A prognostic view on the application of individualized genomics in Parkinson’s disease

Introduction

Since the original description of Parkinson’s disease (PD), and the landmark essay by James Parkinson in 1817 surprisingly little has changed with regards to the clinical features of the disease [1]. Almost two centuries later however we have a much better understanding of the natural history and pathophysiology of the disease [2, 3]. The characterization of the Lewy body as the pathologic lesion observed in the brain of patients with PD and the development of dopamine replacement therapies (levodopa) was the stand out advancements up until the last few years of the twentieth century [4, 5]. As we moved towards the millennium we entered a new era of research on PD which has been driven by genetic discovery. In fact over the last decade, this era of ‘genomics’ has directed both functional biology and in vitro/in vivo modeling of the disease and as we progress will be critical for diagnostic, prognostic and therapeutic outcomes.

The diagnosis of PD relies on the clinical presentation of three of the four cardinal clinical signs; bradykinesia, rest tremor, rigidity and postural instability [6-8]. In addition, a positive response to dopamine replacement therapy is required. These criteria will lead to either a probable or possible diagnosis with definite PD requiring the presence of Lewy body pathology in the surviving neurons of the substantia nigra (referred to using the neuropathological definition as brainstem Lewy body disease; LBD) [9]. Given the many overlapping features with other forms of parkinsonism (PD is the most common form of parkinsonism) such as progressive supranuclear palsy (PSP) and multiple system atrophy (MSA), the clinical diagnosis of PD is not exact [10]. In addition, of course, waiting for the neuropathologic confirmation does not benefit the patient. The application of genetic testing is one approach that although in its infancy may circumvent many of the diagnostic issues and provide the neurologist with a clear path to therapeutic intervention.

Genetic discrimination of those individuals at-risk will be crucial if the degeneration that is observed at the end-stage of the disease begins many decades before the predominant clinical presentation [11]. There are many clinical signs that predate the movement disorder component of PD, in fact upwards of 60% of dopaminergic neurons are lost prior to the presentation of the movement disorder phenotype [12]. These early signs can include anosmia, constipation and REM sleep behavior disorders, however the diagnostic utility of these signs remains controversial as they are present in many other disorders and are not specific to PD [13]. Imaging provides an additional weapon in the arsenal of the neurologist, unfortunately visualization of α-synuclein aggregation is unavailable in vivo at present. The development of amyloid-β plaque (Pittsburgh compound B; PiB) imaging in Alzheimer’s disease (AD) has provided insight into the early stages of neurodegenerative disease. Studies suggest that the initial pathology can be observed approximately 25 years before the clinical onset of the disease [11]. If this holds true in PD it may indicate that disease intervention may need to be performed decades before the movement disorder manifests and present a major obstacle in treatment. Genetic susceptibility profiles may provide one mechanism whereby at-risk individuals can be identified in the prodromal stage, monitored and treated.

To apply clinical genetics to a disease one must first however resolve the variation within the genome that is determining the individual susceptibility to disease. In the context of PD, genetic factors were long thought to play a minimal if any role given the largely sporadic nature of the disease [14]. However we have witnessed a paradigm shift in the field following the identification of the first pathogenic mutation causing an autosomal dominantly inherited form of PD in the gene encoding the α-synuclein protein (SNCA; Table 1) [15]. Subsequently the α-synuclein protein was demonstrated to be the major protein component of the Lewy body which places α-synuclein at the center of PD research [4]. Over the following years it became clear that SNCA point mutations families are rare. The identification of genomic multiplication of the SNCA gene as a dose dependent mechanism of disease postulated the theory that over-expression of the wild-type α-synuclein protein is sufficient to cause both familial and sporadic forms of PD[16, 17]. Indeed subsequent population based genetic studies have shown that common variation at the SNCA locus is a risk for sporadic disease [18, 19]; although caveats remain in identification of the true toxic α-synuclein species [20]. These findings have nominated SNCA knockdown approaches as a viable therapeutic target and the multiplication families may provide an ideal patient group for initial clinical trials [21].

Following the identification of SNCA as a cause of autosomal dominant PD, three genes were identified to cause early-onset forms of autosomal recessive parkinsonism that clinically reflect PD (PARKIN, PINK1 and DJ-1)[22]. Given that only approximately ten to fifteen percent of patients with PD present under the age of 45 years, carriers of mutations within these genes remain few, however clinical genetic testing is available to the neurologist and can be utilized in the diagnosis. An issue within the setting of PD has been the elucidation of common pathways involving the mutated genes; and while many cellular pathways have been implicated evidence has been unconvincing [23]. Recently a novel mechanism for disease in PARKIN and PINK1 mutation carriers has been proposed involving the identification and clearance of damaged mitochondria via mitophagy, a specific form of autophagy [24-26]. Studies have shown that PINK1 (a mitochondrial kinase) recruits Parkin (an E3 ubiquitin ligase) to damaged mitochondria for ubiquitination and then targeted clearance [27]. Although the role of this pathway in the more frequent late-onset form of PD remains unclear, this type of functional readout may be crucial in determining pathogenicity of rare variants.

Perhaps the most clinically relevant gene associated with PD identified to date was found in 2004 [28, 29]. The identification of mutations within the leucine-rich repeat kinase 2 gene (LRRK2) was notable for its clinical and pathologic presentation which is indistinguishable from typical late-onset sporadic PD [30], and the frequency of its most studied mutant the LRRK2 G2019S substitution [31-33]. The LRRK2 G2019S mutant accounts for between 5-6% of familial and 1% of sporadic patients in the North American Caucasian population [34]. Of note, in the Ashkenazi Jewish population and the Berber Arab populations of North Africa this figure can increase upwards of 15% and 40% respectively [35, 36]. The substitution is predicted to increase the kinase activity of the protein and thus provides rationale for drug development [37, 38]. As with SNCA a number of subsequent studies have identified common population variants in the LRRK2 gene which affect the individual’s risk of developing disease [39, 40]. Our study last year working with the global epidemiology of PD consortium (GEO-PD) identified a number of novel risk factors and evidence to support the presence of a protective LRRK2 haplotype [41].

As has been observed for the majority of pathogenic familial mutations, age-at-onset widely varies and as with any late-onset disorder the phenomenon of age-related reduced penetrance is a major clinical issue. Penetrance is modified by other genetic or disease-modifying agents (environmental, stochastic) and is a characteristic feature of PD, observed for both SNCA and LRRK2 mutation carriers [42, 43]. Indeed healthy septuagenarian and octogenarian carriers of SNCA duplications and LRRK2 G2019S have been reported [44, 45]. This wide spread in age-at-onset can be anywhere from 35-95 years in the case of LRRK2 G2019S, and highlights the potential to modify the penetrance of the disease in carriers. These observations provide tangible hopes in the development of disease-modifying therapies if they can be delivered to the correct patients at the correct intervention time in early-stage disease.

The vast majority of patients with PD report no family history and manifest a disease which is sporadic in nature. However, the genetic studies on familial forms of disease have also borne fruit in the nomination of common variants influencing risk of sporadic disease. In fact the most significant findings from population-based association studies nominate common variation in SNCA and microtubule-associated protein tau (MAPT) in which familial mutations have been shown to cause frontotemporal dementia linked to chromosome 17 with tau pathology [46]. Early candidate gene association studies also highlighted variation in SNCA and MAPT [47-49]. For many patients it may be the joint effects of many low penetrant risk and protective factors that determine the individual risk of developing PD. For example a patient with the SNCA risk variants may also harbor the LRRK2 protective haplotype with the combined sum of the effect reflecting the overall risk. Our recent study examined the independent and joint effects of SNCA and MAPT risk variants with members of the GEO-PD consortium [50]. The results suggest the genetic risk alleles at these two loci are independent with an individual harboring both at an increased risk in an additive manner. Identifying the common variants has been a major challenge to the field, with only recent large-scale studies helping to define the genetic architecture of PD.

Early genome-wide association (GWA) efforts were on the whole disappointing; in 2011, the first large meta-analytical approach in GWA studies of PD was published nominating eleven loci (including SNCA, LRRK2 and MAPT) as risk determinants (Table 2) [46, 51]. For the majority of association peaks no individual gene or functional variant/s could be defined therefore measures of population-based risk should be treated with caution. The clinical utility of such observations is yet to be realized and may be irrelevant without the resolution of the specific variants driving the signal. These regions are large, contain numerous genes and given that they were identified in population-based series rather than families make it difficult to select only a small number of patients and controls for sequencing, as functional variants are likely present in both groups but not in everyone. Without a better understanding of the genome-wide susceptibility factors it will be impossible to exploit the advantages of clinical genetic screening or to inform the patient what a positive test predicts regarding the potential of disease, i.e. life-time risk, age-at-onset, clinical phenotype.

Genetic studies are also likely to inform therapeutic intervention strategies and dosage. Unfortunately the current medicines are focused on symptomatic alleviation (e.g. levodopa, dopamine agonists), with no effect on disease progression. As PD is a late-onset disease for the majority of patients (average age-at-onset is approximately 65 years), a therapy which can delay the onset by 10-20 years may be a considerable benefit to many individuals. In addition, presently patients can develop ‘wearing off’ with L-dopa therapy which requires higher doses which results in increased side-effects. The development of dyskinesia, dopamine dysregulation syndrome and impulse control disorders are likely influenced by genetic variation. Identifying these variants and subsequently these individuals at risk will allow the design of tailored drug treatment plans for the individual patient.

The advances in genetics of PD over the last 15 years have been remarkable, however as we enter this new era of genomic medicine the applications of next-generation sequencing approaches including whole-genome sequencing will accelerate gene discovery and provide a better understanding of genetic variation in disease. We witnessed the first gene identified by exome sequencing (exome- the 1% of the genome encoding proteins) in 2011, with two independent groups nominating a mutation in the VPS35 gene as a cause of autosomal dominantly inherited PD [52, 53]. These studies have demonstrated the potential of genetic sequencing and as the cost of these approaches is reduced an ever increasing number of individuals will arrive at the clinician’s door with their personal genome in hand [54].

As we stand at the epoch of individualized medicine driven by genomic profiling we will bear witness to a transformation in patient care (Table 3). Identifying the specific variants that are driving risk of disease, whether through family- or population-based approaches, will act as determinants of diagnosis and treatment [54]. The complex etiology of PD susceptibility involving multiple risk and protective genetic factors acting in unison will likely reflect the therapeutic approach necessary to intervene. For example, those individuals harboring a SNCA driven risk may need a knockdown therapy, those with LRRK2 may need a kinase inhibitor and those with both may need a combination of the two. Most likely these therapeutics with specific targets will form the underlying basis of personalized medicine and the genetic findings in PD will help direct these approaches. However before these drugs can be developed we need a clearer resolution of the underlying functional pathomechanisms of the disease at the cellular level to ensure the correct drug targets and reduce side-effects.