Predictive ability of novel glycovariant biomarkers of CA125 and CA15-3 up to three years prior to ovarian cancer diagnosis: a population-based case-control study

Abstract Background Nanoparticle immunoassays for CA125 and CA15-3 glycovariants are promising tools for epithelial ovarian cancer (EOC) and borderline ovarian tumor (BOT) early detection. Method This retrospective population-based study utilized prospective (< 3 years) plasma samples from 110 cases and 440 controls. Sialyl-Thomsen-nouveau (STn) antibody and macrophage galactose-type lectin (MGL) to detect cancer antigen 125 (CA125) and 15 − 3 (CA15-3) glycoforms were evaluated, using CA125, CA15-3, and HE4 enzyme immunoassay (EIA) references. The area under the receiver operating characteristic curve (AUC), partial AUC (pAUC) at specificities 0.9–1.0, and sensitivity (SN) at 0.98 specificity (SP) for detecting EOC/BOT were calculated. Results The single marker with the highest point estimate for pAUC and SN at 98% SP was CA125 EIA (pAUC 0.71 (0.66, 0.76), SN at 98% SP 0.29 (0.17–0.42 combinations)). The CA15-3 STn + CA125 EIA and CA125 EIA + HE4 EIA improved the detection rate point estimates and provided the highest pAUC values (CA125 EIA + CA15-3 STn : 0.75 (0.70–0.80); CA125 EIA + HE4 EIA : 0.71 (0.66–0.77)). For 0–<1, 1–<2 and 2–<3 years lag time, CA125 EIA and the CA125 glycovariants provided similar pAUC and SN at 98% SP, with largely overlapping confidence intervals. Conclusion This case-control study, which used plasma samples collected up to three years before diagnosis of EOC or BOT, did not provide evidence of a higher discriminatory capacity of the glycovariant biomarkers compared with the clinically established biomarker CA125 EIA in asymptomatic women. Increased plasma volumes and larger cohorts are suggested in future studies.