Early‐phases [18F]florbetapir and [18F]flutemetamol images as proxies of brain metabolism in the Alzheimer’s disease spectrum

Abstract Background Alzheimer’s disease (AD) neuropathologic changes are β‐amyloid (Aβ) deposition, pathologic tau, and neurodegeneration. Dual‐phase amyloid‐PET might be able to evaluate Aβ deposition and neurodegeneration with a single tracer injection. Early‐frames of amyloid‐PET scans provide a proxy for cerebral perfusion, which has shown good correlations with neural dysfunction measured through metabolic consumption, while the late‐frames depict amyloid distribution. Our study aims to assess the comparability between early frames of amyloid‐PET scans and [ 18 F]fluorodeoxyglucose (FDG) PET in their topography and their ability to discriminate patients in the AD continuum. Method 166 subjects evaluated at the Geneva Memory Center (Cognitively Unimpaired, CU = 61; Mild Cognitive Impairment, MCI = 73; Dementia = 21) underwent early‐phase amyloid‐PET – using either [ 18 F]florbetapir (eFBP) (n = 95) or [ 18 F]flutemetamol (eFMM) (n = 72) – and FDG PET. Standardized uptake value ratios (SUVR) were extracted within key regions sensitive for AD (metaROI) to evaluate the correlation of eFBP/eFMM and their respective FDG PET scans. To compare the discriminative power of eFBP, eFMM, and FDG SUVR between Aβ‐negative CU and Aβ‐positive impaired patients, receiver operating characteristic analyses and voxel‐wise t‐tests were performed. SPM single‐subject procedure was applied to investigate hypometabolic and hypoperfusion maps and their spatial overlap by means of DICE. Result Strong positive correlations were found between eFBP/eFMM and FDG SUVR independently of Aβ status and Aβ radiotracers (r>0.85, p<0.001). Both eFBP and eFMM, and FDG SUVR significantly discriminated AD cognitively impaired patients from CU in the metaROI (AUC FBP = 0.768; AUC FMM = 0.778), with FDG performing slightly better than others (AUC FDG = 0.797). For all tracers, voxel‐wise analyses confirmed reduced signals in regions typically belonging to AD hypometabolic pattern in AD patients when compared to controls. SPM single‐subject analysis revealed different hypometabolism and hypoperfusion patterns with a high spatial concordance, differing between stages (DICE average: AD dementia = 0.91; Aβ‐positive MCI = 0.61). Conclusion Distribution of perfusion was comparable to that of metabolism at group level and at single‐subject level. Our findings indicate that eFBP and eFMM imaging provide information closely related to brain regional glucose metabolism supporting their use for individual classification in clinical practice.