Abstract
BACKGROUND: Clinical diagnosis of Alzheimer disease (AD) is only 70% accurate. Reduced cerebral blood flow (CBF) and metabolism in parieto-temporal and posterior cingulate cortex may assist diagnosis. While widely accepted that (18) F-fluoro-2-deoxyglucose positron emission tomography ((18) F-FDG PET) has superior accuracy to CBF-SPECT for AD, there are very limited head-to-head data from clinically relevant populations and these studies relied on clinical diagnosis as the reference standard. AIMS: To compare directly the accuracy of CBF-SPECT and (18) F-FDG PET in patients referred for diagnostic studies in detecting β-amyloid PET confirmed AD. METHODS: A total of 126 patients, 56% with mild cognitive impairment and 44% with dementia, completed both CBF-SPECT and (18) F-FDG PET as part of their diagnostic assessment, and subsequently underwent β-amyloid PET for research purposes. Transaxial slices and Neurostat 3D-SSP analyses of (18) F-FDG PET and CBF-SPECT scans were independently reviewed by five nuclear medicine clinicians blinded to all other data. Operators selected the most likely diagnosis and their diagnostic confidence. Accuracy analysis used final diagnosis incorporating β-amyloid PET as the reference standard. RESULTS: Clinicians reported high diagnostic confidence in 83% of (18) F-FDG PET compared to 67% for CBF-SPECT (P = 0.001). All reviewers showed individually higher accuracy using (18) F-FDG PET. Based on majority read, the combined area under the receiver operating characteristic curve in diagnosing AD was 0.71 for (18) F-FDG PET and 0.61 for CBF-SPECT (P = 0.02). The sensitivity of (18) F-FDG PET and CBF-SPECT was 76% versus 43% (P < 0.001), while specificity was 74% versus 83% (P = 0.45). CONCLUSIONS: (18) F-FDG PET is superior to CBF-SPECT in detecting AD among patients referred for the assessment of cognitive impairment.