Lymph Node Reporting and Data System (LN-RADS)-Retrospective Evaluation for Ultrasound Classification of Superficial Lymph Nodes.

INTRODUCTION

The evaluation of lymph nodes (LNs) in patients with suspected oncological disease is a crucial factor influencing further diagnostics and management. However, there is a lack of a dedicated system for the precise and comprehensive assessment of LNs. To address this gap, we developed the Lymph Node Reporting and Data System (LN-RADS).

METHODS

This retrospective multiparametric analysis included the assessment of 719 LNs in 489 patients. The images were evaluated by three radiologists using the LN-RADS scale, assigning each case to one of six group: 1 (normal), 2 (steatotic), 3 (reactive), 4a (low suspicion of malignancy) vs. 4b (high suspicion of malignancy), and 5 (definitely malignant) and were then correlated with histopathological results. The diagnostic performance of LN-RADS was validated. The analysis of 12 morphological features of LNs was performed to identify predictors of malignancy.

RESULTS

Histopathological analysis confirmed 389 malignant and 330 benign LNs. LN-RADS achieved 89% sensitivity, 85% specificity, and 87% accuracy in the diagnosis of malignant LN. The observed risk of malignancy by group was 0% for LN-RADS 1, 0% for LN-RADS 2, 2% for LN-RADS 3, 31% for LN-RADS 4a, 77% for LN-RADS 4b, and 97% for LN-RADS 5. Cohen's kappa statistic indicated substantial inter-reader agreement. Among the evaluated features, the strongest predictor of malignancy was the cortex thickness diameter, with a threshold value of ≥6 mm (82% accuracy; AUC = 0.894).

CONCLUSIONS

This study demonstrated the high efficacy of the LN-RADS system in distinguishing between benign and malignant lymph nodes and in stratifying malignancy risk. It also showed substantial inter-rater agreement.