Bevacizumab (BEV) continued beyond first progression in patients (pts) with metastatic colorectal cancer (mCRC) previously treated with BEV + chemotherapy (CT): Biomarker findings from ML18147.

Abstract

452 Background: In ML18147, BEV + standard CT was continued as 2nd-line treatment for pts with mCRC progressing after 1st-line BEV + CT. ML18147 met its primary and secondary endpoints and had a comprehensive biomarker programme, including exploratory analyses of immunohistochemistry (IHC) and single nucleotide polymorphisms (SNPs). Methods: Tumour samples were analysed by IHC for VEGFA, VEGFR1, VEGFR2 and neuropilin-1; H-scores were generated. Tumor angiogenesis was determined by stereological analysis of CD31-stained microvessels; 183 SNPs of 30 mainly angiogenesis-related genes were genotyped. We assessed consistency of treatment effect point estimates, i.e. hazard ratios (HRs) on the same side of 1. Analyses were not adjusted for multiple testing or powered to detect statistically significant treatment by biomarker interactions. Results: IHC data were available for 213 pts (26% of ITT population) and SNP data for 274 pts (33% of ITT). There was no significant evidence of biomarker expression levels or their cellular localisation correlating with BEV treatment effect. Point estimates for a CD31 derivative (i.e. vessel or tumour area) suggested a trend for being predictive for overall survival (OS; HR 1.39 and 0.70 for CD31 derivative <median and ≥median, respectively), similar to data from previous studies. Other previously reported correlations (progression-free survival [PFS] with VEGFA and NRP1 in NO16966; OS with NRP1 in AVAGAST) were not seen. SNP analyses confirmed previous correlations for OS (VEGFA, VEGFR2) and suggested additional biomarker candidates (PFS: VEGFA, VEGFR1, VEGFR2, EGLN3; OS: VEGFR2, EGLN1). Five EPAS1 (HIF2α) SNPs were associated with BEV treatment effect, thus confirming previous findings of HIF2α SNPs as a predictive biomarker for BEV treatment effects. Conclusions: Exploratory IHC and SNP analyses in ML18147 revealed potential biomarker candidates for BEV in pts with mCRC. SNP analyses of angiogenesis-relevant genes were consistent with previous findings for a subset of SNPs, consolidating their role as potential predictive biomarkers of BEV treatment across indications. Clinical trial information: NCT00700102.