Plasma protein and tumor tissue gene expression analyses in ovarian cancer reveals differentially co-regulated clusters between benign and malignant conditions

Ovarian cancer is the deadliest of gynecological cancers and surgery is often necessary for a final diagnosis. Benign cases could be managed more conservatively, avoiding the risks and complications associated with surgery, if accurate diagnostic biomarkers existed. Underlying differences between circulating protein biomarkers and tumor gene expression also restricts interpretation and prioritization of potential biomarkers for diagnosis and potential drug targets. Here, high-throughput affinity plasma proteomics data encompassing over 5400 proteins in plasma from 404 women from two independent Swedish cohorts were analyzed alone and combined with total RNA sequencing in corresponding benign and malignant tumor tissue. A subset of 191 proteins previously identified as differentially expressed between benign and malignant conditions were used to perform correlation analyses, revealing similar patterns between groups but much stronger signals in malignant cases. Comparison with known protein interactions from the STRING database revealed a highly interconnected network consisting of 154 proteins in plasma. Differential correlation analysis (DCA) was performed on the full set of 5414 proteins and for their corresponding tumor RNA expression. DCA identified 31 plasma proteins with significant differential correlations (adjusted p < 0.05, {Delta}R > 0.5) and 759 tumor transcript pairs with significantly differentially correlating RNA expression. Distinct protein-protein correlation patterns in plasma were discovered and validated with notable differences between benign and malignant tumors. In general, these patterns were distinct from those detected on gene expression level in tumor tissue. In conclusion, our findings reveal clear differences in plasma protein co-regulation, with distinct correlation patterns between malignant and benign cases. The differences between results obtained in tumor transcriptomics and plasma proteomics results from the same patients warrants further studies into the tumor microenvironment to understand the function of promising protein biomarker candidates and the potential of these as future drug targets.