CervicalMethDx : Precision risk stratification of high-risk HPV samples to reduce unnecessary colposcopy referrals and improve the quality of value-based care

Oncogenic human papillomavirus (hrHPV) screening identifies many transient infections, but few clinically significant precancers, generating high rates of unnecessary colposcopy referrals. In the United States, 70-80% of women referred for hrHPV-positive colposcopy lack actionable precancerous diagnoses. Given that >90% of hrHPV infections clear naturally within 12-24 months, there is an urgent need for objective molecular triage tools to optimize colposcopy-driven biopsy selection and enhance value-based cervical cancer care. The CervicalMethDx precision methylation platform was evaluated using 1,618 samples from IRB-approved studies across Puerto Rico, Latin America, and the United States. Independent datasets and specimen types were tested, including Digene and PreservCyt media, as well as self-collected vaginal swabs. Diagnostic performance metrics (sensitivity, specificity, AUC) were compared across CervicalMethDx assay versions (1.0 singleplex, 1.5 six-gene panel, 2.0 duplex). Cross-assay reproducibility and anatomic-site concordance were assessed using correlation and generalized estimating equation (GEE) models. CervicalMethDx 1.0 detected cervical intraepithelial neoplasia grade 2 or higher (CIN2+) lesions with 96% sensitivity, 84% specificity, and an area under the ROC curve (AUC) of 0.99 in hrHPV-positive Digene samples (n = 108, University of Puerto Rico). In the ESTAMPA Latin American trial (n = 759), CIN3+ detection reached 67% sensitivity, 59% specificity, and an AUC of 0.65. The CervicalMethDx 1.5 six-gene panel identified squamous intraepithelial lesion (SIL) cytology with 94% sensitivity, 57% specificity, and an AUC of 0.85 in CLIA-certified PreservCyt samples (n = 377). No significant methylation differences (p > 0.05) were observed between paired cervical and self-collected vaginal swabs (n = 100 each), demonstrating strong intra-individual concordance. ZNF516 methylation correlated across singleplex and duplex assays (r = 0.51, p = 0.01). GEE modeling showed a negative within-Pearson correlation (ρ = −0.64) between paired cervical and vaginal sites, confirming consistent site-dependent variation and reproducibility. CervicalMethDx provides a robust precision-methylation approach for risk stratification of hrHPV-positive samples with the potential to reduce unnecessary colposcopy referrals and biopsies. The assay's reproducibility across self-collected and clinician-collected specimens supports its potential integration into at-home hrHPV testing workflows, advancing equitable and value-based cervical cancer screening worldwide