Abstract: Predictive central lymph node metastasis on the basis of preoperative features of clinic and ultrasonography in patients with papillary thyroid carcinoma        LIU Wen*，YAN Xue-jing，CHENG Ruo-chuan，et al. *Department of Thyroid Surgery，the First Affiliated Hospital of Kunming Medical University，Kunming 650032，China
Corresponding author：CHENG Ruo-chuan，E-mail：cruochuan@foxmail.com
LIU Wen and YAN Xue-jing are the first authors who contributed equally to the article.
Abstract    Objective    To analyze the predictive value of central lymph node metastases (CNM) in papillary thyroid carcinoma (PTC) by preoperative features of clinic and ultrasonography and establish predictive model. Methods  Consecutive data of 2862 patients with PTC confirmed by postoperative pathology from January 2007 to June 2016 in Department of Thyroid Surgery，the First Affiliated Hospital of Kunming Medical University were analyzed retrospectively. The association between CNM and clinic，ultrasonic features was analyzed，and the predictive model was built by multivariable logistic regression. Results    Of the 2862 PTC patients，the CNM ratio was 43.2%; the resected and metastatic number of central lymph nodes were 8 (0 to 45) and 2 (1 to 38) respectively. The predictive model comprised seven variables: gender，age，tumor size，microcalcification，vascular contorts，resistance index>0.7 and multiple nodular lesions. With the prediction probability as diagnostic indicator，the area under the curve (AUC) of receiver operating characteristic (ROC) was 0.713（95%CI 0.695-0.713）for CNM. Tumor size ≥0.95 cm was the optimal critical point of which specificity and sensitively was 0.649 and 0.677，respectively. Conclusion    CNM in PTC patients could be predicted satisfactorily by predictive model that based on preoperative features of clinic and ultrasonography. It is economic，convenient，noninvasive and could effectively improve the sensitivity of preoperative CNM diagnosis to provide references for operation plan selections.

Key words: papillary thyroid carcinoma, ultrasonography, central lymph node metastasis, microcalcification, predictive model

摘要： 目的    分析甲状腺乳头状癌（PTC）术前临床和超声检查特征对中央区淋巴结转移的预测价值，并建立预测模型。方法    回顾性分析2007年1月至2016年6月昆明医科大学第一附属医院甲状腺疾病诊治中心经术后病理学检查证实为PTC的连续2862例病人的临床资料。由双人复读超声检查结果，分析临床和超声特征与中央区淋巴结转移的关系，并建立诊断模型。结果    2862例PTC病人中央区淋巴结转移发生率为43.2%，清扫中央区淋巴结数目为8（0~45）枚，转移淋巴结数目为2（1~38）枚。建立的预测模型由性别、年龄、肿瘤大小、微钙化、血管走行、阻力指数>0.7、超声多结节病变7个变量组成。以 Logistic回归预测概率作为诊断指标，根据ROC曲线确定界值，概率值预测中央区淋巴结转移的ROC曲线下面积（AUCROC）为0.713（95%CI 0.695~0.732）。确定肿瘤直径≥0.95 cm为ROC曲线的最佳临界点，特异度为0.649，敏感度为0.677。结论    基于术前临床及超声特征的预测模型能够较好地预测PTC病人中央区淋巴结转移，临床应用经济、简便、无创，可有效提高术前诊断敏感度。

关键词: 甲状腺乳头状癌, 超声, 中央区淋巴结转移, 微钙化, 预测模型