Purpose: To develop an algorithm to maximize the diagnostic yield of positron emission tomography (PET)/computed tomog¬raphy (CT) by using defined attenuation and standardized uptake value (SUV) criteria.
Materials and methods: An IRB-approved, HIPAA-compliant retrospective review with waiver of informed consent of data in 1388 consecutive patients who underwent PET/CT for known or suspected lung cancer was completed, and 187 adrenal nodules were identified in 147 patients. Nodules were defined histologically or by size change (malignant, n = 37) or stability for more than 1 year (benign, n = 58). Nodules not sampled for biopsy and with less than 1 year of follow-up were considered indeterminate (n = 92). Diameter, mean attenuation, SUVmax and SUV ratio (nodule SUVmax/liver SUVavg) were compared with t test and receiver operating characteristic analyses. Sensitivity, specificity, posi¬tive predictive value, and negative predictive value were calcu¬lated for diameter > 3 cm, mean attenuation > 10 HU, nodule SUVmax > 3.1, and SUV ratio > 1.0. These were also calculated for higher SUVmax and SUV ratio thresholds that were found to exclude all false-positives. Diagnostic accuracy was compared by using the McNemar test (P
Results: In the study group of 147 patients (aged 42-88 years; mean, 65.5 years; 59 women), combined PET/CT with mean attenuation > 10 HU and SUVmax > 3.1 had 97.3% sensitivity and 86.2% specificity. Combined PET/CT with mean attenuation > 10 HU and SUV ratio > 1.0 had 97.3% sensitivity and 74.1% specificity. The accuracies of these threshold combinations (90.5% and 83.2%, respectively) were significantly different (P = .008). Applying a further cutoff of SUV ratio > 2.5 enabled identification of 22 of 37 metastatic lesions and exclusion of all fluorodeoxyglucose-avid benign nodules.
Conclusion: Definitive identification of many metastases can be accom¬plished by applying an SUV ratio cutoff of greater than 2.5, allowing pragmatic management of adrenal nodules that initially test positive with the combined PET/CT criteria SUVmax > 3.1 and mean attenuation > 10 HU.
