782 Survival in advanced-stage ovarian carcinoma with high VEGF-A expression after addition of bevacizumab. Results from the German ICON7 patient cohort

Introduction/Background* It has previously been demonstrated that bevacizumab may differentially improve ovarian cancer survival. TCGA classification as well as molecular targets of anti-VEGF therapy were shown to be associated with different levels of bevacizumab efficacy. Translation into individualized treatment options or improved disease outcome was yet hindered by inconsistent results across trials and tissue types. It was the aim of this project to validate retrospective analyses of GOG-0218 derived predictive value of microvessel density (CD31) and tumor VEGF-A. Methodology CD31 and VEGF-A immunohistochemistry was performed on whole section FFPE tissue samples from the AGO-OVAR11 (ICON7) trial. Patients were stratified into high and low biomarker-expressing subgroups using median cutoffs. The association between biomarker expression and bevacizumab therapy efficacy was evaluated using a proportional Cox regression model. Efficacy endpoints were progression free survival (PFS) and overall survival (OS). Result(s)* Complete CD31 and VEGF-A immunhistochemical data were available from 387 patients of the German ICON7 trial cohort. Among all biomarker subgroups, only patients with high VEGF-A expression levels had a statistically significant benefit from the addition of bevacizumab to standard chemotherapy. Median PFS and OS of VEGF-A high patients was 23.1 months (95% CI 15.9-30.2) and 64.9 months (median not reached) respectively if bevacizumab was added to standard chemotherapy but only 14.3 (95% CI an 11.2-17) and 47.3 months (95% CI 5.6-36.3) in the control arm. In multivariable analysis, adjusted for age, FIGO and postoperative residual tumor, the anti-angiogenic therapy showed improved PFS (HR: 0.62 [95%CI 0.43-0.89], p=0.011) and OS (HR 0.59 [95%CI 0.39-0.91], p=0.02) among VEGF-A high patients. Patients with low VEGF-A expression levels showed no statistically significant improvement of PFS (HR 0.96 95% CI 0.67-1.38, p=0.83) or OS (HR 1.06 95% CI 0.69-1.62, p=0.80) after addition of bevacizumab. CD31 immunohistochemistry was not predictive for bevacizumab treatment effects in our cohort. Conclusion* A potential predictive value of VEGF-A expression levels was observed in advanced stage ovarian carcinoma patients from the German ICON7 patient cohort, partly confirming GOG-0218 derived findings. Our results may help to develop biomarker stratified anti-VEGF therapy and hold potential to promote personalized treatment strategies in ovarian carcinoma patients.