Head-to-head comparison of tau positron emission tomography tracers [¹⁸F]flortaucipir and [¹⁸F]RO948.

Abstract

[¹⁸F]flortaucipir binds to paired helical filament tau and accurately identifies tau in Alzheimer's disease (AD). However, "off-target" binding interferes with the quantification of [¹⁸F]flortaucipir in several brain regions. Recently, other tau PET tracers have been developed. Here, we compare [¹⁸F]flortaucipir with the novel tau tracer [¹⁸F]RO948 head-to-head in vivo. We included 18 participants with AD, three with amyloid-β-positive amnestic mild cognitive impairment, and four healthy controls. All underwent [¹⁸F]flortaucipir (80-100 min) and [¹⁸F]RO948 (70-90) PET scans within approximately 1 month. Four study participants underwent 0-100-min dynamic scanning. Standardized uptake value ratios (SUVRs) were created using an inferior cerebellar reference region. Neocortical tracer retention was highly comparable using both SUVR and distribution volume ratio-1 values obtained from dynamic scans. However, [¹⁸F]RO948 retention was significantly higher in the entorhinal cortex and lower in the basal ganglia, thalamus, and choroid plexus compared with [¹⁸F]flortaucipir. Increased off-target binding was observed with age for both tracers. Several cases exhibited strong [¹⁸F]RO948 retention in the skull/meninges. This extra-cerebral signal, however, did not affect diagnostic accuracy and remained relatively unchanged when re-examining a subsample after 1 year. Kinetic modeling showed an increase in [¹⁸F]flortaucipir SUVR over the scanning interval, compared with a plateau for [¹⁸F]RO948. [¹⁸F]RO948 and [¹⁸F]flortaucipir bound comparably in neocortical regions, but [¹⁸F]RO948 showed higher retention in the medial temporal lobe and lower intracerebral "off-target" binding. Time-dependent bias of SUVR estimates may prove less of a factor with [¹⁸F]RO948, compared with previous tau ligands.