ABSTRACT
INTRODUCTION: Studies demonstrating limited accuracy of 'positive' and 'negative' lymph nodes on fluorodeoxyglucose (FDG) PET-CT in staging for lung cancer have led to guidelines stating mediastinal nodes enlarged on computed tomography, irrespective of FDG uptake, require endobronchial ultrasound (EBUS)-transbronchial needle aspiration (TBNA). However FDG uptake occurs on a continuous spectrum and the use of standardised uptake value (SUV)max ratios, rather than a binary classification, may have improved diagnostic accuracy. METHODS: This was a retrospective analysis of patients with lung cancer who had PET-CT and EBUS-TBNA in 2015-2018. Results from EBUS and the SUVmax ratio of sampled lymph nodes to mediastinal blood pool (SUVmax LN/MBP) were analysed. RESULTS: From 99 patients 102 malignant and 54 benign nodes were identified. The SUVmax range was 2.5-52 for malignant and 1.6-5.4 for benign nodes. The SUVmax LN/MBP was 1.3-23 for malignant and 0.7-2.3 for benign nodes. All nodes with SUVmax LN/MBP <1.3 were benign with 100% negative predictive value (NPV). All nodes with SUVmax LN/MBP >2.3 were malignant with 100% positive predictive value (PPV). CONCLUSION: In this relatively small sample, SUVmax LN/MBP <1.3 had a NPV of 100% for excluding malignant nodes and SUVmax LN/MBP >2.3 had a PPV of 100% for diagnosing malignant nodes. Using SUVmax ratios could obviate the need for staging EBUS in selected patients with resultant time and cost savings. Selecting different SUVmax ratios, chosen to provide high accuracies for the parameter of interest to change management, is a potentially powerful diagnostic tool that is overlooked when FDG uptake is only classified as 'positive' or 'negative'.
