Introduction/Background*The vascular endothelial growth factor (VEGF) plays an import role in emergence and spread of high-grade serous tubo-ovarian carcinoma (HGSTOC). Bevacizumab, a monoclonal antibody targeting VEGFA, has therefore been added to first-line treatment of advanced HGSTOC. We here map the dynamics of different stromal components of the tumour microenvironment under chemotherapy with or without bevacizumab.

Methodology We performed single-cell RNA-sequencing on 62,461 cells sampled from 6 HGSTOC patients before and after neo-adjuvant platin-based chemotherapy with or without bevacizumab. We identified 44 stromal cell subclusters on which we applied Mixed-effects modelling of Associations of Single Cells to identify cell populations associated with bevacizumab exposure and pathological response using the chemotherapeutic response score.

Result(s)*Our study revealed diverse stromal cell subsets associated with bevacizumab exposure. The addition of bevacizumab to frontline chemotherapy increased the odds of endothelial cell (ECs) prevalence by a 3-fold (OR 2.91, 95%CI:2.36-3.58; p<0.001) in comparison to 13-fold when treated with only chemotherapy (OR 13.42, 95%CI:11.29-15.52; p<0.001). Especially for tip cells, essential for vessel sprouting, a negative odds ratio was found for its association with bevacizumab exposure (OR 0.32, 95%CI:0.21-0.50; p<0.001) while chemotherapy only increased the odds of tip cell recruitment (OR 2.67, 95%CI:1.87-3.80; p<0.001). Tip cell receptors KDR, FLT1 and co-receptor NRP1 were significantly downregulated after bevacizumab exposure. In addition, ECs treated with bevacizumab showed lower scores for hypoxia signatures and demonstrated a significant downregulation of hypoxia-induced genes, including HIF1α. Furthermore, bevacizumab was associated with decreased number of regulatory T cells (OR 0.40, 95%CI:0.32-0.51; p<0.001). Interestingly, the addition of bevacizumab was associated with increased influx of tumour-associated macrophages (TAMs). Especially, PDGFC-expressing TAMs were strongly associated with bevacizumab exposure (OR 21.33, 95%CI:9.01-50.46; p<0.001) and poor response (OR 49.06, 95%CI:18.46-130.39 p<0.001). These cells express platelet-derived growth factor-C (PDGFC) and NRP1 providing a possible escape mechanism to activate KDR in absence of VEGFA.

Conclusion*We here provide initial evidence on the mechanisms underlying early response to bevacizumab and frontline chemotherapy in HGSTOC, including tip cell impairment, reduced hypoxia and a decrease in regulatory T cells. However, bevacizumab exposure increased the influx of PDGFC-expressing macrophages capable to bypass VEGFA-dependent angiogenesis.