Ethnic and geographic differences in mammographic density and their association with breast cancer incidence.

The objective of this pooled analysis was to compare differences in dense areas and percent mammographic densities to breast cancer incidence in populations at different breast cancer risk. The data set included 1,327 women aged 40-80: Caucasians from Norway, Arizona, and Hawaii, Japanese from Hawaii and Japan, Latina from Arizona, and Native Hawaiians from Hawaii. One reader performed computer-assisted quantitative density assessment for all mammographic films. Multiple linear regression models evaluated the influence of the covariates on breast density. Spearman correlation coefficients (r (s)) estimated the association between breast density and breast cancer incidence for the seven populations. After adjustment for covariates, ethnicity, but not location, was significantly associated with breast density. In the full model, 19% of the variation in the dense areas and 46% in the variation of percent densities were explained by measured risk factors. Native Hawaiians had the largest dense areas and women in Japan the smallest, whereas percent densities were highest among Native Hawaiians and Japanese in Hawaii and lowest among Norwegian women. The mean age-adjusted dense area had the strongest association with breast cancer incidence (r (s) = 0.93, P = 0.003); the relation with percent density was considerably weaker (r (s) = 0.32, P = 0.48). The correlation between age-adjusted dense area and breast cancer incidence remained strong after selectively removing individual data points. This comparison of mammographic densities suggests that, on a group level, age-adjusted dense areas may reflect breast cancer incidence better than percent densities.