Performance of DNA methylation analysis in the detection of high-grade cervical intraepithelial neoplasia or worse (CIN3+): a cross-sectional study

Abstract

It is commonly accepted that host genes show high methylation in cervical intraepithelial neoplasia 3 (CIN3) or worse (CIN3+). However, study quality varies, as does the clinical performance of markers in different populations. We aimed to validate candidate gene DNA methylation with standardized testing methods in the same batch of samples. We first compared the performance of 16 DNA methylation markers for detecting CIN3+ in the 82-sample training set, including 24 subjects with ≤ CIN1, 10 subjects with CIN2, 23 subjects with CIN3, and 25 subjects with cervical cancer (CC). Then five methylation markers were selected and subsequently validated among an independent set of 74 subjects, including 47 subjects with ≤ CIN1, 13 subjects with CIN2, 6 subjects with CIN3, and 8 subjects with CC. The results in the validation set revealed that methylation analysis of the SOX1 (SOX1^m) showed a superior level of clinical performance (AUC = 0.879; sensitivity = 85.7%; specificity = 90.0%). SOX1^m had better accuracy than cytology, with a reduced referral rate (23.0% vs. 31.4%) and a lower number of overtreatment (5 vs. 13) cases among high-risk human papillomavirus (hrHPV)-positive women. Importantly, among hrHPV-positive and SOX1^m-negative women, only 1 CIN3 patient was at risk for follow-up after 1 year, whereas 1 CIN3 patient and 1 CC patient were at risk among hrHPV-positive and cytology-negative women. In this investigation, we screened 16 reported methylation markers to provide a basis for future studies related to potential precancerous lesion/cancer methylation markers in the Chinese population. The study also revealed that SOX1^m has optimal CIN3+ detection performance, suggesting that it may be a promising biomarker for detecting CIN3+ in the Chinese population.