Insights from Epidemiology into Dichloromethane and Cancer Risk

Dichloromethane (methylene chloride) is a widely used chlorinated solvent. We review the available epidemiology studies (five cohort studies, 13 case-control studies, including seven of hematopoietic cancers), focusing on specific cancer sites. There was little indication of an increased risk of lung cancer in the cohort studies (standardized mortality ratios ranging from 0.46 to 1.21). These cohorts are relatively small, and variable effects (e.g., point estimates ranging from 0.5 to 2.0) were seen for the rarer forms of cancers such as brain cancer and specific hematopoietic cancers. Three large population-based case-control studies of incident non-Hodgkin lymphoma in Europe and the United States observed odds ratios between 1.5 and 2.2 with dichloromethane exposure (ever exposed or highest category of exposure), with higher risk seen in specific subsets of disease. More limited indications of associations with brain cancer, breast cancer, and liver and biliary cancer were also seen in this collection of studies. Existing cohort studies, given their size and uneven exposure information, are unlikely to resolve questions of cancer risks and dichloromethane exposure. More promising approaches are population-based case-control studies of incident disease, and the combination of data from such studies, with robust exposure assessments that include detailed occupational information and exposure assignment based on industry-wide surveys or direct exposure measurements.


Introduction
Dichloromethane (methylene chloride) has been used extensively as a paint stripper and metal degreaser and as an extraction solvent in the food industry (e.g., in decaffeination of coffee), in the pharmaceutical industry, and in the production of cellulose triacetate film and fiber. Production and use in the United States peaked around 1980, with a production capacity of 830 million pounds per year [1], decreasing more recently to 350 to 650 million pounds per year [2].
Concerns relating to the carcinogenic potential of dichloromethane arose in the 1980s, based in large part on results from a 2-year inhalation exposure experiment conducted by the National Toxicology Program (NTP), in which dichloromethane exposure resulted in an increased incidence of liver and lung tumors in male and female B6C3F 1 mice [3,4]. Metabolism of dichloromethane is hypothesized to involve two primary pathways: a CYP2E1 dependent oxidative pathway producing carbon monoxide, and a glutathione-S transferase-theta 1 (GST-T1)-catalyzed pathway resulting in the production of two highly reactive intermediates, formaldehyde and S-(chloromethyl)glutathione, and carbon dioxide [5][6][7]. The proportion of dichloromethane metabolized via the GST pathway increases at higher exposures. Although GST is considered a detoxification pathway for many chemicals, in the case of dichloromethane it is the GST pathway that has been most strongly implicated in genotoxicity and carcinogenicity [8][9][10][11][12][13][14][15].
The database of epidemiology studies focusing specifically on dichloromethane has expanded considerably with the addition of recent studies of hematopoietic cancers conducted in Europe [16][17][18] and North America [19][20][21][22]. The purpose of this paper is to review the currently available epidemiology studies, focusing on specific cancer sites reported in rodent bioassays and identified through a review of the dichloromethane literature, and highlighting issues that could be addressed in future research.

Methods
We searched the MEDLINE database (last accessed on 30 March 2011) for epidemiologic studies related to dichloromethane and cancer risk, using dichloromethane, methylene chloride, cohort and case-control as search terms. References within relevant reports were also reviewed, and through this process we identified three papers [16,17,20] that were not found through the MEDLINE search strategy because dichloromethane was not used as an indexing term (i.e., the abstract and key words did not include dichloromethane, but dichloromethane-specific data were presented in the analysis). We supplemented this search with a review of cancer epidemiology studies with data on trichloroethylene and tetrachloroethylene, two related solvents with extensive epidemiology databases, to identify any other papers evaluating cancer risk from dichloromethane exposure that had been missed.
Eighteen papers based on epidemiologic studies of cancer risk were identified and included in this evaluation: four cohorts for which the primary solvent exposure was to dichloromethane, one large cohort of civilian employees at a military base with exposures to a variety of solvents, and 13 case-control studies of specific cancers with data on dichloromethane exposure. Publications based on shorter follow-up periods of some of the cohort studies were also identified and reviewed [23][24][25][26][27][28]. Because our focus was dichloromethane-specific results, we did not include studies that only provided estimates of general categories of solvents (e.g., any solvent, or chlorinated solvents).

Cohort Studies
No cohort studies examining cancer incidence and dichloromethane exposure were identified in the literature search; cohort mortality studies with cancer data are summarized in Table 1. Two cohorts were conducted among cellulose triacetate film manufacturing workers in New York State and the United Kingdom, with approximately 1,000 to 1,500 male workers and mean exposure concentrations of 20 to 40 ppm [29,30]. Two other cohort studies were conducted among workers in cellulose triacetate fiber plants in South Carolina and Maryland [28,31,32]. The two fiber plant studies involved higher dichloromethane exposure, and the study in Maryland [32] contained twice as many workers but a shorter mean follow-up period, compared with the two studies of film manufacturing workers. In each of these four studies, the relatively small number of deaths greatly limits their ability to provide insights regarding site-specific cancers other than lung.
Overall, there is little indication of an increased risk of lung cancer among these studies, with estimated SMRs generally <1.0 (ranging from 0.46 to 1.21). One study included discussion of smoking history, obtained from a survey at the facility, that indicated that smoking rates were similar in the workers compared with the general population; thus it is unlikely that differences in smoking could be masking an effect of dichloromethane [29]. Only one of the cohort studies reported an increased risk of liver or biliary tract cancer. In the latest follow-up of the South Carolina fiber plant cohort, the SMR for liver and bile duct cancer, based on four observed cases, was 2.98 (95% CI 0.81-7.63), lower than the SMR of 5.75 (95% CI 1.82-13.8) that was reported in the 1990 analysis based on these same four cases but on a shorter follow-up period (and thus lower number of expected cases) [28,31]. Three of these cases were biliary tract cancers. Elevated SMRs (point estimates greater than 1.0) were seen for brain cancer in both film production cohorts: SMR = 2.16 in New York [29] and 1.45 in the United Kingdom [30], with more mixed results seen in the fiber production cohorts. These estimates are based on a small number of observations (ranging from 1 to 6) and so are relatively imprecise. Although estimates of associations for leukemia (ICD-9 codes 204, 208) were available for all of the film and fiber cohort studies, only one study [29] provided data for non-Hodgkin lymphoma (ICD-9 code 200, 202; SMR 0.49, 95% CI 0.06-1.76, based on two observed cases). The results for leukemia were based on between 0 and 8 observed cases, with one SMR suggesting a doubling of risk (SMR 2.04, 95% CI 0.88-4.03) [29], and the other estimates approximately equal to or less than 1.0. None of the studies provided data on leukemia subtype.
Radican et al. is the latest follow-up of a large cohort study of 14,000 civilian workers, 1,222 of whom were exposed to dichloromethane, employed at Hill Air Force Base in Utah for at least 1 year from 1952 to 1956, with follow-up through 2000 [26,27,33,34]. The most detailed exposure assessment was done for trichloroethylene, the primary focus of the study. Dichloromethane, one of 25 other exposures analyzed, was classified as a dichotomous exposure (ever exposed, never exposed), and dichloromethane associations were reported for only three cancer sites (non-Hodgkin lymphoma, multiple myeloma and female breast cancer) [33]. The rate ratios for non-Hodgkin lymphoma and multiple myeloma in relation to dichloromethane in men were 2.2 (95% CI 0.76-5.42) and 2.58 (95% CI 0.86-7.72), respectively. These rate ratios (particularly those for multiple myeloma) were higher than those for any of the other chemicals examined; the next highest observed rate ratios for multiple myeloma were 2.1, 2.0, and 2.0 for o-dichlorobenzene, Freon, and the "other alcohols" category, respectively. No cases of either of these cancers were observed in women with dichloromethane exposure, but the rate ratio for breast cancer in women was 2.35 (95% CI 0.98-5.65). Associations of similar magnitude or higher (rate ratios of 2.3-2.8) were also seen between breast cancer and some other exposures (Freon, solder flux, isopropyl alcohol, and 1,1,1-trichloroethane). All of these risk estimates were slightly attenuated from the estimates in the earlier examination of this cohort (with follow-up to 1990) described by Blair et al. [27].
The cohort studies with the strongest design are the two triacetate film base production cohorts (Cohort 1 in New York and the United Kingdom cohort, reported in [29] and [30], respectively. These are the cohorts with the most extensive exposure assessment information, and that considered exposure level and duration (as summarized by cumulative exposure) in the analysis. The start of eligibility for cohort entrance corresponds with the beginning of the time when the exposure potential at the work site began, and the follow-up period is relatively long (mean >25 years). Hearne and Pifer [29] also included an analysis of a "Cohort 2", based on 1,013 men employed at least one year between 1964 and 1970. There is some overlap between Cohort 1 and Cohort 2; 707 men were included in both cohorts. Cohort 2 was the focus of previous analyses by Friedlander et al. [23] and Hearne et al. [24,25]. It is not an inception cohort, and would have missed anyone leaving, possibly because of illness or death, before 1964).
Although the exposure levels in the cohorts involved in cellulose triacetate fiber production were higher than those of the film production cohorts, the duration of exposure was relatively short in the South Carolina cohort (56% < 5 years) [31]. In addition, detailed work history information was only available for 475 (37%) of the workers [28], and it is not clear how the exposure assessment was applied to workers with missing job history data. In the Maryland triacetate fiber production plant, duration of exposure was not reported [32], and was not considered in the analysis. Also, the cohort began in 1970, even though production began in 1955, and the missing personnel records made it impossible to recreate an inception cohort. The exposure assessment in the study of civilian Air Force base workers [33] allowed for only a dichotomized classification of dichloromethane and other solvents (with the exception of trichloroethylene). For the breast cancer results, the magnitude of the risk estimates for some of the other solvents (i.e., associations similar to or stronger than that seen with dichloromethane) makes it difficult to determine from the available data how much of the increased risk seen with dichloromethane could be accounted for by potential confounding. The use of mortality rather than incidence data, which is of particular concern for cancers with a relatively good survival rate, such as non-Hodgkin lymphoma, is another limitation of all of these cohort studies. a Data included in a report by Gibbs [35] that included more extensive data for both triacetate fiber cohorts.
b Calculated by summing observed and expected across exposure groups.
c Exposure classification based on a specific job, rather than a summed exposure across all jobs. The 8-hour threshold limit value before 1975 was 500 ppm , and the current OSHA Permissible Exposure Level (PEL) is 25 ppm.

Case-Control Studies
Case-control studies offer the potential for increased statistical power for assessing associations with relatively rare cancers. No case-control studies of liver or lung cancer and dichloromethane exposure were identified in the literature search. Six recent case-control studies of dichloromethane exposure and hematopoietic cancers in adults are summarized in Table 2; this Table also includes one study of dichloromethane exposure and childhood leukemia. These are all population-based studies based on incident cases identified through cancer registries, with the number of cases ranging from 180 [22] to 1,428 [17]. No association was seen in the study of adult leukemia (ICD-9 204,208) in Italy [16], although the results for the subtype of chronic lymphatic leukemia (ICD-9 204.1), based on 150 cases, suggest a possible association in the higher intensity exposure category. This subset was also included in the analysis of non-Hodgkin lymphoma cases presented in Miligi et al. [17], grouped within the subtype of small-cell lymphocytic lymphoma. The three studies of non-Hodgkin lymphoma in Germany, Italy and Connecticut observed ORs between 1.5 and 2.2 with dichloromethane exposure (ever exposed, or highest category of exposure) [17][18][19][20]. In studies that reported more detailed results, there was also some evidence of higher risk among specific subsets of disease, including small-cell lymphocytic lymphoma [17] and diffuse large B-cell lymphoma [19].
Two studies included multiple myeloma cases, but in one of these only four cases were exposed to dichloromethane so effect estimates were not presented [16]. In the study by Gold et al., 180 multiple myeloma cases were selected from two Surveillance, Epidemiology and End Results (SEER) cancer registries [22]. One set of analyses included any possible exposure among the exposed, and a second set of analyses included the low confidence exposure jobs with the unexposed group. In general, somewhat stronger associations or patterns were seen in the second analyses. The OR for ever exposed was 1.5 (95% CI 0.9-2.3) in the first analysis and 2.0 (95% CI 1.2-3.2) with the reclassification of the low confidence jobs. In the second set of analyses, a non-monotonic increasing trend was seen with duration of exposure (OR 1.0, 2.0, 1.1, 2.7, and 2.1, respectively, in the unexposed, 1-4, 5-7, 8-24, and 25-47 years duration groups, trend p = 0.01). Similar patterns were seen with cumulative exposure (trend p = 0.08) and cumulative exposure lagged by 10 years (trend p = 0.06).
In the only study of childhood acute lymphoblastic leukemia, a weak association was seen between any maternal dichloromethane exposure during the 2 years before pregnancy up to the birth (OR 1.34 [95% CI 0.54-3.34]); results were similar when limited to exposures during pregnancy [21]. Stronger associations were seen with probable or definite exposure (OR 3.22 [95% CI 0.88-11.7]) compared with possible or no exposure. The estimates for categories based on concentration and frequency were similar but there was no evidence for an increasing risk with increasing exposure level.
Each of these case-control studies obtained, using a structured interview format, detailed information about all jobs held (or, in the case of the study of childhood leukemia, jobs held in the two years before and during the pregnancy), rather than just the usual or most recent job. This information includes job and industry titles in addition to description of tasks and materials, and was used in conjunction with a job exposure matrix developed to assess intensity and probability of exposure, taking into account temporal changes in solvent use within specific types of workplaces. This exposure assessment procedure is based in large part on expert judgment, for example, by industrial hygienists familiar with specific types of workplaces, and is conducted blinded to case-control status. Five of the studies included supplemental job-specific and industry-specific questionnaire modules focusing on potential exposure to specific solvents [16][17][18]21,22]. The structured and detailed interview format reduces the potential for a differential reporting of exposures by study participants. The addition of specific questionnaire modules designed to obtain more detailed information regarding tasks and exposure conditions also improves the reliability of the assessment.
Case-control studies of dichloromethane exposure and other types of cancers are summarized in Table 3. There was little evidence of associations in the studies of breast cancer [36], pancreatic cancer [37], kidney cancer [38] or rectal cancer [39]. Two case-control studies of dichloromethane exposure and brain cancer have been conducted [40,41]. The stronger of these studies in terms of detail of exposure data was the Heineman et al. study based on cases identified using death certificates from southern Louisiana, northern New Jersey, and the Philadelphia area, with confirmation of diagnosis using hospital records [41]. Controls (frequency matched to cases by age, year of death, and study area) were randomly selected from the death certificates of white males who died of causes other than brain tumors, cerebrovascular disease, epilepsy, suicide, and homicide. Data pertaining to lifetime job history were collected from next-of-kin interviews for cases and controls. There was a trend of increasing risk with increasing probability of exposure to dichloromethane (OR = 1.0, 95% CI 0.7-1.6, for low probability; OR = 1.6, 95% CI 0.8-3.0, for medium probability; OR = 2.4, 95% CI 1.0-5.9, for high probability compared with the referent group of unexposed men; trend p-value < 0.05). The highest risk was seen with the combination of long duration (>20 years) and high intensity (or high probability) exposure. Similar results were seen in additional analyses controlling for age, study area, employment in electronics occupations and industries, and exposure to carbon tetrachloride, tetrachloroethylene, and trichloroethylene. In the analyses adjusting for these other exposures, only dichloromethane exhibited a trend with increasing probability of exposure.
As with the case-control studies of hematopoietic cancers, each of these studies used a job exposure matrix based on expert judgment for the classification of various dimensions of exposure (e.g., probability, intensity, frequency). The job exposure matrix approach is based on work in the 1990s by the National Cancer Institute [42,43]. There is a considerable range among these studies, however, in the detail and quality of the exposure information upon which the classification scheme could be applied, from death certificate occupation data [36,37,40], to interview-based information on most recent and usual jobs [38], to a lifetime job history [39,41]. Dell et al. noted the limitation of the lack of direct exposure measures in this type of assessment, and the difficulty in categorizing jobs with occasional exposure to a specific solvent [44]. The reliability of this procedure is likely to have been improved in the Heineman et al. study [41] by the use of more detailed coding of specific jobs within the industry and occupation code categories to distinguish those of particular relevance to a specific exposure (e.g., production of paint removers within the broader category of production of paints, varnishes, lacquers, enamels, and allied products) [43].

Discussion
Liver cancer has been a major focus of research on dichloromethane. No data from a case-control study of liver cancer are available pertaining to dichloromethane exposure. The cohort study with the higher exposures, the South Carolina triacetate fiber production plant, suggested an increased risk of liver cancer [28,31]. The SMR for liver and biliary tract cancer was 2.98 (95% CI 0.81-7.63) in the latest update of this cohort. This observation was based on four cases; three of these cases were biliary tract cancers, a very rare form of cancer (expected number estimated as 0.15 cases in [28]). No other cohort study has reported an increased risk of liver cancer mortality, although it should be noted that there is no other inception cohort study of a population with exposure levels similar to those of the South Carolina plant.
In the 2-year NTP inhalation exposure study in B6C3F 1 mice (exposure concentrations 0, 2,000, and 4,000 ppm), the liver tumor incidence in male mice increased from 44% in controls to 66% at 4,000 ppm; in females, the incidence rose from 6% to 83% across dose groups (both mortality-adjusted trend p-values < 0.001) [3,4]. The results of an oral exposure (drinking water) study in B6C3F 1 mice are more ambiguous, however [45,46]. There was no indication of an increased incidence of liver tumors in female mice in this study. In males, the incidence of hepatocellular adenomas or carcinomas was 18% and 20% in each of two control groups (combined incidence, 19%), increasing to 26%, 30%, 31%, and 28% in the 60, 125, 185 and 250 mg/kg-day groups, respectively. Serota et al. concluded that there was no dose-related trend and that there were no significant pair-wise differences with the controls, but other interpretations are also supported by the results [46]. Although not provided by Serota et al. [46], the statistical results are presented in the full report of the study (Hazleton Laboratories, [45]): the trend p-value was 0.058; p-values for the pair-wise comparisons with the combined control group were p = 0.071, 0.023, 0.019, and 0.036 for the 50, 125, 185, and 250 mg/kg per day dose groups, respectively. None of the chronic exposure studies in rats have shown a relation between dichloromethane exposure and liver or lung tumors [3,45,[47][48][49].
The relevance of the bioassay studies of dichloromethane in mice to humans in low-exposure scenarios has been questioned, given the high exposure conditions of the genotoxicity studies and animal bioassays, the high background rates of liver cancer in male B6C3F 1 mice, and the relatively high GST activity in mice [50]. Comparisons in mice, rats, and humans of GST enzyme activity in liver and lung tissues indicate a much higher activity in mice. In liver tissue samples, mean GST-T1 activity was 29.7, 18.2, 3.70, 1.60 nmol/min per mg protein in female mice, male mice, rats, and human GST-high conjugator groups, respectively [51]. Another potentially relevant interspecies difference is the localization of GST-T1 within cells. In the mouse, localization is seen in the nuclei of hepatocytes and bile-duct epithelium, while the rat liver does not show preferential nuclear localization of GST-T1. In human liver tissue, some hepatocytes show nuclear localization of GST-T1 and others show localization in cytoplasm, as well as in nuclei of bile duct epithelial cells [52,53].
Consideration of metabolic polymorphisms, both in CYP and GST pathways, however, is important and may modulate susceptibility, as has been shown for trichloroethylene [54]. GST-T1 is expressed at a variety of sites in addition to the liver and lung, including mammary tissue [55], brain [56], and peripheral lymphoctyes [8]. The extent of GSH conjugation and presence of polymorphic phenotypes in these tissues may be significant in understanding the sites of action of dichloromethane, and potential differences in site concordance between species.
The cohort studies pertaining to brain cancer risk are statistically underpowered given the few observed cases, 1 to 6 deaths, and their variable findings are not surprising. The Heineman et al. study, the stronger of the two brain cancer case-control studies in terms of exposure assessment strategy and confirmation of diagnosis, reported relatively strong trends (p < 0.05) wih increasing probability, duration, and intensity measures of exposure, and with the combination of high intensity (or probability) and long (>20 years) duration of exposure (OR 6.1, 95% CI 1.5-28.3) [41]. These strong trends were not seen with the cumulative exposure metric. The difference in patterns seen with cumulative compared to other exposure metrics may reflect a more valid measure of relevant exposures in the brain from the intensity measure, as suggested by the study in rats reported by Savolainen et al. in which dichloromethane levels in the brain were much higher with a higher intensity exposure scenario compared with a constant exposure period with an equivalent time-weighted average [57]. A statistically significant increased incidence of brain or central nervous system tumors has not been observed in any of the animal cancer bioassays, but a 2-year study using relatively low exposure levels (0, 50, 200, and 500 ppm) in Sprague-Dawley rats observed a total of six astrocytoma or glioma (mixed glial cell) tumors in the exposed groups (in females, the incidence was 0, 0, 0, and 2 in the 0, 50, 200, and 500 ppm exposure groups, respectively; in males, the incidence was 0, 1, 2, and 1 in the 0, 50, 200, and 500 ppm exposure groups, respectively; sample size of each group was 70 rats) [49]. These tumors are exceedingly rare in rats, and there are few examples of statistically significant trends in animal bioassays [58].
Large population-based case-control studies of incident non-Hodgkin lymphoma or multiple myeloma in Germany [18], Italy [17] and the United States [19,20,22] observed ORs between 1.5 and 2.2 with dichloromethane exposure (ever exposed, or highest category of exposure), with higher risk among specific subsets of disease. An extensive exposure assessment protocol was used in several of these studies [17,18,22], including job-specific and industry-specific questionnaire modules focusing on potential exposure to specific solvents. Thus although the available epidemiologic studies do not definitively establish an increased cancer risk in relation to dichloromethane exposure, the consistent observations of associations with non-Hodgkin lymphoma indicate that this type of effect is a concern that cannot be dismissed based on available data. Additional studies focusing on specific subtypes of hematopoietic cancers, particularly non-Hodgkin lymphoma, and multiple myeloma, are needed. Childhood leukemia differs from adult-onset hematopoietic cancers with respect to etiologically relevant time window of exposure and potential biological mechanisms [59]. Only one study of childhood leukemia and dichloromethane is available [21]. The results from this study also indicate that further research into this issue is warranted, and would build upon previous research of childhood leukemia and the broader category of parental (paternal or maternal) solvent exposure [60].
Important to any examination of a collection of epidemiology studies are the changes in diagnostic and classification criteria of human lymphoid tumors, particularly non-Hodgkin lymphoma, where classification changes are most significant [61]. A major shift in thinking occurred around 1995 with the Revised European-American Lymphoma (REAL) classification of grouping diseases of the blood and lymphatic tissues along their cell lines compared to previous approaches grouping lymphomas by a cell's physical characteristics. It was increasingly recognized that some non-Hodgkin lymphomas and corresponding lymphoid leukemias were different phases (solid and circulating) of the same disease entity [62]. Diagnostic and classification criteria may not be uniform across studies. Classification differences hinder comparison of consistency across epidemiologic studies of lymphoid cancers and dichloromethane. In addition, the misclassification of disease subtype would be expected to result in attenuated effect estimates, as it is unlikely to be systematically related to exposure. The cohort studies conducted in the 1990s used ICD-9 or ICD-8 classifications, which do not incorporate some of the concepts of contemporary knowledge of lymphomas that have been used in more recent case-control studies.
Based on this review, a number of suggestions for future epidemiology research can be made. Existing cohort studies, given their size and uneven exposure information, are unlikely to resolve questions of cancer risks and dichloromethane exposure; however, further follow-up in the New York film cohort [29] may provide limited additional information. Given the low incidence of brain, liver and hematological cancers, and the small number of women in these cohorts, more promising are case-control studies of incident cases identified from population-based cancer registries, such as the National Cancer Institute's SEER registry. Case-control studies should include robust exposure assessments such as those used in the study of Gold et al. [22] with detailed occupational information and exposure assignment referencing industry-wide surveys (see Bakke et al. [63] and Gold et al. [64] for descriptions of assessments of trichloroethylene and tetrachloroethylene, respectively), or methods that incorporate exposure measures in selected scenarios with questionnaire-based information [65]. Review of task-level information and supplemental questionnaire modules, in addition to industry and occupation codes, as was done in the more recent studies of hematological cancers, can provide valuable information that can improve the sensitivity and specificity of an exposure assessment used in case-control studies [66].
It may also be possible to combine data from the case-control studies of hematological cancers we identified, given the similarities in the exposure assessment methodologies used in the studies, to provide more robust estimates of effects in specific subtypes of these cancers, effects based on different exposure metrics, and effects adjusting for other exposures. The large sample size produced by such an aggregation of studies would be needed to examine gene-environment interactions. For dichloromethane, it is the GST-T1 metabolic pathway that is thought to result in genotoxicity and carcinogenicity. Thus higher risk would be expected among individuals with the GST-T1 +/+ genotype compared with the GST-T1 -/-(null) genotype. Effect modification (i.e., higher risk) would also be expected with genetic variants, or with co-exposures with other CYP substrates (e.g., alcohol, some other solvents), that result in lower CYP2E1 activity and thus higher GST-T1 metabolism of dichloromethane. Barry et al. reported an interaction between the TT genotype of CYP2E1 rs20760673, dichloromethane exposure, and risk of non-Hodgkin lymphoma (OR 4.42, 95% CI 2.03-9.62) [19], but the functional significance of the variant is not known.
The available data from the large cohort of civilian workers at an Air Force base indicate that a number of solvents, including dichloromethane, may be associated with breast cancer risk [33]. Future studies examining this issue should enable control of reproductive risk factors, through the choice of referent group or adjustment in the analysis. Potential confounding by co-exposures (e.g., other solvents) should also be addressed in breast cancer studies.
In summary, developments in exposure assessment for use in population settings, disease classification, and the creation of large-scale cohorts of people at higher risk for some diseases, such as the Sister Study for breast cancer [67], provide a strong foundation for future epidemiological studies of dichloromethane and other solvents. The insights generated from these studies can contribute greatly to our understanding of disease risk, and to the interpretation of animal and mechanistic studies.