Solitary pulmonary nodule (SPN), and even more SPNs incidentally detected in non-oncological subjects, still represents an arduous diagnostic challenge for clinicians. Aim of our study was to evaluate the 18F-Fluoro-deoxy-glucose (18F-FDG) diagnostic performance in one of the largest and strictly selected series of SPNs and in two SPN subgroups (≤1.5cm and >1.5cm).
Three-hundred and sixty-eight 18F-FDG Positron Emission Tomography/Computed Tomography (PET/CT) from 368 non-oncological subjects (233 males; mean age 64.5±11.2 years) with solid SPNs, ranging from 0.8cm to 3cm, incidentally detected and undetermined at diagnostic CT were retrospectively evaluated. SPNs were divided in two subgroups, according to their size: 162 small SPNs (0.8cm-1.5cm) and 206 larger SPNs (>1.5cm-3cm). Histology or radiological follow-up were used as reference standard for malignancy. Diagnostic performance of 18F-FDG visual analysis (uptake > mediastinal activity), ROC curve analysis for SUVmax and Bayesian analysis for 18F-FDG visual evaluation were assessed in all 368 SPNs and in the two subgroups.
From a total of 368 SPNs, 196 were malignant (53%) and 172 benign. In all 368 SPNs, 18F-FDG visual and ROC analysis (SUVmax cut-off=1.5) provided equivalent sensitivity and specificity values of 84% and 80%. For a positive or negative PET result, Bayesian analysis provided post-test probabilities of malignancy >80% and <20%, respectively. Regarding the two SPN groups, 18F-FDG visual analysis provided lower sensitivity (73%) and higher specificity (88%) in 162 small SPNs (≤1.5cm) and, conversely, higher sensitivity (90%) and lower specificity (71%) in 206 larger ones (>1.5cm), with an overall accuracy >80%. ROC analysis provided different optimal SUVmax cut-off (1.3 and 1.9), with sensitivity and specificity of 77% and 82% in SPNs ≤1.5cm and 85% and 79% in SPNs >1.5cm. Positive and negative post-test probabilities of malignancy were 79% and 16% in SPNs ≤1.5cm (vs 38% pre-test probability) and 85% and 22% in SPNs >1.5cm (vs 65% pre-test probability). In both groups, malignant SPNs had significant higher SUV (p<0.0001) than benign ones, whereas only malignant small SPNs were significantly larger (p=0.0054) than benign SPNs.
Our data indicate that 18F-FDG is clinically relevant to rule in or rule out malignancy in undetermined small SPNs detected in non-oncological subjects, as well as in larger ones, suggesting possible different clinical management according to SPN size and post-test probability of malignancy. Moreover, visual analysis can be considered an optimal diagnostic criterion, correctly detecting a wide range of malignant SPNs with different metabolic activity.