Association Between Oral Contraceptive Use and Premenopausal Breast Cancer: Mediated by Hormonal Confounders?

Abstract

To the Editor: Kahlenborn et al1Kahlenborn C Modugno F Potter DM Severs WB Oral contraceptive use as a risk factor for premenopausal breast cancer: a meta-analysis.Mayo Clin Proc. 2006; 81: 1290-1302Abstract Full Text Full Text PDF PubMed Scopus (167) Google Scholar performed a meta-analysis of 34 case-control studies of the relationship between oral contraceptive (OC) use and premenopausal breast cancer. They found that OC use was associated with a risk of breast cancer that was slightly, but significantly, increased (odds ratio, 1.19; 95% confidence interval, 1.09-1.29). Confounding factors are a major issue in the interpretation of all case-control studies. For example, I previously noted that confounding is to be expected in case-control studies of the associations between conditions that are truly caused by high or low levels of steroid hormones and some forms of risky behavior (treated as risk factors).2James WH Hypothesis: gonadal hormones act as confounders in epidemiological studies of the associations between some behavioural risk factors and some pathological conditions.J Theor Biol. 2001; 209: 97-102Crossref PubMed Scopus (17) Google Scholar In the case of premenopausal breast cancer, the disease is thought to be partially caused by high levels of estrogenic (and perhaps androgenic) hormones. Moreover, the behavioral trait of sensation seeking is associated with high levels of both estrogenic and androgenic hormones.3Zuckerman M. Behavioral Expressions and Biosocial Bases of Sensation Seeking. Cambridge University Press, Cambridge, England1994Google Scholar In contrast to age-matched controls, a higher proportion of young female sensation seekers would be expected to choose to engage in voluntary “risky” behavior, eg, smoking, OC use, and abortion. Thus, the association of OCs with breast cancer may simply be a reflection of the independent association of both these factors with high levels of estrogens and/or androgens. The point could be tested by assaying the hormones of control subjects and of young women at the time they first choose to use OCs. I hypothesize that, at the time of OC initiation, users have higher hormone levels than controls. Of course, even if confounding is present, that would not imply that it is the sole cause of the association between OC use and breast cancer. Moreover, the argument would be altered if OC use directly affects women's steroid hormone concentrations.