Clinical performance of ASCL1/LHX8 DNA methylation on first-void urine to screen for cervical cancer

Background: DNA methylation analysis provides a promising triage strategy for cervical intraepithelial neoplasia (CIN) and cancer detection following Human Papillomavirus (HPV) testing on self-collected samples, including urine. Methods: This study aimed to develop an entirely molecular cervical screening approach based on HPV and DNA methylation analysis in at-home collected first-void urine from healthy females (n = 69) and a referral population (n = 385; CIN-cancer). CIN3+ detection was analyzed by multivariate logistic regression. Results: Here we show that urinary ASCL1/LHX8 methylation levels increase significantly in relation to disease severity, with AUC-values for CIN3+ of 0.81 (95% CI: 0.74–0.88) and 0.83 (95% CI: 0.74–0.92) in the training (n = 285) and validation cohort (n = 160), respectively. This corresponds to a validated CIN3+ sensitivity of 73.0% (95% CI: 57.0–84.6%) at 81.9% specificity (95% CI: 73.5–88.1%; <CIN2). Urinary HPV testing is more sensitive (83.8%; 95% CI: 68.9–92.3%) although less specific (59.6%; 95% CI: 50.0–68.5%). For triage of HPV positives, ASCL1/LHX8 methylation and HPV16/18 genotyping have a similar CIN3+ sensitivity (75.0%; 95% CI: 62.8–84.2% vs 73.3%; 95% CI: 61.0–82.9%), with lower genotyping specificity. Combining ASCL1/LHX8 methylation with HPV16/18 genotyping yield a 85.0% sensitivity (95% CI: 73.9–91.9%) at 50.5% specificity (95% CI: 40.8–60.1%). Conclusions: The ASCL1/LHX8 methylation test detects nearly all cancers and a majority of CIN3 in first-void urine, supporting the potential of full molecular screening in urine by primary HPV testing and methylation triage.