Breast and ovarian carcinomas pose major public health problems in most Western countries. Countless attempts have been made to better understand a patient’s lifetime risk of breast cancer. The most significant etiologic risk is increasing age, followed by family history. In addition, hormonal and reproductive factorsie, early menarche and later age at menopause, nulliparity (and, therefore, a greater number of ovulations over the patient’s lifetime), and late age at first pregnancy (greater than age 30 years)also increase a patient’s breast cancer risk.
Obesity (perhaps due to the more affluent high-fat, low-fiber, high-caloric diet) and irradiation of the chest, particularly early in life, have also been incriminated as risk factors for breast cancer. On the other hand, early age at first full-term pregnancy, particularly under age 20, is protective against breast cancer.
Rubinstein et al provide a critical appraisal of several mathematical models for calculating breast cancer risk. The so-called epidemiologic models, based on population studies, include the Gail model and the Claus model. Models derived from logistic regression of risk factors that are better predictors of a genetic etiology are the Shattuck-Eidens, Myriad (Frank), and Couch models. The BRCAPRO (Berry-Parmigiani-Aguilar) model[6,7] is based on Bayesian calculations to assess the probability of a BRCA1 or BRCA2 mutation in concert with the family configuration of affected and unaffected individuals. These genetic models calculate mutation probabilities based on affected family members, and risk is adjusted by Mendelian extrapolation for unaffected relatives.
The Gail model’s main value is its ability to predict the rate of breast cancer in large populations. However, as Rubinstein et al note, this approach tends to overestimate breast cancer risk for young women and underestimate it for older women. Its limitations from the standpoint of genetic risk assessment are the inability to incorporate breast cancer history for more than two first-degree relatives and its failure to consider age at onset of cancer. Also, it does not take second-degree relatives into consideration, thereby ignoring paternal family history.
The Claus model employs segregation analysis on data obtained through the Cancer and Steroid Hormone (CASH) study. This model is useful for the subset of women who have one or two relatives with breast cancer, and it may be more accurate than the Gail model for this particular cohort, especially when there is a paternal family history of breast cancer. Rubinstein et al appropriately suggest that the Gail and Claus models should not be employed in individuals showing a strong family history of cancer. Rather, they should be used cautiously when genetic testing has been performed and the result is negative.
Modified Nuclear Pedigree