PET Imaging of Pathological Tau in Progressive. Supranuclear Palsy

Transcription

1 PET Imaging of Pathological Tau in Progressive Supranuclear Palsy By Sarah Coakeley A thesis submitted in conformity with the requirements for the degree of Master of Science Institute of Medical Science University of Toronto Copyright by Sarah Coakeley 2016

2 PET Imaging of Pathological Tau in Progressive Supranuclear Palsy Sarah Coakeley Abstract Master of Science Institute of Medical Science University of Toronto 2016 PSP is a neurodegenerative movement disorder that is characterized by the pathological accumulation of tau aggregates in the brain. PSP differs neuropathologically from other movement disorders such as PD and MSA, which are classified as synucleinopathies. PSP patients would benefit from a clinically approved tau imaging agent for diagnostic and prognostic purposes. Previous testing of the tau radiotracer [ 18 F]AV-1451 in AD, another tauopathy, suggested that it was capable of distinguishing AD from MCI and healthy controls. The aim of this investigation was to test whether [ 18 F]AV-1451 was able to distinguish PSP patients from PD patients and healthy controls. There were no significant increases in [ 18 F]AV-1451 retention in PSP patients compared to the other two subject groups. These findings may indicate that [ 18 F]AV-1451 is not an effective radiotracer for imaging tau in PSP, or perhaps the method of measuring [ 18 F]AV-1451 retention used in this study was not appropriate. ii

3 Acknowledgments My time at the University of Toronto has been filled with self-discovery and independence. However, I would not have had the success and learning experiences I did without the help of many incredible people. I have had the opportunity to collaborate with amazing minds and work with state of the art technology, and for that I am truly grateful. I would like to begin by acknowledging my mentor and supervisor, Dr. Antonio Strafella, who provided me with the opportunity to complete my master s thesis in his lab. Through his guidance and example, Dr. Strafella has taught me what it means to be a diligent researcher, a compassionate doctor, and a dedicated family man. Dr. Strafella has provided me with exposure to the world of nuclear imaging and thereby the opportunity to develop my analytical, technical, and personal skills. I am so thankful for the support, training, and unique experiences I have been granted. I would like to acknowledge the exceptional guidance, encouragement, and time commitment provided by Dr. Ariel Graff-Guerrero and Dr. Robert Chen. Thank you to Dr. Pablo Rusjan for your countless hours, tremendous guidance, and sharing of your knowledge. Thank you to Alvina Ng, Laura Nguyen, Anusha Ravichandran and Dr. Sylvain Houle for your much-appreciated support. Thank you to the members of my lab for creating a cooperative and welcoming environment. My days working in Dr. Strafella s lab will be remembered fondly, not only from the lessons I learned, but also the friends that I have made. Thank you to Sang Soo iii

4 Cho for being so approachable and dedicated to teaching me about parametric analysis. I will always appreciate your guidance and patience. Thank you to Leigh Christopher and Yuko Koshimori for sharing your knowledge, offering advice, and teaching me the importance of balance in my life. To Marc Jacobs and Marion Criaud, thank you for your guidance, encouragement, and many laughs. Thank you to Christine Ghadery, Crystal Li, Alex Mihaescu, and Rostom Mabrouk for your helpfulness, kindness, and friendships. I would like to acknowledge the friendships, new and old, that have made my time as a graduate student unforgettable. To Sam Fernandes, who came to U of T with me from our undergraduate years, thank you for being an incredible friend, always willing to listen and offer amazing advice. I am so thankful to have met Dunja Knezevic and Anton Rogachov and cherish their support, humor, and friendship over the past two years. Thank you to Zach Lister for your patience, motivation, and countless laughs along the way. I would also like to acknowledge Shannen Busch, Alan Fauteux, and Lisa Shawcross for their invaluable friendships and constant encouragement. I would like to thank my family for their unyielding support and reassurance. To my dad, who dedicated hours to grammatically correcting my manuscripts, and to my mom, who always offered a considerate ear, thank you. I wish to acknowledge my siblings, Joey and Elizabeth, for their comic relief and unparalleled hard work in their respective fields. I know this work would not have been possible without my incredible family. iv

5 Contributions Dr. Antonio Strafella (Supervisor): provided laboratory resources, assistance with study protocol, analysis of results, and guidance to manuscript writing Dr. Ariel Graff-Guerrero: contributed to interpretation of results and guidance to manuscript writing Dr. Robert Chen: contributed to interpretation of results and guidance to manuscript writing Dr. Sang Soo Cho: provided assistance with analysis and direction of manuscript write up Dr. Pablo Rusjan: provided assistance with analysis and direction of manuscript write up Alvina Ng: reconstructed the individual PET images Dr. Alan Wilson: provided expertise in radiosynthesis of [ 18 F]AV-1451 v

6 Table of Contents Abstract. Acknowledgements... Contributions. Table of Contents.. List of Tables List of Figures... List of Abbreviations ii iii v vi x xi xv 1.0 Literature Review Proteinopathies Tauopathies Alzheimer s Disease Corticobasal Degeneration Frontotemporal Dementia Progressive Supranuclear Palsy Synucleinopathies Parkinson s Disease Multiple System Atrophy Progressive Supranuclear Palsy Clinical Presentations & Diagnosis Neuropathology Symptom Management & Treatment vi

7 1.3 Positron Emission Tomography PET Tau Radiotracers FDDNP THK PBB T AV Development & Preclinical Testing Human Testing Significance Aim & Hypothesis Aims Hypothesis Methods Participants & Experimental Design Radiosynthesis of [ 18 F]AV MRI Acquisition PET Acquisition Image Analysis Region of Interest Analysis Standard Uptake Value [ 18 F]AV-1451 Image Analysis 42 vii

8 3.6 Statistical Analysis Results Participant Demographic Region of Interest Analysis Time-Activity Curves SUV SUVR Cerebellum Partial Volume Correction SUVR Corpus Callosum [ 18 F]AV-1451 & MoCA Discussion Overviews of Findings Demographics [ 18 F]AV-1451 Retention Reference Regions Atrophy & Partial Volume Correction Off Target Binding Tau Radiotracers for PSP Experimental Limitations Conclusion Future Directions In Vitro Testing viii

9 7.2 In Vivo Testing Developing New Radiotracers 107 References. 109 ix

10 List of Tables Table 4-1. Participant demographics. 46 Table 4-2. Mean SUV (standard deviation) by ROI across groups from minutes.. 53 Table 4-3. Mean SUV (standard deviation) by ROI across groups from minutes.. 54 Table 4-4. Mean SUVR (standard deviation) from minutes using the cerebellum as a reference region.. 59 Table 4-5. Mean SUVR (standard deviation) from minutes using the cerebellum as a reference region.. 60 Table 4-6. Partial volume corrected mean SUVR (standard deviation) from minutes using the cerebellum as a reference region 76 Table 4-7. Partial volume corrected mean SUVR (standard deviation) from minutes using the cerebellum as a reference region 77 Table 4-8. Mean SUVR (standard deviation) from minutes using the corpus callosum as a reference region.. 88 x

11 List of Figures Figure 1-1. Scoring of neuropathological accumulation of tau in PSP.. 13 Figure 1-2. SUVR images from static scan ( minutes) Figure 1-3. VOI SUVRs ( minutes) for each subject. 30 Figure 4-1. Time-activity curve of cerebellum and putamen in PSP subject 48 Figure 4-2. Time-activity curve of cerebellum and putamen in PD subject.. 49 Figure 4-3. Time-activity curve of cerebellum and putamen in HC subject.. 50 Figure 4-4. Mean SUV of the cerebellum from minutes. 55 Figure 4-5. Mean SUV of the cerebellum from minutes. 55 Figure 4-6. Mean SUV of the corpus callosum from minutes. 56 Figure 4-7. Mean SUV of the corpus callosum from minutes. 56 Figure 4-8. Mean PSP group parametric image of mean SUVR minutes Figure 4-9. Mean PSP group parametric image of mean SUVR minutes Figure Mean PD group parametric image of mean SUVR minutes Figure Mean PD group parametric image of mean SUVR minutes Figure Mean HC group parametric image of mean SUVR minutes.. 63 Figure Mean HC group parametric image of mean SUVR minutes.. 63 Figure Mean SUVRs of the frontal lobe from minutes using the cerebellum as a reference region. 64 Figure Mean SUVRs of the frontal lobe from minutes using the cerebellum as a reference region.. 64 Figure Mean SUVRs of the inferior parietal lobe from minutes using the cerebellum as a reference region 65 xi

12 Figure Mean SUVRs of the inferior parietal lobe from minutes using the cerebellum as a reference region 65 Figure Mean SUVRs of the temporal lobe from minutes using the cerebellum as a reference region.. 66 Figure Mean SUVRs of the temporal lobe from minutes using the cerebellum as a reference region Figure Mean SUVRs of the occipital lobe from minutes using the cerebellum as a reference region.. 67 Figure Mean SUVRs of the occipital lobe from minutes using the cerebellum as a reference region.. 67 Figure Mean SUVRs of the caudate from minutes using the cerebellum as a reference region.. 68 Figure Mean SUVRs of the caudate from minutes using the cerebellum as a reference region.. 68 Figure Mean SUVRs of the putamen from minutes using the cerebellum as a reference region.. 69 Figure Mean SUVRs of the putamen from minutes using the cerebellum as a reference region.. 69 Figure Mean SUVRs of the striatum from minutes using the cerebellum as a reference region.. 70 Figure Mean SUVRs of the striatum from minutes using the cerebellum as a reference region.. 70 Figure Mean SUVRs of the globus pallidus from minutes using the cerebellum as a reference region.. 71 Figure Mean SUVRs of the globus pallidus from minutes using the cerebellum as a reference region.. 71 Figure Mean SUVRs of the substantia nigra from minutes using the cerebellum as a reference region.. 72 Figure Mean SUVRs of the substantia nigra from minutes using the cerebellum as a reference region.. 72 xii

13 Figure Mean SUVRs of the thalamus from minutes using the cerebellum as a reference region.. 73 Figure Mean SUVRs of the thalamus from minutes using the cerebellum as a reference region.. 73 Figure Mean SUVRs of the dentate nucleus from minutes using the cerebellum as a reference region.. 74 Figure Mean SUVRs of the dentate nucleus from minutes using the cerebellum as a reference region.. 74 Figure Mean partial volume corrected SUVRs of the frontal lobe from minutes using the cerebellum as a reference region 78 Figure Mean partial volume corrected SUVRs of the frontal lobe from minutes using the cerebellum as a reference region 78 Figure Mean partial volume corrected SUVRs of the parietal lobe from minutes using the cerebellum as a reference region 79 Figure Mean partial volume corrected SUVRs of the parietal lobe from minutes using the cerebellum as a reference region 79 Figure Mean partial volume corrected SUVRs of the temporal lobe from minutes using the cerebellum as a reference region.. 80 Figure Mean partial volume corrected SUVRs of the temporal lobe from minutes using the cerebellum as a reference region.. 80 Figure Mean partial volume corrected SUVRs of the occipital lobe from minutes using the cerebellum as a reference region.. 81 Figure Mean partial volume corrected SUVRs of the occipital lobe from minutes using the cerebellum as a reference region.. 81 Figure Mean partial volume corrected SUVRs of the caudate from minutes using the cerebellum as a reference region 82 Figure Mean partial volume corrected SUVRs of the caudate from minutes using the cerebellum as a reference region 82 Figure Mean partial volume corrected SUVRs of the putamen from minutes using the cerebellum as a reference region 83 xiii

14 Figure Mean partial volume corrected SUVRs of the putamen from minutes using the cerebellum as a reference region 83 Figure Mean partial volume corrected SUVRs of the globus pallidus from minutes using the cerebellum as a reference region.. 84 Figure Mean partial volume corrected SUVRs of the globus pallidus from minutes using the cerebellum as a reference region.. 84 Figure Mean partial volume corrected SUVRs of the substantia nigra from minutes using the cerebellum as a reference region.. 85 Figure Mean partial volume corrected SUVRs of the substantia nigra from minutes using the cerebellum as a reference region.. 85 Figure Mean partial volume corrected SUVRs of the thalamus from minutes using the cerebellum as a reference region 86 Figure Mean partial volume corrected SUVRs of the thalamus from minutes using the cerebellum as a reference region 86 Figure 5-1. Diagram of tau band variance across different tauopathies 98 Figure 5-2. Electron micrographs of tau filaments 99 xiv

15 List of Abbreviations AD: Alzheimer s disease ANOVA: analysis of variance Aβ: β-amyloid BBB: blood-brain barrier BDI: Beck Depression Inventory CBD: corticobasal degeneration CNS: central nervous system DLB: dementia with Lewy bodies FDDNP: 2-(1-{6-{(2-[ 18 F]fluoroethyl)(methyl)amino]-2- naphthyl}ethylidene)malononitrile AD: Alzheimer s disease FDG: 2-[ 18 F]fluoro-2-deoxy-D-glucose FTD: frontotemporal dementia ID: injected dose MAP: microtubule associated protein MAPT: microtubule associated protein tau MCI: mild cognitive impairment MDS: Movement Disorder Society MMSE: Mini Mental State Examination MoCA: Montreal Cognitive Assessment MRI: magnetic resonance imaging MW: body weight NFT: neurofibrillary tangles xv

16 PBB: phenyl/pyridinyl-butadienyl-benzothiazoles/benzothiazoliums PD: Parkinson s disease PET: positron emission tomography PHF: paired helical filament PIB: Pittsburgh Compound B PiD: Pick s disease PNS: peripheral nervous system PSP: progressive supranuclear palsy PSPRS: Progressive Supranuclear Palsy Rating Scale ROI: region of interest RRT: relative residence time SPM: Statistical Parametric Mapping SUV: standard uptake value SUVR: standard uptake value ratio TAC: time-activity curve THK: 6-(2-fluoroethoxy)-2-(4-aminophenyl)quinolone UPDRS: Unified Parkinson s Disease Rating Scale VOI: volume of interest xvi

17 1.0 LITERATURE REVIEW Tau pathology is found in many neurodegenerative disorders, from dementias such as Alzheimer s disease to movement disorders such as progressive supranuclear palsy (PSP) and corticobasal degeneration (CBD). Among these tauopathies there exists different confirmations and distribution patterns of pathological tau. Tauopathies such as PSP may have clinical signs that overlap with other Parkinsonian disorders; therefore, a method of distinguishing between tauopathies and non-tauopathies would be very useful. Recently, there has been an increased interest in developing positron emission tomography (PET) radiotracers that bind tau in vivo. 1.1 Proteinopathies Proteinopathies are a result of protein aggregation that is usually confined to the central nervous system (CNS). These inclusions may be due to a number of factors, including genetic mutation, altered post-translational modification, or atypical proteolysis (Wenning & Jellinger, 2005). These abnormalities can change the folding and binding of a protein, thereby altering its secondary or tertiary structure and ultimately the function of the protein. Accumulation of protein aggregates over time can cause damage to susceptible brain regions (Wenning & Jellinger, 2005). 1

18 1.1.1 Tauopathies Tau is a microtubule-associated protein (MAP) found in the CNS and peripheral nervous system (PNS). It is primarily located in the axons of neurons in healthy conditions (Avila, Lucas, Perez, & Hernandez, 2004). Similarly to other MAPs, the role of tau is related to microtubule assembly and stabilization (De Silva et al., 2003; Spillantini & Goedert, 2013). Tau is a natively unfolded, hydrophilic phosphoprotein with a long rod structure and possesses a beta-sheet formation (Avila et al., 2004; Iqbal, Liu, Gong, & Grundke- Iqbal, 2010; Spillantini & Goedert, 2013). The MAP tau (MAPT) gene is located on chromosome 17q21.31 and alternative splicing of its mrna yields six different isoforms of tau (Shahani & Brandt, 2002; Spillantini & Goedert, 2013). Exons 2 and 3 can be alternatively spliced to form isoforms with no amino terminal inserts (0N; lacking exons 2 and 3), one amino terminal insert (1N; lacking exon 3), and two amino terminal inserts (2N) (Iqbal et al., 2010). Alternative splicing of exon 10 produces isoforms with 3 amino acid repeats (3R) or 4 amino acid repeats (4R) on the carboxyl terminal. The repeats contain tubulin and microtubule binding domains; therefore, the 4R isoforms have a slightly higher affinity for microtubules than the 3R isoform (Iqbal et al., 2010; Shahani & Brandt, 2002). In a human brain free of tau pathology the ratio of 3R to 4R tau isoforms is approximately equal (Noble, Hanger, Miller, & Lovestone, 2013). Localization of tau in the neuron is largely dependent on post-translational modifications, most notably phosphorylation (Avila et al., 2004). Tau contains numerous serine and tyrosine phosphorylation sites that allow its activity to be regulated. Phosphorylation of tau decreases its affinity for microtubules, thereby decreasing overall tubulin assembly 2

19 (Shahani & Brandt, 2002). Phosphorylated tau is generally sequestered in the soma of neurons, with small traces found in the nucleus possibly involved in the regulation of MAPT mrna transcription. Non-phosphorylated tau is found in distal axonal regions of neurons and is more prone to proteolysis than its phosphorylated counterpart (Avila et al., 2004; Shahani & Brandt, 2002). Though the process of phosphorylation is reversible, hyperphosphorylation of tau can disrupt the structure of neuronal microtubules and lead to pathological conditions known as tauopathies (Avila et al., 2004; Noble et al., 2013). Neurofibrillary tangles (NFT) are formed from phosphorylated tau aggregates and remain intracellular, until the death of the neuron (Avila et al., 2004; Shahani & Brandt, 2002). In many tauopathies tau are not confined to grey matter but also presents in glial cells, namely astrocytes and oligodendrocytes. There are many classes of tauopathies, all categorized by the accumulation of pathological tau in the brain. Differences in tau isoforms, tau pathology, the distribution of pathological tau within the CNS accounts for the varying symptom manifestations across tauopathies; however, there is also frequent symptom overlap between many tauopathies. The most common tauopathies include Alzheimer s disease (AD), corticobasal degeneration (CBD), frontotemporal dementia (FTD), and progressive supranuclear palsy (PSP). 3

20 Alzheimer s Disease AD is the most common cause of early-onset dementia and the most prominent symptom is episodic memory impairment (Ahmed et al., 2014). This neurodegenerative disorder is classified by beta amyloid (Aβ) plaques and NFTs. While plaque density and distribution is not associated with cognitive decline, tau NFT burden has been found to be correlated with neuronal death and cognitive impairment (Ludolph et al., 2009). The initial development of tau NFTs in AD begins with the hyperphosphorylation of tau, followed by the formation of paired helical filaments (PHF) of tau (Iqbal et al., 2010; Lemoine et al., 2015; Thal, Attems, & Ewers, 2014). Braak and colleagues (1993) developed a staging theory of cortical neurodegeneration in AD. Such theory has been used to pathologically diagnose mild to severe AD. In the preclinical stages I and II there are few to many NFTs in the transentorhinal region, which may extend to the entorhinal region and hippocampus (Braak, Braak, & Bohl, 1993). Stages III and IV mark the initiation of the clinical phase, with the presentation of mild to moderate cognitive impairment. Stages III is distinguished by high levels of NTFs in the superficial entorhinal layer. Involvement in the deep entorhinal cortex marks stage IV and affects the relay of information from the hippocampus to the isocortex (Braak et al., 1993). In stages V and VI, neurodegeneration is sufficient to obtain a pathological diagnosis of AD. Pathology has spread to hippocampal areas and association regions of the isocortex (Braak et al., 1993). NFT burden in the grey matter AD cortex is considerably greater than in other tauopathies (Zhukareva et al., 2006). 4

21 All six tau isoforms are present in AD and there is an equal ratio of 3R to 4R isoforms (De Silva et al., 2003; Iqbal et al., 2010; Spillantini & Goedert, 2013; Thal et al., 2014) Corticobasal Degeneration CBD is primarily a sporadic movement disorder that is difficult to diagnose clinically due to its lack of specificity and clinical heterogeneity (Dickson, 1999; Grijalvo-Perez & Litvan, 2014). Approximately 25-56% of cases are accurately diagnosed ante mortem (Grijalvo-Perez & Litvan, 2014). Cardinal signs of CBD include asymmetrical apraxia, rigidity, bradykinesia, and dystonia. Other clinical presentations consist of aphasia, alien limb phenomenon, cortical sensory deficit, focal myoclonus, and a lack of response to levodopa (M. J. Armstrong et al., 2013; Dickson, 1999; Grijalvo-Perez & Litvan, 2014; Ludolph et al., 2009; Poewe & Wenning, 2002). Cognitive impairment can develop during the disease progression, and may lead to a misdiagnosis of AD (M. J. Armstrong et al., 2013; Ludolph et al., 2009). Pathological diagnosis continues to be the gold standard for CBD, with tau aggregation being the primary source of pathology (Grijalvo-Perez & Litvan, 2014; Stamelou & Bhatia, 2015). The main tau isoform present in CBD tau inclusions is 4R (De Silva et al., 2003; Dickson, 1999; Grijalvo-Perez & Litvan, 2014; Liscic, Srulijes, Gröger, Maetzler, & Berg, 2013; Ludolph et al., 2009; Noble et al., 2013; Spillantini & Goedert, 2013; Thal et 5

22 al., 2014). Straight filament NFTs, astrocytic plaques, and coiled bodies often accumulate in CBD in a clustered formation (R. A. Armstrong & Cairns, 2013; De Silva et al., 2003; Takahashi, 2002). Asymmetrical frontal and parietal lobe atrophy is seen in CBD cases (Ahmed et al., 2014; Poewe & Wenning, 2002). Atrophy can also be seen in the thalamus, subthalamic nucleus, pallidum, dentate nucleus, and brainstem nuclei (Poewe & Wenning, 2002). The majority of CBD pathology is found in the grey and white matter of the cortex (Dickson, 1999; McMillan et al., 2013) Frontotemporal Dementia Frontotemporal dementia (FTD) is the clinical term for a heterogeneous group of syndromes that present with non-alzheimer s dementia (Warren, Rohrer, & Rossor, 2013). FTD is present in approximately 15 to 22 per 100,000. Other names that have been replaced with FTD include Pick s disease (PiD) and frontal lobe dementia of the non- Alzheimer s type. PiD is now only referred to cases with pathological confirmation of Pick bodies, which are argyrophilic, tau inclusions (Pressman & Miller, 2013). The primary tau isoforms in these aggregates is 3R (Arai et al., 2001; Spillantini & Goedert, 2013). FTD is divided into three clinical subgroups: behavioural variant frontotemporal dementia (bvftd), nonfluent variant primary progressive aphasia (nfvppa), and semantic variant primary progressive aphasia (svppa) (Pressman & Miller, 2013; Warren et al., 2013). BvFTD is the most common subtype and is marked by a decline in executive skills, 6

23 abnormal behaviour and differed emotional responses. NfvPPA is categorized by impaired speech and language decline, and svppa is marked by a decline in word comprehension and semantic memory (Warren et al., 2013). Pathologically, FTD is associated with three types of protein inclusions, the most common being hyperphosphorylated tau. Mutations in the MAPT gene have been found in approximately 17-32% of cases (Pressman & Miller, 2013) Progressive Supranuclear Palsy (see section 1.2) Synucleinopathies α-synuclein is a presynaptic, 140 amino acid protein whose function is not well understood (Goedert, Trans, Lond, & Goedert, 1999; Wenning & Jellinger, 2005). While there are two other proteins in the synuclein family (β and γ) only α-synuclein forms aggregates in pathological conditions. Genetic mutations, oxidative stress, and environmental factors may trigger or exacerbate the formation of α-synuclein fibrils (Goedert et al., 1999; Wenning & Jellinger, 2005). The most common neurodegenerative disorders associated with α-synuclein aggregation (α-synucleinopathies) are Parkinson s disease (PD), multiple system atrophy (MSA), and dementia with Lewy bodies (DLB) (Wenning & Jellinger, 2005). 7

24 Parkinson s Disease PD is a neurodegenerative movement disorder affecting 1% of North Americans over the age of 60 (De Lau & Breteler, 2006). In most cases PD is sporadic, with certain environment elements and aging being risk factors; however, seven genes have been linked to familial parkinsonism (Lees, Hardy, & Revesz, 2009). The cardinal signs of PD are bradykinesia, resting tremor, rigidity, and an excellent response to levodopa, which is currently the most effective therapy for PD (Goedert et al., 1999; Lees et al., 2009). Early signs of PD include impaired dexterity, fatigue, stiffness, instability, and falls. Urinary incontinence may develop later on in the disease course, along with monotonous speech and freezing of gait. PD progression is generally slow and unilateral (Lees et al., 2009). Pathologically, PD is distinguished by Lewy bodies, pale bodies, and Lewy neurites. α- synuclein inclusions are the main components of Lewy bodies and neurites (Goedert et al., 1999; Lees et al., 2009). Loss of dopaminergic neurons and depigmentation is seen in the pars compact of the substantia nigra. Atrophy can be present in the locus coeruleus, dorsal nucleus of the vagus, and the raphe nuclei (Lees et al., 2009). 8

25 Multiple System Atrophy MSA is rare atypical parkinsonian syndrome, with a prevalence of approximately cases per 100,000 (Wenning, Colosimo, Geser, & Poewe, 2004). There are two clinical subtypes of MSA: MSA-P which is dominated by Parkinsonian features, and MSA-C with cerebellar ataxia as the primary motor presentation (Poewe & Wenning, 2002; Wenning et al., 2004). The clinical features of MSA-P include akinesia, rigidity, and orofacial dystonia (Poewe & Wenning, 2002). Less common is postural instability that lead to falls, similar to PSP (Liscic et al., 2013; Wenning et al., 2004). MSA-C presents with gait ataxia, limb kinetic ataxia, scanning dysarthria, and cerebellar oculomotor disturbances. MSA patients often do not respond well to levodopa medications (Poewe & Wenning, 2002). Although clinical diagnosis has relatively good specificity, it lacks sensitivity and many cases of MSA are misdiagnosed or undiagnosed (Wenning et al., 2004). Pathologically, MSA is characterized by the aggregation of α-synuclein (Poewe & Wenning, 2002; Wenning et al., 2004). Inclusions can be found in neurons and glial cells, with depigmentation of the substantia nigra and atrophy of the putamen (Wenning et al., 2004). 9

26 1.2 Progressive Supranuclear Palsy Progressive Supranuclear Paly (PSP) is the second most common atypical parkinsonian disorder, affecting approximately 5-6/100,000 North Americans (Golbe, 2014b; Liscic et al., 2013). This neurodegenerative disorder often presents in middle to late age (55 to 70 years) and disease duration is approximately 4 to 5.5 years (Osaki et al., 2004) Clinical Presentations & Diagnosis of Progressive Supranuclear Palsy Onset of PSP symptoms is insidious, the earliest sign of PSP is most commonly unexplained falls, followed by deteriorating postural instability. The second most common clinical manifestation of PSP is dysarthria (Litvan et al., 1996). Approximately 60% of cases present with gait difficulty, and bradykinesia similar to PD is also frequent (Golbe, 2014a; Litvan et al., 1996). The hallmark of PSP is supranuclear vertical gaze palsy and horizontal gaze may be affected later on in the disease course (Liscic et al., 2013; Litvan et al., 1996). Cognitive changes and eventually dementia can appear later on in the disease progression (Kobylecki et al., 2015; Litvan et al., 1996). Misdiagnosis is a common problem in PSP and up to 50% of cases are misdiagnosed or undiagnosed. The most common misdiagnosis of PSP is PD (Osaki et al., 2004). 10

27 1.2.2 Neuropathology of Progressive Supranuclear Palsy Tau aggregates are the main pathology in PSP; therefore, it is classified as a primary tauopathy. Tau inclusions can accumulate in both neurons and glial cells. In neurons, NFTs present as straight filaments and are formed primarily by 4R tau inclusions (Arai et al., 2001; Dickson, 1999; Zhukareva et al., 2006). Aggregation also occurs in glial cells in astrocytes these are termed tufted astrocytes and in oligodendrocytes coiled bodies. The regional distribution of lesions is not clustered and generally presents as random dispersions (R. A. Armstrong & Cairns, 2013; Tawanna & Ramsden, 2001). The majority of NFT tau pathology is situated in subcortical regions, primarily in the basal ganglia, with involvement specifically in the striatum, caudate, putamen, subthalamic nucleus, substantia nigra and also dentate nucleus of the cerebellum (R. A. Armstrong & Cairns, 2013; Togo & Dickson, 2002; Wray, Saxton, Anderton, & Hanger, 2008). Brainstem areas commonly implicated include the pontine nuclei and tegmentum, the locus coeruleus, oculomotor nuclei, the periaqueductal grey matter and the superior colliculus (Dickson, 1999). Tau pathology may also be present in the premotor and motor cortices (R. A. Armstrong & Cairns, 2013). Tufted astrocytes are primarily seen in the frontal lobe, putamen and cerebellar white matter, and may have cognitive implications (Tawanna & Ramsden, 2001; Zhukareva et al., 2006). Williams and colleagues (2007) established a scoring system (12 points) that grades the severity of PSP based on tau pathology in the substantia nigra, caudate, and dentate nucleus. See Figure

28 Atrophy primarily occurs in the majorly affected midbrain and subcortical regions. Midbrain atrophy is very common and can be accompanied by atrophy of the aqueduct of Sylvius (Dickson, 1999; Tawanna & Ramsden, 2001). The subthamalic nucleus, globus pallidus and the ventrolateral substantia nigra present with the brunt of neuronal loss. Depigmentation is seen in the substantia nigra and locus coeruleus (Dickson, 1999). The dentate nucleus, superior cerebellar peduncle, and brainstem tegmentum may also appear smaller. Up to 40% of the striatum has been reported to undergo neurodegeneration. Cortical atrophy may also occur, specifically in the frontal cortex spanning to the precentral gyrus (R. A. Armstrong & Cairns, 2013; Dickson, 1999; Tawanna & Ramsden, 2001). 12

29 Figure 1-1. Scoring of neuropathological accumulation of tau in PSP. Legend (A). Scoring 0-1 (B): minor involvement of the basal ganglia and pre-motor cortex. Scores

30 (C): increased pathology in the basal ganglia, pontine nuclei, and dentate nucleus. Scores 4-5 (D): severity increased in basal ganglia and dentate nucleus, increased involvement in white matter of the cerebellum and frontal and parietal lobes. Scores 6-7 (E): moderately severe involvement in the substantia nigra, globus pallidus, subthalamic nucleus, pontine nuclei and white matter of the cerebellum, increased involvement in frontal and parietal lobes. Scores >7 (F): severe pathology in subthalamic nucleus, increased involvement in neocortical regions. Reproduced with permission from (Williams et al., 2007) Treatment & Symptom Management of Progressive Supranuclear Palsy Currently there are no therapies for PSP symptom management, or treatments to reduce tau burden or prevent further accumulation. Levodopa and other pharmaceuticals that act on the dopaminergic system, as well as the cholinergic and GABAergic systems, historically do not generally aid with PSP symptoms (Koros & Stamelou, 2016). In attempt to address tau burden in tauopathies, inhibition of tau phosphorylation has been of large interest recently (Noble et al., 2013). Tideglusib, a glycogen synthase kinase 3 (GSK 3) inhibitor, was tested in PSP in an effort to reduce tau hyperphosphorylation. There was no significant difference between the tideglusib groups and the placebo group in terms of the PSP rating scale; however, patients in the drug groups showed slowed atrophy (Hoglinger et al., 2014). Other efforts have been made in animal studies to develop 14

31 microtubule stabilizing therapeutics, drugs targeting heat shock proteins, and tau aggregation inhibitors (Gerson, Castillo-Carranza, & Kayed, 2014; Schroeder, Joly- Amado, Gordon, & Morgan, 2015). Investigations into active and passive tau immunization are also currently being formed on animal models (Schroeder et al., 2015). 1.3 Positron Emission Tomography Positron emission tomography (PET) is a non-invasive imaging modality that uses radioligands (or radiotracers) to detect targets in vivo. Ligands are typically labeled with either fluorine-18 or carbon-11, radioisotopes with relatively short half-lives (110 minutes and 20 minutes, respectively). Following a venous injection, the radiotracer enters the bloodstream and is transported to the brain (or any other target organ/region) where it interacts with its target protein. The radioligand s unstable nucleus undergoes positron decay and when the emitted positron collides with a neighbouring electron an annihilation event occurs. The result of an annihilation event is the release of two photons of equal energy (511 kev) in opposite directions (exactly 180º from one another). These photons are detected by scintillation detectors that line the PET scanner in a cylindrical configuration. Detection of two photons within a coincidence window (~ 6 nsec) results in a coincidence event that is plotted on a sinogram. After the scan is complete, the sinogram undergoes scatter correction and attenuation correction, using the brief transmission prior 15

32 to the emission scan. The PET image is reconstructed based on the sinogram plots and radioligand dead time and decay correction should be performed Positron Emission Tomography Tau Radiotracers Brain PET radiotracers must meet many of the same requirements of drugs with targets in the brain, in order to passively cross the blood brain barrier (BBB) and be effective as neuroimaging agents. Lipophilicity is one of the most important factors that determines whether or not a radioligand will successfully cross the BBB. A LogP of approximately is optimal because the compound is lipophilic enough to rapidly cross the bilipid membrane, but not so lipophilic that the compound binds to plasma proteins, P- glycoprotein, or other nonspecific targets (Pike, 2010; Shah & Catafau, 2014; Villemagne & Okamura, 2014). Another condition for passively crossing the BBB is a low molecular weight (<500Da) (Pike, 2010). An optimal radiotracer will have rapid uptake and washout from the brain. For the majority of effective radioligands 5% of the injected dose is taken up into the brain 2 to 5 minute post-injection (Villemagne & Okamura, 2014). It is important that the metabolism of a radiotracer occur outside the brain and that the metabolites are less lipophilic than the parent compound to avoid nonspecific binding in the brain and thus, greater noise (Pike, 2010; Shah & Catafau, 2014). 16

33 Tau as a radioligand target poses many challenges. It is an intracellular protein; therefore, the radiotracer must not only be capable of passing through the BBB, but also neuronal and glial plasma membranes (Shah & Catafau, 2014; Villemagne & Okamura, 2014). Due to the various concentrations of tau in different brain regions, an ideal radiotracer would have very high affinity for tau in order to limit the amount of radioactivity that must be injected into the subject. The tau radioligand should have a very high specificity for tau because it is often co-localized with other proteins that have β-sheet secondary structures (Shah & Catafau, 2014; Villemagne & Okamura, 2014). Due to the involvement of pathological tau in glial cells, the radioligand should also have very low nonspecific binding in white matter (Shah & Catafau, 2014) FDDNP The radiotracer 18 F-(2-(1-{6-{(2-[ 18 F]fluoroethyl)(methyl)amino]-2- naphthyl}ethylidene)malononitrile) ([ 18 F]FDDNP) was developed as the first non-invasive biomarker to image NFTs and Aβ plaques in the brains of AD patients (Shoghi-Jadid et al., 2002). The first human testing of [ 18 F]FDDNP occurred in 9 AD subjects (7 probable AD, 2 possible AD) and 16 neurologically healthy controls. Participants also underwent a 2-[ 18 F]fluoro-2-deoxy-D-glucose (FDG) PET scan to test for glucose metabolism across brain regions. ROIs were delineated manually and relative residence time (RRT) was calculated for each ROI. This study demonstrated that the [ 18 F]FDDNP parent compound crossed the BBB quite rapidly and appeared to be metabolized peripherally with no signs 17

34 of metabolites crossing the BBB (Shoghi-Jadid et al., 2002). [ 18 F]FDDNP binding appeared to be inversely correlated with glucose metabolism. Using autoradiographic methods, it was determined that [ 18 F]FDDNP binding in vivo was consistent with immunohistochemical staining of NFTs and Aβ. Interestingly, [ 18 F]FDDNP was inversely correlated with participants cognitive scores (Shoghi-Jadid et al., 2002). These results are consistent with previous findings that tau is associated with cognitive decline (Ludolph et al., 2009). Although this was a small sample size and kinetic modeling is required to validate the distribution of [ 18 F]FDDNP, this preliminary study demonstrated the radiotracer s ability to label NFTs and Aβ in vivo in AD. A later [ 18 F]FDDNP study included AD patients (n=25), MCI patients (n=28), and healthy control participants (n=30) (Small et al., 2006). MCI patients exhibited significantly increased uptake of [ 18 F]FDDNP compared to healthy controls, while uptake in AD patients was significantly higher than both MCI and healthy controls (Small et al., 2006). Areas with particularly high uptake included the frontal lobe, temporal lobe, parietal lobe, and posterior cingulate. These regions are known to have tangle and plaque involvement in AD. A neuropathological evaluation was performed on one participant who passed away 14 months following baseline testing. It was found that areas high in [ 18 F]FDDNP binding were associated with high NFTs and plaque immunoreactivity. Particularly the hippocampus and entorhinal cortex, which were high in NFT concentration, displayed high [ 18 F]FDDNP binding. This study supports previous findings in human [ 18 F]FDDNP studies and suggests that this radiotracer is capable of distinguishing AD, MCI, and healthy controls (Small et al., 2006). 18

35 A review written by Shin and colleagues (2011) outlines the utility and limitations of imaging AD using [ 18 F]FDDNP. This radiotracer is capable of labelling tau inclusion in MAPT transgenic mice, expressing human pathological tau. When tested in MCI and AD subjects, [ 18 F]FDDNP signal was correlated with both Mini Mental State Examination (MMSE) scores and AD brain pathology progression described by Braak & Braak (1993) (Shin, Kepe, Barrio, & Small, 2011). Furthermore, when AD patients were imaged with [ 18 F]FDDNP and [ 11 C]PIB, higher binding of [ 18 F]FDDNP but not [ 11 C]PIB was seen in the medial temporal cortex, a region associated with memory and cognitive impairment in AD. These findings support the theory that tau aggregate accumulation is associated with cognitive decline. Patients with frontotemporal dementia also showed increased uptake of [ 18 F-FDDNP. In PSP, [ 18 F]-FDDNP binding was seen in the caudate, putamen, thalamus, and cortical regions as disease severity increased (Shin et al., 2011). The most recent [ 18 F]FDDNP study performed with PSP recruited 15 PSP, 9 PD, and 5 healthy age-matched controls (Kepe et al., 2013). All participants were imaged for 65 minutes following a [ 18 F]FDDNP injection. Results demonstrated significantly increased binding of [ 18 F]FDDNP in the midbrain, subthalamic region, and cerebellar white matter in PSP patients. Additionally, patients with more severe PSP had greater [ 18 F]FDDNP binding in cortical regions (Kepe et al., 2013). These subcortical regions are consistent with PSP pathology and cortical involvement is indicative of PSP progression. The primary concern with imaging PSP using [ 18 F]FDDNP is its lack of specificity for tau over other proteins with β-sheet structures. 19

36 THK A group of tau radiotracers have been developed from arylquinone derivatives. The first in its group to be tested was [ 18 F]THK-523. Fodero-Tavoletti and colleagues (2011) investigated the appropriateness of [ 18 F]THK-523 as a PET radiotracer and its ability to detect tau pathology with high affinity and selectivity. Favourable properties of [ 18 F]THK-523 as a radiotracer include low molecular weight, capability of being labeled with 18 F at a high specific radioactivity, and an acceptable lipophilicity to cross the BBB. In vitro binding studies using AD and age-matched control brain slices and in vivo transgenic mice studies demonstrated that [ 18 F]THK-523 selectively bound tau deposits over Aβ plaques (Fodero-Tavoletti et al., 2011; Harada et al., 2013). However, testing in non-ad tauopathy brain slices (PiD, PSP, and CBD) revealed no THK523 fluorescence (Fodero-Tavoletti & Furumoto, 2014). Lack of binding in these straight filament tauopathies may be due to a specificity of THK-523 for PHF tau conformation. Two derivatives of THK-523, THK-5105 and THK-5117, were optimized for PET use and tested in vitro in AD brain slices (Okamura et al., 2013). Fluorescent tissue staining and autoradiography revealed that THK-5105 and THK5117 labeling of NFTs in the hippocampus was consistent with immunohistochemistry results. There were significantly different binding patterns of [ 18 F]THK-5105 and [ 18 F]THK-5117 compared to [ 11 C]PIB, suggesting that the arylquinone derivatives do not exhibit Aβ plaque binding (Okamura et al., 2013). 20

37 In a clinical study the uptakes of [ 18 F]THK5117 and [ 11 C]PIB were compared in 8 AD patients and 6 age-matched controls (Harada et al., 2015). SUVRs were calculated for both radiotracers using the cerebellar cortex as a reference region. Regional distribution of [ 18 F]THK-5117 differed from that of [ 11 C]PIB, indicating that [ 18 F]THK-5117 binding does not reflect Aβ localization (Harada et al., 2015). The [ 18 F]THK-5117 SUVRs of the temporal cortex were significantly higher in AD patients compared to healthy controls. As previously mentioned, high pathological tau load in the temporal cortex is associated with cognitive decline. [ 18 F]THK-5117 demonstrated excellent pharmacokinetics, with rapid entry into the brain and reaching a plateau in healthy controls around 50 minutes postinjection. However, there was substantial retention in the subcortical white matter, which may represent off-target binding (Harada et al., 2015). A longitudinal [ 18 F]THK-5117 study was performed by Ishiki and colleagues (2015) to track pathological tau in 5 cases of AD. Patients were diagnosed based on the National Institute of Neurological and Communicative Disorders and Stroke, and the AD Related Disorders Association s criteria. All AD patients and 5 age-matched controls underwent a [ 11 C]PIB scan to test for Aβ deposits. Healthy controls did not show plaques on their PIB scan. A following [ 18 F]THK-5117 scan was performed and volumes of interest (VOIs) were delineated automatically using PMOD software (Ishiki et al., 2015). Baseline scans revealed significantly higher SUVRs in the temporal lobes of AD patients compared to healthy controls. Additionally, regional distribution of [ 18 F]THK-5117 in AD patients was consistent with previous post-mortem findings. Annual changes in the retention of [ 18 F]THK-5117 in AD brains were seen in the middle and inferior temporal gyri and the 21

38 fusiform gyrus, suggesting that tau pathology originates in the medial temporal cortex and accumulates laterally (Ishiki et al., 2015) PBB3 Phenyl/pyridinyl-butadienyl-benzothiazoles/benzothiazoliums (PBBs) are a group of compounds developed for tau labeling based on fluorescent screening for β-sheet binding capability (Maruyama et al., 2013). Testing in the brain stem of PS19 mice (4R isoform mutation) demonstrated the ability of PBBs to bind tau positive NFTs. Furthermore, fluorescent microscopy confirmed the high affinity of PBBs for tau aggregates (Maruyama et al., 2013). Additional optimization yielded [ 11 C]PBB3 for PET use. Using PS19 mice and micropet [ 11 C]PBB3 rapidly crossed the BBB and bound tau inclusions. Retention was significantly greater in PS19 mice compared to age-matched control mice (Maruyama et al., 2013). The first human testing of [ 11 C]PBB3 was measured in comparison to [ 11 C]PIB signal. The SUVR retention patterns of [ 11 C]PBB3 differed significantly from that of [ 11 C]PIB, suggesting that [ 11 C]PBB3 binds tau selectively over Aβ (Maruyama et al., 2013). Accumulation of [ 11 C]PBB3 was seen in the medial and lateral temporal cortices, and the frontal cortex consistent with the Braak Staging Theory (Braak et al., 1993; Maruyama et al., 2013). When tested in a corticobasal syndrome patient [ 11 C]PBB3 retention was high in the neocortex and subcortical structures (Maruyama et al., 2013). 22

39 A recent clinical trial was performed on 7 AD patients and 7 healthy control participants (Kimura et al., 2015). Participants underwent both [ 11 C]PBB3 and [ 11 C]PIB scans on the same day. All AD subjects were Aβ positive and all healthy controls were Aβ negative. Arterial samples were taken during the [ 11 C]PBB3 scans which revealed rapid metabolism of the parent compound to a more lipophilic metabolite that has been shown to cross the BBB in mouse brains (Kimura et al., 2015). As previously mentioned, entry of a metabolite into the brain decreases the signal to noise ratio T808 In attempt to develop a PET radiotracer designed to bind PHF tau, benzo[4,5]imidazole[1,2-a]pyrimidines fluorescent compounds were screened and T557 was a hit for PHF tau binding (Zhang et al., 2012). When tested against brain slices with tau pathology and Aβ plaques, T557 selectively bound to tau over Aβ. However, this compound demonstrated poor brain uptake; therefore, T808 was developed through optimization (Zhang et al., 2012). Autoradiography on 33 AD brain tissues (pathologically confirmed diagnosis) and 12 non-ad healthy control cases was performed with [ 18 F]T808. Strong signals were observed in the tau-rich/aβ-rich regions of the AD brains, but not in the tau-poor/aβ-rich regions of the AD brains, nor the tau-poor/aβ-poor tissues in of the non-ad cases (Zhang et al., 2012). This demonstrated selectivity of [ 18 F]T808 for tau over Aβ. [ 18 F]T808 has an appropriate binding affinity (Kd=22nM) to PHF tau and its 23

40 pharmacokinetics are ideal for crossing the BBB. Testing in mouse models revealed fast uptake and washout from the brain. When tested against 72 of the most common CNS targets, [ 18 F]T808 showed no inhibition at clinical concentrations, though there was mild inhibition of the norepinephrine and monoamine transporter (Zhang et al., 2012). The first human testing of [ 18 F]T808 was performed in 8 AD patients and 3 healthy controls (Chien et al., 2014). There was rapid uptake and distribution throughout the brain. SUVRs in reference to the cerebellum were calculated for each VOI. These VOIs included the frontal, parietal, lateral temporal, mesial temporal, and occipital lobes, the cerebellum, the genu region of the white matter, and the approximate area of the hippocampus (Chien et al., 2014). Due to the fast washout of the radiotracer from the cerebellum, SUVRs were analyzed at both minutes and minutes. While the values from minutes static frames were higher, the trends across the participants were similar for both time frames. In all healthy control cases, SUVRs of the cortical regions were low. Mild AD subjects had mild to moderate retention of [ 18 F]T808 in the hippocampal area, mesial temporal, lateral temporal, and parietal lobes. In the minute time frame there was also a moderate increase in frontal lobe retention. The SUVRs in the parietal, lateral temporal, and frontal lobes, and hippocampal areas were increased in the moderate to severe AD patients (Chien et al., 2014). These distribution patterns are reflective of PHFtau accumulation in AD (Braak et al., 1993; Chien et al., 2014). 24

41 [ 18 F]AV Development & Preclinical Testing Another benzo[4,5]imidazole[1,2-a]pyrimidine-derived PET radioligand is [ 18 F]AV-1451 ([ 18 F]T807). The first preclinical testing of [ 18 F]AV-1451 was published in 2013 by Xia and colleagues. Screening of fluorescent compounds against PHF tau in AD brains yielded T726, which demonstrated selective binding to tau over Aβ plaques. Further optimization of T726 produced the non-fluorescent analog T807 (AV-1451) (Xia et al., 2013). Autoradiography was performed on frontal brain sections of 8 PHF-tau/Aβ rich slices, 9 PHF-tau poor/aβ rich slices, and 9 PHF-tau/Aβ poor slices in conjunction with immunohistochemistry staining to determine [ 18 F]AV-1451 binding specificity. Significant signal was only seen in the PHF-tau rich slices, indicating a >25 fold selective binding of [ 18 F]AV-1451 to tau over Aβ plaques (Xia et al., 2013). Pharmacokinetic evaluation revealed promising properties to cross the BBB. [ 18 F]AV-1451 has a low molecular weight (262.1g/mol), is sufficiently lipophilic (logpoct of 3.4, logp of 1.67), and possesses a high specific radioactivity of 9.36 Ci/µmol. Competitive binding assays with 72 of the most common CNS targets revealed minimal non-specific binding, with the exception of the norepinephrine transporter and monoamine transporter. When tested in 6 mice, [ 18 F]AV-1451 demonstrated rapid uptake into the brain and fast washout with some retention in the bone, thought to be due to defluorination. The majority of [ 18 F]AV-1451 was distributed to the kidneys and bladder, although there was a smaller portion delivered 25

42 to the liver. While four metabolites of the parent compound were noted, there were no detectable metabolites in the brain (Xia et al., 2013). Overall preclinical evaluation of this radiotracer yielded promising results that [ 18 F]AV-1451 would effectively bind PHF-tau in human brains. A post-mortem study by Marquie and colleagues (2015) performed further autoradiography screenings and included brain slices from an array of tauopathies. Brain slices from 3 AD, 3 PiD, 3 PSP, 2 CBD, and 2 control cases were tested. A phosphor screen [ 18 F]AV-1451 autoradiography indicated strong signals in the entorhinal, frontal, temporal, parietal, and occipital cortices of AD brain slices with tau positive NFTs. Nonradiolabeled AV-1451 was able to competitively inhibited the [ 18 F]AV-1451 signals (Marquie et al., 2015). As expected, there was no signal in the control brain slices, with the exception of the substantia nigra, which was confirmed to be neuromelanin containingcells. Non-PHF tauopathies (PiD, PSP, CBD, and transgenic mouse model) exhibited no signal of [ 18 F]AV-1451 in brain regions afflicted with their respective tau pathology. However, in the PSP cases there was a signal in the entorhinal cortex, suggesting that [ 18 F]AV-1451 was labeling age-related tau accumulation according the Braak Staging Theory (Braak et al., 1993; Marquie et al., 2015). Nuclear emulsion [ 18 F]AV-1451 autoradiography results were also consistent with these findings. Silver grains accumulated around both intracellular and extracellular PHF-tau in brain slices of AD, but did not accumulate to any significant degree around Pick bodies of a PiD case, tufted astrocytes in PSP, nor coiled bodies of CBD brain slices. It was concluded from this preclinical study that [ 18 F]AV-1451 bound with selectivity to PHF-tau in AD and age- 26

43 related tau accumulation, but did not bind to any significant degree to straight filaments of tau that are present in non-ad tauopathies (PiD, PSP, and CBD), and aggregates containing β-amyloid and α-synuclein. Additional binding to neuromelanin-containing cells, such as those in the substantia nigra pars compacta was observed (Marquie et al., 2015). Another post-mortem examination of [ 18 F]AV-1451 in AD (n=5), PSP (n=6), PiD (n=5), and CBD (n=4) was performed by Sander and colleagues (2016). [ 18 F]AV-1451 signal was consistent with immunohistochemistry staining of tau in AD and PiD slices. However, there was no substantial signal in PSP and CBD brain slices, consistent with the previous post-mortem study (Sander et al., 2016) Human Testing [ 18 F]AV-1451 was first tested clinically in 2 AD, 1 MCI, and 3 healthy controls (Chien et al., 2013). The MMSE was used to test for cognitive ability in all participants. All 3 healthy controls scored 28 and above, the MCI patient had a score of 26, and the AD patients were classified as mild and severe based on their scores of 21 and 7, respectively. The MCI and AD patients underwent a [ 18 F]Florbetapir (Amyvid ) PET scan to establish the Aβ positive pathology (Chien et al., 2013). All participants underwent a dynamic PET scan from 0-60min following the injection of [ 18 F]AV-1451 and a subsequent min static scan was also acquired. The mean activity injected for all participants was 27

44 10.11mCi, and peak activity was reached at 4-10min post injection in all areas of the brain. VOIs were drawn manually on the CT scan acquired on the PET/CT scanner prior to the PET scan. VOIs included the frontal, parietal, lateral temporal, mesial temporal, and occipital lobes, the cerebellum, the genu region of the white matter, and the approximate area of the hippocampus (Chien et al., 2013). Mean SUVs and SUVRs in reference to the cerebellum were calculated for each VOI using the min static scan. In the MCI patient SUVR was increased in the hippocampal area, and the parietal, mesial temporal, and lateral temporal lobes. The mild AD patient had the highest SUVR in the lateral temporal lobe, and additionally had high SUVRs in the mesial temporal, parietal, frontal and occipital lobes, and the hippocampal area. The VOIs with the most elevated SUVRs in the severe AD patient were the parietal and lateral temporal lobes, followed by the mesial temporal, frontal, and occipital lobes, and the hippocampal area (Chien et al., 2013). The severe AD patient had a significantly higher SUVR in the parietal lobe compared to the mild AD and the MCI patients. The pattern of [ 18 F]AV-1451 uptake in the severe AD patient reflected the pathological tau deposition outline in Braak stages V- VI (Braak et al., 1993; Chien et al., 2013). The mild AD patient had significantly higher SUVR in the lateral temporal, mesial temporal lobes, and the hippocampal area compared to the MCI patient and healthy controls. This retention distribution is characterized by Braak stages III-IV (Braak et al., 1993; Chien et al., 2013). Interestingly, the oldest healthy control (67 years old) had a higher SUV in the cerebellum, as well as the cortical VOIs compared to the two younger healthy controls (56 and 58 years old). This is thought to be due a an age-related accumulation of PHF tau (Chien et al., 2013). 28

45 Figure 1-2. SUVR images from static scan ( minutes). Little to no retention is seen in the healthy control. Mild retention in MCI and moderate retention in mild AD, with increased cortical retention is seen in the severe AD patient. Reproduced with permission from (Chien et al., 2013). 29

46 Figure 1-3. VOI SUVRs ( minutes) for each subject. Cortical uptake differs from AD to MCI and HC. Reproduced with permission from (Chien et al., 2013). 30

47 Another human study was performed with AD patients (44), MCI patients (87) and healthy older adults (42) to test whether [ 18 F]AV-1451 binding reflected the Braak histopathological staging (Braak et al., 1993; Schwarz et al., 2016). The older control group was defined as individuals 50 years old. Distribution of [ 18 F]AV-1451 SUVR ( min) in all groups was in agreement with Braak staging patterns and previous neuropathological studies. Of all the subjects, only 7 presented with signals that did not conform to the neuropathological staging of tau deposition (Schwarz et al., 2016). The post-mortem study performed by Marquie and colleagues (2015) prompted a human study that aimed to further investigate the off-target binding of [ 18 F]AV-1451 to neuromelanin-containing cells in the substantia nigra (Hansen et al., 2016). PD is neuropathologically marked by the depigmentation of the substantia nigra due to denervation of dopaminergic neurons. PD patients (17) and healthy controls (16) underwent a [ 18 F]AV-1451 PET scan. There was a 30% decrease in [ 18 F]AV-1451 midbrain signal in PD patients compared to control, which is consistent with the postmortem literature that a 30% reduction of dopaminergic neurons in the substantia nigra pars compacta occurs before symptom presentation in PD. Additionally, there was no signal in the nigra of pig and rat brains, because the nigra of non-primate species do not contain neuromelanin (Hansen et al., 2016). [ 18 F]AV-1451 is the first PET radiotracer to bind neuromelanin and has the potential to measure depigmentation of the substantia nigra in vivo in PD and other parkinsonian syndromes. 31

48 Significance PSP is difficult to diagnose clinically, especially in the early stages due to its clinical heterogeneity. Symptom overlap with other parkinsonian disorders, namely PD, and tauopathies contribute to the misdiagnosis of PSP. A misdiagnosis could lead to an inaccurate prognosis and prescription of inappropriate medication. [ 18 F]AV-1451 could serve as a new biomarker for the detection of PSP. Additionally, PET is an excellent tool for monitoring disease progression and severity. New treatment options that decrease tau burden can be assessed using the PET tracer to determine their molecular efficacy, and reduce the large sample size and funding required by most clinical trials. There are many other tauopathies that could also benefit from this radiotracer; therefore, validation of [ 18 F]AV-1451 in PSP patients offers diverse applications. 32

49 2.0 AIM & HYPOTHESIS There are currently no PET radiotracers available to image pathological tau in PSP patients in a clinical setting. While many tau radiotracers are still in the clinical trial stages, most these studies have focused on AD, as this is the most common tauopathy. PSP patients must be included in trials testing tau imaging agents in order to establish an appropriate radiotracer for this patient population. Individuals with PSP could benefit from the diagnostic and prognostic potential of a tau radiotracer due to the insidious and aggressive nature of this neurodegenerative movement disorder. In vitro and in vivo investigations using [ 18 F]AV-1451 in AD and MCI patients have yielded promising preliminary results, demonstrating its capability of to image tau pathology in these disorders. However, further testing must be done to determine this radiotracer s efficacy in other tauopathies. AD has specific tau pathology (PHF tau) that differs from PSP and other straight-filament tauopathies. Therefore, tau radiotracers that may effectively image tau in AD are not necessarily useful for imaging tau pathology in PSP. 2.1 Aim Determine whether [ 18 F]AV-1451 is an appropriate PET radiotracer for imaging pathological tau in PSP using PET and MRI technology. 33

50 2.2 Hypothesis [ 18 F]AV-1451 retention will be significantly elevated in the cortical and subcortical brain regions of PSP patients, compared to PD and healthy controls. Specifically, elevation in brain regions including the substantia nigra, caudate, putamen, globus pallidus, dentate nucleus, frontal lobe, and parietal lobe are of interest to differentiate PSP from PD and HC. 34

51 3.0 METHODS 3.1 Participants and Experimental Design Twenty-two participants were recruited for this study: 6 PSP patients, 6 PD patients, and 10 healthy age-matched controls. PSP diagnoses were confirmed using the National Institute for Neurological Disorder and Society (NNDS-SPSP) criteria. PSP disease severity was measured using the Progressive Supranuclear Palsy Rating Scale (PSPRS) and the Unified Parkinson s Disease Rating Scale Part III (UPDRS-III). Patients with PD met the UK Parkinson s Disease Society Brain Bank Criteria and were also assessed with the UPDRS-III to measure motor symptom severity. The Unified Parkinson s Disease Rating Scale (UPDRS) was originally developed during the 1980 s to assess the severity of motor and non-motor symptoms associated with PD. In 2001, the Movement Disorders Society (MDS) evaluated, critiqued, and changed aspects of the scale, renaming it the MDS-UPDRS (Goetz et al., 2008). Specifically, the MDS- UPDRS III ( motor examination ) contains 18 items, some divided into left, right, and other distributions for a total of 33 scores. Motor aspects tested include: finger taps, hand movement, pronation/supination, toe tapping, leg agility, arising from chair, gait, freezing of gait, postural stability, posture, global spontaneity of movement, postural tremor of hands, kinetic tremor of hands, rest tremor amplitude, and constancy of rest tremor (Goetz et al., 2008). Detailed instructions have also been included in the scale to ensure uniform 35

52 administration. When tested on 877 PD patients with 69 raters the new MDS-UPDRS III had high internal consistency (Goetz et al., 2008). The UPDRS III was tested on 175 PSP patients and while scores correlated with Hoehn and Yahr staging, there were many PSP clinical features missing from the scale (Cubo et al., 2000). In order to quantitatively assess PSP severity, Golbe and Ohman-Strickland (2007) developed the PSP Rating Scale (PSPRS). This is a 6 part scale: history, mentation, bulbar, ocular motor, limb motor, and gait and midline. There are 28 items, scored either 0-2 or 0-4, depending on the item (Golbe & Ohman-Strickland, 2007). The PSPRS demonstrated a general increase in scores over time as PSP patient disability increased. This rating scale also had high reliability when used by a clinician (Golbe & Ohman- Strickland, 2007). The National Institute for Neurological Disorder and Society for PSP developed criteria (NNDS-SPSP) for diagnosing possible, probable, and definite PSP (Litvan et al., 1996). All diagnoses required the disease onset after 40 years of age, a gradual progression of the disorder, and no evidence of another disorder that could explain clinical symptoms. Possible PSP necessitates vertical gaze palsy, OR slowing of vertical saccades and postural instability accompanied by falls. A possible PSP diagnosis using these criteria has a sensitivity of 83% and specificity of 93%. Probable PSP requires vertical gaze palsy, AND slowing of vertical saccades and postural instability accompanied by falls. The probable PSP criteria results in a highly specific diagnoses (100%); however, it has a relatively low sensitivity (83%). Probable PSP criteria are appropriate for diagnoses in 36

53 research studies when other disorders must be excluded. A definite PSP diagnosis can only be obtained with a clinically probable or possible PSP diagnosis AND histopathological evidence (Litvan et al., 1996). The Montreal Cognitive Assessment (MoCA) was used to score the general cognitive abilities of all participants. In order to assess subjects depression level two weeks prior and including the day of the screening, the Beck Depression Inventory (BDI) was administered to all those enrolled in the study. Exclusion criteria for all participants are as follows: no other history of neurodegenerative disorder(s), no alcohol or drug dependency/abuse, unable or unwilling to undergo PET and/or MRI scan. Patients on parkinsonian medication were required to undergo a 12-hour withdraw prior to their PET scan. There was no withdrawal from psychotropic drugs. Additionally, control subjects must not have had any history of neurological or psychiatric disorders. Written informed consent was obtained from all participants. This study was approved by the Research Ethics Board at the Centre for Addiction and Mental Health, University of Toronto. In general, participants underwent PET and MRI scans on separate visits to limit stress and fatigue, with the exception of one HC and one PD patient who had the MRI and PET scans on the same day. No subjects were excluded due to any structural findings on the MRI. 37

54 3.2 Radiosynthesis of [ 18 F]AV-1451 Clinical doses of [ 18 F]AV-1451 were produced by Centre for Addiction and Mental Health (Toronto, ON). [ 18 F]fluoride was produced using the cyclotron on site. The labeling precursor, T807 (7-(6-fluoropyridin-3-yl)-5H-pyridol[4,3-b]indole), was radiolabeled with 18 F in the presence of K222, K2CO3 and DMSO at 130ºC for 10 minutes. The mixture was then cooled to 50ºC. A water wash was performed to get rid of the DMSO, any excess [ 18 F]fluoride, and impurities. The reaction mixture was transferred to an HPLC loop and placed onto a semipreparative column (X-select HSS T3, 250 x 10.00mm, 5µ). To purify the crude reaction mixture, 18% aqueous ethanol with a ph of 2 (controlled by HCL) was used at a 5mL/min flow rate. Ultraviolet (λ=254nm) and radiochemical detectors monitored the eluent for the 22 minute retention period. The collected HPLC fraction containing the major radiochemical product was diluted with 8.4% sodium bicarbonate for injection and 20mL of sterile water for injection. The product was then washed with water to remove any salts, CH3CN, and [ 18 F]fluoride. It was then eluted with 1 ml of ethanol and 10mL of 0.9% sodium chloride for injection. The final solution was comprised of 10% ethanol in 0.9% sodium chloride. This method of [ 18 F]AV-1451 radiosynthesis had been tested to yield 14% (uncorrected) in an average 60 minutes (Shoup et al., 2013). 38

55 3.3 MRI Acquisition A 3.0 T GE Discovery MR750 MRI system (General Electric, Milwaukee, WI) was used to acquire whole-brain proton density-weighted and T1-weighted MRI scans. The MRI system was equipped with a GE Standard 8-Channel head coil. Proton density-weighted 2D images (oblique plane, 84 slices; matrix of 256 x 192; 22cm FOV; 2.0cm slice thickness; TE = Min Full; TR = 6000 ms; flip angle = 8 ) and T1-weighted 2D images (sagittal plane, 200 slices; matrix of 256 x 230; 24cm field of view (FOV); 0.9cm slice thickness; inversion time (TI) = 650ms; echo time (TE) = 3000ms; repetition time (TR) = 6700ms; flip angle = 8 ). Proton density-weighted MRIs were obtained to provide anatomical reference for ROI delineation of the PET scans. T1 weighted MRIs were used for spatial normalization of parametric PET images. 3.4 PET Acquisition PET scans were acquired on a high resolution PET/CT Siemens-Biograph HiRez XVI (Siemens Molecular Imaging Knoxville, TN, U.S.A.). Prior to the PET scan, a low dose (0.2mSv) CT scan was performed and used for attenuation correction. In order to prevent head movement during the PET scan, a thermoplastic facemask was custom-fitted to each participant and attached to a head-fixation system (Tru-Scan Imaging, Annapolis). A bolus 39

56 of [ 18 F]AV-1451 was injected into the antecubital vein via an intravenous line. Injected amount of [ 18 F]AV-1451 ranged from mci, with an average specific activity of mCi/µmol (+/ mci/µmol) and chemical purity of a least 95%. Following the injection, a 90-minute PET scan was acquired. Emission list mode data was then rebinned into a series of 3D sinograms that were corrected for attenuation and scatter. The PET acquisition from 0 to 90 minutes was binned into 28 frames (8 x 15s, 3 x 60s, 5 x 120s, 5 x 300s, 5 x 600s). Fourier rebinning was then applied to convert the 3D sinograms to 2D sinograms (Defrise et al., 1999). 2D filter back projection was used to reconstruct the 2D sinograms into image space. Images were reconstructed with a spatial resolution of 2 2 2mm (x y z). 3.5 Image Analysis Regions of Interest Analysis Analysis of brain PET images can be performed on a voxel-based method, or a region of interest (ROI)-based method. Rusjan and colleagues (2006) developed an automated program to delineate ROIs and extract time-activity curves (TACs) for these regions. The subject s MR image (T1-weighted or proton density weighted) is used to transform a brain template previously fitted to standard brain atlases (Kabani, Collins, & Evans, 1998; Talairach & Tournoux, 1988) for all subjects. The brain template contains a series of 40

57 predetermined ROIs and these ROIs are altered and refined based on the probability of grey matter voxels in the individual s MRI. The subject s MRI is then coregistered to the PET image and the refined ROIs are converted to PET space and applied to the individual s PET image (Rusjan et al., 2006). The extracted TACs following this ROI delineation were very reliable and were not subject to the intra and inter-operator variability of manual ROI delineation (Rusjan et al., 2006) Semi-quantification of Radioligand Uptake In order to determine radiotracer plasma concentration and the rate constant for entry into the brain from the plasma, a full kinetic compartmental analysis must be performed. If such data is not currently available for a radioligand, a method of semi-quantitatively measuring uptake and retention in various areas of the brain is calculating the standard uptake value (SUV). SUV over time in a target region is based on the relationship between the concentration of radioactivity in the region (Cimg, kbq/ml), the injected dose into the subject (ID, MBq), and the subject s body weight (BW, Kg). (1) SUV (t) = CPET (t) ID/BW To compare target regions, SUV can be normalized based on a reference region that does not contain any imaging target and therefore, any radiotracer binding in this region 41

58 represents nonspecific or off-target binding. Normalization using the SUV of a reference region is called the standard uptake value ratio (SUVR). (2) SUVR (t) = SUVtarget SUVreference [ 18 F]AV-1451 Analysis Standard uptake value ratios (SUVRs) were calculated in reference to the cerebellum and corpus callosum for each ROI. ROIs were delineated using ROMI software, an automated program designed by Rusjan and colleagues (2006) based on standardized atlases (Kabani et al., 1998; Talairach & Tournoux, 1988). Using Statistical Parametric Mapping software (SPM8, Welcome Department of Imaging Neuroscience, London, UK) each individual s MRI (GE 3T, proton density weighted, 1mm slice thickness) was used to non-linearly transform a standardized brain template (International Consortium for Brain Mapping/Montreal Neurological Institute 152 MRI) with predefined cortical and subcortical ROIs. The individual ROI template was then further refined using SPM8 based on grey matter probability of the segmented MRI. The refined individual ROIs were aligned and resliced using a normalized mutual information algorithm to match the individual s PET scan. Time-activity curves (TACs) were then extracted for each ROI. ROIs used for analysis were chosen based on brain regions implicated with tau pathology in PSP post mortem studies; these included the substantia nigra, striatum, caudate, 42

59 putamen, globus pallidus, and thalamus (Williams et al., 2007). All cortical regions (frontal, temporal, parietal, and occipital) were also included in the analysis. The ROI template used in ROMI did not contain the dentate nucleus; therefore, this ROI was draw manually and TACs were extracted using MarsBaR toolbox (Brett, Anton, Valabregue, & Poline, 2002). SUVs for each ROI were calculated using the TAC values. SUVR for each region was determined in reference to the cerebellum and corpus callosum, regions considered free of pathological tau in PSP (Williams et al., 2007). Based on visual inspection of the TACs, the 30 to 60 minute time point represented the pseudo-equilibrium in which the radiotracer was both bound and unbound to the tau target; therefore, SUVR values were averaged from 30 to 60 minutes post injection. Partial volume effect correction was performed in order to control for differences in radiotracer uptake due to atrophy in the patient groups (Rousset, Ma, & Evans, 1998). Parametric PET images were created for each group by calculating mean SUVR (30-60 minutes) in SPM12 (Welcome Department of Imaging Neuroscience, London, UK) for visualization purposes. 3.6 Statistical Analysis [ 18 F]AV-1451 SUV and SUVR were compared across groups using an independent samples analysis of variance (ANOVA) and Bonferroni post hoc testing in IMB SPSS Statistics 20 to measure significant differences in uptake between PSP, PD, and HC. 43

60 Linear regression was used to test whether age and MoCA scores were predictors of [ 18 F]AV-1451 uptake across the ROIs. Demographic factors were also analyzed to test for differences between groups. An ANOVA and Bonferroni post hoc testing were performed on age, MoCA scores, and BDI across all three participant groups. An independent samples t-test was performed on disease duration and UPDRS to test for differences between the PD and PSP patient groups. A threshold of p<0.05 was implemented to determine significance. 44

61 4.0 RESULTS 4.1 Participant Demographics Participant demographics are shown in table 4-1. There were 6 PSP patients, 6 PD patients, and 10 HC participants enrolled in this study. Although PSP patients were, on average, older than PD and HC subjects, there were no significant differences in age between groups. Gender ratio across all three groups also did not differ significantly. As expected, the mean MoCA scores of PSP patients were significantly lower than the mean MoCA scores of PD patients and HC (p<0.001). All PSP MoCA scores were below 26, indicating a certain level of cognitive decline. These results are consistent with the previously reported cognitive impairment in PSP and the association of tau with dementia. The mean BDI score of the PSP group was significantly greater than that of the PD group (p<0.05) and HC group (p<0.001). PD patients had, on average a higher BDI than HC (p<0.001). Elevated BDI scores in the patient groups were expected due to their physically and mentally debilitating disorders. The mean UPDRS-III score (motor examination) of PSP patients was significantly increased compared to PD patients (p<0.001), indicating that PSP patients enrolled in this study tended to have more severe motor symptoms than that of PD patients. Disease duration of PSP patients was, on average, shorter than PD disease duration; however, this difference is not significant. These scores are consistent with the aggressive nature of PSP onset. There were no significant differences in amount of [ 18 F]AV-1451 injected across the three groups. 45

62 Age in years Gender (M/F) MoCA score BDI score UPDRS-III score PSPRS score Disease duration in years Amount injected in mci PSP (n=6) PD (n=6) HC (n=10) 72.2 (6.77) (9.61) 65.9 (9.93) 2/4 3/3 2/ (2.39) a, b 28.3 (.715) 26.6 (1.51) 14.8 (2.32) b, c 11.2 (2.23) d 2.56 (2.30) 60.7 (7.58) a 26.3 (3.01) (9.81) (1.41) 5.50 (2.43) (.263) 4.72 (.261) 4.95 (.458) Table 4-1. Participant demographics. Mean values (standard deviation). a PSP is significantly different from PD at p< b PSP is significantly different from HC at p< c PSP is significantly different from PD at p<0.05. d PD is significantly different from HC at p<

63 4.2 ROI Analysis Time-Activity Curves [ 18 F]AV-1451 uptake was rapid, reaching a peak between 5 to 12 minutes for most ROIs. By the end of the 90-minute scan the cerebellum and cortical area activities were below 50 nci/cc, whereas some of the subcortical region activity remained higher. Examples of the time activity curves are present in figures 4-1 to

64 Activity (nci/cc) 350 PSP Time-Activity Curve Cerebellum Putamen Time (s) Figure 4-1. Time-activity curve of cerebellum and putamen in PSP subject 48

65 Activity (nci/cc) 300 PD Time-Activity Curve Cerebellum Putamen Time (s) Figure 4-2. Time-activity curve of cerebellum and putamen in PD subject. 49

66 Activity (nci/cc) 250 HC Time-Activity Curve Cerebellum Putamen Time (s) Figure 4-3. Time-activity curve of cerebellum and putamen in HC subject. 50

67 4.2.2 SUV Mean SUV values (30-60 minutes) for each ROI across the three groups are presented in table 4-2. The highest SUVs were found in the subcortical regions across all participant groups, specifically the striatum, caudate, and putamen. The highest SUV was in the putamen for PSP (1.93), PD (1.73), and HC (1.90). In PSP, all subcortical regions had elevated SUVs compared to cortical region SUVs. As mentioned the putamen had the highest SUV (1.93), followed by the striatum (1.86), the globus pallidus (1.81), the caudate (1.75), and the dentate nucleus (1.71). The ROIs with the most elevated SUVs in PD included the putamen (1.73), the striatum (1.72), and the caudate (1.72). HC had the highest SUVs in the putamen (1.90), the striatum (1.89), the caudate (1.96), and the substantia nigra (1.75). Cortical SUVs in PSP ranged from , in PD these values ranged from , and in HC The lowest SUV in PSP was the frontal lobe (1.30). In the PD group the lowest SUVs were found in the frontal lobe (1.31), the inferior parietal lobe (1.32), and the occipital lobe (1.32). Similarly, the ROIs with the lowest SUVs in HC were the occipital lobe (1.37) inferior parietal lobe (1.40), and the frontal lobe (1.42). In every ROI, PD had the smallest SUV compared to PSP and HC. For the HC and PSP, instead, the group with the highest SUV varied across the ROIs. The SUVs of the reference regions (e.g. the cerebellum and corpus callosum), are presented graphically in figures 4-4 and 4-6. Results for the cerebellar SUVs indicated that PSP had the most uptake (1.46), followed by HC (1.38), and PD (1.24). Although there are differences in SUV in this reference region across the three participant groups, these 51

68 variances are not significantly different (p>0.05). Within groups cerebellar SUVs there is also variance. PSP SUVs range from to 1.86, PD range is from to 1.72 and HC from to SUVs of the other reference region, the corpus callosum, also differed across PSP (1.14), PD (0.978), and HC (1.08), though not significantly (p>0.05). Within groups, the range of corpus callosum SUVs was slightly narrower than that of the cerebellar SUVs. PSP ranged from to1.47, PD from to 1.34, and HC from 0.54 to SUVs were also calculated from minutes post-injection. Values are presented in table 4-3 and figures 4-5 and 4-7. Due to washout, the values are smaller from minutes compared to the minute time frame. Neither time frame demonstrated any significant differences when comparing ROI SUVs across the three participant groups. 52

69 Frontal lobe Inferior parietal lobe Temporal lobe Occipital lobe Striatum Caudate Putamen Globus pallidus Substantia nigra Thalamus Dentate nucleus Cerebellum Corpus callosum PSP PD HC 1.30 (.260) 1.31 (.343) 1.42 (.321) 1.43 (.322) 1.32 (.336) 1.40 (.293) 1.48 (.314) 1.38 (.326) 1.44 (.297) 1.53 (.299) 1.32 (.235) 1.37 (.253) 1.86 (.381) 1.72 (.432) 1.89 (.463) 1.75 (.377) 1.72 (.453) 1.86 (.486) 1.93 (.434) 1.73 (.431) 1.90 (.453) 1.81 (.543) 1.52 (.442) 1.66 (.398) 1.54 (.292) 1.31 (.354) 1.75 (.385) 1.67 (.343) 1.51 (.345) 1.69 (.351) 1.71 (.314) 1.41 (.321) 1.58 (.364) 1.46 (.328) 1.24 (.294) 1.38 (.330) 1.14 (.240).978 (.238) 1.08 (.229) Table 4-2. Mean SUV (standard deviation) by ROI across groups from minutes. No significant differences between groups. 53

70 Frontal lobe Inferior parietal lobe Temporal lobe Occipital lobe Striatum Caudate Putamen Globus pallidus Substantia nigra Thalamus Dentate nucleus Cerebellum Corpus callosum PSP PD HC.817 (.151).819 (.242).861 (.204).927 (.223).827 (.243).861 (.184).955 (.209).852 (.233).888 (.189) 1.07 (.223).879 (.198).890 (.166) 1.17 (.262) 1.05 (.322) 1.14 (.308) 1.07 (.327) 1.06 (.380) 1.09 (.319) 1.24 (.304) 1.05 (.300) 1.16 (.304) 1.32 (.356) 1.01 (.323) 1.12 (.287) 1.14 (.338).916 (.333) 1.26 (.350) 1.07 (.245).869 (.239).983 (.227) 1.07 (.224).894 (.176).997 (.288).911 (.227).753 (.213).813 (.223).918 (.203).768 (.206).850 (.192) Table 4-3. Mean SUV (standard deviation) by ROI across groups from minutes. No significant differences between groups. 54

71 Figure 4-4. Mean SUV of the cerebellum from minutes. Figure 4-5. Mean SUV of the cerebellum from minutes. 55

72 Figure 4-6. Mean SUV of the corpus callosum from minutes. Figure 4-7. Mean SUV of the corpus callosum from minutes. 56

73 4.2.3 SUVR Cerebellum Mean SUVR values (30-60 minutes) were calculated using the cerebellum as a reference region. Parametric images (figures 4-8, 4-10, 4-12) were created for each group to visually depict the uptake in PSP, PD, and HC using SUVR. The subcortical regions, primarily the striatal area, have the strongest signals across all three groups. In PSP, there was a notable amount of atrophy. This atrophy was confirmed by visual inspection of the individual MRIs. It appeared as though there was some uptake in the temporal and occipital lobes in PSP, PD, and HC. Frontal lobe signal can be seen in PD and HC; however, there is little signal in PSP. This may be due to frontal lobe atrophy that was seen on the individual PSP MRIs. Partial volume correction was performed to address the possible effect of frontal and ventricular atrophy on the uptake in PSP patients. ROI analysis of SUVRs is presented across groups in table 4-4. SUVRs for PSP ranged from to 1.33 (frontal lobe and putamen, respectively). As with the SUVs, the subcortical regions: putamen (1.33), striatum (1.29), caudate (1.25), and globus pallidus (1.22), had the greatest SUVRs in PSP. The frontal and inferior parietal lobes (0.985) were the only two regions in the PSP group that have SUVRs less than 1, indicating less binding than the cerebellar reference region. No regions in PD and HC had SUVRs less than 1. SUVRs for PD ranged from 1.06 (frontal lobe) to 1.40 (putamen). The putamen, striatum (1.39), and caudate (1.39) presented with the highest SUVRs. Regions with the lowest SUVR in PD included the frontal lobe (1.06), inferior parietal lobe (1.07), and the substantia nigra (1.07). The frontal and inferior parietal lobes in the HC group also had the 57

74 lowest SUVRs (1.03). The highest SUVR in HC was the putamen (1.39), followed by the striatum (1.38) and the substantia nigra (1.33). There were no ROIs that had a significantly higher SUVR in PSP patients compared to PD and HC. The frontal lobe SUVR in PD (1.06) and HC (1.03) groups were significantly higher than the frontal SUVR in PSP (0.896) (p<0.05). The substantia nigra SUVR in the HC group (1.33) was also significantly increased compared to PSP (1.09) and PD (1.07) (p<0.05). In the majority of ROIs, PD SUVRs are elevated (though not significantly) compared to PSP, with the exception of the globus pallidus, the substantia nigra, and the dentate nucleus. Similarly, HC SUVRs were higher than PSP SUVRs in all regions, except the occipital lobe and the dentate nucleus. Cerebellar SUVRs were also calculated from the minute time point and presented in table 4-5 and figures 4-9, 4-11, and Almost all SUVRs were increased compared to the minutes SUVRs. This is due to the lower cerebellar SUVs. The only change in significant differences across the three groups was that HC substantia nigra SUVR was not significantly elevated compared to PSP, but remained significantly elevated compared to PD. 58

75 Frontal lobe Inferior parietal lobe Temporal lobe Occipital lobe Striatum Caudate Putamen Globus pallidus Substantia nigra Thalamus Dentate nucleus Cerebellum PSP PD HC.896 (.0707) 1.06 (.0685) a 1.03 (.0701) a.985 (.0671) 1.07 (0.553) 1.03 (.0736) 1.03 (.0544) 1.12 (.0707) 1.06 (.0610) 1.07 (.0934) 1.09 (.0784) 1.02 (.0845) 1.29 (.112) 1.39 (0.737) 1.38 (.126) 1.22 (.213) 1.39 (0.672) 1.35 (.143) 1.33 (.0747) 1.40 (.0857) 1.39 (.126) 1.25 (.206) 1.24 (.152) 1.26 (.217) 1.09 (.146) 1.07 (.114) 1.33 (.161) a,b 1.16 (.120) 1.22 (.0945) 1.24 (.0611) 1.19 (.143) 1.16 (.0395) 1.16 (.115) Table 4-4. Mean SUVR (standard deviation) from minutes using the cerebellum as a reference region. ANOVA and post hoc (Bonferroni) results. a SUVR was significantly elevated compared to PSP SUVR at p<0.05 b SUVR was significantly elevated compared to PD SUVR at p<

76 Frontal lobe Inferior parietal lobe Temporal lobe Occipital lobe Striatum Caudate Putamen Globus pallidus Substantia nigra Thalamus Dentate nucleus Cerebellum PSP PD HC.912 (.0921) 1.09 (.101) a 1.07 (.121) a 1.02 (.0583) 1.10 (.0817) 1.08 (.108) 1.06 (.0500) 1.14 (.0988) 1.11 (.0979) 1.19 (.0992) 1.19 (.135) 1.12 (.115) 1.30 (.213) 1.38 (.0830) 1.40 (.182) 1.21 (.383) 1.39 (.141) 1.34 (.194) 1.36 (.142) 1.39 (.0594) 1.43 (.184) 1.44 (.274) 1.33 (.155) 1.39 (.174) 1.27 (.271) 1.21 (.258) 1.57 (.235) b 1.18 (.0750) 1.15 (.0984) 1.22 (.0945) 1.19 (.113) 1.21 (.122) 1.22 (.180) Table 4-5. Mean SUVR (standard deviation) from minutes using the cerebellum as a reference region. ANOVA and post hoc (Bonferroni) results. a SUVR was significantly elevated compared to PSP SUVR at p<0.05 b SUVR was significantly elevated compared to PD SUVR at p<

77 Figure 4-8. Mean PSP group parametric image of mean SUVR minutes. Figure 4-9. Mean PSP group parametric image of mean SUVR minutes. 61

78 Figure Mean PD group parametric image of mean SUVR minutes. Figure Mean PD group parametric image of mean SUVR minutes. 62

79 Figure Mean HC group parametric image of mean SUVR minutes. Figure Mean HC group parametric image of mean SUVR minutes. 63

80 Figure Mean SUVRs of the frontal lobe from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the frontal lobe from minutes using the cerebellum as a reference region. 64

81 Figure Mean SUVRs of the inferior parietal lobe from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the inferior parietal lobe from minutes using the cerebellum as a reference region. 65

82 Figure Mean SUVRs of the temporal lobe from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the temporal lobe from minutes using the cerebellum as a reference region. 66

83 Figure Mean SUVRs of the occipital lobe from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the occipital lobe from minutes using the cerebellum as a reference region. 67

84 Figure Mean SUVRs of the caudate from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the caudate from minutes using the cerebellum as a reference region. 68

85 Figure Mean SUVRs of the putamen from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the putamen from minutes using the cerebellum as a reference region. 69

86 Figure Mean SUVRs of the striatum from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the striatum from minutes using the cerebellum as a reference region. 70

87 Figure Mean SUVRs of the globus pallidus from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the globus pallidus from minutes using the cerebellum as a reference region. 71

88 Figure Mean SUVRs of the substantia nigra from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the substantia nigra from minutes using the cerebellum as a reference region. 72

89 Figure Mean SUVRs of the thalamus from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the thalamus from minutes using the cerebellum as a reference region. 73

90 Figure Mean SUVRs of the dentate nucleus from minutes using the cerebellum as a reference region. Figure Mean SUVRs of the dentate nucleus from minutes using the cerebellum as a reference region. 74

91 4.2.4 Partial Volume Correction Partial volume correction was performed in order to account for any effects on uptake due to atrophy and small ROIs. Across all three groups and in every ROI the SUVR was increased due to partial volume correction. Values from minutes are presented in table 4-6. As expected, partial volume correction eliminated the significant differences between the PSP frontal lobe SUVR and that of PD and HC. Partial volume correction also greatly affected the caudate, increasing the PSP SUVR from 1.22 to 2.41, the PD SUVR from 1.39 to 2.41, and the HC SUVR from 1.35 to The significant increase in the HC SUVR of the substantia nigra compared to PSP and PD was removed. PSP had the highest substantia nigra SUVR (2.61), followed by HC (2.34), and PD (1.83). Partial volume correction did introduce one new significant different not seen in the uncorrected data. The occipital SUVR in the PD group (1.64) was significantly higher than that of the HC group (1.41). As a general trend, partial volume correction reduced the difference in uptake between the cortical regions and the subcortical regions. There was no longer the distinction of low SUVR in the cortical areas and higher SUVR in the subcortical areas. With partial volume correction, values varied from 1.32 to 2.61 in PSP, 1.34 to 2.41 in PD, and 1.24 to 2.34 in HC. The majority of PD and HC ROI SUVRs were no longer elevated compared to PSP. 75

92 The partial volume effect corrected SUVRs were also calculated from minutes and presented in table 4-7. With the exception of the thalamus, caudate, and putamen, all other ROIs had higher SUVRs compared to the SUVRs. The only significant difference between groups remains in the occipital lobe. Frontal lobe Parietal lobe Temporal lobe Occipital lobe Caudate Putamen Globus pallidus Substantia nigra Thalamus PSP PD HC 1.66 (.124) 1.65 (.184) 1.52 (.178) 1.84 (.188) 2.00 (.352) 1.74 (.155) 1.32 (.0676) 1.34 (.126) 1.24 (.108) 1.57 (.174) 1.64 (.172) a 1.41 (0.101) 2.41 (.508) 2.41 (.312) 2.32 (.603) 1.33 (.0508) 1.48 (.121) 1.34 (.188) 1.47 (.153) 1.36 (.348) 1.52 (.135) 2.61 (.694) 1.83 (.427) 2.34 (.460) 1.97 (.150) 1.85 (.148) 1.83 (.166) Table 4-6. Partial volume corrected mean SUVR (standard deviation) from minutes using the cerebellum as a reference region. a SUVR was significantly elevated compared to HC at p<

93 Frontal lobe Parietal lobe Temporal lobe Occipital lobe Caudate Putamen Globus pallidus Substantia nigra Thalamus PSP PD HC 1.68 (.215) 1.70 (.232) 1.51 (.238) 1.88 (.236) 2.04 (.402) 1.72 (.240) 1.36 (.0893) 1.38 (.178) 1.24 (.133) 1.75 (.211) 1.79 (.319) a 1.49 (0.132) 2.17 (.590) 2.39 (.301) 2.13 (.583) 1.31 (.215) 1.43 (.0723) 1.37 (.243) 1.71 (.203) 1.57 (.484) 1.47 (.278) 3.29 (1.48) 1.94 (.866) 2.59 (.597) 1.91 (.219) 1.72 (.191) 1.70 (.153) Table 4-7. Partial volume corrected mean SUVR (standard deviation) from minutes using the cerebellum as a reference region. a SUVR was significantly elevated compared to HC at p<

94 Figure Mean partial volume corrected SUVRs of the frontal lobe from minutes using the cerebellum as a reference region. Figure Mean partial volume corrected SUVRs of the frontal lobe from minutes using the cerebellum as a reference region. 78

95 Figure Mean partial volume corrected SUVRs of the parietal lobe from minutes using the cerebellum as a reference region. Figure Mean partial volume corrected SUVRs of the parietal lobe from minutes using the cerebellum as a reference region. 79

96 Figure Mean partial volume corrected SUVRs of the temporal lobe from minutes using the cerebellum as a reference region. Figure Mean partial volume corrected SUVRs of the temporal lobe from minutes using the cerebellum as a reference region. 80

97 Figure Mean partial volume corrected SUVRs of the occipital lobe from minutes using the cerebellum as a reference region. Figure Mean partial volume corrected SUVRs of the occipital lobe from minutes using the cerebellum as a reference region. 81

98 Figure Mean partial volume corrected SUVRs of the caudate from minutes using the cerebellum as a reference region. Figure Mean partial volume corrected SUVRs of the caudate from minutes using the cerebellum as a reference region. 82

99 Figure Mean partial volume corrected SUVRs of the putamen from minutes using the cerebellum as a reference region. Figure Mean partial volume corrected SUVRs of the putamen from minutes using the cerebellum as a reference region. 83

100 Figure Mean partial volume corrected SUVRs of the globus pallidus from minutes using the cerebellum as a reference region. Figure Mean partial volume corrected SUVRs of the globus pallidus from minutes using the cerebellum as a reference region. 84

101 Figure Mean partial volume corrected SUVRs of the substantia nigra from minutes using the cerebellum as a reference region. Figure Mean partial volume corrected SUVRs of the substantia nigra from minutes using the cerebellum as a reference region. 85

102 Figure Mean partial volume corrected SUVRs of the thalamus from minutes using the cerebellum as a reference region. Figure Mean partial volume corrected SUVRs of the thalamus from minutes using the cerebellum as a reference region. 86

103 4.2.5 SUVR Corpus Callosum Mean SUVR values (30-60 minutes) were calculated using the corpus callosum as a reference region and presented in table 4-8. The SUVRs using the corpus callosum were higher than the SUVRs using the cerebellum as a reference region because the corpus callosum SUVs were lower than that of the cerebellum. The highest SUVRs in the PSP group included the striatum (1.63), the putamen (1.61), the globus pallidus (1.59), and the caudate (1.56). In PD patients the putamen (1.75), the striatum (1.74, and the caudate (1.74) had the highest SUVRs. The ROIs with the most elevated SUVR were the putamen (1.74), the striatum (1.72), and the caudate (1.70) In no cases across the three groups was the SUVR below 1, indicating that all ROIs had a greater uptake of [ 18 F]AV-1451 compared to the corpus callosum. PSP and PD did not differ significantly across any SUVRs. The HC group SUVR was significantly elevated in the substantia nigra (1.62) compared to PSP (1.35) and PD (1.33) (p<0.05). 87

104 Frontal lobe Inferior parietal lobe Temporal lobe Occipital lobe Striatum Caudate Putamen Globus pallidus Substantia nigra Thalamus Dentate nucleus Cerebellum PSP PD HC 1.20 (.200) 1.33 (.0485) 1.30 (.153) 1.28 (.134) 1.34 (.0450) 1.29 (.149) 1.31 (.0887) 1.39 (.0560) 1.33 (.134) 1.30 (.0938) 1.36 (.0961) 1.27 (.161) 1.63 (.143) 1.74 (.0628) 1.72 (.234) 1.56 (.299) 1.74 (.0651) 1.70 (.266) 1.61 (.237) 1.75 (.0798) 1.74 (.225) 1.59 (.269) 1.53 (.160) 1.52 (.149) 1.35 (.122) 1.33 (.103) 1.62 (.258) a,b 1.39 (.143) 1.53 (.0903) 1.55 (.147) 1.47 (.114) 1.44 (.0787) 1.44 (.0883) Table 4-8. Mean SUVR (standard deviation) from minutes using the corpus callosum as a reference region. a SUVR was significantly elevated compared to PSP SUVR at p<0.05 b SUVR was significantly elevated compared to PD SUVR at p<

105 4.3 [ 18 F]AV-1451 and MoCA Due to the cognitive component that accompanies many tauopathies, MoCA scores were tested against [ 18 F]AV-1451 uptake across all subjects. Logistic regression results indicated that MoCA scores were not predictors of SUVRs using the cerebellum as a reference region (both partial volume effect corrected and uncorrected) in any of the ROIs analyzed. 89

106 5.0 DISCUSSION 5.1 Overview of Findings This study was designed to contribute to the ever-growing research of tau in vivo imaging. While tau is involved in many neurodegenerative disorders, PSP is one that has been underrepresented in many tau neuroimaging studies. In order to advance the current knowledge regarding tau and PSP, we investigated the PET radiotracer [ 18 F]AV-1451 in this patient population. It was predicted that [ 18 F]AV-1451 uptake in cortical and subcortical regions would be significantly greater in PSP patients compared to PD patients and HC. Relationships between [ 18 F]AV-1451 uptake and demographic information were also explored to further test the efficacy of the radiotracer Demographics Analysis of demographic characteristics revealed several significant differences between groups. PSP patients had, on average, a lower MoCA scores than PD and HC participants, suggesting decreased general cognition. All PSP patients had scores less than 26 on the MoCA scale, indicating below average cognition and potentially cognitive impairment, though further testing would have to be done for confirmation (Zadikoff et al., 2008). Significantly lower MoCA scores are consistent with previous literature, connecting 90

107 cognitive impairment with both tau pathology and as a clinical manifestation in some PSP cases (Brown et al., 2010). Lower MoCA scores were anticipated to be correlated with increased tau load, and therefore, increased [ 18 F]AV-1451 uptake; however, MoCA scores were not significant predictors of radiotracer retention. Another significant demographic difference between PSP, PD, and HC were BDI scores. The PSP group had a significantly higher mean BDI score compared to both PD and HC. PD patients had, on average, higher BDI scores than HC scores. In fact, all PSP and PD scores were above all HC scores. This indicates that PSP patients have a significantly higher incidence of depressive symptoms, as measured by mood and activity patterns in the two weeks prior to the screening date. Due to the degenerative nature of PSP and PD, it is expected that self-esteem and self-worth are affected as motor and cognitive symptoms worsen and activities of daily living become more challenging. Difficulty adjusting to one s current situation may be a factor that reduces the general mood of PSP and PD patients and lead to increased BDI scores compared to HC subjects. The final demographic characteristic that differed significantly between groups was the UPDRS-III scores of PSP and PD patients. All PSP patients scored higher than the PD patients, and the PSP group average UPDRS-III score was significantly elevated compared to the PD average score, indicating increased motor symptom severity in PSP patients. This finding is consistent with previous reports of PSP as an aggressive disease with rapid onset of motor symptoms (Liscic et al., 2013). Another finding that supports 91

108 the aggressive nature of PSP was that the mean disease duration of PSP was shorter than that of PD; however, this difference was not statistically significant [ 18 F]AV-1451 Retention Using SUVR as a semi-quantitative method of measuring [ 18 F]AV-1451 uptake in the brains of PSP, PD, and HC subjects revealed some changes in a few ROIs, though not the differences that were expected. SUVR was not significantly higher in PSP compared to PD and HC in any of the ROIs analyzed. These results are consistent with the findings of Marquie and colleagues (2015). [ 18 F]AV-1451 phosphor screen autoradiography of the frontal, parietal, temporal, and occipital cortices, hippocampus, entorhinal cortex, cingulate, basal ganglia, midbrain, and the cerebellum was performed on three pathologically confirmed PSP brains. Tau immunoreactivity was tested on adjacent brain slices and compared to the phosphor screen results. Although the brain slices analyzes displayed straight filaments comprised of tau inclusions, the phosphor screen results displayed no [ 18 F]AV-1451 signal in PSP brain regions with non-phf-tau aggregations (Marquie et al., 2015). Interestingly, there were two ROIs in which PD and HC had a significantly greater [ 18 F]AV-1451 retention compared to the PSP group. The mean frontal lobe SUVR of PD and HC using the cerebellum as a reference region was significantly higher than that of PSP. However, this statistically significant difference was not present in the corpus 92

109 callosum SUVR analysis and is not consistent with pathological studies of PSP post mortem brains; therefore, this finding may be due to one of the experimental limitations of this study (Williams et al., 2007). In both cerebellar and corpus callosum analysis, the mean HC substantia nigra SUVR was significantly higher than both PSP and PD groups. Partial volume correction removed this difference (see section 5.1.4) Reference Regions Our results using both the cerebellum and the corpus callosum as reference regions demonstrated higher SUVRs in the subcortical regions compared to cortical regions. The corpus callosum had lower SUVs in all ROIs compared to the cerebellum; therefore, SUVRs using the corpus callosum were all higher than SUVRs using the cerebellum as a reference region. The reason for disparities among the reference regions could be due to the fact that one is grey matter (cerebellum) and the other is white matter (corpus callosum); therefore, the amount of non-specific binding differs. It is for this reason that analysis was primarily performed using the cerebellum as a reference region because all ROI analyzed were also grey matter. The corpus callosum analysis served as a comparison of retention patterns, and was consistent with the cerebellar results. The mean PSP cerebellar SUV was higher than that of PD and HC groups; however, the mean PSP corpus callosum SUV was also elevated compared to PD and HC, and all three groups demonstrated an increased cerebellar SUV compared to their respective corpus callosum SUV. If the cerebellum or the corpus callosum contained any specific binding (i.e. tau 93

110 aggregates) then they would not be appropriate reference regions and the retention in PSP patients would underrepresented by their SUVR calculation. This possibility is further discussed in section 5.3 (experimental limitations) and section 7.0 (future directions) Atrophy and Partial Volume Correction Inspection of individual MRIs revealed considerable atrophy in the ventricles and frontal lobes of PSP patients compared to PD and HC MRIs. This finding is consistent with previous MRI studies involving PSP cases (Giordano et al., 2013; Massey et al., 2012). Atrophy can reduce the size of ROIs, making it difficult to segment (the grey matter, white matter, and CSF) and normalize the MRI, thereby risking the possibility of inaccurate ROI delineation. Successful segmentation and normalization was achieved for all subjects and the ROI template was carefully matched to each individual PET scan. Another concern with smaller ROIs due to atrophy is the partial volume effect. This phenomenon causes spill in and/or spill out of measured radioactivity in any given ROI. Spill out is of particular interest in the cases of small ROIs with high activity because the measured activity within the ROI spills out into neighbouring areas, making the small ROI appear to have less retention of the radiotracer than is in fact true. Due to the amount of atrophy that was seen in the PSP MRIs we believed that partial volume correction should be applied in order to control for the potential spill out effect. 94

111 Partial volume effect correction was performed on all PET scans, regardless of the subject group, to ensure objectivity and consistency between the three groups and also within groups. As expected, partial volume effect correction increased the SUVRs of some PSP ROIs. However, this correction also increased the SUVRs of the same PD and HC ROIs; therefore, there were still no significant increases in PSP SUVR compared to PD and HC. Partial volume effect correction did eliminate the distinction between low cortical SUVRs and higher subcortical SUVRs. Increased frontal lobe SUVR, particularly in PSP was predicted, as this region is involved in the later stages of tau pathology development, and presented with substantial atrophy, thereby diminishing the relative activity measured in the frontal lobe (Williams et al., 2007). Significant increases in caudal SUVRs were also seen across all three groups. While there may have been atrophy in the caudate of PSP patients, it is unlikely that there was the same amount of atrophy in all groups; therefore, this increase in SUVR is not believed to be due to presence of tau pathology in PSP. The substantia nigra was another ROI that demonstrated a substantial increase in SUVR following partial volume effect correction. It is unclear how accurate [ 18 F]AV-1451 binding is in this area due to previously reported data. [ 18 F]AV-1451 phosphor screen autoradiography revealed signals on tauopathy (PSP and AD) and non tauopathy (control and dementia with Lewy bodies) substantia nigra brain slices (Marquie et al., 2015). This may indicate nonspecific binding of [ 18 F]AV-1451 in this brain area and signals in the control brain slices is consistent with our findings in the non-partial volume effect corrected data. Partial volume effect correction removed the significant increase of HC substantia nigra SUVR compared to PSP and PD. Based on these results and previously 95

112 reported data it is unclear how reliable [ 18 F]AV-1451 is at selectively imaging pathological tau in the substantia nigra Off Target Binding In previous post mortem and human studies, [ 18 F]AV-1451 was found to bind melanin and neuromelanin-containing cells (Hansen et al., 2016; Marquie et al., 2015). This study demonstrated consistent results. [ 18 F]AV-1451 SUVR in the substantia nigra of HC was significantly higher than SUVR in PD and PSP from minutes, and significantly greater than PD from minutes. While this significant difference did not stand PVC, HC still tended to have a higher mean SUVR when compared to PD and PSP. These results, along with results from in vitro and in vivo studies, suggest that [ 18 F]AV-1451 is capable of binding neuromelanin-containing cells in the substantia nigra and may be useful for imaging the depigmentation of the substantia nigra that occurs in PD and other parkinsonian disorders. 5.2 Tau Radiotracers for PSP Tauopathies differ from each other in many ways: symptomology, tau load pattern, etc. One of the most important characteristics to successfully image tau in vivo in all tauopathies is the ability of an imaging agent to bind all conformations of tau. Tauopathies 96

113 contain varying ratios of tau isoforms, the ratio in AD is similar to that of a healthy brain with an equal 3R to 4R ratio, as indicated by its major tau bands of 60, 64 and 68 kda and minor band of 72 kda illustrated in figure 5-1. PSP and CBD tau aggregates run as major bands of 64 and 68 kda and minor band at 72 kda are primarily composed of 4R isoforms, while Pick bodies in PiD are generally formed by 3R isoforms, as demonstrated by the major tau bands of 60 and 64 kda and a minor band of 68 kda. An ideal radiotracer for all tauopathies would be capable of imaging both isoforms. Previous in vitro and in vivo work with [ 18 F]AV-1451 on AD brain slices and AD patients suggests that this radioligand is capable of binding both 3R and 4R tau isoforms (Chien et al., 2013; Schwarz et al., 2016; Xia et al., 2013). Additionally when tested in post mortem tissue, [ 18 F]AV-1451 phosphor screen autoradiography demonstrated signal in AD but no signal in PSP and CBD (4R), and PiD (3R). Therefore, [ 18 F]AV-1451 does not seem to preferentially image one tau isoform over the other. Lack of signal in PSP brain slices and the absence of differences between our groups is not likely due to the 4R tau isoform present in PSP tau inclusions. Another differential between AD and other tauopathies is the structure of its tau aggregates. AD tau inclusions form primarily PHF conformations, with only approximately 5% of total tau load represented by straight filaments (Spillantini, Bird, & Ghetti, 1998). PHFs have a diameter of approximately 8-20 nm and periodicity of 80 nm. PSP and other tauopathies (CBD and PiD) present primarily with straight filaments and to a much lesser degree PHFs. Differences between PHF tau and straight filaments is illustrated in figure

114 Figure 5-1. Diagram of tau band variance across different tauopathies. Type I: major tau bands of 60, 64 and 68 kda and a minor tau band of 72 kda. Type II: major tau bands of 60 and 64 kda and a minor tau band of 68 kda. Type III: major tau bands of 64 and 68 kda and a minor tau band of 72 kda. Reproduced with permission from (Spillantini et al., 1998). 98

115 Figure 5-2. Electron micrographs of tau filaments (scale bar: 100nm). A: PHF in AD. B: straight filament in PSP. C: PHF in Down syndrome. D: straight filament in PiD. E: PHF in Seattle family. F: twisted filament in familial MSTD. Reproduced with permission from (Spillantini et al., 1998). 99