Multiomics characterization of in vivo resistance to anti-VEGF therapy in ovarian carcinoma

Epithelial ovarian cancer (OC) is the most lethal gynecologic malignancy in developed countries. Due to the lack of a screening strategy, the disease is usually diagnosed at an advanced stage and has a poor prognosis. In the advanced stage, antiangiogenic therapy combined with platinum-based chemotherapy represents a first-line clinical treatment option, but despite initial clinical benefit, resistance eventually occurs. The aim of this study is to investigate the mechanisms behind anti-VEGF therapy resistance in OC, by generating and analyzing multiomics data obtained from patient-derived xenografts (PDXs). We used a panel of seven PDXs, originally derived from the ascitic fluid of high-grade serous OC (HGSOC) patients and propagated in immune deficient mice. To a subset of the mice in each PDX group, we administered bevacizumab, a monoclonal antibody targeting VEGFA. To comprehensively profile the treated and wild-type mice within each group, we performed RNA-Seq, mass spectrometry (MS) proteomics, and whole exome sequencing (WES). In line with clinical data, the in vivo experiment reveals two mechanisms of resistance: three PDXs were intrinsically resistant to anti-VEGF therapy and the other four PDXs responded in the beginning but eventually developed resistance. To understand the reasons of these two mechanisms, we explore and integrate the three different layers of omics data. We analyze the mutational profiles of the different samples and identify genes differentially mutated in treated vs wild-type samples, and employ a proteogenomics approach to enhance the accuracy of peptide identification from MS data. We also examine the transcriptomics and proteomics profiles across tumors and replicates and identify genes potentially involved in the mechanisms of resistance. Genomic data reveals high mutational rates in all PDX models, as described in other studies on HGSOC. Mutations are particularly enriched in MUC, HLA and T53 genes. No significant mutations are enriched or selected during antiangiogenic therapy, suggesting that methylation might play an important role. Transcriptomics and proteomics data show that non-responders" models have a poor modulation at the transcriptome and proteome levels. Conversely, acquired resistant tumors display a wide transcriptome and proteome rewiring.

Our data indicates that acquired resistance models have a strong upregulation of interferon signatures at both transcriptomics and proteomics levels. In addition, acquired resistance models downregulate cell cycle and DNA packaging signatures as a mechanism of drug adaptation. RNA splicing analysis uncovers that one model is producing a splicing VEGFA variant as a mechanism to escape VEGFA inhibition, while other models adapt to antiangiogenic therapy by modulating their metabolism. After disclosing the mechanisms of resistance, we will validate our findings in vivo through combination treatment and survival analysis.