Should women with a family history of breast cancer avoid use of oral contraceptives?

Abstract

To the Editor: We previously published results from a small pilot study suggesting that young women with BRCA1/2 mutations who have used oral contraceptives (OCs) are at much increased risk of breast cancer. 1 BRCA1/2 mutations account for the majority of hereditary breast cancers, 2 so that if OCs are associated with a particularly high risk of breast cancer in BRCA1/2 mutation carriers, then we would expect this effect to be reflected in the extensive epidemiologic literature on OCs and breast cancer in women with a family history of breast cancer, in particular in young women. We have listed in Table 1 the studies that reported risk of early onset breast cancer (<45) associated with duration of OC use by stratum of family history of breast cancer.3–5 We estimated the odds ratios (ORs) of breast cancer in women with and without family history with and without long-term OC use using women with no family history who had not used OCs as a reference. These ORs were used to determine whether the interaction between family history and long-term OC use exceeds additivity.6 All three studies suggest an excess over additive effects for family history and long-term OC use in the youngest age group. In Rosenberg et al ’s study, the interaction was only apparent for women under age 35. A fourth study suggested no interaction between ever/never OC use and family history,7 but did not present data on duration of OC use. A fifth study suggested an interaction between family history and long-term OC use,8 but contained inadequate information for the odds ratios to be estimated. Table 1: The Effect of Family History of Breast Cancer and Long-Term Oral Contraceptive (OC) Use on Breast Cancer Risk in Young WomenThese data therefore suggest that family history modifies the effect of OC use on early onset breast cancer occurrence, and this result is supported by the finding that cell proliferation may be particularly increased in OC users with a family history of breast cancer. 9 These data thus support indirectly our finding that OC use increases breast cancer risk in young women with BRCA1/2 mutations. 1 It may be, however, that the results in the Table 1 are a biased sample of the studies that have data on the relation between OC use and breast cancer and how it is affected by family history. That is, a form of publication bias may exist in which authors are more likely to report on this particular relation to family history if the result indicates a positive effect. We, for example, did not report the data separately by family history in our report of a study of early breast cancer conducted in Los Angeles, where we found no effect modification by family history. 10 The question of whether family history modifies the effect of OC use on breast cancer in young women is important. The collaborative study on breast cancer 11 only reported the relative risk by family history in all ages combined. A further analysis of their data restricted to young women would help settle this issue. Giske Ursin Chaoyang Li Malcolm C. Pike