|| Checking for direct PDF access through Ovid
Lymph node involvement in non-small-cell lung cancer (NSCLC) is a major factor in determining management and prognosis. We aimed to evaluate the accuracy of fluorine-18-fluorodeoxyglucose-PET/computed tomography (CT) for the assessment of nodal involvement in patients with NSCLC.In this retrospective study, we included 61 patients with suspected or confirmed resectable NSCLC over a 2-year period from April 2013 to April 2015. 221 nodes with pathological staging from surgery or endobronchial ultrasound-guided transbronchial needle aspiration were assessed using a nodal station-based analysis with original clinical reports and three different cut-offs: mediastinal blood pool (MBP), liver background and tumour standardized uptake value maximal (SUVmax)/2.Using nodal station-based analysis for activity more than tumour SUVmax/2, the sensitivity was 45%, the specificity was 89% and the negative predictive value (NPV) was 87%. For activity more than MBP, the sensitivity was 93%, the specificity was 72% and NPV was 98%. For activity more than liver background, the sensitivity was 83%, the specificity was 84% and NPV was 96%. Using a nodal staging-based analysis for accuracy at detecting N2/3 disease, for activity more than tumour SUVmax/2, the sensitivity was 59%, the specificity was 85% and NPV was 80%. For activity more than MBP, the sensitivity was 95%, the specificity was 61% and NPV was 96%. For activity more than liver background, the sensitivity was 86%, the specificity was 81% and NPV was 92%. Receiver-operating characteristic analysis showed the optimal nodal SUVmax to be more than 6.4 with a sensitivity of 45% and a specificity of 95%, with an area under the curve of 0.85.Activity more than MBP was the most sensitive cut-off with the highest sensitivity and NPV. Activity more than primary tumour SUVmax/2 was the most specific cut-off. Nodal SUVmax more than 6.4 has a high specificity of 95%.