Ductal Lavage for Breast Cancer Risk Assessment

Lisa A. Newman, MD, MPH, FACS, Cassann Blake, MD


Cancer Control. 2002;9(6) 

In This Article

Risk Assessment Models

The established breast cancer risk assessment models have several limitations resulting in increased motivation to investigate more accurate alternative means of identifying high-risk individuals. The widely used Gail breast cancer risk model[24] is utilized to determine eligibility for chemoprevention trials.[25,26] The Gail model is a logistic regression model that was developed by analyzing breast cancer risk factors from a case-control subset of participants of the Breast Cancer Detection and Demonstration Project (BCDDP), a mammography screening program. This model estimates the likelihood that an individual woman will develop breast cancer over a 5-year period and over her projected lifetime by accounting for the relative risks conferred by age at evaluation, expected longevity, first-degree family history of breast cancer, age at menarche, nulliparity or age at live birth, and number of prior breast biopsies. The Gail model has been modified[27] since the original version was published in 1989[24] to estimate risk for invasive disease only and to account for ethnicity (African American vs white American) and whether atypical hyperplasia was present in any prior biopsies. A relative risk of 1.8 is included in the model calculation for a woman with a history of a breast biopsy demonstrating atypical hyperplasia.

The Gail model has been validated as being accurate in predicting numbers of cancer cases likely to develop in a given cohort of patients,[27,28,29,30] yet its generalizability has been challenged because of several issues. Until recently,[31] the model was not validated in a cohort of non-white American women; it does not account for risk conferred by the paternal or extended family history; and it does not account for well-known pathologic indices of increased breast cancer risk such as lobular carcinoma in situ. Furthermore, although hormone replacement therapy (HRT) was a significant breast cancer risk factor in the BCDDP patient population from which the model was derived, the number of appropriately exposed women was insufficient for this feature to be included into the model.[24] Recently published results from the placebo-controlled phase III study of HRT in the Women's Health Initiative[30] demonstrated that after a median follow-up of 5.2 years, exogenous hormones were associated with a hazard ratio of
1.26 (95% confidence interval, 1.00-1.59). These findings raise the potential concern that the model may not be adequately accounting for the degree of risk conferred by an exposure that has increased in prevalence since the era of accrual to the BCDDP.

As demonstrated by Rockhill et al,[31] the model's discriminatory accuracy at the individual level is particularly suboptimal. Detection of proliferative breast hyperplasia, especially with atypia, may be a useful means of refining short-term risk estimates for the individual patient based on epidemiologic models. In the only study where the model's validity was assessed in a non-white patient population, its accuracy was questioned in a subset of young African American women (ie, younger than 45 years old).[32] This study also demonstrated weakness of the model at the individual level.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.