Analysis of risk factors and construction and validation of a predictive model for determining the risk of endometrial cancer in postmenopausal patients with abnormal uterine bleeding

Abstract

The data of 174 postmenopausal patients with abnormal uterine bleeding admitted were assessed to determine associated risk factors and develop and validate a prediction model to evaluate the risk of endometrial cancer in these patients. The patients were divided into a study group and a control group, among which 62 patients were diagnosed with endometrial cancer. A binary logistic regression analysis model using multifactorial regression analysis was established, and a column line graph of the prediction model was created using the R software. The model’s goodness-of-fit test was performed using the Hosmer-Lemeshow test, and SPSS (version 27, International Business Machines Corporation, Armonk, NY, USA) was used to plot the receiver operating characteristic (ROC) curve to evaluate the model’s predictive value. Binary logistic multifactorial regression analysis revealed that elevated body mass index (BMI), human epididymal protein 4 (HE4), cancer antigen 125 (CA125), combined fibroids and thickened endometrial cancer were risk factors for endometrial cancer in patients with abnormal postmenopausal uterine bleeding, based on which a probability model for predicting the risk of developing endometrial cancer in patients with abnormal postmenopausal uterine bleeding was constructed, and represented as P = 1/[1 + exp (4.227 − 4.594X1 − 2.029X5 − 1.165X6 − 1.817X7 − 2.080X8)]. In addition, the goodness-of-fit test, assessed using Hosmer and Lemeshow, yielded an χ2 value of 14.253 and a p-value of 0.075. Furthermore, the ROC curve analysis demonstrated an area under the curve (AUC) of 0.993 (95% confidence interval (CI), 0.892–0.974; p < 0.05). In conclusion, elevated BMI, HE4 and CA125, along with the presence of combined fibroids and thickened endometrial lining, were identified as significant risk factors for endometrial cancer in postmenopausal patients with abnormal uterine bleeding. The risk prediction model developed in this study provides a scientifically sound approach to assess the risk of endometrial cancer in these patients.