Mammographic density (MD) is one of the strongest risk factors for breast cancer, but largely underused for risk assessment. Recent data have shown that incorporating a single 'baseline'MD measure into the well known Gail model only slightly improves the model's discriminatory power. Whether integration of trajectories of longitudinal change in MD further enhances the model's predictive power has not been explored. We hypothesize that the general population is a mixture of heterogeneous sub-groups with regard to the developmental profile of MD, and the trajectories of change may not be completely captured by a single MD measurement. We further hypothesize that integration of longitudinal changes in MD into the Gail model will improve the model's predictive power of individual risk. This study builds upon an ongoing pilot study where 655 breast cancer cases and 627 frequency-matched controls with 3 or more screening mammograms within the last 14 years are being recruited. All archived screening mammograms will be processed using a validated computer-assisted "interactive thresholding'algorithm to assess patterns of longitudinal change in MD. We will use a novel latent growth mixture model to examine the association of longitudinal changes of MD with risk of breast cancer, and to evaluate the risk prediction power of a refined Gail model that incorporates information on the longitudinal developmental profiles of MD. This study may have important implication for risk prediction. Refined Gail model may enhance the model's predictive ability to identify high-risk individuals, and better guide the initiation of chemoprevention and interventions.
Project Narrative This study builds upon an ongoing pilot study where 655 breast cancer cases and 627 healthy controls with 3 or more screening mammograms within the last 14 years are being recruited. This unique study population, each participant with at least 3 or more screening mammograms, will be readily available and allow us to explore whether integration of longitudinal changes in mammographic density over time would enhance the discriminatory power of the well known Gail model for breast cancer risk assessment.