临床和双能CT特征对甲状腺乳头状癌侧颈区淋巴结转移的预测价值

Clinical and Dual-Energy CT Features for Predicting Lateral Cervical Lymph Node Metastasis from Papillary Thyroid Cancer

目的探讨临床和双能CT特征对甲状腺乳头状癌(PTC)侧颈区淋巴结转移(LLNM)的预测价值。方法回顾性分析术前行双能CT检查并经病理证实的263例PTC患者的临床和影像资料。纳入分析的临床和CT图像特征包括年龄、性别、甲状腺功能指标、病灶位置、病灶大小、病灶是否合并钙化及囊变。测量平扫和动静脉期双能CT定量参数,包括碘浓度(IC)、有效原子序数(Z_(eff))、标准化碘浓度(NIC)、标准化有效原子序数(NZ_(eff))、双能指数(DEI)和能谱曲线斜率(λ_(HU))。比较LLNM组和非LLNM组间临床和双能CT特征差异。应用受试者工作特性(ROC)曲线评估临床特征、双能CT特征和两者联合模型的效能,并计算曲线下面积(AUC)、敏感度和特异度。结果男性、甲状腺球蛋白异常、甲状腺球蛋白抗体异常、病灶位于上极、病灶最大短轴径>10 mm、病灶合并钙化的PTC患者更容易出现LLNM(均P<0.05)。LLNM组平扫NIC显著低于非LLNM组,LLNM组动静脉期IC、NIC和DEI均显著高于非LLNM组(均P<0.05)。联合模型预测效能最高(AUC为0.773,敏感度84.1%,特异度65.3%),其次是双能CT特征模型(AUC为0.745,敏感度79.9%,特异度67.4%),两者均优于临床特征模型(AUC为0.642,敏感度62.6%,特异度65.3%,均P<0.05)。结论相较于临床特征,双能CT特征对PTC患者LLNM具有更高的预测价值。

Objective To explore the value of clinical and dual energy CT(DECT)features for predicting lateral cervical lymph node metastasis(LLNM)in papillary thyroid carcinoma(PTC).Methods Clinical and imaging data of 263 patients with pathologically proven PTC were retrospectively analyzed.Clinical and CT features including age,sex,thyroid function indicators,location,size,calcification,and cystic change were collected.Quantitative DECT parameters including iodine concentration(IC),effective atomic number(Z_(eff)),normalized IC(NIC),normalized Zeff(NZ_(eff)),dual energy index(DEI)and slope of energy spectrum curve(λ_(HU))were calculated.The differences in clinical and DECT features between LLNM and non-LLNM group were compared.Receiver operating characteristic(ROC)curve was used to evaluate the efficacy of clinical features,DECT features,and the combined model.The area under the ROC curve(AUC),sensitivity and specificity were calculated simultaneously.Results PTC patients who are male,with abnormal Tg and TgAb indicators,lesions located at the upper pole,maximum short axis diameter of lesions>10 mm,and lesions with calcification were prone to LLNM(P<0.05).Non-contrast NIC in LLNM group were significantly lower than those in non-LLNM group,while arterial and venous phase IC,NIC,and DEI in LLNM group were significantly higher than those in non-LLNM group(P<0.05).Combination of clinical and DECT features showed the optimal predictive performance(AUC,0.773;sensitivity,84.1%;specificity,65.3%),followed by DECT features(AUC,0.745;sensitivity,79.9%;specificity,67.4%).Both of them outperformed clinical features alone(AUC,0.642;sensitivity,62.6%;specificity,65.3%;P all<0.05).Conclusion Compared to clinical features,DECT features showed higher predictive value for LLNM in PTC patients.