VMAT2 Availability in Parkinson’s Disease with REM Sleep Behaviour Disorder

Mikaeel Valli (  mikaeel.valli@mail.utoronto.ca ) University of Toronto https://orcid.org/0000-0002-4059-2333 Sang Soo Cho Seoul National University Carme Uribe CAMH: Centre for Addiction and Mental Health Mario Masellis Sunnybrook and Women\'s College Health Sciences Centre: Sunnybrook Health Sciences Centre Robert Chen Toronto Western Hospital Alexander Mihaescu CAMH: Centre for Addiction and Mental Health Antonio P. Strafella CAMH: Centre for Addiction and Mental Health


Introduction
Rapid eye movement (REM) sleep behavior disorder (RBD) is a parasomnia characterized by the loss of normal skeletal muscle atonia during REM sleep (1). This results in dream enacting behaviours that are often violent or aggressive in nature. Increasing evidence points to the link between idiopathic RBD and alpha-synucleinopathies such as Parkinson's disease (PD) (2,3). RBD is one of the hallmark prodromal complications that manifests as early as 20 years before o cial PD diagnosis (4). The risk estimate of developing PD is 33% at 5 years since RBD diagnosis. This risk sharply increases to 91% at 14 years (5).
This underscores the importance of RBD in PD pathology as it is a strong clinical predictor of PD and is co-morbid in nearly half of the PD patient population (5).
In the time of PD diagnosis, there is already advanced degeneration within the nigrostriatal dopaminergic system, with estimates of 50-70% dopaminergic terminal loss within the putamen based on a postmortem study (6) and in vivo neuroimaging of early-stage PD patients with unilateral motor impairment (7). In the prodromal phase between the time of RBD diagnosis and clinical manifestation of PD, there are multiple neuroimaging evidence that shows presynaptic striatal degeneration, particularly by indexing the dopamine transporter (DAT) density. DAT is a symporter primarily found on the presynaptic terminals of nigrostriatal neurons. It carries the function of clearing released dopamine from the synaptic cleft back into the presynaptic terminals-thereby modulating dopaminergic transmission (8). Neuroimaging studies report consistent striatal DAT depletion in 20-40% of RBD patients relative to healthy controls, particularly within the putamen (5,(9)(10)(11)(12). The decline of striatal DAT has been demonstrated from healthy controls to sub-clinical RBD (REM sleep without atonia on polysomnography, but without abnormal nocturnal behaviours) to manifest RBD to PD (13,14). This worsening pattern continues in a subset of PD patients with RBD where they show greater DAT depletion in the caudate and putamen compared to PD patients without RBD (15,16). Consistently, another study by Arnaldi and colleagues (2015) showed DAT levels in the putamen progressively decreased from idiopathic RBD to PD without RBD to PD with RBD. Taken together, these studies implicated that DAT imaging may play a contributory role in detection of RBD and in PD patients with RBD.
In addition to measuring DAT density as an index for presynaptic dopaminergic integrity, quanti cation of the type 2 vesicular monoamine transporter (VMAT2) is also another approach. VMAT2 is an integral membrane protein that is responsible for shuttling monoamine neurotransmitters including dopamine from the cytosol to the synaptic vesicles (8). Evidence shows that VMAT2 is sensitive to changes in vesicular dopamine concentration (17). However, its binding site was shown to be less sensitive to changes induced by medication or compensatory mechanisms associated with the loss of dopaminergic neurons relative to DAT (18). Hence, quantifying VMAT2 levels allow for a more accurate measurement of the dopaminergic terminal integrity compared to measuring DAT levels (19 (20). Similar results were shown in a smaller older study using [ 11 C]DTBZ where the authors observed reduced VMAT2 availability in the posterior putamen in RBD patients in relation to healthy controls (21). As presented here, there are limited studies that focused on VMAT2 imaging, but these investigations indicate that VMAT2 imaging may also have a certain role in detecting RBDs.
In summary, the literature seems to suggest a certain contribution of the presynaptic dopaminergic system in RBD. However, the characterization of VMAT2 has been rarely explored-especially in PD patients with RBD. By examining VMAT2 as an in vivo molecular target with its unique properties, we aim to elucidate VMAT2 levels in PD patients who are comorbid with RBD. We achieved this by using [ 11 C]DTBZ PET imaging in PD patients with and without RBD, while having age-matched healthy controls.
We hypothesize that PD patients with RBD to show reduced VMAT2 availability in striatal regions including caudate, putamen and ventral striatum relative to healthy controls and PD patients without RBD. This hypothesis would be in line with reductions of the DAT observed in the striatal regions seen in idiopathic RBD and PD patients with RBD relative to controls and PD patients.

Participants
A total of 45 participants were enrolled in this study: 15 PD patients without RBD, 15 PD patients with RBD, and 15 age-matched healthy controls ( Table 1). Some of the imaging, demographic, cognitive and psychological data from these participants have been reported previously (22,23). Both PD patient groups met the criteria for the UK Parkinson Disease Society Brain Bank. All participants had no evidence of other neurological or psychiatric conditions or any other medical conditions that precluded them from the PET and MR imaging. The severity of PD motor symptoms was tested through the Hoehn and Yahr Scale and the United Parkinson Disease Rating Scale (UPDRS-III) while patients were on medication. Levodopa equivalent daily dose (LEDD) calculation for each patient has been previously described by Evans et al. (24). All participants were age-matched; and PD patients were matched for disease severity based on UPDRS-III score and LEDD. Identi cation of probable RBD symptoms was completed through using the informant-based response to the rst question on the Mayo Sleep Questionnaire: "Have you ever seen the patient appear to act out his/her dreams while sleeping." Patients were classi ed as clinically probable RBD if the patients' sleeping partner answered yes to this question (25). This singular question has been validated against polysomnography, with a sensitivity of 98% and speci city of 74%, in a multicenter prospective cohort study of healthy older adults and suspected neurodegenerative disease (25,26).
In order to prepare for the PET scan, PD patients performed an overnight 12-hour withdrawal from antiparkinsonian medication to minimize the effect of medication during the scans while maintaining patient comfort and functioning (27). To prevent excessive fatigue, the PET and structural MRI scans were completed on separate days. Montreal Cognitive Assessment (MoCA; 28) and Beck Depression Inventory (BDI; (29) were obtained on all participants to assess general cognitive capabilities and depression levels, respectively. All participants provided informed written consent prior to beginning study procedures which were approved by the research ethics committees for the Centre of Addictions and Mental Health and the University Health Network of the University of Toronto.

Imaging Acquisition
The preparation of the [ 11 C]DTBZ radioligand was described previously (30). PET scans were collected using a three-dimensional (3D) high resolution research tomograph (HRRT) scanner (Siemens, Knoxville, TN). This equipment allows the measurement of radioactivity in 207 brain slices, with a thickness of 1.22 mm each (22). The detectors of the HRRT are a lutetium oxyorthosilicate/lutetium-yttrium oxyorthosilicate phoswich, with each crystal element measuring 2 × 2 × 10 mm 3 . To minimize head motion, a customized thermoplastic facemask was provided to each participant prior to the HRRT PET scan, and the facemask was secured through the head-xation system (Tru-Scan Imaging, Annapolis). After securing participants within the PET scanner, a transmission scan was rst completed using a single photon point source, 137 Cs (t 1/2 = 30.2 years, E γ = 662 keV), which had a duration of 6 minutes and 9 seconds. This transmission scan was immediately followed by the acquisition of the emission scan to correct for attenuation (where frame durations were: 1 × background; 15 frames × 60 seconds; and 15 frames × 300 seconds). Subsequently, the [ 11 C]DTBZ radioligand was injected as a bolus into an intravenous line placed in the antecubital vein. Emission data were collected in list mode for 60 minutes while subjects were at rest.
The emission list mode data were re-binned into a series of 3D sinograms. The 3D sinograms were gap lled, scatter corrected and Fourier re-binned into 2-dimensional (2D) sinograms. The images were reconstructed from the 2D sinograms using a 2D ltered-back projection algorithm. The reconstructed images had 256 × 256 × 207 cubic voxels that measured 1.22 × 1.22 × 1.22 mm 3 . The dynamic images were then reconstructed into 17 frames. The rst frame was variable as it was dependent on the time between the start of acquisition and the introduction of the [ 11 C]DTBZ radioligand in the tomograph eld of view. The following frames were de ned as: 1× ≥ 22 seconds, 4 × 60 seconds, 3 × 120 seconds, 8 × 300 seconds, and 1 × 600 seconds.
To provide anatomical reference for the parametric PET image analysis and to rule out structural lesions, a whole-brain T1-weighted MR image was acquired from each participant using GE Signa HD × MRI system (GE Discovery MR750 3T; T1-weighted images, fast spoiled gradient echo with repletion time = 6.7 milliseconds, echo time = 3.0 milliseconds, ip angle = 8 mm, slice thickness = 1 mm, number of excitations = 1, and matrix size = 256 × 192).

Imaging Analysis
Image preprocessing was completed using an in-house software, Regions of Mental Interest (ROMI; 31).
This software was used to obtain the time activity curve (TAC) for the reference region-the occipital lobe -for all participants. ROMI used Statistical Parametric Mapping (SPM8, Welcome Department of Imaging Neuroscience, London, UK), where each participant's MR image was used to nonlinearly transform a standardized brain template (International Consortium for Brain Mapping/Montreal Neurological Institute 152 MRI) with prede ned regions of interests (ROIs). The individual ROI template underwent further re nement based on the gray matter probability of the segmented MRI. The re nement of each individual's ROIs were then aligned and resliced using a normalized mutual information algorithm (32) to match the individuals PET scan. Subsequently, the TAC for the occipital lobe was obtained from the dynamic [ 11 C]DTBZ PET image in the native space.
Upon the completion of the pre-registration procedure with ROMI, [ 11 C]DTBZ PET parametric nondisplaceable binding potential (BP ND ) maps were generated in the native PET space with simpli ed reference tissue model (33) using the occipital cortex TAC value as reference region (obtained through ROMI). This was completed using Receptor Parametric Mapping software (RPM; 34) within MATLAB R2015a (version 8.5.0.197613; MathWorks). Using SPM12 (version 7487) within MATLAB, the parametric BP ND images were transformed into standardized stereotaxic space using each participants' individual MRI. These normalized images were then smoothed with a Gaussian function at 8 mm full width halfmaximum.
These ROIs were transformed into a parametric [ 11 C]DTBZ PET BP ND map, and the BP ND values were extracted using MATLAB based REX toolbox (http://web.mit.edu/swg/software.htm). We used the BP ND values obtained through REX for each ROI for statistical analysis.
Statistical analysis Demographic characteristics were tested for differences between the three groups (i.e., healthy controls, PD patients without RBD, and PD patients with RBD) using ANOVA. Speci cally, we performed ANOVA and Bonferroni post-hoc testing on age, MoCA score, BDI, UPDRS-III score, Hoehn and Yahr score, LEDD amount, quality and quantity of injected radioligand across all three participant groups. To assess for differences in sex proportions between groups, a chi-squared analysis was performed. Statistical outliers was investigated using the interquartile range method (36).
Mixed effects model was used to compare the extracted [ 11 C]DTBZ BP ND between the three groups for each ROI. The xed factors within the model were group (i.e., healthy controls, PD patients without RBD, and PD patients with RBD) and side (i.e., left vs. right ROI); the participants was kept as the random factor. The model also co-varied for sex, MoCA and BDI score. Post hoc independent sample t tests were used to assess for differences between groups and were corrected for multiple comparisons using the Bonferroni method.
Multiple regression analyses were used to correlate [ 11 C]DTBZ BP ND within each ROI against clinical measures including UPDRS-III score, Hoehn and Yahr score, LEDD amount, and disease duration while factoring patient group condition (i.e., PD patients with and without RBD). This regression model included sex, MoCA and BDI as co-variates. All tests were completed using SPSS (version 21; Chicago, IL); and the alpha level was set to 0.05 as a cut-off to determine signi cance.

Results
The demographic, clinical and psychological characteristics for each group are summarized in Table 1. There were no differences between all participant groups (i.e., healthy controls, PD patients without RBD, and PD patients with RBD) for age, radiotracer injected dose, injected mass, and speci c activity. PD patients with and without RBD were comparable in relation to their disease duration, UPDRS-III score, Hoehn and Yahr score, and LEDD amount. However, differences were observed between groups for sex, MoCA, and BDI. These three variables were included as co-variates in subsequent analyses. There were no statistical outliers using the 1.5 interquartile range method.
As there were main effects of MoCA [F (2, 42) = 7.29, p = 0.002] and BDI [F (2, 42) = 3.89, p = 0.028], we followed up with Bonferroni post hoc t-tests to detect where the group differences occurred. These particular results were reported previously by Valli et al. (23). In summary for BDI, patients with RBD had the highest depression score, then PD patients without RBD, then healthy controls. Only PD patients with RBD were statistically higher than healthy controls (t = -2.68, p = 0.03). Regarding MoCA, in comparison to health controls, PD patients with RBD had the lowest cognitive score (t = 3.70, p = 0.003), followed by patients without RBD (t = 2.66, p = 0.0165). There were no statistical differences between the two PD subgroups on the MoCA measure. patients with and without RBD were lower than healthy controls for all these signi cant ROIs. Summary of the Bonferroni corrected post-hoc t-test results between PD patients with RBD versus healthy controls and PD patients without RBD versus healthy controls is displayed in Table 2. We were not able to detect statistical differences between PD patients with and without RBD for these signi cant regions (Fig. 1). In summary, the signi cant ndings from the mixed effects model was driven mainly by the differences between healthy controls and both PD subgroups, not the differences between PD patients with and without RBD. Column B on the right shows results between PD patients with RBD (PD-RBD+) and healthy controls.

Results
Multiple regressions were computed to predict clinical measures (i.e., disease duration, UPDRS-III score, Hoehn and Yahr score, and LEDD amount) from group condition (i.e., PD patients with and without RBD), and [ 11 C]DTBZ BP ND of all basal ganglia ROIs explored, while covarying for sex, MoCA and BDI score. We were able to observe a signi cant interaction between left caudate BP ND and group condition where these two measures predicted UPDRS-III score (F (1, 23) = 11.13, p = 0.003, R 2 = 0.59). Figure 2

Discussion
This study aimed to characterize the presynaptic dopaminergic integrity in PD patients with and without RBD by indexing VMAT2 availability using PET imaging. We observed that both PD patient subgroups had lower tracer binding compared to healthy controls within the basal ganglia. There were no indications to show [ 11 C]DTBZ binding differences between PD patients with and without RBD, which did not con rm our hypothesis. The reduced radioligand signal compared to healthy controls re ects lower VMAT2 availability in both patient subgroups-which implies nigrostriatal denervation as a result of PD neurodegeneration. In the group of PD patients without RBD, there was a strong negative relationship between left caudate VMAT2 availability and UPDRS-III score. In other words, with the worsening of motor severity in patients without RBD, it correlated with more presynaptic denervation in the left caudate. However, this relationship was not present in PD patients with RBD.
The observed reduction of VMAT2 density as measured by [ 11 C]DTBZ in our PD patient sample in the putamen, caudate, substantia nigra, and globus pallidus relative to controls is consistent with previous neuroimaging studies of VMAT2 (7,(37)(38)(39)(40). This reinforces the notion that [ 11 C]DTBZ PET imaging holds the potential to effectively differentiate PD patients from controls (41,42). The observed signi cant correlation between motor severity for PD patients and left caudate was also consistent with previous studies in their sample of PD patients with no reports of other co-morbidities (41,42). This relationship is in line with the view that motor disability in PD is primarily associated with impairment of subcortical structures within the nigrostriatal pathway (43). In contrast, no signi cant interactions were observed between radiotracer binding and clinical measures in PD patients with RBD, thus implying that other neurochemical abnormalities may account for this clinical complication.
Unlike previous DAT studies (14,16,44), we found no evidence of VMAT2 level differences between groups of PD patients with and without RBD. This observation is consistent with a previous investigation that used [ 11 C]DTBZ PET imaging as a secondary focus to their study objectives in PD patients with and without RBD (45). Speci cally, they were unable to differentiate VMAT2 levels of PD patients with RBD from PD patients without RBD within the caudate and putamen (45). Our current study was different from this previous study where we used age-matched healthy controls and explored more brain regions within the basal ganglia, beyond just the caudate and putamen. There are several possible reasons that could explain the lack of differences in VMAT2 levels between PD patients with and without RBD that we observed. VMAT2 has been demonstrated to be a stable marker for presynaptic nigrostriatal integrity as it is less prone to changes induced by medications or compensatory mechanisms associated with the loss of dopaminergic neurons (18). Our ndings re ect that VMAT2 remained unaffected by the co-morbidity of RBD in conjunction with presynaptic changes associated with medication or compensatory mechanisms that would normally occur in PD (18). This implies that the PD pathophysiology is the primary driving force depleting VMAT2 availability in both groups.
An alternative explanation may involve the relationship between the activities and availability of VMAT2 and DAT in the presynaptic terminals (46). Previous literature has consistently shown that PD patients with RBD have lower DAT availability relative to PD patients and controls (14,16,44). This observed reduction of DAT levels would in turn lower the neural ability to reuptake dopamine back into the presynaptic terminals for vesicular repackaging and subsequent reutilization (17,47,48). The lower levels of presynaptic vesicles with dopamine should in turn result in a reduction of the VMAT2 levels in PD with RBD. However, this rationale was de ed by the lack of differences in VMAT2 levels between PD patients with and without RBD. It could be possible that there were varying endogenous intravesicular dopamine levels in PD patients with and without RBD that may have netted in negligible differences in BP ND signalling between the two groups (17).
Previous literature shows other molecular structures and neurotransmitters may play a larger role contributing to RBD in PD apart from DAT (14,16,44). Outside the striatum, our group showed a negative relationship where D2 receptor availability within the uncus parahippocampus decreased with increasing disease severity in PD patients with RBD relative to PD patients without RBD (23). Other studies showed altered cholinergic (45) and noradrenaline levels (49), along with changes in glucose metabolic activity (15) in PD patients with RBD relative to patients without RBD, suggesting RBD pathophysiology in PD is multi-systemic that impacts regions beyond the striatum.
This study has some limitations to consider. Our patients were screened for probable RBD through the rst question of the Mayo Sleep Questionnaire as part of their routine neurology clinic visit. These patients were not con rmed through the sleep polysomnography test. Despite this, the questionnaire has been validated in two studies. The rst study validated the questionnaire against polysomnography in a multi-centre prospective cohort trial that included patients suspected to have neurodegenerative disease and healthy older adults. This study achieved a sensitivity of 98% and speci city of 74% (25). A follow up validation study of the Mayo Sleep Questionnaire was carried out in a community-based sample of 128 participants who had underwent a previous polysomnography test, and resulted in a sensitivity of 100% and speci city of 95% (26). These evidences makes the sleep questionnaire a useful research tool in settings where polysomnography is not readily available (50).
Another limitation is that patients were assessed on several clinical measures while "ON" dopamine replacement therapy and they were "OFF" medication during the PET scan. In turn, the relationship between motor severity and dopaminergic degeneration within the basal ganglia may not be fully representative even though VMAT2 was shown to be less prone to medication in uences (37). Although we showed a relationship between nigrostriatal innervation and motor score in PD patients without RBD, investigating motor features while "ON" and "OFF" medication and changes with VMAT2 levels may provide more insight about the true relationship observed between PD with and without RBD and motor severity in relation to the left caudate VMAT2 availability.

Conclusion
In comparison to age-matched controls, the current study revealed that PD patients with and without RBD had lower VMAT2 levels within the subcortical structures of the basal ganglia which re ects denervation within the nigrostriatal pathway. No signi cant interactions were detected between the BP ND and clinical scores in PD patients with RBD. These ndings collectively suggest that VMAT2 may not be a signi cant contributor to the pathophysiology of RBD in PD patients. Future studies are encouraged to explore other underlying neural chemistry mechanisms to better understand the driving force of RBD in PD patients.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Ethics approval and consent to participate All participants provided informed written consent prior to beginning study procedures which were approved by the research ethics committees for the Centre of Addictions and Mental Health and the University Health Network of the University of Toronto.

Consent for publication
Not applicable.  This interaction plot is a result from the regression analysis. Patient group moderates the left caudate BP in PD-RBD−: as the UPDRS-III score increases, PD-RBD− patients have lower VMAT2 availability. However, this relationship is not existent for PD-RBD+. This gure was plotted using marginal means that accounted for the included co-variates: sex, MoCA and BDI. The grey band for each line represents the 95% con dence interval.