| Summary: | Ovarian cancer is a highly heterogeneous disease and its formation is affected by many epidemiological factors. It has typical lack of early signs and symptoms, and almost 70% of ovarian cancers are diagnosed in advanced stages. Robust, early and non-invasive ovarian cancer diagnosis will certainly be beneficial. Herein we analysed the regulatory sequence methylation profiles of the <i>RASSF1</i>, <i>PTEN</i>, <i>CDH1</i> and <i>PAX1</i> tumour suppressor genes by pyrosequencing in healthy, benign and malignant ovarian tissues, and corresponding plasma samples. We recorded statistically significant higher methylation levels (<i>p</i> < 0.05) in the <i>CDH1</i> and <i>PAX1</i> genes in malignant tissues than in controls (39.06 ± 18.78 versus 24.22 ± 6.93; 13.55 ± 10.65 versus 5.73 ± 2.19). Higher values in the <i>CDH1</i> gene were also found in plasma samples (22.25 ± 14.13 versus 46.42 ± 20.91). A similar methylation pattern with positive correlation between plasma and benign lesions was noted in the <i>CDH1</i> gene (<i>r</i> = 0.886, <i>p</i> = 0.019) and malignant lesions in the <i>PAX1</i> gene (<i>r</i> = 0.771, <i>p</i> < 0.001). The random forest algorithm combining methylation indices of all four genes and age determined 0.932 AUC (area under the receiver operating characteristic (ROC) curve) prediction power in the model classifying malignant lesions and controls. Our study results indicate the effects of methylation changes in ovarian cancer development and suggest that the <i>CDH1</i> gene is a potential candidate for non-invasive diagnosis of ovarian cancer.
|
|---|
