Study objective: To compare the performance of CT, radio-labeled 18F-fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET) blinded to CT, and FDG-PET visually correlated with CT, in the detection of N2 metastatic mediastinal lymph nodes (MLN) in patients with non-small cell lung cancer (NSCLC) and to hypothesize how PET could influence our actual mediastinal staging procedures.
Setting: Tertiary university hospital.
Patients and methods: In 50 patients with potentially operable NSCLC, thoracic CT, PET, and invasive surgical staging were performed. Blinded prospective interpretation was performed for each test and compared with surgical pathology results. Abnormalities on each of these staging examinations were recorded on a standard MLN map.
Results: The sensitivity, specificity, and accuracy in detecting N2 disease of CT was 67%, 59%, and 64%, respectively. Results of PET blinded to CT were significantly better (p=0.004): 67%, 97%, and 88%, respectively. For PET visually correlated with CT, this was 93%, 97%, and 96%, respectively. In 22 patients, both CT and PET were normal, and this was correct in all cases.
Conclusions: PET was significantly more accurate than CT in the MLN staging in NSCLC. Both examinations were complementary, since visual correlation with the anatomic information on CT improved the reader's ability to discriminate between hilar vs subaortic MLN FDG uptake, and between paramediastinal tumor vs tracheobronchial MLN FDG uptake. If the results can be confirmed in larger numbers of patients, PET could reduce the need for invasive surgical staging remarkably.