Vascular endothelial growth factor A (VEGF-A) amplification and long-term response to ramucirumab (ram) in metastatic gastric cancer (mGC): The VERA study

3143 Background: The anti-VEGFR-2 monoclonal antibody ram, alone or with paclitaxel, is a cornerstone of second-line treatment of mGC. Even if about half patients do not benefit from ram, no predictive biomarkers have been identified so far. In TCGA, VEGF-A amplification was found in 7% of cases, almost exclusively in chromosomal instability subtype. We hypothesize that VEGF-A amplification in tumor cells could lead to autocrine/paracrine stimulation of tumor growth beside angiogenesis, potentially identifying a patients’ subgroup with exceptional responses to ram. Methods: VERA was a multicentric, prospective study based on a translational hypothesis. mGC patients were included according to the following criteria: 1) complete (CR) or partial response (PR) to single-agent ram; 2) >6 months PFS to single-agent ram; 3) >10 months PFS to paclitaxel+ram. According to a Fleming single-stage design, hypothesizing a prevalence of VEGF-A amplification of 1% and 15% among all-comers and exceptional responders, 20 exceptional responders were required to reject the null hypothesis of low prevalence of VEGF-A amplification, with alpha- and beta- errors of 0.05 and 0.10, respectively. VEGF-A amplification (defined as >10% tumor cells with ≥10 VEGF-A copies, variably sized signal clusters or a ratio of VEGF-A gene to centromere of ≥2) was centrally assessed through fluorescent in situ hybridization on pre-treatment FFPE tumor tissue. Results: At 7 Italian Centers, we included 20 patients satisfying the 1st (n=1), 2nd (n=2), or 3rd (n=17) criterion. Clinical-pathological features were: M/F, 11/9; median age 63 years; gastric/GEJ, 17/3; intestinal/diffuse, 14/6, HER2+/HER2-, 4/16. Median PFS and overall survival to ram-based treatment were 15.6 and 25.7 months, with best response: CR/PR/SD, 0/10/10. VERA met its primary endpoint, revealing 3/20 (15%) tumors with VEGF-A amplification (1 case presenting big clusters, 1 small clusters and 1 with >10% tumor cells with ≥10 VEGF-A copies). Conclusions: Validation analyses of first- and second-line randomized trials could confirm VEGF-A amplification as a biomarker of long-term response to ram-based treatment in mGC patients, advancing treatment personalization.