Abstract P3-07-42: Predicting outcome and benefit to first-line bevacizumab in advanced/metastatic hormone receptor (HR)+/HER2-negative breast cancer (BC) treated with endocrine therapy: A correlative science study from the LEA phase III clinical trial (GEICAM/2006-11_GBG 051)

Abstract

Introduction: The role of bevacizumab in combination with chemotherapy in metastatic BC is controversial, and no biomarker exists as of today that predicts benefit to this agent. In the LEA clinical trial, a numerical, statistically non-significant benefit from the addition of bevacizumab to endocrine therapy (ET) was observed in the first-line metastatic setting (18.4 vs. 13.8 months of Progression-Free Survival (PFS), p=0.14). Here, we explored various gene expression-based predictors of outcome and benefit to bevacizumab. Methods and materials: LEA trial randomized 380 patients with HR+/HER2- advanced disease to bevacizumab in combination with ET (ET+B) vs. ET alone. Primary endpoint was PFS. Expression of BC selected genes was evaluated in formalin-fixed paraffin-embedded (FFPE) primary tumors using the nCounter platform from patients randomized in Spain that consent for biomarker analyses. The following variables were evaluated: 1) research-based PAM50 intrinsic subtypes (categorical variable; Luminal A, Luminal B, HER2-enriched, Basal-like and Normal-like); 2) research-based PAM50 signatures (continuous variable; scores showing the distant of the gene expression values of an individual sample compared to the centroid gene values for each PAM50 intrinsic subtype); 3) risk of recurrence (ROR) groups (low, medium and high); 4) the 13-gene hypoxia/VEGF signature (continuous); and 5) Ki67 by immunohistochemistry (continuous). Uni- and multivariable Cox models for PFS were used to test the prognostic significance of each variable. To determine whether each variable is predictive of bevacizumab benefit, we tested the interaction term of each variable by treatment arm in a Cox model. Results: Tumor samples from 103 patients were analyzed: 55 (53%) in ET+B arm and 49 (47%) in ET arm. Subtype distribution was as follows: 57 (55.3%) Luminal A, 32 (31.1%) Luminal B, 5 (4.9%) HER2-enriched, 1 (1.0%) Basal-like, and 8 (7.8%) normal-like. In a univariate analysis, Luminal B tumors had a poorer outcome using Luminal A as reference (13.8 vs. 21.3 months, respectively; (hazard ratio, HR=1.80, 95% CI 1.10-2.95, p=0.019). Concordant with this finding, Luminal A signature was associated with a better outcome. Similarly, ROR-P high group showed a poorer outcome than ROR-P low group (8.5 vs. 19.4 months; HR=2.88, 95% CI 1.30-6.35, p=0.009). Neither VEGF-13 signature nor Ki67 were found to be associated with PFS. Similar findings were obtained after adjustment for treatment, age, previous ET, ECOG, visceral disease and number of metastatic sites. In terms of treatment benefit, the HER2-enriched signature was the only variable found predictive of bevacizumab PFS benefit in univariate (p=0.010) and multivariate (p=0.015) analyses. Conclusions: In advanced HR+/HER2- disease, intrinsic subtype (i.e. Luminal A vs. B) independently predicts PFS following first-line ET. In addition, HR+/HER2-negative tumors with high expression of the HER2-enriched signature, a biomarker of estrogen-independence, benefit the most from bevacizumab. Further validation of these prognostic and predictive biomarkers is warranted. Citation Format: Prat A, de la Haba-Rodriguez J, Guerrero Á, García-Sáenz JA, Morales S, Antón A, Muñoz M, Ramos M, Martínez-Jáñez N, Margelí M, Servitja S, Rojo F, Galván P, González S, Cruz J, Sánchez-Rovira P, Perelló A, Rodríguez-Martin C, Casas M, Carrasco E, Caballero R, Martín M. Predicting outcome and benefit to first-line bevacizumab in advanced/metastatic hormone receptor (HR)+/HER2-negative breast cancer (BC) treated with endocrine therapy: A correlative science study from the LEA phase III clinical trial (GEICAM/2006-11_GBG 051). [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 P3-07-42.