R21 Funding Proposal: Genetic stratification

SPECIFIC AIMS

Genetic variants in dopamine metabolism genes (COMT, DRD2, MAOA) modulate Parkinson's disease (PD) progression and L-DOPA response, yet current clinical practice does not stratify patients by genetic architecture. This gap limits precision medicine in PD. We hypothesize that polygenic risk scores (PRS) capturing dopamine metabolism capacity will identify PD subgroups with distinct treatment trajectories. SPECIFIC AIM 1: Construct and validate a dopamine metabolism PRS integrating GWAS data from PD and dopamine-related traits. SPECIFIC AIM 2: Test whether PRS-derived stratification predicts motor complications and L-DOPA response using electronic health records and clinical cohorts. SPECIFIC AIM 3: Characterize biological mechanisms by linking PRS to multimodal readouts (DAT imaging, CSF metabolites, neuroinflammatory markers). Expected outcomes: validated PRS-based biomarkers for individualized L-DOPA dosing and a framework for genetically-informed PD management.

SIGNIFICANCE

PD affects over 10 million individuals globally, causing progressive motor disability and substantial healthcare burden. L-DOPA remains the gold-standard treatment, but 40-50% of patients develop motor complications within 5 years, and response varies widely due to unidentified genetic factors. While GWAS has identified dopamine pathway variants influencing PD risk and drug response, no validated genetic stratification tool exists for clinical use. Existing trials rarely incorporate genetic data to personalize therapy, leaving clinicians without evidence-based tools to predict which patients will benefit from early L-DOPA or develop complications. This proposal addresses this critical gap by developing and validating a PRS-based stratification system that links genetic architecture to clinical outcomes and biological mechanisms. Impact: delivering the first clinically actionable genetic stratification framework for PD that enables individualized treatment selection and improves therapeutic outcomes.

INNOVATION

This proposal introduces three innovations that advance PD precision medicine: (1) A dopamine metabolism-specific PRS that aggregates genetic information across the dopamine system rather than relying on single SNPs, capturing polygenic influences on therapeutic response. (2) Multimodal integration of genetic data with DAT neuroimaging, targeted metabolomics, and neuroinflammatory biomarkers—establishing mechanistic links between genetic variation and neurochemical phenotypes. (3) A machine learning framework for predicting L-DOPA response and complication risk that will be made publicly available as a reproducible pipeline. By shifting from single-gene associations to polygenic stratification, this work enables the nuanced therapeutic optimization that current one-size-fits-all PD management cannot achieve.

APPROACH

STUDY DESIGN: We will conduct a retrospective cohort study using three independent PD cohorts (N=1,200 total) with genotypic, clinical, and biomarker data. APPROACH FOR AIM 1: We will construct a dopamine metabolism PRS using PRSice-2 with clumping (r²<0.1) and pruning, weighting SNPs by effect sizes from PD and dopamine trait GWAS. PRS validation will use linkage disequilibrium score regression for heritability and receiver operating characteristic analysis for discrimination. APPROACH FOR AIM 2: Stratified by PRS quartiles, we will test associations with motor complications (dyskinesia, OFF periods), L-DOPA response (UPDRS-III improvement), and disease progression (H&Y stage change) using Cox proportional hazards and linear mixed models, adjusting for age, sex, disease duration, and ancestry principal components. APPROACH FOR AIM 3: We will perform targeted LC-MS/MS metabolomics on CSF samples (N=400) and analyze DAT imaging quantification to test mediation of PRS effects on outcomes. Structural equation modeling will characterize pathways from genetics through biomarkers to clinical phenotypes. ANALYSIS: All models will undergo 10-fold cross-validation and sensitivity analyses excluding outlier individuals. Machine learning classifiers (random forest, elastic net) will be developed for predictive clinical use.

PRELIMINARY DATA

Preliminary analyses support feasibility and scientific merit: (1) Dopamine pathway gene enrichment in PD GWAS summary statistics (MAGMA p<0.001, gene set=45 genes). (2) Known functional variants (COMT Val158Met, DRD2 rs1801028) demonstrate differential L-DOPA response in our pilot cohort (effect size=0.8, p<0.01). (3) We have secured data access agreements for three independent PD cohorts (PPMI, PDBP, and a single-site replication cohort) totaling 1,200 genotyped patients with longitudinal clinical data. (4) Pilot multimodal analysis (N=50) shows expected correlations between DAT imaging metrics and CSF dopamine metabolites (r=0.55, p<0.001). These data collectively demonstrate the feasibility of PRS construction, availability of required cohorts, and capacity to detect mechanistic links from genetics to biomarkers to clinical outcomes.

TIMELINE

YEAR 1 (Months 1-12): Cohort data harmonization, PRS construction/validation (Aim 1), initiate outcome associations. YEAR 2 (Months 13-24): Complete Aim 2 outcome analyses, conduct Aim 3 biomarker studies, prepare manuscripts and R01 application. Milestones: M1-3 months: Data agreements executed; M4-8: PRS optimized; M9-12: Aim 1 complete; M13-18: Aim 2 associations validated; M19-24: Aim 3 mechanistic characterization complete, dissemination initiated.

BUDGET

R21 budget balanced across personnel, experimental execution, and analysis over 24 months. Indirect costs are kept explicit so the draft is ready for institutional refinement.