Abstract P1-01-08: Changes in breast density with mammography and breast MRI among high-risk postmenopausal women

Abstract

Abstract Background: Mammographic density (MD) represents one of the strongest predictors of breast cancer risk. Tamoxifen, a selective estrogen receptor modulator (SERM) and proven chemopreventive agent, causes reductions in MD which correlate with subsequent breast cancer risk. Current methods for measuring MD, such as the Cumulus technique, are labor intensive and not fully automated, which limits utility in the clinical setting. Breast MRI is being used with increasing frequency in high-risk women for breast cancer screening and yields 3-dimensional views of breast volume. Breast MRI fibroglandular tissue volume (MRIV) has been correlated with MD in cross-sectional studies, but less is known about changes in breast density over time on MRI and its relation to breast cancer risk. One study reported a significant reduction in breast MRIV but not MD with the SERM, raloxifene, in high-risk women. We implemented a novel MRI segmentation method to measure MRIV compared to MD in 20 high-risk postmenopausal women that received high-dose vitamin D supplementation for 1 year. Methods: Twenty high-risk postmenopausal women [defined as a 5-year breast cancer risk per the Gail model of ≥1.67%, lobular or ductal carcinoma in situ (LCIS/DCIS)] were assigned to a 1-year intervention of vitamin D3 20,000 IU or 30,000 IU weekly. Other eligibility criteria included baseline MD ≥25% (as assessed qualitatively by BIRADS category) and no current SERM use. Women underwent a digital mammogram and breast MRI at baseline and 12 months. MD was evaluated by the Cumulus technique and MRIV was measured using a novel segmentation method on noncontrast T1-weighted images from breast MRIs. The method incorporated two well-established image analysis techniques, the region-based active contour model and watershed algorithm, into a unified framework. Specific anatomical landmarks were introduced into the algorithm to consistently identify breast boundaries on longitudinal MRI slices. For fibroglandular tissue segmentation, an algorithm combining region-based active contours and level-set approach was used to allow for easy initialization and fast speed. Correlation coefficients and paired t-tests were used to compare MD and MRIV and changes from baseline to 12 months. Results: From Nov 2007 to Jan 2011, 20 postmenopausal women were enrolled and 16 were evaluable at 12 months. Mean baseline breast density measurements were lower for MD compared to MRIV, 12.89% (SD 11.99) vs. 26.64% (SD 6.95), respectively. After 1 year of high-dose vitamin D, there was no significant change in MD or MRIV, mean absolute changes of -0.14% vs. -1.12%, respectively. We did not observe a significant correlation between MD and MRIV density measurements (correlation coefficient= -0.31, p=0.08). Discussion: We observed greater changes over time in breast density on serial breast MRI compared to mammography, however, no significant changes in MD or MRIV was noted after a 1-year intervention of vitamin D. Longer follow-up may be required in order to detect changes in breast density with non-hormonal interventions. Citation Format: Parijatham S Thomas, Mary Beth Terry, Xiaotao Guo, Binsheng Zhao, Richard Ha, Swathi Namburi, Tong Xiao, Dawn Hershman, Susan Refice, Katherine D Crew. Changes in breast density with mammography and breast MRI among high-risk postmenopausal women [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P1-01-08.