中高危单侧甲状腺乳头状癌对侧中央区淋巴结转移危险因素分析及预测模型建立

Risk factors analysis and prediction model establishment of contralateral central lymph node metastasis in intermediate-to-high risk unilateral papillary thyroid carcinoma

目的探讨中高危单侧甲状腺乳头状癌(uPTC)发生对侧中央区淋巴结(Cont-CLNs)转移的危险因素,并构建预测模型。方法回顾性分析2021年1月至2023年6月在南通大学附属医院接受甲状腺癌手术的206例患者的临床资料,其中男50例,女156例,年龄[M(Q_(1),Q_(3))]49.0(33.8,57.0)岁。采用单因素分析和多因素logistic回归分析筛选出Cont-CLNs转移的危险因素,构建中高危uPTC发生Cont-CLNs转移的预测模型列线图,分别采用受试者工作特征(ROC)曲线下面积(AUC)、校准曲线和决策曲线(DCA)对模型的预测能力、准确性和临床适用性进行评估。采用R语言随机筛选出70%的患者设立验证组对该模型进行内部验证。结果根据患者是否发生Cont-CLNs转移分为转移组(n=56)和未转移组(n=150),两组患者年龄分别为39.0(28.0,56.8)岁和51.0(38.8,57.0)岁。两组在性别、肿瘤最大径(>1 cm)、同侧中央区淋巴结(Ipsi-CLNs)转移、Ipsi-CLNs转移数目(≥4)及患侧侧颈区淋巴结转移与Cont-CLNs转移方面差异有统计学意义(均P<0.05)。多因素logistic回归分析显示,男性(OR=2.926,95%CI:1.063~8.051)、肿瘤最大径>1 cm(OR=4.471,95%CI:1.344~14.877)、Ipsi-CLNs转移数目≥4(OR=5.011,95%CI:1.815~13.834)是中高危uPTC发生Cont-CLNs转移的危险因素(均P<0.05)。建模组中预测模型预测中高危uPTC发生Cont-CLNs转移的ROC曲线AUC、灵敏度、特异度分别为0.821(95%CI:0.744~0.898)、82.5%、63.4%;内部验证组中预测模型预测中高危uPTC发生Cont-CLNs转移的ROC曲线AUC、灵敏度、特异度分别为0.810(95%CI:0.717~0.902)、63.3%、83.7%。建模组与验证组校正曲线表明该模型具有较好的校准能力。建模组与验证组DCA曲线提示预测模型具有良好的临床适用性。结论该研究构建的预测模型对中高危uPTC发生Cont-CLNs转移具有较好的预测能力,当中高危uPTC患者为男性、肿瘤最大径>1 cm、Ipsi-CLNs转移数目≥4时应警惕Cont-CLNs转移,可考虑行双侧中央区淋巴结清扫。

Objective To explore the risk factors of contralateral central lymph nodes(Cont-CLNs)metastasis in intermediate-to-high risk unilateral papillary thyroid carcinoma and establish a prediction model.Methods The clinical data of 206 patients receiving thyroid cancer surgery at Nantong University Affiliated Hospital between January 2021 and June 2023 were retrospectively analyzed,including 50 males and 156 females,with an age of[M(Q_(1),Q_(3))]49.0(33.8,57.0)years old.The risk factors of Cont-CLNs metastasis were screened by univariate analysis and multivariate logistic regression analysis.A nomogram was constructed for predicting Cont-CLNs metastasis in intermediate-to-high risk uPTC.The area under the receiver operating characteristic(ROC)curve(AUC),calibration curve,and decision curve analysis(DCA)were used to evaluate the model′s predictive ability,accuracy,and clinical applicability,respectively.R language was used to randomly select 70%of the patients to establish a validation group for internal validation of the model.Results Patients were divided into a metastasis group(n=56)and a non-metastasis group(n=150)based on the occurrence of Cont-CLNs metastasis.The ages of the two groups were 39.0(28.0,56.8)years and 51.0(38.8,57.0)years,respectively.There were statistically significant differences in gender,maximum tumor diameter(>1 cm),ipsilateral central lymph nodes(Ipsi-CLNs)metastasis,number of Ipsi-CLNs metastases(≥4),and lateral lymph node metastasis and Cont-CLNs metastasis between the two groups(all P<0.05).The results of multivariate logistic regression analyses showed that males(OR=2.926,95%CI:1.063-8.051),maximum tumor diameter>1 cm(OR=4.471,95%CI:1.344-14.877),and number of Ipsi-CLNs metastases≥4(OR=5.011,95%CI:1.815-13.834)were risk factors for Cont-CLNs metastasis(all P<0.05).The AUC of the ROC curve,sensitivity,and specificity for predicting Cont-CLNs metastasis in intermediate-to-high risk uPTC by the prediction model in the modeling group were 0.821(95%CI:0.744-0.898),82.5%,and 63.4%,respectively.In the internal validation group,the AUC of the ROC curve,sensitivity,and specificity for predicting Cont-CLNs metastasis in intermediate-to-high risk uPTC by the prediction model were 0.810(95%CI:0.717-0.902),63.3%,and 83.7%,respectively.The calibration curves of the modeling group and the validation group showed that the model had good calibration ability.The DCA curves of the modeling group and the validation group indicated that the prediction model had good clinical adaptability.Conclusions The prediction model constructed in this study has good predictive performance for Cont-CLNs metastasis in intermediate-to-high uPTC.When patient with intermediate-to-high risk uPTC is male,with maximum tumor diameter>1 cm,and the number of Ipsi-CLNs metastases≥4 should be alert to Cont-CLNs metastasis,and bilateral central lymph node dissection may be considered.