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

ObjectiveTo investigate the correlation between the ultrasonogrophic characteristics of thyroid microcarcinoma wall configuration (≤1.0 cm in the maximum diameter of the carcinoma) and cervical lymph node metastasis by binary logistic regression. MethodsA 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 selected for multivariable analysis, namely the margin, border, periphery and calcification of the wall configuration, to establish a regression model. Then, the odds ratio （OR） of variables in the equation was compared. ResultsAccording to multivariable 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. ConclusionsLogistic 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.