REM (risk of endometrial malignancy): a proposal for a new scoring system to evaluate risk of endometrial malignancy.

Abstract

It is often difficult to distinguish a benign endometrial disease from a malignancy and tools to help the physician are needed to triage patients into high and low risk of endometrial cancer. The purpose of this study was to obtain a predictive model to assess the risk of endometrial malignancy (REM) in women with ultrasound endometrial abnormalities. Women, between ages 45 to 80 years, diagnosed through ultrasound with endometrial abnormalities and scheduled to have surgery were enrolled on a prospective study at the Department of Gynaecologic Oncology of Campus Bio-Medico, University of Rome. Preoperative clinical, ultrasound and laboratory characteristics were taken into account. Logistic regression algorithm was used to categorize patients into low- and high-risk groups for endometrial cancer. A total of 675 patients were considered for the analysis: 88 with endometrial cancer and 587 with benign endometrial disease. We divided the patients into two groups: training set (TS) and verification set (VS). Preoperative age, symptom, HE4 levels, and ultrasound endometrial thickness were found statistically significant, and were included into a multivariate logistic regression model to determine the probability to have endometrial cancer. In the TS, REM reported 93.3% sensitivity and 97.1% specificity [positive predictive value (PPV), 0.83; negative predictive value (NPV), 0.98; AUC, 0.957; 95% confidence interval (CI), 0.908-0.984]. In the VS, REM reported 89.3% sensitivity and 95.4% specificity (PPV, 0.73; NPV, 0.98; AUC, 0.919; 95% CI, 0.829-0.970). Our data support the use of REM to triage patients into low- and high-risk groups for endometrial cancer, even if an external validation of the model is needed.