CervicalMethDx: A Precision DNA Methylation Test to Identify Risk of High-Grade Intraepithelial Lesions in Cervical Cancer Screening Algorithms.

UNLABELLED: Cervical cancer is one of the most common cancers in women. Despite progress in prevention and success in early detection through cytologic screening and human papillomavirus (HPV) detection, there remains a challenge in triaging women appropriately to colposcopy and biopsy. We sought to validate the CervicalMethDx test, a precision DNA methylation classifier for cervical cancer detection, as a reflex test in women with HPV-positive samples. A blinded retrospective study was performed on well-characterized samples in PreservCyt media from a large referral clinical laboratory in the United States. DNA methylation was assessed in three gene promoters (ZNF516, FKBP6, and INTS1) and a control gene (β-actin) by quantitative real-time methylation-specific PCR (qMSP) analysis, using machine learning algorithms. We compared DNA methylation levels in HPV-positive patients presenting with lesions in the Pap test and cervical intraepithelial neoplasia grade 2 (CIN2) or CIN3 histologic diagnosis with DNA methylation levels in HPV-positive patients with lesions in the Pap test but no intraepithelial lesion or malignancy. The CervicalMethDx test correctly classified 95% of the CIN2 samples (n = 210), with 91% sensitivity, 100% specificity, and an area under the ROC curve (AUC) of 0.96, and 94% of CIN3 samples (n = 141), with 90% sensitivity, 100% specificity, and an AUC of 0.96. Moreover, the CervicalMethDx test correctly classified 94% of combined CIN2/CIN3 samples (n = 351), with 93% sensitivity, 97% specificity, and an AUC of 0.96. CervicalMethDx demonstrated strong discriminatory power for identifying CIN2/CIN3 risk and may complement current triage strategies for colposcopy referral. Prospective, population-based studies, including those in low-resource settings, are needed for further evaluation.

PREVENTION RELEVANCE: The CervicalMethDx test integrates DNA methylation analysis and machine learning to improve early detection of high-grade cervical lesions (high-grade squamous intraepithelial lesions), offering a noninvasive, cost-effective screening tool. Enhanced risk stratification and overtreatment reduction expand equitable access to precision prevention programs. Further validation will clarify CervicalMethDx's alignment with global cervical cancer prevention strategies.