Why are thromboembolic events so common in ovarian cancer? An investigation of tumor-mediated molecular markers (236)

Abstract

<b>Objectives:</b> It is well known that cancer patients with concurrent venous thromboembolism (VTE) have poorer oncologic outcomes. Even though 25-40% of ovarian cancer patients develop a VTE, the role of tumor-specific molecular factors in VTE development is not known. Therefore, we aimed to perform a comprehensive multi-platform omics analysis to identify factors in ovarian cancer that could be related to the development of VTE or the associated impact on patient outcomes. <b>Methods:</b> Patients diagnosed with high-grade serous ovarian carcinoma (HGSOC) between 2013 and 2019 with frozen tissue and matched germline blood samples available were screened for a concurrent diagnosis of VTE. We performed a multiplatform omics analysis, including bulk RNA sequencing (RNA-seq), proteomics, and DNA whole-exome sequencing (WES), followed by integrated analysis on a total of 32 patient samples (<i>n</i>=16, VTE; <i>n</i>=16, non-VTE as controls). Bioinformatics analyses were then carried out to identify potential molecular abnormalities that could contribute to the pathology of VTE in ovarian cancer. <b>Results:</b> The timing of VTE varied across the 16 cases of VTE, ranging from the onset at the time of diagnosis of ovarian cancer (<i>n</i>=9; 56%), during neoadjuvant chemotherapy (NACT; <i>n</i>=2, 12.5%), and within 28 days of surgery or during adjuvant chemotherapy <i>(n</i>=5; 31.3%). The average overall survival (OS) of patients with VTE was 2.05 years (SD: 1.35); however, OS for the control group could not be calculated since more than 80% (<i>n</i>=13, 81.3%) are still alive with the disease after a mean of 3.5 years (SD: 1.07). RNA-seq analysis revealed 130 differentially expressed genes in VTE versus control. Most notably, the <i>F3</i> gene, which encodes tissue factors, was upregulated in nearly all VTE samples compared to controls. A Cox regression using age, grade, and gene expression indicated an association with worse overall survival for <i>GAPDHP65</i> and <i>NRSN2-AS1</i> with hazard ratios of >1.5 (p=0.018 and p<0.001, respectively). On proteomics analysis, we quantified over 7,300 proteins in VTE versus control, of which 255 were significantly altered (p<0.01). Principal component analysis showed that these altered proteins explained 47.5% of the variance between groups. They did not demonstrate a relationship with the onset of VTE timing, suggesting tumor-mediated changes independent of ovarian cancer treatment status. Among the significantly altered putative drug targets, we identified platelet-derived growth factor subunit B (PDGFB) elevated nearly tenfold in the VTE versus the control group. We also observed a profound trend of increased collagen 1A1 and 3A1 in the VTE group. The proteins elevated in the VTE group were biased towards an expression within the extracellular space. Integrated analyses with RNA-seq data indicate a strong correlation of protein and transcript abundance in 64 candidate proteins (Spearman Rho=0.677, p<0.0001). WES analysis revealed predominantly missense mutations in the form of single-nucleotidepolymorphisms (SNPs). Several distinct genes were mutated within the VTE cohort compared to the control group; however, these did not reach statistical significance, likely due to the small sample size. <b>Conclusions:</b> We provide the first multiplatform comprehensive multi-omics analysis of HGSOC based on VTE development. These findings suggest substantial molecular differences in tumors from patients who develop VTE compared to those that do not. These findings could have implications for risk stratification and preventative and treatment strategies for VTE.