A Nomogram to Predict Benign/Malignant Mediastinal Lymph Nodes Based on EBUS Sonographic Features

Area of Science:

• Pulmonology
• Oncology
• Medical Imaging

Background:

• Endobronchial ultrasound (EBUS) sonographic features are crucial for differentiating benign from malignant lymph nodes during transbronchial needle aspiration (TBNA).
• Accurate risk stratification of lymph nodes is essential for guiding diagnostic and therapeutic decisions in thoracic oncology.

Purpose of the Study:

• To identify key EBUS sonographic risk factors for lymph node malignancy.
• To develop and validate nomograms for predicting malignancy risk in lymph nodes.

Main Methods:

• A total of 1082 lymph nodes from 625 patients were analyzed, with data randomly allocated to training and validation sets.
• Logistic regression identified independent predictors of malignancy, including long-axis, short-axis, echogenicity, fusion, and central hilar structure (CHS).
• Two nomograms were constructed based on these predictors and validated for their discriminatory performance.

Main Results:

• Multivariate analysis identified five EBUS features (long-axis, short-axis, echogenicity, fusion, CHS) as independent predictors of malignancy.
• The 5-feature nomogram demonstrated good discrimination in both training (AUC=0.880) and validation (AUC=0.905) sets.
• A separate nomogram for EBUS-TBNA negative nodes, using long-axis, echogenicity, and CHS, also showed strong predictive power (AUCs of 0.890 and 0.834).

Conclusions:

• A novel scoring system utilizing two distinct nomograms can effectively predict lymph node malignancy.
• These nomograms offer a valuable tool for risk assessment in patients undergoing EBUS-TBNA procedures.