Is longitudinal tau PET ready for use in Alzheimer’s disease clinical trials?

Abstract

This scientific commentary refers to ‘Longitudinal tau PET in ageing and Alzheimer’s disease’, by Jack, Jr et al. (doi:10.1093/brain/awy059).

This scientific commentary refers to ‘Longitudinal tau PET in ageing and Alzheimer’s disease’, by Jack, Jr et al. (doi:10.1093/brain/awy059).

Tau PET imaging, using radiotracers like ¹⁸F-flortaucipir, has been shown to reliably detect the tau-containing paired helical filaments typical of Alzheimer’s disease dementia in vitro and in vivo (Smith et al., 2016). ¹⁸F-flortaucipir may also bind to tau aggregates in other tauopathies, including corticobasal degeneration and progressive supranuclear palsy, albeit to a much lesser extent compared to Alzheimer’s disease (Smith et al., 2017). Given the central role that accumulation of tau pathology is thought to play in the progression and clinical manifestation of Alzheimer’s disease, the ability to measure this pathology in vivo represents a major step forward and provides opportunities for early diagnosis and assessment of target engagement in clinical trials. As tau PET is a novel technology, longitudinal studies determining the change over time in the regional uptake of ¹⁸F-flortaucipir have not been available to date. In this issue of Brain, Jack and co-workers evaluate the change over 1 year in tau PET signal in a relatively large group of clinically unimpaired individuals with differing amyloid-β status (59 amyloid-β-negative and 37 amyloid-β-positive), as well as 30 amyloid-β-positive individuals with mild cognitive impairment or Alzheimer’s disease dementia (Jack et al., 2018).

The results reveal significant tau accumulation in the amyloid-β-positive clinically unimpaired group, in a pattern that is not restricted to the medial temporal lobe but also encompasses medial parietal areas, including the posterior cingulate cortex. This suggests that the initial accumulation of tau aggregates in Alzheimer’s disease may not be restricted to the medial temporal lobes as much as implied by Braak staging. Instead, Jack et al. demonstrate that early longitudinal tau aggregation coincides with regions showing accumulation of amyloid-β fibrils during the earliest stages of preclinical Alzheimer’s disease (Palmqvist et al., 2017). Although there was no significant change at the group level among the amyloid-β-negative unimpaired group, previous cross-sectional studies have revealed mild associations between older age and increased retention of ¹⁸F-flortaucipir in the medial temporal lobe (Lowe et al., 2018). This might be congruent with a very slow and subtle accumulation of tau aggregates in these regions over decades in amyloid-β-negative individuals. However, the observation that amyloid-β positivity was the strongest predictor of tau accumulation among unimpaired individuals reiterates that accumulation of amyloid-β fibrils is an important driver for the build-up of tau aggregates during preclinical Alzheimer’s disease (Price and Morris, 1999), at least to reach the levels detectable with tau PET imaging.

It is possible that PET imaging with tracers like ¹⁸F-flortaucipir or ¹⁸F-florbetapir may not be able to detect the earliest accumulation of either tau or amyloid-β, as these tracers reveal only certain aspects of the pathology that occur later in the disease process (Fig. 1). Specifically, the amyloid tracers (¹¹C-PiB and the ¹⁸F ligands) were developed from Thioflavin-T, and measure predominantly amyloid-β fibrils in cored/neuritic plaques, and more variably diffuse plaques. Thus, amyloid PET imaging techniques likely do not measure intraneuronal accumulation of amyloid-β in endosomes/multivesicular bodies, or the intracellular as well as extracellular amyloid-β oligomers, which probably precede the formation of plaques. ¹⁸F-flortaucipir was identified by screening different compounds using intact brain tissue sections containing paired helical filament tau from patients with Alzheimer’s disease. However, this tracer does not detect abnormal phosphorylation of tau or other post-translational modifications, which likely precede the formation of tau aggregates. Combining PET imaging with measurements of amyloid-β and tau variants in CSF may be a fruitful approach to assess very early changes related to amyloid-β and tau, and to understand how these events relate to the regional patterns detected with tau PET at different stages of the disease (Mattsson et al., 2017).

Figure 1.

Hypothetical sequence of events related to the progression of Alzheimer’s disease pathology and the ability to measure these changes with PET. PET images for three clinically normal older individuals are displayed, with amyloid PET (florbetapir) shown on the top and tau PET (AV1451) on the bottom for each participant. (A) Intracellular changes related to amyloid clearance and the formation of oligomers, as well as phosphorylated tau, occur before an abnormal signal is detected with PET. (B) Initial changes detected with amyloid PET are likely focal (posterior cingulate and medial prefrontal cortex) and capture extracellular cored amyloid plaques, whereas diffuse plaques likely occur earlier but are not detected with amyloid PET agents. (C) As participants become clearly positive on amyloid PET, many will also show evidence of elevated tau signal with PET in the medial temporal lobe and other cortical regions. Example clinical trial scenarios are shown in boxes alongside this progression. Here, we suggest how biomarkers could be used to determine target engagement of a therapeutic during phases 1–2. In phase 3, we posit that the outcomes that are most relevant to clinical change are those that reflect a downstream event relative to the drug target. For instance, a therapeutic that is able to act directly on amyloid plaques should also show an effect on tau PET and markers of neurodegeneration (ND) in addition to lowering the amyloid PET signal. Showing that a downstream event is altered will boost confidence that the drug is thwarting the Alzheimer’s disease dementia pathway and thus increasing the likelihood of clinically meaningful change. Aβ = amyloid-β.

It is as yet unknown whether high levels of amyloid-β pathology are required during all stages of the development and progression of tau pathology in Alzheimer’s disease. Alternatively, amyloid-β pathology might only be required for the initial build-up of toxic tau aggregates in certain susceptible cortical regions; the subsequent progression of tau accumulation throughout the brain may then at some stage become independent of amyloid-β (Hyman, 2011). Therapeutic interventions aimed at reducing amyloid-β aggregates could provide key insights into the interplay between amyloid-β and tau pathologies, by revealing whether lowering levels of toxic amyloid-β species prevents further build-up and spread of tau pathology. As more clinical trials integrate PET imaging and CSF measures of amyloid-β and tau, it is anticipated that answers to these questions will emerge within the next decade (Fig. 1).

It is also interesting to note from the paper by Jack et al. that the retention of ¹⁸F-flortaucipir increases simultaneously in large parts of the neocortex in amyloid-β-positive clinically impaired individuals. This might seem to argue against the assumption that tau pathology in Alzheimer’s disease spreads in a prion-like fashion. However, it could still be the case that tau spreads from neuron to neuron within highly connected neocortical regions when the concentrations of tau aggregates are very low, followed by a gradual accumulation of tau paired helical filaments in many regions simultaneously later in the disease. It is also possible that patients show heterogeneity in patterns of tau deposition (Pontecorvo et al., 2017); variance that would be masked at the group level and when averaged would appear as a distributed global pattern. Follow-up analyses that define subject-specific maps of elevated tau PET signal and examine these patterns over time may be able to address this possibility. Nevertheless, the pattern of results at the group level speaks against the idea that large amounts of tau fibrils are needed in a certain region before they can spread to the next region, and instead suggests a more distributed and continuous build-up of tau aggregates.

Besides giving clues about the evolution of tau pathology in Alzheimer’s disease, the study by Jack et al. also provides evidence that longitudinal tau PET might be used in clinical trials evaluating disease-modifying therapies. To evaluate target engagement and help accelerate trials, early studies in phase 1–2 using agents directed against tau-containing filaments should evaluate whether the tau PET signal decreases after 6–12 months of treatment. Additionally, tau PET may have an important role in evaluating therapies directed against amyloid-β. Numerous neuropathology and PET studies have shown that tau pathology is more closely linked to overt cognitive impairment and neurodegeneration, compared to amyloid-β accumulation (Pontecorvo et al., 2017), and the dominant theories of Alzheimer’s disease suggest that tau pathology mediates the impact of amyloid

Glossary

¹¹C-Pittsburgh compound B (PiB): A radiotracer originally derived from thioflavin-T that can be imaged using PET to reveal the abundance of insoluble β-sheet amyloid fibrils.

¹⁸F-florbetapir: A radiotracer to measure β-sheet amyloid fibrils that was developed after PiB. This compound is more convenient than PiB for clinical use given the longer half-life of ¹⁸F compared to ¹¹C.

¹⁸F-flortaucipir (¹⁸F-AV-1451): A radiotracer originally identified when screening compounds binding to neurofibrillary tangles in brain tissue from patients with Alzheimer’s disease that can be used to quantify tau aggregates in vivo with PET.

on cognition (Bennett et al., 2004). Consequently, it is quite likely that amyloid-β therapies need to slow down, stop or even reverse the cortical build-up of tau aggregates in order to prevent the clinical manifestation of this disease. Jack and colleagues demonstrate that the estimated sample sizes required to detect a significant change in tau PET over time are clearly smaller than those needed to detect a change in memory and other clinical indices. The sample size estimates to detect change were even smaller for MRI (cortical thickness). However, tau PET is a rather new methodology and work is still needed to optimize the postprocessing of longitudinally obtained tau PET images, including but not limited to methods for partial volume effect correction, definition of target regions of interest and the region of interest used for normalization (Southekal et al., 2017). Furthermore, changes in grey matter atrophy detected with MRI may not be specific to Alzheimer’s disease (Jack et al., 2013), and may instead be influenced by a confluence of age-related aetiologies such as vascular changes or aggregation of TAR DNA-binding protein 43 (TDP-43). Thus, a therapeutic agent that specifically targets amyloid-β or tau pathology may have a greater impact on the tau PET signal than on slowing brain atrophy (assuming that only some proportion of grey matter atrophy is attributable to amyloid-β and tau). We are therefore convinced that tau PET should be incorporated into most clinical trials recruiting individuals throughout the Alzheimer’s disease spectrum (clinically unimpaired and impaired), and that changes in tau PET may be a more important indicator of drug success than changes in amyloid PET in phase 3 trials, even when evaluating anti-amyloid-β therapies. Armed with the ability to measure accumulation of tau aggregates throughout the brain in vivo, we will now be able to visualize for the first time whether therapeutics reduce tau accumulation and importantly, determine whether these reductions are associated with clinical improvement (Fig. 1). This new combination of tools provides an unprecedented opportunity to directly test the hypothetical cascade of events that is proposed to underlie Alzheimer’s disease dementia.