschler et al. (1995) or Vamvakas et al. (1998) were
7 Interviews with relatives are termed "next-of-kin interviews" and are often obtained when cases or controls are
deceased. The lack of direct and accurate information confirming whether subjects had exposure to TCE and at what
levels, as may occur in interviews with relatives or other next-of-kin, can potentially introduce misclassification bias.
-------
included in this report. Information on occupational TCE exposure was self assessed, obtained
from face-to-face interviews with all cases and controls, except for 21 case subjects for whom
information was obtained from next-of-kin interviews or inferred from information on job
history using either expert-based JEM methods or an occupational exposure database. Brtining
et al. (2003) reported "the logistic regression results, adjusted for age, gender, and smoking,
confirmed a TCE-related renal cell carcinoma risk in this region." Using the occupational
exposure database for a comparison of industries with and without TCE exposure, a significant
excess risk was estimated for the longest held job in TCE-exposing industries (OR = 1.8,95%
CI: 1.01,3.2). Any exposure in "metal degreasing" was an RCC risk factor (OR = 5.6, 95% CI:
2.3, 13.3). Self-reported narcotic symptoms, which Briining et al. (2003) attribute to peak
exposures, were associated with an excess risk (OR ~ 3.7, 95% CI: 1.8, 7.5). Both studies
further suggest associations between RCC and TCE exposure but importantly, as Pesche et al.
(2004) discuss, each study by itself does not provide clear evidence of an association.
The identification of altered excretion of kidney enzymes in TCE-exposed individuals
with kidney cancer (RCC) and histopathological characteristics and gene mutations in these
tumors can provide biological support or plausibility for observations in case-control studies, A
high prevalence of multiple mutations of the Von Hippel-Lindau (VHL) gene has been
previously reported in a series of renal cell tumors of TCE-exposed subjects (Briining et al.,
1997; Brauch et al., 1999), with Brauch et al. (1999) reporting a C > T transition at nucleotide
454, a mutation not found in peripheral lymphocytes or normal renal tissue from these
individuals. A transition of nucleotide 454 has not been observed in other large mutational
surveys of RCCs (Brauch et al., 1999), but a few observations of this mutation have been
documented in VHL families (Weirich et al., 2002; Hes et al., 2000). The report of Brauch et al.
(2004) extends their work with mutational analysis of 21 of the 39 non-TCE-exposed RCCs in
Vamvakas et al. (1998), comparing findings in this group to mutational spectra analysis of 17 of
the 19 original cases in Vamvakas et al. (1998) identified with TCE exposure reported in their
1999 publication. These reports are suggestive of genetic alterations as associated with RCC
risk, and TCE-exposed cases show a higher percentage of alterations, some of which may
indicate very specific changes to the VHL gene. The reports of Bolt et al. (2004) and Green et
al. (2004) of a mixed pattern of increased concentrations of tubular maker proteins in the urine of
TCE-exposed compared to non-exposed subjects suggest altered integrity of renal tubular cells,
but such changes may not necessarily predict RCC.
2.3. COMMUNITY STUDIES
Community epidemiologic studies can augment observations in analytical cohort or case-
control studies. Community studies on TCE.exposure are of residents with exposure determined
-------
by place of residence of water supply; these studies can include individuals of varying ages and
susceptibilities, as compared to an occupational population that is of a more restricted age range
- and health status. Community studies may be helpful for generating hypotheses about life-stage,
genetic, and environmental contributions to disease outcomes such as cancer. These aspects may
be more difficult to assess in occupational cohort studies affected by the well-recognized healthy
work bias (Pearce et al., 1986). Community studies on TCE are of particular interest for two
reasons. First, they are of the oral exposure route, in contrast to the inhalation exposure route of
occupational cohort studies, which is important because the pharmacokinetics of oral and
inhalation TCE exposures differ with respect to delivered dose. Second, community studies
have relatively high statistical power even though exposure levels are relatively low. The
community studies are of multiple solvent exposures, assessed at the ecological (community)
level rather than the individual level. Contemporaneous or retrospective assessment of disease
relative to exposure and aggregation bias compromise their interpretability (Wartenberg et al.,
2000), as does their limited or, in some cases, lack of adjustment for potential confounders.
Many of the community studies on TCE have assessed relationships between drinking
water exposures and leukemia. Age at first exposure may be an important risk factor for
subsequent development of leukemia association with drinking water exposure to chlorinated
solvents that included TCE. A recent study of Costas et al. (2002) investigated factors
potentially responsible for the leukemia cluster in Woburn, MA. This analytical study was a
follow up to the studies of the Massachusetts Department of Public Health (1997) and Lakagos
et al. (1986). Costas et al. (2002) reported that their study suggested an association with
developing childhood leukemia among children whose mothers were likely to have consumed
water from wells contaminated with TCE at concentrations an order of magnitude higher than
perchloroethylene during pregnancy than for those who did not; OR^8 = 8.3 (95% Cl: 0.8,
94.7). A statistically significant exposure-response relationship was identified for the period
during pregnancy9. In contrast, the child's potential for exposure from birth to diagnosis showed
no association with leukemia risk. Morgan and Cassady (2002) investigated a 10-year cancer
incidence, including leukemia in children younger than 15 years of age, in 13 contiguous census
tracts in the city of Redlands, California, with TCE and perchlorate and did not report
associations with leukemia or other site-specific cancers. This leukemia finding is similar to the
8 Odds ratio (OR) from proportional hazards model and adj usted to control for SES, maternal smoking during
pregnancy, maternal age at birth of child, and breast feeding using a composite covariate.
" Never exposed category, OR = 1.0 (9 cases); least exposure category, OR = 3.5 (95% CI: 0.2, 58.1) (3 cases); most
exposed category, OR = 14.3 (95% CI: 0.9,224.5) (7 cases); chi-square test for trend, p < 0.05. OR adjusted to
control for SES, maternal smoking during pregnancy, maternal age at birth of child, and breast feeding using a
composite covariate.
-------
finding of Costas et al. (2002) for the exposure period from birth to diagnosis. Unlike Costas et
al. (2002), however, Morgan and Cassady (2002) did not investigate critical time periods of
susceptibility..
The study of Lee et al. (2003) is one of the few studies available evaluating cancer other
than leukemia. Lee et al. (2003) assessed liver, stomach, colorectal, and lung cancers in
residents in two villages. This study is suggestive of an association between exposure to
chlorinated hydrocarbons and liver cancer in males. The exposed village was located
downstream from an electronic factory and received water contaminated with chlorinated
solvents. TCE concentrations in well water of the downstream community were an order of
magnitude higher than perchloroethylene and 1,1-dichloroethylene concentrations. Monitoring
of well water from the control village located upstream from the factory did not shown
contamination by chlorinated solvents. Lee et al. (2003) reported increased mortality odds ratios
(MORs) among males for all cancer, and liver cancer for the periods after 10 years of latency,
namely, 1980-89, and 1990-97. The adjusted MOR for liver cancer in males was 2.6 (95% CI:
1.2, 5.5) with a significant linear trend for the period effect. No associations were observed
between other cancer sites and residency in the village containing solvent-contaminated well
water. The MOR was calculated with cardiovascular-cerebrovascular diseases as the reference
diseases, and multiple logistic regression analyses were performed to estimate the effects of
exposure and calendar period after adjustment for age. Ecologic studies such as these lack
individual information on groundwater exposure and, for this reason, provide only limited
inferences about cause-effect associations. This study also did not address potential confounding
related to hepatitis viral infection status, a risk factor for liver cancer, or potential
misclassification due to the inclusion of secondary liver cancer among the case series.
3. ISSUES RELATED TO TCE EPIDEMIOLOGIC EVIDENCE
Epidemiologic studies clearly provide the most direct information for identification of
human hazards (NRC, 1994) and for quantitative dose-response analyses. All studies of
acceptable quality, whether yielding positive or null results, or even suggesting protective
carcinogenic effects, should be considered in assessing the totality of the human evidence.
Analysis and interpretation of the epidemio logic body of evidence, ideally studies of different
populations and investigative methods, is carried out with attention to patterns and trends in
response and is a critical step in identifying a hazard. An important aspect of the analysis is an
evaluation of risk as reflecting a false positive result or due to chance, confounding, or possibly
bias in a direction opposite to the null. Equally important in an assessment of risk are false
-------
negative findings attributable to a number of biases, including misclassification and selection
biases, the healthy worker effect, or study design limitations. Other anomalous findings can
result from insufficient power, small numbers of site-specific cases or deaths, insufficient latent
period between exposure and measurement of cancer, insufficient followup in cohort studies of
the exposed population, or low exposure prevalence in case-control studies.
The epidemiologic conclusions in EPA's 2001 draft assessment (U.S. EPA, 2001) drew
from the state-of-the-science paper of Wartenberg et al. (2000), which presents a comprehensive
review of more than 80 published papers and letters to editors on the cancer epidemiology of
people exposed to TCE. Wartenberg et al. (2000) found cohort, case-control, and community
studies as supporting an overall summary of the epidemiologic evidence as consistent with
previous conclusions of some evidence of liver cancer and NHL (1ARC, 1995) but more strongly
suggested associations with TCE exposure. The authors also found moderate support in terms of
aspects of causation as discussed by Sir Bradford Hill (Hill, 1965). The evidence showed a
temporal relationship of disease preceding exposure as supported by evidence from cohort
studies, which received greater weight in the critical analysis. Relative risks for several cause-
specific cancers were between 1.7 and 1.9 and were of moderate strength; elevated cause-
specific risks were observed in many studies, suggesting some consistency across studies; the
findings of associations with multiple cause-specific cancers implied that TCE was not specific
but, rather, was a multisite carcinogen10; examinations of biological gradients were limited by
data; the coherence or the biology of cancer was not believed to conflict with a TCE etiology;
and experimental evidence existed in the animal bioassay literature.
A number of perspectives maybe found on the treatment of the epidemiologic studies in
the EPA draft assessment, in general, and in Wartenberg's analysis, specifically, and include: (1)
EPA's Science Advisory Board (SAB) report (U.S. EPA, 2002a) summarizing their review of
the 2001 draft assessment; (2) public comments submitted to EPA on their 2001 draft
assessment; and (3) published literature as letters to editors. Moreover, a number of
epidemiologic studies have been reported in the open literature subsequent to Wartenberg's
comprehensive review and provide additional information, particularly for kidney cancer and
NHL.
The SAB comments on the epidemiologic analysis were contained in their responses to
Charge Question 2, on whether EPA's conclusion in the cancer weight-of-evidence
characterization of TCE as "likely" carcinogenic to humans was adequately supported.
Suggested areas of inquiry for the SAB panel discussion included the characterization of the
strength of the epidemiologic evidence and the Wartenberg analysis, particularly its inclusion of
'" Traditionally, specificity has been defined in term of one cause, one disease (Hill, 1965).
10
-------
the Henschler et al, (1995) study. The panel believed that EPA's overall characterization of TCE
was reasonable, based in part, on epidemiologic evidence in humans showing associations
between TCE exposure and several cancers, including several at the same sites seen in animal
bioassays (U.S. EPA, 2002a). The SAB also noted "among the epidemiological studies, the data
appeared strongest overall for liver cancer and, to some degree, for lymphoma." Moreover, the
SAB comments endorsed the division of the cohort studies into three tiers and recommended that
EPA explicitly weight more strongly the studies with more precise estimates of personal
exposure, including case-control studies that specifically focus on TCE, than the other studies
that involved exposure to a variety of chemicals.
The SAB report advised EPA to "identity more clearly and then explicitly apply criteria
for the selection of the epidemiologic studies, that EPA should select the broadest possible array
of studies for each endpoint meeting those criteria, taking into consideration study design,
availability of exposure estimates, and the goal of protecting health." The SAB panel's specific
suggestions for addressing the conflicting human epidemiologic evidence for kidney tumors
focusing primarily on the study by Henschler et al. (U.S. EPA, 2002a) are discussed in more
detail below. The SAB report also included opinions of individual panel members, although this
advice did not represent a majority or unanimous opinion among panel members on including
discussions of prostate cancer and childhood cancer "as there is limited epidemiologic evidence
to support both of these endpoints." Several members also gave advice on the Wartenberg
analysis by offering either specific comments or opinions on performing a formal meta-analysis
of the TCE cancer studies.
Drawing conclusions from epidemiologic data on exposure and disease relationships
involves complex judgments about the body of evidence. Several methodological issues need
consideration in an integrated analysis of the epidemiologic body of evidence; these issues and
perspectives are discussed more fully below. These scientific issues expand upon the general
SAB comments discussed above and incorporate discussion of both relevant scientific literature
and comments submitted by the public or published in the open literature. NAS input on how
EPA can address these issues in integrating the body of epidemiologic evidence would help to
strengthen EPA's revised assessment.
3.1. NARRATIVE AND QUANTITATIVE METHODS
The practice of causal inference in environmental epidemiology relies upon three
approaches: systematic narrative reviews, criteria-based inference methods, and, increasingly,
meta-analysis (Weed, 2002). All three have been employed in various analyses of the
epidemiologic literature on cancer and TCE exposure. Interpreting observations in
epidemiologic evidence can involve both the quantitative method of meta-analysis and the
ll
-------
qualitative criteria-based method of causal inference (Weed, 2000). The epidemiologic
conclusions in EPA's 2001 draft assessment (U.S. EPA, 2001) drew from the state-of-the-
science paper of Wartenberg et al. (2000). Wartenberg et aK (2000) present a comprehensive
review of the epidemiologic studies evaluating possible associations between TCE and cancer
based on studies available at that time, adopting statistical methods for a critical review of these
studies. The authors considered cohort studies to be the most reliable design of the studies
reviewed and categorized all cohort studies into the following three tiers according to the
specificity of the exposure information:
• Tier I-High-quality studies that included direct assessment of exposure using
U-TCA, a biological marker of TCE exposure, or exposure reconstruction using
JEMs and job histories.
• Tier II-Cohort studies that used indirect methods of exposure assessment such as
job titles and other general information to infer potential exposure to TCE; this
tier also included a small number of PMR studies.
• Tier Ill-Studies of dry-cleaner and laundry workers; this tier also included a
small number of PMR studies.
Studies were further grouped by incidence or mortality within each tier. Wartenberg et
al. (2000) summarized the studies separately for each tier, providing an estimate of the average
or weighted relative risk across studies in a tier using a meta-analysis approach. The evaluation
of case-control studies and community studies consisted of a descriptive summary, a critical
review, and a presentation of risks for each study. Thus one of the major issues for synthesizing
epidemiologic evidence is:
What qualitative and/or quantitative approaches can best inform causal
inferences of the TCE body of epidemiologic evidence?
The narrative review approach is a common method used in epidemiologic and other
disciplines. The purpose of the narrative review is to summarize the available evidence and to
make causal association conclusions about exposure-disease relationships (Weed, 2002). Many
narrative reviews do not assess quality and weight of associations reported within a published
paper or for the literature as a whole. Causal claims can arise from a "tally" or count of positive
and nonpositive findings. Frequently, studies are mistakenly cited as "negative" that are
12
-------
nonpositive or "null" and, in this case, the narrative reflects a strength-of-the-evidence rather
than a weight-of-the-evidence determination, which includes an evaluation of study attributes
and limitations. Furthermore, a recent analysis of narrative reviews in seven widely read
epidemiologic journals revealed that a large proportion of these narrative reviews were of
questionable quality, lacking a stated purpose, clear literature search criteria, inclusion and
exclusion criteria for the studies summarized in the review, and clear descriptions of the causal
criteria used to interpret epidemiologic evidence (Breslow et al., 1998).
Causal claims using criteria-based approaches are made according to a set of criteria or
standards applied to the evidence summarized within the systematic narrative review (Weed,
2002). Many, if not most, of the elements first proposed by Sir Bradford Hill (Hill, 1965) form
the backbone of the criteria. General guidance on interpreting epidemiologic evidence widely
cite these criteria (WHO, 2000; U.S. EPA, 1999; Federal Focus, 1996; Pastides et al., 1991).
None of the criteria are conclusive in isolation, and the only criterion that is essential is the
temporal relationship. Although a single study may be indicative of a cause-effect relationship,
confidence in inferring a causal relationship is increased when several independent studies are
concordant in showing the association, when the association is strong, and when other criteria
for causality are also met (U.S. EPA, 1999). Some criteria-based approaches may also include
quantitative analyses, but this is not a general rule.
Meta-analysis is increasingly adopted as a formal statistical method for reviewing and
summing a body of evidence. It is especially useful when results from several studies disagree
with regard to direction of effect, or when sample sizes are individually too small to detect an
effect (Stangl and Berry, 2000). Common meta-analytic methods can include fitting of meta-
regression models, such as a fixed-effects or random effects model; linear regression analysis to
assess dose-response; and pooled analyses (Crump et al., 2003; Greenland, 1994; NRC, 1988).
Compared with a traditional narrative review, a review incorporating a meta-analysis can be less
subjective (Wong and Raabe, 1996). Additionally, benefits of conducting a meta-analysis in a
review include enhancing statistical power, an important consideration when original studies are
limited by small numbers of site-specific events; more precise risk estimates; fewer statistical
comparisons with the benefit of reducing issues associated with multiple comparisons; and the
ability to examine heterogeneity across studies (Wong and Raabe, 1996; Morris, 1994). Use of
meta-analysis methods is intended to introduce consistency and comprehensiveness into what
otherwise might be a more subjective review of the literature. 'It can contribute to hazard
characterization when viewed as a quantitative review of the literature, a "study of studies"
(WHO, 2000). The value of such an analysis is dependent upon a systematic review of the
literature that uses transparent criteria of inclusion and exclusion. In interpreting such analyses,
it is important to consider a number of factors, including the effects of differences in study
13
-------
quality and publication bias. Blair et al. (1995) discuss the use of meta-analysis in
environmental health studies and provide a decision framework for considering whether or not to
carry out an analysis.
A number of causal association claims have been provided in reviews of the TCE cancer
epidemiologic literature (IARC, 1995; Weiss, 1996; McLaughlin and Blot, 1997; Lynge et al.,
1997; McLaughlin and Lipworth, 2000; Briining and Bolt, 2000; Wartenberg et al., 2000; Lavin
et al., 2000; Mandel and Kelsh, 2001; IOM, 2003; NTP, 2002; and Wong, 2004). Table 2
presents conclusions drawn in these reviews. Several reviews focus on one site, such as the
kidney (McLaughlin and Blot, 1997; Mandel and Kelsh, 2001; Bruning and Bolt, 2000). Lavin
et al. (2000) focus on a few sites, such as liver, lung, kidney, and all-site cancers. The IARC
(1995), Lavin et al. (2000), IOM (2003), and NTP (2002) reviews of the TCE literature adopt a
criteria-based approach. NTP (2002) also drew upon the criteria-based review of Wartenberg et
al. (2000) to support the overall evaluation of TCE exposure and human cancer.
Application of meta-analysis methods to TCE cancer epidemiologic studies appears
feasible because a number of studies present summary relative risks using these methods (Table
3). These studies suggest that, with one exception, relative risk measures are homogeneous; that
is, heterogeneity or systematic variation in the relative risk measure was not present for an
individual cancer site across studies. The exception was for kidney cancer and the inclusion of
Henschler et al. (1995), who observed a large relative risk compared with observed relative risks
in other cohort studies. Many of the epidemiologic studies in Table 3 examined heterogeneity as
part of sensitivity analyses or analyses that examine whether the same or different results are
achieved using different statistical methods (Delgado-Rodriquez, 2001), but not all studies
present full details of the analytical methods or criteria for study inclusion. OjajSrvi et al. (2001)
present a meta-analysis of pancreatic cancer and occupational exposure to chlorinated
hydrocarbon solvents and contained an analysis of the TCE cohort and case-control studies. As
part of public comment on the U.S. EPA external review draft assessment, Kelsh et al. (2002)
employ meta-analysis methods in a sensitivity analysis to examine studies reporting kidney
cancer. In a presentation at the Toxicology Forum meetings, Kelsh (2003) expands this analysis
to include the liver, liver and biliary passages, and NHL. Both Raaschou-Nielsen et al. (2003)
and Morgan et al. (1998) present site-specific summary relative risks using another meta-
analysis approach of summing observed and expected numbers of site-specific cancers. Axelson
(2004) expands the analysis of Morgen et al. (1998), adding information from the studies of
Blair et al.j(1998), Boice et al. (1999), and Hansen et al. (2001) that were published following
the analysis of Morgen et al. (1998). Furthermore, a comparison of the kidney cancer, liver
cancer, and NHL findings in Table 3, the only sites found in common across analyses, suggests
several factors as influencing the magnitude and statistical significance of the joint or meta-
14
-------
relative risk. These factors include choices regarding study selection, cancer site (site-specific
cause such as primary liver cancer or a broader category), and endpoint studied (incidence or
both incidence and mortality). It is also important to note that some consistency is apparent
across studies in the magnitude of the relative risk, suggesting that uncertainties associated with
the choice of statistical modeling methods are likely to be small.
3.2. SOUNDNESS OF DATABASE
The body of epidemiologic evidence on TCE exposure is large compared with the body
of epidemiologic evidence for other chemicals assessed by EPA. Furthermore, most of the
epidemiologic studies on TCE exposure have become available within the past 10 years, and
several new studies have been published since EPA's draft assessment. These newer studies
examine possible relationships between exposure to TCE and occurrence of liver and kidney
cancer, leukemia, and NHL, sites that EPA's draft assessment had associated with TCE
exposure. NAS input on the strengths and limitations of the epidemiologic body of data would
help to strengthen EPA's revised assessment. Specifically,
Which studies carry a greater weight in the hazard evaluation?
Consideration of a number of issues this issue paper has identified previously is
important for answering this question. However, EPA asks the NAS panel to focus its attention
on the two issues more fully discussed in Sections 3.2.1 and 3.2.2.
3.2.1. Studies of Morbidity or Mortality
Both incidence and mortality rates are measures of risk, and members of EPA staff have
used both measures in identifying hazards and in performing dose-response analyses for
estimating risk associated with a unit of exposure. A large body of studies exists that has
described mortality in occupational cohorts; fewer studies of cancer incidence in TCE-exposed
workers exist Wartenberg et al. (2000) present joint relative risk estimates for incidence studies
and mortality studies separately in an attempt to reduce heterogeneity. For both qualitative and
quantitative approaches to synthesizing the epidemiologic evidence for making causal
inferences, an important issue is therefore:
For which site-specific cancers can studies of incidence carry greater weight
for hazard identification than studies based on death certificates?
15
-------
Incidence rates give an accurate indication of the risk of a disease in a population. In the
absence of incidence data, epidemiologic studies rely on mortality data to assess exposure-
disease associations. An understanding of the accuracy of death certificate information as a
surrogate for incidence data is important for evaluating observations in the mortality studies.
Death certificate inaccuracies would obscure exposure-disease associations toward the null and
may explain apparent inconsistencies in observed findings between epidemiologic studies using
incidence and those based on death certifications. In diseases for which survival is poor and
mortality is 95% or 100% within a short time period (e.g., lung cancer, pancreatic cancer), the
mortality rate as inferred from death certificates may be a good surrogate for incidence rate.
Conversely, it is also possible that site-specific cancers with a high survival rate may not be
accurately recorded on the death certificate as an underlying cause of death. Death
certifications, in this case, may be a poorer surrogate for recording incidence for the following
reasons. First, incidence can be under reported or miscoded on the death certification, i.e.,
inconsistencies are seen between incidence and cause of death. Several reasons may explain
inconsistencies between incidence data inferred from histological and pathological reports and
the causes of death on the death certificate. Factors such as higher 5-year survival rates and
changes over time in categories by which site-specific cancers are identified in the International
Classification of Diseases (ICD) have been shown as important to the accuracy of death
certificate coding (Percy et al., 1990,1981). •
Second, even if a site-specific cancer may be accurately recorded as an underlying cause
of death, mortality studies may underestimate the risk of the disease due to better survival.
Several site-specific cancers of interest evaluated in the TCE epidemiologic studies have high
survival rates; other sites of interest have much lower survival rates. For the U.S. population, the
following 5-year survival rates were identified using incidence recorded in the Surveillance,
Epidemiology, and End Results (SEER) Program (Ries et al., 2004): invasive cervix uteri
cancer, 93%; kidney and renal pelvis cancer, 64%; NHL, 59%; invasive liver and intrahepatic
bile duct cancer, 8%; and pancreatic cancer, 4%.
The mortality of liver cancer can be either underestimated or overestimated depending on
which disease classification categories are used. The extent of miscoding on death certificates
varies by site. For this reason, greater uncertainties resulting from misclassification bias
accompany cohort mortality studies that assess associations between exposure and these
endpoints. Percy et al. (1990) show that cause-specific deaths with lower survival rates such as
liver cancercan also be extensively miscoded on death certificates. Percy et al. (1990), using
data from the SEER Program, compared primary liver cancer in 2,388 cases to that recorded on
the death certificate. Their study showed that only 53% of the deaths were attributed on death
certificates to primary liver cancer. Furthermore, only 24% of the 156 incident cases with
16
-------
intrahepatic bile duct liver cancer had died with a recorded underlying cause of intrahepatic bile
duct cancer. This study also showed inaccuracies in death certificates as a marker of disease
diagnosis. Percy et al. (1990), furthermore, showed that of 2,977 death certificates with a
underlying cause of death of primary liver cancer, 83% were of patients with a hospital diagnosis
of liver cancer; 17% of the deaths had a different primary cancer site at diagnosis. For the 537
deaths from intrahepatic bile duct cancer on the death certification, only 18% of the patients had
hospital diagnoses of cancer at this site; 29% were diagnosed as having primary liver cancer, and
28% were found to have extrahepatic bile duct cancers when originally diagnosed (Percy et al,
1990).
The extent of miscoding kidney cancer and NHL on death certificates is less than that for
liver cancer and for gallbladder-biliary cancer. Percy et al. (1981) reported detection rates11 of
greater than 80% for both kidney cancer and NHL. Furthermore, Selikoff and Seidman (1992),
in their study of lung and kidney cancers, mesothelioma, and asbestosis among asbestos-worker
deaths, showed death certificates with kidney cancer as a cause of death as highly accurate of a
clinical or histopathologic diagnosis of kidney cancer.
3.2.2. Exposure Assessment Issues in Epidemiologic Studies
Adequate characterization of exposure is a characteristic that is important in
epidemiologic studies, with greater weight generally given to studies with more precise and
specific exposure estimates. Careful evaluation of a study's exposure assessment method is
important in the evaluation of a body of epidemiologic data, particularly if divergent
observations may be due to exposure misclassification bias. The methods by which exposure is
assessed in epidemiologic studies of TCE are diverse but include a number of studies in which
biomonitoring data are utilized. As discussed above, the draft EPA assessment placed greatest
weight on cohort studies with the more precise estimate of TCE exposure to individual cohort
subjects. One of the important issues for both qualitative and quantitative syntheses of the body
of epidemiologic evidence for TCE is thus:
What advice can NAS provide EPA on weighting the different exposure
assessment approaches adopted in the TCE epidemiologic studies?
In many cases, actual exposure measurements are lacking and surrogates such as
available current or historical monitoring data are often used to reconstruct exposure parameters.
Use of surrogates carries a potential for misclassification, i.e., an individual may be placed in an
1' Detection rate is the proportion of hospital diagnoses with cancer of a certain site in which the cause of death
reflects the same as the hospital diagnosis.
17
-------
incorrect exposure group. All exposure estimation methods, whether used by subjects or
experts, can have low validity and reliability and therefore can lead to misclassification bias. If
the misclassification is nondifferential, the resulting risk measure could be reduced to the null.
Exposure estimation methods need to be carefully designed using evidence about techniques that
improve performance and, where possible, tested (Teschke et al., 2002). Sensitivity analysis of
exposure uncertainties on effect measures can also provide information (HEI, 1999).
Inaccurate exposure and dose estimates lead to exposure misclassification, a
nondifferential bias that can mask possible causal association relationships toward the null and
dampen exposure-response gradients (Stayner et al. 1999; Gomez et al., 1994). Biological
markers potentially offer excellent measures of exposure (Hulka and Margolin, 1992).
Biological markers for dose can take into account individual pharmacokinetic characteristics that
can modify the absorption, metabolism, and delivery of the exposure to target tissues. Biological
markers can, therefore, improve estimates of dose. The three Nordic cohorts of Axelson et al.
(1994), Anttila et al. (1995), and Hansen et al. (2001) identify study subjects using the TCE
biological marker of U-TCA. On average, 2.2 (Hansen et al., 2001) and 2.5 (Anttila et al., 1995)
samples were available per subject. In most cases, biological monitoring was carried out for one
solvent, TCE. In these studies, an estimate of lifetime exposure to individual study subjects is
not possible in the absence of duration of exposure information. Although this is not an issue for
identifying a hazard, the lack of information on TCE exposure received over the full duration of
employment limits inferences about dose-response relationships from these cohorts. For this
reason, it is not surprising that Hansen et al. (2001) did not observe strong concentration-
response relationships. Additionally, analyses of these studies using EPA's cancer risk
estimation procedures will produce overestimates of lifetime excess cancer risks if TCE
exposure concentrations had decreased since the period of biological monitoring.
Other cohort studies adopt a number of approaches for exposure assessment Several
studies assign TCE exposure to study subjects using surrogate information on patterns of TCE
use by job title obtained from historical job descriptions, from historical industrial hygiene
surveys, or from personal interviews to develop job exposure matrices. Raaschou-Nielsen et al.
(2003) lack TCE exposure attribution to individual study subjects but use such surrogate
information to increase the likelihood of correctly identifying subjects as having TCE exposure.
For several cohorts, industrial hygiene measurements were either absent before the 1970s (Boice
et al., 1999; Morgan et al., 1998) or were quite limited (Blair et al., 1998; Stewart et al., 1991).
Furthermore, both Ritz (1999) and Greenland (1994) classify study subjects as TCE exposed
using information obtained from personal interviews in the absence of historical monitoring.
Many more cohort studies identify TCE as one of a number of potential exposures but do
not identify individual subjects with TCE exposure. The main shortcoming of analyses by
18
-------
occupation and industry is that they do not identify specific agents as risk factors,(Teschke et ah,
2002). For example, Garabrant et al. (1988) identify TCE exposure in 37% of the jobs held by
70 of 14,067 study subjects employed at an aircraft manufacturing plant. While offering limited
information about this cohort's exposure to TCE, this analysis does not provide information on
exposures of the roughly 14,000 remaining study subjects to TCE. Another shortcoming of this
type of study is that the lack of an association with a particular job or industry may mask the
effect of exposure to a specific chemical to which only some individuals in the job are exposed
(Teschke et al., 2002). In this case, relative risks are close to the null value. EPA allots greater
weight to studies with more precise and specific exposure estimates for these reasons.
As reviewed in Teschke et al. (2002), case-control studies adopt a number of approaches
to occupational exposure assessment. Some case-control studies adopting JEMs or JTEMs to
identify jobs to assign potential TCE exposure lack industrial hygiene monitoring data for the
industries from which cases and control arose. Attribution of TCE exposure in these studies is
based on self-reported information or job lifetime history codes according to standard
occupational classification for either recent, usual occupation or industry. Several studies assign
TCE exposure to cases and controls using a JEM that incorporates concepts such as probability
and intensity of exposure, and assignment of exposure by time to account for changing patterns
of chemical use. These features are intended to increase the accuracy of exposure assessment
and were first included in the JEM of Gomez et al. (1994) and then adopted in the studies of
Dosemeci et al. (1999) and Heineman et al. (1994). Generic JEMs have also been adopted in the
RCC case-control studies of Bruning et al. (2003) and Pesche et al. (2000). An issue associated
with the use of generic JEMs is the sensitivity, or ability to identify study subjects as exposed,
and specificity, or ability to identify study subjects as not exposed. The JEMs used by Bruning
et al. (2003), Dosemeci et al. (1999), and Pesche et al. (2000) may be sensitive for solvent
exposure. Dewer et al. (1991), in their study of Canadian jobs and specific chemical exposures
comparing information from expert interviews and generic JEMs, show that a generic JEM has a
high sensitivity (84%) and specificity (97%) for solvent exposure; TCE exposure was not
uniquely examined. Both generic and study-specific JEMs are able to show associations where
expected, although only the study-specific expert assessment produced clear exposure-response
trends (Teschke et al., 2002). This may be one reason for the observed absence of an exposure-
response relationship reported by Pesche et al, (2000).
A number of solvents have been documented in TCE cohort studies (Marano et al., 2000;
Morgan et al., 2000, 1998; Boice et al., 1999; Blair et al., .1998; Stewart et al., 1991) and TCE-
exposed subjects likely have potential exposures to other chlorinated solvents, including
perchloroethylene and 1,1,1-trichloroethane. Potential exposure to multiple chlorinated solvents
other than TCE is an important consideration in the TCE epidemiologic studies for two reasons.
19
-------
First, these chemicals can share a metabolic profile similar to that of TCE, and, second, some
epidemiologic studies also report associations between exposure to these other solvents and
cancer. Two other EPA staff papers discuss interactions of TCE, its metabolites, and other
chemical exposures: Issues in Trichloroethylene Pharmacokinetics and Interactions of TCE, Its
Metabolites, and Other Chemical Exposures. Little information is presented for individual
subjects on the concentration of exposure, the duration of exposure, and the time period of
exposure to these other solvents. Exposure to solvents other than TCE should not be thought of
as confounding the association between TCE and cancer, as might be considered in traditional
epidemiology. The metabolic profile in humans for several of the solvent exposures is similar, at
least qualitatively but not quantitatively, to that of TCE. hi this case, the metabolite represents
an integrated measure, i.e., cumulative exposure, and an evaluation of observed associations in
the epidemiologic studies also should take into consideration mode-of-action hypotheses.
3.3. NON-HODGKIN'S LYMPHOMA
NHL is an endpoint of interest in the TCE epidemiologic studies. Conclusions drawn by
1ARC from the epidemiologic evidence in 1995 noted associations with NHL in the TCE cohort
studies, as did EPA's draft assessment (U.S. EPA, 2001). More recently, NTP (2002) and
Wartenberg and Scott (2002) discussed the evidence supporting associations between TCE
exposure and NHL. Furthermore, both Hansen et al. (2001) and Raaschou-Nielsen et al. (2003),
two studies published since EPA's draft assessment, observed statistically significant
associations between NHL and TCE exposure.
Malignant lymphomas include a diverse group of diseases, such as NHL, Hodgkin's
disease, multiple myeloma, and leukemia. Lymphomas are characterized not only on
histological features but also on cell surface markers, and cytogenetic and clinical features (Seto,
2004). Moreover, NHL, like the larger lymphoma category, is a disease composed of numerous,
etiologically distinct neoplasms (Herrinton, 1998). NHL incidence has been increasing and now
represents the sixth leading cause of cancer deaths in males in the United States (Fisher, 2003).
In contrast, deaths due to liver and intrahepatic bile duct cancer and to kidney cancer in U.S.
males are ranked lower than NHL as the eighth and tenth leading causes of death, respectively
(ACS, 2004).
Several issues arise that may affect interpretation of NHL associations in the TCE
epidemiologic studies and that also may be important to examinations of consistency, or lack
thereof, across studies. NAS input on how EPA can weigh NHL observations in the
epidemiologic studies would help to strengthen EPA's revised assessment.
20
-------
What advice can NAS provide EPA on factors that may influence
interpretation of the epidemiologic observations on NHL?
First, most studies did not use the most current classification scheme to code incident
cases or deaths. Epidemiologists code cause-specific incident cases or mortality diagnoses
according to the ICD. The ICD has been revised several times during the followup period in
cohort studies. Furthermore, the classification system used by pathologists for lymphomas has
undergone a number of significant revisions within the past 10 years, and revisions to the ICD
reflect the changes in the pathology classification system. Many concepts of contemporary
knowledge of lymphomas have been incorporated into the International Lymphoma Group's
Revised European-American Lymphoma (REAL) classification that was proposed in 1994 and
adapted into the International Classification of Diseases (ICDrO) (Herrinton, 1998). More
recently, the World Health Organization (WHO) has proposed a classification scheme that may
be considered an updated version of the REAL classification (Cogliatti and Schmid, 2002).
Before the REAL and WHO classifications, the National Cancer Institute Working Formulation
was designed to provide a common basis for interpreting older classifications, such as the
Rappaport, Luke-Collins, Kiel, and WHO lymphoma systems (Fisher, 2003).
The TCE epidemiologic studies evaluating NHL have used a number of different ICD
classifications. All four Nordic studies classified NHL according to the 7th revisipn of the ICD
(ICD-7) and all reported consistent findings. Other revisions of the ICD were used in the more
recent studies of Blair et al. (1998) [ICDA-8], Boice et al. (1999) [ICD-9], Garabrant et al.
(1988) [ICD in effect at date of death; ICD-7, ICDA-8, or ICD-9], Morgan et al. (2000,1998)
[ICD in effect at date of death; ICD-7, ICDA-8, or ICD-9], and Ritz (1999) [ICD-9]. Few case-
control studies on lymphoma are available. NHL cases in Hardell et al. (1994) were
histologically verified and were classified using the Rappaport system. Persson et al. (1989)
does not identify the system used to classify NHL cases in their study.
Epidemiologic studies are only now beginning to examine associations between exposure
and cell-specific malignancies such as B-lymphocyte lymphoma. An example is the recent study
of Bukowski et al. (2003). None of the TCE studies have examined associations by cell type.
Lymphomas are many distinct, definable diseases that are grouped together (Fisher, 2003)
defined broadly as B-cell and T-cell lymphomas with further divisions into precursor neoplasms
and mature neoplasms (Cogliatti and Schmid, 2002). This implies that although lymphomas
have been classified in the past into distinct categories, they can share common biological
properties and differentiation pathways. For example, a B-cell lymphoma in origin may be
classified under older schemes into several distinct categories, e.g., NHL, multiple myeloma, or
leukemia. Epidemiologic studies would appear to lack consistency and specificity, that is,
exposure is not associated with a unique endpoint, if associations were shown with several
21
-------
cause-specific cancers. Thus, individual TCE study observations of excess risks of NHL and
multiple myeloma may not be inconsistent with a mode-of-action argument, although the EPA's
draft TCE assessment (U.S. EPA, 2001) noted that mode-of-action information for tbxicity in the
lymphopoietic system was more limited than that for the liver, kidney, and lung.
3.4. CONSIDERATIONS FOR META-ANALYSIS
In addition to the issues discussed above that are applicable to both qualitative and
quantitative syntheses of epidemiologic data, decisions on whether and how to undertake a meta-
analysis entails the consideration of a number of additional issues. Blair et al. (1995) discuss a
number of desirable and undesirable attributes of meta-analysis. Specifically, EPA would
benefit from input from the NAS as to:
How appropriate and useful is applying meta-analysis methods to synthesize
epidemiologic data on TCE, and what strategies could be employed to
implement such an analysis?
Public comments submitted to U.S. EPA after the release of its draft TCE assessment
emphasize that application of meta-analyses is informative when studies are homogeneous
(HSIA, 2002; DOD, 2002). An assumption is made in meta-analysis that summary estimates of
effects are homogeneous or equal across individual study findings (Petitti, 1994), and an analysis
of heterogeneity or variation in study results can be an important function of meta-analysis
(Greenland, 1998). Blair et al. (1995) further note that possible sources of inconsistencies or
heterogeneity in the relative risk or other effect measure in individual studies should be
examined in the meta-analysis. Variability in exposure characterization, study design, and
control for confounding may explain observed inconsistencies in the relative risks from multiple
studies and because of this may limit or even preclude the statistical combination of multiple
studies.
A pooled analysis, an approach that combines studies of a similar design and of similar
exposure patterns and concentrations, is compatible with this perspective. The summed SIRs
reported by Raaschou-Nielson et al. (2003) come closest to a pooled analysis of the Nordic
cohorts. It is also important to consider which studies to include in the meta-analysis and how to
group these studies. The epidemiologic studies included in the meta-analysis mentioned above
are not consistently .treated across reports. Each of the analyses in Table 3 has based its findings
on a combination of different studies; no one analysis appears to share a common set of studies,
although some individual studies are commonly found. Most reports in Table 3 include the
incidence studies of Axelson et al. (1994) and Anttila et al. (1995) with the more current ones,
including Hansen et al. (2001), Blair et al. (1998) (or the earlier report of this cohort by Spirtas et
22
-------
al. [1991]), Morgan et al. (2000, 1998), and Boice et al. (1999). All studies except those of
Axelson et al. (1994), Anttila et al. (1995), and Hansen et al. (2001) assessed mortality. The
meta-analyses in Table 3 report summary risk estimates and do not consider differences in the
qualities of the exposure assessment or measured outcome. For example, the incidence study of
Anttila et al. (1995) with information on TCE exposure for individual study subjects as well as
the mortality study of Garabrant et al. (1988), which lacks individual-specific TCE exposure
information, have been placed into the same grouping in one analysis. This treatment is in
contrast to the statistical approach taken by Wartenberg et al. (2000), who considered exposure
assessment characteristics and measured outcome as leading to possible heterogeneity. A
stratified analysis is the most common tool in meta-analysis for minimizing potential
heterogeneity in relative risks or effect measures resulting from study design factors (Morris,
1994).
As discussed above, Wartenberg and Scott (2002) show a lack of heterogeneity in cause-
or site-specific relative risks except that for kidney cancer in the studies considered to have more
robust inferences of TCE exposure (Tier I incidence studies), with analyses examining fewer
sites drawing similar conclusions (Kelsh, 2003; Borak et al., 2000). Heterogeneity in kidney
cancer risk is introduced from the inclusion of Henschler et al. (1995), who reported a strong
association for kidney cancer incidence. Public comments (DOD, 2002; HSIA, 2002; Rhomberg
et al., 2002a) to EPA on the external draft TCE synthesis and characterization (U.S. EPA, 2001) •
note criticisms in the scientific literature of the study of Henschler et al. (1995) and suggest this
study should not be included in a quantitative analysis. Much of the criticism focused on the
study's development originating from a cluster of kidney cancer cases (U.S. EPA, 2002b) and
lacking an a priori hypothesis (Rhomberg, 2002b). Concerns mentioned by the SAB peer
review panel included the variability of underlying population rates for kidney cancer in German
and Danish cancer registries; the magnitude of the indicated risk, which was far out of
proportion to risks observed in most other studies; and the significance of the study in light of
the whole epidemiologic database (U.S. EPA, 2002a). hi their letter to the editor to the journal
where Wartenberg et al. (2000) was published, Borak et al. (2000) wrote that the contribution to
the overall joint relative risk of Henschler et al. (1995) was not proportional to the size of the
study (the increase in the combined study population by 169 exposed and 359 total subjects,
1.3% and 2.0%, respectively, was shown to increase the relative risk by nearly 74%).
This leads to the question as to why the relative risks in Henschler et al. (1995) are
heterogeneous and larger than relative risks reported in other Tier I incidence studies. Exposure
conditions may be different for these subjects compared with other cohorts. Cherrie et al. (2001)
developed exposure concentrations for this study using information provided in Henschler et al.
(1995) to examine this hypothesis. Cherrie et al. (2001) estimated long-term average exposure
23
-------
during cleaning activities as between 10 ppm and 225 ppm, the lower end of this range similar to
average time-weighted-average concentrations reported for the Nordic cohorts (Hansen et al.,
2001; Anttila et al., 1995; Axelson et al., 1994); however, peak exposure concentrations were of
approximately 1,800 ppm to 4,000 ppm, which is consistent with the reports of narcotic effects
in cardboard workers studied by Henschler et al. (1995). Other possible explanations are
identified by the SAB panel (U.S. EPA, 2002a) and include variation in cause-specific cancer
rates in the reference or comparison population and the use of sonographic screening of exposed
workers, which would increase the ability to detect renal cell carcinomas in the exposed group
compared with the comparison population.
Overall, SAB panel members agreed that the inclusion of Henschler et al. (1995) in an
overall meta-analysis would introduce significant heterogeneity. The SAB panel also suggested
that this study may provide some information and recommended that EPA consider this study in
their revision of the TCE assessment. The SAB panel recommended that EPA include this study
in the overall analysis of the epidemiologic body of evidence, taking into account information on
exposure levels and potential biases in this study (U.S. EPA, 2002a). It is important to point out
that this study was published roughly 10 years ago and that more information has become
available. Additionally, the recent case-control studies of Pesche et al. (2000) and Bruning et al.
(2003) may have a greater statistical power than cohort studies to evaluate renal cell carcinoma
and TCE exposure. EPA seeks guidance from the NAS panel as to how Henschler et al. (1995)
can inform EPA's overall weight-of-evidence judgments.
Approaches for conducting a meta-analysis that use fixed or random effects models
weigh the contribution of an individual study to the overall summary relative risk using the
inverse of the study's variance. However, the TCE database limitations was for this approach.
Several studies lack symmetry on a log scale between the point estimate and the upper or lower
confidence bound, due to either rounding error or alternative calculation algorithms. A variance
derived from the difference between the natural log of the upper 95% confidence interval and the
natural log of the point estimate cannot be calculated in two cases: (1) where no cause- or site-
specific cancer cases were observed in an individual study, and (2) where a lower 95%
confidence interval is reported as zero; the lognormal value of zero is undefined. However,
public and other comments (DOD, 2002; Boice and McLaughlin, 2001) discuss the use of the
lower 95% confidence interval as an alternative for estimating variance. Several approaches
may be employed to examine asymmetry of the confidence interval and the overall impact on a
meta-analysis of choices concerning confidence intervals. First, EPA can ask the principal
investigator to provide more information such as the number of site-specific cancers observed
and expected in the study if asymmetry is a result of rounding error resulting from insufficient
information presented in the published study. In most cases, this information will reduce
24
-------
rounding error with greater adherence to statistical demands for symmetry. In cases of zero
reported site-specific cases, Bayesian methods can be examined to obtain an estimate of the
observed number with a confidence interval developed using the posterior information.
Sensitivity analysis can examine the overall influence of variances calculated from both the
upper and lower 95% confidence intervals on the overall site-specific finding. Finally, methods
using a summed value of observed and expected site-specific cases can be examined, although
this approach excludes information contributed by studies with internal referents, such as Blair et
al. (1998).
4. SUMMARY
The issues raised in the above sections suggest a number of potential approaches for
evaluating the TCE cancer epidemiologic evidence. EPA proposes to carry out a systematic
review of the epidemiologic body of literature, carefully considering the utility of criteria-based
approaches and meta-analysis approaches to an appropriate subset of studies. This review will
be characterized according to well-characterized principles for evaluating studies,such as clearly
stated purpose, careful literature searches, explicit inclusion and exclusion criteria, assessment of
study validity and bias, and well-articulated definitions and rules of inference for selected causal
criteria. EPA will use these criteria in weighing the overall epidemiologic evidence and provide
tables summarizing critical information for each key study including type of study, number of
subjects, sources of exposure information, years and estimated levels of exposure, and basis for
the estimated exposure levels.
25
-------
Table 1. Epidemiologic studies assessing cancer and trichloroethylene exposure
Study
I. Cohort Studies
Anttila et al. (1995) (Tola et al., 1980; Axelson et al., 1978)
Axelsonetal. (1994)
Blair (1980)
Blair etal. (1989)
Blair et al. (1998) (Spirtas et al., 1991)
Boice etal. (1999)
Chang et al. (2003a,b)
Dubrow and Gute (1987)
Henschler et al. (1995) (Swaen, 1995;
Bloemen and Tomenson, 1995)
Garabrantetal. (1988)
Hansen et al. (2001)
Morgan et al. (2000,1998)
Raaschou-Nielsen et al. (2003)
Ritz(1999)
Sinks etal. (1992)
Shannon etal. (1988)
Shindell and Ulrich (1985)
Stem etal. (2001)
II. Case-Control Studies
Brtoing et al. (2003)
Doseraecietal. (1999)
Fredriksson et al. (1989)
Fritschi and Siemiatycki (1996)
Greenland (1994)
Hardell etal. (1981)
Hardell et al. (1994)
Heineman etal. (1994)
Persson etal. (1989)
Pesche et al. (2000)
Vamvakas et al. (1998) (Green and Lash, 1999;
Vamvakas et al., 2000; Mandel, 2001; Vamvakas
Outcome
Incidence
Incidence
Mortality
Mortality
Mortality/Incidence
Mortality
Mortality
Mortality (PMR)
Site-cancer for mortality, kidney cancer incidence
Mortality
Incidence
Mortality
Incidence
Mortality
Mortality/Incidence
Mortality
Mortality
Mortality (PMR)
Renal cell carcinoma
Renal cell carcinoma
Colon cancer
Melanoma
Several cancer sites
Lymphoma
Lymphoma
Astrocytic brain cancer
Lymphoma
Renal cell carcinoma
Renal cell carcinoma
etal., 2001)
26
-------
Table 1. Epidemiologic studies assessing cancer and trichloroethylene exposure
(continued)
III. Community Studies
Cohnetal. (1994)
Costas et al. (2002) (MaDPH, 1997; Lagakos et al., 1986)
Faglianoetal.(1990)
Flood and Chapin (1988)
Flood etal. (1990)
Isacsonetal. (1985)
Lee et al. (2003)
Mallin(1990)
Morgan and Cassady (2002)
Turnbulletal. (1990)
Waller etal. (1992)
Waller and Turnbull (1993)
Vartianinen et al. (1993)
Leukemia, NHL
Leukemia (case-control)
Leukemia
Leukemia
Leukemia
A number of site-specific cancers
Liver, stomach, colorectal, and lung cancers
Bladder cancer
Total cancer and 12 site-specific cancers
Leukemia
Leukemia
Leukemia
Liver, Hodgkin's disease, NHL, leukemia
Note: Epidemiologic studies were identified through a search of the National Library of Medicine's PubMed
database (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi) using the keywords "trichloroethylene cancer
epidemiology" or were cited in Wartenberg et al. (2000).
PMR = proportional mortality ratio; NHL = non-Hodgkin's lymphoma.
27
-------
Table 2. Causal inferences as reported in reviews of TCE body of evidence
IARC (1995)
"Overall, the important observations are the elevated risk for cancer of the liver
and biliary tract and the modestly elevated risk for NHL in all three of the most
informative cohort studies. The suggested marginally increased risk for NHL in
areas with TCE-contaminated groundwater is noted."
Weiss (1996)
"It is clear from a review of the data that, both in terras of the small relative
increases seen and the small number of observations upon which those increases
are based, the evidence currently available in support of a causal hypothesis is
quite limited."
McLaughlin and Blot (1997)
"There is little evidence of an increased risk of renal-cell cancer and exposure to
TCE or perchloroethylene." "Although it is virtually impossible using
epidemiology data to rule out conclusively a small increase in risk of renal-cell
cancer, the totality of data reviewed herein are not remotely close to suggesting a
cause-effect relationship between renal cancer and TCE and perchloroethylene."
Lynge et al. (1997) 'The three most informative cohorts consistently indicated an excess relative
risk for cancer of the liver and biliary tract with 23 observed cases and 12.9
expected. The risk for these cancers also was elevated in one of the two less
informative studies." "The results for NHL were consistent across the three
most informative studies indicating a modest excess relative risk, with 27
observed versus 18.9 expected cases." "There is evidence for increased risks of
cancer following exposure to: trichloroethylene (for the liver and biliary tract
and for NHL)...."
McLaughlin and Lipworth (2000) "There have been 6 published cohort studies of TCE-exposed workers, with
none reporting a significantly increased risk of renal cell cancer. There have
been 6 case-control studies that tried to evaluate TCE specifically and none has
reported a significant association. Thus, the weight of the epidemiologic
evidence provides no credible support for the hypothesis that TCE causes renal
cell cancer in humans."
Brttaing and Bolt (2000)
Wartenberg et al. (2000)
"In general, the human nephrocarcinogencitiy of trichloroethylene appears
practically to be a high-dose phenomenon."
"Overall, our analysis is consistent with that of IARC (13) and Weiss (6) but
suggests more strongly an association of TCE exposure with kidney and liver
cancer and some support for Hodgkin's disease and NHL. There is also a
possible association of cervical cancer with TCE and PER exposure. Some data
suggest associations between TCE exposure and multiple myeloma and prostate,
laryngeal, and colon cancers." "In terms of Hill's aspects of causation, we find
moderate support."
28
-------
Table 2. Causal inferences as reported in reviews of TCE body of evidence (continued)
Lavin et al. (2000)
"In summary, the epidemiological evidence as a whole does not support a causal
association between human occupational TCE exposure and an increased
incidence of lung, liver, or kidney cancers."
Institute of Medicine (2003)
NTP (2002)
Consensus Not Reached on Category of Association
• Trichloroethylene and colon cancer
• Trichloroethylene and cervical cancer
Summary of the Committee's Consensus Conclusions
Inadequate/Insufficient Evidence to Determine Whether An Association Exists:
Cancers
• Specific solvents other than tetrachloroethylene and dry-cleaning
solvents and bladder cancer
Specific solvents other than tetrachloroethylene and dry-cleaning
solvents and kidney cancer
Specific solvents other than benzene and brain and other central
nervous system cancers
Specific solvents other than benzene and NHL
Specific solvents other than benzene and acute and adult leukemia
"Trichloroethylene (TCE) is reasonably anticipated to be a human carcinogen
based on limited evidence of carcinogenicity from studies in humans, sufficient
evidence of carcinogenicity from studies in experimental animals, and which
indicates there is an increased incidence of malignant and/or a combination of
malignant and benign tumors at multiple tissue sites in multiple species of
experimental animals, and information suggesting TCE acts through
mechanisms that indicate it would likely cause cancer in humans." "Evidence
for the carcinogenicity of TCE in humans comes from seven cohort studies with
specific TCE exposure well characterized for individual study subjects. A meta-
analysis of these cohort studies found that occupational exposures to TCE was
associated with excess hi liver cancer, non-Hodgkin's lymphoma, prostate
cancer, and multiple myeloma, with the strongest evidence for the first three
cancers (Wartenberg et al, 2000)." "Findings from other cohort studies, with
less accurate assessment of TCE exposures, have more variable results.
Exposure to TCE was assessed less accurately in case-control studies; in many
studies, TCE exposure was estimated from exposure to solvents in general.
These studies typically reported higher cancer rates for tumor site similar to
those observed in the cohort studies."
29
-------
Table 2. Causal inferences as reported in reviews of TCE body of evidence (continued)
Wong (2004)
"A review of epidemiologic studies of workers exposed to.TCE does not support
a causal association between exposure to the solvent TCE and an increased risk
of any site-specific cancer (including cancer of the liver and biliary passages and
NHL)." "Ingestion of water with typical TCE contamination levels does not
result in any significant increase in cancer risk."
NHL = non-Hodgkin's lymphoma; IARC = International Agency for Research on Cancer; PER = perchloroethylene
30
-------
V
S2
U
VI
Pv
-------
Qjj
S)
•3
o
5
I
.2
w
t
I
i
SJ>
•a
.3
•§
4>
•a
et
2
•s
^
If)
s
^
^g
a
••C
«
"5
IB
1
S
t
s
IX
"5
_>
*«
e
R
£
J
•£
55
•S
«
=
k
e
«
^H
S
'•*
a
e
5
o.
I
£)
£ * -
•i > X
2 3 1,
(o1 ^" C7
— (S <*i
vo' rf ff
2^ i O
O OQ «•«
— — rJ
f
ON
X^1
•a
ts
d
1
U
1
^S
Anttilaetal-. (1995),
Hansenetal. (20011
•o
V
y
1
1
W
1
SW
¥
•a
s
i__i
en
^H
*.
o^
o
o
ibcohort
=i
(A
PJ
S
1
Anttilaetal. (1995)
•a
-H
U
t
=8
>
-0
o
1
K
B1
^^
^
I— 1
of
o
(S
X-N
Axelsonetal. (199^
!
wj
,^
Os
0
»•
^r
"j
1
TCEsub
I
/***
00
0V
OS
K
Morgen et al. (2000
1
U_l
R
oT
o
*M
Blair etal. (1998),
jjj1
x-^v
°T
rf
>— i
VO
1
J3
3
UJ
U
i
Boiceetal. (1999)
—»,
.Hansenetal. (2001
r
•a
X ffi
32
-------
REFERENCES
ACS (American Cancer Society). (2004) Estimated US Cancer Deaths. Available at http://www.cancer.org/.
Anttila, A; Pukkala, E; Sallmen, M; et al. (1995) Cancer incidence among Finnish workers exposed to halogenated
hydrocarbons. J Occup Environ Med 37:797-806.
Axelson, O. (2004) Negative and non-positive epidemiological studies. Int J Occup Med Environ Health
17:115-121.
Axelson, O; Andersson, K; Hogstedt, C; et al. (1978). A cohort study on trichloroethylene exposure and cancer
mortality. J Occup Med 20:194-196.
Axelson, 0; Selden, A; Andersson, K; et al. (1994) Updated and expanded Swedish cohort study of trichloroethylene
and cancer risk. J Occup Med 36:556-562.
Blair, A. (1980) Mortality among workers in the metal polishing and plating industry, 1951 -1969. J Occup Med
22:158-162.
Blair, A; Haas, T; Prosser, R; et al. (1989) Mortality among United States Coast Guary marine inspectors. Arc
Environ Health 44:150-156.
Blair, A; Burg, J; Foran, J; et al. (1995) Guidelines for application of meta-analysis in environmental epidemiology.
Regul Toxicol Pharmacol 22:189-197.
Blair, A; Hartage, P; Stewart, PA; et al. (1998) Mortality and cancer incidence of aircraft maintenance workers
exposed to trichloroethylene and other organic solvents and chemicals: extended followup. Occup Environ Med
55:161-171.
Bloemen, L; Tomenson, J. (1995) Increased incidence of renal cell tumours in a cohort of cardboard workers
exposed to trichloroethylene. Letter to the Editor. Arch Toxicol 70:129-130.
Boice, JD, Jr; Marano, DE; Fryzek, JP; et al. (1999) Mortality among aircraft manufacturing workers. Occup
Environ Med 56:581-597.
Boice, JD; McLaughlin, JK. (2001) Error in TCE analysis. Letter to Editor. Environ Health Persepct 109:A108.
Bolt, H; Lammert, M; Selinski, S; et al. (2004). Urinary alpha 1-microglobulin excretion as biomarker of renal
toxicity in trichloroethylene-exposed persons. Int Arch Occup Environ Health. 77:186-190.
Borak, J; Russi, M; Puglisi, JP. (2000) Meta-analysis of TCE carcinogenicity. Letter to the Editor. Environ Health
Perspect 108:A542-544.
Brauch, H; Weirich, G; Horaauer, MA; et al. (1999) Trichloroethylene exposure and specific somatic mutations in
patients with renal cell carcinoma. J Natl Cancer Inst. 91:854-861.
Brauch, H; Weirich, G; Klein, B; et al. (2004) VHL mutations in renal cell cancer: dose occupational exposure to
trichloroethylene make a difference? Toxicol Lett 151:301-310.
Breslow, NE; Day, NE. (1980) Stat'stcal method in Cancer Research. Vol 1: The Analysis of Case-Control Studies.
IARC Scientific Publications. No. 32. ISBN 92-832-0132-9.
Breslow, RA; Ross, SA; Weed, DL. (1998) Quality of reviews in epidemiology. Am J Public Health 88:475-477.
Bruning, T; Bolt, HM. (2000) Renal toxicity and carcinogenicity of trichloroethylene: key results, mechanisms, and
controversies. Crit Rev Toxciol 30:253-285.
33
-------
BrOning, T; Weirich, G; Hornauer, MA; et al. (1997) Renal cell carcinomas in trichloroethene (TRI) exposed
persons are associated with somatic mutations in the von Hippel-Lindau (VHL) tumour suppressor gene. Arch
Toxicol 71:332-335.
Brttaing, T; Pesche, B; Wiesenhtltter, B; et al. (2003) Renal cell cancer risk and occupational exposure to
trichloroethylene: Results of a consecutive case-control study in Arnsberg, Germany. Am J Ind Med 43:274-285.
Bukowski, JA; Huebner, WW; Schnatter, R; et al. (2003) An analysis of the risk of B-lymphocyte malignancies in
industrial cohorts. J Toxicol Environ Health (Part A) 66:581-597.
Chang, Y; Tai, CF; Lin, RS; et al. (2003a) A proportionate cancer morbidity ratio study of workers exposed to
chlorinated organic solvents in Taiwan. Ind Health 41:77-87.
Chang, Y-M; Tai, C-F; Yang, S-C; et al. (2003b) A cohort mortality study of workers exposed to chlorinated organic
solvents in Taiwan. Ann Epidemiol 13:652-660.
Cherrie, JW; Kromhout, H; Semple, S. (2001) The importance of reliable exposure estimate in deciding whether
trichloroethylene can cause kidney cancer. Letter to the Editor. J Cancer Res Clin Oncol 127:400-402.
Cogliatti, SB; Schmid, U. (2002) Who is WHO and what was REAL? Swiss Med Wkly 132:607-617.
Conn, P; Klotz, J; Bove, F; et al. (1994) Drinking water contamination and the incidence of leukemia and non-
Hodgkin's lymphoma. Environ Health Perspect 102:556-561.
Costas, K; Knorr, RS; Condon, SK. (2002) A case-control study of childhood leukemia in Wobum, Massachusetts:
the relationship between leukemia incidence and exposure to public drinking water. Sci Total Environ 300:23-35.
Crew, KD; Neugut, Al. (2004) Epidemiology of upper gastrointestinal malignancies. Semin Oncol 31:450-464.
Crump, KS; Canady, R; Kogevinas, M. (2003) Meta-analysis of dioxin cancer dose response for three occupational
cohorts. Environ Health Perspect 111:681-687.
Delgado-Rodriquez, M. (2001) Glossary on meta-analysis. J Epidemiol Community Health 55:534-536.
Dewer, R; Siemiatycki, J; Gerin M. (1991). Loss of statistical power associated with the use of a job-exposure
matrix in occupational case-control studies. Appl Occup Environ Hyg 6:508-515.
DOD (Department of Defense). (2002) Public comments to US EPA. Letter from John Paul Woodley, Jr., Assistant
Deputy Under Secretary of Defense (Environment) to Jim Cogliano, Ph.D. January 18,2002. Available at
http://docket.epa.gov/edkpub/do/EDKStaffltemDetailView?objectId=090007d48037bc02 .
Dosemeci, M; Cocco, P; Chow, WH. (1999) Gender differences in risk of renal cell carcinoma and occupational
exposures to chlorinated aliphatic hydrocarbons. Am J Ind Med 36:54-59.
Dubrow, R; Gute, DM. (1987) Cause-specific mortality among Rhode Island jewelry workers. Am J Ind Med
12:579-593.
Fagliano, J; Berry, M; Bove, F; et al. (1990) Drinking water contamination and the incidence of leukemia: an
ecologic study. Am J Public Health 80:1290-1212.
Federal Focus, Inc. (1996) Principles for evaluating epidemic logic data in regulatory risk assessment. Expert Panel
Report. Washington, DC: Federal Focus, Inc.
Fisher RI. (2003) Overview of non-Hodgkin's lymphoma: biology, staging, and treatment. Semin Oncol 30:3-9.
Flood, T; Chapin, C. (1988) Report on mortality in Maricopa County. Phoenix, Arizona: Arizona Department of
Health Services.
34
-------
Flood, T; Aickin, M; Lucier, J; et al. (1990) Incidence study of childhood cancer in Maricopa County: 1965-1986,
Phoenix, AZ: Arizona Department of Health Services.
Fredriksson, M; Bengtsson, NO; Hardell, L; et al. (1989) Colon cancer, physical activity, and occupational
exposures: a case-control study. Cancer 63:1838-1842.
Fritschi, L; Siemiatycki, J. (1996). Melanoma and occupational: results of a case-control study. Occup Environ
Med 53:168-173.
Garabrant, DH; Held, J; Langholz, B; et al. (1988) Mortality of aircraft manufacturing workers in Southern
California. Am JInd Med 13:683-693.
Garcia, AM; Checkoway, H. (2003) A glossary for research in occupational health. J Epidemiol Community Health
57:7-10,
Gomez, MR; Cocco, P; Dosemeci, M; et al. (1994) Occupational exposure to chlorinated aliphatic hydrocarbons:
Job exposure matrix. Am J Ind Med 26:171-183.
Green, LC; Lash, TL. (1999) Re: "Renal cell cancer correlated with occupational exposure to trichloroethylene,"
Letter to the Editor. J Cancer Res Clin Oncol 125:430-432.
Green, T; Dow, J; Ong, CN; et al. (2004) Biological monitoring of kidney function among workers occupationally
exposed to trichloroethylene. Occup Environ Med 61:312-317.
Greenland, S. (1994) Invited commentary: A critical look at some popular meta-analytic methods. Am J Epidemiol
140:290-296.
Greenland, S. (1998) Chapter 32, Meta-Analysis. In: Rothmann, KJ; Greenland, S, eds. Modern epidemiology, 2nd
ed. Philadelphia, PA: Lippincott Williams and Wilkins.
Hansen, J. (2004) Cohort studies of cancer risk among Danish workers exposed to TCE. Presentation at Symposium
on New Scientific Research Related to the Health Effects of Trichloroethylene, February 26-27,2004, at the
Renaissance Mayflower Hotel, Washington, DC). Available online at http://cfbub2.epa.gov/ncea/cfm/recordisplay
-cfm?deid=75934.
Hansen, J; Raaschou-Nielsen, O; Christensen, JM; et al (2001) Cancer incidence among Danish workers exposed to
trichloroethylene. J Occup Environ Med 43:133-139.
Hardell, L; Eriksson, M; Lenner, P; et al. (1981) Malignant lymphoma and exposure to chemicals, especially organic
solvetns, chlorophenols and phenoxy acids: a case-control study. Br J Cancer 43:169-176.
Hardell, L; Eriksson, M; Degerman, A. (1994) Exposure of phenoxyacetic acids, chlorophenols, or organic solvents
in relation to histopathology, stage, and anatomical localization of non-Hodgkin's lymphoma. Cancer Res
54:2386-2389.
HEI (Health Effects Institute). 1999) Diesel emissions and lung cancer; epidemiology and quantitative risk
assessment. Cambridge, MA: Health Effects Institute.
Heineman, EF; Cocco, P; Gomez, MR; et al. (1994) Occupational exposure to chlorinated aliphatic hydrocarbons
and risk of astrocytic brain cancer. Am J Ind Med 26:155-169.
Henschler, D; Vamvakas, S;Lamraert, M; etal. (1995) Increased incidence of renal cell tumors in a cohort of
cardboard workers exposed to trichloroethene. Arch Toxicol 69:291-299.
Herrinton, LJ. (1998) Epidemiology of the Revised European-American Lymphoma classification subtypes.
Epidemiol Rev 20:187-203.
35
-------
Hertz-Picciotto I. (1995) Epidemiology and quantitative risk assessment: a bridge from science to policy. Am J
Public Health 85:484-491.
Hes, FJ; McKee, S; Taphoorn, MJB; et al. (2000) Cryptic von Hippel-Lindau disease: germline mutations in patients
with haemangioblastoma only. J Med Genet 37:939-943.
Hill AB. (1965) The environment and disease: association or causation? Proc R Soc Med 58:295-300.
HSIA (Halogenated Solvents Industry Alliance). (2002) Public comments. Letter from PH Dugard, to the U.S. EPA
dated January 18, 2002. Available online at http://docket.epa.gov/edkpub/do/EDKStafiItemDeteilView?objectId=
090007d480381590.
Hulka, BS; Margolin, BH. (1992) Methodologic issues in epidemiologic studies using biological markers. Am J
Epidemiol 135:200-209.
IARC (International Agency for Research on Cancer). (1995) IARC Monographs on the Evaluation of Carcinogenic
Risks to Humans. Vol 63. Dry cleaning, some chlorinated solvents and other industrial chemicals. Lyon, France:
International Agency for Research on Cancer.
IOM (Institute of Medicine of the National Academies). (2003) Gulf war and health. Volume 2. Insecticides and
Solvents. Washington, DC: The National Academies Press.
Isacson, P; Bean, JA; Splinter, R; et al. (1985) Drinking water and cancer incidence in Iowa. III. Association of
cancer with indices of contamination. Am J Epidemiol 121:856-869.
Kelsh, M. (2003) What can we learn about TCE carcinogenicity from epidemiology studies? Presentation at the 29lk
Annual Toxicology Forum, Summer Meeting. Aspen, Colorado, July 13-17.
Kelsh, M; Cher, D; Exuzides, A; et al. (2002) Critique of the Wartenberg et al. .review on the epidemiologic
evidence of TCE carcinogenicity. Submitted as public comment to the U.S. Environmental Protection Agency by
the Halogenated Solvent Industry Alliance. Available online at http://docket.epa.gov/edkpub/do/
EDKStaffltemDetailView?objectId=090007d480381590 .
Lagakos, SW; Wessen, BJ; Zelen, J. (1986) An analysis of contaminated well water and health effects in Woburn,
Massachusetts. J Am Stat Assoc 81:583-5 96.
Lavin, AL; Jacobson, CF; DeSesso, JM. (2000) An assessment of the carcinogenicity potential of trichloroethylene.
Hum Ecol Risk Assess 6:565-641.
Lee, J-H; Chung, C-W; Ma, Y-C; et al. (2003) Increased mortality odds ratio of male liver cancer in a community
contaminated by chlorinated hydrocarbons in groundwater. Occup Environ Med 60:364-369.
Lynge, E; Anttila, A; Hemminki, K. (1997) Organic solvents and cancer. Cancer Causes Control 8:406-419,
MaDPH (Massachusetts Department of Public Health). (1997) Woburn childhood leukemia follow-up study. Boston
MA: Bureau of Environmental Health A.
Mallin, K. (1990) Investigation of a bladder cancer cluster in northwestern Illinois. Am J Epidemiol 132:S96-S106.
Mandel, J. (2001) Renal cell cancer correlated with occupational exposure to trichloroethylene. Letter to the Editor.
J Cancer Res Clin Oncol 127:265-266.
Mandel, J; Kelsh, M. (2001) A review of the epidemiology of trichloroethylene and kidney cancer. Hum Ecol Risk
Assess 7:727-735.
Marano, DE; Boice, JD; Fryzek, JP; et al. (2000) Exposure assessment for a large epidemiological study of aircraft
manufacturing workers. Appl Occup Environ Hyg 15:644-656.
36
-------
McLaughlin, J; Blot, W- (1997) A critical review of epidemiology studies of trichloroethylene and perchloroethylene
and risk of renal cell cancer. Int Arc Occup Environ Health 70:222-231.
McLaughlin, JK; Lipworth, L. (2000) Epideraiologic aspects of renal cell cancer. Semin Oncol 27:1115-123.
Morgan, JW; Cassady, RE. (2002) Community cancer assessment in response to long-time exposure to perchlorate
and trichloroethylene in drinking water. J Occup Environ Med 44:616-621.
Morgan, RW; Kelsh, MA; Zhao, K; et al. (1998) Mortality of aerospace workers exposed to trichloroethylene.
Epidemiology 9:424-431.
Morgan, RW; Kelsh, MA; Zhao, K; et al. (2000) Erratum. Epidemiology 11:360.
Monson, R. (1980) Occupational Epidemiology. Boca Raton, FL: CRC Press Inc.
Morris, RD. (1994) Meta-analysis in cancer epidemiology. Environ Health Perspect l02(Suppl 8):61-66.
NRC (National Research Council). (1983) Risk assessment in the federal government: managing the process.
Washington, DC: National Academy Press.
NRC (National Research Council). (1988) Health risks of radon and other internally deposited alpha-emitters. BEIR
IV. Washington, DC: National Academy Press.
NRC (National Research Council). (1994) Science and Judgment in risk assessment. Washington, DC: National
Academy Press.
NTP (National Toxicology Program). (2002) 10th report on carcinogens. Public Health Service, U.S. Department of
Health and Human Services, Research Triangle Park, NC. Available online at http://ehp.niehs.nih.gov/roc/tenth
/intro.pdf.
Ojajarvi, A; Partanen, T; Anlbom, A; et al. (2001) Risk of pancreatic cancer in workers exposed to chlorinated
hydrocarbon solvents and related compounds: a meta-analysis. Am J Epidemiol 153:841-850.
Pastides, H; Mundt, KA. (1991) Occupational epidemiology resource manual. Chemical Manufacturers Association.
Pearce, N; Checkoway, H; Shy, C. (1986) Time-related factors as potential confounders and effect modifiers in
studies based on an occupational cohort. Scand J Work Environ Health 12:97-107.
Percy, C; Stanek, E; Gloeckler, L. (1981) Accuracy of cancer death certificates and its effect on cancer mortality
statistics. Am J Public Health 71:242-250.
Percy, C; Ries, LG; Van Holten, VD. (1990) The accuracy of liver cancer as the underlying cause of death on death
certificates. Public Health Rep 105:361-367.
Persson, B; Dahlander, A-M; Fredriksson, M; et al. (1989) Malignant lymphoma and occupational exposures. Br J
Ind Med 46:516-520.
Pesche, B; Brttning, T. (2004) Renal cell cancer risk and occupational exposure to trichloroethylene. Presentation at
the Symposium on New Scientific Research Related to the Health Effects of Trichloroethylene, February 26-27,
2004 at the Renaissance Mayflower Hotel, Washington, DC. Available online at http://cfpub2.epa.gov/ncea/cfm
/recordisplay.cfm?deid=75934.
Pesche, B; Haerting, J; Ranft, U; et al. (2000) Occupational risk factors for renal cell carcinoma: agent-specific
results from a case-control study in Germany. Int J Epidemiol 29:1014-1024.
Petitti, DB. (1994) Meta-analysis, decision analysis, and cost-effectiveness analysis. New York, NY: Oxford
University Press.
37
-------
Raaschou-Nielsen, O; Hansen, J; Christensen, JM; et al. (2001) Urinary concentrations of trichloroacetic acid in
Danish workers exposed to trichloroethylene, 1947-1985. Am J Ind Med 39:320-327.
Raaschou-Nielsen, 0; Hansen, J; Thomson, BL; et al. (2002) Exposure of Danish workers to trichloroethylene,
1947-1989. ApplOccup Environ Hyg 17:693-703.
J
Raaschou-Nielsen, 0; Hansen, J; McLaughlin, JK; et al. (2003) Cancer risk among workers at Danish companies
using trichloroethylene: a cohort study. Am J Epidemiol 158:1182-1192.
Rhomberg, L. (2002a) Summary of oral comments. Presentation at the June 18, 2002, meeting of U.S. EPA Science
Advisory Board.
Rhomberg, L. (2002b) The a posteriori probability of a kidney cancer cluster attributed to trichloroethylene
exposure. Letter to the Editor. Environ Health Perspect 110: A232.
Ries, LAG; Eisner, MP; Kosary, CL; et al., eds. (2004) SEER cancer statistics review, 1975-2001. Bethesday, MD:
National Cancer Institute. Available online at http://www.seer.cancer.gov/csr/! 975_2001/.
Ritz, B. (1999) Cancer mortality among workers exposed to chemicals during uranium processing. J Occup Environ
Med 41:556-566.
Selikoff, 1J; Seidman, H. (1992) Use of death certificates in epideraiological studies, including occupational hazards:
variations in discordance of different asbestos-associated diseases on best evidence ascertainment. Am J Ind Med
22:481-492.
Seto, M. (2004) Genetic and epigenetic factors involved in B-cell lymphomagenesis. Cancer Sci 95:704-710.
Shannon, HS; Haines, T; Bernholz, C; et al. (1988) Cancer morbidity in lamp manufacturing workers. Am J Ind
Med 14:281-290.
Shindell, S; Ulrich, S. (1985) A cohort study of employees of a manufacturing plant using trichloroethylene. J
Occup Med 27:577-579.
Siemiatycki, J. (1991) Risk factors for cancer in the workplace. Boca Raton, FL: CRC Press.
Siemiatycki, J; Richardson, L; Straif, K; et al. (2004) Listing occupational carcinogens. Environ Health Perspect
112:1447-1459.
Sinks, T; Lushniak, B; Haussler, JB; et al. (1992) Renal cell cancer among paperboard printing workers.
Epidemiology 3:483-489.
Spirtas, R; Stewart, PA; Lee, JS; et al. (1991) Retrospective cohort mortality study of workers at an aircraft
maintenance facility. I. Epideraiological results. BrJ Ind Med 48:515-530.
Stangl, DK: Berry, DA. (2000) Meta-analysis in medicine and health policy. New York, NY: Marcel Dekkar, Inc.
Stayner, L; Bailer, AJ; Smith, R; et al. (1999) Sources of uncertainty in dose-response modeling of epidemiological
data for cancer risk assessment. Ann NY Acad Sci 895:212-222.
Stern, F; Lehman, E; Ruder, A. (2001) Mortality among unionized construction plasterers and cement masons.
Am J Ind Med 39:373-388.
Stewart, PA; Lee, JS; Marano, DE; et al. (1991) Retrospective cohort mortality study of workers at an aircraft
maintenance facility. II. Exposures and their assessment. Br J Ind Med 48:531-537.
Swaen GMH. (1995) Increased incidence of renal cell tumours in a cohort of cardboard workers exposed to
trichloroethylene. Letter to the Editor. Arch Toxicol 70:127-128.
38
-------
Taubes, G. (1995) Epidemiology reaches its limits. Science 269:164-169.
Teschke, K; Olshan, AF; Daniels, JL; et al. (2002) Occupational exposure assessment in case-control studies:
Opportunities for improvement. Occup Environ Med 59:575-594.
Tola, S; Vilhunen, R; Jarvinen, E; et al. (1980) A cohort study on workers exposed to trichloroethylene.
J Occup Med 22:737-40.
Tumbull, BW; Iwano, EJ; Burnett, WJ; et al. (1990) Monitoring for clusters of disease: application to leukemia
incidence in upstate New York Am JEpidemiol 132:814-322.
U.S. EPA (Environmental Protection Agency). (1999) Guidelines for carcinogen risk assessment [review draft].
Risk Assessment Forum, Washington, DC: NCEA-F-0644. Available online at
http ://www.epa.gov/ncea/raf7rafguid .htm.
U.S. EPA (Environmental Protection Agency). (2001) Trichloroethylene health risk assessment: synthesis and
characterization. Washington, DC; EPA/600/P-01/002A. Available online at http://www.epa.gov/ncea.
U.S. EPA (Environmental Protection Agency). (2002a) Review of the draft trichloroethylene health risk assessment:
synthesis and characterization: An EPA Science Advisory Board Report. Washington, DC; EPA-SAB-EHC-03-002.
U.S. EPA (Environmental Protection Agency). (2002b) Summary minutes of public meeting of Science Advisory
Board Environmental Health Committee Trichloroethylene Health Risk Assessment: Synthesis and Characterization
Review Panel; June 18-19; Washington D.C. Available online at http://www.epa.gov/sab/pdf7tcedft061802m.pdf.
Vamvakas, S; Bruning, T; Bolt, HM; et al. (1998) Renal cell cancer correlated with occupational exposure to
trichloroethene. J Cancer Res Clin Oncol. 124:374-382.
Vamvakas, S; Bruning, T; Bolt, HM; et al. (2000) Renal cell cancer correlated with occupational exposure to
trichloroethene. Reply. J Cancer Res Clin Oncol. 126:178-180.
Vamvakas, S; Bruning, T; Bolt, HM; et al. (2001) Renal cell cancer correlated with occupational exposure to
trichloroethene: reply to letter to the editors.. J Cancer Res Clin Oncol. 127:267-268.
Vartianinen, T; Pukkala, E; Reinoja, T; et al. (1993) Population exposure to tri- and tetrachloroethene and cancer
risk: two cases of drinking water pollution. Chemosphere 27:1171-1181.
Waller, LA; Turnbull, BW. (1993) The effects of scale on tests for disease clustering. Stat Med 12:1869-1884.
Waller, LA; Turnbull, BW; Clark, LC; et al. (1992) Chronic disease surveillance and testing of clustering of disease
and exposure: application to leukemia incidence and TCE-contaminated dump-sites in upstate New York.
Environmetrics 3:281-300.
Wartenberg, D; Scott, CS. (2002) Carcinogenicity of trichloroethylene. Letter to the editor. Environ Health
Perspect HO:A13-A14.
Wartenberg, D; Reyner, D; Scott, CS. (2000) Trichloroethylene and cancer: epidemiologic evidence. Environ
Health Perspect 108(suppl 2):161-176.
Wartenberg, D; Reyner, D; Scott, CS. (2001) Errors in TCE analysis: response. Environ Health Perspect
109:A108-A109.
Weed, DL. (1997) Methodologic guidelines for review papers. J Natl Cancer Inst 89:6-7.
Weed, DL. (2000) Interpreting epidemiological evidence: how meta-analysis and causal inference methods are
related. IntJEpidemiol29:387-390.
WeedDL. (2002) Environmental epidemiology Basics and proof of cause-effect. Toxicol 181-182:399-403.
39
-------
Weirich, G; Klein, B; W6hl, T; et al. (2002) VHL2C phenotype in a German von Hippel-Lindau family with
concurrent VHL germline mutations P81S and L188V. J Clin Endocrinol Metab 87:5241-5246.
Weiss,N. (1996) Cancer in relationship to occupational exposure to trichloroethylene. Occup Environ Med 53:1-5.
WHO (World Health Organization). (2000) Evaluation and use of epidemiological evidence for environmental
health risk assessment: WHO Guideline Document. Environ Health Perspect 108:997-1002.
Wong, O. (2004) Carcinogenicity of trichloroethylene: an epidemiologic assessment. Clin Occup Environ Med
4:537-589.
Wong, O; Raabe, G. (1996) Application of meta-analysis in reviewing occupational cohort studies. Occup Environ
Med 53:793-800.
40
-------
------- |