Correlation between five-year overall survival by PREDICT and 21 gene recurrence score (RS) in a Mexican private institution.

Abstract

e12543 Background: In México the 5-year overall survival of Early Breast Cancer (EBC) reaches to 82-97%. Oncotype Dx evaluates the expression of 21 genes associated with recurrence and classifies patients into risk groups. PREDICT (https://breast.predict.nhs.uk/tool) is an online tool which assesses 5 and 10 year overall survival in breast cancer patients who receive adjuvant chemotherapy, adjuvant hormonal therapy and trastuzumab therapy for HER2(+) patients. Methods: Retrospective review of medical records of early stage breast cancer patients with (HR+) and HER2 (-) treated at our institution. Clinicopathological characteristics and (RS) were collected. 5-year overall survival was calculated with the PREDICT online tool. Both scores were compared and the correlation was estimated with Spearman’s Rho and global agreement with intraclass correlation coefficient (CCI), statistical software: STATA SE ver11.0 (StataCorp LLC Texas,USA). Results: From January 2008 to December 2018, 136 patients with EBC (IA-IIB), HR(+) HER2(-), N0-1 were included. The Median age at diagnosis was 55.03 years (30-80). Stage IA patients accounted for 68.38% of the population. Patients were classified into risks according to the original description of Oncotype. 72 patients (53%) were classified as low risk (LR), 49 (36%) at intermediate risk (IR) and 15 (11%) patients at high risk (HR). When reclassifying the risk categories using the cut-off values in TAILORx trial, the population distribution was modified, with a notable increase in the population in IR with 86 (63.2%) patients in this group, 28 patients in the LR group ( 20.5%) and 22 patients (16.1%) in HR. We decided to use TAILORx cut-off values for the aim of this work. Mean overall survival established by the 21 gene RS was 98% for the overall population. The mean overall survival calculated by PREDICT was 93%. Spearman’s correlation coefficient was 0.16 (Spearman’s Rho = 0.16 p = 0.065) with intraclass correlation coefficient (ICC) = 0.04 (IC 95% -0.15 – 0.22, p = 0.33). Conclusions: The results do not show a clear correlation between the tests. Lack of such correlation may be due to a low number of patients. In our population PREDICT couldn't replace the RS test for therapeutic decision making.