Using epidemiological studies to check the consistency of the cancer riskspredicted by high-dose animal experiments: A methodological review

Epidemiological studies have been cited in the literature as evidence both for and against the human cancer risks predicted by high-exposure rodent st udies. However, there has been little overall consistency in the ways that these animal-to-human comparisons have been made. This review examines some examples of these types of comparisons and describes the methods and techn iques used by different investigators. Eleven "key decision areas" that nee d to be addressed are identified and recommendations for consistent, logica l, and statistically appropriate approaches that might be taken to standard ize the process are provided. In general, it is suggested that investigator s provide the most useful information when they use logical, transparent, a nd statistically valid comparisons to pursue limited and focused objectives , such as directly testing the validity of an existing regulatory guidance value. Other recommendations include selecting biologically plausible extra polative models that fit the data and drawing conclusions that are consiste nt with the study results and objectives.