Establishment Of A Model For Predicting The Risk Of Lymphatic Vascular Space Invasion In Endometrial Carcinoma

Objective:To explore the related risk factors that affect the occurrence of lymphatic vascular space infiltration(LVSI)in patients with endometrial cancer(EC),screen out independent risk factors as predictors,and construct a prediction model,which can be used to predict the risk of LVSI before comprehensive staging surgery,and then risk stratification,so as to realize individual evaluation and treatment of patients.Method:Selected patients who were hospitalized in the Department of Gynecology of the First Hospital of Lanzhou University from January 2014 to December 2020 and were confirmed to be EC by postoperative pathological examinations.A total of 268 eligible patients were screened,and the clinicopathological data of the above patients were collected.Then,the regression model was established based on single and multi-factor analysis;the independent variables corresponding to the independent risk factors selected by the multi-factor analysis as predictors,and the R language was used to build a nomogram model;additional,evaluate the above two models separately.Finally,the Yoden index is used for risk stratification.Results:1.Univariate and multivariate analysis: Age ≥ 54 years old,CA125 > 35U/ ml,tumor diameter > 20 mm,non endometrioid adenocarcinoma,poor differentiation,myometrial invasion depth ≥ 1 / 2,cervical stromal involvement and parametrial metastasis were related risk factors for LVSI in EC patients(P < 0.05),of which CA125 > 35 U / ml(P = 0.012),myometrial invasion depth ≥ 1 / 2(P = 0.000),non endometrioid adenocarcinoma(P = 0.007)and poor differentiation(P =0.000)were independent risk factors for LVSI.2.Four independent risk factors were included in the Logistic regression model,and the corresponding regression coefficients were obtained.The prediction equation was Y =-1.697 + 0.991 * preoperative CA125 + 1.578 * myometrial invasion + 1.513* pathological type + 1.871 * histological grade.When ROC curve was drawn,the area under the curve was the largest,which was 0.865(95% CI: 0.812～0.918).A nomogram model for predicting LVSI risk of EC patients was successfully constructed.The c-index value obtained by internal validation was 0.865,and the mean absolute error of calibration curve was 0.029.3.The total score <82.5 is low risk,82.5<total score <173 is medium risk,and total score>173 is high risk.Conclusion:1.The individual prediction model of LVSI improves the ability of identifying endometrial cancer patients with high risk of lymphatic vascular space invasion before comprehensive staging surgery.2.In clinical work,preoperative CA125 > 35U/ml,muscular infiltration depth≥1/2,non-endometrioid adenocarcinoma and poorly differentiated EC patients can be classified by using this nomogram,and they are medium/high risk patients,which should be paid attention to.3.Combined CA125,muscular infiltration depth,pathological type and histological classification of the four clinical pathological indicators established the Logistic regression model can be used to predict the risk of LVSI in patients with endometrial carcinoma,and it can make through the nomogram prediction results more intuitive and visualization,facilitate assessment and interpretation,to assist the clinical work has a certain value.