Non-basal subtype defined by FOXC1 expression as an independent predictor of capecitabine efficacy in the triple negative breast cancer GEICAM/2003-11_CIBOMA/2004-01 trial.

Abstract

516 Background: In a prespecified correlative analysis of the GEICAM/2003-11_CIBOMA/2004-01 phase III clinical trial (CIBOMA trial), PAM50 non-basal status identified triple-negative breast cancer (TNBC) patients (pts) most likely to benefit from adjuvant capecitabine (cap). FOXC1 cell plasticity/metastasis transcriptional driver assessment by standardized Veresca FOXC1 Immunohistochemistry (IHC) BC test (Veresca) has shown rigorous capacity to identify basal-like breast cancer (BLBC) subtype throughout TNBC cohorts. We sought to explore the concordance of BLBC identification by FOXC1 with PAM50 molecular subtyping and its prognostic/predictive value in the CIBOMA trial (NCT00130533). Methods: We analyzed FOXC1 expression (Veresca, Onconostic Technologies) and the Ki-67 Antigen MIB-1 (Dako) by IHC on tumors from 705 TNBC pts randomized to maintenance with cap (357, 51%) vs observation (obs) (348, 49%) after standard (neo)adjuvant chemotherapy (CT). Veresca score (VFOXC1) was defined as % of tumor cells with positive nuclear staining (Proportion Score 0-5) plus staining intensity (Intensity Score 0-3). BLBC subtyping by Veresca (cutoff ≥4) and PAM50 signature (Nanostring) were compared using ROC analysis and Kappa index (KI). Ki67 ≥20% cutoff was applied. Cox regression models were assessed to predict DRFS (primary endpoint), DFS and OS (secondary endpoints). Results: VFOXC1≥4 identified 460 (65%) BLBC pts in this CIBOMA trial cohort (229 (49.8%) treated with cap, and 231 (50.2%) in obs). VFOXC1 detection was not associated with any clinicopathological characteristic. VFOXC1 expression significantly correlated with PAM50 BLBC subtyping (ROC analysis AUC 0.874, KI 0.43) but moderately with IHC subtyping (AUC 0.548). VFOXC1 BLBC subtype was not associated with DRFS (HR=0.94; 95%CI 0.68-1.30; p=0.713) in the entire cohort. However, VFOXC1 non-BLBC subtype was a strong independent predictor of cap benefit for DRFS (univariate analysis, interaction p-value p=0.117; HR=0.53; 95%CI 0.31-0.90; p=0.019; multivariate analysis, HR=0.44; 95%CI 0.25–0.76; p=0.003). This predictive effect of VFOXC1 non-BLBC on cap benefit was also confirmed at DFS (HR=0.47; 95%CI 0.28–0.78; p=0.003) and OS (HR=0.48; 95%CI 0.24–0.96; p=0.038), and was independent of Ki67 proliferation status (multivariate DRFS; Ki67<20%, HR=0.41; 95%CI 0.20-0.82; p=0.012; Ki67≥20%, HR=0.31; 95%CI 0.10-0.97; p=0.045). VFOXC1 non-BLBC subtype had no association with clinical outcome in obs arm. Conclusions: In the CIBOMA TNBC trial, the non-BLBC definition by a single biomarker (FOXC1 Veresca) provided prognostic and predictive value of cap benefit after (neo)adjuvant CT, corroborating our previous findings by PAM50 and IHC non-BLBC subtyping. This is a pragmatic option to effectively apply findings from this trial in the real-world setting. Clinical trial information: NCT00130533 .