Decker, Paul

7008 Background: Fluoro-deoxyglucose positron emission tomography (FDG-PET) is recommended for diagnosis and staging of known or suspected NSCLC. Meta-analyses examining the accuracy of FDG-PET to diagnose lung cancer demonstrated high sensitivity 94% and specificity 83% but were performed in select centers. The purpose of this study is to evaluate the accuracy of FDG-PET to diagnose NSCLC and examine enrolling site differences in the national prospective ACOSOG Z4031 trial. Methods: 1,074 patients with clinical stage I (cT1-2N0M0) known or suspected NSCLC were enrolled between 2004 and 2006 in the Z4031 study and underwent surgical resection. The final diagnosis was determined by pathological examination. FDG-PET results were abstracted from radiology interpretations included in the case report forms. FDG-PET avidity was categorized based on either radiologist description or reported maximum standard uptake value (SUV). The four categories were: not avid and not cancerous (SUV=0), low avidity and likely not cancerous (SUV>0 and <2.5), avid and probably cancerous (SUV>2.5 and <5) and highly avid and likely cancerous (SUV>5). Sensitivity analysis of FDG-PET diagnostic accuracy was performed for varying levels of avidity and by preoperative lesion size. Differences in accuracy by enrolling site were examined. Results: There were 51 enrolling sites in 39 cities with 969 eligible participants. Preoperative FDG-PET results were available for 682 participants. Lung cancer prevalence was 83%. FDG-PET sensitivity was 82% (95% CI: 79-85), and specificity was 31% (95% CI: 23-40). Positive and negative predictive values were 85% and 26%, respectively and accuracy improved with lesion size. There were 80 false positive scans and 69% were granulomas. False negative scans occurred in 101 patients (11were ≤10mm) with adenocarcinoma, squamous, bronchoalveolar cell and neuroendocrine tumors responsible for 64%, 12%, 10% and 8% of false negative results, respectively. Specificity did not differ between the 8 sites with >25 patients (p=0.74). Conclusions: In a national surgical population with clinical stage I NSCLC, FDG-PET to diagnose lung cancer performed poorly compared to published studies. Reasons for poor test performance should be explored.