Abstract P3-14-06: The Utility of Margin Index to Predict Residual DCIS Following Breast Conserving Surgery

Abstract

Abstract Background: Although breast conserving surgery (BCS) is commonly performed for ductal carcinoma in situ (DCIS), patients may require re-excision for close or positive margins to rule out residual disease. The concept of margin index (defined as margin distance/tumor size) has been found to predict residual disease in patients with invasive breast cancer, but this concept has yet to be evaluated for patients with DCIS. We sought to determine if DCIS margin index (defined as the closest margin of DCIS/extent of DCIS) was predictive of residual disease. Methods: We performed a chart review of all in situ breast cancers treated with BCS at an academic breast center in 2009. DCIS margin index was calculated as closest margin distance of DCIS (mm)/extent of DCIS (mm). Non-parametric statistical analyses were performed to evaluate the relationship between margin index and residual DCIS. Results: 177 patients with DCIS were treated with BCS in 2009. Of these, 87 (49.1%) underwent resection of additional tissue for close margins or as routine practice, and had recorded quantitative margin distances and tumor sizes such that margin index could be calculated. Of these, 18 (20.7%) had residual DCIS in the additional tissue excised. Median patient age was 60.4 years (range; 31.9–91.4). Of the 50 patients with available immunohistochemistry analysis, 42 (84%) were ER-positive and 36 (72%) were PR-positive respectively. 12 (13.8%) had low grade DCIS, 53 (60.9%) intermediate grade DCIS, and 22 (25.3%) high grade DCIS. Median tumor size was 16mm (range; 1–80 mm); and median margin distance was 2 mm (range; 0–18 mm). Median margin index was 0.125 (range; 0 to 18). Factors correlating with residual DCIS on univariate analysis were high tumor grade (33.3% vs. 23.2%, p = 0.044), PR-negativity (66.7% vs. 22.7%, p = 0.044), increasing tumor size (median 27.5 mm vs. 15.0 mm, p = 0.003), decreasing margin distance (median 1.0 mm vs. 2.1 mm, p = 0.004), and lower margin index (median 0.032 vs. 0.200, p < 0.001). Patient age and ER status were not correlated with residual DCIS. In a multivariate model controlling for grade and PR-status, margin index was no longer a significant predictor of residual disease, although a trend was noted (p = 0.076). To determine the relative strength of margin index (vs. extent of DCIS and margin distance) in predicting residual DCIS, a second multivariate model including PR-status, grade, extent of DCIS and margin distance was created. In this second model, extent of DCIS (p = 0.634) and margin distance (p = 0.152) were less predictive of residual disease than margin index was in the first model. PR-negativity remained the only predictor of residual disease in both. Conclusions: Margin index is a stronger predictor of residual DCIS than either margin distance or extent of DCIS, but loses significance when taking PR status into account. That PR-negativity is strongly associated with residual DCIS (independent of tumor size, grade and margin status) is intriguing and warrants further research. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P3-14-06.