A proteomic analysis of differential protein expression between platinum-sensitive and platinum-resistant high grade serous ovarian cancer

Introduction: Ovarian cancer is the most lethal gynecologic malignancy. Platinum-based therapy is the mainstay of treatment, with 70% chance of treatment sensitivity. Resistance to platinum therapy is the leading cause of mortality in advanced ovarian cancer. There are limited studies using proteomics to evaluate platinum resistance in ovarian cancer. This study aims to perform proteomic evaluation of treatment naive tissue from high grade serous ovarian cancer (HGSOC) tumors that are platinum-sensitive (PS) versus resistant (PR). Methods: Formalin-fixed paraffin-embedded (FFPE) tumor tissue collected from 63 patients with HGSOC were analyzed, including 40 PS & 23 PR. Platinum sensitivity was defined as disease response without evidence of recurrence after 6 months of treatment. Samples were analyzed using a high-throughput liquid chromatography-tandem mass spectrometry (LC-MS/MS) protocol. Differential expression (DE) analysis was performed to identify proteins distinguishing between the two groups. Unsupervised machine learning techniques such as least absolute shrinkage and selection operator (LASSO) regression analysis, linear discriminant analysis (LDA), and weighted gene co-expression network analysis (WCGNA) were employed to create a differentially expressed protein profile between PS and PR. Results: Among 4048 proteins identified, 1,604 met quality thresholds for further analysis, appearing in >50% of samples. Thirty-seven proteins were significantly differentially expressed with four upregulated in PS and 33 upregulated in PR. With LASSO regression analysis, an additional 61 DE proteins were identified. LDA revealed 7 proteins (DPM1, INF2, ISYNA1, RBM12B, ATP5F1C, GNL1, UBA7) to have a predictive potential with 85% sensitivity and 100% specificity. Functional enrichment analysis implicated pathways related to RNA binding, epigenetic regulation, spliceosome activity, glutathione metabolism, and metabolic reprogramming in platinum sensitivity. Conclusion: With the combination of traditional statistics and unsupervised machine learning, this study generated a list of DE proteins and facilitated a 7-protein panel that can be used to further investigate and understand the complex tumor dynamics behind platinum sensitivity and resistance.