Precision estimates of Relative and Absolute Cerebral Blood Flow in Alzheimer’s disease and Cognitively Normal individuals

Abstract

AbstractBackgroundAlzheimer’s disease is characterized by regional reductions in cerebral blood flow (CBF) and the gold standard for measuring CBF is [15O]H2O PET. Recently, proxies of relative CBF, derived from the early distribution phase of amyloid‐β and tau tracers, have gained attention. It has been hypothesized that relative CBF is likely to have higher precision (but not accuracy) than absolute CBF, as global variations in CBF are cancelled out. The present study assessed precision of [15O]H2O PET derived relative and absolute CBF measures, and compared them with relative CBF proxies estimated from [18F]florbetapir and [18F]flortaucipir scans.MethodDynamic [15O]H2O, [18F]florbetapir and [18F]flortaucipir test‐retest (TRT) PET datasets were included, consisting of eleven cognitively normal (CN) individuals for [15O]H2O, four CN and five AD participants for [18F]florbetapir, and six CN and eight AD participants for [18F]flortaucipir. [15O]H2O data were analysed using the standard single tissue compartment model and [18F]florbetapir and [18F]flortaucipir using the simplified reference tissue model (SRTM). Relative CBF (i.e. K1/K1’) was derived from absolute CBF (i.e. K1 of [15O]H2O) or estimated using SRTM (i.e. R1), in all cases using cerebellar cortex as reference tissue. CBF values were obtained for 22 cortical grey matter regions and TRT variability (eq.1) was calculated for all metrics. Differences (%) between test and retest measures were assessed using Bland‐Altman analysis and TRT variability for different tracers and between diagnostic groups were compared using Wilcoxon Signed‐Rank and Mann‐Whitney U‐tests.ResultAverage differences (M±SD) between test and retest measures were small for relative measures, [15O]H2O K1/K1’:‐2.0±7.8%, [18F]florbetapir R1:‐1.7±4.1%, [18F]flortaucipir R1:‐0.1±4.2%, but somewhat larger for [15O]H2O K1:‐5.7±14.6%,(Fig.1). In case of [15O]H2O, TRT variability was lower for K1/K1’ than forK1 (p<0.001). Furthermore, [18F]florbetapir and [18F]flortaucipir derived R1 showed lower TRT variability than [15O]H2O K1/K1’ (p<0.001). Only for [18F]flortaucipir, TRT variability was higher in the AD compared with the CN group.ConclusionRelative CBF has higher precision than absolute CBF derived from [15O]H2O PET scans. Furthermore, relative CBF proxies derived from commonly used amyloid‐β and tau tracers appear to have an even higher precision, possibly due to the compensatory effect of extraction fraction and/or the older scanner used for [15O]H2O scans.