Analysis Of Risk Factors For Postmenopausal Endometrial Cancer And Establishment Of Nomogram Prediction Model

Background:Endometrial cancer（EC）is one of the common malignant tumors of female genital tract.Although most EC are diagnosed at an early stage and have a good prognosis,20-30% of patients are found at an advanced stage with a poor prognosis.About 70-75% of EC occurs in postmenopausal women.With the gradual deepening of the aging of the population in China,the incidence of postmenopausal EC is increasing year by year.Therefore,establishing a reasonable and effective screening method for postmenopausal EC to identify high-risk groups,and then the early diagnosis and treatment of EC is the key to reducing its mortality.Objective:In this study,by analyzing the risk factors of EC in postmenopausal women,a nomogram prediction model was constructed to provide an economical and effective screening method for postmenopausal EC,which improved the early diagnosis rate and survival rate,and reduced unnecessary invasive diagnosis operation.Methods:1.Retrospective analysis was performed on the clinical data of postmenopausal EC patients（n=310）and non-EC patients（n=248）admitted to the Hospital from January 2018 to February 2022,which were divided into training cohort（n=444）and validation cohort（n=114）according to the visit time.2.Univariate logistic analysis was performed on the general clinical data,laboratory serological and imaging clinical indicators of the of patients in the training cohort（P<0.05）.Then statistically significant risk factors were selected out for LASSO regression screening.The significant variables in LASSO regression were included in multivariate logistic analysis and the independent risk factors for EC in postmenopausal women were finally screened out（P<0.05）.3.Use R4.1.2 software to build the nomogram prediction model of postmenopausal EC risk.The predictive performance of the model was evaluated by ROC curve and AUC.The fitting degree,calibration degree and the clinical value of the model were evaluated by Hosmer-Lemeshow test,calibration curve respectively and DCA curve,respectively.Finally,internal validation（Bootstrap method with selfsampling 1000 times）and external validation（validation cohort）were used to verify the prediction performance and clinical practice value of the model.Results:1.Through univariate and multivariate logistic analysis,vaginal bleeding（OR=10.39,95%CI: 6.13-17.61,P<0.001）,diabetes（OR=2.18,95%CI: 1.16-4.12,P=0.016）,total cholesterol>5.17mmol/L（OR=1.95,95%CI: 1.11-3.43,P=0.020）,CA125>35U/ml（OR=2.22,95%CI: 1.06-4.64,P=0.034）,CA199>37U/ml（OR=3.52,95%CI: 1.59-7.79,P=0.002）,endometrial thickness≥0.5cm（OR=2.18,95%CI:1.30-3.64,P=0.003）were independent risk factors for endometrial cancer in postmenopausal women.2.Based on BMI,high blood pressure and other 6 risk factors mentioned above,a nomogram prediction model for the risk of postmenopausal endometrial cancer was successfully constructed.The area under the ROC curve（AUC）of the model in the training and validation cohort was 0.868（95% CI: 0.834-0.901）and0.816（95%CI:0.740-0.893）,respectively,which indicates that the model has good predictive performance for the risk of postmenopausal EC.Setting the cutoff value of the model in the ROC curve of the training cohort as 0.619,the sensitivity and the specificity is78.9% and 81.9%,respectively.The P value of the model Hosmer-Lemeshow test in the training and validation cohort was 0.591 and 0.918,respectively,which means the good fitting degree.The calibration curves of the model in the training and validation cohort showed that the model predicted the risk of postmenopausal endometrial cancer in good agreement with the actual risk.The DCA of the model in the training and validation cohort showed that the model has certain clinical practice value.Conclusion:1.Vaginal bleeding,diabetes,total cholesterol>5.17mmol/L,CA125>35U/ml,CA199>37U/ml,and endometrial thickness≥0.5cm were independent risk factors for endometrial cancer in postmenopausal women.2.The constructed nomogram prediction model of postmenopausal endometrial cancer incidence risk has a good predictive ability and a clinical practice value for the early screening of the disease.This model provides a convenient,non-invasive,economical and effective screening method for postmenopausal endometrial cancer,which has important social significance.