基于超声造影特征构建列线图预测甲状腺微小乳头状癌颈淋巴结转移风险

Risk of cervical lymph node metastasis of thyroid micropapillary carcinoma predicted by constructing a nomogram based on contrast-enhanced ultrasound features

目的通过分析甲状腺微小乳头状癌(papillary thyroid microcarcinoma,PTMC)患者临床资料、结节超声和超声造影(contrast-enhanced ultrasound,CEUS)特征建立预测颈淋巴结转移(lymph node metastasis,LNM)风险的列线图模型,为合理规范的临床决策提供依据。方法回顾性分析2020年12月1日至2021年12月31日在新疆医科大学第一附属医院行甲状腺手术治疗的PTMC患者404个结节临床资料,应用随机函数按照7∶3分为建模组(n=282)与验证组(n=122)。应用Logistic回归分析筛选PTMC颈淋巴结转移的独立相关因素,构建列线图,以曲线下面积(area under curve,AUC)评估模型诊断效能,应用验证组数据进行外部验证。结果构建模型显示结节边缘、声晕、多灶性、包膜下生长或侵犯包膜、消退模式是颈淋巴结转移的危险因素(P<0.05);建模组AUC为0.747(0.690~0.804),最佳cut-off值为0.430,灵敏度0.65,特异度0.73;验证组AUC为0.778(0.697~0.860),最佳cut-off值为0.419,灵敏度0.64,特异度0.81。结论本研究构建的列线图可个体化预测PTMC颈淋巴结转移的风险,超声和超声造影特征有助于指导高风险人群的临床决策。

Objective To construct a nomogram model for predicting the risk of cervical lymph node metastasis(LNM)by analyzing the clinical data and contrast-enhanced ultrasound(CEUS)features of the patients with papillary thyroidmicrocarcinoma(PTMC),so as toprovide a basis for rational and standardized clinical decision-making.Methods Nodal data were collected from the patients with PTMC who underwent surgical treatment at the hospital from December 1,2020 to December 31,2021.A total of 404 nodules were randomly divided into 282 in the modeling group and 122 in the validation group according to 7∶3 by random function.Logistic regression analysis was used to screen for factors associated with cervical lymph node metastasis in PTMC.The diagnostic efficacy of the model was assessed in terms of area under curve(AUC),and the validation group data were applied for external validation.Results The model showed that nodule margin,mulifocality,subcapsular growth or capsular invasion,halo and wash-out pattern were the risk factors for cervical lymph node metastasis(P<0.05).The AUC of themodeling group was 0.747(0.690-0.804),with an optimal cut-off value of 0.430,corresponding to a sensitivity of 0.65 and a specificity of 0.73.The AUC of thevalidation group was 0.778(0.697-0.860),with an optimal cut-off value of 0.419,corresponding to a sensitivity of 0.64 and a specificity of 0.81.Conclusion Thenomogram constructed in this study can individually predict the risk of cervical lymph node metastasis in PTMC.Routine ultrasound and CEUS features can help to identify risk groups early and guide clinical decision-making.