jnci.oxfordjournals.org JNCI | Correspondence 277 CORRESPONDENCE Response

( 1) is intriguing but is complex and will prove difficult to confirm. We should remember that the modification of the effect of ovarian hormones on cancer risk by age of exposure is not restricted to breast cancer or to women at high risk for the disease. For example, oral contraceptives are associated with a reduced risk of endo-metrial cancer, but estrogen hormone replacement therapy is a risk factor. Oral contraceptives are also associated with a reduced risk of ovarian cancer, but again hormone replacement is a risk factor ( 2). At least three possible explanations exist for these relationships: 1) different drugs and different doses, 2) different effects of exogenous hormones in women with and without endogenous circulating estrogens, and 3) age-specific effects, that is, hormone exposure is associated with latent and prolonged reduced risk in young women but with increased short-term risk in older women. However, regarding the first point, the formulations for oral contraceptives and hormone replacement therapies are actually more similar than dissimilar, and the asso-ciations of oral contraceptives with reduced cancer risk do not appear to differ by formulation ( 3). Regarding the second point, the study of young women with early surgical menopause would be informative. Regarding the third point, if estrogen exposure were to advance the clinical presentation of a preexisting cancer or a precancerous lesion, and if the rate of tumor growth was rapid, then we would expect to see an adverse effect among current users and a rapid decline in risk after cessation of therapy. This expectatio