Re: tamoxifen prevention of breast cancer: an instance of the fingerpost.

In their commentary on tamoxifen, Lippman and Brown [(1), p. 1813] cite figures from Fisher (2) and state that “these figures demonstrate that tamoxifen prevention potentially could impart a substantial net benefit to the public health.” In light of the article presented by Gail et al. (3) in the same issue of the Journal and some of our own calculations given below, we wish to offer a different perspective. According to Fisher’s cited commentary (2), approximately 29 million women in the United States would meet the eligibility criteria for the Breast Cancer Prevention Trial (BCPT) (4), and in this group of 29 million, 1.4 million cases of breast cancer might be expected during the course of 5 years, of which 700 000 would potentially be prevented by prophylactic use of tamoxifen. First, there are not 1.4 million cases of breast cancer in the entire population of U.S. women over a 5-year period. For the past 5 years, there have been between 175 000 and 185 000 cases of breast cancer per year (5). Second, not all breast cancer cases are going to occur among women eligible for tamoxifen chemoprevention. Women were eligible for the BCPT if they fell into one of the following categories (4): aged 35–59 years with a 5-year risk of at least 1.67% (the 5-year risk of the “average” 60-year-old U.S. woman) according to the model of Gail et al. (6), age greater than or equal to 60 years, or history of lobular carcinoma in situ. Because of concern over whether all healthy women 60 years of age and older should be advised to consider tamoxifen (7), the U.S. Food and Drug Administration (FDA) indications for tamoxifen chemoprevention are slightly different from the BCPT eligibility criteria: aged 35 years of age or older and a 5-year risk of at least 1.67%. In the Nurses’ Health Study cohort, only 30% of women aged 60–69 years had a 5-year risk of 1.67% or higher, as estimated by the model of Gail et al. In this cohort, among women aged 35–69 years who developed breast cancer during a 5-year period, fewer than one-quarter of cases occurred in women who would have been eligible for tamoxifen under the FDA indications. For several important reasons, therefore, it is misleading to state that 50% of 1.4 million cases of breast cancer during a 5-year period could be prevented by tamoxifen. More important, the statement by Lippman and Brown (1) that there might be substantial public health gains from prophylactic tamoxifen use disregards the central message of the analysis by Gail et al. (3). This analysis shows that it is unlikely that chemoprevention with tamoxifen will lead to substantial public health benefits. Indeed, the calculations by Gail et al. (3) indicate that widespread tamoxifen use would lead to net public health losses precisely in the very large segment of the general population that contributes the bulk of breast cancer cases, i.e., women aged 50–79 years (both black and white) with intact uteri and with breast cancer risk that is not very different from “average.” Benefit outweighs harm for white women aged 50–59 years and 60–69 years only at very high levels of breast cancer risk (ù3.5% for women 50–59 years of age and ù6.5% for women 60–69 years of age). The proportion of white women in these age-risk groups is exceedingly small in the general population, unlike their representation in the BCPT. In the Nurses’ Health Study cohort, less than 1% of women aged 50–59 years had a 5-year breast cancer risk greater than or equal to 3.5% as estimated by the Gail et al. model; less than 0.5% of women 60– 69 years of age had a 5-year risk greater than or equal to 6.5%. Less than 4% of all breast cancer cases that occurred in the cohort over a 5-year period arose among these high-risk women. For black women, who face higher risks of stroke, pulmonary embolism, and deep-