US risk stratification system based on tumor burden for lateral lymph node metastasis in papillary thyroid carcinoma.

[OBJECTIVES] To develop a tumor burden score (TBS)-based risk stratification system (RSS) for preoperative discrimination of lateral lymph node metastasis (LLNM) and compare it with conventional US and TBS-only models.

[MATERIALS AND METHODS] This retrospective study included consecutive 404 adult patients diagnosed with PTC who underwent thyroid US and lateral lymph node dissection between January 2021 and October 2023. A prospective validation cohort was established with 174 patients enrolled from November 2023 to August 2024. Two novel tumor burden parameters, including D1/D2 and TBS score, were proposed. Logistic regression analyses were performed to assess associations between variables and LLNM in the training cohort. Features with p < 0.05 were incorporated into the LLNM RSS based on the Framingham risk score equation. Area under the receiver operating characteristic curve (AUC) was compared among the RSS, conventional US model, and TBS model in both cohorts.

[RESULTS] In both training and validation cohorts, the RSS showed improved performance for identifying LLNM compared with the TBS model (AUC, 0.809 vs. 0.703, p < 0.01 and AUC, 0.752 vs. 0.634, p = 0.03), and in the training cohort, the AUC of the RSS was superior compared with the conventional US model (AUC, 0.809 vs. 0.732, p < 0.01). Furthermore, we identified clustered punctate echogenic foci (PEF) distributed along the periphery of primary tumors as a robust independent predictor of LLNM (OR = 3.157, 95% CI: 1.520-6.557).

[CONCLUSION] The user-friendly RSS constructed based on the tumor burden concept outperformed conventional US and TBS models for distinguishing LLNM in PTC preoperatively.

[KEY POINTS] Question To develop a risk stratification system (RSS) for the prediction of lateral lymph node metastasis (LLNM) in papillary thyroid carcinoma (PTC), addressing the limitations of first-line ultrasound. Findings The tumor burden-based RSS achieved superior accuracy (AUC = 0.81) versus the conventional ultrasound-based model (AUC = 0.70), effectively evaluating the risk of LLNM preoperatively. Clinical relevance The RSS incorporating tumor burden parameters and peripheral punctate echogenic foci sign demonstrates strong predictive value for LLNM and enables risk-stratified surgical planning, avoiding unnecessary lymph node dissection in low-risk PTC while identifying occult metastases in high-risk cases.