儿童及青少年甲状腺乳头状癌颈侧区淋巴结转移风险预测模型研究

Risk model analysis of lateral lymph node metastasis in papillary thyroid carcinoma of children and adolescents

目的 探讨儿童及青少年甲状腺乳头状癌(papillary thyroid carcinoma in children and adolescents,CAA-PTC)颈侧区淋巴结转移(lateral lymph node metastasis,LLNM)的高危特征以及预测模型。方法 回顾性分析1990年1月~2017年12月在浙江省肿瘤医院头颈外科接受初次手术的108例CAA-PTC患者的病例资料,采用单因素分析和多因素Logistic回归的方法研究LLNM与CAA-PTC患者各项临床病理特征的关系,构建LLNM的风险预测模型,并制作受试者工作特征曲线分析其预测价值。结果 108例CAA-PTC发生LLNM共71例,转移率为65.74%(71/108)。单因素分析,LLNM与CAA-PTC患者的年龄、双侧累及、多发病灶、病灶大小以及腺外侵犯有关(P<0.05)。多因素Logistic回归分析,年龄(<14岁)、病灶大小(≥2.5 cm)以及腺外侵犯是CAA-PTC发生LLNM的独立预测因素(P均<0.05),这三项高危特征构建的数学模型预测LLNM的曲线下面积为0.806(95%CI:0.721~0.892,P<0.001),高于单项特征,预测的敏感性为71.80%,特异性为75.69%。结论 年龄(<14岁)、病灶大小(≥2.5 cm)以及腺外侵犯是CAA-PTC发生LLNM的高危特征,三者构建的的数学模型对预测CAA-PTC患者LLNM具有较好的临床价值。

OBJECTIVE To investigate the risk factors of lateral lymph node metastasis(LLNM) in papillary thyroid carcinoma of children and adolescents(CAA-PTC) and establish a risk prediction model.METHODS The clinical data of 108 CAA-PTC patients who underwent primary surgery in the Department of Head and Neck Surgery of Zhejiang Cancer Hospital from January 1990 to December 2017 were retrospectively analyzed.Univariate analysis and multivariate Logistic regression were used to study the relationship between LLNM and various clinicopathological features of CAA-PTC patients.The risk prediction model of LLNM was constructed,and the receiver operating characteristic curve was made to evaluate its predictive value.RESULTS There were 71 cases of LLNM in 108 CAA-PTC patients,and the LLNM rate was 65.74%(71/108).Univariate analysis showed that LLNM was related to CAA-PTC patients’ age,bilateral involvement,multiple lesions,lesion size and extrathyroidal infiltration(all P<0.05).Multivariate Logistic regression analysis showed that age(<14 years),lesion size(≥2.5 cm) and extrathyroidal infiltration were independent predictors of LLNM in CAA-PTC(all P<0.05).The area under the curve of the mathematical model constructed by these three high-risk characteristics for predicting LLNM was 0.806(95% Confidence Interval:0.721-0.892,P<0.001),while the sensitivity and specificity was 71.80% and 75.69%,respectively.CONCLUSIONAge(<14 years),lesion size(≥2.5 cm) and extrathyroidal infiltration are independent risk factors for LLNM in CAA-PTC patients,and the mathematical model constructed by the three features has good clinical value for predicting LLNM.