The Predictive Value Of Clinicopathologic Parameters Combined With Tumor Markers In Lymph Node Metastasis Of Endometrial Cancer

[Objective]To analyze the risk factors and explore the predictors of lymph node metastasis in endometrial cancer.And then to establish the lymph node metastasis prodiction model of endometrial cancer preliminarily,realizing individualized treatment.[Methods]Clinical data of 246 patients who underwent comprehensive staging surgery for endometrial cancer in Northern Jiangsu People's hospital from July 2014 to July 2019 were retrospectively analyzed,including 31 patients with lymph node metastasis and 215 patients without lymph node metastasis.To collect the information of all patients,such as the general clinical data(age,menopausal status,body mass index(BMI),history of pregnancy,history of hypertension or diabetes),the data of surgical pathology(tumor diameter,pathological type,histological grading,muscular invasion,lymphatic vascular infiltration,cervical interstitial involvement,abdominal cavity flushing fluid)and the preoperative levels of tumor markers(carbohydrate antigen 125(CA125),carbohydrate antigen 199(CA199),carcinoembryonic antigen(CEA),alpha fetal protein(AFP),human epididymis protein 4(HE4)).The risk factors of lymph node metastasis in endometrial carcinoma were analyzed by univariate analysis using?~2test,Fisher's exact probability test and rank sum test.ROC was drawn to explore the critical value,sensitivity and specificity of tumor markers in lymph node metastasis of endometrial cancer.Multivariate Logistic regression analysis was used to screen the independent predictors in lymph node metastasis of endometrial cancer.In the end,we constructed a risk prediction model preliminarily and evaluated the discrimination and calibration of the prediction model by AUC and Hosmer-Lemeshow goodness of fit test.[Results]1.Univariate analysis:general clinical data:postmenopausa women(?~2=4.921,P=0.027),history of pregnancy less than 2 times(?~2=6.370,P=0.012)were risk factors for lymph node metastasis of endometrial cancer.Age,BMI,hypertension and diabetes were not significantly correlated with lymph node metastasis(P>0.05).Clinicopathological parameters:Tumor diameter 2 cm or more(?~2=15.509,P<0.001),non-endometrial adenocarcinoma(?~2=21.557,P<0.001),the organization in low differentiation(?~2=16.602,P<0.001),deep muscle layer infiltration(?~2=14.732,P<0.001)and lymphatic vascular infiltration(?~2=38.825,P<0.001)are risk factors for endometrial cancer lymph node metastasis.Cervical interstitial involvement and the positive of abdominal cavity flushing fluid had no significant correlation with lymph node metastasis(P>0.05).Preoperative multiple serum tumor markers:Comparing the five tumor markers'overall levels between positive and negative lymph node metastasis,the result illustrated that the raised preoperative levels of three tumor markers,CA125(Z=-2.421,P=0.015),CA199(Z=-2.505,P=0.012)and HE4(Z=-3.280,P=0.001),are related to the lymph node metastasis of endometrial cancer.While there was no statistically significant difference between the two groups in CEA and AFP.According to the ROC,when the optimal critical value of CA125 was 21.75 U/m L,the sensitivity and specificity was 64.5%and 61.6%.And the area under the ROC curve was 0.647.When the optimal critical value of CA199 was 20.16 U/m L,the sensitivity and specificity was 51.6%and 80.2%.And the area under the ROC curve was 0.652.When the optimal critical value of HE4 was 98.80 pmol/L,the sensitivity and specificity was 61.3%and 81.4%.And the area under the ROC curve was 0.699.Regarding CA125?21.75 U/ml(?~2=7.766,P=0.012),CA199?20.16 U/ml(?~2=15.472,P<0.001),and HE4?98.80 pmol/L(?~2=27.077,P<0.001)as the risk factors of lymph node metastasis is statistically significant difference.2.Multivariate analysis:The above mentioned parameters,which were statistically signifcant differences,were incorporated into the multivariate logistic regression analysis.The tumor diameter 2 cm or more(P=0.009),non-endometrioid adenocarcinoma(P=0.020),lymphatic vascular infiltration(P=0.022)and HE4?98.80 pmol/L(P<0.001)were screened as independent predictors of lymph node metastasis in endometrial cancer by stepwise forward selection.3.The risk factor prediction model is established initially:The area under the ROC of the prediction model is 0.861,indicating that the prediction model has good ability to discriminate.The predictive value of the model is higher than other predictors detected independently.The Hosmer-Lemeshow goodness of fit test prompted that the difference between the predicted value of the model and the actual observed value was not statistically significant difference(?~2=3.822,P=0.575).The prediction model had good calibration ability and high accuracy.[Conclusion]1.The optimal critical values of CA125,CA199 and HE4 were 21.75 U/m L,20.16U/m L and 98.80 pmol/L,respectively.The diagnostic efficacy of HE4 in lymph node metastasis of endometrial cancer was higher than that of CA125 and CA199.2.Menopause,pregnancy history<2 times,tumor diameter?2cm,non-endometrioid adenocarcinoma,the organization in low differentiation,deep muscle infiltration,lymphatic vascular infiltration,CA125?21.75 U/ml,CA199?20.16 U/ml,HE4?98.80 pmol/L are risk factors for lymph node metastasis of endometrial cancer.3.Tumor diameter?2cm,non-endometrioid adenocarcinoma,lymphatic vascular infiltration and HE4?98.80 pmol/L were independent predictors of lymph node metastasis of endometrial cancer.4.To establish a prediction model for lymph node metastasis of endometrial cancer initially by Logistic regression analysis.The prediction model has good ability of discrimination and calibration,conduciving to individualized treatment of endometrial cancer.