甲状腺癌患者术后发生喉返神经损伤危险因素分析以及相关预测模型的构建

Analysis of Risk Factors for Postoperative Recurrent Laryngeal Nerve Injury in Patients with Thyroid Cancer and Construction of Related Prediction Models

目的探讨甲状腺癌患者术后发生喉返神经(RLN)损伤的危险因素,并构建相关预测模型。方法纳入2019年1月-2023年1月就诊于广西壮族自治区人民医院的380例甲状腺癌患者作为研究对象,按术后是否发生RLN损伤分为RLN损伤组23例和N-RLN损伤组357例,比较两组患者一般临床资料以及手术相关资料,多因素Logistic回归分析获得影响甲状腺癌患者术后发生RLN损伤的独立预测因素,基于独立预测因素建立预测甲状腺癌患者术后出现RLN损伤的列线图模型。以Bootstrap法和校准曲线进行列线图模型的内部验证,绘制预测甲状腺癌患者术后发生RLN损伤的ROC曲线,分析独立预测因素的预测效能。结果多因素Logistic回归分析结果显示,肿瘤位置、手术次数及术中神经监测为影响甲状腺癌患者术后发生RLN损伤的独立预测因素(P<0.05)。基于多因素分析结果获得的3个独立预测因素:肿瘤位置、手术次数及术中神经监测,构建预测甲状腺癌患者术后发生RLN损伤的列线图模型。列线图模型的区分度评价指标C指数(C-Index)为0.869(95%CI:0.823~0.958);拟合优度(H-L)检验结果显示,甲状腺癌患者术后发生RLN损伤概率的预测值与实际观测值符合度良好(P>0.05)。ROC分析结果显示,肿瘤位置、手术次数及术中神经监测预测甲状腺癌患者术后发生RLN损伤概率的曲线下面积(AUC)分别为0.779(95%CI:0.712~0.834)、0.756(95%CI:0.709~0.816);以及0.887(95%CI:0.835~0.935),而三者联合预测的AUC为0.937(95%CI:0.887~0.976)。结论肿瘤位置、手术次数及术中神经监测为影响甲状腺癌患者术后发生RLN损伤的独立预测因素,基于独立预测因素构建的列线图模型用于预测甲状腺癌患者术后发生RLN损伤具有较高的预测价值。

Objective To investigate the risk factors of recurrent laryngeal nerve(RLN)injury in patients with thyroid cancer after surgery,and to construct a relevant prediction model.Methods A total of 380 patients with thyroid cancer who were admitted to the People's Hospital of Guangxi Zhuang Autonomous Region from January 2019 to January 2023 were included in the study.According to whether RLN injury occurred aftersurgery,they were divided into RLN injury group(n=23)and N-RLN injury group(n=357).The general clinical data and operation-related data of the two groups were compared.Multivariate Logistic regression analysis was used to obtain independent predictors of RLN injury in patients with thyroid cancer after operation.Based on independent predictors,a nomogram model for predicting RLN injury in patients with thyroid cancer after operation was established.Bootstrap method and calibration curve were used to verify the nomogram model.The ROC curve for predicting RLN injury in patients with thyroid cancer after operation was drawn,and the predictive efficacy of independent predictors was analyzed.Results Multivariate Logistic regression analysis showed that tumor location,number of operations and intraoperative nerve monitoring were independent predictors of RLN injury in patients with thyroid cancer after surgery(P<0.05).Based on the results of multivariate analysis,three independent predictors were obtained,namely tumor location,number of operations and intraoperative nerve monitoring,and a nomogram model was constructed to predict RLN injury in patients with thyroid cancer after operation.The discrimination evaluation index C index(C-Index)of the nomogram model was 0.869(95%CI:0.823-0.958).The goodness of fit(H-L)test results showed that the predicted value of RLN injury probability in patients with thyroid cancer after operation was in good agreement with the actual observed value(P>0.05).The results of ROC analysis showed that the area under the curve(AUC)of tumor location,number of operations and intraoperative nerve monitoring for predicting the probability of RLN injury in patients with thyroid cancer were 0.779(95%CI:0.712-0.834),0.756(95%CI:0.709-0.816)and 0.887(95%CI:0.835-0.935),respectively.The AUC of the combined prediction of the three was 0.937(95%CI:0.887-0.976).Conclusion Tumor location,number of operations and intraoperative nerve monitoring are independent predictors of RLN injury in patients with thyroid cancer after surgery.The nomogram model based on independent predictors has a high predictive value for predicting recurrent laryngeal nerve injury in patients with thyroid cancer after surgery.