An angiogenic tumor phenotype predicts poor prognosis in ovarian cancer

Abstract

ObjectiveEpithelial ovarian cancer (OC) is the deadliest gynecological malignancy worldwide. Blocking angiogenesis with bevacizumab, an antibody targeting vascular endothelial growth factor (VEGF), shows efficacy in different lines of OC therapy. This study investigates the clinical impact of tumoral expression of angiogenesis-related genes and their association with bevacizumab response in OC in retrospective analysis of three independent cohorts. MethodsmRNA expression of seven angiogenic genes (VEGF, VEGFR2, PDGFA, PDGFB, PDGFRA, PDGFRB, KIT) was quantified in an inception OC cohort (n = 195) and a transcriptional tumor angiogenesis score from 0 to 3 was established and linked to progression-free survival (PFS) and overall survival (OS). This score was corroborated in an independent publicly available cohort from The Cancer Genome Atlas (TCGA, n = 582) and prediction of therapeutic efficacy of bevacizumab by the angiogenesis score was analyzed in the Gene Expression Omnibus (GEO) dataset GSE140082 (n = 380) from the ICON7-trial. ResultsThe tumor angiogenesis score prognosticated PFS and OS in patients with OC from the inception cohort (p < 0.001, respectively). Tumoral PDGFA expression (PFS: HR 2.46, p = 0.005; OS: HR 2.26, p = 0.011) and a high tumoral transcriptional angiogenesis score (PFS: HR 1.41, p = 0.018) were identified as independent predictors of clinical outcome. The transcriptional angiogenesis score exhibited a significant though smaller effect size on PFS in the TCGA cohort. However, in the ICON7-trial, the angiogenesis score was not associated with benefit of bevacizumab treatment. ConclusionsOur study indicates that tumoral expression of angiogenic genes is unfavorable in OC. The established score could be used to identify patients who respond to targeted angiogenic therapies, a concept that warrants prospective controlled clinical trials.