Abstract P2-08-10: Validation of prediction of local recurrence (LR) by Prosigna® (PAM50) in a Danish breast cancer cooperative group (DBCG) cohort of hormone receptor positive (HR+), postmenopausal early breast cancer (EBC) patients allocated to 5yr of endocrine therapy (ET)

Abstract

Abstract Background: HR+EBC patients are routinely treated with both adjuvant radiation therapy (RT) and ET. RT is considered an important tool for achieving local control of disease. A limited number of biomarkers have been demonstrated to predict LR. In a previously performed retrospective analysis of a randomized trial, Prosigna (PAM50) risk of recurrence (ROR) score identified low risk patients with a local recurrence rate of 1.6% at 9.5yr median follow-up. In this study, we seek to validate the ability of ROR to predict LR in a comprehensive nationwide cohort from Denmark. Methods: Using the population based DBCG database primary FFPE tumor blocks and follow-up data were collected from all postmenopausal Danish women diagnosed from 2000-2003 with HR+EBC (N=2,722). Prosigna (PAM50) on the NanoString nCounter® Dx Analysis System assigned each patient an ROR score and associated risk group based on pre-specified cutoffs. Patients are also assigned an intrinsic subtype (Luminal A, Luminal B, Her2-Enriched, Basal-Like) based on gene expression. Univariate and multivariate analyses were performed to assess the ability of Prosigna (PAM50) to predict LR. Results: 48 local recurrences were observed with median follow-up of 9.25 yr. Continuous ROR was significantly associated with LR in univariate and multivariate models including node status (0, 1, 2, or 3 positive nodes), tumor size (≤2 vs.>2cm), grade (I, II, III, or unknown), age (≤65 vs.>65yr), and local treatment (mastectomy (MX), lumpectomy+RT, or MX+RT) (p=0.036 and p=0.049, respectively). Clinicopathologic variables were not significant in the multivariate model alone or in combination (p=0.85 for full model excluding ROR). Utilization of a pre-specified LR cutoff, hazard ratio (HR) and [95%CI, p-value] for high risk vs. low risk patients in a univariate analysis was 1.96 [1.11-3.46, p=0.0205] and 2.04 [1.08-3.83, p=0.0275] in the multivariate analysis. 10-year cumulative incidence of LR for low risk patients was 1.7% [1.1%-2.6%]. Similarly, 10-year cumulative incidence of LR for Luminal A patients was 1.7% [1.1%-2.6%]. 10-year cumulative incidence for high risk patients was 2.3% [1.3%-3.2%]. Conclusions: In a large population-based study of n=2,722 patients, Prosigna (PAM50) predicted LR over standard variables. These data validate a pre-specified cutoff separating patients at high and low risk of LR. Additional studies of Prosigna (PAM50) in RT-untreated populations are ongoing. Citation Format: Ole Eriksen J, Jensen M-B, Laenkholm A-V, Kibøll T, Bruun Rasmussen B, Knoop AS, Ferree S, Haffner T, Buckingham W, Schaper C, Ejlertsen B. Validation of prediction of local recurrence (LR) by Prosigna® (PAM50) in a Danish breast cancer cooperative group (DBCG) cohort of hormone receptor positive (HR+), postmenopausal early breast cancer (EBC) patients allocated to 5yr of endocrine therapy (ET). [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P2-08-10.