Purpose: To evaluate whether PET/low-dose CT (ldCT) using (18)F-fluorodeoxyglucose (FDG) improves characterization of indeterminate single pulmonary nodules (SPNs) in patients at high risk of lung cancer.
Methods: Retrospective analysis of 307 patients who underwent FDG-PET/CT for indeterminate SPNs identified 93 (70 men, age range 46-90 years) at high risk of lung cancer (age >40 years, minimum 10 pack-year smokers). SPNs were evaluated for the presence and intensity of FDG avidity and ldCT patterns. The performance of visual and semiquantitative FDG-PET/ldCT algorithms for characterization of SPNs was compared to that of ldCT. Incongruent FDG-PET and ldCT patterns were analyzed for significance in further patient management.
Results: Malignancy was diagnosed in 38% patients. FDG avidity defined 33 SPNs as true-positive (TP) and 2 as false-negative (FN) (malignant), and 41 as true-negative (TN) and 17 as false-positive (FP) (benign). For SUVmax of 2.2 (by ROC analysis) there were 27 TP, 8 FN, 48 TN and 10 FP SPNs. LdCT defined 34 TP, 1 FN, 28 TN and 30 FP lesions. Of the FP lesions on ldCT, 60% were FDG-negative. Visual PET/ldCT analysis had a sensitivity of 94%, a specificity of 70%, an accuracy of 80%, a positive predictive value (PPV) of 66%, and a negative predictive value (NPV) of 95% as compared to 77%, 83%, 81%, 73%, 86% for semiquantitative PET/ldCT and 97%, 48%, 66%, 53%, 96% for ldCT, respectively. Both PET/ldCT algorithms had statistically significantly higher specificity and accuracy than ldCT. Semiquantitative analysis showed significantly higher PPV and lower sensitivity and NPV than found with ldCT.
Conclusion: A single screening procedure encompassing FDG-PET and ldCT may improve screening for lung cancer in high-risk patients. The significantly improved specificity may potentially reduce FP ldCT results and further unnecessary invasive procedures.