Visual and Semiquantitative Analyses for F-18 Fluorodeoxyglucose PET Scanning in Pulmonary Nodules 1 cm to 3 cm in Size

Abstract

While visual assessment is the simplest way to evaluate positron emission tomography (PET) with fluorodeoxyglucose (FDG), faintly positive nodules are often difficult to evaluate. We performed visual and semiquantitative analyses of pulmonary nodules from 1 to 3 cm in size to determine the optimal method of analysis for PET data, especially for faintly positive nodules on visual assessment. Positron emission tomography data were analyzed for 161 pulmonary nodules from 1 to 3 cm in size (108 malignant and 53 benign nodules). On visual assessment, FDG uptake by the nodules was classified into three grades in comparison with mediastinal blood pool, (ie, definitely positive, faintly positive, and negative). In addition, FDG uptake was measured by the standard uptake value (SUV), the contrast ratio to the contralateral lung (CR-lung), and the contrast ratio to the cerebellum (CR-brain). Cut-off values for each variable were determined from receiver operating characteristics (ROC) curves, and the values were 2.5 for the SUV, 0.4 for the CR-lung, and 0.25 for the CR-brain. Nodules with FDG uptake above these cut-off values were defined as positive in each method. Visual assessment showed definitely positive for 80 nodules (64 malignant and 16 benign), faintly positive for 22 (17 malignant and 5 benign), and negative for 59 nodules (27 malignant and 32 benign). In the 139 nodules that were definitely positive or negative by visual assessment, there were no significant differences of sensitivity and specificity among the four methods (visual assessment, SUV, CR-lung, and CR-brain). In the 17 malignant nodules that were faintly positive, the SUV did not detect any true-positive nodules, but the CR-lung and CR-brain showed 9 and 5 true-positive nodules, respectively, resulting in a significantly higher sensitivity than the SUV (p < 0.001 and p = 0.02, respectively). Fifteen of 17 faintly positive malignant nodules (88%) were histologically well- or moderately-differentiated adenocarcinoma. The semiquantitative methods (SUV, CR-lung, and CR-brain) do not improve the efficacy of visual assessment alone for nodules graded as definitely positive or negative. However, faintly positive nodules on visual assessment should be evaluated by the CR-lung or CR-brain rather than the SUV, and are often differentiated adenocarcinomas.