88: Assessing the risk of concurrent endometrial cancer among women with endometrial hyperplasia

Abstract

To identify the independent risk factors of concurrent endometrial cancer among women with endometrial hyperplasia and build the clinical prediction model incorporating the clinical and pathological markers. This cross-sectional retrospective observational study included data from 379 women who had a hysterectomy at the two institutes, Sisters of Charity Hospital and Roswell Park Cancer Institute, of Buffalo, New York from January 2005 to December 2016 following a diagnosis of endometrial hyperplasia with or without atypia. All women with a diagnosis of endometrial hyperplasia who subsequently received hysterectomy with or without staging operation were included in the study and analyzed. Women with a preoperative diagnosis of endometrial cancer prior to hysterectomy were excluded from the study. The variables chosen for the multivariable logistic regression analysis were ‘endometrial hyperplasia (with atypia vs. without atypia), age (≤ 53 vs. > 53), BMI (≤ 35 vs. > 35), the length of endometrial cavity (< 7cm vs. ≥ 7cm), the progestin treatment before hysterectomy (yes vs. no), smoking, diabetes, hypertension, and hyperlipidemia. Then, a stepwise backward regression model to determine the predictors of endometrial cancer was performed. The performance of predictive models was assessed with receiver operating characteristic (ROC) curve and Hosmer-Lemeshow test. Fifty-five (14.51%) of a total of 379 women with endometrial hyperplasia was diagnosed with endometrial cancer after hysterectomy. In bivariate regression analysis, the risk factors significantly associated with concurrent endometrial cancer were atypical endometrial hyperplasia, age ≥ 50, postmenopausal bleeding, the use of progestin, diabetes mellitus, hypertension, and hyperlipidemia. In adjusted multivariate logistic regression analysis, the presence of atypical endometrial hyperplasia was associated with a significant increase in the risk of being diagnosed with concurrent endometrial cancer in the final pathology of hysterectomy specimen (OR, 56.708, 95% CI 7.601-423.056). Using a stepwise backward multivariate regression (alpha set at 0.1) analysis, the model with atypical endometrial hyperplasia, hypertension, smoking, and postmenopausal bleeding was constructed. This model predicted the risk of being diagnosed as endometrial cancer better, in comparison to the model with atypical endometrial hyperplasia alone or the model controlling simultaneously for the variables selected in the bivariate analyses indicated above (stepwise backward regression model, AUC=0.8749; the model with atypical endometrial hyperplasia, AUC=0.7983; multivariate regression model, AUC=0.8696, p-value = 0.0001). The addition of postmenopausal bleeding, hypertension, and smoking to atypical endometrial hyperplasia in the risk prediction model improves the risk assessment of concurrent endometrial cancer.View Large Image Figure ViewerDownload Hi-res image Download (PPT)View Large Image Figure ViewerDownload Hi-res image Download (PPT)