Cervical cancer is one of the most common gynecological malignancies and it is preventable through the yearly diagnosis and management of pre-cancerous cervical disease. The profile of miRNA expression in cervical epithelium cells is altered with cervical dysplasia development and further progression. The NOVAprep-miR-CERVIX is a new approach for the assessment of cervical dysplasia through the analysis of six marker miRNAs. This study aims to evaluate theperformance and diagnostic potency of the new method. Cytological smears from 226 women (NILM, n.114; HSIL, n.112) were included in the study. A VPH test was performed with RealBest DNAHPV HR screen Kit, six marker miRNAs (miR-21, -29b, -145, -451a, -1246, -1290) were assayed using NOVAprep-miR-CERVIX kit. Obtained data were analyzed using the Delta Ct method and random forest machine learning algorithm. The results of the quantitative analysis of six microRNAs were expressed as a miR-CERVIX parameter, which ranged from 0 to 1, where "0" corresponded to the healthy cervical epithelium, while "1" corresponded to high-grade squamous intraepithelial dysplasia. The average value of miR-CERVIX differed in groups of NILM and HSIL samples (0.34 vs. 0.72; p < 0.000005). An estimation of miR-CERVIX allowed for the differentiation between healthy and pre-cancerous samples with sensitivity of 0.79 and specificity of 0.79, as well as to confirm HSIL with specificity of 0.98. Interestingly, the HSIL group included HPV(+) and HPV(-) samples, which were statistically significantly different in terms of miR-CERVIX value. Analysis of CC-associated miRNAs in material of cervical smear might serve as an additional method for the evaluation of cervical dysplasia severity.
Keywords: RT-PCR; cervical cancer; diagnosis; intraepithelial neoplasia; miRNA; random forest.