Diagnostic Efficacy And Prediction Model Construction Of Ultrasonography And Blood Indexes For Endometrial Carcinoma Caused By Abnormal Uterine Bleeding Before Menopause

Objective:By discussing the clinical feature,metabolic indexes,inflammatory indexes and tumor indexes of premenopausal women with abnormal uterine bleeding in combination with the results of pelvic ultrasound examination,the risk factors of endometrial cancer were identified,the diagnostic efficacy of each risk factor in endometrial cancer was evaluated,and the risk profile model of endometrial cancer caused by abnormal uterine bleeding in premenopause was constructed.To provide reference for the early diagnosis of endometrial cancer.Methods:By retrospective analysis,a total of 495 cases were selected in the first hospital of Lanzhou University from January 2018 to January 2022 with premenopausal patients with abnormal uterine bleed,246 patients with pathologically confirmed endometrial malignancies as the study group and 249 patients with benign endometrial lesions as the control group through the standard.Basic clinical data,metabolic indexes,inflammatory indexes,tumor indexes and pelvic ultrasound results were collected.SPSS 27.0 statistical software was used to compare and analyze the indicators of the two groups to identify the independent risk factors for endometrial cancer caused by abnormal uterine bleeding before menopause.The Receiver operating characteristic curve(ROC)was drawn to evaluate the diagnostic effectiveness of each indicator on endometrial cancer,and the risk profile model of endometrial cancer caused by abnormal uterine bleeding before menopause was established by R language.The ROC was drawn to evaluate the predictive value of the model.The accuracy of the model was tested by calibration curve.The clinical practical value of the line graph model was further verified by Decision curve analysis(DCA).Results:1.There were statistically significant differences in age,BMI,menstrual history,history of hypertension,history of diabetes and abnormal uterine bleeding time between endometrial cancer group and endometrial control group(P<0.001).There were significant and statistically significant differences in total cholesterol(TC),triglyceride(TG)and low-density lipoprotein(LDL)in metabolic indexes(P<0.05),and significant and statistically significant differences in CA125 and HE4 in tumor indexes(P<0.001).There were significant differences in endometrial thickness(ET)and endometrial clarity in pelvic ultrasound results(P<0.001).2.Multivariate analysis suggested age(OR=1.131,P<0.001),BMI(OR=1.144,P<0.05),abnormal uterine bleeding time(OR=1.059,P<0.05),CA125(OR=7.533,P<0.001),TC(OR=1.319,P<0.05),ET(OR=1.294,P<0.001),endometrial resolution(OR=4.509,P<0.001),diabetes history(OR=2.502,P<0.05)were independent risk factors for endometrial cancer.3.The ROC curve showed that the AUC of age was 0.669(95%CI: 0.621-0.718,P<0.001),the sensitivity was 58%,the specificity was 75%,and the AUC of BMI was0.598(95%CI:0.548-0.648,P<0.001)when the Yoden index was 0.175,the sensitivity was 57%,the specificity was 61%,and the AUC for abnormal uterine bleeding time was 0.665(95%CI: 0.617-0.713,P<0.001),when the Yoden index was0.288,the sensitivity was 57%,the specificity was 72%,the AUC of TC was 0.609(95%CI: 0.560-0.659,P<0.001),when the Yoden index was 0.189,the sensitivity was67%,the specificity was 52%,the AUC of ET was 0.761(95%CI:0.719-0.803,P<0.001),when the Yoden index was 0.427,the sensitivity was 84%,the specificity was 59%,the above indexes had certain diagnostic efficacy for endometrial cancer.4.A line graph model of endometrial cancer was established based on independent factors,and the C index of the ROC curve was 0.891(95%CI :0.862-0.919,P<0.001).The AUC value of the line graph model was greater than 0.8,the sensitivity was 86%,and the specificity was 77%.The C index and AUC showed that the model had strong prediction efficiency.The calibration curve showed that the actual curve was close to the calibration line,which indicated that the accuracy of the line graph was high,and the decision curve display model had good clinical practical value.Conclusions:1.Age,BMI,menstrual history,history of diabetes,history of hypertension,abnormal uterine bleeding time,TC,TG,LDL,CA125,HE4,ET and endometrial resolution were risk factors for endometrial cancer.Age,BMI,diabetes history,abnormal uterine bleeding time,TC,CA125,ET and endometrial resolution were independent risk factors for endometrial cancer.Drawing ROC curve to evaluate the above independent risk factors was of certain diagnostic value for endometrial cancer.2.In this study,the risk profile model constructed can evaluate and predict the probability of endometrial cancer in patients with premenopausal abnormal uterine bleeding,with good prediction performance,confirming that the occurrence of endometrial cancer is correlated with various factors,providing a simple,convenient,safe and effective prediction method for clinical patients with premenopausal abnormal uterine bleeding,and also providing reference for clinical obstetricians and gynecologists to diagnose and treat such patients.It has a good prospect of clinical application.