medRxivGenetic & Genomic MedicinePreprint
Novel Parkinson's Disease Genetic Risk Factors Within and Across European Populations
The Global Parkinson’s Genetics Program (GP2), Hampton L. LeonardiD
Abstract
IntroductionWe conducted a meta-analysis of Parkinsons disease genome-wide association study summary statistics, stratified by source (clinically-recruited case-control cohorts versus population biobanks) and by general European versus European isolate ancestries. This study included 63,555 cases, 17,700 proxy cases with a family history of Parkinsons disease, and 1,746,386 controls, making it the largest investigation of Parkinsons disease genetic risk to date. MethodsMeta-analyses were performed using standard fixed and random effect models for the European sub-populations, the case-control studies, and the population biobanks separately. Finally, all of the European ancestries for all study types as well as proxy cases were combined in our final cross-European meta-analysis. We estimated heritable risk across ancestry groups, investigated tissue and cell-type enrichment, and prioritized risk genes using public data to facilitate functional follow-up efforts. ResultsThe final combined cross-European meta-analysis identified 134 risk loci (59 novel), with a total of 157 independent signals, significantly expanding our understanding of Parkinsons disease risk. Multi-omic data integration revealed that expression of the nominated risk genes are highly enriched in brain tissues, particularly in neuronal and astrocyte cell types. Additionally, we prioritized 33 high-confidence genes across these 134 loci for future follow-up studies. ConclusionsBy integrating diverse European populations and leveraging harmonized data from the Global Parkinsons Genetics Program (GP2), we reveal new insight into the genetic architecture of Parkinsons disease. We identified a total of 134 risk loci, expanding the number of known loci associated with PD by approximately 24%. We also provided an initial layer of biological context to these results through follow-up analyses in an effort to facilitate follow-up studies and precision medicine efforts with the goal of advancing Parkinsons disease research.
- organism: human
- assay: GWAS summary statistics
- research focus: GWAS / association mapping
- research focus: fine-mapping
- approach: statistical model
- approach: matrix factorization
In-depth summary
Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 1 (17 Mar 2025) · CC0 1.0
A cross-European Parkinson's disease GWAS meta-analysis of 63,555 cases, 17,700 proxy cases and 1,746,386 controls finds 134 risk loci (59 novel), 157 independent signals, and brain-enriched risk genes.
Earlier collaborative Parkinson's disease GWAS meta-analyses had nominated 102 risk loci across ancestries. This one adds Global Parkinson's Genetics Program (GP2) cohorts, all genotyped on the NeuroBooster array and processed with one pipeline, and separates general European, Icelandic, Finnish and Ashkenazi Jewish strata, and clinical versus biobank ascertainment, to compare effects and heritability.
Key results
ShownThe joint cross-European meta-analysis yields 134 genome-wide significant Parkinson's disease loci, 59 of them novel, and conditional analysis raises the count of independent signals to 157.
Loci at P < 5e-8 with FUMA LD clumping (r^2 > 0.6, 250 kb windows); novel = more than ±250 kb from loci in Nalls et al., Foo et al. or Kim et al. COJO conditioning (GP2 European reference, 37,916 individuals) added 23 signals, including 3 coding signals at GBA1, 3 at TMEM175 (2 before) and 3 extra at LRRK2. Scaled genomic inflation stayed below 1.03. · Abstract; GWAS and Meta-analyses; Identification of novel and independent loci…
ShownMost previously reported loci replicate at genome-wide significance, and effect sizes at earlier lead variants agree closely with the 2019 meta-analysis by Nalls et al.
Replicated at P < 5E-8 within ±250 kb: 66/78 loci from Nalls et al. (84.6%), 10/11 from Foo et al. (90.9%) and 65/78 from Kim et al. (83.3%). Most loci that did not replicate still showed association (75 loci at 5E-6, 83 at 5E-4). Effects at the 90 Nalls et al. lead variants correlate with the joint analysis at r = 0.97 (P = 3.94E-52). · Replication in previous GWAS studies
ShownEffects vary across European strata. Genetic correlation is high between clinical and biobank meta-analyses and between general European, Finnish and Icelandic data, but low between general European and Ashkenazi Jewish data.
LDSC rG, clinical vs biobank within EUR 0.892 (SE 0.0372); EUR vs FIN 0.8672 (SE 0.0726); EUR vs ICE 0.9988 (P = 0.0184); EUR vs AJ 0.4058 (P = 0.0048); positive control GP2 vs IPDGC EUR 0.9866. In the joint analysis 38 loci had I2 > 30% (28.79% of loci, vs 15.6% in Nalls et al. 2019). An Icelandic-only hit near PTTG (rs10031495, P = 1.32E-08) did not replicate jointly (P = 0.1013). · Identification of novel and independent loci…; Variant associations and heritability are…; DISCUSSION
SuggestedLiability-scaled SNP heritability differs across strata and is lower in biobank than in clinical case-control data. The Ashkenazi Jewish estimate is the highest but very imprecise, and the authors link it to a strong LRRK2 signal.
Liability-scaled h2: AJ 28.13% (SE given as 25.08 and, scaled, 16.44), EUR 8.5% (SE 0.0062), FIN 7.67%, ICE 1.54%; biobank estimates fall below clinical case-control ones. The discussion cites a roughly 2-fold larger LRRK2 effect in AJ; the low ICE value is attributed to rare LRRK2, GBA1 and ITSN1 variants that LDSC does not capture. · Variant associations and heritability are…; DISCUSSION
SuggestedRisk genes are enriched for brain expression, mainly in neurons and astrocytes, and gene prioritization nominates 32 high-confidence candidates for follow-up (the abstract says 33).
FUMA: fourteen brain tissues enriched after multiple-testing correction (mainly cerebellum and frontal cortex; also substantia nigra, spinal cord), none outside the brain. Neurons predominate among cell types; astrocytes, with GABAergic neurons, in human prefrontal cortex. Genes chosen: Bonferroni-significant in MAGMA and PoPS ≥ 90th percentile. · Abstract; Prioritization of tissues, cell types, and genes…; Functional mapping identifies genes underlying…
Methods, data and reproducibility
Per-cohort GWAS (Plink2; covariates were principal components, sex and age where available) on GP2, IPDGC, UK Biobank and Fox Insight genotypes, plus FinnGen, deCODE and MVP summary statistics, were combined by fixed- and random-effects meta-analysis (Plink1.9) per ancestry stratum, per ascertainment type and jointly. Follow-up used COJO conditioning, LDSC, FUMA/MAGMA enrichment and PoPS gene prioritization.
- Final cross-European joint meta-analysis (21 studies; EUR, ICE, FIN and AJ strata; clinical and biobank ascertainment, plus proxy cases) · human · n = 63,555 cases, 17,700 proxy cases, 1,746,386 controls
- GP2 cohorts genotyped on the NeuroBooster array, quality-controlled with GenoTools and imputed with TOPMed-r3 · human · n = 22,048 cases (26.98%) in the final meta-analysis
- IPDGC clinical case-control datasets, with samples duplicated in or related to GP2 removed (KING); EUR-predicted samples only · human
- UK Biobank case-control GWAS (code p42032) and parental-history proxy-case GWAS, EUR-predicted samples · human
- FinnGen release 11 summary statistics, G6_PARKINSON endpoint (labelled FIN by country of origin) · human
- deCODE Genetics summary statistics for clinically defined Icelandic cases (labelled ICE) · human
- Fox Insight Genetics Study cases (a subset of 23andMe PD samples) matched to NCI population controls on the Illumina GSA array · human
- Million Veteran Program PD summary statistics, European ancestry as estimated by MVP · human · phs002453.v1.p1
- Methods & tools
- Plink2, Plink1.9, GenoTools, KING, TOPMed-r3 imputation reference panel, NeuroBooster array, FUMA, MAGMA, COJO (Genome-wide Complex Trait Analysis), LDSC, Ensembl VEP, PoPS
- Data & accessions
- https://ndkp.hugeamp.org/research.html?pageid=a2f_downloads_280
- Software & environment
- Plink2 (per-cohort GWAS) and Plink1.9 (meta-analysis); TOPMed-r3 imputation panel; FinnGen release 11. Software identifiers are stated to be listed in the GitHub repository, archived on Zenodo (DOI 10.5281/zenodo.15013321).
- Reproducibility gaps
- Released data are ancestry-level summary statistics on NDPK; access to individual-level data is not described. One availability section says code awaits GP2 DCD review, the other links the GitHub repository. 23andMe summary statistics were excluded (data sharing limits).
Limitations (6)
- Only samples of European ancestry are analysed; the authors expect upcoming GP2 studies to be more inclusive as datasets grow. · stated by the authors
- LDSC heritability relied on a European reference panel that represents AJ, FIN and ICE poorly. LDSC can also be unstable and biased at low sample sizes, and the AJ estimate has a large SE and needs follow-up. · stated by the authors
- Random-effects models can be conservative when heterogeneity is driven by ancestry, so loci significant only under fixed effects may be less generalizable across ancestries. · stated by the authors
- Some samples could not be re-genotyped or resequenced, which may add noise despite the uniform single-array processing of GP2 samples. · stated by the authors
- No independent replication cohort is described for the 59 novel loci; all available samples went into the discovery meta-analysis. · noted in this summary
- Some counts are inconsistent: 33 prioritized genes in the abstract vs 32 in results and discussion, and 102 vs 92 previously known loci in the introduction and the replication section. · noted in this summary
Read this if you need updated Parkinson's disease risk loci from European cohorts, Icelandic, Finnish and Ashkenazi Jewish comparisons, or prioritized genes and summary statistics for functional follow-up.
- Citations
- 61
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Multi-ancestry genome-wide meta-analysis in Parkinson's disease
Jonggeol Jeffrey Kim, Dan Vitale, … Piu Chan·Nature GeneticsPeer-reviewed·Dec 28, 2023
- Expanding Parkinson’s disease genetics: novel risk loci, genomic context, causal insights and heritable risk
Mike A. Nalls, Cornelis Blauwendraat, … Ciara García·The Lancet NeurologyPeer-reviewed·Nov 6, 2019
- The Global Landscape of Genetic Variation in Parkinson's disease: Multi-Ancestry Insights into Established Disease Genes and their Translational Relevance
Lara Mariah Lange, Zih‐Hua Fang, … the Global Parkinson’s Genetic Program (GP2)·medRxiv·Jul 8, 2025
- Shared and Sex-Specific Genetic Risk for Parkinson's Disease Risk Across European Populations
the Global Parkinson’s Genetic Program (GP2), Hampton L. Leonard·medRxiv·Jan 2, 2026
- Insights Into Parkinsons Disease Genetics in African Populations: Expanded GWAS Identifies Ancestry-Specific and Cross-Population Risk Loci
N. Okubadejo, O. O. Ojo, … Asli·medRxiv·Mar 3, 2026
- Genetic Modifiers of Age at Onset for Parkinson's Disease in European Populations
the Global Parkinson’s Genetic Program (GP2), Erin R. Voss·medRxiv·Oct 26, 2026