Multiple system atrophy as emerging template for accelerated drug discovery in α-synucleinopathies.

There is evidence that the α-synucleinopathies Parkinson's disease (PD) and the Parkinson variant of multiple system atrophy (MSA-P) overlap at multiple levels. Both disorders are characterized by deposition of abnormally phosphorylated fibrillar α-synuclein within the central nervous system suggesting shared pathophysiological mechanisms. Despite the considerable clinical overlap in the early disease stages, MSA-P, in contrast to PD, is fatal and rapidly progressive. Moreover recent clinical studies have shown that surrogate markers of disease progression can be quantified easily and may reliably depict the rapid course of MSA. We therefore posit that, MSA-P may be exploited as a filter barrier in the development of disease-modifying therapeutic strategies targeting common pathophysiological mechanisms of α-synucleinopathies. This approach might reduce the number of negative phase III clinical trials, and, in turn, shift the available resources to earlier development stages, thereby increasing the number of candidate compounds validated.

Introduction

Parkinson's disease (PD) and multiple system atrophy (MSA) are two distinctive neurodegenerative disorders characterized by α-synuclein (αSYN) inclusion pathology, although some genetic PD variants show only little or no αSYN inclusion pathology, particularly PARK2 and LRRK2 mutation carriers [1]. During the last two decades numerous candidate neuroprotective agents have been evaluated in large randomized controlled PD trials. Unfortunately, none of them proved effective either reflecting a lack of neuroprotective efficacy or shortcomings of preclinical models, outcome measures including surrogate markers and trial design issues [2]. Novel strategies to overcome the deadlock in PD interventional trials are therefore needed. We here posit, based on overlapping features [3], that drug discovery in PD can be accelerated by conceptualizing the parkinsonian variant of MSA (MSA-P) as a template for the validation of joint therapeutic strategies targeting αSYN aggregation. Hereinafter, we will review overlaps in genetics, pathology, etiopathogenesis, and clinical presentation, discuss trial considerations and finally suggest a modified drug development approach based on the quick win, fast fail paradigm [4].

Genetic overlap

Rare familial forms of PD generated important insights into the genetic underpinnings of parkinsonism. The seminal description by Polymeropoulos et al. demonstrating that mutations in SNCA, coding for αSYN, are a cause of PD has revolutionized the field for three reasons [5]. First, it clearly disproved the long held hypothesis that PD pathogenesis is not genetic in nature; second, it provided a robust target to generate experimental cell and animal models and third, it offered the first clue that αSYN plays a pivotal role in neurodegeneration.

Systematic evaluations of the SNCA locus in PD and MSA have further advanced our understanding of α-synucleinopathies substantially. In addition to missense mutations, multiplication (triplication or duplication) of the entire SNCA locus has been demonstrated to cause familial PD [6-8]. Although these mutations have not been detected in MSA, the presence of GCI-like inclusions as well as clinical phenotypes reminiscent of MSA have been noted in some SNCA multiplication family members suggesting that an increase in αSYN expression level could be involved in MSA pathogenesis [8,9]. This notion is supported by observations in transgenic mouse models overexpressing αSYN under oligodendroglial promoters, where GCIs as well as neurodegeneration and clinical features of MSA are observed [10]. However, to date gene expression studies of small size did not reveal SNCA expression changes in pathology-proven MSA cases [11], and the role of changes in αSYN expression or degradation in MSA pathogenesis remains elusive.

A genome-wide association study (GWAS) has linked single nucleotide polymorphisms at the SNCA locus to an increased risk for PD [12]. Subsequently, the top risk variants from this study were tested in about 500 MSA cases and 4000 controls and significant association for two variants in SNCA were detected [13]. However, preliminary results of the first MSA GWAS suggest that the SNCA risk loci will not reach genome-wide significance reflecting either insufficient power of the MSA GWAS or ethnic confounders in previous studies [14]. More recently, Kiely et al. [15] reported a SNCA p.G51D mutation in a British family with autosomal-dominant inheritance sharing neuropathological features characteristic of both PD and MSA and thereby providing a possible link between MSA and PD.

These findings indicate that αSYN is presumably a key player in neurodegenerative processes in PD and MSA. Furthermore, the intriguing observation of SNCA multiplications causing MSA-like glial pathology point towards common αSYN-related pathophysiological mechanism that we will discuss below.

Pathologic overlap

Immunohistochemical studies revealed that abnormal αSYN aggregates are found in neurons and axons in PD brains [8], whereas, αSYN aggregates in MSA are predominantly seen in the cytoplasm of oligodendrocytes. The term α-synucleinopathies was coined to embrace these clinicopathological entities [16]. While the cellular distribution of the aggregates is clearly different, there is still important pathological overlap. In both, PD and MSA-P, neurodegeneration is associated with the Lewy body and GCI burden as well as the increase of soluble αSYN in substantia nigra and striatum [17]. In addition, apart from the apparent overlap in degeneration of the dopaminergic nigrostriatal system, involvement of the central and peripheral autonomic nervous system including the dorsal motor vagal nucleus, autonomic parts of the spinal cord, as well as the peripheral autonomic nervous system (e.g. the cardiac and enteric systems), is common to both PD and MSA-P [17-20].

Intriguingly, brainstem LBs – a classical pathological hallmark of PD – were also reported in MSA and, vice versa, GCI pathology occurred in familial PD cases with rapid disease progression [8,9,21]. The molecular basis of these co-existing pathologies remains to be elucidated.

Overall, these neuropathological similarities provide further evidence that αSYN fragments are able to interfere with physiological processes thereby triggering neurodegenerative processes irrespective of the underlying inclusion pathology.

Common pathogenic pathways

The exact mechanisms of αSYN-triggered neurodegeneration are not completely understood, but studies point to common principles that may underlie disease development and progression. At the molecular level αSYN is a natively unfolded, soluble cytosolic protein. Biochemical analyses of αSYN deposits in MSA and PD revealed abnormal phosphorylation at the serine-129 residue, a mechanism that is known to increase αSYN aggregation propensity, as well as the precipitation of ubiquitinated fibrils made up of proteinase K resistant, β-sheeted αSYN [22,23]. Experimental evidence suggests that misfolded αSYN can influence the viability of cells affected by aggregation pathology in numerous fashions including the disruption of cytoskeleton integrity, dysfunction of the ubiquitin-proteasome system and autophagy pathways, by causing mitochondrial dysfunction, or by disruption of the endoplasmic reticulum-Golgi vesicular trafficking and membrane integrity [24,25]. In addition, it was recently proposed that presynaptic αSYN deposits generate a dying back type of neuronal loss [26-28]. In parallel, loss of dendritic spines appears to accompany the presynaptic αSYN pathology in Dementia with Lewy Bodies [29] and loss of dendritic spines was also reported for striatal neurons in post-mortem PD brains [30]. These findings are consistent with a notion that αSYN aggregate-related synaptic dysfunction may constitute the starting point of neurodegeneration in LB disorders [29]. In contrast, MSA-associated neuronal cell death appears secondary to the αSYN-related oligodendroglial dysfunction [31]. Remarkably, glial filamentous inclusions in PD brains affecting oligodendroglial as well as astroglial cell populations were immunoreactive for αSYN and correlated with the severity of nigral cell loss [32].

The temporal evolution of neuropathology in PD [33], combined with the post-mortem observation of αSYN pathology within fetal grafts of transplanted PD patients [34,35] implicate that a transmission of αSYN (or its toxic truncated forms) from cell to cell may be a critical mechanism underlying disease spread in α-synucleinopathies. Interestingly, converging preclinical evidence indicates the ability of αSYN to spread throughout the PD brain in a prion-like fashion [36,37]. In MSA, the origin of oligodendroglial αSYN remains elusive as illustrated by lack of evidence of αSYN expression in oligodendrocytes of healthy and MSA brains [38]. This gives rise to the notion that αSYN is taken up into oligodendrocytes from extracellular space with αSYN most likely being secreted by neurons.

In addition to direct neurotoxic effects of intra- and extracellular αSYN accumulation, it is proposed that αSYN may trigger gliosis which is associated with enhanced oxidative stress. Oxidative and nitrative stress seem to play a pivotal role in both PD and MSA pathogenesis either directly affecting neuronal survival or resulting in posttranslational modifications of αSYN which consequently leads to accelerated protein aggregation and enhanced toxicity [39-41]. In PD an increasing body of evidence delineates the role of microglial and astroglial responses [41]. Similarly, MSA in vivo and post-mortem data also support the involvement of glial mechanisms and transgenic models confirm the role of inflammatory signaling in the pathogenesis of GCI-related neurodegeneration [42]. However, further aspects of pro-inflammatory mechanisms in the pathogenesis of PD and MSA need to be addressed to define potent candidate targets for neuroprotection [39,41].

In summary, despite the obvious notion of αSYN being involved in neurodegeneration in PD and MSA, several molecular mechanisms overlap. Both disorders are characterized by spreading of αSYN pathology with PD showing a pattern of neuron-to-neuron αSYN propagation. In MSA, neuronal αSYN is passed to oligodendrocytes as suggested by the lack of SNCA expression in glial cells [38]. Furthermore, posttranslational modifications of αSYN and neuroinflammatory processes are comparable in MSA and PD. Major genetic, morphologic, and pathogenic overlaps between idiopathic PD and MSA-P are also summarized in Fig. 1.

Fig. 1

Overlapping and discriminating features of MSA and idiopathic PD. The figure shows distinguishing and overlapping features of MSA-P and Parkinson's disease at the pathogenic, pathologic and clinical level. Unique MSA-P features are emphasized by a red highlight, PD features are shaded in blue and overlapping features are presented on a purple background.

Clinical overlap

Alongside the apparent overlap in motor features, clinicopathological and prospective clinical studies indicate that MSA-P and PD share non-motor symptoms as well [43]. Due to this overlap, PD and MSA-P are frequently indistinguishable in the early disease stages – even when formal criteria are applied [3]. Distinguishing and overlapping clinical features are shown in Fig. 1.

Motor features

The diagnosis of PD hinges on the presence of bradykinesia together with one out of three additional hallmark features including rest tremor, rigidity, and postural instability [44]. In MSA-P all of these features may occur although pill-rolling rest tremor is uncommon [45] and, in contrast to PD, postural instability emerge early within the first years into the disease-course in MSA-P [46]. One of the key-characteristics of PD is the marked and persistent levodopa response with drug-induced motor complications at later disease stages [44]. Despite MSA-P being considered levodopa-unresponsive, there is an increasing body of evidence suggesting that MSA patients receiving levodopa may show a beneficial (albeit mostly self-limited) motor response and some of these patients will experience levodopa-induced motor complications [3,45]. These findings suggest that a subgroup of MSA-P patients may present with a PD-like syndrome. The latter notion is further supported by a case series of pathologically proven MSA-P patients exhibiting slow progression and prolonged survival of more than 15 years [47].

Non-motor features

Non-motor features are common to both disorders involving autonomic, sleep and neuropsychiatric domains. As suggested for premotor MSA [43], orthostatic hypotension (OH) may precede the onset of motor symptoms in some PD patients as well [48]. Urinary problems emerge almost universally in PD and MSA. Although time course and severity of urinary symptoms are obviously different, both PD and MSA-P share similar urological features including frequency, urgency, urge incontinence and nocturia reflecting a prominent sphincter-detrusor dyssynergia [49]. Erectile dysfunction in men and genital hyposensitivity in women are the earliest, but usually not the presenting features of MSA-P [50], and sexual dysfunction may also occur as an early symptom in PD [51]. Intriguingly, a retrospective analysis of a large cohort of men demonstrated that erectile dysfunction is associated with an increased risk of developing PD [52]. Also both MSA-P and PD patients suffer from reduced bowel frequency, difficulties in evacuation and rarely fecal incontinence [53]. Nevertheless, a pattern of widespread, severe and rapidly progressive generalized autonomic failure with adrenergic dysfunction clearly points towards a diagnosis of MSA [54].

Sleep disorders including rapid eye movement sleep behavior disorder (RBD) are common in PD and MSA-P. In fact, it appears as if α-synucleinopathies share a common RBD phenotype often with premotor onset [55,56].

In contrast to the long-held belief, clinicopathological evidence proved that MSA patients may develop cognitive dysfunction and dementia [57]. Additionally, a recently published prospective neuropsychological study found that approximately 20% of MSA patients score below the fifth percentile on the Mattis dementia rating scale [58]. Nevertheless, dementia is much more common in PD compared with MSA [59], however, the profile of cognitive impairment is comparable in PD and MSA as both disorders exhibit prominent impairment in frontal executive functions [58,59].

Another distinction between MSA and PD is the olfactory sense with MSA patients showing intact or mildly impaired olfaction only, whereas most PD patients are hyposmic or sometimes anosmic with deficits affecting odor detection, discrimination and identification. Even more interestingly, olfactory disturbances precede the motor-onset of PD [44,60].

Disease progression

Rapid disease progression in MSA-P usually leads to loss of independent ambulation within the first few years of disease while PD patients usually maintain mobility and functional independence for more than 10 years with appropriate management [57]. Although natural motor progression of PD is not well documented, data on the short-term rate of progression are available due to placebo-controlled trials suggesting accelerated progression of motor dysfunction in the early course of the untreated disease. These data are consistent with observations from the pre-levodopa era reporting latencies to Hoehn and Yahr stage IV and V of 9 and 14 years, respectively. The introduction of levodopa has significantly delayed the disability milestones and reduced mortality in PD [61]. In contrast, MSA-P patients experience rapid motor decline regardless of levodopa therapy [45], most likely reflecting the widespread striatal involvement. Furthermore, wheelchair-dependency was reported to occur on average after 6.7 years from disease-onset in MSA patients. Other clinical milestones of pre-terminal disease appeared in rapid succession at 5.3 years after disease-onset [57]. Another characteristic feature occurring in both disorders are falls. In contrast to MSA, PD patients usually suffer from falls in advanced stages only, however, a retrospective clinicopathological study demonstrated that latency to onset of recurrent falls but not their duration differentiated PD from MSA [46]. Finally, disease progression in PD is characterized by the increasing prevalence of non-motor symptoms, in particular neuropsychiatric features and autonomic failure [59]. In MSA, autonomic failure including urogenital disturbances and OH occurs early in the course of the disease and in a substantial proportion of cases it even precedes motor onset [43]. Furthermore autonomic failure, particularly urogenital symptoms, appear to worsen over time [45,62].

Clinicopathological studies in neurodegenerative diseases are limited by the inability to follow patients longitudinally and by the difficulty of associating pre-mortem clinical symptoms to post-mortem findings. In contrast, brain imaging has the potential to study lesion progression in vivo. This has been repetitively performed in parkinsonian disorders [63-65]. In PD, functional imaging studies using both fluorodopa- and β-CIT single photon emission computerized tomography (SPECT) estimated an annual loss of signal in the range of 5%–10% per year [66,67] which is considerably slower than progression rates determined in MSA-P [68]. Atrophy as determined by magnetic resonance imaging (MRI) is more rapid in MSA-P as compared with PD with regional atrophy rates ranging from 1.0 to 2.5% decline per year in MSA and 0.3–0.8% decline per year in PD [63]. In addition, a significant increase in diffusivity over time was observed in MSA patients only [69-71].

Trial considerations

In the absence of accurate biomarkers that are able to monitor PD disease progression, neuroprotection and disease-modification trials have been restricted to clinical outcomes which have important shortcomings including the fact that disease-specific rating scales are nonlinear in nature and may fail to detect progression of important disease features [2]. To this end, interventional trials utilizing clinical milestones as primary outcome measures would be helpful. Obviously, latter studies are not feasible in PD with latencies to clinical milestones exceeding a decade. However, such trials may succeed in MSA with pre-terminal clinical milestones being reported to appear in rapid succession shortly after diagnosis [57]. In addition, PD progression as measured by validated rating scales is slow [61], whereas the motor decline in MSA as determined by the disease-specific Unified MSA Rating Scale is much more rapid [45]. As mentioned before, trials aiming at disease-modification in PD require extensive study periods of at least two years due to the slow progression. In that regard and with excellent symptomatic treatment options at our disposal, long-term placebo-controlled clinical trials in PD have raised serious ethical concerns.

Serial brain imaging could be instrumental in the context of future interventional trials because monitoring progression by imaging parameters has been shown to reduce the required sample size. The progression of structural imaging in PD is slow hampering its utility in PD disease-modification trials [63], whereas the continuous signal decline in presynaptic dopamine transporter imaging warranted the use of dopamine transporter SPECT in PD trials. Dopamine transporter SPECT was exploited to document nigrostriatal degeneration in two independent clinical trials suggesting disease-modifying effects in both of the two studied compounds [72,73]. However, the signal alterations were subsequently attributed to dopamine transporter expression dynamics in the levodopa treated control group [74-76]. By contrast, functional and morphologic imaging parameters were shown to be rapidly progressive in MSA. Thus far, three MSA trials used neuroimaging as a secondary outcome measure. An efficacy and safety study on lithium (NCT00997672) compared micro- and macrostructural MRI parameters before and after the treatment, including voxel-based morphometry, resting functional MRI, diffusion tensor imaging, and magnetic resonance spectroscopy, but was terminated prematurely due to safety concerns. Another clinical trial, testing the efficacy of rasagiline in MSA-P (EudraCT 2009-014644-11), compared rates of progression of putaminal abnormalities in an imaging sub-study exploiting diffusion-weighted MRI. The imaging results showed progressive increase in putaminal apparent diffusion coefficient values over time and supported the clinical observation of lacking treatment effect [77]. A recent study (NCT00911365) on mesenchymal stem cells (MSC) in patients with MSA defined changes in cerebral glucose metabolism and gray matter density as determined by fluorodeoxyglucose positron emission tomography and MRI-based morphometry, respectively, as secondary outcome measures. The results revealed a more pronounced decrease in cerebral glucose metabolism and gray matter density at 360 days relative to baseline in the cerebellum and the cerebral cortical areas in the placebo group compared to the MSC group [78].

Finally, bearing recent experimental approaches counteracting intra- and extracellular αSYN toxicity in mind, in vivo visualization of αSYN deposition has become a recognized research target. MSA seems to be an ideal candidate for these imaging approaches possibly showing a better signal-to-noise ratio due to the greater αSYN load. In fact, Kikuchi and co-workers [79] reported that GCIs may be visualized in vivo in MSA patients using positron emission tomography and the amyloid ligand 11C-2-[2-(2-dimethylaminothiazol-5-yl)ethenyl]-6-[2-(fluoro)ethoxy]benzoxazole. Nevertheless, additional studies are urgently required to establish imaging of intracellular αSYN deposits as surrogate marker in future interventional trials.

The rapid clinical progression of MSA-P and the lack of confounding symptomatic therapies in MSA as well as the presence of rapidly progressive and reliable surrogate markers are unique advantages of using MSA patients to screen for disease-modifying agents.

Discussion

Clinical development (phases I–III) account for approximately 63% of the costs for each new compound launched with 53% arising from phase II until launch [4]. It seems natural that clarifying the neuroprotective potential earlier in the drug development process would increase the number of evaluated interventional treatments. Based on the trial considerations delineated above, we suggest that MSA-P could be employed to clinically validate compounds, which were shown to interfere with joint pathophysiological mechanisms in preclinical α-synucleinopathy models. To this end, we posit that a modified drug development approach (Fig. 2) based on the quick win, fast fail paradigm [4] exploiting MSA-P phase II proof of concept trials as a template disease for medications targeting αSYN could accelerate drug development in α-synucleinopathies. Employing such a MSA filter barrier could allow an earlier decision whether a candidate compound effectively mediates neuroprotection and, therefore, warrant further clinical validation (phase III trials). Consequently, the number of phase III trials prone to failure would decrease which, in turn, would increase the resources that could be allocated to earlier drug development stages. Overall, we suggest that an early MSA proof of concept trial has two major implications for accelerating development of neuroprotective strategies in PD with αSYN inclusion pathology: (1) due to the time- and cost-savings generated by reducing the number of negative phase III trials, more resources would be available to earlier drug development phases which consequently would increase the number of candidate compounds being validated. (2) Rapidly progressive surrogate measures and the lack of effective symptomatic treatment would enable relatively short-term, parallel group design MSA trials to prove neuroprotective efficacy. In this context, it is crucial to reliably identify MSA-P patients early in the disease course at the level of independent ambulation. This can be achieved with current consensus criteria for possible MSA [54] which show an excellent positive predictive value of 95% at first clinic visit for a post-mortem MSA diagnosis [80]. Nevertheless, due to the suboptimal sensitivity of these criteria [80] the development and validation of highly sensitive and specific imaging and cerebrospinal fluid or plasma biomarker is a high priority objective in MSA research.

Fig. 2

Modified quick win, fast fail drug development approach The figure illustrates possible time- and cost-savings of a modified quick-win, fast fail proof of concept (PoC) approach (B) in comparison to the current drug development and validation strategy (A). We posit that introduction of a MSA interventional trial as PoC filter barrier would be useful to pre-select the most promising candidate compounds and thereby substantially reduce the number of negative phase III trials being carried out.

However, it has to be acknowledged that the described paradigm has certain shortcomings including the risk of false negative MSA trials. If any doubt on the face validity of such a proof of concept trial remains and a strong preclinical rational for a disease-modifying efficacy exists, a repetition of the trial in a PD setting would be required to exclude a PD-specific mechanism of neuroprotection. Vice versa, a positive MSA study does not necessarily predict that the same intervention will work in PD and in every case a phase III study in PD will be required to confirm efficacy specifically in PD. Difficulties in patient recruitment is another major factor in slowing drug development. This is even more problematic in rare disease, however, recruiting sufficient patient numbers for MSA is still possible as proven by the rifampicin and rasagiline MSA trials [77,81]. Finally, the typical patient diagnosed with MSA is likely to be considerably more disabled than the patient initially diagnosed with PD and it is in newly diagnosed PD patients that disease modifying therapies would be most useful. This dissimilarity in patient severity may limit the generalizability of findings in trials between MSA and PD. However, if disease-modifying efficacy of interventions targeting the effects of α-synucleinopathy lesions can be shown in MSA subjects, this finding would strengthen further therapeutic developments for PD. However, such MSA findings can only be extrapolated to those PD patients who suffer from an αSYN inclusion pathology.

Disclosures

Florian Krismer, Kurt A. Jellinger, Sonja W. Scholz and Angelo Antonini report no conflicts of interest. Klaus Seppi served on advisory boards for Astra Zeneca, Teva, received payment for lectures including service on speakers bureaus: UCB, GlaxoSmithKline, Boehringer-Ingelheim, Lundbeck, AOP Orphan Pharmaceuticals AG, Movement Disorder Society and received funding from Michael J. Fox Foundation, Medical University Innsbruck, Oesterreichische Nationalbank, FWF Austrian Science Fund. Nadia Stefanova received honoraria for consultancies from Biogen and Merz and received research funding from the Austrian Science Fund; Astra Zeneca; Lundbeck. Werner Poewe has received personal fees from AbbVie, Astra-Zeneca, Teva, Novartis, GlaxoSmithKline, Boehringer-Ingelheim, UCB, Orion, Merck Serono, and Merz Pharmaceuticals (consultancy and lecture fees for Parkinson's disease clinical drug development programmes). Gregor K. Wenning received personal fees from Chelsea Therapeutic, Lundbeck, Teva (consultancy and lecture fees) and research funding from the Medical University Innsbruck, Oesterreichische Nationalbank, FWF Austrian Science Fund.

Study funding

This manuscript was supported by funds of the Austrian Science Fund (FWF): F04404-B19. The present manuscript was not industry-sponsored.