A review of the flortaucipir literature for positron emission tomography imaging of tau neurofibrillary tangles

Abstract Alzheimer’s disease is defined by the presence of β-amyloid plaques and neurofibrillary tau tangles potentially preceding clinical symptoms by many years. Previously only detectable post-mortem, these pathological hallmarks are now identifiable using biomarkers, permitting an in vivo definitive diagnosis of Alzheimer’s disease. 18F-flortaucipir (previously known as 18F-T807; 18F-AV-1451) was the first tau positron emission tomography tracer to be introduced and is the only Food and Drug Administration-approved tau positron emission tomography tracer (Tauvid™). It has been widely adopted and validated in a number of independent research and clinical settings. In this review, we present an overview of the published literature on flortaucipir for positron emission tomography imaging of neurofibrillary tau tangles. We considered all accessible peer-reviewed literature pertaining to flortaucipir through 30 April 2022. We found 474 relevant peer-reviewed publications, which were organized into the following categories based on their primary focus: typical Alzheimer’s disease, mild cognitive impairment and pre-symptomatic populations; atypical Alzheimer’s disease; non-Alzheimer’s disease neurodegenerative conditions; head-to-head comparisons with other Tau positron emission tomography tracers; and technical considerations. The available flortaucipir literature provides substantial evidence for the use of this positron emission tomography tracer in assessing neurofibrillary tau tangles in Alzheimer’s disease and limited support for its use in other neurodegenerative disorders. Visual interpretation and quantitation approaches, although heterogeneous, mostly converge and demonstrate the high diagnostic and prognostic value of flortaucipir in Alzheimer’s disease.


Graphical Abstract Introduction
Alzheimer's disease is a clinicopathological entity 1,2 characterized by progressive memory loss, cognitive dysfunction and the inability to perform activities of daily living.4][5][6] The defining hallmark features of Alzheimer's disease are β-amyloid (Aβ) plaques and tau neurofibrillary tangles (NFTs). 1,3Previously only detectable post-mortem, positron emission tomography (PET) biomarkers 1,7,8 now permit a pathological diagnosis before death.These biomarkers aid in diagnosis, staging of disease, evaluating disease progression and assessing the efficacy of potential therapeutics.Flortaucipir (previously 18 F-T807; 18 F-AV-1451) is currently approved by the U.S. Food and Drug Administration (FDA) as Tauvid™: 'a radioactive diagnostic agent indicated for PET imaging of the brain to estimate the density and distribution of aggregated tau neurofibrillary tangles (NFTs) in adult patients with cognitive impairment who are being evaluated for Alzheimer's disease.' 9 It was the first tau PET tracer to be introduced and has been widely adopted and validated in several research and clinical settings.Here, we present an overview of the published literature on flortaucipir PET imaging of NFTs.We considered all accessible peer-reviewed literature from a PubMed search of the term '(Flortaucipir) OR (Tauvid) OR (AV1451) OR ("AV 1451") OR ("AV-1451") OR (T807) OR (T-807) OR ("T 807")' covering all available dates until 30 April 2022; 605 papers were identified of which 462 were included and 143 were excluded as they were not original research articles or relevant to flortaucipir.An additional 12 papers that did not meet search criteria were included as they were considered to be relevant to this review.

Flortaucipir in typical Alzheimer's disease, MCI, and pre-symptomatic populations
Much of the published literature [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28] indicates that the binding of flortaucipir to tau NFTs parallels NFT accumulation described as Braak stages, 10,[29][30][31] especially for Aβ-positive subjects, 30,32 and correlates with clinical symptoms and measures of neurodegeneration. 33Braak Stages I and II are used when NFT involvement is predominant in transentorhinal, entorhinal and hippocampal regions, Stages III and IV when limbic regions, medial and inferior temporal lobes are involved and Stages V and VI when NFTs may be found widespread in the neocortex. 29Distinct Braak stage-specific patterns of flortaucipir binding were reported in a proportion like that seen at autopsy in patients with Alzheimer's disease. 34Regional flortaucipir distribution in most cases paralleled the expected (Braak-like) spatial progression and in turn was significantly associated with amyloid status, diagnostic category and measures of global cognition. 35t should be noted that throughout the review, references to clinical syndromes (mild cognitive impairment [MCI], dementia, etc.) may or may not be supported by evidence of the underlying aetiology (Alzheimer's pathology, etc.).In general, the terminology reflects that utilized in the specific literature to ensure clarity in the reporting of those findings.

Flortaucipir positron emission tomography and pathology at autopsy
Flortaucipir penetrates the blood-brain barrier to bind NFTs, allowing in vivo visualization of the density and distribution of neocortical NFT accumulation.Its correlation with NFT pathology at autopsy has been demonstrated by several studies.
A study 36 evaluating flortaucipir and autopsy findings in 64 patients had a short mean difference of 2.6 months between flortaucipir scan and autopsy.The standard of truth was tau pathology consistent with Braak Stages V and VI or high Alzheimer's disease neuropathological change (ADNC) consistent with Braak Stages V and VI NFT pathology and extensive Aβ plaque pathology.The ADNC score can be 'none, low, intermediate and high', 37 and such determination is based on three different pathology scores capturing (i) amyloid plaques with the Thal phases method 38 ; (ii) NFT stage with the Braak method 29 and (iii) neuritic Aβ score with the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) method. 39A high ADNC score corresponds to advanced Thal (Thal Phase 4/5) and Braak stages (Braak V or VI), with at least moderate CERAD score. 37Majority read of the visual evaluation by five raters found that flortaucipir had a sensitivity of 92.3% and specificity of 80.0% for identifying Braak Stages V and VI NFT pathology as well as a sensitivity of 94.7% and specificity of 80.8% for identifying high ADNC. 36he study was replicated with a separate five readers and an additional 18 patients (n = 82) with similar findings.
A second study 40 evaluated autopsy findings and quantitative flortaucipir (n = 26) using the standard uptake value of flortaucipir in a meta-region of interest (ROI) consisting of amygdala, entorhinal cortex, fusiform, parahippocampal as well as inferior temporal and middle temporal gyri and were normalized to the cerebellar crus to obtain standardized uptake value ratios (SUVrs).All participants with NFT pathology of Braak Stage IV or greater had an elevated (abnormal) flortaucipir SUVr, which identified Alzheimer's disease spectrum pathology with a sensitivity of 87% and specificity of 82%.This study only had two participants at Braak Stage IV, which were close to the internally derived threshold, and in a follow-up study with 143 subjects, it was stated that elevated flortaucipir SUVr was associated with Braak V/VI. 41Similarly, high correspondence was found between in vivo flortaucipir retention and neuropathology in Alzheimer's disease patients (n = 8) with flortaucipir SUVr quantitation reliably detecting advanced Alzheimer's disease pathology (Braak VI). 42However, participants with earlier Braak stages (Braak I-IV) did not show elevated tracer uptake compared with young, tau-negative controls.
In addition to autopsy measures of NFT stage and Aβ, flortaucipir also correlates with immunohistology measures of tau.These important results support the finding that flortaucipir maps to the density as well as the distribution of NFTs in the brain.Further, quantitation of cortical regional flortaucipir binding was found to be significantly correlated with quantitative P-tau immunohistochemistry, 40,43,44 In this case study, flortaucipir also significantly (P < 0.001) correlated with other post-mortem findings, including density of tau-positive neurites, intrasomal tau tangles and total tau burden.Another study (n = 22) 45 provided evidence that flortaucipir bound in regions precisely mirroring the pattern of P-tau immunohistology staining across different Braak stages.

Flortaucipir and amyloid positron emission tomography
Aβ plaques, another Alzheimer's disease hallmark feature, can be assessed in vivo using FDA and European Medicines Agency-approved PET imaging tracers ( 18 F-florbetapir, 18 F-florbetaben and 18 F-flutemetamol) and research-only tracers [ 18 F-flutafuranol (NAV4694) and 11 C-Pittsburgh Compound B].][48][49][50][51][52] A recent study specifically investigated whether elevated global flortaucipir signal would be an accurate predictor of amyloid-PET positive status, finding 98% accuracy. 53everal other studies have provided data in support of these observations, overall showing that regardless of the method adopted (e.g. the threshold or quantitation approach used), elevated neocortical flortaucipir signal was rarely observed in patients without concomitant elevated neocortical amyloid-PET signal.][56][57][58][59][60] Even in a predominantly MCI cohort, 72% of those with elevated flortaucipir signal limited to the medial temporal region (likely an earlier measure of tau positivity than neocortical signal) also had elevated neocortical amyloid-PET signal. 61These findings imply that elevated flortaucipir signal may support the identification of both NFT and Aβ plaque pathology, consistent with the autopsy results (above).That is, elevated flortaucipir indicated both Braak V/VI NFT levels and high ADNC (Braak V/VI plus elevated amyloid).This observation is consistent with neuropathology data from autopsy studies, showing that an advanced Braak V/VI NFT stage is in most cases associated with widespread and dense amyloid plaque accumulation. 62hese findings also align with suggestions that global tau, specifically tau beyond the medial temporal lobe, is associated with, and may be dependent on, Aβ accumulation. 55,63,64ome investigators postulate that increased Aβ accumulation might affect subsequent spread of tau via intrinsic neuronal networks. 13,23,636][67] Thus, a sequential cascade of events is hypothesized where local, medial temporal tau precedes global Aβ pathology, which, in turn, precedes global tau pathology.This cascade is supported by reports of the rate of Aβ accumulation 68 and baseline Aβ [69][70][71][72] predicting flortaucipir accumulation.In addition, antecedent elevations in Aβ, primarily in inferior temporal and parietal regions, were observed ∼5 years prior to elevated flortaucipir, primarily in inferior temporal and parietal regions 18,73,74 Although elevated flortaucipir signal was consistently shown to be associated with an elevated amyloid-PET scan, the reverse was not true-an elevated amyloid-PET scan was not always associated with an elevated flortaucipir scan.Similarly, a negative flortaucipir PET scan was not always associated with a negative amyloid-PET scan.It should be noted, however, that one study suggested high levels of Aβ binding (Centiloid >50CL) at baseline were indicative (80% sensitivity and specificity) of an elevated flortaucipir scan 5 years later. 74ven in non-Alzheimer's disease dementia and cognitively unimpaired (CU) participants, elevated amyloid-PET SUVr was shown to be the strongest predictor of subsequent elevated flortaucipir signal. 75tudies have found that flortaucipir SUVr was significantly higher for Aβ+ MCI and Alzheimer's disease relative to the respective Aβ− diagnostic groups. 55Flortaucipir SUVr was higher for the Aβ+ versus Aβ− groups irrespective of neurodegeneration status 76 and among subjects without elevated tau (tau−). 69However, in the neurodegeneration negative group, flortaucipir deposition was limited to the medial and lateral temporal regions. 76One study showed that, in comparison with younger CU subjects, older Aβ− CU subjects showed increased flortaucipir signal restricted to the medial temporal lobe, but no difference in other neocortical regions (e.g.suggestive of PART).There were no differences in flortaucipir uptake among older impaired versus unimpaired Aβ− subjects. 551][82] Further, there are conflicting reports on the distribution and density of flortaucipir binding with different CSF and PET profiles. 83,84n a head-to-head cross-over study comparing the clinical utility of both flortaucipir and amyloid-PET evaluations, 85 both had a significant and similar impact on diagnosis and diagnostic confidence, and adding one to the other further improved diagnostic confidence.This agreed with findings that combining amyloid and tau PET may help with differential diagnosis and prognosis. 86 discussion on the association of flortaucipir and nonamyloid PET tracers is provided in Supplementary material.

Flortaucipir and CSF
CSF and plasma markers (discussed below) provide estimates of soluble circulating protein levels, such as amyloid and tau proteins. 87For amyloid, common markers include Aβ42 and the Aβ42/Aβ40 ratio. 2,88,89The most common and currently validated tau markers are total tau protein (t-tau) and tau phosphorylated at threonine 181 (P-tau181), threonine 217 (P-tau217) or threonine 231 (P-tau231). 2,90CSF t-tau assays measure the amount of tau protein in CSF, regardless of phosphorylation, and are generally considered the markers of neurodegeneration.Phosphorylated tau is a measure of dysregulated tau metabolism and processing and is representative of disease state (intensity of the disease process).It is hypothesized that abnormal phosphorylated tau contributes to NFT accumulation. 7lzheimer's disease/MCI individuals have abnormally low levels of CSF Aβ42, lower Aβ42/Aβ40 CSF ratios and higher levels of CSF P-tau, t-tau and correspondingly high flortaucipir PET signal compared with CU. 3,54,[91][92][93][94][95] Several groups have shown that flortaucipir global SUVr significantly correlates with CSF t-tau (r = 0.37-0.80),P-tau181 (r = 0.29-0.75),P-tau217 (r = 0.78) and P-tau181/Aβ42 (r = 0.73); however mixed findings are reported for CSF Aβ42 (r = −0.49[98][99][100][101][102] Studies indicated that relative to flortaucipir PET, CSF P-tau181 had concordance of 71-87% 25,[103][104][105][106] and CSF P-tau217 had concordance of 74-80%. 103,106In head-to-head studies, both correlation 78 and concordance 106 of flortaucipir are reported to be higher with P-tau217 compared with P-tau181.P-tau181/Aβ40 ratio has improved performance over P-tau181 alone. 107Evaluations of discordance suggest that tau CSF+/PET− is more common than the reverse, 104,108 and discordance is more prevalent in earlier disease stages, 105 supporting the sequential cascade hypothesis that tau phosphorylation precedes NFT deposition. 109,110Flortaucipir, CSF P-tau181 and CSF t-tau all had high accuracy in identifying dementia due to Alzheimer's disease or MCI due to Alzheimer's disease (area under the curve [AUC] 0.84-1.00). 24,25,100Flortaucipir is, however, reportedly a better predictor of neurodegeneration, cognitive performance and cognitive decline in comparison with CSF t-tau and P-tau181 or 217 measures. 25,100,111,112This aligns with tau-PET being an estimate of the overall burden of aggregated tau pathology and being downstream of peripheral phosphorylated tau changes in the sequential cascade of biomarkers. 10,113SF t-tau and P-tau181 are reported to correlate significantly with flortaucipir PET signal in entorhinal, limbic, parahippocampal, inferior temporal, temporoparietal, medial prefrontal and medial temporal regions (partial r = 0.53-0.73). 49,56,96,100,102,103,111,114Lower CSF Aβ42 predicted higher flortaucipir binding in temporal lobe but not in limbic regions, 100 perhaps consistent with the hypothesis that tau accumulation begins in the mesial temporal lobe but the development of tau beyond the mesial (limbic) temporal lobe is associated with, and may be dependent on, amyloid accumulation. 55,62,63The variations in correlation and concordance between flortaucipir and CSF measures are at least partially reflective of different populations, assays, thresholds and imaging methods.For example, the relationships of CSF t-tau or P-tau181 are specifically stated to differ based on the stage of disease and region of flortaucipir uptake. 10,114The distribution and density of flortaucipir signal is also reported to be different between those who are positive for Aβ as assessed by either PET or CSF. 84SF markers other than Aβ and tau have also been compared with flortaucipir.These CSF markers include glucose, which was associated with lower flortaucipir binding in Braak Stage I/II in women and APOE ɛ4 carriers, 115 apoB levels, which correlated with flortaucipir binding in the entorhinal, parahippocampal and fusiform regions, 116 tumor necrosis factor alpha (TNF-α) and interlleukin 8 (IL-8), which were negatively correlated with flortaucipir in the medial temporal lobe, 117 and neurofilament light chain (NfL) which was correlated with higher flortaucipir binding. 118Synaptosomal-associated protein, 25 kDa, growth-associated protein 43 (GAP-43) and neurogranin were not shown to correlate with flortaucipir. 118

Flortaucipir and blood biomarkers
Like CSF, the primary blood markers of interest in Alzheimer's disease pathology are amyloid (Aβ42, Aβ42/Aβ40 ratio) and tau (P-tau, t-tau).There is an emerging literature showing relationships between these plasma markers and tau-PET imaging.Plasma markers for Alzheimer's disease pathology, however, are relatively new, and their performance is likely to be impacted by the given marker (e.g.Aβ42/Aβ40, P-tau181, P-tau217, P-tau231), as well as by which assay/platform the analyses are conducted on (e.g.mass spectrometry versus immunoassay) 119 ; however, overall the findings support the hypothesis that blood (and CSF) markers become abnormal prior to PET biomarkers.Lower plasma Aβ42/Aβ40 ratios are associated with increased flortaucipir binding in Alzheimer's disease-related regions, 120 and plasma Aβ42/ Aβ40, P-tau181 and P-tau217, but not NfL, had robust discrimination performance reported for identifying elevated flortaucipir binding. 121,122Plasma P-tau, t-tau, P-tau/Aβ42 and t-tau/Aβ1-42 ratios had significant correlations with flortaucipir binding in both cross-sectional and longitudinal analyses. 99,101,103,121Additionally, plasma P-tau181, as well as P-tau231, correlates with flortaucipir SUVr in medial, temporal, parietal and inferior temporal regions, and plasma P-tau231 also correlates with flortaucipir SUVr in the lateral parietal and occipital cortices. 123It was also reported that plasma P-tau181 was correlated with cortical flortaucipir SUVr as well as frontal, precuneus, temporoparietal and posterior cingulate SUVrs. 121he relationship between plasma P-tau181 and flortaucipir was stronger with increasing clinical disease severity. 101ne study 109 reports that plasma P-tau181 becomes abnormal simultaneously with CSF P-tau181, but before flortaucipir, and this finding was further supported by a second study. 124Similar to the CSF findings, reported above, this suggests that plasma P-tau markers become abnormal prior to flortaucipir signal and aligns with findings that plasma P-tau correlates with both amyloid-PET and tau-PET, 121 further supporting the sequential cascade of biomarker abnormality hypothesized in Alzheimer's disease progression. 113dditional studies looking at exploratory markers found that plasma N-terminal tau (NT1) was associated with flortaucipir binding. 125In contrast, evaluations of flortaucipir and serum uric acid, 126 serum zinc, 127 haemoglobin 128 or homocysteine 129 found no associations.
Blood biomarkers for Alzheimer's identification represent a fast-moving landscape, and much has changed since the cut-off date for this literature review.It should be noted that new horizon biomarkers, such as MTBR species, have been proposed that may have a higher concordance with tau-PET. 130

Flortaucipir and cognition
Several cross-sectional studies report that increasing global flortaucipir is associated with worse performance on global cognition, 11,16,19,23,46,55,131 more severe functional impairment 21,131,132 as well as more severe memory impairment. 16,23,55,131,133Similar observations have been described in CU subjects, where higher levels of flortaucipir were associated with worse performance on memory 52,134,135 and digital cognition tests. 136dditionally, a greater extent of binding was associated cross-sectionally with worse performance on global cognition. 11,59,95Others have reported on associations between regional flortaucipir binding and specific cognitive domains or tasks.Elevated flortaucipir binding in the medial temporal lobe, especially the entorhinal cortex, is associated with poorer performance on measures of global cognition [137][138][139] and memory. 10,14,17,46,48,61,138,140,141Visuospatial impairment has been observed to associate with elevated flortaucipir in the occipital lobe and right temporal regions, 14 whereas language impairment has been associated with elevated flortaucipir in parts of the language network (including lateral parietal and inferior frontal regions), as well as with occipital and temporal areas, 14,140 and executive/attentional deficits were found to be associated with elevated flortaucipir in lateral temporal, parietal and frontal regions. 1428][149][150] In CU populations, increased temporal flortaucipir binding is associated with greater memory decline, 147,151,152 but not with change in executive function. 153Of note, such association between elevated flortaucipir binding and longitudinal memory decline in CU participants was only observed within amyloid-PET positive subjects. 153Similarly, higher flortaucipir binding was associated with decreased learning (practice effects) 154,155 and increased reporting of subjective cognitive deficit in CU subjects. 156Even within participants with subjective memory complaints, increased medial temporal lobe flortaucipir correlated with greater subjective cognitive decline as well as Aβ burden. 157,158lortaucipir, amyloid-PET, CSF biomarkers and atrophy are all associated with cognitive performance and decline; however, flortaucipir has a stronger relationship with cognition than the other biomarkers, 58,59,94,159 with one contrasting report where amyloid-PET was more strongly associated with the Montreal Cognitive Assessment in 20 MCI participants. 491][162] One report found significant negative correlations between flortaucipir uptake in early tau accumulation regions (e.g.medial temporal areas) and cognition, in patients without dementia, across memory, executive and language domains.These associations were particularly significant in A+/T+ subjects who were both amyloid and tau-PET positive, but remained significant, although to a lower magnitude, even when restricting the analyses to amyloid-PET-negative, tau-PETpositive subjects. 614][165][166][167][168][169][170][171][172][173][174] Braak Stage I/II ROI flortaucipir SUVr has an accuracy of 78% in identifying participants with affective symptoms. 167hese studies are all consistent with the neuropathological literature that the spatial location and density of tau may indicate the characteristics of the cognitive deficits, as well as the degree of neurodegeneration and synaptic dysfunction.

Flortaucipir and structural MRI
6][177][178][179][180][181][182][183][184][185] These studies suggest a relationship between flortaucipir and neurodegeneration and also that each contribute independently to measures of cognitive impairment. 146Specifically, flortaucipir is more sensitive than volumetric measures at detecting early cognitive changes in pre-clinical Alzheimer's disease, but both measures correlate strongly with cognition at the clinical stages of Alzheimer's disease. 58lthough negative associations between flortaucipir binding and cortical thickness are reported, 49,186,187 the relationship with atrophy is more complex with regional differences reported. 49,167,188,189However, baseline flortaucipir measures were shown to predict longitudinal changes in atrophy, 69,183,184,188,190 with the distribution of flortaucipir binding being predictive of the topology of future atrophy. 191,192orrelations between flortaucipir and glucose hypometabolism are generally stronger than between atrophy and hypometabolism. 181,184,193Further, correlations among flortaucipir uptake, hypometabolism and cortical thickness are strongest in participants with greater cortical tau burden. 181,184,193inally, CU individuals with a high degree of enlarged centrum semiovale perivascular spaces had an odds ratio of 2.24 for elevated flortaucipir signal in comparison with those without a high degree of enlarged centrum semiovale perivascular spaces. 194

Flortaucipir and other imaging modalities
Functional connectivity MRI or diffusion tensor imaging studies 176,181,[195][196][197][198][199][200][201][202][203][204][205][206][207] suggest that more densely interconnected brain regions have higher flortaucipir binding in Alzheimer's disease dementia and are associated with reduced functional connectivity.Entorhinal flortaucipir binding is correlated with hippocampal activation during visual item memory encoding. 208Higher functional connectivity was associated with higher covariance in flortaucipir SUVr, regardless of the presence of cognitive impairment (MCI or dementia), amyloid deposition or small vessel disease (SVD). 200Of note, this study demonstrated that brain regional functional connectivity patterns were predictive of regional tau-PET signal patterns, potentially supporting the notion that functional connectivity facilitates the seeding and propagation of tau. 200Although in one study of individuals with elevated Aβ, connectivity was lower in those with elevated flortaucipir binding. 201Another study 209 reported reductions in the hippocampal connectome correlated with flortaucipir binding in the hippocampus in both MCI and Alzheimer's disease dementia participants, whereas another study 210 reported that flortaucipir binding was not associated with functional MRI response amplitude.
It was found that a dual-phase relationship exists in amyloid-positive individuals with increased connectivity with low neocortical tau and decreased connectivity with elevated flortaucipir levels, suggesting hyperconnectivity followed by hypoconnectivity phases during pre-clinical Alzheimer's disease. 201Flortaucipir binding did not impact the association between default mode network and cortical thinning in participants with abnormal amyloid-PET. 211he relationship between Alzheimer's disease pathology and measures of vascular comorbidity is still not well understood, although a number of MRI studies have started to investigate such relationships.One study 204 observed that flortaucipir, but not amyloid-PET, was associated with white matter integrity loss.Another study 212 reports that regions with white matter hyperintensity also exhibited higher flortaucipir binding.Cerebral microhaemorrhages were associated with global flortaucipir binding, and more extensive microhaemorrhages predicted future parietal flortaucipir binding. 213Increased flortaucipir binding was associated with lower cerebral blood flow in the entorhinal cortex measured through arterial spin-labelled MRI, 214 as well as in the lateral temporal and parietal regions. 142Increased flortaucipir binding was also associated with lower global blood flow measured by pseudocontinuous arterial spin-labelled MRI. 215Additionally, quantitative susceptibility mapping was significantly associated with flortaucipir uptake in the basal ganglia, 216 and increased tissue sodium levels, measured through 7T sodium MRI, were associated with higher flortaucipir binding. 217verall, these studies seem to support the notion that vascular changes are associated with tau pathology, potentially at difference with amyloid pathology, as also demonstrated by autopsy data. 218Future studies systematically investigating the relationship between cerebrovascular disease and Alzheimer's disease pathology 219 are warranted to potentially guide the design of future clinical trials and/or interventions.
Few studies have investigated the relationship between flortaucipir-PET and magnetoencephalography signals in Alzheimer's disease patients. 220,221These studies demonstrated how flortaucipir-PET signal elevation co-localizes with and modulates different alterations in neural synchrony, 220 as well as how flortaucipir-PET signal elevation associates with abnormal neural oscillations. 221

Flortaucipir with clinical treatment trials
Flortaucipir has been used as both an inclusion criterion and secondary end-point in clinical treatment trials.Flortaucipir was used to identify a low-medium tau population for inclusion in the anti-amyloid therapy donanemab Phase 2 trial, which returned positive results. 222Further, it was used to stratify patients in the donanemab Phase 3 trial, which also returned positive results. 223Data from the Phase 3 donanemab trial suggested that patients with a lower disease burden as measured by tau-PET received a greater benefit from the treatment.Donanemab is an antibody directed at the pyroglutamate modification of the third amino acid of the amyloid beta epitope that is present only in brain amyloid plaques.Treatment with donanemab was shown to reduce cerebral amyloid plaques and slow the rate of disease progression.
3][224] No impact on flortaucipir binding was reported for treatment with LY3202626 (a β-secretase 1 inhibitor), 225 rasagiline 226 or lanabecestat. 224,227Mixed findings were reported for cholinesterase inhibitors with one study showing slower longitudinal increases of flortaucipir binding with treatment 228 and a second suggesting that donepezil treatment was reported to be associated with higher flortaucipir binding in MCI patients. 229ncreased change from baseline in flortaucipir binding was found with aducanumab treatment in a single case study. 230hese studies highlight the ongoing use and importance of flortaucipir to monitor and elucidate the consequences of putative therapeutic treatments for Alzheimer's disease 229 ; further, it is anticipated that flortaucipir may be even more useful as an outcome marker for tau targeting therapies.

Flortaucipir and other measures
APOE ɛ4 carriers generally have greater flortaucipir binding and/or accumulation in comparison with non-carriers, 14,[231][232][233][234] and this is reported to follow a medial temporal lobe predominant pattern. 235In addition, APOE ɛ4 carriers with elevated Aβ had increased flortaucipir binding; with homozygotes, additionally having a more widespread pattern of binding. 50The relationship between APOE ɛ4 carriage and flortaucipir is reported to be both independent 74,231 and dependent on interactions with sex, [236][237][238][239] Aβ 50,232 and/or age 190 in conflicting reports.Women APOE ɛ4 carriers were reported to have higher flortaucipir binding in comparison with women non-carriers 236 and men, 236,239 and men homozygotes were reported to have higher flortaucipir binding than men heterozygotes. 239A three-way interaction among age, Aβ, and APOE ɛ4 carriage was also reported 240 where younger (66-70-year-old) APOE ɛ4 carriers with elevated Aβ had significantly higher flortaucipir binding than non-carriers, but that older (74-78 year old) APOE ɛ4 carriers with elevated Aβ did not have elevated binding in comparison with non-carriers.
There is a large body of evidence in support of a complex relationship between age and flortaucipir PET signal.This complexity may be due to the entanglement of the role of tau in the natural history of Alzheimer's disease pathogenesis and in the normal ageing process.2][243][244][245][246][247][248] Our hypothesis is that patients with a younger age of onset may present with a higher burden of NFT pathology in comparison with those with a later disease onset.Contrastingly, in CU individuals, increasing age has been predominantly observed to positively correlate with focal increases in medial temporal flortaucipir PET signal, with some evidence for focal neocortical associations in amyloid-positive CU subjects. 12,14,23,102,152,249Flortaucipir binding is greater in women compared with men, 72,151,233,[250][251][252][253][254] although flortaucipir binding also has a strong relationship with the interaction between sex and APOE ɛ4 genotype. 255No differences in flortaucipir by race (African American and white) were reported. 256In patients of equal cognitive impairment, education was correlated with flortaucipir binding, [257][258][259] suggesting that with higher levels of education more tau pathology is necessary to induce cognitive deficit, consistent with the cognitive reserve hypothesis. 259,260here have been studies examining associations between flortaucipir binding with obstructive sleep apnoea, 261,262 higher cardiovascular risk metrics, 263 lower self-reported sleep, 264 lower self-reported physical activity, 265 birth season, 266 visual contrast sensitivity, 267 more fragmented networks measured with electro-magnetoencephalography 268 and head injury especially in those who also reported a loss of consciousness. 269It was also reported to have an association with lower olfactory scores in two reports 270,271 but no association with olfactory identification in a second report. 272Flortaucipir binding had no reported association with air particulate matter that had an aerodynamic diameter <10 μm. 273n addition to APOE ɛ4, bridging integrator 1 rs744373 risk allele carriers showed a higher 274,275 and faster accumulation of flortaucipir binding, which was exacerbated in the presence of elevated Aβ, 276 and higher flortaucipir binding was reported to be associated with the PPP2R2B (rs76752255) and IGFBP3 (rs117402302) single-nucleotide polymorphisms, 277 'axon-related' and lipid metabolism genetic profiles as well as the microtubule-associated protein tau (MAPT) gene. 278Flortaucipir binding had no reported association with a total cholesterol polygenic risk score. 279In twin pairs, the intensity and extent of flortaucipir binding was reported to be similar suggesting a large role for genetics in the accumulation and spreading of tau. 280o fully understand the impact of some of these measures, particularly age, APOE ɛ4 and their interaction with Aβ as well as sex, race, socio-economic factors and comorbidities, larger longitudinal studies over longer time periods are required.

Flortaucipir in atypical Alzheimer's disease
In addition to typical late-onset amnestic Alzheimer's disease dementia, flortaucipir has also been used to investigate binding in atypical variants of Alzheimer's disease, including logopenic variant of primary progressive aphasia (lvPPA), posterior cortical atrophy (PCA) and the behavioural Alzheimer's disease (bvAD) phenotype. 112,281,282In general, these atypical phenotypes tend to occur more frequently in younger Alzheimer's disease patients, accounting for about a third of patients with early-onset compared with ∼6% of patients with late-onset Alzheimer's disease 283 The topography of flortaucipir binding patterns varies by phenotypes, with a topography consistent within the clinical presentation.289][290][291] As for bvAD, flortaucipir binding has been shown to be more heterogeneous but still consistent with the typical Alzheimer's disease temporoparietal pattern. 14,286,292,293In addition, flortaucipir binding patterns in atrophy-defined subtypes of Alzheimer's disease (e.g.'typical', 'limbic-predominant', 'hippocampal-sparing', and 'mild atrophy') also differ, with the greatest neocortical binding observed in hippocampal sparing and typical patients, while limbic predominant patients showed particularly high entorhinal binding. 176,294,29543][244][245]248 Across atypical variants, as with typical late-onset amnestic Alzheimer's disease, elevated flortaucipir binding, global, regional or voxelwise, is associated with cognitive impairment/decline, 16,140,190,242,284,287 prospective atrophy, 191 longitudinal florbetapir accumulation, 181,285,296 neurodegeneration measured with structural MRI, 27,181,184,190,284,287 synaptic dysfunction measured with FDG-PET, 14,181,184,190,286,290,293 fractional anisotropy 181 and aberrant functional connectivity. 297lortaucipir binding patterns in autosomal dominant Alzheimer's disease (ADAD) have also been investigated. 298- 305These studies overall showed flortaucipir binding elevation in signature Alzheimer's disease temporo-parietal regions, with a topography consistent with available evidence in typical, late-onset Alzheimer's disease.However, ADAD cases tended to show a greater magnitude of flortaucipir binding compared with sporadic typical late-onset Alzheimer's disease. 301,305gain, medial and inferior lateral temporal flortaucipir binding was associated with cognitive impairment. 300,302,3039][300] Of note, in PSEN1 E280A ADAD, early entorhinal flortaucipir binding elevation was associated with striatal amyloid-PET binding. 304In Down syndrome (DS), the trisomy 21 causes a duplication of the amyloid precursor protein gene, leading to over-production of amyloid beta. 306lortaucipir studies in DS have reported elevated binding spatially consistent with observed findings in typical Alzheimer's disease [307][308][309][310][311] and that neocortical flortaucipir binding is associated with cognitive decline, 309,311 even when limited to subthreshold amyloid accumulation. 308

Flortaucipir in non-Alzheimer's disease neurodegenerative conditions
While flortaucipir is not intended for diagnostic use in neurodegenerative and neurological conditions other than Alzheimer's disease dementia, it has been used for research purposes. 308Overall, flortaucipir binding elevation has been described in regions in which tau pathology or neurodegeneration would be expected, e.g.basal ganglia in PSP or anterior temporal lobes in semantic variant primary progressive aphasia (svPPA; see below).Several caveats, however, should be considered for a comprehensive interpretation of such findings.First, post-mortem autopsy and tissue studies have repeatedly demonstrated weak/null flortaucipir binding to non-Alzheimer's disease pathology aggregates, suggesting that in vivo findings may be capturing processes that colocalize or associate with the given underlying pathology.Elevated flortaucipir binding in non-Alzheimer's disease conditions, with some exceptions such as in PART, should thus not be interpreted as evidence for tau pathology.Second, such elevations are usually observed at the group-level, with consistently lower effect sizes compared with findings in Alzheimer's disease, not being informative or specific at the single-subject level.Third, such elevations, when present, are observed in regions not relevant for Alzheimer's disease.Similarly, regional flortaucipir PET signal has been shown to accurately discriminate (reaching an AUC of 0.99) between high ADNC and frontotemporal lobar degeneration (FTLD) at autopsy. 41Fourth, flortaucipir binding elevation in non-Alzheimer's disease neurodegenerative conditions in most cases does not track with disease severity, unlike the robust associations observed with clinical and cognitive decline in Alzheimer's disease.

Suspected or definite non-Azheimer's disease tauopathies
One of the most intense areas of flortaucipir research has been non-Alzheimer's disease tauopathies.Flortaucipir is useful and reliable in detecting tau NFTs in Alzheimer's disease, thus is thought to be specific for binding to tangles with a paired helical filament (PHF) morphology and a balanced combination of 3R/4R tau isoforms.In non-Alzheimer's disease tauopathies, tau aggregations may present in different conformations (e.g.straight filaments) and with different isoforms ratios [e.g.3R-dominant in Pick's disease or 4R-dominant in progressive supranuclear palsy (PSP) and corticobasal degeneration (CBD), as well as with mixed 3R/4R tauopathy such as in PART or in chronic traumatic encephalopathy (CTE)]. 312,313ithin these conditions, flortaucipir PET-to-autopsy studies have provided contrasting evidence, and some studies noting a correlation between flortaucipir in vivo and post-mortem tau neuropathology in patients with CBD, 314 partially in argyrophilic Grain disease, 315 but not in PSP, 316 and with contrasting evidence in CTE. 317Of note, in vitro studies have more consistently demonstrated weak/null flortaucipir binding to various non-Alzheimer's disease pathologies.

3R-or 4R dominant non-Alzheimer's disease tauopathies
Overall, in vivo topography of flortaucipir binding elevation in PSP has been observed in basal ganglia and subcortical structures across clinical variants, [318][319][320][321][322][323][324][325][326][327][328][329][330] correlating with volume loss, cortical thinning and 331 gait abnormalities 332 and associating with neuroinflammation assessed by 11 C-PK11195-PET. 333,334Similar findings have been observed in cortico-basal syndrome (CBS, suspect CBD), with elevation mostly observed in cortico-spinal and motor cortical/frontoparietal areas, as well as in basal ganglia and subcortical structures, 325,330,[335][336][337][338][339] although to a lower magnitude compared with Alzheimer's disease, 339,340 or not significantly different from CN. 329 Recently, a study conducted on a mixed cohort of suspected 4R-dominant tauopathies (i.e.PSP and CBS) described a significant negative correlation between flortaucipir binding and synaptic density with 11 C-UCB-J-PET. 330Similar flortaucipir binding patterns involving subcortical and frontal/temporal regions, including white matter, have been observed in behavioral variant frontotemporal dementia (BvFTD) 315,341,342 and in non-fluent variant primary progressive aphasia (nfPPA), 288,343 as well as in nfPPA/apraxia of speech patients with definite Pick's disease. 203,325,344In keeping with cross-sectional evidence, longitudinal studies showed higher rates of change for flortaucipir in PSP, although not correlating with functional decline 345 and in MAPT Q351R 346 with an expected topography.Such flortaucipir changes were also observed in patients with apraxia of speech, either primary progressive or in the context of nfPPA, 343,347 although no difference in binding elevation for PSP versus controls is reported. 348

3R/4r non-Alzheimer's disease tauopathies
One study showed evidence of elevated flortaucipir binding in medial temporal lobe structures in some suspected non-Alzheimer's disease pathophysiology cases (SNAP), likely due to PART.A follow-up flortaucipir scan, however, showed that such binding did not increase over time. 349Another study found significant and similar associations between medial temporal lobe flortaucipir binding and atrophy, regional thickness and cognition in both amyloid-positive and amyloid-negative (likely PART) groups, suggesting that flortaucipir may be sensitive to tau pathology in PART. 350However, other reports suggest that flortaucipir in the medial temporal lobe does not reliably detect NFT in patients with definite PART as secondary pathology at autopsy. 325Overall, the available data do not support the use of flortaucipir for the detection of PART, at the same time suggest that the presence of PART is unlikely to have considerable impact on a flortaucipir signal pattern.
Flortaucipir binding has been studied in patients with head trauma, traumatic brain injury (TBI) and repetitive TBI, as well as CTE.In participants with war-related TBI and/or PTSD, mild and diffuse flortaucipir was reported across frontal, 351,352 parietal 352 and temporal 351,352 cortices as well as in the cerebellum. 351Additionally, significant correlations were reported between flortaucipir binding and neuropsychological impairment in subjects with PTSD or PTSD + TBI. 352n war veterans as well as in individuals with repetitive TBI and suspected or at-risk CTE, patchy, dot-like patterns of flortaucipir binding were reported across the cortex or at the grey matter/white matter junction, a typical observation in CTE pathology post-mortem, [353][354][355] as well as in superior frontal, medial temporal and left parietal areas. 355,356Participants with TBI and suspected or at-risk CTE had group-level elevated flortaucipir binding in the striatum and substantia nigra 357 ; however, overall they exhibited a high degree of heterogeneity at the individual level. 353,354,357Additionally, war veterans with elevated flortaucipir binding also presented with elevated plasma NfL levels, an axonal injury marker. 355n a study including American football players, flortaucipir binding in cortical regions was found to associate with years of play at the group-level. 356Furthermore, flortaucipir binding in similar contact sport populations has been reported to be higher in APOE ɛ4 but not in MAPT haplotype H1H1 carriers. 358In some subjects with elevated flortaucipir binding and TBI, including sports or accident-related TBI, flortaucipir binding was found to associate with functional connectivity and white matter abnormalities. 359In amyloid-positive cases with a history of TBI, flortaucipir binding consistently follows Alzheimer's disease -like patterns of topography and magnitude. 354,360Adding to the complexity of flortaucipir findings in CTE, a recent case-study described a lack of correlation between in vivo flortaucipir binding and postmortem tau pathology in a case with confirmed CTE. 317espite several studies in CTE describing flortaucipir binding elevation at the group-level, effect sizes are usually weak, with non-specific topographies, inconsistent clinical/cognitive correlations and no evidence of correlation with post-mortem tau pathology.In keeping with these limitations, flortaucipir is specifically not indicated for use in CTE in the USA. 308

Microtubule-associated protein tau mutation carriers
Several studies have investigated flortaucipir binding in MAPT mutation carriers.2][363][364] Flortaucipir binding showed an expected topography involving frontotemporal and subcortical areas in MAPT mutations known to lead to a dominant 4R tauopathy, such as MAPT 10 + 16, MAPT N279K, MAPT S305N, MAPT P301L and MAPT S305I, but with a much lower magnitude to that seen in 3R:4R tauopathies. 315,363,365,366One study including a symptomatic carrier of a MAPT G272V mutation showed extensive cortical flortaucipir binding, with autopsy data showing a combination of 3R tauopathy and encephalitis. 364As for pre-symptomatic carriers, there is some indication that focal flortaucipir binding elevation can be observed in medial temporal lobe in MAPT R406W, in frontal areas in MAPT L315R, encompassing insula and frontoparietal regions in MAPT P301. 364In the same study, however, no binding elevation was observed for other pre-symptomatic MAPT P301L and MAPT S320F carriers. 364inally, the MAPT rs242557 variant has been associated with increased flortaucipir binding in the hippocampus. 367

Suspected or definite TDP-43 proteinopathies
Several independent studies have demonstrated elevated flortaucipir binding in svPPA, typically associated with transactive response DNA binding protein 43 (TDP-43) Type-C pathology, 368 suggesting flortaucipir may bind to targets or processes associated with pathologies other than Alzheimer's disease tauopathy.These studies have consistently demonstrated asymmetric elevated binding in anterior temporal lobes, often in temporal poles, consistent with the known atrophy and synaptic dysfunction pattern in svPPA, 288,315,[369][370][371][372][373] and potentially associated with neuroinflammation at translocator protein PET. 342In contrast to volume loss, flortaucipir temporal binding in svPPA remained stable over time. 374TDP-43 pathology is also observed in patients with a C9ORF72 or progranulin (GRN) gene mutation.Several studies have investigated flortaucipir binding in C9ORF72, overall providing mixed evidence of flortaucipir binding restricted to punctate foci, 315 frontotemporal, 375 entorhinal cortex, 376 mild inferior 372 or heterogeneous frontotemporal, parietal, occipital and cerebellar regions. 366Similar heterogenous findings have been observed in GRN mutation carriers, with reports of elevated frontotemporal and parietal binding, 315 in contrast to null or very mild inferior temporal/cerebellar binding. 366

Suspected or definite synucleinopathies
Several studies have investigated flortaucipir binding in conditions associated with underlying α-synuclein pathology, including Parkinson's disease, dementia with Lewy bodies and multiple system atrophy (MSA).386][387][388][389][390][391] Such flortaucipir binding in dementia with Lewy bodies has been observed to associate with Aβ positivity, with relatively higher flortaucipir binding in dementia with Lewy bodies cases with greater amyloid burden, 380,382,386,390 cognitive impairment 382,385 or functional connectivity abnormalities, 392 but not with synaptic density. 388au positivity, as measured by flortaucipir or CSF P-tau181, was also found to be associated with lower frequency of Parkinsonism and rapid eye movement (REM) sleep behaviour disorder in dementia with Lewy bodies. 391Furthermore, flortaucipir binding in dementia with Lewy bodies was found to generally associate with both CSF t-tau and neuropathological tau (both total tau and regional histology NFTs) at autopsy. 390ne longitudinal study has additionally demonstrated that flortaucipir binding in dementia with Lewy bodies increases over time, especially in posterior areas, correlating with concomitant neurodegeneration and cognitive worsening. 393imited evidence is available for flortaucipir binding in MSA, with one study describing putaminal binding elevation associated with concomitant dopaminergic degeneration. 394

Vascular cognitive impairment and dementia
Few studies have investigated flortaucipir binding patterns in subcortical vascular cognitive impairment (SVCI), SVD or vascular dementia.One study demonstrated that subjects with SVCI, even when amyloid-negative, showed increased flortaucipir binding in inferior temporal areas. 395Within subjects with SVCI, higher inferior temporal flortaucipir binding was also found to associate with higher cerebral SVD scores, with medial parietal and temporal flortaucipir binding overall associating with cognitive function.Of note, several of these associations between flortaucipir and cognition were also significant when focusing on the amyloid-negative SVCI patients.Another study reported that prevalence of tau positivity, as measured by flortaucipir with quantitation in Braak III/IV regions, was observed to be lower in Aβ+ SVCI subjects when compared with Aβ+ Alzheimer's disease subjects (25.9% versus 70%). 396urthermore, SVCI subjects who were both Aβ+ and flortauci-pir+ showed steeper cognitive decline, greater global cortical thinning, but not decreased hippocampal volume, compared with Aβ+ and flortaucipir negative subjects. 396An independent study found a similar prevalence of flortaucipir positivity in a group of patients with SVD and cerebral amyloid angiopathy (CAA), with 5/15 (33.3%) participants showing elevated flortaucipir binding at a visual read. 397Another study showed that subjects with probable CAA and an amnestic neuropsychological profile tended to show higher flortaucipir binding in Alzheimer's disease signature temporoparietal regions compared with CAA subjects with nonamnestic profiles, although no differences in amyloid-PET were observed. 398lortaucipir literature associated with other neurodegenerative and neurological conditions, including prion diseases, is included in Supplementary material.

In vitro
Several studies have compared flortaucipir binding with binding of other compounds in tissue specimens.Autoradiography studies showed that in Alzheimer's disease, flortaucipir mostly binds to lesions of tau aggregated in PHFs, including intra-neuronal and extra-neuronal tangles and dystrophic neurites, 400,401 replicating Braak staging. 45he maturity of tau pathology may also have an impact on the strength of flortaucipir binding. 406Flortaucipir did not significantly bind to tau aggregated in 4R-dominant straight filaments, 406,407 beta-amyloid, 401 synuclein 406,408 or TDP-43. 399,401,406Some reports have highlighted weak flortaucipir binding to 3R-dominant Pick's disease 399,400,406 and FTD with Parkinsonism-17 400 tau lesions.Very similar properties were observed in head-to-head studies comparing flortaucipir with 18 F-florquinitau (MK6240), 409 18 F-THK5351, 399,410 18 F-THK5117 410 and 11 C-PBB3, 402,410 with perhaps some evidence for relatively stronger binding of PBB3 in both 4R-dominant and 3R-dominant tauopathies. 402

In vivo
A few previous studies have compared flortaucipir binding patterns with the patterns observed for other novel tau-PET radioligands in vivo in the same patients, including THK5351, 411 18 F-RO948 412 and florquinitau (MK6240). 413ompared with THK5351, flortaucipir showed correlated, but higher, binding in Alzheimer's disease patients and lower binding in FTD.The higher binding showed by THK5351 in non-Alzheimer's disease was thought to reflect other neurodegenerative processes, given that THK5351 strongly binds to monoaminoxidase (MAO) B. 411,414 Notably, flortaucipir showed less off-target binding in brainstem, white matter and striatum when compared with THK5351.Compared with RO948 in a cohort mainly of Alzheimer's disease dementia patients, flortaucipir showed lower binding in the entorhinal cortex and higher binding in the striatum, thalamus and choroid plexus, whereas skull/meningeal off-target tracer binding was noticeably higher with RO948. 412Of note, neocortical binding was very highly correlated across the neocortex, peaking in the Braak III/IV stage ROI (Spearman rho = 0.96).Both tracers showed increased off-target binding with older age. 412When compared with florquinitau in a cohort mostly including cognitively normal controls, the SUVrs generally were highly correlated (R 2 ≥ 0.92), except in the Braak II regions, possibly due to higher adjacent off-target binding evident with flortaucipir.Of note, flortaucipir showed higher striatal/choroid plexus off-target binding, whereas florquinitau showed higher meningeal off-target binding. 413Similar relationships between flortaucipir and florquinitau were observed in a study on ADAD. 415lthough not coming from head-to-head studies, supporting evidence for similar properties across these radioligands is provided by a cross-sectional study pooling flortaucipir, RO948 and florquinitau data. 416This study demonstrates that tau-PET binding across several regions, e.g.entorhinal cortex or a temporal composite ROI, provides similar accuracy in differentiating Alzheimer's disease versus non-Alzheimer's disease or cognitively normal participants across tracers. 416lortaucipir was used as a comparison standard to evaluate performances of a new putative radioligand, 18 F-N-methyl-lansoprazole, for which a favourable clinical translation was not exhibited. 417

Flortaucipir technical considerations
Flortaucipir-PET scans can be evaluated either with a qualitative visual read, assessing binding elevation topography and magnitude, or with different types of quantitation, providing precise binding estimates.

Visual read
The FDA-approved method to visually evaluate a flortaucipir scan enables the evaluation of both the presence and extent of elevated binding.This protocol includes reorientation of the image, delineation of a cerebellar ROI and relative windowing of the colour scale to evaluate the presence of binding 1.65 times greater than the cerebellum.When such binding is consistently observed in the posterior lateral temporal/occipital or parietal areas, the image is read as positive, regardless of frontal involvement.This method has been developed to improve the signal-to-background ratio and improve accuracy in detecting flortaucipir signal elevation, demonstrating high sensitivity, specificity and accuracy in predicting Alzheimer's disease pathology at autopsy. 36ased on the localization of the elevated binding, the Alzheimer's disease pattern can be classified as moderate (elevation limited to lateral posterior temporal/occipital) or advanced (elevation involving parietal regions or lateral posterior temporal/occipital extending to frontal regions). 36,52he advanced Alzheimer's disease pattern is associated with a greater risk of cognitive decline over 18 months in patients with MCI or Alzheimer's disease dementia. 52A recently described visual read protocol for flortaucipir includes both a quantitative score based on binding magnitude (from 0 for no binding to 2 for intense binding) in seven different ROIs and a qualitative pattern evaluation. 418The visual scores were associated with flortaucipir cortical SUVrs, clinical diagnosis and amyloid status, with the Alzheimer's disease patterns also demonstrating efficacy for identifying amyloid-positive MCI and Alzheimer's disease dementia patients. 418

Quantitation
Quantitation approaches are currently the standards of choice in research and clinical trials, both for estimation of individual-level and group-level changes in longitudinal data.Flortaucipir quantitation is generally performed with an ROI-based approach, generating a ratio between average binding in a target region and average binding in a reference region (SUVr).Target regions should be the areas indicative of underlying Alzheimer's disease pathology (e.g.areas included in the Braak NFT staging 29 ), and reference regions should be unaffected by the disease process.Common target regions include areas involved earlier in disease stage, (e.g.entorhinal cortex or inferior temporal areas), as well as larger regions involved in later disease phases [e.g. the temporal meta-ROI 289 or the multi-block barycentric discriminant analysis ROI]. 419The most used reference region is the cerebellar grey matter/crus, 420 although sometimes only inferior and composite white matter areas (PERSI-defined reference 421,422 or the centrum semiovale 423 ) were employed.Different implications for cross-sectional or longitudinal studies should be considered for the different reference regions. 424One recent study also evaluated the possibility of a dual reference tissue approach where cerebellar grey matter and subcortical white matter are used to derive different parameters from dynamic flortaucipir scans [relative delivery (R1) and distribution volume ratios (DVRs), respectively]. 425hile the choice of target/reference region is usually largely consistent across groups, there are several sources of variability intrinsically associated with quantitation, which may limit the comparability of findings.These include (but are not limited to): (i) native versus template-space processing; (ii) use of atlas-defined ROIs and the use of different atlases; (iii) MRI-based versus PET-only pipelines 426,427 ; (iv) different structural MRI segmentation software (e.g.Statistical Parametric Mapping, FreeSurfer); (v) utilization of partial volume correction approaches and (vi) use of different statistical approaches (e.g.median versus mean).Specifically for longitudinal data and the ability to measure change over time, a recent study comparing >400 combinations of methodological choices indicated that the use of larger temporal meta-ROIs, longitudinal analytical pipelines, composite WM reference regions, two-class partial volume correction and median statistics were the optimal choices. 428n addition to SUVr values derived from short scans (e.g.20 min), quantitation has also been performed on longer dynamic scans to derive other flortaucipir uptake parameters such as binding potentials (BPs) and DVRs.0][431] Further, several groups have employed voxelwise analyses to investigate flortaucipir binding patterns in Alzheimer's disease and related conditions, both cross-sectionally and longitudinally, and their relationship with cognitive impairment and decline. 16,63,97,145,178,181,235,432More recent approaches under development include whole-brain pattern analysis and machine learning techniques, e.g.Tau IQ 433 or convolutional neural network approaches. 434

Off-target binding
Flortaucipir-PET studies showed significant off-target binding, hypothesized to be associated with: iron deposits, bleeding, calcifications, neuromelanin, leptomeningeal melanin, MAO A/B or their combination.In flortaucipir human scans, off-target binding is visually evident in the striatum, brainstem and choroid plexus and less frequently in the meninges and dorsal cerebellum (see 435,436 for reviews).PET/MRI studies revealed a possible association between off-target binding and iron deposits/accumulation. 437,438Regarding cerebral bleeding, flortaucipir binding was found to associate with microbleeds, 438 chronic infarctions 439 and subacute infarction, 437 with negative findings in patients with unilateral occlusion of pre-cerebral arteries. 440A few studies have reported on possible elevation of flortaucipir binding in meningiomas. 439,441Autoradiography studies have showed saturated flortaucipir binding in substantia nigra in melanoma cells 442 ; strong off-target binding to neuromelanin and melanin-containing cells, some weak binding to areas of haemorrhage 409 ; binding to neuromelanin-containing neurons in substantia nigra, 408 as well as in leptomeningeal melanocytes adjacent to ventricles and midbrain. 408In the same study, binding to choroid plexus was only observed in subjects with leptomeningeal melanocytes within choroidal stroma. 408Another study demonstrated that epithelial cells in the choroid plexus actually include tangle-like structures (corresponding morphologically to Biondi 'ring' tangles), potentially qualifying flortaucipir binding in the choroid plexus as on-target. 443As for MAO, although contrasting evidence has been reported, 409,410,[444][445][446][447][448] recent reports have demonstrated that MAO-A binding is not a significant contributor to the observed PET signal in Alzheimer's disease and that there is only weak binding of flortaucipir to MAO-B in vitro at micromolar concentrations. 448As for MAO-A, despite low nanomolar affinity, flortaucipir binding displays fast dissociation rates, leading to a non-significant contribution to the observed PET signal in areas with significant tau pathology burden. 448Overall, it is suggested that its contribution to off-target binding for flortaucipir is very weak or minimal, 448 especially considering the difference in flortaucipir binding patterns in different conditions such as Alzheimer's disease versus PSP or Parkinson's disease.
Off-target binding, however, might be problematic in specific circumstances.For instance, off-target signal in the basal ganglia may impact the quantitation of basal ganglia signal elevations in non-Alzheimer's disease neurodegenerative conditions such as PSP or CBD, which is likely to already be affected by the lower affinity of flortaucipir for non-Alzheimer's disease tau.Off-target signal in the choroid plexus, on the other hand, might impact quantitation in medial temporal lobe regions aimed at earliest detection of Alzheimer's disease tau accumulation.Specifically for the hippocampus, several studies have attempted to correct for off-target binding to provide more accurate flortaucipir binding estimates. 138,449,450Regardless of localization, the relative impact of off-target binding in imaging studies should also be considered in light of its relationships with age, 451 sex 452 and race. 453

Summary
A thorough evaluation of the literature pertaining to flortaucipir has been presented.A number of autopsy studies evidence that flortaucipir represents an accurate in vivo detection and quantitation of the density and distribution of neocortical NFT accumulation.Flortaucipir binding is elevated in cases of typical Alzheimer's disease and MCI due to Alzheimer's disease as well as in individuals who are amyloid positive.In fact, evidence suggests that an elevated flortaucipir scan has a high positive predictive value for an elevated amyloid-PET scan and could be used to indicate the presence of both NFT and Aβ plaque pathology.
In addition to amyloid PET markers, the literature suggests moderate to strong concordance for flortaucipir with both CSF and blood-based biomarkers of Aβ and tau and particularly with phosphorylated tau compounds.Increased density and extent of flortaucipir binding have been shown to correlate with worse deficits on cognitive tests.Additionally, elevated flortaucipir has been shown to be predictive of future cognitive decline as well as clinical progression from CU to MCI or Alzheimer's disease, suggesting a potential prognostic role for flortaucipir in Alzheimer's disease clinical evaluations.There are also reports that the region of elevated flortaucipir binding provides further prognostic information regarding the domain of cognitive deficit, with several reports identifying elevated flortaucipir binding in the medial temporal lobe associated with memory loss.Reports suggest that flortaucipir has a stronger correlation with neurodegeneration and cognition compared with amyloid-PET, supporting the amyloid cascade hypothesis where elevated global amyloid precedes elevated tau which in turn precedes neurodegeneration.
There are conflicting or sparse reports on the impact of factors such as APOE genotype, sex, race, age, socioeconomic factors and comorbidities on the level of flortaucipir binding observed.None of the literature, however, appears to suggest any impact of these factors on the relationship between flortaucipir PET signal and the distribution and density of tau pathology.Increased numbers as well as more longitudinal studies are required to fully understand how these factors and their interplay affect flortaucipir binding.In cases of atypical Alzheimer's disease, the pattern of flortaucipir signal elevation has overall been reported to be consistent with the expected topography and patterns based on autopsy data.In cases of non-Alzheimer's disease neurodegenerative conditions, the findings are mixed and depend on the tauopathy or condition under study.The use of flortaucipir is considered to be limited in these studies.In general, flortaucipir appears to bind more consistently, as expected, in 3R/4R tauopathies, still to a lower extent compared with evidence available in Alzheimer's disease.However, it should be reiterated that flortaucipir is not indicated for non-Alzheimer's disease conditions.Flortaucipir demonstrates comparable performance in comparison with other tau PET tracers, including the second-generation tracers.All tracers considered had off-target binding, albeit in different regions, that needs to be addressed.
Overall, the findings reported in this comprehensive review of the literature indicate that flortaucipir is an accurate and useful in vivo marker of tau pathology that can improve the diagnostic, and potentially prognostic, landscape for Alzheimer's disease.