Circulating extracellular vesicles protein expression for early prediction of platinum-resistance in high-grade serous ovarian cancer

Abstract

Platinum resistance in high-grade serous ovarian carcinoma (HGSOC) portends a poor prognosis. Although initial platinum-based chemotherapy response rates are high, 15-20% of patients demonstrate primary resistance to platinum therapy and almost all patients will develop platinum resistance in the recurrent setting. No predictive or diagnostic biomarkers have been utilized specific to platinum resistance. This study aimed to identify candidate biomarkers for platinum resistance in HGSOC using an extracellular vesicle (EV) based approach. We found differentially expressed and distinct EV proteins, namely TMEM205 and CFH, in patients with platinum-resistant (PR) HGSOC compared to those of platinum-sensitive (PS) patients, utilizing liquid chromatography-tandem mass spectrometry (LC-MS/MS). Expression of these EV proteins were validated in patient-derived PR cell lines as well as in clinically relevant mouse models of HGSOC post-platinum therapy. We corroborated these findings using serum samples from patients with PS and PR-HGSOC. Both EV CFH and EV TMEM205 exhibited excellent diagnostic capability for PR as noted by receiver operating characteristic curves with area under the curve values of 0.95 and 0.84, respectively. The high diagnostic performance of TMEM205 and CFH within EVs compared to the relatively poor performance of conventional serum proteins such as Ca125 suggests their robust potential as non-invasive biomarkers for detecting platinum resistance in HGSOC. Furthermore, the ROC curve for the combined biomarker demonstrated excellent diagnostic performance, with an AUC of 0.973, a true positive rate (TPR) of 0.938, and a false positive rate (FPR) of 0.062. Incorporating this multi-protein biomarker panel alongside established biomarkers further enhances diagnostic accuracy. Serum EV CFH and TMEM205 are promising biomarkers for early detection of platinum resistance in HGSOC and may highlight underlying chemoresistance mechanisms, offering potential future therapeutic targets.

Fig. 1. Increased EVs secretion with proteins in platinum-resistant HGSOC.

A Image Stream Flow Cytometry (ISF) demonstrating the isolation of extracellular vesicles (EVs) using Exo-quick kit and SEC column in patient serum with platinum-sensitive (PS) and platinum-esistant (PR) high-grade serous ovarian carcinoma (HGSOC). B ISF-based quantification of EVs in both PS and PR HGSOC patient serum samples (n = 6, *p < 0.01). C Protein analysis of EVs from platinum-sensitive (PS) and platinum-resistant (PR) HGSOC patient serum samples using LC-MS/MS (n = 6). Candidate proteins were selected based on differential expression with statistical significance (p* < 0.05) in PR-HGSOC compared to PS patients. D EVs isolation by ultracentrifugation in PS and PR cells and ISF-based quantification of EVs in PS (PEOC1) and PR (PEOC4) patient derived cell lines (n = 5, p < 0.04). E Confirmation of selected candidate protein expression by western blot in two distinct patient-derived cell lines representing PS (PEOC1) and PR (PEOC4) phenotypes. F Carboplatin treatment enhances candidate protein expression on the membrane of platinum-resistant HGSOC cells in vitro. A significant increase in the expression of TMEM205, CFH, FAS, STAT3 and CETP in platinum-resistant cells (TR127) following a 24-hour carboplatin treatment, as determined by ELISA (n = 4, *p < 0.05). G Visualization of candidate protein localization in HGSOC platinum-resistant cells (TR127) treated with carboplatin for 24 h, using confocal microscopy.

Fig. 2. Identification of extracellular vesicle (EV) candidate expression during early carboplatin treatment in HGSOC mouse models.

A BALB/c nude mice were intrabursal-injected with platinum-resistant HGSOC cells (TR127), recapitulating the clinical presentation of HGSOC with aggressive ovarian tumor growth and metastatic spread to the mesentery, diaphragm, and pelvic regions. After tumor growth confirmation at the 2nd week, intraperitoneal carboplatin (CP, 2 mg/kg) treatment was administered for 5 weeks. B Mouse serum samples were collected at the end of each treatment cycle before the subsequent CP cycle, and EV protein expression was assessed using ELISA (n = 4; **p ≤ 0.005; *p ≤ 0.01 compared to pre-CP treatment or the 1st cycle of CP treatment). C–E Evaluation of candidate proteins TMEM205, CFH, CD1B, and FAS with established biomarkers CA125, HE4 and Mesothelin in HGSOC patient serum and serum-derived extracellular vesicles (EVs) using ELISA. This analysis was conducted on samples from both PS and PR samples in a training cohort of n = 12 patients, ** revealing statistically significant differences (p < 0.001).

Fig. 3. Identifying the sensitivity and specificity of EVs candidate proteins TMEM205 and CFH.

A, B Receiver Operating Characteristic (ROC) curves depicting the diagnostic performance of CA125, HE4, CFH, and TMEM205, as determined by ELISA results from EVs isolated from serum samples of platinum-sensitive and platinum-resistant HGSOC patients (n = 32). Our candidate proteins, CFH and TMEM205, exhibited an impressive Area Under the Curve (AUC) exceeding 0.95 and 0.84, in contrast to CA125 and HE4 with an AUC of 0.53 and 0.55, respectively, in EV samples. Serum sample AUCs are also presented in (B). C Validation of candidate protein expression (CFH and TMEM205) in EVs derived from serum samples of platinum-sensitive and platinum-resistant HGSOC patients using ELISA (p < 0.0001). D Correlation plot of considered biomarkers (CFH, TMEM205, CA125 and HE4) across all patients. E Evaluation of out-of-sample prediction performance for PS and PR HGSOC patient samples. The optimal biomarker combination, determined by maximizing the smoothed empirical estimate of utility (SHUM), alongside the performance (true positive rate (TPR), false positive rate (FPR), and Mathew’s correlation coefficient (MCC)) of individual biomarkers through leave-one-out (LOO) analysis in the ELISA dataset, is presented. The coefficients of biomarkers within the optimal combination for enhanced diagnostic accuracy are detailed, with bootstrap standard errors (from 500 samples). F Boxplots of the optimal combination vector scores are shown across the outcome categories for all patients. Optimal cut point = 0.173 and Youden index is 0.877 (identified as PR if score is greater than cut-off). G Receiver operating characteristic (ROC) curve for diagnosing PR HGSOC via optimally combined biomarker (AUC = 0.973; TPR = 0.938; and FPR = 0.062).