BACKGROUND: To evaluate the association between automatically measured mammographic breast density (MBD) and osteoporosis in postmenopausal women and determine whether this relationship differs across body mass index (BMI). METHODS: This retrospective cross-sectional study included 7,143 postmenopausal women who underwent both digital mammography and dual-energy X-ray absorptiometry (DXA). MBD (%) was automatically quantified using the Laboratory for Individualized Breast Radiodensity Assessment (LIBRA). Osteoporosis was defined as a DXA T-score ≤ -2.5. BMI-dependent associations were evaluated using logistic regression with a BMI×MBD interaction (continuous model) and prespecified categorical analyses (median split). Discrimination (AUC) was compared between models with and without MBD. RESULTS: Osteoporosis was present in 12.8% (912/7143) of participants. In the continuous interaction model including age, BMI, MBD, and a BMI×MBD interaction, the BMI×MBD interaction was significant (P = 0.007). The estimated odds ratio (OR) for osteoporosis per 1% higher MBD was 0.989 (95% CI 0.981-0.997) at BMI 20, 0.976 (95% CI 0.964-0.989) at BMI 25, and 0.963 (95% CI 0.943-0.984) at BMI 30. In a prespecified median split (MBD < 25.3% vs. ≥ 25.3%), the absolute risk difference was larger in women with BMI ≥ 25 (9.28% vs. 4.59%). Adding MBD to an age + BMI model minimally changed AUC (0.682 to 0.684). CONCLUSION: Higher automated mammographic breast density was associated with lower odds of osteoporosis, and this inverse association was stronger at higher BMI. Although mammographic breast density added little incremental discrimination beyond age and BMI, these findings may provide biologically and epidemiologically relevant insight but do not support clinical decision-making based on mammographic breast density alone.
Lee et al. (Thu,) studied this question.