Abstract
Recent discoveries with respect to the genetic and molecular basis of Parkinson’s disease have led to an increasing recognition of the etiologic heterogeneity of the disorder. As in other complex diseases, it is hoped that this knowledge will allow the identification of novel therapeutic targets that will eventually lead to a more efficient treatment, based on the patient’s individual genetic predispositions.
‘Personalized medicine’ (PM, a term often used interchangeably with ‘individualized medicine’, ‘stratified medicine’, or ‘precision medicine’) is a widely and often controversially discussed concept which states that treatment of many diseases can be more effective and less prone to side effects, if it is specifically tailored to a patient’s “personal”, i.e., genetic, anatomical, physiological, or clinical characteristics (see for example https://www.nih.gov/precision-medicine-initiative-cohort-program).
While some have criticized the term as almost trivial (surgeons have always taken the individual anatomical characteristics of their patients into account), others argue that the concept will change the treatment of complex diseases such as cancers, inflammatory or neurodegenerative diseases, because most of them do not represent true disease entities, but should rather be seen as syndromes with overlapping clinical and pathologic characteristics and heterogeneous etiologies.
A number of success stories have already been written, like the improved treatment of lung cancers based on the detection of mutations in the EGFR gene [1], or the effective treatment of cystic fibrosis caused by a single specific mutation (D551G) by the drug ivacaftor [2]. While these examples highlight the promise of genetic stratification for the development of new causative treatments, it is clear that other aspects of PM approaches, including for example individual lifestyle choices, clinimetric characteristics or patient preferences, may be of equal importance, but beyond the scope of this short commentary.
Parkinson’s disease (PD), like lung cancer, is clearly not a single entity etiologically. While dopamine deficiency is a common feature and thus, dopamine replacement reliably ameliorates symptoms in practically all patients, genetics tells us that underlying causes of dopaminergic neurodegeneration can be diverse, and thus should be expected to benefit from specific targeted interventions more than from a “one size fits all” approach.
However, despite considerable progress in the understanding of the genetic basis of PD, the underlying cause is known in only a small minority of patients. Less than 5% of PD is caused by highly penetrant mutations in single genes with dominant or recessive Mendelian inheritance, and probably a slightly larger proportion carry genetic risk variants of moderate effect, such as the LRRK2 G2019S mutation or mutations of the glucocerebrosidase gene (GBA) [3].
The most obvious candidates for “personalized” treatment are rare monogenic forms of PD. However, in most cases it is still unclear how exactly mutations lead to disease and how this process could be modulated, as transgenic animal models often fail to recapitulate cardinal features of the human disease. The study of patient derived neuronal cells in vitro, differentiated from induced pluripotent stem cells is a promising approach to close this knowledge gap [4].
It is often thought to be disappointing that any “personalized” treatment aimed to correct the effect of a mutation in a monogenic form of PD would benefit only a small number of patients. But this may not be the end of the story. There is a growing body of evidence that suggests that even sporadic patients can potentially be stratified, according to the most prominently affected molecular pathways, for which the monogenic cases may be seen as models. For example, mitochondrial dysfunction, the hallmark of recessive early-onset PD, has been demonstrated in a subset of sporadic late-onset cases, and they might particularly benefit from a strategy aimed at restoring mitochondrial integrity.
A similar scenario appears to evolve for the LRRK2: although the molecular mechanism by which mutations cause or predispose to PD is still not clear, auto-phosphorylation of a serine residue at position 1298 of the LRRK2 protein, which can be measured in urine exosomes, emerges as a biomarker of increased LRRK2 kinase activity which is linked to its pathogenicity [5]. Interestingly, a subset of sporadic PD patients without LRRK2 mutations also have an increased ratio of p1298 / total LRRK2, indicating that they may belong to the same pathogenetic subgroup, and thus might benefit from an LRRK2-modifying treatment, if one becomes available.
Even genetically homogeneous groups of patients may vary considerably in their clinical phenotype, which may be due to additional genetic, epigenetic, or non-genetic modifying factors. Therefore, another goal of PM approaches is the improvement of efficacy and the reduction of side effects of symptomatic treatments [6]. Complications of dopaminergic treatment, including impulse control disorder, disabling dyskinesias or dopa-induced psychosis do not occur in all patients at the same dose. There is evidence that knowledge of genetic variability, in combination with known clinical risk factors, can improve prediction of the occurrence of ICD and fluctuations [7, 8],and thus may guide choice of treatment, while the results on the genetic basis of drug induced psychosis in PD are still more controversial [9]. Similarly, the efficacy of COMT inhibition with entacapone may depend on genetic variability in the COMT gene [10]. These observations still rely on a limited number of relatively small studies, but multiple longitudinal and deeply phenotyped and genotyped PD cohorts have been built in recent years, which may allow to systematically address these questions.
The most promising approach to achieve disease modification in a genetically stratified subgroup may be to address pathways linked to genetic risk factors of moderate effect, like heterozygous GBA-mutations, where the affected pathway is already relatively well understood [11]. These variants are associated, depending on the type of mutation, with an odds ratio (increase of risk to develop PD) between 3 and 15, which roughly translates to a penetrance between 10 and 30% [12]. As the majority of carriers of these variants actually do not develop the disease, either the biological effects of the mutations are not so strong and thus might be overcome by interventions, or there are powerful protective or compensatory mechanisms, which could potentially be exploited, if known. Advantages of focusing on patients with these risk factors is that they are sufficiently common to allow appropriately powered studies and that they hold the promise to allow treatment in pre-motor disease stages. A number of clinical studies are being prepared using different approaches [13, 14]. A fundamental prerequisite for all of them is the availability of sufficiently large patient cohorts which are both characterized clinically and genotyped to a sufficient depth to allow to address emerging questions of genetic heterogeneity and modifiers.
If currently available cohorts and the power of genetics are combined in large collaborative studies,personalized approaches in PD may enter clinical practice sooner than many have anticipated.