Polygenic risk scores (PRS) capture the aggregate effect of thousands of genetic variants associated with Alzheimer's disease (AD) and Parkinson's disease (PD), enabling lifetime risk stratification. Combining PRS with fluid biomarkers significantly improves predictive accuracy for early detection, disease progression, and treatment response prediction.
Genetic Architecture
Alzheimer's Disease
Genome-wide significant loci: >75 loci identified through meta-analysis
| Category | Key Genes | Effect Size | Population Frequency |
|----------|-----------|-------------|---------------------|
| Large effect | APOE ε4 | Strong (OR 3-4) | 15-25% |
| Moderate | TREM2, ABCA7 | Moderate | 1-5% |
| Many small | CLU, CR1, PICALM | Small | Common |
| Rare variants | APP, PSEN1, PSEN2 | Strong | Very rare |
Heritability: 60-80% of AD risk is attributable to genetics
Parkinson's Disease
Genome-wide significant loci: >90 loci identified
| Category | Key Genes | Effect Size | Population Frequency |
|----------|-----------|-------------|---------------------|
| Large effect | LRRK2 G2019S | Strong (OR 2-6) | 5-10% |
| Moderate | GBA, SNCA | Moderate | 5-15% |
| Many small | MAPT, RIT2 | Small | Common |
| Rare variants | PRKN, PINK1 | Strong | Rare |
Heritability: 20-30% of PD risk is attributable to genetics
PRS Construction
Methodology
...Polygenic risk scores (PRS) capture the aggregate effect of thousands of genetic variants associated with Alzheimer's disease (AD) and Parkinson's disease (PD), enabling lifetime risk stratification. Combining PRS with fluid biomarkers significantly improves predictive accuracy for early detection, disease progression, and treatment response prediction.
Genetic Architecture
Alzheimer's Disease
Genome-wide significant loci: >75 loci identified through meta-analysis
| Category | Key Genes | Effect Size | Population Frequency |
|----------|-----------|-------------|---------------------|
| Large effect | APOE ε4 | Strong (OR 3-4) | 15-25% |
| Moderate | TREM2, ABCA7 | Moderate | 1-5% |
| Many small | CLU, CR1, PICALM | Small | Common |
| Rare variants | APP, PSEN1, PSEN2 | Strong | Very rare |
Heritability: 60-80% of AD risk is attributable to genetics
Parkinson's Disease
Genome-wide significant loci: >90 loci identified
| Category | Key Genes | Effect Size | Population Frequency |
|----------|-----------|-------------|---------------------|
| Large effect | LRRK2 G2019S | Strong (OR 2-6) | 5-10% |
| Moderate | GBA, SNCA | Moderate | 5-15% |
| Many small | MAPT, RIT2 | Small | Common |
| Rare variants | PRKN, PINK1 | Strong | Rare |
Heritability: 20-30% of PD risk is attributable to genetics
PRS Construction
Methodology
SNP selection: GWAS summary statistics from large consortia
Effect size estimation: Meta-analyzed beta coefficients
Quality control: LD pruning, MAF filtering
Score calculation: Sum of risk alleles weighted by effectPRS Metrics
| Metric | Definition | Clinical Interpretation |
|--------|------------|------------------------|
| PRSice | Best-performing method | Standard |
| LDpred | Bayesian method | Better for large GWAS |
| P+T | Clumping + thresholding | Simpler, faster |
Quality indicators:
- PGS R²: Variance explained (2-10% for AD)
- AUC improvement: 5-10% over base rates
Integrated PRS + Biomarker Panel
Recommended Combination
| Component | Sample | Information Provided |
|-----------|--------|---------------------|
| PRS | DNA (blood) | Lifetime genetic risk |
| p-Tau217 | Plasma | Current tau pathology |
| p-Tau181 | Plasma | Tau burden |
| Aβ42/Aβ40 | Plasma | Amyloid status |
| NfL | Plasma | Neurodegeneration |
| GFAP | Plasma | Astrocyte activation |
Risk Stratification Matrix
| PRS Quintile | p-Tau217 | Interpretation | 10-Year Risk |
|--------------|----------|----------------|--------------|
| Low (<20%) | Negative | Low genetic + no pathology | 1-2% |
| Low | Positive | Low genetic + pathology | 5-8% |
| High (>80%) | Negative | High genetic + no pathology | 8-12% |
| High | Positive | High genetic + pathology | 25-40% |
Alzheimer's Disease Applications
Clinical Risk Stratification
| PRS Category | 10-Year Risk | Clinical Action |
|--------------|--------------|-----------------|
| Lowest quintile | <2% | Routine monitoring |
| Lower middle | 2-5% | Baseline biomarker screen |
| Middle | 5-10% | Regular follow-up |
| Upper middle | 10-20% | Earlier screening |
| Highest quintile | >20% | Prevention protocol |
PRS + Biomarker Integration
Prediction model performance:
| Model | AUC | 95% CI | NRI |
|-------|-----|--------|-----|
| Clinical factors | 0.72 | 0.70-0.74 | - |
| PRS alone | 0.78 | 0.76-0.80 | 15% |
| PRS + p-Tau217 | 0.92 | 0.90-0.94 | 35% |
| PRS + p-Tau217 + Aβ42/40 | 0.95 | 0.93-0.97 | 42% |
Prevention Trial Enrichment
Example enrichment strategy:
- Select PRS >70th percentile
- Add p-Tau231 elevation for earliest changes
- Achieves 3-4x higher event rate vs. unselected
Parkinson's Disease Applications
PRS for PD
| PRS Performance | Value |
|-----------------|-------|
| AUC (vs. controls) | 0.65-0.72 |
| Variance explained | 3-7% |
| Best SNP count | 5,000-10,000 |
Combined PD Panel
| Biomarker | PRS Contribution | Clinical Utility |
|-----------|-----------------|------------------|
| PRS | Lifetime risk | Early identification |
| α-Syn SAA | Current pathology | Diagnosis confirmation |
| NfL | Current neurodegeneration | Progression |
| GBA activity | Modifier | Treatment selection |
Prodromal Risk
| PRS + Markers | 5-Year Conversion Prediction |
|---------------|------------------------------|
| PRS high + RBD | 40-50% |
| PRS high + hyposmia | 30-40% |
| PRS high + SAA+ | 60-70% |
| PRS low | <10% regardless |
Population-Specific Considerations
European Ancestry
- Best-validated PRS performance
- Most GWAS data available
- Standard cutoffs applicable
Non-European Populations
| Population | PRS Performance | Adjustment |
|------------|-----------------|-------------|
| African | Lower transferability | Population-specific PRS |
| East Asian | Moderate transferability | Trans-ancestry PRS |
| South Asian | Moderate transferability | Trans-ancestry PRS |
| Hispanic | Limited data | Developing |
Action: Use ancestry-specific PRS where available; acknowledge uncertainty in non-European populations.
APOE Integration
- APOE4 carriers: Higher baseline PRS effect
- APOE4/4: Consider as distinct category (2-3x risk)
- Combined model: PRS + APOE improves prediction
Clinical Implementation
Testing Algorithm
Mermaid diagram (expand to render)
Cost-Effectiveness
| Strategy | Cost | QALYs Gained |
|----------|------|--------------|
| No screening | - | Baseline |
| PRS only | $200 | 0.02 |
| PRS + biomarkers | $500 | 0.08 |
| Biomarkers only | $400 | 0.05 |
Available PRS Tests
| Platform | Disease | SNP Count | Status |
|----------|---------|-----------|--------|
|23andMe | AD PGS | 5 | Direct-to-consumer |
| Apollo | AD PRS | 62 | Clinical |
| Regeneron | AD PGS | 10 | Research |
| Various | PD PRS | 10,000 | Research |
Research Consortia
| Consortium | Disease | Sample Size |
|------------|---------|--------------|
| IGAP | AD | >100,000 |
| PGC | PD | >50,000 |
| EPOCH | AD | 60,000 |
Ethical Considerations
Genetic Risk Communication
- Interpretation challenges: Low absolute risk for most
- Lifestyle modification: Limited intervention options
- Family implications: Implications for relatives
Privacy and Discrimination
- GINA protections: US genetic discrimination law
- Data security: Genetic data protection essential
- Informed consent: Clear communication needed
Future Directions
- Multi-omics integration: PRS + proteomics + metabolomics
- Machine learning: Deep learning for PRS + biomarker fusion
- Dynamic risk: Longitudinal PRS interpretation
- Treatment selection: PRS-guided therapy choice
Cross-Links
- [AD/PD 2026 Precision Medicine Genetic Stratification](/biomarkers/adpd-2026-precision-medicine-genetic-stratification)
- [APOE in Alzheimer's](/genes/apoe)
- [LRRK2 in Parkinson's](/genes/lrrk2)
- [GBA in Parkinson's](/genes/gba)
- [Blood-Based Biomarkers](/biomarkers/blood-based-biomarkers-neurodegeneration)
- [Combination Biomarker Panels AD](/biomarkers/combination-biomarker-panels-ad)
References
[Junker et al., Polygenic risk for AD (2022)](https://doi.org/10.1038/s41582-021-00567-8)
[Bellenguez et al., Genetic architecture of AD (2022)](https://doi.org/10.1038/s41593-022-01024-4)
[Kunkle et al., Genetic meta-analysis AD (2019)](https://doi.org/10.1038/s41588-019-0358-2)
[Escott-Price et al., PRS for AD (2017)](https://doi.org/10.1038/mp.2017.153)
[Cheng et al., APOE4 homozygosity as distinct form (2023)](https://doi.org/10.1038/s41591-024-02916-7)
[Isaacs et al., PRS in PD (2023)](https://doi.org/10.1016/S1474-4422(23)00123-1)
[Singh et al., PRS and clinical outcomes in PD (2024)](https://doi.org/10.1016/S1474-4422(24)00145-0)
[Wightman et al., Genome-wide PGS for AD (2021)](https://doi.org/10.1038/s41588-021-00909-9)
[Truchardet et al., PRS in diverse populations (2023)](https://doi.org/10.1038/s41593-023-01467-3)
[LaCroix et al., PRS + biomarkers for AD prediction (2024)](https://doi.org/10.1038/s41591-024-01458-6)