Genome-wide DNA methylation profiling identifies two novel genes in cervical neoplasia.

ABSTRACT

DNA methylation analysis may improve risk stratification in cervical screening. We used a pan-epigenomic approach to identify new methylation markers along the continuum of cervical intraepithelial neoplasia (CIN) to cervical cancer. Physician-collected samples (54 normal, 50 CIN1, 40 CIN2 and 42 CIN3) were randomly selected from women at a single-center colposcopy clinic. Extracted DNA was subjected to Illumina Infinium EPIC array analysis, and methylation was assessed blinded to histopathological and clinical data. CpG sites whose state of methylation correlated with lesion grade were assessed (Spearman correlation), and a weighted methylation score was calculated comparing normal to CIN3. Validation of the top selected genes was performed in an independent cohort (100 normal, 50 CIN1, 50 CIN2, 50 CIN3 and 8 cervical cancers) of new patients, referred for colposcopic examination at three hospitals, using targeted DNA methylation Illumina amplicon sequencing. The relationship between a combined weighted marker score and progression from normal through precancerous lesions and cervical cancer was compared using one-way ANOVA. Our analyses revealed 7,715 CpGs whose methylation level correlated with progression (from normal to CIN1, CIN2 and CIN3), with a significant trend of increased methylation with lesion grade. We shortlisted a bigenic (hyaluronan synthase 1, HAS1 and ATPase phospholipid transporting 10A, ATP10A corresponding to cg03419058 and cg13944175 sites) marker set; r = 0.55, p < 0.0001. Validation of the four most discriminating genes (CA10, DPP10, FMN2 and HAS1) showed a significant correlation between methylation levels and disease progression (p-value < 2.2 × 10-16 , adjusted R2 = 0.952). Translational research of the identified genes to future clinical applications is warranted.