Mammographic density: use in risk assessment and as a biomarker in prevention trials.

Introduction Mammographic breast density is positively associated with risk of developing breast cancer for both preand postmenopausal women. Breast Imaging Reporting and Data System (BIRADS) or similar semiquantitative pattern analyses are inexpensive and suitable for large epidemiologic studies, but intrareader reliability in interpretation is suboptimal. Continuous, computerassisted measurements have greater intraobserver reliability and are more suitable for prevention trials. There are many technical and physiologic factors that can affect density readings including film exposure, positioning, compression, change in hormonal milieu, weight, and alcohol intake. If mammographic breast density is to be used as a response biomarker, change in these factors must be minimized over the course of the prevention trial. Although drugs (or other interventions) that result in reduced density are likely to result in reduced risk for breast cancer, it is quite possible that there will be interventions (such as weight loss) that will result in reduced risk of breast cancer but will not reduce proportional breast density or will actually increase it. Furthermore, there is no current evidence that change in breast density alters individual breast cancer risk. More studies are needed that examine the comparative discriminatory ability of mammographic density or nondiagnostic tissue sampling (i.e., random periareolar fine needle aspiration) when added to the Gail or other epidemiologic risk models.