Pathology and PET-FDG Correlation Studies Determine False Positive and Negative Malignancies of the Lung
C Chuang, D Raymond, JC Wandtke, H Xu. University of Rcohester Medical Center, Rochester
Background: Positron emission tomography (PET) with glucose analog F-18-flurodeoxyglucose (FDG-PET) is used to screen for tumors of the lung. It has been recognized that benign lung lesions can have an elevated FDG uptake with a standardized uptake value (SUV) >2.5 and low-grade cancers can have a SUV<2.5, but extensive analysis of neoplastic and non-neoplastic lesions of the lung correlated with FDG-PET studies are limited.
Design: 218 resected lung specimens from 1/2005 to 12/2007 were examined with their corresponding PET-CT scans. Of the 218, 34 were non-neoplastic [16 necrotizing granulomas (NG), 1 non-necrotizing granuloma (NNG), 14 pneumonias, 2 hemorrhages, 1 hyalinized nodule], 26 benign to low grade malignancies [1 hamartoma, 2 solitary fibrous tumors, 2 lymphomatoid granulomatosis, 7 carcinoids, 8 BACs, 7 well-differentiated (WD) adenocarcinomas (AC)] and 158 were moderately (MD) to poorly (PD) differentiated malignancies [77 ACs, 53 squamous carcinomas (SC), 6 adenosquamous carcinomas, 17 large cell carcinomas, 1 carcinosarcoma, 3 metastatic pleomorphic sarcomas]. SUV and size of the lesion determined by CT were obtained. Sensitivity and specificity were determined based on SUV>2.5, which is radiologically suspicious for malignancy. One-sided ANOVA correlated the histologic grade with the corresponding SUV.
Results: Sensitivity and specificity for suspected malignancies using SUV>2.5 cut-off was 0.87 (95% CI 0.81-0.91) and 0.58 (95% CI 0.41-0.74), respectively. Fifteen false positives (8 NGs, 1 NNG, 5 pneumonias, 1 hyalinized nodule) were non-neoplastic. There were 24 false negatives (3 carcinoids, 5 BACs, 2 WD ACs, 12 MD ACs, 2 MD SCs). A SUV>7.6 was 100% specific for malignancy. Utilizing criteria of SUV<2.5 and lesion size>12 mm as a predictor of benign pathology, the positive predictive value was 0.36 (95% CI 0.21-0.55) and the negative predictive value was 0.88 (95% CI 0.82-0.92). Mean SUVs of WD, MD, and PD carcinomas were 3.5 ( 3.2), 6.7 ( 4.2), 10.8 ( 5.9), respectively. One-sided ANOVA analysis of SUVs comparing the histologic grades of ACs and SCs showed a 3.2 SUV (95% CL 0.7-5.8) increase from WD to MD carcinoma and a 4.1 SUV (95% CL 2.2-5.9) increase from MD to PD carcinoma with a p value <0.01.
Conclusions: SUV utilization in the lung as a screening tool, although effective, has numerous pitfalls. False positives were often observed in necrotizing granulomas and pneumonias. False negatives not only included low-grade neoplasms but also high-grade malignancies.
Monday, March 9, 2009 9:30 AM
Poster Session I Stowell-Orbison/Autopsy Award # 228, Monday Morning