Cancer Causes & Control

, Volume 24, Issue 12, pp 2037–2049

Occupational exposure to methylene chloride and risk of cancer: a meta-analysis

Authors

  • Tao Liu
    • Department of Occupational HealthHuzhou Center for Disease Control and Prevention
  • Qin-er Xu
    • Department of Occupational HealthHuzhou Center for Disease Control and Prevention
  • Chuan-hui Zhang
    • Department of Occupational HealthHuzhou Center for Disease Control and Prevention
    • Department of Occupational HealthHuzhou Center for Disease Control and Prevention
Review paper

DOI: 10.1007/s10552-013-0283-0

Cite this article as:
Liu, T., Xu, Q., Zhang, C. et al. Cancer Causes Control (2013) 24: 2037. doi:10.1007/s10552-013-0283-0

Abstract

Methods

We searched MEDLINE and EMBASE for epidemiologic studies on occupational exposure to methylene chloride and risk of cancer. Estimates of study-specific odds ratios (ORs) were calculated using inverse-variance-weighted fixed-effects models and random-effects models. Statistical tests for heterogeneity were applied.

Results

We summarized data from five cohort studies and 13 case–control studies. The pooled OR for multiple myeloma was (OR 2.04; 95 % CI 1.31–3.17) in relation to occupational exposure to methylene chloride but not for non-Hodgkin’s lymphoma, leukemia, breast, bronchus, trachea and lung, brain and other CNS, biliary passages and liver, prostate, pancreas, and rectum. Furthermore, we focused on specific outcomes for non-Hodgkin’s lymphoma and multiple myeloma because of exposure misclassification. The pooling OR for non-Hodgkin’s lymphoma and multiple myeloma was 1.42 (95 % CI 1.10–1.83) with moderate degree of heterogeneity among the studies (I2 = 26.9 %, p = 0.205).

Conclusions

We found an excess risk of multiple myeloma. The non-Hodgkin’s lymphoma and leukemia that have shown weak effects should be investigated further.

Keywords

Methylene chlorideMeta-analysisCancerHydrocarbons, chlorinatedOccupational exposure

Introduction

Methylene chloride (dichloromethane) is used as a solvent in paint strippers and removers (30 %), in adhesives (20 %), as a propellant in aerosols (10 %), as a solvent in the manufacture of pharmaceuticals and drugs (10 %), in chemical processing (10 %), as a metal cleaning and finishing solvent (10 %), and in urethane foam blowing (5 %) [1]. Other uses make up the remaining 5 %. Methylene chloride is also widely used in applications such as metal cleaning and degreasing, polyurethane foam manufacturing, triacetate film and fiber manufacturing, food extraction, and aerosol propellants [2]. Current household products that may contain methylene chloride include lubricants, valve cleaners, and degreasers for automobiles, adhesive and varnish removers, paint strippers, and one household herbicide [3]. Methylene chloride is present in these products at percentages ranging from 1 to 90 %.

Methylene chloride is extensively metabolized in mammalian species through two competing pathways: (1) oxidation by the mixed function oxidase enzymes and (2) conjugation with glutathione catalyzed by glutathione-S-transferase(s) (GST). Interest in methylene chloride as a possible human carcinogen was prompted by animal bioassays, reporting an increased incidence of lung and liver tumors in mice inhaling high concentrations of methylene chloride vapor for the majority of their natural lifetime, but not rats, exposed to methylene chloride [4]. Earlier inhalation studies by Burek et al. [5] in rats and hamsters exposed at similar concentrations had shown no evidence of lung or liver tumors. The International Agency for Research on Cancer (IARC) has stated that methylene chloride is possibly carcinogenic to humans (group 2B) [6]. The evaluation was based on a combination of mechanistic data and sufficient evidence in experimental animals, while the evidence for carcinogenicity in humans was assessed as limited.

Several reviews have commented on the association between methylene chloride and risk of cancers in human [710]. However, no meta-analysis has been carried out so far. The aim of this paper is to summarize the epidemiological evidence of an association between occupational exposure to methylene chloride and risk of cancer.

Materials and methods

Search strategy

We searched for studies of workers exposed to methylene chloride published in any language using PubMed software to search MEDLINE (US National Library of Medicine, Bethesda, MD) and EMBASE. Combinations of the following keywords were used: “dichloromethane,” “methylene chloride,” “chlorinated solvent,” “solvent,” “cancer,” “carcinogen,” “mortality,” “neoplasm,” “case–control,” “cohort,” “epidemiology.” The search period was 1 January 1990 through 31 October 2012. In addition, we manually reviewed the reference lists from relevant original research and review articles and documents. Unpublished studies were not considered. All searches were carried out independently by two investigators (T.L and Q.E.X), and results were merged.

Study inclusion criteria and data collection

Studies were included if (1) they used a cohort or case–control study design; (2) the outcome of interest was clearly defined as cancer of an anatomical site; (3) they specifically identified methylene chloride exposure by reference to industrial hygiene records, individual biomarkers, job-exposure matrices, or industrial processes that involved the use of methylene chloride (cohort studies), or included questions regarding methylene chloride exposure (case–control studies); (4) they provided information that can be used to estimate the relation between methylene chloride and cancer risk in terms of odd ratio (OR), relative risk (RR), standardized mortality ratio (SMR), standardized relative risk (SRR), cumulative incidence ratio (CIR), or standardized incidence rate ratio (SIR) and their confidence intervals (CIs) or provided enough data to calculate them (raw data, p value, or variance estimate). Case–control studies that collected non-specific or general exposure information such as “solvents” or “chemicals” or that analyzed findings by a single job title were not included. We excluded case reports and case series. When risks of cancer of different anatomical sites were available in the same publication, we considered each cancer separately.

We developed a questionnaire and recorded study name, year of publication, study design, sample size (cases and controls or cohort size), methylene chloride exposure assessment, study outcomes, duration of follow-up, variables used for adjustment, or matching.

Quality assessment

The quality of the selected studies was assessed independently by two authors (T.L and Q.E.X) using the Newcastle–Ottawa Scale (NOS) [11]. The NOS uses two different tools for case–control and cohort studies and consists of three parameters of quality: selection, comparability, and exposure/outcome assessment. The NOS assigns a maximum of four points for selection, two points for comparability, and three points for exposure or outcome. For stratification purposes, studies with scores of six points or more were considered to be of moderate to good study quality. However, all studied were used for analysis, irrespective of NOS score. Any discrepancies were addressed by a joint reevaluation of the original article.

Statistical analysis

Based on the reported CIs, we estimated the standard errors [seln(effect size)] for the ln(effect size) given by the formula seln(effect size) = [ln(upper limit)−ln(lower limit)] ÷ (2*Z1−α/2), where for a 95 % CI, Z1−α/2 equals 1.96 [12]. For the studies for which the 95 % CI was not reported, we calculated them by the Fischer’s exact method using the observed deaths and expected deaths reported in the articles [13].

Overall pooled RRs estimates and their corresponding 95 % CIs were obtained using fixed-effects (Mantel–Haenszel method) and random-effects (DerSimonian and Laird method) methods [14]. Given the significant amount of heterogeneity, only the random-effects estimates are presented. Between-study heterogeneity was assessed using the Q and I2 statistics, with PQ < 0.10 or I2 > 25 % indicating significant heterogeneity [15, 16].

The meta-analyses were performed with Stata software (version 10; StataCorp LP, College Station, TX, USA) using a combination of available macros [17]. A p value <0.05 was considered statistically significant for all tests except for the heterogeneity.

Results

Search results

The MEDLINE and EMBASE search identified a total of 3,086 articles. Based on titles and abstracts, 3,004 articles were excluded at first screening because they did not meet the eligibility criteria, such as no original data, human research, methylene chloride data, cancer site, or duplicates of the same study. Full-text copies of the remaining 82 potentially relevant studies were obtained. Sixty-four studies were excluded because they did not meet the inclusion criteria. Five articles have been updated (Lanes et al. [18] updated Lanes et al. [19]. Radican et al. [20] updates Blair et al. [21], Spirtas et al. [22]. Tomenson [23] updates Tomenson et al. [24]. Hearne et al. [25] updates Hearne et al. [26]). Finally, 18 studies (five cohort studies [18, 20, 23, 25, 27] and 13 case–control studies [2840]) that assessed the association of methylene chloride with risk of cancer met criteria for inclusion in the meta-analysis (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig1_HTML.gif
Fig. 1

Flow chart of the meta-analysis

Study characteristics and study quality

We found six studies on non-Hodgkin’s lymphoma, three studies on multiple myeloma, three studies on leukemia, three studies on breast cancer, five studies on brain and other CNS cancer, five studies on bronchus, trachea, and lung, three studies on rectum cancer, five studies on pancreas cancer, three studies on prostate cancer, and three studies on biliary passages and liver.

Cohort studies were published between 1993 and 2011. Three studies originated from USA, one from UK, and one from Canada. According to the NOS, two studies (40 %) were scored six points or more, indicating a moderate study quality (Table 1). Case–control studies were published between 1994 and 2012. Eight studies originated from America, two from Italy, one from Germany, and two from Canada. According to the NOS, 10 studies (77 %) were scored six points or more, indicating a moderate to good study quality.
Table 1

Characteristics of 13 case–control studies

Source

Country

Case size (events)

Control size (events), source

Gender

Type of cancer

Main risk estimate (95 % CI)

Adjustment, matching, and restriction factors

Ascertainment of exposure

Ascertainment of outcome

Heineman et al. [28]

USA

741 (119)

741 (108), population

Male

Brain and other CNS

1.3 (0.9–1.8)

Age, study area, and employment in electronics-related occupations or industries

Job-exposure matrices

Death registry

Cantor et al. [29]

USA

33,509 (8,427)

117,794 (42,755), population

Female

Breast

0.8 (0.78–0.83)

Age, race, and socioeconomic status

Job-exposure matrices

Death registry

Cocco et al. 30]

USA

12,980 (867)

51,920 (3,239), population

Female

Brain and other CNS

1.2 (1.1–1.3)

Marital status, socioeconomic status, and age

Job-exposure matrices

Death registry

Kernan et al. [31]

USA

63,097 (10,745)

252,386 (304,738), population

Male/female

Pancreas

0.97 (0.93–1.01)

Age, metropolitan status, region of residence, and marital status, race, and gender

Job-exposure matrices

Death registry

Dumas et al. [32]

Canada

257(7)

1,259 (29), hospital

Male

Rectum

1.2 (0.5–2.8)

Age, education, respondent status, cigarette smoking, beer consumption, and body mass index

Expert assessment

Cancer registry

Miligi et al. [33]

Italy

1,135 (36)

1,246 (36), population

Male/female

Non-Hodgkin’s lymphoma

1.13 (0.70–1.81)

Age, sex, education, and area

Job-exposure matrices

Cancer registry

Seidler et al. [34]

Germany

710 (20)

710 (29), population

 

Non-Hodgkin’s lymphoma

0.8 (0.5–1.5)

Age, sex, region, smoking, and alcohol consumption

Expert assessment

Cancer registry

Costantini et al. [38]

Italy

586 (9)

1,278 (29), population

Male/female

Leukemia

0.7 (0.3–1.5)

Age, gender, education, and area

Job-exposure matrices

Cancer register

Wang et al. [35]

USA

601(52)

717 (46), population

Female

Non-Hodgkin’s lymphoma

1.5 (1.0–2.3)

Age, family history of hematopoietic cancer, alcohol consumption and race, education, smoking, medical history of immune-related disease, and annual family income

Job-exposure matrices

Cancer register

Barry et al. [36]

USA

518 (47)

597 (33), population

Female

Non-Hodgkin’s lymphoma

1.69 (1.06–2.69)

Age and race

Job-exposure matrices

Cancer register

Gold et al. [37]

USA

180 (37)

481 (54), hospital

Male/female

Multiple myeloma

2.0 (1.2–3.2)

Age, race, study site, gender, and years of education

Job-exposure matrices

Cancer register

Vizcaya et al. [40]

Canada

2,016 (32)

2,001 (27), population

Male/female

Lung

0.9 (0.3–2.6)

Age, smoking habit, educational attainment, socioeconomic status, ethnicity, and exposure to eight known carcinogens

Expert assessment

Cancer register

Neta et al. [39]

USA

686 (21)

799(53), hospital

Male/female

Brain and other CNS

0.5 (0.3–0.9)

Age, race, sex, hospital site, and proximity of residence to the hospital

Job-exposure matrices

Cancer register

The exposure assessment of ten studies was based on a job-exposure matrix. Eight studies used self-reported exposure, occupational histories, and cumulative exposure.

Statistical analysis and pooling

Non-Hodgkin’s lymphoma

Figure 2 shows the OR and 95 % CI from the individual studies and the pooled OR based on a random-effects model. The average pooled estimate for non-Hodgkin’s lymphoma among methylene chloride exposed workers was 1.28 (95 % CI 0.96–1.70), with a moderate degree of heterogeneity among the studies (I2 = 29.1 %, p = 0.217). There was no significant association when evaluating cohort studies (OR 1.19; 95 % CI 0.31–4.55) and case–control studies (OR 1.27; 95 % CI 0.93–1.72) separately (Table 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig2_HTML.gif
Fig. 2

OR and 95 % CIs of non-Hodgkin’s lymphoma associated with occupational exposure to methylene chloride: random-effects meta-analysis

Table 2

Characteristics of 5 cohort studies

Source

Country

Total cohort

Gender

Year of follow-up

Type of cancer

Main risk estimate (95 % CI)

Adjustment, matching, and restriction factors

Ascertainment of exposure

Ascertainment of outcome

Lanes et al. [18]

USA

1,271

Male/female

14

Breast

0.54 (0.11–1.57)

Age, race, gender, and calendar period

Self-reported occupational histories

Death register

Bronchus, trachea and lung

0.8 (0.43–1.37)

Pancreas

0.83 (0.10–2.99)

Biliary passages and liver

2.98 (0.81–7.63)

Gibbs et al. [27]

USA

3,211

Male/female

20

Prostate

0.76 (0.23–2.55)

Age, race, gender, and calendar period, social, economic, ethnic, and cultural factors related to disease

Self-reported occupational histories

Death register

Bronchus, trachea and lung

0.79 (0.34–1.87)

Hearne et al. [25]

USA

1,311

Male

25

Non-Hodgkin’s lymphoma

0.49 (0.06–1.78)

Age, calendar year, and time-from-first exposure

Self-reported occupational histories

Death register

Multiple myeloma

0.68 (0.01–3.79)

Brain and other CNS

2.16 (0.79–4.69)

Bronchus, trachea, and lung

0.75 (0.49–1.09)

Rectum

1.15 (0.64–1.90)

Pancreas

0.92 (0.30–2.14)

Biliary passages and liver

0.42 (0.01–2.36)

Leukemia

2.04 (0.88–4.03)

Prostate

1.0 (0.40–2.07)

Radican et al. [20]

USA

1,222

Male/female

35–45

Non-Hodgkin’s lymphoma

2.02 (0.76–5.42)

Age, gender, calendar time

Self-reported occupational histories

Death register

Multiple myeloma

2.58 (0.86–7.72)

Breast

2.35 (0.98–5.65)

Tomenson et al. [23]

United Kingdom

1,785

Male

62

Leukemia

1.06 (0.12–9.10)

Unspecified

Self-reported occupational histories

Death register

Brain and other CNS

1.70 (0.21–13.63)

Bronchus, trachea, and lung

1.92 (0.58–6.38)

Rectum

0.42 (0.04–4.68)

Pancreas

0.53 (0.10–2.73)

Prostate

0.63 (0.20–1.97)

Multiple myeloma

The fixed-effects OR for multiple myeloma was elevated at 2.04 (95 % CI 1.31–3.17, I2 = 0.0 %, p > 0.1, Fig. 3). No association was found when evaluating cohort studies. Furthermore, according to the Glass et al. study [41], we focus on specific outcomes for non-Hodgkin’s lymphoma and multiple myeloma because of exposure misclassification. The pooling OR for non-Hodgkin’s lymphoma and multiple myeloma were 1.42 (95 % CI 1.10–1.83) with moderate degree of heterogeneity among the studies (I2 = 26.9 %, p = 0.205), 1.38 (95 % CI 1.03–1.85) for case–control studies, 1.69 (95 % CI 0.88–3.26) for cohort studies (data not shown).
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig3_HTML.gif
Fig. 3

OR and 95 % CIs of multiple myeloma associated with occupational exposure to methylene chloride: fixed-effects meta-analysis

Leukemia

The OR for leukemia was 1.19 (95 % CI 0.54–2.65), with a moderate degree of heterogeneity among the studies (I2 = 44.5 %, p = 0.165). There was no significant association when evaluating cohort studies (OR 1.90; 95 % CI 0.93–3.89) (Fig. 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig4_HTML.gif
Fig. 4

OR and 95 % CIs of leukemia associated with occupational exposure to methylene chloride: random-effects meta-analysis

Breast cancer

Figure 5 shows the OR and 95 % CI from the individual studies and the pooled OR based on a random-effects model. The OR for breast cancer was 1.02 (95 % CI 0.48–2.16), with a moderate degree of heterogeneity among the studies (I2 = 67.4 %, p = 0.046).
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig5_HTML.gif
Fig. 5

OR and 95 % CIs of breast cancer associated with occupational exposure to methylene chloride: random-effects meta-analysis

Brain and other CNS cancer

The random-effects OR for brain and CNS cancer was 1.11 (95 % CI 0.78–1.58, I2 = 65.8 %, p < 0.1, Fig. 6). No association was found when evaluating cohort studies and case–control studies separately.
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig6_HTML.gif
Fig. 6

OR and 95 % CIs of brain and other CNS cancer associated with occupational exposure to methylene chloride: random-effects meta-analysis

Bronchus, trachea, and lung cancer

The fixed-effects OR for bronchus, trachea, and lung cancer was 0.82 (95 % CI 0.62–1.09, I2 = 0.0 %, p > 0.1, Fig. 7). No association was found when evaluating cohort studies.
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig7_HTML.gif
Fig. 7

OR and 95 % CIs of bronchus, trachea and lung cancer associated with occupational exposure to methylene chloride: fixed-effects meta-analysis

Biliary passages and liver cancer

Figure 8 shows the OR and 95 % CI from the individual studies and the pooled OR based on a random-effects model. The OR for biliary passages and liver cancer was 1.69 (95 % CI 0.30–9.64), with a moderate degree of heterogeneity among the studies (I2 = 40.9 %, p = 0.193).
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig8_HTML.gif
Fig. 8

OR and 95 % CIs of biliary passages and liver cancer associated with occupational exposure to methylene chloride: random-effects meta-analysis

Rectum cancer

The fixed-effects OR for rectum cancer was 1.12 (95 % CI 0.71–1.76, I2 = 0.0 %, p > 0.1, Fig. 9). No association was found when evaluating cohort studies.
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig9_HTML.gif
Fig. 9

OR and 95 % CIs of rectum cancer associated with occupational exposure to methylene chloride: fixed-effects meta-analysis

Pancreas cancer

The fixed-effects OR for pancreas cancer was 0.97 (95 % CI 0.93–1.01, I2 = 0.0 %, p > 0.1, Fig. 10) among all the studies. The OR for cohort studies was 0.80 (95 % CI 0.38–1.71).
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig10_HTML.gif
Fig. 10

OR and 95 % CIs of pancreas cancer associated with occupational exposure to methylene chloride: fixed-effects meta-analysis

Prostate cancer

The fixed-effects OR for prostate cancer was 0.76 (95 % CI 0.51–1.14, I2 = 0.0 %, p > 0.1, Fig. 11).
https://static-content.springer.com/image/art%3A10.1007%2Fs10552-013-0283-0/MediaObjects/10552_2013_283_Fig11_HTML.gif
Fig. 11

OR and 95 % CIs of prostate cancer associated with occupational exposure to methylene chloride: fixed-effects meta-analysis

Discussion

The association between occupational exposure to methylene chloride and risk of cancer has been assessed. Our results demonstrate an increase in the pooled estimate (OR 2.04; 95 % CI 1.31–3.17) for multiple myeloma in relation to occupational exposure to methylene chloride but not for non-Hodgkin’s lymphoma, leukemia, breast, bronchus, trachea and lung, brain and other CNS, biliary passages and liver, prostate, pancreas, and rectum. Furthermore, we focus on specific outcomes for non-Hodgkin’s lymphoma and multiple myeloma because of exposure misclassification, according to the Glass et al. study [41]. The pooling OR for non-Hodgkin’s lymphoma and multiple myeloma were 1.42 (95 % CI 1.10–1.83) with moderate degree of heterogeneity among the studies (I2 = 26.9 %, p = 0.205), 1.38 (95 % CI 1.03–1.85) for case–control studies, and 1.69 (95 % CI 0.88–3.26) for cohort studies (data not shown).

Cytochrome P450 2E1 (CYP2E1) and glutathione-S-transferase theta 1 (GSTT1) are two major metabolic pathways in the biotransformation of methylene chloride [36, 42]. Oxidation by P450 pathway via formyl chloride finally leads to the formation of CO, induces cellular toxicity [43]. Barry et al. [36] indicated that CYP2E1 activity was expected to influence the toxicity of methylene chloride. The OR for non-Hodgkin’s lymphoma among women TT for rs2070673 in the CYP2E1 was 4.42 (95 % CI 2.03–9.62). However, Olvera-Bello et al. [44] suggested that methylene chloride was highly cytotoxic in human peripheral blood mononuclear cells, even at doses within the safety range, which could potentially cause an increase in the accumulation of reactive oxygen species via GSTT1 pathway, leading to cell cycle regulation and the mitogenic response.

Exposure misclassification has been a critical limitation among epidemiologic studies. Misclassification can occur with all study designs, and it can concern any of basic elements of an epidemiologic study: the outcome, exposure, or the covariates (confounder, modifiers). From the point of view of exposure assessment, exposure misclassification is of major concern [45]. Most studies assess occupational exposure to specific solvents by means of the job-exposure matrices, self-reported occupational histories, and expert assessment of exposure [46]. Ten of eighteen studies in this meta-analysis have relied on some type of semiquantitative job-exposure matrix (JEM), and sources of data in these studies were based on death certificates or cancer registries. The most recent occupation and type of industry held by the decedent have been reported on death certificates. But no information on most known (or suspected) risk factors, duration of employment, job title, and other occupations were recorded, such as age at menarche, age at first birth, age at menopause, history of benign breast disease, and family history for breast cancer. Thus, the exposure assessment based on usual occupation and type of industry may not accurately reflect the exposures related to the cause of death.

For the occupation data from cancer registries, information collected directly from study participants rather than a proxy respondent is generally a more reliable source of information for job histories. Six studies benefited from an expert’s translation of lifetime occupational history into specific exposures. Nevertheless, retrospective exposure assessment procedure was not based on active measurement, and concentration of exposure could not be estimated in absolute terms. Moreover, the complexity of use of solvents that have been used interchangeably and at times together has made the evaluation of specific exposure difficult. But because of blindly with respect to disease status and controlled nature of the computer-assisted personal interview, the error involved in applying a JEM is non-differential with respect to case/control status and therefore can lead to some attenuation of true odds ratios.

Potential known or unknown confounders may lead to biased results. It has shown that non-Hodgkin’s lymphoma among women exposed to methylene chloride is restricted to the specific genotype/phenotype of CYP2E1, an enzyme involved in the activating/detoxification of methylene chloride [36]. This and other genetic factors are potential effect modifiers that were not addressed in the individual studies. Furthermore, four of the studies explicitly adjusted for smoking [32, 34, 35, 40], although some authors decided not to adjust for it after having ruled out confounding in their data. Thus, lack of control for smoking may have upwardly biased the relative risk estimates in the cohort studies that used external comparisons, but may not have affected case–control studies or cohort studies using internal comparison groups.

Our study identified third implications for the future. First, our meta-analysis provides new information for a carcinogenic effect of methylene chloride in humans, which contributes to the IARC classification of methylene chloride. Second, it is important to establish an increased cancer risk in relation to methylene chloride exposure, particularly non-Hodgkin’s lymphoma and multiple myeloma, as classification differences and independent control groups hinder comparison of consistency across epidemiologic studies of lymphoid cancers and methylene chloride. Third, more than two-thirds of the studies were case–control studies, given semiquantitative exposure assessment, and the cohort studies were of small sizes and uneven exposure information; therefore, the future researches designed with large population-based case–control or cohort studies are needed in order to make reliable inferences on cause–effect mechanisms.

In conclusion, we found an excess risk of multiple myeloma. The non-Hodgkin’s lymphoma and leukemia that have shown weak effects should be investigated further.

Conflict of interest

None.

Copyright information

© Springer Science+Business Media Dordrecht 2013