Platelet RNA Signature Enables Early and Accurate Detection of Ovarian Cancer: An Intercontinental, Biomarker Identification Study

Abstract

Background: Timely, accurate, and non-invasive detection of ovarian cancer in women with adnexal masses presents a significant clinical challenge. Morpho-physiological alternations accompanied by increased platelet counts was found by the long-term clinical practice and provided a rationale for investigating the potential of platelet RNA profiles to detect ovarian cancer. Methods: In this study, 928 eligible participants (ovarian cancer, n=465; benign adnexal masses, n=296; healthy women, n=167) from three Chinese and six European medical facilities were included between September 2016 and May 2019. A 102-platelet-RNA classifier, named tumour-educated platelet-derived gene panel of ovarian cancer (TEPOC), was developed by the least absolute shrinkage and selection operator (LASSO) and minimum redundancy maximum relevance (mRMR) feature selection, followed a support vector machine (SVM) model in the training cohort (TC) and validated in two domestic (VC1 and VC2) and one European (VC3) external validation cohorts. The performance of TEPOC was compared with that of carbohydrate antigen 125 (CA125), using histological diagnosis as the reference standard. Findings: TEPOC detected OC in VC1, VC2, and VC3 with the area under the receiver operating characteristic curve (AUC) values of 0·923 (0·855–0·990), 0·918 (0·872–0·963), and 0·887 (0·813–0·960), respectively, which were superior to those of CA125 (VC1: 0·735; VC2: 0·836; VC3: 0·824), P < 0.05. Compared to CA125, TEPOC exhibited higher accuracy and specificity and non-inferior sensitivity in detecting early-stage OC, non-epithelial ovarian malignancies, and borderline ovarian tumours, and discriminating OC from endometriosis. Combining the use of CA125 with TEPOC further enhanced its diagnostic performance (AUC, VC1: 0·955; VC2: 0·939; VC3: 0·917). Interpretation: TEPOC enabled early and accurate non-invasive detection of ovarian cancer and harboured robustness, compatibility, and universality that withstood validation across different races, pathological subtypes of ovarian cancer, and other malignancies. Clinical Trials: Chinese clinical trial registry (ChiCTR2100046452). Funding: National Major Science and Technology Project and Natural Science Foundation of China. Declaration of Interest: None to declare. Ethical Approval: This study was approved by the Ethical Committee of Tongji Medical College, HUST (S253) and the Ethical Committee of the LUMC (#P16.229).