Improving diagnostic performance of 18F-FDG-PET/CT for assessment of regional nodal involvement in non-small cell lung cancer

To assess the diagnostic performance of combined 2-[18F]-fluoro-2-deoxy-d-glucose (18F-FDG) positron-emission tomography (PET)/computed tomography (CT) mediastinal blood pool (MBP) activity cut-off for staging nodal involvement, and to examine other variables that may improve the diagnostic performance of PET/CT in non-small cell lung cancer (NSCLC). All patients diagnosed with NSCLC who underwent endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) and 18F-FDG-PET/CT between June 2016 and August 2018 were included. Nodal station and nodal staging-based analyses were performed, comparing the MBP cut-off and five other PET/CT parameters (node maximum standardised uptake value [SUVmax], node/MBP SUVmax ratio, node/tumour SUVmax ratio, node short axis diameter, and node SUVmax/node short axis diameter ratio) with histopathology results. The optimal cut-off value for each PET/CT parameter was determined using receiver operating characteristic curve analysis. One hundred and thirteen patients with a total of 321 nodes with pathological sampling were included. Nodal activity above MBP on PET/CT demonstrated 97.4% sensitivity, 35.8% specificity, 32.8% positive predictive value, and 97.8% negative predictive value. Of the five other PET/CT parameters examined, the two most promising were node SUVmax and node/MBP SUVmax. The node SUVmax cut-off of 3.9 demonstrated 90.9% sensitivity and 61.9% specificity, and the node/MBP SUVmax cut-off of 1.7 demonstrated 90.9% sensitivity and 60.7% specificity. Compared to the MBP cut-off, use of a higher node/MBP SUVmax ratio cut-off and use of other PET/CT variables can improve the diagnostic performance of PET/CT for NSCLC nodal staging. In particular, specificity for detecting malignant nodal involvement is improved while maintaining high sensitivity. •Using different PET/CT cut-offs can improve diagnostic performance in NSCLC staging.•SUVmax cut-off 3.9 demonstrated sensitivity 90.9% and specificity 61.9%.•Node/MBP SUVmax cut-off 1.7 demonstrated sensitivity 90.9% and specificity 60.7%.