Identification of a novel hypermethylation marker, ZSCAN18, and construction of a diagnostic model in cervical cancer

Abstract Background Cervical cancer is a common malignant tumor in women that seriously threatens women’s lives and health. DNA methylation markers have been reported to be related to cervical cancer. Therefore, to find more biomarkers, we adopt a “dry- and wet-lab” strategy that combines bioinformatics, machine learning and experimental methods for novel methylation biomarker identification. Results According to the data of cervical cancer samples from TCGA and GEO, 18 differentially methylated CpGs (DMCs) were obtained by integration analysis of the methylome and transcriptome. After assessment through the ROC curve in both the identification cohort (n = 349) and validation cohort (n = 414) from datasets, 5/18 top CpG sites were obtained as potential methylation biomarkers. Subsequently, a series of validation experiments were performed on 3/5 CpG sites which were all located in the promoter of ZSCAN18. First, pyrosequencing results showed that the methylation level of the ZSCAN18 promoter was significantly higher in cervical cancer tissues than in normal tissues (△β value > 0.33, P < 0.05). Then, the methylation PCR (MSP) results showed that the hypermethylation rate in cervical cancer tissues was 80%, which was significantly more than 30% in normal tissues ( P < 0.05). Eventually, the quantitative methylation PCR (QMSP) results in cervical thinprep cytologic test (TCT) samples of different lesion stages showed that both the level and positivity of ZSCAN18 methylation increased with the grade of cervical lesions, and the positivity rate was up to 77.8% (21/27) in cancer samples. Further diagnosis models showed that the ridge regression model (RR) had the best performance of the six machine learning models, with AUC areas of 0.9421 and 1.0000 in the validation and mock test cohorts, respectively. Functional analysis demonstrated that overexpression of ZSCAN18 repressed the proliferation of cervical cancer cells ( P < 0.05). Conclusions In this study, we established a rapid, effective and systemic research strategy to screen novel methylation markers for cervical cancer. The level of ZSCAN18 promoter methylation increases with the severity of cervical lesions and can be used as a DNA methylation biomarker for cervical cancer. The diagnostic model can improve the diagnostic ability.