评估甲状腺乳头状癌术后患者颈部淋巴结转移规律、超声检查特征及诊断效能，为术后随访监测及临床决策提供参考依据。

回顾性收集2023年1月至2025年4月于四川省肿瘤医院接受甲状腺术后随访、超声检查发现可疑颈部淋巴结转移并经病理证实的97例甲状腺乳头状癌患者。依据术后病理，将超声可疑恶性淋巴结分为恶性组与良性组。记录转移淋巴结的颈部分区（侧颈区：Ⅱ、Ⅲ、Ⅳ、Ⅴ区；中央区：Ⅵ区及Ⅶ区）、超声表现及临床资料（如淋巴结清扫术式），并采用单因素分析方法（独立样本t检验、χ²检验或Fisher确切概率法、Kruskal-Wallis H检验）比较恶性组与良性组、中央区组与侧颈区组、不同淋巴结清扫术式组间差异。计算常规超声对各分区淋巴结的诊断效能，采用受试者操作特征（ROC）曲线进行分析，并采用DeLong检验比较曲线下面积（AUC）的差异。

97例患者中，淋巴结转移以Ⅳ区（60.82%）、Ⅲ区（57.73%）最为常见。超声检查发现的188枚可疑恶性淋巴结中，166枚经病理证实为恶性，22枚为良性。良、恶性淋巴结超声特征比较，恶性组中混合型/周边型血流模式（48.80%/24.70% vs 18.18%/0）及2~3级血流分级（61.45% vs 18.18%）比例更高，差异具有统计学意义（P＜0.001；χ²=14.36，P＜0.001）。中央区转移恶性淋巴结共40枚，侧颈区转移恶性淋巴结共178枚，中央区转移淋巴结直径更小[（12.36±5.22）mm vs（16.96±6.93）mm，t=4.64，P＜0.001]，且高回声灶、混合型/周边型血流及2~3级血流分级等典型恶性超声特征患者比例更低（10.00% vs 33.71%；7.50%/35.00% vs 26.97%/48.88%；35.00% vs 66.29%），差异具有统计学意义（χ²=8.15、P=0.004；χ²=21.69、P＜0.001；χ²=12.63、P＜0.001）。超声对Ⅲ区、Ⅳ区的诊断敏感度较高（87.50%、89.83%），对Ⅱ区及Ⅴ区的诊断敏感度较低（55.00%、52.17%）；ROC曲线分析显示，超声对Ⅲ区（AUC=0.901）、Ⅳ区（AUC=0.897）及Ⅵ区（AUC=0.866）诊断效能优于Ⅱ区（AUC=0.731），差异具有统计学意义（Z=3.172、3.033、2.306，P=0.002、0.002、0.021），且对Ⅲ区、Ⅳ区的诊断效能优于Ⅴ区（AUC=0.741），差异具有统计学意义（Z=2.570、2.466，P=0.010、0.014）。初次仅行中央区淋巴结清扫的患者Ⅱ区复发率高于中央区联合侧颈区淋巴结清扫者（55.00% vs 31.58%，χ²=5.36，P=0.021）。

甲状腺乳头状癌术后转移恶性淋巴结多见于Ⅲ区、Ⅳ区，且超声诊断效能优越。混合型血流模式及高血流分级是鉴别良、恶性淋巴结的重要指标。但超声对Ⅱ区及Ⅴ区淋巴结转移诊断效能相对有限，应在随访时进一步结合多区域征象综合判断，以避免漏诊。

To investigate the distribution patterns and ultrasonographic features of cervical lymph node metastasis in postoperative papillary thyroid carcinoma (PTC) patients, as well as the diagnostic performance of ultrasound, in order to provide evidence for postoperative surveillance and clinical decision-making.

A retrospective analysis was conducted on 97 patients with papillary thyroid carcinoma who underwent postoperative follow-up at Sichuan Cancer Hospital from January 2023 to April 2025 and were found to have suspicious cervical lymph node metastasis on ultrasound and subsequently confirmed by pathology. Based on the pathology, the ultrasound-identified suspiciously malignant lymph nodes were divided into the malignant group and the benign group. The regional distribution of recurrent lymph nodes, ultrasonographic features, and clinical data were recorded. For intergroup comparisons, the t-test or Kruskal-Wallis H test was used for continuous variables, and the χ² test or Fisher's exact test (when applicable) was used for categorical variables. The diagnostic performance of conventional ultrasound for each cervical compartment, the impact of initial surgical dissection extent on subsequent recurrence sites, the ultrasonographic features of benign versus malignant lymph nodes, and the differences in ultrasonographic features of malignant lymph nodes between different compartments were evaluated. Receiver operating characteristic (ROC) curve analysis was performed, and the area under the curve (AUC) values were compared using the DeLong test.

Among the 97 patients included, lymph node metastases were most commonly observed in levelⅣ (60.82%) and levelⅢ (57.73%). Vascular patterns were effective indicators for differentiating benign from malignant lymph nodes: the malignant group exhibited a higher proportion of mixed/peripheral vascular pattern (Fisher's exact test, P＜0.001) and Adler grade 2-3 blood flow (χ²=14.36, P＜0.001). Metastatic lymph nodes in the central compartment had smaller diameters [(12.36±5.22) mm vs (16.96±6.93)mm, t=4.64, P＜0.001] and lower proportions of typical malignant ultrasonographic features, including hyperechoic foci (10.00% vs 33.71%; χ²=8.15, P=0.004), mixed/peripheral vascular pattern (7.50%/35.00% vs 26.97%/48.88%; χ²=21.69, P＜0.001), and Adler grade 2-3 blood flow (35.00% vs 66.29%; χ²=12.63, P＜0.001). Ultrasound demonstrated a high diagnostic sensitivity for levelⅢ (87.50%) and levelⅣ (89.83%), but a low sensitivity for levelⅡ (55.00%) and levelⅤ (52.17%). ROC curve analysis showed that the diagnostic performance for levelⅢ (AUC=0.901), levelⅣ (AUC=0.897), and levelⅥ (AUC=0.866) was superior to that for levelⅡ (AUC=0.731) (Z=3.172, 3.033, and 2.306; P=0.002, 0.002, and 0.021, respectively), and the performance for levelⅢand levelⅣ was also superior to that for levelⅤ (AUC=0.741) (Z=2.570 and 2.466; P=0.010 and 0.014, respectively). Patients who initially underwent central compartment dissection alone had a significantly higher recurrence rate in levelⅡ compared to those who received combined lateral neck dissection (55.00% vs 31.58%, P=0.021).

Postoperative recurrent lymph nodes in PTC are predominantly located in levels ⅢandⅣ, where ultrasound exhibits excellent diagnostic performance. Mixed vascular pattern and high Adler grade are important indicators for differentiating malignant from benign lymph nodes. However, the diagnostic efficacy of ultrasound for levelsⅡandⅤ is relatively limited, necessitating comprehensive evaluation incorporating multi-level signs during follow-up to avoid missed diagnosis.