Lysosomal polygenic risk score in Parkinson's disease across populations.

Technical Summary: Lysosomal Polygenic Risk Score in Parkinson's Disease Across Populations

Problem Statement
Parkinson's disease (PD) is a clinically heterogeneous neurodegenerative disorder with complex etiology involving genetic, environmental, and aging factors. While genome-wide association studies (GWAS) have identified numerous risk variants, they explain only a fraction of heritability. Lysosomal dysfunction is a known pathogenic mechanism in PD, particularly involving the GBA1 gene. However, the contribution of common genetic variation within the broader lysosomal pathway (excluding specific pathogenic GBA1 mutations) to PD risk and clinical phenotypes remains under-explored, especially across diverse ancestral populations. Existing polygenic risk scores (PRS) often lack transferability across ancestries, and pathway-specific PRS models for lysosomal function have not been comprehensively evaluated in large, multi-ancestry cohorts.

Methodology
The authors constructed a lysosomal polygenic risk score (lyso-PRS) and evaluated its performance using data from the Global Parkinson's Genetics Program (GP2) Release 11.

• Cohorts: The study analyzed 32,054 idiopathic PD (iPD) cases and 18,188 healthy controls (HC) across ten distinct ancestries (European, African American, African, Ashkenazi Jewish, Latino/Indigenous Americas, Central Asian, East Asian, Middle Eastern, South Asian, and Complex Admixture History).
• Genetic Data: Genotypes were derived from the Neurobooster array, imputed and quality-controlled using Genools.
• Phenotype Definition:
  • iPD: Defined by excluding individuals with known pathogenic variants in established PD genes (including GBA1, LRRK2, SNCA, etc.).
  • GBA1-PD: Individuals with pathogenic/likely pathogenic GBA1 variants.
  • Non-Manifesting Carriers (NMC): Individuals with GBA1 variants but no PD diagnosis.
• PRS Construction:
  • Lysosomal genes were defined using the Molecular Signatures Database (MSigDB), yielding 519 unique genes. GBA1 and LRRK2 were excluded from the gene set to isolate the polygenic burden of the remaining lysosomal pathway.
  • Allele weights were derived from 2025 European PD GWAS summary statistics.
  • Variants with minor allele frequency (MAF) < 0.01 were excluded. After filtering, 217,053 SNPs were retained.
  • The score was calculated using PRSice-2 with linkage disequilibrium (LD) clumping ($r^2 = 0.1$, 250kb).
• Statistical Analysis:
  • Performance was assessed via Area Under the Curve (AUC), Odds Ratios (OR), and balanced accuracy.
  • Associations with clinical outcomes (age at onset, UPDRS scores, cognitive function) were tested using regression models adjusted for sex, age, and principal components.
  • The European cohort was split into a training set (European-GP2) and an independent validation set (European-TUEPAC). Non-European cohorts utilized default p-value thresholds due to limited sample sizes.

Key Results

1. Ancestry-Specific Performance:

   • The lyso-PRS demonstrated the strongest predictive performance in European populations. In the European-GP2 training cohort, the AUC was 0.602 ($p = 2.02 \times 10^{-172}$), and in the European-TUEPAC validation cohort, it was 0.632 ($p = 5.34 \times 10^{-20}$).
   • Performance was significantly lower in non-European populations (AUC range: 0.51–0.59). The lowest performance was observed in African ancestry (AUC = 0.509).
   • In non-European groups, lyso-PRS scores were generally higher in iPD than HC, but significance was not reached in African American and African cohorts.

2. Differentiation of PD Subtypes in Europeans:

   • The lyso-PRS successfully distinguished GBA1-PD from both healthy controls (AUC = 0.621) and non-manifesting carriers (AUC = 0.621).
   • However, the score showed weak discriminatory power between GBA1-PD and idiopathic PD (iPD) (AUC = 0.515), suggesting the score captures general PD risk rather than GBA1 mutation status specifically.

3. Clinical Associations (European Cohorts):

   • Age at Onset: Higher lyso-PRS was associated with earlier age at onset in the European-GP2 cohort ($\beta = -0.21$, $p = 0.022$), though this was not replicated in the European-TUEPAC cohort ($p = 0.14$), despite consistent effect direction.
   • Motor Symptoms: Higher lyso-PRS was associated with worse motor experiences of daily living (UPDRS-II) and motor severity (UPDRS-III) in the training cohort, but these findings were not replicated in the validation cohort.
   • Motor Complications: A consistent positive association was found between higher lyso-PRS and UPDRS-IV (motor complications) in both the training ($\beta = 0.32$, $p = 0.0088$) and validation ($\beta = 0.22$, $p = 0.045$) cohorts.
   • Cognition: In the validation cohort, higher lyso-PRS was associated with worse cognitive performance ($\beta = -0.17$, $p = 3.58 \times 10^{-3}$), whereas no significant association was found in the training cohort.
   • Sex: No significant association was found between lyso-PRS and sex in either cohort.

Significance and Claims
The authors claim this study represents the largest and first comprehensive evaluation of lysosomal polygenic risk across ten populations. The primary significance lies in demonstrating that:

• Pathway-Specific Risk: Common genetic variation in the lysosomal pathway (independent of GBA1 mutations) contributes to PD susceptibility and can stratify risk, distinguishing GBA1-PD from non-manifesting carriers.
• Additive Risk Model: The findings suggest an additive effect where GBA1 mutations and the broader lysosomal polygenic burden independently increase PD risk.
• Clinical Relevance: Lysosomal polygenic burden may influence specific clinical features, particularly motor complications (UPDRS-IV), potentially linking lysosomal dysfunction to treatment-related motor issues.
• Ancestry Limitations: The study highlights a critical disparity in PRS performance across ancestries, attributing reduced accuracy in non-European populations to the reliance on European GWAS summary statistics and differences in linkage disequilibrium patterns. The authors emphasize the urgent need for population-specific GWAS data to improve polygenic risk prediction accuracy globally.

The paper concludes that while lyso-PRS shows promise for biological stratification and identifying high-risk patients for lysosomal-targeted therapies, its current utility is limited to European populations, and further large-scale, diverse studies are required to validate clinical associations and improve cross-ancestry transferability.