A methodology for assessing carcinogenic hazards of chemicals

Assessment of carcinogenic hazards of substances involves evaluation and classification of qualitative and quantitative data, including the attendant uncertainty. Recently proposed EPA guidelines represent a significant step towards a uniform carcinogenic risk assessment procedure and its appropriate use. The Methodology for Assessing Carcinogenic Hazards of Chemicals proposed here shares common characteristics with the EPA guidelines but it also focuses on other aspects of carcinogen assessment, not addressed by the guidelines, such as: folding together quantitative and qualitative data into a combined hazard index; classification of limited, incomplete or flawed qualitative data; utilization of quantitative risk estimates based on weak qualitative evidence. The elements are used to assess carcinogenic hazards of chemicals, qualitative and quantitative. The qualitative assessment consists of stratification into one of five categories of weight-of-evidence, whereas quantitative assessment consists of stratification into one of four potency categories on the basis of carcinogenic unit risk estimate. Carcinogenic hazard index, codified in letters A to E, is derived from the quantitative and qualitative elements of assessment in a two-dimensional matrix. The matrix is designed so that weaker evidence for carcinogenicity requires a correspondingly higher unit risk value for a given score. Assessment of 100 chemicals using our methodology shows a good distribution of scores from A to E. Group B, which can be described as moderately hazardous, is most frequently represented. The methodology represents a mixture of two possible approaches to assessing hazards of carcinogens: one which gives the highest rank to those likely to produce the greatest risk, the other which ranks highest those most likely to produce cancer in humans. By using this combined approach, it is ensured that a chemical hazardous on either count is not missed. The methodology also makes maximum use of all available data without over-interpreting some highly uncertain procedures such as quantitative risk estimation or extrapolation of animal response to humans.