A theory of the cancer age-specific incidence data based on extreme value distributions

The incidence of cancers varies with age, if normalized this is called the age-specific incidence. A mathematical model that describes this variation should provide a better understanding of how cancers develop. We suggest that the age-specific incidence should follow an extreme value distribution, based on three widely accepted assumptions: (1) a tumor develops from a single cell, (2) many potential tumor progenitor cells exist in a tissue, and (3) cancer is diagnosed when the first of these many potential tumor cells develops into a tumor. We tested this by comparing the predicted distribution to the age-specific incidence data for colon and prostate carcinomas collected by the Surveillance, Epidemiology and End Results network of 17 cancer registries. We found that colon carcinoma age-specific incidence data is consistent with an extreme value distribution, while prostate carcinomas age-specific incidence data generally follows the distribution. This model indicates that both colon and prostate carcinomas only occur in a subset of the population (22% for prostate and 13.5% for colon.) Because of their very general nature, extreme value distributions might be applicable to understanding other chronic human diseases.