In vivo assessment in cognitively normal individuals
In order to assess the specificity of tau radiotracers in vivo, PET studies have investigated their retention pattern in healthy subjects. Tau PET studies in cognitively normal (CN) elderly individuals using [
18F]THK tracers have shown that cortical retention, although above reference levels, was relatively low and mainly confined to the medial aspect of the temporal lobe [
45,
53]. Similar cortical findings were obtained using [
18F]AV-1451, with all studies conducted thus far showing some degree of retention located within temporal regions [
54‐
65].
In all these studies, however, locally high tracer retention was seen in a number of cerebral regions in CN subjects, both elderly and young, which seems to be off-target binding. For instance, studies have shown extensive in vivo binding of [
18F]AV-1451 and [
18F]THK tracers in the midbrain and basal ganglia, and of [
18F]AV-1451, but not for [
18F]THK5351, in the choroid plexus of CN subjects [
26,
45,
53,
61]. As reported in in vitro studies (see above), this is likely to reflect off-target binding to various entities such as MAO-A [
23], or pigmented or mineralizad vascular structures [
31,
37]. In addition, high subcortical retention in the white matter was noted with [
18F]THK5117, probably as a result of nonspecific binding to β-sheet structures present in myelin basic proteins [
30]. This was greatly diminished, however, with the
(S)-form of the tracer, [
18F]THK5317, and with the more recently developed [
18F]THK5351 [
26,
53]. Lastly, high retention of [
11C]PBB3 was reported in the dural venous sinuses of CN subjects [
28]; it is not yet clear, however, whether this reflects off-target binding.
In vivo assessment in Alzheimer’s disease
Several clinical stages have been defined in AD, including preclinical, symptomatic pre-dementia (prodromal), and dementia. With the development of molecular imaging, specific diagnostic criteria integrating amyloid PET imaging have been recently proposed to better define these stages [
66,
67]. It seems, however, that amyloid PET imaging alone does not discriminate well between symptomatic (prodromal and demented) stages of AD. There is thus a strong interest in investigating the regional retention of tau PET tracer in vivo at different stages of the pathology.
In patients with Alzheimer’s disease dementia: A fast growing number of in vivo studies aimed to assess the retention pattern of tau PET tracers in patients with a diagnosis of probable AD, in comparison to CN individuals. Most of the published studies in humans have thus far focused on the THK tracers or [
18F]AV-1451; one study compared the in vivo retention of the radiotracer [
11C]PBB3, however, in three patients with AD dementia and three CN subjects [
28], reporting higher tracer accumulation in patients compared to controls in several brain areas, predominantly medial temporal regions.
The first THK radiotracers developed (the racemic forms of [
18F]THK523, [
18F]THK5105, and [
18F]THK5117) showed important limitations, such as substantial overlap between clinical groups [
68,
69] or high retention in white matter [
30,
68,
69], which precluded simple visual assessment and prevented their future use in clinical settings. So far, the most promising radiotracers from this family appear to be [
18F]THK5317 and [
18F]THK5351. In vivo studies in AD dementia patients using these tracers have shown cortical uptake matching the distribution of tau deposits reported from histopathological studies, with retention in the inferior temporal region providing the best discrimination between patients and CN subjects [
26,
53]. [
18F]THK5351, however, has more favorable pharmacokinetics, less white matter binding, and a higher target-to-reference signal than [
18F]THK5317 [
70]. Other groups using [
18F]AV-1451 in vivo substantiated these findings by reporting good discrimination between AD dementia patients and CN subjects, with greater cortical retention in patients, mostly within the temporal cortex [
22,
45,
56,
59,
61]. The pattern of cortical retention in patients was again in agreement with the expected pattern of tau deposition in AD. Across studies, retention was predominant in the temporal cortex, with the inferior temporal gyrus appearing to be the best region for discriminating between AD dementia patients and CN subjects (Table
4).
Table 4
Cerebral regions showing significant group differences between AD patients and controls across studies
[18F]T807 | Johnson et al., 2016 [ 59] | 56 | 6 | | NS | + | + | | | + | | | |
| 20 | 20 | + | + | + | + | + | a | + | + | + | NS |
[18F]THK5117 | | 8 | 8 | + | NS | + | + | + | NS | + | + | NS | NS |
[18F]THK5317 | data from Chiotis et al., 2016 [ 53] | 9 | 9 | + | NS | + | + | NS | + | + | + | NS | NS |
[18F]THK5351 | Lockhart et al., 2016 [ 45] | 6 | 10 | | NS | NS | + | | a | + | + | NS | NS |
There is also an interest in the relationship between the patterns of tau deposition assessed in vivo and the symptomatology of clinical variants of sporadic AD, such as posterior cortical atrophy, logopenic variant of primary progressive aphasia, or behavioral/dysexecutive variant. Pathological studies have indicated that while these atypical forms share the pathological hallmarks of AD, they present with distinct neurodegenerative patterns, matching the symptomatology [
71,
72]. Case series describing the retention of [
18F]AV-1451 in vivo in posterior cortical atrophy, logopenic variant of primary progressive aphasia, and a behavioral variant of AD [
61,
73,
74] as well as in one non-amnestic AD patient [
61], have reported a neuroanatomical correspondence between the retention of the tracer and the clinical presentation for all variants, with [
18F]AV-1451 retention most prominent in the clinically affected regions.
In prodromal Alzheimer’s disease and mild cognitive impairment: Beyond the ability to discriminate AD dementia patients from CN subjects, a major challenge for tau radiotracers is their efficacy as early biomarkers, that is, their use as a sensitive tool for detecting early stages of AD tau pathology. In a recent study using [
18F]THK5317, the authors reported that not only patients with AD dementia but also prodromal AD patients (patients with mild cognitive impairment – MCI – and positive Aβ PET) had significantly greater cortical retention than CN subjects [
53]. There was however no statistical difference between prodromal AD and AD dementia patients in this sample, although a greater proportion of patients with AD dementia showed high [
18F]THK5317 retention in cerebral regions that are expected to be affected by tau pathology only late in the disease course. Other studies have reported that [
18F]AV-1451 retention best discriminated MCI patients from CN subjects in mesial temporal regions (parahippocampal cortex, and entorhinal cortex) [
56,
59]. As for the hippocampus, interestingly, some authors reported significant group differences [
56] while others did not [
59]. This discrepancy was probably due to differences between the studies in quantification methods and the studied populations: other than the differences in recruitment criteria, not all MCI patients in these two studies were amyloid positive (77 and 67%, respectively), meaning that a significant proportion were unlikely to be at an early stage of AD. In addition to these findings, Pontecorvo et al. [
75] reported that younger AD patients (i.e. under 75) had greater [
18F]AV-1451 cortical retention than older AD patients, and Cho et al. [
57] reported that patients with early-onset AD (i.e. < 65 years) had greater [
18F]AV-1451 cortical retention than patients with late-onset AD, as described in
post-mortem histopathology studies on NFTs and neuritic plaques [
76]. Of note, the same off-target binding reported in CN subjects was also observed in AD patients for all tracers [
26,
28,
53,
61].
Relationship between the retention of the tracers and clinical impairment: Several studies using [
18F]AV-1451 or THK radiotracers have started investigating the relationship between the regional tracer retention and concomitant cognitive performance in AD patients. They have reported a significant negative relationship between global cortical tracer retention and global cognitive status [
56,
68], and also between retention in the temporal cortex and global cognition [
30,
59,
77]. One longitudinal study also reported a significant positive relationship between increased [
18F]THK5117 retention in the temporal cortex and cognitive decline [
78]. Retention in the temporal cortex was also found to correlate with memory impairment in AD patients (across both prodromal and dementia stages) [
57,
77]. Specifically, it appears that worse performance on domain-specific tests was associated with greater retention in key regions implicated in the involved cognitive domain [
56,
61].
In preclinical Alzheimer’s disease: Conceptual and biomarker advances over the past decade have led to the identification of a preclinical phase of AD, recently formalized by new diagnostic criteria that integrate biomarkers for brain amyloidosis (i.e. CSF Aβ
42 and Aβ PET) and neurodegeneration (CSF tau, regional atrophy, and [
18F]fluorodeoxyglucose ([
18F]FDG) PET) [
66,
67,
79,
80]. Though these criteria for preclinical AD have not been formally applied in all studies that have thus far used tau PET imaging to investigate CN older adults, Aβ-negative subjects had only localized increases in medial temporal lobe retention, while Aβ-positive subjects, believed to be within the AD preclinical pathway, showed more extensive tracer retention, including in AD signature regions [
54,
64]. Comparison between Aβ-positive and Aβ-negative subjects, however, showed no group differences in hippocampal retention [
64]. A further study involving sub-classification of subjects into preclinical stage 1 (Aβ-positive, neurodegeneration-negative) and preclinical stage 2 (both Aβ- and neurodegeneration-positive) showed higher [
18F]AV-1451 retention in medial temporal regions at both stages 1 and 2, relative to Aβ-negative and neurodegeneration-negative subjects (stage 0), and higher levels in the inferior temporal gyrus at stage 2, relative to stages 0–1 [
60].
Of note, a highly interesting population to study preclinical stages of AD in is presymptomatic individuals carrying mutations involved in autosomal dominant AD. These individuals have been the focus of many research groups over the past years, as they will eventually develop AD, and thus offer the opportunity to assess in vivo the progression of pathological features before the onset of symptoms [
81,
82]. There are to date, however, no published reports on tau PET in presymptomatic cases of autosomal dominant AD.
In vivo assessment in non-AD proteinopathies
CBD and PSP, two diseases in the spectrum of frontotemporal lobar degeneration, which are characterized by atypical parkinsonism and substantial clinicopathological overlap [
83,
84], have received increased attention with the emergence of tau PET imaging. Both diseases are characterized by the deposition of abnormally hyperphosphorylated tau, mostly 4R, in tubular or straight filaments, in contrast to the PHFs in AD. Moreover, the spatial distribution of tau deposits in these diseases is distinct from that seen in AD [
85,
86]. High tau deposition (measured with [
18F]AV-1451, [
18F]THK5317 or [
18F]THK5351 PET) was observed in patients with a clinical diagnosis of PSP, in areas expected based on the neuropathological literature: the basal ganglia, thalamus, dentate nucleus of the cerebellum, and midbrain [
36,
53,
87‐
89]. The association between [
18F]AV-1451 retention in the basal ganglia and clinical deterioration in these PSP patients was not consistently reported. Concordance with pathological patterns of tau deposition was also found in patients with CBD: case-reports of Aβ-negative patients with clinical diagnoses in the CBD spectrum revealed increased tau deposition, as measured by [
11C]PBB3, [
18F]THK5317 and [
18F]THK5351, predominantly in white matter and the basal ganglia, but also in other cortical areas [
28,
35,
53].
Dementia with Lewy bodies and Parkinson’s disease are characterized by the presence of α-synuclein aggregates, although the presence of tau deposits similar to those in AD pathology are also commonly found [
90‐
92]. [
18F]AV-1451 retention in patients with dementia with Lewy bodies and Parkinson’s disease-related cognitive impairment, but not in cognitively unimpaired Parkinson’s disease patients, was found to be higher than in controls, although greatly variable [
93]; the [
18F]AV-1451 retention was negatively related to global cognitive function but not to the concomitant Aβ load. Another study comparing patients with dementia with Lewy bodies and patients with AD dementia reported a much lower cerebral retention of [
18F]AV-1451 in AD, and revealed that the retention in the medial temporal lobe could discriminate between the two disease groups [
94]. Though further studies are required, and while keeping in mind that the clinical distinction between dementia with Lewy bodies and AD can be challenging, these findings highlight the potential utility of tau imaging in the context of differential diagnosis.
Following a different approach, Hansen et al. and Cho et al. [
88,
95] took advantage of the reported off-target binding of [
18F]AV-1451 to neuromelanin [
37], and aimed at imaging the loss of dopaminergic neurons in the substantia nigra of patients with Parkinson’s disease [
88,
95]. Lower [
18F]AV-1451 nigral retention was observed in patients with Parkinson’s disease, in comparison to a control group, although the overlap between patients and controls limits the clinical translation of the findings. Further, nigral retention in patients with Parkinson’s disease did not correlate with dopamine transporter levels in the basal ganglia (measured by [
123I]FP-CIT single photon emission computed tomography), motor disability, age, or time since diagnosis.
In vivo retention of [
18F]AV-1451 was also assessed in cases carrying mutations of the MAPT gene: Bevan-Jones et al. [
96] described, in a patient with familial frontotemporal dementia due to a MAPT mutation (MAPT 10 + 16C > T), a retention pattern in agreement with the regional pattern of 4R tau pathology observed in the brain of the deceased father, carrier of the same mutation. Smith at al. [
97] studied the in vivo retention of [
18F]AV-1451 in three symptomatic patients (two with MCI, one demented) carrying a MAPT mutation (p. R406W); the latter mutation is pathologically characterized by the presence of cortical NFTs. Here again, the [
18F]AV-1451 retention pattern was in agreement with reported post-mortem findings on tau deposits, showing involvement of temporal and frontal regions with sparing parietal and occipital lobes [
98]. The authors suggested a progression pattern of tau in this mutation, although this requires further investigation in studies with a longitudinal design and larger sample sizes.
Taken together, these studies suggest that the developed tau PET tracers can image the expected regional distribution of tau pathology outside the AD spectrum, especially in tauopathies. This is, however, at odds with in vitro findings mentioned earlier, which suggests that [
18F]AV-1451 might not bind substantially to, or might bind only to a small fraction of, the 4R tau burden [
31,
37].
In vivo assessment in suspected non-AD pathophysiology
Operationalization of the National Institute on Aging-Alzheimer’s Association (NIA-AA) criteria for preclinical AD [
79] led to the identification of Aβ-negative CN individuals with positive neuronal injury biomarkers [
99]. Believed to represent non-AD etiologies, this group has been labeled “suspected non-AD pathophysiology” (SNAP). SNAP is thought to represent the in vivo equivalent of the recently described “primary age-related tauopathy” (PART), a concept currently under debate [
100], introduced to describe the frequent observation in autopsy studies of focal NFTs pathology, despite the absence or minimal presence of Aβ plaques [
101]. Several investigations using tau PET have made reference to SNAP as a possible explanation for the high percentage of Aβ-negative cases in CN individuals with an estimated Braak stage of I-II [
56] and for focally elevated cortical [
18F]AV-1451 retention [
57,
63]. Additional studies have described cases possibly representative of PART [
62,
95], although these also raised the possibility that AD pathology might be masking PART in preclinical individuals, with Aβ pathology below the detection threshold of Aβ PET imaging. Findings from the Harvard Aging Brain Study, however, do not support the hypothesis that SNAP is the in vivo counterpart of PART, as mean retention of [
18F]AV-1451 within the medial temporal lobe among SNAP individuals was almost identical to that seen in stage 0 subjects (CN, Aβ- and neurodegeneration-negative) and lower than levels in subjects at preclinical stages 1–2 [
60]. Importantly, this study highlights discordance between tau PET and neurodegenerative biomarkers used to define SNAP (i.e. hippocampal volume and [
18F]FDG PET), a finding that carries implications for staging criteria for both SNAP and preclinical AD.