• breast cancer
• mammographic density
• repeated measurements
• risk model

In post-menopausal breast cancer (pBC), early detection is a key strategy to improve overall survival. An accurate risk prediction model can help targeting screening programs and therefore improve timely cancer detection. Among the factors associated with pBC risk, we focused on percent mammographic density (PMD) and body mass index (BMI). We used data from a case-control study nested in the E3N cohort, including 247 cases and 290 controls. We described the intra-individual variation of PMD and BMI over time and estimated the contribution of their measurements taken before and after the age of 55 in modeling the risk of pBC.
The correlations between the two measurements (median time interval 12 years) were 0.73 and 0.83 for PMD and BMI, respectively. Larger changes in PMD were observed with larger between-assessment intervals. Post-menopausal BC risk was independently associated with PMD, BMI, their variations, and age.
Assessing premenopausal BMI and PMD, along with their changes over time, did not significantly improve the AUC of the logistic regression model for predicting pBC risk compared to using postmenopausal measurements alone (0.64, 95%CI from 0.60 to 0.69 for both). Both approaches outperformed the model using only the premenopausal measurements (0.61, 95%CI from 0.56 to 0.69). The risk estimates of the model including baseline assessments and their changes across menopausal transition where similar to those from the model including baseline assessments alone when PMD changes where similar to those expected based on premenopausal measurements, age, and hormone replacement therapy (HRT), but not when PMD changes where lower or higher.
Our findings suggest that assessing PMD in premenopausal women may enhance pBC risk estimation, particularly for HRT users who might experience atypical patterns of PMD variation.