Abstract P1-07-28: Retrospective analysis of clinicopathologic features predictive of oncotype DX discordance in estrogen receptor positive, node negative breast cancer patients

Abstract

Abstract BACKGROUND: Oncotype DX (ODX) is a validated recurrence score (RS) used to predict the risk of recurrence and benefit of chemotherapy in ER positive, node negative breast cancer patients. Prior to ODX, treatment recommendations regarding adjuvant chemotherapy and mortality approximation have taken into account clinical and pathologic risk factors. A discordance rate of 7-19% between risk allocating pathologic factors and ODX RS has been previously reported with progesterone receptor (PR) negativity noted as a defining clinical feature in numerous cases. The association between other clinicopathologic features and discordance is less certain. METHODS: ODX data and clinicopathologic features were retrospectively reviewed for 724 breast cancer tumors belonging to 704 patients between 2006 and 2016. ODX discordance was defined as either 1-step discordance or 2-step discordance between ODX risk group (low, intermediate, high) and tumor grade (TG) (well differentiated, moderately differentiated, poorly differentiated). Tumors with 1-step discordance received a discordance score (DS) of 1 while those with 2-step discordance received a DS of 2. The database was subsequently analyzed using Paik's RS cutoffs as well as those outlined in the TAILORx trial. An odds ratio (OR) of >1 was consistent with discordance. RESULTS: Among 724 tumor samples, ODX ER score (p=0.000), ODX PR score (p=0.000), ODX HER2 score (p=0.000), TG (p=0.000), mitotic count (MC) (p=0.0012), DCIS grade (p=0.0046), DCIS type (comedo necrosis vs. non-comedo necrosis) (p=0.0335) and micropapillary features (p=0.0044) were significantly associated with RS. Median age of cohort was 59 years and median tumor size was 1.2 cm. Of 724 tumors, 619 from 604 subjects were eligible for assessment of discordance. Median RS was 16. Using Paik's RS cutoffs, 64.3% discordance was observed: 52.5% 1-step discordance (DS 1) and 11.8% 2-step discordance (DS 2). The TAILORx categorization yielded a discordance rate of 44.3%: 40.1% 1-step discordance and 4.2% 2-step discordance. On univariate analysis and using Paik's RS cutoffs, young age (p= 0.0240), high MC (p=0.0006), large tumor size (>20 mm) (p=0.0209), the presence of DCIS (p=0.0480), high DCIS grade (p= 0.0033), and high ODX PR and ER scores (p= 0.0000) were significant clinicopathologic features predictive of discordance. On multivariate analysis, high MC (p= 0.0000), high ODX PR and ER scores (p=0.0000) remained significant as well as premenopausal status (p=0.026). Per TAILORx cutoffs, univariate analysis revealed younger age (p= 0.0060), high MC (p= 0.0270), premenopausal status (p= 0.0124), and high ODX PR and ER scores (p= 0.0000) as significant for discordance. On multivariate analysis, high ODX PR and ER scores (p= 0.0000) remained significant. CONCLUSION: In this retrospective ODX database, premenopausal status, high MC, high ODX PR and ER scores as per Paik's RS cutoffs were significant predictors for ODX discordance while high ODX PR and ER scores were significant predictors per the RS's outlined in the TAILORx trial. RS cutoffs per the TAILORX trial appear to create less discordance between RS and TG than the original cutoffs outlined by Paik and colleagues. Citation Format: Cascetta KP, Zimmerman BS, Eggert L, Molot MC, Ru M, Nayak A, Bleiweiss I, Tiersten A. Retrospective analysis of clinicopathologic features predictive of oncotype DX discordance in estrogen receptor positive, node negative breast cancer patients [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P1-07-28.