Epidemiological approaches in the investigation of environmental causes of cancer: the case of dioxins and water disinfection by-products

Abstract

I will refer in this paper to difficulties in research in environmental causes of cancer using as examples research on dioxins and on drinking water disinfection by-products (DBPs) that have created considerable controversy in the scientific and wider community. Dioxins are highly toxic chemicals that are animal carcinogens. For many years, evaluation of the carcinogenicity of dioxins in humans was based on case-control or registry based studies. The development of methods to measure dioxins in blood indicated that these studies suffered from extreme exposure misclassification. The conduct of large cohort studies of workers with widely contrasted exposures together with the use of biomarkers and models for exposure assessment, led to convincing evidence on the carcinogenicity of dioxins in humans. The high toxicity of a few dioxin congeners, the availability of a scheme to characterize the toxicity of a mixture of dioxins and related compounds and the long half-life of these compounds facilitated epidemiological research. Contrary to dioxins, trihalomethanes (THMs) and most of the hundreds of DBPs in drinking water are chemicals of low toxicity. For more than 15 years, the main evidence on the carcinogenicity of DBPs was through ecological or death certificate studies. More recent studies based on individual assessment confirmed increases in bladder cancer risk. However even those studies ignored the toxicological evidence on the importance of routes of exposure to DBPs other than ingestion and, probably, underestimated the risk. Persistence of weak study designs together with delays in advanced exposure assessment models led to delays in confirming early evidence on the carcinogenicity of DBPs. The evaluation of only a few chemicals when exposure is to a complex mixture remains a major problem in exposure assessment for DBPs. The success of epidemiological studies in identifying increased risks lies primarily on the wide contrast of exposure to DBPs in the general population that overcomes the significant exposure misclassification. Exposure assessment has been the Achilles heel for studies on dioxins and DBPs and cancer. The combination of powerful study designs, advanced exposure assessment together with a better understanding of mechanisms of disease and the use of biomarkers of exposure, led to the strengthening of the epidemiological evidence.

Figure 1

Time trends in publications in occupational, environmental and genetic cancer epidemiology 1995-2009 (no limits, hits retrieved from PubMed)

Figure 2

Odds Ratios for soft-tissue sarcoma and exposure to phenoxy herbicides, chlorophenols and dioxins. 15 case-control studies (adapted from E Johnson 1993, ref. [34])

Figure 3

Concentration of TCDD in serum of New Zealand professional sprayers (applicators) in relation to total months spent spraying 2, 4, 5-T (Smith et al, JNCI. 1992, ref. [15])

Figure 4

Total cancer mortality and urbanization for parishes without drinking water from the Mississippi (closed circles) and those with some al all water from the Mississippi (stars). The latter are expected to have higher exposure to THMs (from Page et al 1976, ref. [19])

Figure 5

Pooled analysis of case-control studies on bladder cancer and exposure to Trihalomethanes in drinking water. Log odds ratio for bladder cancer and average exposure to THMs (mg/L) using natural splines (3df), both sexes (from Villanueva et al 2004, ref. [22])