B-308 Analytical and clinical performance of a novel ovarian cancer screening test in asymptomatic women

Abstract

Abstract Background Ovarian cancer (OC) ranks among the deadliest gynecological cancers, often diagnosed at advanced stages when treatment options are limited. Costly sequencing methods and established immunoassay and imaging-based methods lack early-stage sensitivity, highlighting the need for a cost-effective test with improved performance. Extracellular vesicles (EVs) produced in abundance by all cells can potentially improve test sensitivity and detection of multiple colocalized cancer specific markers can improve test specificity. The randomized controlled trial UKCTOCS (United Kingdom Collaborative Trial of Ovarian Cancer Screening) where over 200,000 postmenopausal asymptomatic women were randomly assigned to three arms revealed no significant reduction in mortality with OC screening. We report on an EV-based OC Test performance in serum samples from asymptomatic postmenopausal women in UKCTOCS. Methods The OC Test was previously locked (biomarkers, method, cutoff) and validated in an independent cohort. We evaluated the OC Test in a blinded case-control study nested within the no screening (NS) and annual ultrasound screening (USS) arms of UKCTOCS. All UKCTOCS participants provided a blood sample at study entry. Cases were all women who developed high grade serous ovarian cancer (HGSOC) within 36 months following sample donation (spun within 28 hours). They were matched by age at collection (+/-2 years) to women who had no OC detected either before or within the trial follow up period (controls). Where a case had >1 sample within the 0-36 months period, the control sample was matched to the case sample closest to diagnosis. The OC Test was run with a locked method and classifier algorithm utilizing 3 independent biomarker combinations tested in duplicate. OC Test results were calculated based on a previously determined cutoff prior to unblinding of clinical outcomes. Run controls were prepared by harvesting conditioned media from COV413A cell cultures (an ovarian epithelial-serous carcinoma cell line), purifying EVs and spiking into pre-screened OC Test negative human plasma at three levels. Run controls were run twice per day, one replicate per run for 17 days. Results We retrospectively assessed 116 samples from 108 cases (63 clinical Dx NS, 15 clinical Dx USS, 30 screen detected USS) women with HGSOC and 1310 samples from 1310 controls. OC test sensitivity for HGSOC was 84% (36/43; 95% CI 69%-93%), Stage I/II sensitivity was 89% (8/9; 95% CI 51%-99%) and the AUC (area under the curve) was 0.93 for case samples drawn 0-12 months prior to clinical Dx and all control samples. The OC Test specificity was 98% (1099/1176, 95% CI 97%-98%). Within laboratory CVs were 20.9%, 12.7% and 0.6% for low negative, high negative and high positive run controls respectively. Data from 143 samples (10%, 9 cases, 134 controls) were excluded due to operator error, low sample volume or quality issues. Conclusions The OC Test is capable of highly sensitive detection of HGSOC within one year of clinical diagnosis including at the earliest stages of disease while maintaining exceptionally high specificity. The results demonstrate the potential application of the OC Test for ovarian cancer screening in asymptomatic women.