lt is generally considered that the initiating event in the process of chemical carcinogenesis is the induction of a mutation in the genetic material (DNA) of somatie cells (i.e., cells other than ova or sperm). Because this genotoxic mechanism theoretically does not have a threshold, there is a probability of harm at any level of exposure. Therefore, development of a TDI is considered inappropriate for nonthreshold chemicals, and mathematical low-dose risk extrapolation is applied. On the other hand, some carcinogens acting through an indirect mechanism are capable of producing tumors in animals or humans without exerting genotoxic activity. lt is generally believed that a threshold dose exists for these nongenotoxic carcinogens.

In order to make the distinction regarding the underlying mechanism of carcinogenicity, each compound that has been shown to be a carcinogen was evaluated on a case-by-case basis, taking into account the evidence of genotoxicity, the range of species affected, and the relevance to humans of the tumors observed in experimental animals.

In establishing GVs for drinking water quality, IARC categorization of chemical carcinogenicity was used. IARC assigns chemical substances with respect to their potential carcinogenicity into the following categories:

Group 1: The agent is carcinogenic to humans.
Group 2A: The agent is probably carcinogenic to humans.
Group 2B: The agent is possibly carcinogenic to humans.
Group 3: The agent is not classifiable as to its carcinogenicity to humans.
Group 4: The agent is probably not carcinogenic to humans.

For compounds that show nonthreshold effects, such as genotoxic carcinogens, the linearized multistage model, a conservative model, was generally adopted for use in developing GVs. Estimated risks are based on a 60-kg person drinking 2 L water/day over a 70-year life span. Cancer risk estimates are based on body weight extrapolation (i.e., a surface area-to-body weight correction was not applied since excessively conservative estimates would result).

The GV is the concentration in drinking water corresponding to an estimated risk of one additional cancer case per 100,000 (10 3) people exposed over a lifetime, using the upper 95% confidence limit of the dose-response curve as the zero dose or no exposure is approached. The excess lifetime cancer risk of 10-1 is an arbitrarily selected value, and it is up to each country to determine its own acceptable excess cancer risk. Concentrations associated with estimated excess lifetime cancer risk of 10' or lo 6 can be calculated by multiplying or dividing, respectively, the GV by 10. In cases where the concentration associated with a 10-5 lifetime excess risk is not achievable because of inadequate analytical or treatment technology, a provisional GV is set at a higher practicable level, and the estimated associated cancer risk is presentes.

lt should be emphasized that GVs for carcinogenic substances computed using mathematical models must be considered, at best, as rough estimates of cancer risk. Uncertainties involved may be two orders of magnitude (i.e., from 0.1 to 10 times the GV). However, the linearized multistage model tends to be conservative in that at low doses it will give higher estimates of cancer risk than will many other models (NRC 1980); the actual risk might be zero. Moderate short-term exposure to levels exceeding the GV for carcinogens does not significantly affect the risk.

A summary of the risk assessment of selected disinfectants and DBPs is given in Table 3.
Table 3. Risk assessment of selected disinfectants and disinfection by-products