Localized Fibroglandular Tissue as a Predictor of Future Tumor Location within the Breast

Background: Mammographic density (MD) is a strong marker of breast cancer risk, but it is unclear whether tumors arise specifically within dense tissue. Methods: In 231 British women diagnosed with breast cancer after at least one negative annual screening during a mammographic screening trial, we assessed whether tumor location was related to localized MD 5 years prior to diagnosis. Radiologists identified tumor locations on digitised films. We used a validated algorithm to align serial images from the same woman to locate the corresponding point on the prediagnostic film. A virtual 1 cm square grid was overlaid on prediagnostic films and MD calculated for each square within a woman's breast (mean = 271 squares/film). Conditional logistic regression, matching on a woman's breast, was used to estimate the odds of a tumor arising in a square in relation to its prediagnostic square-specific MD. Results: Median (interquartile range) prediagnostic MD was 98.2% (46.8%–100%) in 1 cm-squares that subsequently contained the tumor and 41.0% (31.5%–53.9%) for the whole breast. The odds of a tumor arising in a 1 cm-square were, respectively, 6.1 (95% CI: 1.9–20.1), 16.6 (5.2–53.2), and 25.5-fold (8.1–80.3) higher for squares in the second, third, and fourth quartiles of prediagnostic MD relative to those in the lowest quartile within that breast (Ptrend < 0.001). The corresponding odds ratios were 2.3 (1.3–4.0), 3.9 (2.3–6.4), and 4.6 (2.8–7.6) if a 3 cm-square grid was used. Conclusion: Tumors arise predominantly within the radiodense breast tissue. Impact: Localized MD may be used as a predictor of subsequent tumor location within the breast. Cancer Epidemiol Biomarkers Prev; 20(8); 1718–25. ©2011 AACR.