Research Report on the Risk Assessment of Mixtures of Disinfection By-Products (DBPS) in Drinking Water

This report presents a number of manuscripts and progress reports on statistical and biological research pertaining to the health risk assessment of simple DBP mixtures. Research has been conducted to generate efficient experimental designs to test specific mixtures for departures from additivity by building joint, dose-response threshold additivity models using single chemical data. To test the usefulness of these statistical models, EPA laboratory scientists generated health effects data for hepatotoxicity and nephrotoxicity in female CD-1 mice exposed to mixtures of DBPs, specifically the four trihalomethanes (THMs). Mixtures toxicity data were also developed for the THMs on carcinogenic, developmental, and reproductive effects in Japanese medaka (Oryzias latipes) fish embryos. In addition, risk assessment methodologies were proposed to characterize the health risks from exposure to mixtures of DBPs across various treatment scenarios. This collaborative research was conducted to address multiple statistical, toxicologic, and risk assessment issues including the testing and refinement of quantitative methods, and the development of data on the toxicity and potential interactions of mixtures of DBPs. Results are briefly summarize below:

1) The THMs have been investigated as priority drinking water chemicals whose toxicity and joint action is important to understanding potential human health risks. In CD-1 mice experiments, real world mixing ratios of THMs were tested, comparing an ozonation process and a chlorination process and showing no difference in the toxicity. However, further testing is needed using chemicals other than the THMs, such as bromate, that are uniquely or more typically produced by a certain treatment process. Continued development of these data will aid in the estimation of health risks from exposure to multiple DBPs for specific treatment processes.

2) Evidence of additivity at relatively low doses has been produced by applying the threshold additivity model to laboratory data in CD-1 mice and in the medaka. Such evidence is important to risk assessment because it supports the use of assumptions of additivity as default procedures in the risk assessment of multiple DBPs at low doses.

3) Because hundreds of DBPs exist that are potential candidates for toxicologic testing, such efficient experimental designs are necessary, and inexpensive bioassays for the screening of combinations of chemicals are essential.

4) Use of the medaka fish has been explored for screening effects in mixture studies, such as cancer, developmental defects, and systemic effects (e.g., liver, kidney, thyroid, and cardiac effects). Results for the medaka are preliminary, as the protocols for the testing procedures are still being developed; however, use of the medaka as a screening tool is promising. This is particularly true because cancer and developmental effects, the endpoints of most concern in the drinking water epidemiologic literature, are relevant endpoints observed in the medaka experiments.

5) The binary experiments underway in the CD-1 mice have been designed not only for use in the development of threshold additivity models, but also to test the proportional response addition model and the interaction-based hazard index approach. Both of these methods require experimental data to test their relevance for use in human health risk assessment and to refine the models and parameters being proposed. Some of this testing has begun using the binary information and whole mixture data currently available, with results anticipated when all the laboratory studies are completed.


U.S. EPA. Research Report on the Risk Assessment of Mixtures of Disinfection By-Products (DBPS) in Drinking Water. U.S. Environmental Protection Agency, Washington, DC, EPA/600/R-03/039, 1999.