Diet and cancer. Evidence from associations of multiple primary cancers in the SEER program.

The occurrence of multiple primary cancers may reflect common etiologic factors. We investigated the extent to which the diet and cancer hypothesis was supported by data from the Surveillance, Epidemiology, and End Results (SEER) Program on multiple primary associations. Cancers of the colon/rectum and prostate in men, and those of the breast, colon/rectum, and uterine corpus in women, were hypothesized a priori to be diet-related cancers. Of the eight multiple primary associations among diet-related cancers that were possible in men and women, relative risks (RR) of a second diet-related primary cancer developing after a first diet-related primary ranged from 1.06 to 1.43. The lower bound of the 99% confidence intervals (CI) for five of these associations exceeded 1.00, and fell between 0.95 and 0.99 for the other three associations. The observed multiple primary associations were compatible with the existence of common etiologic dietary elements. However, hormonal, immunologic, and medical care factors shared by these malignancies must be considered as alternative explanations for these findings.