运用Logistic回归分析模型评价甲状腺微小癌壁结构超声声像图特征与颈部淋巴结转移的相关性研究

Evaluation of the correlation between the ultrasonogrophic characteristics of thyroid microcarcinoma wall configuration and cervical lymph node metastasis by binary logistic regression

目的应用Logistic回归分析评价甲状腺微小癌（肿瘤最大直径≤1．0cm）壁结构超声声像图特征与颈部淋巴结转移的相关性。方法经手术病理证实的结节最大直径≤1．0cm的甲状腺微小癌患者98例共102个结节，将其分为颈部淋巴结转移组32例（34个）和非转移组66例（68个）。所有入选病例的影像资料被完整保存在影像归档和通信系统（PACS）中，对图片资料进行回顾性分析。将前期研究中Logistic分析所得的4个危险因素（壁结构中的边缘、边界、周边及钙化）纳入多因素分析，建立回归模型，比较进入方程中的变量优势比（OR值）。结果通过多因素分析的Logistic回归分析显示，4个危险因素都是甲状腺微小癌颈部淋巴结转移的独立危险因素（P〈0．05），其中边缘分型的OR值高于其他变量。结论Logistic回归分析模型筛选出对甲状腺微小结节良、恶性有鉴别诊断的特征变量中，甲状腺微小癌壁结构的边缘较其他超声特征对预测淋巴结转移更有优势。

Objective To investigate the correlation between the uhrasonogrophic characteristics of thyroid micro- carcinoma wall configuration （≤ 1.0 cm in the maximum diameter of the carcinoma） and cervical lymph node metastasis by binary logistic regression. Methods A total of 98 thyroid microcarcinoma patients with 102 nodes in total （with the maximum diameter less than 1.0 cm） which were pathologically confirmed were enrolled into the current study. They were divided into two groups： a cervical lymph node metastasis group （32 patients with 34 nodes） and a cervical lymph node non - metastasis group （66 patients with 68 nodes）. All the image data were kept in the hospital＇s picture archiving and communication systems （PACS） for retrospective analysis. According to previous studies, four risk factors were se- lected for multivariable analysis, namely the margin, border, periphery and calcification of the wall configuration, to es- tablish a regression model. Then, the odds ratio （OR） of variables in the equation was compared. Results According to muhivariable analysis, all the four risk factors were the independent factor of cervical lymph node metastasis in thyroid microcarcinoma patients （ P 〈 0.05 ）, whose OR of the margin were higher than other independent variables. Conclu- sions Logistic regression analysis can screen out the valuable variables for differential diagnosis of benign and malignant thyroid small nodules. The margin of thyroid microcarcinoma wall configuration is better to predict lymph node metastasis than other ultrasound characteristics.