Abstract P5-16-04: Predictive biomarkers for invasion on final pathology in patients with preoperative diagnosis of ductal carcinoma in situ of the breast by needle biopsy

Abstract

Abstract Background Predictors of synchronous invasive breast cancer in patients diagnosed with only ductal carcinoma in situ (DCIS) in pre-operative needle biopsies have not been well-defined. Establishing such predictors of invasion has potential to significantly alter management by identifying those patients for whom surgery may be avoidable en lieu of conservative management. This study aims to identify clinicopathologic factors from pre-operative needle biopsies that are predictive of invasive cancer on subsequent surgical excision. Methods The study population consisted of 69 breasts from 67 patients initially diagnosed with only DCIS on needle biopsy (core needle or mammotome) at St. Luke's International Hospital, Japan from 2006 until 2008. Parameters analyzed included presenting clinical features, DCIS nuclear grade and morphologic pattern, and immunohistochemical expression of estrogen receptor (ER), progesterone receptor (PR), epidermal growth factor receptor-2 (HER2), Ki-67 antigen, p16, p53 and cyclooxygenase-2 (COX2) in biopsy specimens. These immunohistochemical markers were previously identified to foretell invasive carcinoma subsequent to DCIS (Kerlikowske, JNCI 2010). Associations between clinical, pathological, and immunohistochemical findings in initial biopsy specimens and the presence of invasive cancer on subsequent excision were analyzed for significance using univariate and multivariate analysis. Results Of 69 breasts with only DCIS on initial needle biopsy, subsequent surgical excision revealed pure DCIS in 46 (66.7%), microinvasive carcinoma in 4 (5.8%), and invasive carcinoma in 19 (27.5%) cases. Sentinel node biopsy was performed in 57 (82.6%) of 69 cases, and 53 (93.0%) of these showed no evidence of lymph node metastases. All 4 cases with lymph node metastases revealed invasive carcinoma in surgical excisions. By univariate analysis, pre-operative factors significantly associated with invasion on surgical excision included detection of a lump by palpation (p<0.05), sampling by core needle biopsy rather than mammotome (p<0.01), p53 positivity (p<0.01), and lack of ER expression (p<0.05) in needle biopsy samples. Combined absence of p16 and COX2 with low Ki-67 expression in needle biopsies was associated with pure DCIS (and absence of invasive cancer) in surgical excisions (p<0.05). Multivariate analysis using all five significant univariate variables revealed two independent pre-operative predictors of invasive cancer on final pathology, namely sampling by core needle biopsy (odds ratio 7.0; 95% CI 1.2-41.7; p<0.05) and p53 positivity (odds ratio 7.1; 95% CI 1.1-44.9; p<0.05). Conclusion If confirmed in a larger sample, predictive clinical and biomarker parameters can help identify those patients diagnosed with only DCIS in needle biopsy who are at high risk of harboring unsampled invasive cancer on final pathology. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P5-16-04.