This well-written article can benefit only from reinforcement of a few of its major points, some supplemental discussion about the important role of biologic models in understanding and managing breast cancer development, and a note about the critical need for research and perspectives from the social sciences concerning this subject. I say "only" because this article beautifully and clearly explores some of the language of epidemiology critical to the subject, language which is becoming increasingly important in routine medical practice. Practitioners and, increasingly, the public (medical "consumers") are concerned with risks and numbers.
Three Important Messages
The article has three messages that deserve emphasis: First, women overestimate their risks for breast cancer, both their lifetime probabilities and, more importantly, their immediate future--say, 5-year--risks. The adverse consequences of these overestimates do demand our careful clinical attention. As Vogel nicely shows, we can use available models to estimate risks for individuals, but, for those at greatest risk, for example, those with BRCA1 mutations, interpreting the genetic test results is challenging because of allelic heterogeneity.
A second critical message is that the optimal details or components of the counseling encounter are not yet defined. Experienced genetic counselors suggest that several hours of individual counseling is the norm for women with a high prior probability of testing positive for a BRCA1 mutation. In my experience, in contrast to the suggestion Vogel makes, the period around the diagnosis of breast cancer in a close relative is a bad time to work with an at-risk patient, possibly because most women are emotionally in turmoil at this time.
Finally, management also is tricky. Vogel does not mention the uncertainties over the risks and benefits of mammography in premenopausal women, even those at increased risk for breast cancer. Given the complexities and uncertainties inherent in managing and counseling these women, I would strongly second his suggestion for referral for counseling and possibly genetic testing when prophylactic surgery is under consideration.
Biologic Models of Breast Cancer Development
Well-read clinicians will recognize the limitations of the models Vogel describes. His discussion does not bring these "risk factors" together into a rational biologic model. To supplement this article, let me add these perspectives. Clearly, genetic factors, that is, familial (presumably primarily inherited) mutated genes, are the most powerful conferrers of risk for breast cancer. At present, the mechanisms through which these genes exert their powerful influences are unknown.
Beyond genetic influences, there are essentially two, not mutually exclusive, biologic models for breast cancer development. The first is a differentiation model. The major observation in humans that provides support for this model is the linear relationship of age at first birth and risk of breast cancer so definitively described a quarter-century ago by MacMahon et al (Figure 1). How one states that relationship governs how numerically significant it seems. In contrast to Vogel's suggestion, I believe that Russo et al have provided us with an explanation for this observation: Lobular maturation and terminal-end bud differentiation consequent to a full-term pregnancy decrease the numbers of undifferentiated cells in the mammary glands of rodents.
The second model is essentially a hormone exposure model. The associations of breast cancer risk with age at menarche, age at menopause (natural or artificially induced, as by surgical oophorectomy), obesity, anovulatory cycles (increased by exercise), hormone density, and hormone replacement therapy in postmenopausal women (a weak association) all support such a model. Pike et al have emphasized this model, calculating the significant consequences of premature termination of ovarian function. In their calculations, they have used the repeated observation that (for breast cancer prior to menopause) the graphed age incidence curve for major malignancies on log/log scales is linear. The details of which hormones are most significant and timing of exposures are unclear, but this should not distract us from appreciating the broad picture.
What Vogel's discussion and the comments above suggest is that we need to devote greater attention to the construction of models of breast cancer development. We must then use available clinical data and observations to test these models and to develop, whenever possible, model-based studies.
Need for Involvement of the Social Sciences
Finally, it is critical to note that breast cancer risk is not a subject solely for medical clinicians or geneticists. Perspectives and research from a broad range of social sciences, from historians and anthropologists through journalists and educators, are needed. Only 3% of the budget of the Human Genome Project is earmarked for such research, and there is only one social science member on the Task Force on Genetic Testing of the National Center for Human Genome Research. In my view, we can achieve the best results for individuals and our society only by giving significantly greater recognition to the importance of conducting social science research in this area and by committing further resources to that end.