甲状腺微小乳头状癌颈侧区淋巴结转移风险预测模型研究

A predictive model research of the risk of lateral lymph node metastasis in papillary thyroid microcarcinoma

目的探讨甲状腺微小乳头状癌(PTMC)颈侧区淋巴结转移(LLNM)的危险因素并构建临床预测模型。方法回顾性分析2020年6月至2021年6月郑州大学第一附属医院甲状腺外科首次收治的748例PTMC病人的病例资料,根据是否存在LLNM将其分为转移组(n=126例)和非转移组(n=622例)。使用R软件在同一坐标系中,绘制中央区淋巴结阳性个数(PCLNN)及中央区阳性淋巴结对数(LODDS)的受试者工作特征(ROC)曲线,选择能更好反映中央区淋巴结转移程度的自变量纳入模型。采用单因素及多因素Logistic回归分析PTMC病人发生LLNM的高危因素。构建预测LLNM的列线图模型,并评价模型的预测效能。结果LODDS预测LLNM的ROC曲线下面积(AUC)为0.878,高于PCLNN的AUC值(0.802)。多因素Logistic回归分析显示,年龄、最大癌灶直径、双侧、突破包膜、浸润性生长以及LODDS是PTMC发生LLNM的独立预测因素。列线图模型的ROC曲线AUC值为0.898(95%CI:0.869~0.927),Calibration曲线证明其一致性较好。结论年龄、最大癌灶直径、双侧、突破包膜、浸润性生长以及LODDS是PTMC病人发生LLNM的独立危险因素;建立的列线图可有效预测LLNM的风险,具有一定的临床使用价值。

Objective To evaluate the risk factors of lateral lymph node metastasis(LLNM)in papillary thyroid microcarcinoma(PTMC)and to construct a clinical prediction model.Methods We retrospectively collected data pertaining to 748 patients with PTMC who underwent treatment in the Department of Thyroid Surgery of the First Affiliated Hospital of Zhengzhou University from 2020 to 2021,and these patients were divided into the following 2groups:a LLNM-positive group and a LLNM-negative group.The receiver operating characteristic(ROC)curves of the number of examined central lymph node(PCLNN)and the log odds of positive lymph node(LODDS)were plotted in the same coordinate system by R software,and an independent variable that better reflects the degree of cervical lymph node metastasis was selected for inclusion in the model.Univariate analysis and multivariate Logistic regression were used to analyze the risk factors for LLNM in PTMC patients.Construct a nomogram model for predicting LLNM,and evaluate its prediction efficiency.Results The results showed that the area under the ROC curve(AUC)of LODDS predicted LLNM was 0.878,which was higher than the AUC value(0.802)of PCLNN.Multivariate Logistic regression analysis showed that age,maximum tumor diameter,bilateral,capsule involvement,invasive extension and LODDS were independent predictors of LLNM in PTMC.The AUC value of the ROC curve of the nomogram model was 0.898(95%CI:0.869~0.927),and the result of Calibration curve demonstrated its good consistency.Conclusion Age,maximum tumor diameter,bilateral,capsule involvement,invasive extension and LODDS were independent risk factors of LLNM in PTMC patients.The established nomogram can effectively predict the risk of LLNM,which has certain clinical use value.