Abstract
The histological criteria for classifying endometrial hyperplasia (EH) are based on architectural crowding and nuclear atypia; however, diagnostic agreement among pathologists is poor. We investigated molecular biomarkers of endometrial cancer (EC) risk in women with simple hyperplasia or complex hyperplasia without atypia (SH/CH-nonA). Forty-nine patients with EC preceded by SH/CH-nonA were identified, of which 23 were excluded (15 with complex atypical hyperplasia (CAH), six not consenting, one with a diagnosis <6 months prior, and one lost to follow-up). The EH tissues of these patients were compared with those of patients with SH/CH-nonA that did not progress to EC (control) through microRNA (miRNA) array analysis, and the results were verified in an expanded cohort through reverse transcription-quantitative polymerase chain reaction (RT-qPCR). MiRNA arrays analyses revealed 20 miRNAs that differed significantly (p < 0.05, fold change >4) between the control (n = 12) and case (n = 6) patients. Multiplex RT-qPCR for the 20 miRNAs in the expanded cohort (94 control and 25 case patients) led to the validation of miR-30a-3p (p = 0.0009), miR-141 (p < 0.0001), miR-200a (p < 0.0001), and miR-200b (p < 0.0001) as relevant biomarkers, among which miR-141, miR-200a, and miR-200b regulate the expression of phosphatase and tensin homolog (PTEN). For the prediction of EC, the area under the curve for miR-30a-3p, miR-141, miR-200a, and miR-200b was 0.623, 0.754, 0.783, and 0.704, respectively. The percentage of complete PTEN loss was significantly higher in the case group than in the control group (24% vs. 0%, p < 0.001, Fisher’s exact test). A combination of complete PTEN loss and miR-200a provided optimal prediction performance (sensitivity = 0.760; specificity = 1.000; positive predictive value = 1.000; negative predictive value = 0.937; accuracy = 0.947). MiR-30a-3p, miR-141, miR-200a, miR-200b, and complete PTEN loss may be useful tissue biomarkers for predicting EC risk among patients with SH/CH-nonA.