Molecular Imaging of Extrapyramidal Movement Disorders with Dementia: The 4R Tauopathies

Abstract

Two pathologically distinct neurodegenerative conditions, progressive supranuclear palsy and corticobasal degeneration, share in common deposits of tau proteins that differ both molecularly and ultrastructurally from the common tau deposits diagnostic of Alzheimer disease. The proteinopathy in these disorders is characterized by fibrillary aggregates of 4R tau proteins. The clinical presentations of progressive supranuclear palsy and of corticobasal degeneration are often confused with more common disorders such as Parkinson disease or subtypes of frontotemporal lobar degeneration. Neither of these 4R tau disorders has effective therapy, and while there are emerging molecular imaging approaches to identify patients earlier in the course of disease, there are as yet no reliably sensitive and specific approaches to diagnoses in life. In this review, aspects of the clinical syndromes, neuropathology and molecular biomarker imaging studies applicable to progressive supranuclear palsy and to corticobasal degeneration will be presented. Future development of more accurate molecular imaging approaches is proposed.

INTRODUCTION

Several extrapyramidal movement disorders are progressive neurodegenerative conditions that may evolve to include cognitive decline and dementia. These conditions include Parkinson disease (PD) as well as atypical parkinsonisms (atypical parkinsonian syndromes - APS) {^1,2}, the latter consisting of multiple system atrophy (MSA), progressive supranuclear palsy (PSP) and corticobasal degeneration (CBD). The most prevalent extrapyramidal disorders with prominent cognitive impairment are the Lewy body dementias (LBD) comprised of dementia with Lewy bodies (DLB) and of PD with dementia (PDD). Combined, the estimated prevalence of the Lewy body dementias is approximately 200/100,000, and may be 3-fold more common in the population aged over 65 years (calculated from data in {³}). The LBD conditions are the subject of an accompanying review in this issue {⁴}. Here, we will consider two additional disorders among the APS with prominent cognitive impairment: PSP and CBD {^5,6}. Estimated prevalence of PSP is approximately 4/100.000, while that of CBD is not presently available (it is considered “rare”) {³}. PDD, MSA, PSP am CBD have common symptomatic onset as disorders of voluntary movement, sometimes without conspicuous cognitive impairment. The movement disturbance at onset is classified as extrapyramidal; lacking in overt weakness or spasticity, but with prominent Parkinsonism, including rigidity and bradykinesia. PSP, CBD and MSA constitute the major subtypes of APS, with lack of objective and sustained response to dopamine replacement therapy (i.e. levodopa or dopamine agonists). Dementia is rare in MSA, and this syndrome will not be considered further in this review.

The clinical diagnoses of PD versus APS remains challenging (accuracy in the range of 75% {^7–9}), especially at symptomatic onset {¹⁰}. With the evolution of time and progression of symptoms and signs, parkinsonian disorders are often diagnosed correctly by movement disorder specialists {¹¹}. Characteristic clinical signs specific to PSP or to CBD often develop, however, even then, definite diagnosis rests on neuropathological examination {^12,13}. Unfortunately, these disorders have no effective symptomatic therapies, and like other neurodegenerations, effective disease-modifying therapies are also unavailable. This latter problem is not easily addressed in design of potential clinical trials, as early interventions are limited by diagnostic uncertainty and misdiagnoses, and later course interventions may be ineffective due to advanced progression of the degenerative pathology. Thus, improved approaches to early diagnostic classification of parkinsonism are necessary. Molecular neuroimaging approaches offer the possibility of accurate early diagnostic classification and may contribute to determination of underlying neuropathological changes and their progression. These aspects will be discussed in the following sections of this review.

CLINICAL SYNDROMES IN PSP AND CBD

Several clinical syndromes are now recognized as related to underlying pathological deposits of the microtubule associated protein tau, including both PSP and CBD {¹⁴}. The former is the most frequent of these relatively uncommon APS conditions. However, the clinical syndromic presentations of these pathological disorders are varied, and they overlap with pathologically distinct alternative diagnoses, limiting the specificity of clinical diagnoses.

PSP

PSP (known also as the Steele-Richardson-Olszewski syndrome) is probably the most frequent APS (see initial description in {¹⁵}). “Typical” clinical features consist of generalized bradykinesia with most prominent involvement and rigidity of axial and proximal limb musculature, especially involving the neck. Loss of postural reflexes, resulting in frequent falls and difficulty with ambulation are frequent diagnostic clues {¹⁶}, usually absent within the first year of initial presentation of PD {^11,17}. Impairment in PSP often involves dysarthria and dysphagia early in the course. While there may be a subjective response to initial levodopa therapy in a minority of PSP patients, this may not be as objectively evident as is the response in PD, and as symptoms progress, lack of benefit becomes more obvious. The most convincing clinical sign of PSP is the presence of impaired eye movements, particularly involving vertical down-gaze, termed Richardson Syndrome (PSP-Richardson). Patients are unable to voluntarily direct gaze from point-to-point, and tracking of visual targets is similarly impaired. Eye movements can be elicited on examination, however, by instructing patient to maintain visual fixation on a target during passive movement of the head by the examiner. Thus, examiner extension of the patient’s neck produces vertical depression of the eyes, most likely related to engagement of the vestibular inputs to brainstem gaze centers. Unfortunately, this diagnostically specific feature may develop late in the progression of symptoms, and may be absent in some patients. Initial clinical presentation of PSP may include patients with prominent cervical dystonia and gaze impairments (Richardson syndrome), or with prominent postural and gait instability, or occasionally with cognitive impairment resembling the behavioral variant of frontotemporal lobar degeneration (bvFTLD; see {¹⁸} for these dementia criteria) {^19–21}. As symptoms progress, most PSP patients eventually develop dementia, although this can sometimes be difficult to appreciate clinically in the face of severe movement limitations affecting also speech production. Patients with the prominent postural instability and gait freezing or with the prominent cognitive disturbances are often not diagnosed initially with PSP {¹²}, leading to a relative under-diagnosis at clinical onset. Diagnosis of definite PSP continues to rely on histopathologic evaluation {²¹}. PSP has significantly more rapid decline and shorter survival than typical of PD {²²}, further emphasizing the need for more accurate clinical characterization of PD versus APS on initial presentation.

CBS

Corticobasal degeneration may also present clinically as an APS. In CBD, the “typical” extrapyramidal movement onset is with unilateral limb dystonic rigidity accompanied by apraxia, clinically referred to as corticobasal syndrome (CBS) {^23,24}. As in other APS, there is no reliable sustained response to dopamine replacement therapy. As symptoms progress, rigidity may generalize and spread to involve the contralateral side of the body, but there is most often relative preservation of side-to-side asymmetry of severity reflecting the features at onset. The diagnosis of CBS is supported by several unusual features that may be individually or collectively expressed. A cerebrocortical sensory loss involving the dystonic limb may be present. Some CBS patients demonstrate a somatosensory stimulation triggering of myoclonus (reflex myoclonus) in the involved limb. Some CBS patients demonstrate the “alien limb” phenomenon, where the involved limb may move to involuntarily interfere with tasks performed with the contralateral limb {²³}. However, the initial presentation of CBS often lacks these suggestive features, and frequently involves signs and symptoms of cognitive impairment: progressive non-fluent aphasia; executive movement disturbance (prominent apraxia); or behavioral variant frontotemporal lobar degeneration syndrome (loss of insight and lack of behavioral regulation) {^25–28}. Some patients with pathologically confirmed CBD may present clinically with an apparent PSP clinical phenotype. Cognitive impairment and dementia are usually present in advanced stages of CBD, but may be among the clinical problems at presentation. Again, diagnosis of definite CBD requires histopathologic evaluation. As many as 40% of patients with clinical diagnoses of CBS may instead demonstrate Alzheimer disease, frontotemporal lobar degeneration (usually with TDP-43 depositions) or Lewy body dementia changes at autopsy {^13,27}.

NEUROPATHOLOGY

Both PSP and CBD are characterized at a cellular level by pathologic accumulations of misfolded protein aggregates consisting of the microtubule-associated protein, tau (MAPT or tau) {¹⁴}. Tau functions physiologically as a stabilizer of microtubules through binding to polymerized tubulin, particularly important for axonal maintenance in neurons (see {²⁹} for review). Tau is encoded by a single human gene on chromosome 17, and is translated ultimately in several distinct primary sequence variants on the basis of post-transcriptional mRNA splicing. The MAPT gene consists of 16 exons, variably included in mRNA species. Alternative splicing with inclusion / exclusion of exons 1 and 2 gives rise to 3 variants: one inclusive of both exons, another with exclusion of exon 2 and a third with exclusions of both exons 1 and 2. Additional transcriptional variation is imposed by variable inclusion of exon10. This region is one of several repeated sequence domains (R domains) which code protein regions where tau interacts with microtubules. Tau mRNA species may include 4 R domains (termed 4 repeat tau variants, 4R tau), or may exclude exon 10, resulting in mRNAs encoding 3 repeat tau variants (3R tau). A range of neurodegenerative conditions are characterized by specific deposits of aggregated tau proteins with apparent selective involvement of tau isoforms and differing protein deposit ultrastructures {^30–34}. The most well-recognized pathologic tau deposits are the paired helical filaments (PHF) seen in cerebrocortical apical dendrites in Alzheimer disease (AD), required for pathologic diagnosis of definite AD. The PHF deposits consist of combined 3R and 4R tau proteins with prominent hyper-phosphorylation compared to the physiologic, soluble tau expressions.

It is currently appreciated that several neurodegenerations beyond AD are associated with pathologic tau depositions, and these are may be further characterized by the ultrastructural characteristics of the deposits {³⁴} as well as by the tau isoforms involved. Immunochemical histologic staining with antibodies selective for 3R or 4R tau species identify the protein isoforms associated with pathologic inclusions {^30,33}. Both PSP and CBD as well as some cases of bvFTLD are characterized pathologically by deposits of hyper-phosphorylated 4R tau {^30,31,35}, ultrastructurally consisting of both straight and twisted filaments. The neuropathological features associated with PSP and CBD and with their distinction are based on the neuroanatomic distributions of the 4R tau pathology and especially on the microscopic cellular appearances of the lesions, as summarized below.

PSP

The neuropathology most consistently observed in PSP is neuronal loss and gliosis involving the posterior fossa, basal ganglia and frontal lobe cerebral cortices {^36,37}. Specific nuclei and sites in these regions are selectively involved, including: the superior colliculus, midbrain periaqueductal grey matter, substantia nigra, locus coeruleus and cerebellar dentate nuclei; the globus pallidus magnocellular basal forebrain and subthalamic nucleus; precentral cerebral cortex. Conspicuous atrophy of the midbrain, superior cerebellar peduncle is typical. It is appreciated as well that the distribution of neuropathology in PSP cases is heterogeneous, related to the variety of clinical syndromic presentations {^{20,21,38–40}}. Overall, cognitive impairment is associated with the severity of tau pathology {⁴¹}. On a microscopic level, fibrillary deposits of 4R tau are present in involved neurons. Astrocytes demonstrate specific somewhat amorphous tau protein deposits, resulting in “tufted” astrocytic cytoplasm – specific to the diagnosis of PSP {³⁷}. Oligodendroglia display cytoplasmic abnormalities with filamentous inclusions termed “coiled bodies”.

CBD

Neuropathological changes in CBD are most often found in the posterior frontal and anterior parietal cerebral cortices, basal ganglia and the substantia nigra {^{29,30,33,37,42}}. Frequently, focal cerebrocortical atrophy is present in the regions demonstrating most intense neuronal loss and gliosis. Affected neurons may demonstrate a swollen or “ballooned” cytoplasmic morphology. Pathologic changes are observed also in glia, with clusters of astrocytic processes demonstrating fibrillary 4R tau deposits, termed “astrocytic plaques” – the diagnostic hallmark of CBD {^37,42,43}. Fibrillary deposits are seen in gray matter as well as in subjacent cerebral white matter structures. Although the ultrastructure of 4R tau deposits in CBD resemble those characteristic of PSP, detailed pathologic analyses indicate that the tufted astrocytes typical of PSP are not co-expressed in the brains of cases demonstrating the astrocytic plaques typical of CBD {^44,45}. Thus, the entities appear pathologically distinct {⁴⁶}. Further, the involvement of cerebral cortices is often more evident in CBD than PSP, and the conspicuous atrophy of the midbrain and superior cerebellar peduncle seen in PSP are absent in CBD. It is proposed that the astrocytic changes characteristic of CBD evolve earlier than do the neuronal pathologic changes on the basis of neuropathology in asymptomatic, presumed early cases {⁴⁷}.

CEREBRAL GLUCOSE METABOLIC IMAGING

PSP

Mapping functional neuronal activity utilizing [¹⁸F]fluorodeoxyglucose positron emission tomography (FDG-PET) has been reported by several investigators in subjects with APD, including PSP and CBD. The majority of studies have selected subjects with PSP-Richardson syndrome or with CBS, with relatively uncommon availability of postmortem confirmation of underlying neuropathology. Despite use of varying analytical approaches, the most common features defining PSP in FDG-PET are relative hypometabolism of the dorsal and dorsolateral prefrontal and medial prefrontal / anterior cingulate cortices {^48,49}. Voxel-based correlation analysis reveals anterior cingulate cortical hypometabolism associated with vertical gaze palsy and cerebellar vermis hypometabolism associated with nystagmus {⁵⁰}. Hypometabolism in the striatum, affecting predominantly the caudate nucleus, is a common additional feature in autopsy-confirmed PSP {⁵¹}. Metabolic abnormalities are most often bilateral and symmetrical in involvement of these structures. More recent studies, perhaps owing to improved spatial resolution in contemporary PET imaging, identify also conspicuous hypometabolism in the brain stem, particularly involving the dorsal midbrain {⁵¹}. Metabolic activity is relatively preserved in parietal and temporal neocortices and the posterior cingulate cortex, serving to distinguish PSP from the typical pattern seen in AD. Subjects with PD are also distinguished from PSP by preservation of precentral and prefrontal cortical activity and by relative hypermetabolism in the striatum, most conspicuous in the putamen of PD {⁵²}. Often, changes characteristic of the PSP metabolic pattern are evident on subjective review of individual cases (Fig 1), however, improved sensitivity and specificity of classification are afforded by statistical procedures employing voxel-level comparisons of patients with normal subject databases or by identifying subtle covariate patterns of abnormal activity that may be difficult to recognize in some cases (see discussion below).

Figure 1:

Cerebral glucose metabolic examples of subjects with clinical syndromes associated with Parkinsonism and dementia. In each section, the top row of images depicts surface-rendering of cerebrocortical FDG activity. Colored voxels represent regions of significantly reduced metabolism compared to a database of normal subjects according to the key at bottom right of the figure: −1.5 SD (light blue) to −5 SD (dark violet). The second row of images in each section depicts selected transaxial levels from dorsal (image left) to ventral (image right). Regions of reduced metabolic activity are indicated by blue arrowheads; regions of increased activity are depicted by red arrowheads. A: Corticobasal Syndrome. Glucose hypometabolism in the right primary somatosensory cortex and the right thalamus. B: Progressive Supranuclear Palsy – Richardson Syndrome. Glucose hypometabolism in bilateral sensorimotor cortices, striatum, thalamus and in the midbrain. C: Dementia with Lewy Bodies. Glucose hypometabolism in bilateral temporoparietal association cortices and in left greater than right primary visual cortices. Glucose metabolism is increased in the striatum bilaterally as can be seen in nigrostriatal degeneration in the absence of dopamine replacement therapy. D: Primary Progressive Non-Fluent Aphasia. Glucose hypometabolism in the left mid-frontal cerebral cortex (Broca’s area). E: Behavioral Variant Frontotemporal Lobar Degeneration. Glucose hypometabolism in bilateral prefrontal association cortices.

CBS

Subjects with CBS most often demonstrate lateralized FDG hypometabolism centered on the peri-Rolandic cortices, with additional involvement of the thalamus {^53–56} (Fig 1). The striatum may be additionally asymmetrically involved. The metabolic decreases are observed with most prominent abnormality in the cerebral hemisphere contralateral to the most clinically-involved limbs. Again, the pattern of cortical hypometabolism is distinct from that typical of AD, affording good distinction among subjects undergoing evaluation for cognitive impairment.

Differential Classification of APD

Qualitative analyses of FDG-PET scans, assisted by voxel-level comparisons to normal subject scans, demonstrate ability to classify APD subjects according to clinical diagnoses at follow-up {^57–62}. Most studies are not supported by neuropathological confirmation of definite diagnoses, however, distinctions of PSP/CBD from PD and MSA are most likely to be correct. Importantly, prognosis and overall survival in Parkinsonism is predicted by FDG-PET classification {⁶³}. Accurate correlation of FDG-PET patterns with PSP-Richardson versus CBS is also likely (see {⁶⁴} for review). An informative autopsy-based study identified interhemispheric asymmetric peri-Rolandic frontoparietal cortical and striatal plus thalamic hypometabolism in CBS due to CBD, while CBS due to AD pathology demonstrated asymmetric posterior parietal and temporal plus posterior cingulate cortex hypometabolism and CBS due to PSP pathology demonstrated more frontal cortex predominant hypometabolism involving also the anterior cingulate {⁶⁵}. Within the spectrum of PSP / CBD, owing to considerable overlap of clinical syndromes and underlying neuropathologies, the latter are not likely to be entirely resolved by FDG-PET imaging.

Multivariate Image Analyses

Advanced, voxel-based multivariate statistical algorithms for analyses of FDG-PET studies have been employed by several investigators in analyses of scans obtained in cases of Parkinsonism. This approach was initially introduced and identified a pattern of latent covariations in FDG activity in subjects with PD, not necessarily evident in subjective scan interpretations or in region-of-interest based univariate approaches. {⁵²}. The approach has the theoretical, and realized, advantage of detecting patterns of metabolic change that are consistent, yet potentially within the overall range of normal variability. This permits feasibility of extending FDG analyses to patients with earlier stage, minor expression of symptoms and signs. Extension of the principal component analytical approach to groups of parkinsonian patients resulted in identification of distinct covariate patterns of activity characteristic of PD, MSA, PSP and CBD {^66–75}. In virtually all reported applications of these analytical approaches, correct subject classification compared to subsequent clinical outcome diagnosis is achieved in approximately 90% of cases, and distinctions may be identified early in symptomatic presentation. A relative limitation of these investigations remains that the comparison clinical diagnosis is achieved over time at the bedside – there are relatively few instances of neuropathological confirmatory validation. Nevertheless, the principle component analysis approach affords opportunity to make an early characterization of the underlying neurologic dysfunction, and may significantly contribute to refining clinical diagnosis of probable PSP or CBD.

SYNAPTIC NEUROCHEMICAL MARKERS

Presynaptic Nigrostriatal Terminals

Many investigators have studied and reported on synaptic neurochemical markers in PD and in APS, including PSP and CBS {see {^62,76,77,78} for reviews). The majority of interest has focused on imaging of nigrostriatal dopaminergic synapses as representative of the major recognized pathology in PD. A number of distinct presynaptic neurochemical targets have been successfully translated for assessment of dopamine terminals in the striatum, beginning with [¹⁸F]fluorodopa (FDOPA), acting a s false neurotransmitter synthetic substrate {^79,80}. This was followed by introduction of ligands targeting the presynaptic dopamine re-uptake transporter (DAT) {⁸¹} and by ligands targeting the monoaminergic synaptic vesicular transporter (VMAT2) {⁸²}. All of these imaging approaches depict the significant nigrostriatal lesions in PD subjects with similar overall sensitivity. Initial extension to imaging APD was reported with the use of FDOPA {⁸³}. Subjects with PD, multiple system atrophy (MSA) and PSP all demonstrated FDOPA defects, with PD demonstrating significantly greatest reductions in the posterior putamen relative to the caudate nucleus. Similar overall likelihood of abnormal scans was found in the APS subjects, with apparent more prominent abnormality in the caudate nucleus, comparable to the putamen abnormality, in PSP and MSA. Side-to-side asymmetry may be more conspicuous in PD versus APS. In clinical translation, however, to the use of [¹²³I]ioflupane SPECT DAT imaging, overall scan appearances do not reliably distinguish between PD and APS patients - all demonstrate significant reduction in striatal tracer localization {^84,85,86}. Striatal dopaminergic denervation correlates with the severity of covariate FDG patterns of hypometabolism in PD and CBS, but not in PSP, suggesting that better diagnostic distinction may be achieved on the basis of FDG imaging {⁸⁷}.

Striatal Dopamine Receptors

In addition to DAT imaging, studies depicting predominantly post-synaptic dopamine D2 receptors on intrinsic striatal neurons may characterize APD syndromes. With presence and progression of intrinsic striatal neuropathology in APS, D2 receptor binding may be reduced on the basis of SPECT imaging of [¹²³I]iodobenzamide (IBZM). Several investigators have employed this imaging strategy, identifying losses of DAT in PD and APS, but with maintained normal D2 receptors in the PD striatum {^88–90}. However, the majority of these studies have included 3 most common APS types: PSP, CBS and MSA. Of these, the most consistent expression of intrinsic striatal neurodegeneration is in MSA. Thus, distinction of PD versus APS is often driven by inclusion of MSA subjects. In fact, studies that sub-classify APS often identify less severe nigrostriatal degeneration in PSP and CBD than in MSA {^88,89}. Reduction in striatal D2 receptors is insignificant in occasional cases of PSP and many cases of CBS {^88–91}. In distinction of PD versus APS, diagnostic abnormalities of FDG metabolism are more frequent than are changes in IBZM SPECT imaging of striatal D2 receptors {^90,92}.

Acetylcholinergic Markers

Approaches to molecular imaging of presynaptic cholinergic synapses have been developed and applied to APS, including PSP and CBS. Initially introduced were methods for imaging of acetylcholinesterase (AChE) activity on the basis of enzyme substrate tracers that undergo conversion to polar, locally trapped brain metabolites, resulting in images depicting enzymatic activity {^93,94}. A report describing a variety of Parkinsonian syndromes studied with one such tracer, N-[¹¹C]-methylpiperidinyl proprionate (PMP), found marked reduction of activity in the striatum, thalamus, brain stem and cerebellum of PSP-Richardson subjects compared to controls {⁹⁵}. Significant reductions in PD subjects were observed in the same regions, but were less severe. Additionally, there was widespread reduction of activity throughout the cerebral cortices of similar magnitude in PD and PSP (average 13–15% reductions from controls). Studies employing the AChE substrate tracer [¹¹C]N-methyl-4-piperidyl acetate (MP4A) also revealed reduction of activity in the thalamus and striatum of PSP relative to normal subjects and mild, diffuse cerebrocortical reduction. Comparison subjects with PD demonstrated more severe cerebrocortical reductions and less severe reductions in the thalamus and striatum {⁹⁶}. An additional report employing MP4A again revealed substantial activity reduction in PSP thalamus and mild reduction throughout the cerebral cortex {⁹⁷}. Patients with CBS, in comparison, had relatively normal thalamic activity, but more severe cerebrocortical reductions. Additional comparison to bvFTLD subjects revealed relative preservation of both thalamic and cerebrocortical activities. The reduced PET AChE activity in PSP thalamus is consistent with losses of brain stem ascending projections from the pedunculopontine and laterodorsal tegmental nuclei, documented in postmortem analyses {^98,99). Striatal losses of AChE PET activity are consistent with known decreases in intrinsic striatal cholinergic neurons {¹⁰⁰}. The neocortical AChE PET reductions are expected on the bases of neuronal losses in the basal forebrain cholinergic nuclei that are more severe in CBD than in PSP (99).

IMAGING PROTEINOPATHY

Important recent advances in molecular imaging of neurodegenerative disease are based on the design and translation of radioligands targeting the characteristic pathologic protein depositions conferring neuropathological diagnoses (see {¹⁰¹} for recent review). Initially{¹⁰²}, introduction of ligands targeting fibrillary deposits of Aβ amyloid characteristic of Alzheimer disease (AD) demonstrated feasibility of the approach, and have contributed important and unique insight into the role of amyloid deposition in the development of AD, the population-based natural history of cerebral Aβ deposition {^103,104} and its association with clinical dementia syndromes {¹⁰⁵}.

Building on this evidence and approach, more recent investigations have sought similar radioligand approaches to the detection of other proteinopathies, particularly tau deposits {^34,106–109}. Again, interest focused initially on detection of the protein depositions characteristic of PHF in AD. Early suggestion of NFT-tau radioligand targeting was reported for [¹⁸F]FDDNP, an agent initially developed for Aβ amyloid depiction {¹¹⁰}. However, it became apparent that the ligand targeted also tissues with high NFT burden, but without Aβ (e.g. hippocampal formation in AD) on the basis of binding both to Aβ plaques and to NFT deposits. Many laboratories conducted searches for suitable small molecule ligands that would be appropriate to specific imaging of PHF in AD. These investigations have highlighted an important limitation - unlike fibrillary Aβ deposits where a small molecule histologic stain (thioflavin-T) was available as an exemplar, there are no prior small molecules with defined specificity for localizing NFT or other tau protein accumulations. Neuropathological detection of tau deposits relies on silver staining approaches to tissue preparation, buttressed by use of immunohistochemistry employing antibodies specific to tau protein epitopes. These histological approaches do not readily translate to the design of small molecular radioligands. The approach used by most studies in search of potential tau radioligands has been use of synthetic tau fibrils, and/or tau fibrils isolated from human tissue samples, in high-throughput screens of large libraries of compounds. This approach has successfully identified numerous NFT tau imaging candidates, however, is limited by difficulty in establishing the detailed specificity of ligand interactions with potentially confounding off-target sites.

T807 / AV-1451 / Flortaucipir

Among the early-identified radioligands selectively targeting NFT tau deposits in AD is the benzamide initially designated [¹⁸F]T807 (subsequently renamed to AV-1451) {¹¹¹}, and now designated as flortaucipir-F18 after recent US FDA approval (Figure 2). Multiple studies have demonstrated flortaucipir uptake and retention in brain regions known to be involved with PHF tau depositions in AD, correlating generally in anatomic extent with clinical severity of cognitive impairment and dementia {¹¹²}.

Figure 2:

Structures of radioligands reported in PET investigations of 4R tauopathies. T807 is additionally known as AV-1451 or as Flortaucipir-F18.

The possibility of flortaucipir labeling of other tau aggregates, specifically 4R fibrils, has been studied by several investigators in PSP and CBS subjects, with mixed results. Several studies report increased flortaucipir accumulation in PSP relative to neurologically-normal subjects in the cerebellar dentate nucleus, brain stem, subthalamus and thalamus and in the basal ganglia {^113–118}. Most distinct increases are reported in the globus pallidus and the cerebellar dentate nucleus. Results demonstrate considerable overlap of tracer retention measures between groups. These findings are generally in keeping with the expected distribution of pathology from prior autopsy investigations. However, other reports fail to support significant differences in tracer accumulation in basal ganglia (including globus pallidus) or cerebellum (including dentate nucleus) {¹¹⁹}. Studies fail also to identify tracer intensity or distribution correlates of PSP symptomatic severity {¹¹⁷} and do not confirm at autopsy that PET binding is associated with in vitro tracer localization to tau deposits {^118,120}. Interestingly, studies comparing PSP with PD and normal subjects identify apparent reduction of uptake in the PD substantia nigra, attributed to off-target tracer binding to neuromelanin, and depicting its reduction in PD {^121,122}. Investigators have also studied flortaucipir uptake in subjects with CBD, indicating asymmetric increased tracer retention in peri-Rolandic cerebral cortex and in the corticospinal subcortical white matter and in the striatum contralateral to the most involved limbs {¹²³}. These abnormalities were distinct in comparison to PSP and AD subjects, but the effect sizes of the changes were minor.

The aforementioned in vivo tracer localization studies, however, are complicated by direct in vitro ligand binding interaction studies performed in tissues from a range of neurodegenerative conditions. Several laboratories confirm high-affinity binding of flortaucipir to NFT-tau in tissue sections and homogenates from patients with AD. Tissues bearing 4R tau aggregates in PSP and CBD do not express such binding {^124–126}. Binding is identified also in tissues with elevated neuromelanin content, especially in the substantia nigra, indicating off-target (non NFT-tau) binding {¹²⁴}. There is additional evidence supporting the possibility of ligand interaction with MAO-B {¹²⁷}, although in vivo evidence appears contradictory to a significant MAO component of flortaucipir uptake in PET imaging {¹²⁸}. In summary, it appears that flortaucipir may bind with relatively low affinity to 4R tau deposits in PSP and CBD, but the intensity of these changes in PET images is limited, and in vitro evidence does not support significant high affinity tracer binding to 4R tau deposits.

THK5351

Another family of chemical ligand structures, quinoline derivatives, have been investigated for their ability to detect tau deposits with PET imaging {¹²⁹}. Emerging from a series of structure-activity modifications with progressively improving kinetic and regional brain biodistribution properties in AD patients is the ligand designated THK5351 {¹³⁰} (Figure 2). Investigators report increased ligand retention in the globus pallidus and midbrain of a PSP-Richardson syndrome subject {¹³¹} and asymmetrically in the frontoparietal cortex and globus pallidus of several subjects with CBD {¹³²}. However, it has become apparent that most of the high-affinity binding of THK5351 depends on MAO-B, not on tau deposits {^133,134}. Thus, the anatomic distribution properties in PSP and CBD are most likely attributable to increased MAO-B expression in reactive astrocytes in areas of cellular injury and ongoing neurodegeneration {¹³⁵}.

PBB3

A final chemical class of ligands developed for potential PET imaging of tau deposits are benzothiazole derivatives, screened for in vitro interactions with AD brain tissue and for in vivo uptake in a transgenic mouse model expressing human mutant MAPT (P301S FTDP-17 mutation), resulting in 4R tau deposits in the brain stem and spinal cord (PS19 mouse). The selected ligand, designated PBB3 (Figure 2), demonstrated binding to NFT-tau in AD tissue and increased binding in vitro and in vivo to the PS19 mouse brain stem deposits {¹³⁶}. Human clinical translation demonstrates increased PBB3 retention in the cerebral cortices of AD patients, but also conspicuous subcortical retention in a CBD patient {¹³⁶}. Subsequent study identified increased ligand retention in PET studies of PSP patients, with abnormalities identified in basal ganglia, subthalamic nucleus and midbrain {^137,138}. There was correlation of severity of Parkinsonism with the intensity of increased uptake in the subcortical white matter, and of cognitive impairment with increased uptake in cerebrocortical gray matter {¹³⁷}. Additional subjects with MSA and familial PD due to SNCA duplication demonstrated increased tracer binding in basal ganglia, thalamus and brain stem, suggesting the possibility of off-target binding to alpha-synuclein deposits in these disorders {¹³⁸}. In vitro binding studies indicate that PBB3 binding to synuclein deposits in MSA could be detectable in PET imaging {¹³⁹}. Comparative in vitro binding studies reveal that PBB3 and flortaucipir each bind with high affinity to NFT in AD tissues, while PBB3 binds differentially to 4r and 3R tau deposits in other neurodegenerations, not interacting with flortaucipir with high affinity {¹⁴⁰}. Further comparative in vitro assays demonstrate high affinity binding of PBB3, flortaucipir and THK5351 in AD tissue, however, the PBB3 binding was not cross-inhibited by flortaucipir or THK5351, indicating distinct interactions with PHF tau structures {¹⁴¹}. MAO-B blocking with deprenyl substantially reduced high-affinity binding of THK5351, but not that of flortaucipir or PBB3.

SUMMARY and FUTURE DIRECTIONS

The 4R tauopathies, reviewed here consisting of PSP and CBD, represent relatively uncommon neurodegenerative syndromes compared to AD and PD, but confer significant patient disability without available effective therapy. At symptomatic presentation, the challenge to these disorders is diagnosis. Initial clinical abnormalities can mimic PD or a primary dementia syndrome (often a FTLD subtype). Unfortunately, there is currently no established diagnostic approach to identify patients with PSP or CBD in the earliest stages of the disorders, limiting not only correct diagnosis and prognosis, but importantly also, preventing the design and evaluation of potential novel therapeutic interventions. As discussed previously, PSP and CBD remain neuropathological diagnoses, and even the most suggestive clinical syndromes (i.e. PSP-Richardson or CBS) may demonstrate non-4R tau final pathologic diagnoses in some patients.

Improved application of molecular imaging biomarkers may improve early clinical diagnostic accuracy. In particular, use of FDG-PET with advanced multivariate analyses appears most favorable. Current studies, however, are limited by relatively scarce data to confirm underlying neuropathology in the majority of these studies - most have relied on early scan analyses to predict later clinical diagnostic classifications. The latter, as discussed previously, can be misleading compared to final pathologic diagnoses. Studies advancing FDG-PET to potentially identify even more specific features that accurately predict neuropathological diagnoses are needed. Other molecular imaging approaches to study of defined neurochemical populations (dopaminergic or acetylcholinergic) do not offer distinction of the 4R tauopathies from more common movement disorder and cognitive disorder neurodegenerations.

Emerging evidence suggests the possibility of developing targeted interventions in PSP and CBD on the basis of misfolding and misprocessing of 4R tau in both disorders. Future investigation into the mechanism(s) underlying these protein processing abnormalities could yield important clues to new therapeutic approaches. Such future progress could be supported and accelerated by the development of novel radiotracers specific for binding to 4r tau straight and twisted fibrils characteristic of the deposits in PSP and CBD. The potential for this concept is suggested by the apparently distinct binding interactions of flortaucipir versus PBB3 in PFH, suggesting differing high-affinity sites for the tracers. Ongoing studies are seeking to identify selective sites on 4R deposits that are not shared by PHF or other protein deposits. If successful, such specific 4R tau ligands could serve not only to identify patients, but also to follow progression versus improvement of depositions in therapeutic trials.