Abstract
(11)C-RO-963, (11)C-RO-643, and (18)F-RO-948 (previously referred to as (11)C-RO6924963, (11)C-RO6931643, and (18)F-RO6958948, respectively) have been reported as promising PET tracers for tau imaging based on in vitro and preclinical PET data. Here we describe the first, to our knowledge, human evaluation of these novel radiotracers. Methods: Amyloid PET-positive Alzheimer disease (AD) subjects and younger controls each received 2 different tau tracers. Dynamic 90-min scans were obtained after bolus injection of (11)C-RO-963, (11)C-RO-643, or (18)F-RO-948. Arterial blood sampling was performed on 11 healthy controls and 11 AD subjects. Regions were defined on MR images, and PET data were quantified by plasma reference graphical analysis (for total distribution volume) and target cerebellum ratio (SUV ratios of 60- to 90-min frames). SUV ratio images were also analyzed voxelwise. Five older controls each underwent 2 scans with (18)F-RO-948 for evaluation of test-retest variability. Four AD subjects underwent a repeated (18)F-RO-948 scan 6-22 mo after the first scan. Six additional healthy controls (3 men and 3 women; age range, 41-67 y) each underwent 1 whole-body dosimetry scan with (18)F-RO-948. Results: In younger controls, SUV(peak) was observed in the temporal lobe with values of approximately 3.0 for (11)C-RO-963, 1.5 for (11)C-RO-643, and 3.5 for (18)F-RO-948. Over all brain regions and subjects, the trend was for (18)F-RO-948 to have the highest SUV(peak), followed by (11)C-RO-963 and then (11)C-RO-643. Regional analysis of SUV ratio and total distribution volume for (11)C-RO-643 and (18)F-RO-948 clearly discriminated the AD group from the healthy control groups. Compartmental modeling confirmed that (11)C-RO-643 had lower brain entry than either (11)C-RO-963 or (18)F-RO-948 and that (18)F-RO-948 showed better contrast between (predicted) areas of high versus low tau accumulation. Thus, our subsequent analysis focused on (18)F-RO-948. Both voxelwise and region-based analysis of (18)F-RO-948 binding in healthy controls versus AD subjects revealed multiple areas where AD subjects significantly differed from healthy controls. Of 22 high-binding regions, 13 showed a significant group difference (after ANOVA, F((1,21)) = 45, P < 10(-5)). Voxelwise analysis also revealed a set of symmetric clusters where AD subjects had higher binding than healthy controls (threshold of P < 0.001, cluster size > 50). Conclusion:(18)F-RO-948 demonstrates characteristics superior to (11)C-RO-643 and (11)C-RO-963 for characterization of tau pathology in AD. Regional binding data and kinetic properties of (18)F-RO-948 compare favorably with other existing tau PET tracers.