Trichloroethylene and cancer: systematic and quantitative review of epidemiologic evidence for identifying hazards

Abstract

We conducted a meta-analysis focusing on studies with high potential for trichloroethylene (TCE) exposure to provide quantitative evaluations of the evidence for associations between TCE exposure and kidney, liver, and non-Hodgkin lymphoma (NHL) cancers. A systematic review documenting essential design features, exposure assessment approaches, statistical analyses, and potential sources of confounding and bias identified twenty-four cohort and case-control studies on TCE and the three cancers of interest with high potential for exposure, including five recently published case-control studies of kidney cancer or NHL. Fixed- and random-effects models were fitted to the data on overall exposure and on the highest exposure group. Sensitivity analyses examined the influence of individual studies and of alternative risk estimate selections. For overall TCE exposure and kidney cancer, the summary relative risk (RRm) estimate from the random effects model was 1.27 (95% CI: 1.13, 1.43), with a higher RRm for the highest exposure groups (1.58, 95% CI: 1.28, 1.96). The RRm estimates were not overly sensitive to alternative risk estimate selections or to removal of an individual study. There was no apparent heterogeneity or publication bias. For NHL, RRm estimates for overall exposure and for the highest exposure group, respectively, were 1.23 (95% CI: 1.07, 1.42) and 1.43 (95% CI: 1.13, 1.82) and, for liver cancer, 1.29 (95% CI: 1.07, 1.56) and 1.28 (95% CI: 0.93, 1.77). Our findings provide strong support for a causal association between TCE exposure and kidney cancer. The support is strong but less robust for NHL, where issues of study heterogeneity, potential publication bias, and weaker exposure-response results contribute uncertainty, and more limited for liver cancer, where only cohort studies with small numbers of cases were available.

Keywords: NHL; kidney cancer; liver cancer; meta-analysis; occupational exposure; trichloroethylene.

Figure 1

Forest plot of overall TCE exposure and kidney cancer from random-effects meta-analysis. Individual study results are plotted with 95% confidence intervals. Symbol sizes reflect relative weights of the 15 cohort and case-control studies.

Figure 2

Forest plot of highest TCE exposure group and kidney cancer from random-effects meta-analysis. Individual study results are plotted with 95% confidence intervals. A risk estimate of 1.0 is assigned for highest exposure in Anttila et al. [38], Axelson et al. [37] and Hansen et al. [39] to account for presumed reporting bias. Symbol sizes reflect relative weights of the 13 cohort and case-control studies.

Figure 3

Forest plot of overall TCE exposure and liver cancer from random-effects meta-analysis. Individual study results are plotted with 95% confidence intervals. Symbol sizes reflect relative weights of the 9 cohort studies.

Figure 4

Forest plot of highest TCE exposure group and liver cancer from random-effects meta-analysis. Individual study results are plotted with 95% confidence intervals. A risk estimate of 1.0 is assigned for highest exposure in Hansen et al. [39] and Zhao et al. [25] to account for presumed reporting bias. Symbol sizes reflect relative weights of the 8 cohort studies.

Figure 5

Forest plot of overall TCE exposure and non-Hodgkin lymphoma from random-effects meta-analysis. Individual study results are plotted with 95% confidence intervals. Symbol sizes reflect relative weights of the 17 cohort and case-control studies.

Figure 6

Forest plot of highest TCE exposure group and non-Hodgkin lymphoma from random-effects meta-analysis. Individual study results are plotted with 95% confidence intervals. Symbol sizes reflect relative weights of the 13 cohort and case-control studies.