探讨超声对原发性甲状腺淋巴瘤（PTL）与甲状腺未分化癌（ATC）的鉴别诊断。

回顾性分析2004年9月至2019年11月四川大学华西医院73例PTL和28例ATC患者的临床、病理和超声表现。年龄、病灶最大径及杨氏模量值最大值（Emax）、杨氏模量值最小值（Emin）、杨氏模量值平均值（Emean）、甲状腺病灶与正常组织的弹性评估值之间的比值（Ratio）等计量资料，组间比较采用t检验或Mann-Whitney U检验。性别、临床症状、超声分型、形态、血流分级、增强程度等计数资料，组间比较采用χ²检验或者Fisher精确检验。采用Logistic回归分析剪切波弹性成像（SWE）对诊断PTL、ATC的优势参数，并对优势参数进行受试者操作特征（ROC）曲线分析。另外，分析PTL的各病理亚型占比，并采用χ²检验或者Fisher精确检验比较最常见的2种病理类型的超声分型、形态、血流分级等的差异。

ATC声音嘶哑发生率高于PTL（28.6% vs 5.5%），差异有统计学意义（χ²=8.220，P=0.004），两者在年龄、性别、颈部包块短期增大、呼吸困难、吞咽困难方面比较，差异均无统计学意义（P均＞0.05）。PTL病灶最大径大于ATC患者［43.50（42.50）mm vs 26.00（26.00）mm］，差异有统计学意义（Z=2.636，P=0.008）。PTL内部回声极低、后方回声增强、伴有慢性甲状腺炎背景比例高于ATC（60.3% vs 21.4%；45.2% vs 0；97.3% vs 3.6%），差异均有统计学意义（χ²=12.439、25.128、86.786，P=0.002、＜0.001、＜0.001）。ATC结节型、形态不规则、边缘模糊、内部低回声、钙化比例高于PTL（100% vs 68.5%；96.4% vs 69.9%；82.1% vs 56.2%；64.3% vs 34.2%；60.7% vs 2.7%），差异有统计学意义（χ²=12.294、8.121、5.884、12.439、44.536，P=0.001、=0.004、=0.015、=0.002、＜0.001）。两者最常见Adler血流分级均为2级，差异无统计学意义（P＞0.05）。两者增强程度及灌注缺损比较差异均无统计学意义（P均＞0.05）。Emin值PTL组小于ATC组［3.70（8.85）kPa vs 25.60（17.90）kPa］，差异具有统计学意义（Z=2.591，P=0.010），2组间Emean、Emax、Ratio值比较差异均无统计学意义（P均＞0.05）。Logistic回归分析发现Emin与诊断PTL、ATC呈独立相关性，对Emin进行ROC曲线绘制，ROC曲线下面积为0.863，最大约登指数为0.774，对应的截断值为11.950 kPa，敏感度为85.7%，特异度为91.7%。本研究中PTL共8种病理亚型，最常见的是弥漫性大B细胞淋巴瘤（DLBCL），其次是黏膜相关淋巴组织结外边缘区B细胞淋巴瘤（MALT）；DLBCL结节型、边缘模糊、形态不规则、不均匀低回声、Adler血流2级比例高于MALT（72.5% vs 33.3%；87.5% vs 33.3%；67.5% vs 38.1%；60.0% vs 19.0%），差异均具有统计学意义（χ²=15.190、4.040、18.840、9.273、17.693，P＜0.001、=0.044、＜0.001、=0.007、＜0.001），后方回声特征比较中，差异无统计学意义（P＞0.05）。

PTL与ATC均具有特征性的超声表现，超声有助于两者的鉴别诊断。

To assess the value of ultrasonography in the diagnosis and differential diagnosis of primary thyroid lymphoma (PTL) and anaplastic thyroid carcinoma (ATC).

The clinical, pathological, and ultrasonic manifestations of 73 cases of PTL and 28 cases of ATC were analyzed retrospectively at West China Hospital of Sichuan University from September 2004 to November 2019. For numerical data such as age, maximum diameter of the lesion, the maximum (Emax) and minimum (Emin) values of Young's modulus, mean (Emean) value of Young's modulus, and the ratio between elastic evaluation values of thyroid lesion and normal tissue (Ratio), t-test or Mann Whitney U test was used for comparisons between groups. Gender, clinical symptoms, ultrasound classification, morphology, blood flow classification, enhancement degree, and other count data were compared between groups by chi square test or Fisher exact test. Logistic regression analysis was used to analyze the dominant parameters of shear wave elastography (SWE) in the diagnosis of PTL and ATC, and the ROC curve of the dominant parameters was analyzed. In addition, the proportion of each pathological subtype of PTL was compared, and the two most common pathological types of ultrasound classification, morphology, blood flow classification, and other count data were compared by chi square test or Fisher exact test.

Hoarseness occurred significantly more frequently in ATC than in PTL (28.6% vs 5.5%, χ²=8.220, P=0.004), while there was no significant difference in age, gender, short-term enlargement of neck mass, dyspnea, or dysphagia (P＞0.05 for all). The largest diameter of PTL lesion was larger than that of ATC [43.50 (42.50) mm vs 26.00 (26.00) mm, Z=2.636, P=0.008]. The background proportion of acoustic enhancement and concomitant chronic thyroiditis was higher in PTL than in ATC (60.3% vs 21.4%, χ²=12.439, P=0.002; 45.2% vs 0, χ²=25.128, P＜0.001; 97.3% vs 3.6%, χ²=86.786, P＜0.001). The rates of nodular type, irregular morphology, indistinct margins, internal hypoechogenicity, and calcification were significantly higher in ATC than in PTL (100% vs 68.5%, χ²=12.294, P=0.001; 96.4% vs 69.9%, χ²=8.121, P=0.004; 82.1% vs 56.2%, χ²=5.884, P=0.015; 64.3% vs 34.2%,χ²=12.439, P =0.002; 60.7% vs 2.7%, χ²=44.536, P＜0.001). The most common Adler flow grade was grade 2, which was not significantly different between the two entities (P＞0.05). Neither the degree of enhancement nor the perfusion defect was significantly different between ATC and PTL (P＞0.05). Emin was significantly smaller in PTL than in ATC [3.70 (8.85) kPa vs 25.60 (17.90) kPa, Z=2.591, P=0.010], while Emean, Emax, and Ratio were not significantly different between the two entities (P＞0.05 for all). Logistic regression analysis showed that Emin was independently correlated with the diagnosis of PTL and ATC. The ROC curve of Emin was drawn; the area under the ROC curve was 0.863, the maximum Youden index was 0.774, and the corresponding cut-off value was 11.950 kPa, with a sensitivity and specificity of 85.7% and 91.7%, respectively. In this study, there were eight pathologic subtypes of PTL, with the most common being diffuse large B-cell lymphoma (DLBCL), followed by mucosa associated lymphoid tissue extranodal marginal zone B-cell lymphoma (MALT). DLBCL was characterized by nodular type, indistinct margins, irregular morphology, heterogeneous hypoechogenicity, and higher proportion of grade 2 Adler blood flow than that in MALT (72.5% vs 33.3%, χ²=15.190, P＜0.001; 65.0% vs 38.1%, χ²=4.040, P=0.044; 87.5% vs 33.3%, χ²=18.840, P＜0.001; 67.5% vs 38.1%, χ²=9.273, P=0.007; 60.0% vs 19.0%, χ²=17.693, P＜0.001), but there was no significant difference in the posterior echo characteristics (P＞0.05).

Both PTL and ATC have characteristic ultrasonographic findings. Ultrasound is helpful in their diagnosis and differential diagnosis.