From Genomic Puzzles to Network Solutions: Charting Parkinson’s Path using Systematic Review/ Meta Analysis as the Compass

In Europe about 1% of people over the age of 60 are diagnosed with Parkinson’s disease (PD) with a tendency to increase, representing a major socioeconomic challenge ( 1, 2). PD is characterized by impaired movement control and progressively degenerating neurons in the substantia nigra of the brain. As a result, the amount of dopamine in the corpus striatum is lowered. Treatment options mainly center around substituting dopamine, a principle introduced in the 1960’s. Since then, no innovation has occurred. We question whether PD is a single disease. We hypothesize that PD’s comorbidities such as cognitive impairment, neuropsychiatric and autonomic symptoms and other comorbidities may lead towards a much better understanding of the presumably different underlying causal mechanisms. Histologically, PD is correlates with the accumulation of alpha-synuclein aggregates called Lewy Bodies that may spread in a characteristic fashion within the brain ( 5). Lewy Bodies are thought to contribute to the loss of dopaminergic neurons in the substantia nigra. However, whether this is a uniform and causal mechanism, its origin and role in disease progression are unclear ( 6, 7). Not knowing the underlying causality makes both early-stage diagnosis of PD and curative treatments impossible ( 8). Here we propose to define so-called disease modules ( 9) causally involved in different phenotypes of PD. Beginning with identified risk or PD associated genes, the resulting protein-protein interaction modules within the interactome, different signaling modules are constructed ( 10-12). The crucial point for building those disease modules is to identify the right set of seeds to begin with. This appears to be quite challenging due to the arbitrariness of different disease definitions and differences in genetic databases even for uniform disease ontologies. Even when using the same disease term, different databases including the Comparative Toxicogenomics Database (CTD), DisGeNET, OMIM, GWAS catalog, Open Target Genetics or Platform (OTG, OTP), the Catalogue of Somatic Mutations in Cancer (COSMIC) and the Cancer Genome Atlas Program (TCGA), yield different results, and thus suggesting entirely different starting points to define the disease modules. Using the PD term, six of these databases contain different number of genes: 129 (OTG), 386 (GWAS catalog), 435 (OTP), 474 (OMIM), 2,185 (DisGeNET) and 28,574 (CTD). Intersecting them shows generally low overlap between different databases and only 5 genes are shared across all databases. However, we know from other examples that even the consensus among different databases is misleading when compared to the gold standard, a systematic review followed by quantitative meta-analysis (SR/MA). Therefore, we propose to use a SR/MA of the entire literature also for PD to identify and rank disease genes, that can be further utilized for network analyses. In addition, the SR/MA can be further used to evaluate different databases.