Abstract
455
Objectives Cerebral tau neurofibrillary tangles (NFTs) are linked with pathogenesis and disease severity in Alzheimer disease (AD). New positron emission tomography (PET) tracers, including [18F]AV-1451 (also known as [18F]T807) (1,2), show promise for quantifying tau deposits in vivo. This study evaluates several image-based methods for quantifying [18F]AV-1451 PET based on neuropathologic staging systems (3-5), comparing kinetic analysis with simpler quantification methods for future studies.
Methods Dynamic [18F]AV-1451 PET imaging (data provided by Avid Radiopharmaceuticals, Inc.) from 10 subjects (4 control [CON], 3 mild cognitive impairment [MCI], 3 AD) were analyzed using an automated T1 MRI-based atlas to obtain volumes-of-interest (VOI) and time activity curves (TAC). Atlas VOIs were combined into five separate VOIs following the Braak staging system for cerebral tau deposition (5), starting with VOI Stage I-II encompassing the hippocampus and parahippocampal / ambient gyri, and subsequent VOI stages III through VI encompassing the additional structures that match the known typical histopathologic pattern of spread of NFTs. Arterial input functions (AIF) for compartmental modeling were based upon petrous carotid artery TAC corrected for tracer clearance from plasma using first order kinetics based on published mice data (1). Tracer uptake in brain regions was assessed using compartmental modeling (2-tissue 4-parameter; 2T-4k), reference Logan graphical analysis, Simplified Reference Tissue Model 2 (SRTM2), and Ichise's Multilinear Reference Tissue Model 2 (MRTM2) to obtain distribution volume ratios (DVR). These measures were compared to simpler, clinically practical uptake measures using PET data from 80-100 min to calculate standardized uptake value ratio (SUVR). Cerebellum was the reference VOI for all analyses. Agreement between uptake measures was analyzed using Pearson correlation. Group differences in tracer uptake were assessed using the Kruskal-Wallis test.
Results Subjects showed a range of tau deposition, with higher uptake in MCI versus controls and AD versus MCI. For example, mean Stage III SUVR measured 1.0±0.1 for CON, 1.65±0.3 for MCI, and 2.3±1 for AD. 2T-4k DVR showed very high correlation with simpler DVR measures (Logan, SRTM2, MRTM2) and SUVR (r2=0.94-0.99). Correlation was also very high among the simpler DVR measures (r2=0.98-0.99) and between the simpler DVR measures and SUVR (r2=0.89-0.94). Relative to 2T-4k DVR, Logan, SRTM2, and MRTM2 DVR showed ~20% negative bias (linear regression slope ≍ 0.8 with y-intercept ≍ 0.2). In this select small population, SUVR showed significant difference across subject groups for all VOIs (p=0.035-0.038).
Conclusions The very high correlations among uptake measures that range from 2-tissue 4-parameter compartmental model to SUVR not only help to validate these methods for quantifying [18F]AV-1451 retention, but also suggest that the clearance of [18F]AV-1451 from plasma in humans may be consistent across subjects and similar to mice. This study indicates that SUVR is a reasonable method to quantify [18F]AV-1451 based upon its high correlations with more rigorous DVR measures and its ability to distinguish CON, MCI, and AD subjects in this early analysis.