Association of imaging-based body fat distribution and mammographic density in the multiethnic cohort adiposity phenotype study

Abstract

Background: As the stronger association of obesity with postmenopausal breast cancer in Asian than white women may be due to body fat distribution, we examined the relation of adiposity measures with percent mammographic density (PMD), a strong predictor of breast cancer incidence. Methods: Atotal of 938 women from five ethnic groups (69.1-2.7 years) in the Adiposity Phenotype Study (APS) underwent DXA and MRI imaging. PMD was assessed in routine mammograms using a computer-assisted method. Spearman correlation coefficients were computed and general linear models were applied to estimate regression coefficients (b) for PMD per 0.5 SD units of adiposity measures while adjusting for known confounders, includingDXA total body fat. Results: For 701 (75%) of the participants (69.1<2.7 years), valid mammograms were obtained. Whereas total body fat, the trunk-toperiphery fat ratio (TPFR), visceral fat (VAT), and subcutaneous fat (SAT) were inversely correlated with PMD (P < 0.0001), the VAT/ SAT ratio correlated positively (rspearman 1/4 0.10; P 1/4 0.01). In fully adjusted models,PMDremained inversely related to TPFR and SAT and disappeared for VAT, while it was strengthened for VAT/SAT (b 1/4 0.51; P 1/4 0.009). This relation was stronger in Japanese Americans than other ethnic groups. Conclusions: This is the first study to show an association of a high VAT/SAT ratio with greater PMD, a marker of breast cancer risk after taking into account total body fat. Impact: The results indicate a link between the propensity to accumulate VAT and the amount of fat in the breast (1-PMD), which may influence the relation of obesity with breast cancer incidence.