2882 - Performance of Node-RADS Scoring System for a Standardized Assessment of Normal-Sized Cervical Lymph Nodes in Nasopharyngeal Carcinoma

Purpose/Objective(s): Accurate cervical lymph nodes (CLNs) diagnosis is pivotal for nasopharyngeal carcinoma (NPC) staging and treatment. Current radiologic diagnostic criteria for CLNs define a short-axis diameter of =10 mm as positive, based on pathological analyses of metastatic CLNs from other head and neck cancers. However, the criteria may lack sensitivity, potentially leading to the oversight of some normal-sized malignant LNs. The node reporting and data system (Node-RADS), a structured approach proposed to enhance diagnostic accuracy and consistency across all anatomical lymph node sites. This study aims to assess the performance of Node-RADS in the context of NPC, focusing on its utility in identifying malignant normal-sized LNs.

Materials/Methods: Patients with confirmed NPC who underwent cervical fine needle aspiration cytology (FNAC) and magnetic resonance imaging (MRI) from March 2014 to April 2023 were retrospectively included. Two experienced radiologists, blinded to pathological results, scored the LNs using Node-RADS. Pathological results was the gold standard. Diagnostic performance was evaluated through receiver operating characteristic (ROC), sensitivity and specificity. The correlation between scores, configuration criteria (texture, border, shape) and pathological results was statistically analyzed.

Results: A total of 272 patients were finally included, with 346 CLNs with pathological results were analyzed. Of the 346 CLNs, 216 were normalized sizes and 130 were met the current criteria. Node-RADS scores correlated strongly with histological metastases across all analyzed LNs (?² = 271.7, Cramer's V = 0.9, P < 0.001) and the normal-size LNs (?² = 123.8, Cramer's V = 0.8, P < 0.001). Among configuration features, texture showed the strongest association with metastases (?² = 155.3, Cramer's V = 0.7, P < 0.001). Node-RADS demonstrated superior diagnostic performance compared to current criteria (N=346, AUC: 0.919, se: 84.2%, sp: 99.5% VS AUC: 0.791, se: 71.2%, sp: 87%). For 216 normal-sized LNs, 42 (19.4%) were pathologic positive, and the positive rate of pathology is increased with the Node-RADS score: Score 1: 0/20 (0%), Score 2: 0/102 (0%), Score 3: 20/72 (28%), Score 4: 22/22 (100%), notably, score 4 exhibit 100% pathological positivity for metastasis. In detail, there is 10 LNs located in level I (short axis range:8.1-9.8mm), 8 in level II (5.7-9.5mm), 1 in level III (6mm), 2 in level V (4.6-7.8mm), 1 in level VIII (4.9mm).

Conclusion: Node-RADS significantly improves diagnostic accuracy for lymph node metastases in NPC, particularly for normal-sized LNs. Our study demonstrates that Node-RADS scores of 4 in normal- sized LNs can be classified as metastatic nodes (specificity 100%), scores of 3 will be suggested FNAC to confirm pathologic result, and scores of 1–2 just follow-up observation. Node-RADS minimizes unnecessary invasive biopsies while maintaining diagnostic precision, supported by the system’s high specificity for standardized LN assessment.