妇科恶性肿瘤淋巴结清扫术后淋巴漏发生的logistics回归分析及预测模型构建

Logistics regression analysis and construction of prediction model for lymphatic leakage after radical resection of gynecological malignant tumors

目的探讨妇科恶性肿瘤淋巴结清扫术后淋巴漏发生的危险因素,建立预测模型,为妇科恶性肿瘤术后淋巴漏的发生预测提供参考。方法选取许昌市中心医院2017年2月—2022年7月700例妇科恶性肿瘤行淋巴结清扫术患者,将患者按照5∶2的比例随机分为建模组(n=500例)和验证组(n=200例),依照术后淋巴漏发生情况将建模组分为淋巴漏组(n=43例)与非淋巴漏组(n=457例)两个亚组,采用logistic回归分析筛选妇科恶性肿瘤淋巴结清扫术后淋巴漏发生的危险因素,基于回归系数构建预测模型,采用受试者工作曲线(ROC)评估模型区分度。结果logistic回归分析显示超重、术中出血量大、淋巴结清扫范围为盆腔+腹主动脉旁、淋巴结清扫数目多、术前贫血、术前低蛋白血症是妇科恶性肿瘤根治术患者术后淋巴漏发生的独立危险因素;妇科恶性肿瘤淋巴结清扫术后发生淋巴漏概率值回归方程为:P=1/[1+e^(-(-6.315+0.944×BMI指数+0.881×术中出血量+1.496×淋巴结清扫范围+0.794×淋巴结清扫数目+0.772×贫血+0.740×低蛋白血症))],Hosmer-Lemeshow检验回归方程拟合优度(P=0.890);通过建模组数据进行内部验证ROC曲线下面积0.827[95%CI(0.761~0.892)],灵敏度为81.40%,特异度为75.71%,据Youden指数最大原则cut-off点0.094;利用验证组数据进行外部验证,验证组共有15例患者发生术后淋巴漏,以上述cut-off点0.094为截点值,ROC曲线下面积为0.817[95%CI(0.710~0.924)],灵敏度为86.67%,特异度为76.76%。结论建立的预测模型判别效度良好,可用于识别妇科恶性肿瘤淋巴结清扫术后淋巴漏发生高危人群。

Objective To explore the risk factors of lymphatic leakage after radical resection of gynecological malignant tumors and construct the prediction model so as to provide reference for the prediction of lymphatic leakage.Methods A total of 700 patients with gynecological malignant tumors undergoing radical resection in Xuchang Central Hospital were enrolled between February 2017 and July 2022,and they were randomly divided into modeling group(n=500)and verification group(n=200).According to presence or absence of postoperativelymphatic leakage,patients in modeling group were divided into leakage group(n=43)and non-leakage(n=457).The risk factors of lymphatic leakage after radical resection of gynecological malignant tumors were screened out by logistic regression analysis.The prediction model was constructed based on regression coefficients,and the model differentiation was evaluated by receiver operating characteristics(ROC)curves.Results Logistic regression analysis showed that overweight,much intraoperative blood loss,pelvic+para-aortic lymph node dissection,much lymph nodes dissection,preoperative anemia and hypoproteinemia were independent risk factors of lymph leakage after radical resection of gynecological malignant tumors.The regression equation to predict lymphatic leakage was as follow:P=1/[1+e^(-(-6.315+0.944×BMI+0.881×intraoperative blood loss+1.496×lymph node dissection range+0.794×number of lymph nodes dissection+0.772×anemia+0.740×hypoproteinemia))],the goodness of fit of the regression equation was detected by Hosmer-Lemeshow(P=0.890).The internal verification of data was conducted in modeling group,and its area under ROC curve(AUC),sensitivity,specificity and the cut-off value based on the maximum Youden index were 0.827[95%CI(0.761-0.892)],81.40%,75.71%and 0.094,respectively.The external verification of data was conducted in verification group.In verification group,there were 15 cases with postoperative lymphatic leakage.Taking the above cut-off point as the cut-off value,AUC,sensitivity and specificity were 0.817[95%CI(0.710-0.924)],86.67%and 76.76%,respectively.Conclusion The constructed prediction model has good discriminate validity,and which can be applied to identify the high-risk group with lymphatic leakage after radical discriminate of gynecological malignant tumors.