9
General Review of Epidemiologic Evidence Pertaining to Cancer
The Environmental Protection Agency (EPA) draft Integrated Risk Information System (IRIS) assessment of tetrachloroethylene characterizes the epidemiologic literature as supportive of classifying tetrachloroethylene as a likely carcinogen. That classification is based primarily on reported associations with hematopoietic, lymphopoietic, and esophageal cancers. There is a substantial epidemiologic literature on the potential association of exposure to tetrachloroethylene with selected malignancies. However, the committee believes that a balanced and critical review of the human epidemiologic literature provides only limited evidence that tetrachloroethylene is carcinogenic in humans. The challenges of obtaining valid estimates of exposure, in addition to the challenges inherent in observational epidemiology, make it difficult to draw conclusions about causal associations between tetrachloroethylene and cancer in humans.
The epidemiologic literature relating tetrachloroethylene to cancer is notable in three ways: a number of studies show associations with a variety of cancers, there is limited consistency between studies with respect to the associations, and few studies were able to quantify or even identify specific tetrachloroethylene exposure. The latter point, not uncommon in studies of occupational and environmental causes of cancer, makes interpretation of the literature particularly difficult. Several positive associations are reported in the literature, but the inconsistency among studies raises concern, so a consistent critical review of the literature is needed. The draft IRIS assessment does not provide the detail and methodology used for evaluating literature. Overall, it appears that the procedure was to accept the results of positive studies with little critical evaluation of validity and to dismiss null studies of similar or better methodologic rigor as flawed. If it is EPA’s intention to err on the side of protecting public health when reviewing the literature, that should be stated clearly in the document. Otherwise, a clearer discussion of criteria used to identify studies of merit and a more balanced critique would strengthen the draft IRIS assessment.
The draft’s critiques of studies are often uneven; studies that found no association are criticized more often than studies that found a positive association even if they had similar methodologic limitations. An example is the discussion of case-control studies on page 4-150, lines 19-31. Several of the criticized features of the case-control design that are mentioned are not inherent in the design, such as that associations may be nonlinear (this design does not require categorical exposure measures) or that duration and cumulative exposure do not address age at first exposure (this information can simply be asked of participants). Many of the studies suffered from a lack of statistical power—a common problem in studying rare cancers and exposures. However, the concern over power is uneven. On page 4-149, the absence of an association between employment in dry-cleaning and death due to lymphatic and hematopoietic cancer (Ruder et al. 2001) is attributed to lack of power. In contrast, a positive association between exposure to tetrachloroethylene and multiple myeloma in aircraft maintenance workers was based on only two deaths and is described only as noteworthy but imprecise (page 4-148, lines 6-9). There is little discussion of the potentially important limitations of proportionate-mortality studies, such as inaccuracies in death certification and the inability to adjust for potential confounders. There is some discussion of confounders in relation to the standardized mortality ratio (SMR) studies of esophageal cancer on page 4-153, but it is also unbalanced in that it focuses on adjustment for smoking but does not mention the absence of adjustment for alcohol; in addition, the effect of adjustment for smoking is derived from estimates for lung cancer and may not translate directly to esophageal cancer.
A number of errors suggest an incomplete understanding of epidemiologic and statistical methods. Such errors reduce confidence in the draft’s conclusions. For example, EPA summed observed and expected cases from studies with diverse types of end points (incidence and mortality) and, using different approaches to calculating the expected values, calculated a ratio of the summed observed and expected values. Expected numbers from different studies can be added only if they are derived from the same external rates, but mortality and incidence are different. One of the most troubling misunderstandings is related to the dismissal of the results of the 2006 study by Lynge et al. In reference to that study’s findings on non-Hodgkin lymphoma (and later on bladder cancer), EPA notes that exposure information was not available on about 20% of cases and of controls and that much of the exposure information came from next of kin. It then uses that to explain why Lynge et al. found no risk associated with tetrachloroethylene exposure and suggests an automatic bias toward the null due to misclassification. In the first instance, missing exposure data are analogous to nonresponse in that the subjects are not included in any classification group. Nonresponse will not introduce bias if it is nondifferential; if it is differential, it could bias an effect measure either toward or away from the null. In the second instance, exposure information from next of kin make it more likely that hazardous exposures will be overreported by the families of workers who developed cancer than by families of workers who did not; this would have resulted in
overestimation, not attenuation, of the association. Similar arguments regarding the study are incorrectly made for other cancer sites, and the draft refers to the study as “uninformative.” It is unclear why Lynge et al. (2006) received such critical review and papers that were methodologically less sound were accepted with little comment.
The draft IRIS assessment indicated that the strongest evidence linking tetrachloroethylene to cancer consisted of observed associations with esophageal cancer and lymphoma (page 4-184, lines 6-17). Evidence on other cancer end points—including renal, bladder, cervical, and lung cancers—is less certain and does not weigh as heavily in the assessment (page 4-184, lines 25-33). After a brief and uncritical discussion of the epidemiologic literature that references the criteria for causation outlined by Hill (1965), the document concludes that “together, the evidence on tetrachloroethylene partially fulfills several of these criteria and is suggestive of a cause and effect relationship between tetrachloroethylene and human cancer. The body of human evidence is not sufficient to regard tetrachloroethylene as a known human carcinogen” (p. 44-187; emphasis added). In contrast, in Chapter 6 of the draft (“Characterization of Hazard and Dose-Response”), the evidence associating tetrachloroethylene exposure with cancer is stated more confidently (page 6-5, lines 31-35; page 6-6, lines 1-5; page 6-10, lines 27-29 and 31-35; and page 6-11, lines 1-6). It is difficult to reconcile the discussion in Chapter 4 with the conclusion in Chapter 6.
ESOPHAGEAL CANCER
The draft IRIS assessment emphasizes the association between tetrachloroethylene and esophageal cancer primarily because of the results of three studies: by Vaughan et al. (1997), Ruder et al. (2001), and Blair et al. (2003). The work by Blair et al. and Ruder et al. were mortality studies of dry-cleaner union members, and the latter was a community-based case-control study. It is interesting to compare the results of the two studies. With the same methods, the populations were enumerated from similar sources and followed for similar periods. Blair et al. followed 5,369 union members in St. Louis who worked for at least 1 year during 1948-1993. The population studied by Ruder et al. included 1,708 workers selected from union rosters in California, Illinois, Michigan, and New York. Both studies reported an excess risk of death from esophageal cancer; Blair et al. reported an SMR of 2.2 (95% confidence interval [CI], 1.5-3.3) and Ruder et al. an SMR of 2.47 (95% CI, 1.35-4.14). The excess in the paper by Ruder et al. was limited to workers with at least 20 years since first employment and was highest in those with at least 5 years of exposure (SMR, 5.03; 95% CI, 2.41-9.47). Blair et al. reported similar SMRs in workers with little or no exposure (SMR, 2.1; 95% CI, 0.9-4.4) and those with medium or high exposure (SMR, 2.2; 95% CI, 1.1-3.5).
Esophageal cancer is also associated with smoking and alcohol consumption, which are difficult to control for in mortality studies because the data are
often not available. The studies of Blair et al. and Ruder et al. also reported an excess of deaths from other causes associated with smoking, including lung cancer, emphysema, and heart disease. EPA’s draft IRIS assessment discounts potential confounding by smoking but does not adequately support its conclusion in the section on esophageal cancer (page 4-153, lines 30-33). In contrast with the findings of Blair et al. and Ruder et al., a large mortality study (Boice et al. 1999) in a population of aircraft manufacturers (N = 77,965) had an appreciable number of workers with routine (N = 2,631) and intermittent (N = 3,199) exposure to tetrachloroethylene but reported no association between that exposure and esophageal cancer. The case-control study by Vaughan et al. (1997) reported an increased but not significant odds ratio (OR) for dry-cleaning work, which was adjusted for smoking habit and alcohol consumption. That estimate was based on only two exposed cases, however, and, particularly when multiple covariates were adjusted for, was too statistically unstable to be informative (OR 3.6; 95% CI, 0.5-27.0). A methodologically sound nested case-control study by Lynge et al. (2006) reported no association between working as a dry-cleaner and esophageal cancer. Those negative findings were dismissed by EPA because some of the population could not be classified by exposure. As discussed earlier, this does not preclude the use of results based on subjects on whom exposure data were available.
Overall, there is limited evidence to support an association between tetrachloroethylene and esophageal cancer. The two mortality studies of dry-cleaners are suggestive of an association, but the potential for confounding by smoking and alcohol consumption is appreciable. Thus, the committee therefore concluded that the epidemiologic literature is not sufficient to support an association between tetrachloroethylene and esophageal cancer.
LYMPHOID CANCERS
EPA’s draft IRIS assessment concludes that the epidemiologic data “suggested an association between lymphoma and tetrachloroethylene” (p. 4-184). The committee concurs with that conclusion but adds that the data are relatively weak and inconsistent. The rationale for the committee’s conclusion is discussed in detail in Chapter 8.
Epidemiologic studies of the association between exposure to tetrachloroethylene and lymphoid cancers vary in design, validity, specificity of exposure assessment, type of population studied, outcome, and sample size, all of which contribute to the inconsistency of results and reduce confidence in conclusions that are drawn from the data.
OTHER CANCERS
A number of studies have reported associations between tetrachloroethylene and other cancers, including cervical, lung, and bladder cancer. The results
for those cancers are less consistent but should not be dismissed. The draft IRIS assessment considered those end points but did not weigh them heavily in the classification of tetrachloroethylene as a human carcinogen. That is appropriate.
GENERAL COMMENTS ON THE ENVIRONMENTAL PROTECTION AGENCY’S PRESENTATION OF EPIDEMIOLOGIC EVIDENCE ON CANCER
One of the biggest difficulties in assessing the cogency of the EPA’s assessment related to cancer is how the data are organized in the tables and some parts of the text. It would be much easier to evaluate the overall picture of results regarding tetrachloroethylene and a particular cancer if the tables were organized by cancer type as opposed to the current format, which organizes them by study design. The current format requires the reader to jump between sections for cohort mortality, incidence, and case-control studies. Studies are sometimes further categorized as to the type of worker included (for example, drycleaner vs degreaser); this makes it extremely difficulty to evaluate the overall consistency or lack of consistency in results related to specific cancers.
Errors in reporting results also occur occasionally. For example, the draft reports (on page 4-150, lines 1-3), in relation to Hodgkin disease, “a statistically significantly elevated risk for male [sic] with a job title of dry cleaner or laundry worker (Costantini et al. 2001).” The result from Costantini et al. for that group in relation to Hodgkin disease was an OR of 2.5 (95% CI, 0.3-24.6), which is not significant and was based on a single case.
The overall impression is that data are presented to support a positive association between tetrachloroethylene and cancer and that studies that found no such association are criticized or minimized. EPA should provide a clearer discussion of criteria used to identify studies of merit and a more balanced critique to strengthen the draft IRIS assessment.
RESEARCH RECOMMENDATIONS
Population-based studies, preferably in well-defined occupational cohorts, that can measure both cancer incidence and mortality and have sophisticated exposure reconstruction components that are specific to tetrachloroethylene would add significantly to the literature. The studies must also be adequately controlled for the effects of smoking and alcohol consumption to address the lingering questions of the association between tetrachloroethylene and esophageal cancer. In the absence of data to control for these confounders, sensitivity analyses should be conducted to estimate the exposure effect after adjustment under reasonable sets of assumptions regarding smoking prevalence and the strength of smoking effects. Further research that classifies exposure only by occupational title will not add to the literature.