Abstract 213: Evaluating high risk BI-RADS 4 mammographic lesions: a pilot trial of textural analysis (TA) as a supplement to digital mammography (DM)

Abstract

Abstract Background: Although mammographic lesions classified as BI-RADS 4 are considered suspicious for carcinoma - by definition - and biopsy (bx) is therefore always recommended, only 20-30% of such lesions ultimately yield malignancy at biopsy. TA is an ancillary non-subjective quantitative analysis platform that can be performed as an add-on to routine mammographic analysis. The ability of TA to characterize lesions with high rates of heterogeneity led us to hypothesize that TA, as a supplement to routine DM analysis, might aid in selecting patients (pts) for bx vs. serial observation. Methods: A retrospective analysis was performed on 40 screening pts with no history of cancer, with newly identified mammographic densities characterized as BI-RADS 4. Lesions measured 1-3 cm size and had no associated microcalcifications. Twenty pts had confirmed malignant biopsy while the other 20 pts served as age matched controls with similar lesion characteristics but ultimate benign biopsy result. For each lesion, TA was performed on a 1×1cm region of interest (ROI), along with a secondary control ROI of normal breast tissue. Results: Pts with malignant and benign bx had a median age of 53 years (range 28-83) vs. 52 years (range 32-75) (p = NS) and median nodule size 1.5 cm (range 1.0-2.9 cm) vs. 1.3 cm (range 1.0-2.5 cm) (p = NS), respectively. No differences between the groups were seen in menopausal status nor parenchymal density. The mean positive pixel (MPP) value was most predictive of malignancies at a Spatial Scale Frequency (SSF) filter level of 0.8. The median MPP for malignant lesions was 577 vs 730 for benign lesions (p = 0.033). There was no difference between MPP or other TA parameters in normal tissue of either group. Conclusion: TA, as a supplement to DM, may help stratify which women with high-risk BI-RADS 4 densities that better benefit from biopsy vs. those that might be safely followed with serial surveillance. Larger scale studies will be needed to validate these findings and are currently ongoing. Citation Format: Melissa R. Gordon, Erkut Borazanci, Daniel D. Maki, Ron L. Korn. Evaluating high risk BI-RADS 4 mammographic lesions: a pilot trial of textural analysis (TA) as a supplement to digital mammography (DM). [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 213. doi:10.1158/1538-7445.AM2015-213