Women with radiologically dense breasts are at substantially greater risk of developing breast cancer than women with mostly fatty tissue. However, little is known about the biological mechanisms underlying this relation. We evaluated the association between breast density and hormone receptors using tissue microarrays (TMA). Breast cancer cases were recruited from a nested case-control study, a substudy of the Multiethnic Cohort. All women had provided information on reproductive characteristics and hormone replacement therapy (HRT) at cohort entry and as part of the breast density study. Prediagnostic digitized mammograms were assessed for densities using a computer-assisted method. Paraffin-embedded breast tissue was available through the Hawaii Tumor Registry. After review of pathologic slides for selection of representative malignant and non-malignant tissue from each case, TMAs were prepared including 4 cores from tumor tissue and 4 from benign tissue. Immunostaining for progesterone receptor (PR) and estrogen receptor alpha (ERa) was performed. Image analysis was used to measure stained cells. We created categorical variables indicating positive staining of samples. General linear models with adjustment for potential confounders were applied and least square means of breast density by receptor expression were computed. We obtained consent and tissue blocks for 283 women; sufficient tissue to be placed on TMAs was available for 279 cases. Although some cores were lost or unreadable, all except two subjects had at least one PR or ERa result. Mean percent densities were 49% for the 62 premenopausal and 34% for the 217 postmenopausal women. For malignant samples, 191 out of 276 had PR expression in at least one core; of the benign samples, 123 out of 255 were stained. For ERa, 229 out of 276 malignant samples showed expression in at least one core; the respective values for benign tissue were 134 out of 250. Kappa values for PR and ERa expression were 0.36 in malignant tissue and 0.34 in benign tissue. For women whose tumor tissue showed no ERa expression, mean densities were 28% as compared to 36% for women with any ERa expression (p = 0.01). This difference was present for pre- (38 vs. 53%; p = 0.08) and postmenopausal women (25 vs. 33%; p = 0.04). After stratification by HRT use, the differences persisted among estrogen users (20 vs. 35%; p = 0.02) and combined HRT users (24 vs. 37%; p = 0.12), but we observed no difference among non-users (27 vs. 25%; p = 0.80). In benign tissue, there was no difference in breast density by ERa expression (35 vs. 36%; p = 0.54). We also observed no relation between PR expression and mammographic densities in malignant or benign samples. These preliminary findings suggest that women with hormone dependent breast cancers, as assessed by ERa expression in tumor tissue, have higher mammographic densities. This relation appears to be limited to premenopausal women and postmenopausal HRT users.
99th AACR Annual Meeting-- Apr 12-16, 2008; San Diego, CA