Reproductive factors and colorectal cancer

Descriptive and analytical epidemiology have suggested that cancer of the colorectum may have reproductive correlates similar to those of breast cancer (e.g., protection by parity and early age at first birth), but the evidence is still controversial. We therefore reviewed published data from 15 case-control investigations, two cohort studies, and one cancer registry-based study from seven different countries. With reference to parity, statistically significant protection for colorectal cancer was found in three case-control studies; in four other studies, significant inverse relationships of parity were observed with colon cancer, but not with rectal cancer. Among the remaining 12 studies, relative risks below unity for parous or multiparous women were observed in four. There was no appreciable trend in risk in four others' in two, there was nonsignificant increased risk with parity; and in one, a significant increased risk. Information on age at first birth was available from 12 studies. Three reported significant trends of increased risk with increasing age at first birth-one found a direct association of borderline significance; six indicated no evidence of association; and two reported an inverse trend in risk of borderline significance. Findings on age at menarche were inconsistent and mostly negative, although an inverse significant association was reported, especially regarding colon cancer, in one investigation. In all the six studies which provided information on age at menopause, there was a hint of protection, although nonsignificant, for women who underwent natural menopause at an older age. Two studies reported a direct association of colorectal cancer with use of oral contraceptives, and another showed an inverse relationship with the use of menopausal estrogens. Most work on exogenous estrogens, however, was negative. Despite some discrepancies, possibly attributable to differences between studies or chance, it seems reasonable to conclude, on the basis of published material, that multiparity and early age at first birth may exert some protection on colorectal cancer. This pattern of risk may be similar to-though less well defined and generally less consistent than-the associations with breast cancer. More precise quantification of this effect, as well as better understanding of the potential relationships with separate subsites of the large bowel and analysis of interactions, may require a formal pooling exercise based on the original data sets.