GWAS of plasma NfL in East Asian cohort

PRIMARY OUTCOME

plasma NfL levels

EXPECTED OUTCOMES

1. **APOE ε4 dose-response:** Each ε4 allele associated with β=0.12 (95% CI: 0.08-0.16) increase in log-NfL, p=1.2×10⁻¹²; explains ~6% variance in plasma NfL 2. **KLHDC4 sentinel SNP:** rs12345 variant (MAF=0.15) associated with β=0.08 per allele (SE=0.01), p=4.5×10⁻⁹; replicated in Asian meta-analysis (p=2.1×10⁻⁶) 3. **Heritability estimate:** SNP-based heritability h²SNP=0.18 (SE=0.03) on observed scale; λGC=1.02 indicating no residual population stratification 4. **Novel loci:** 3 novel loci reaching genome-wide significance (p<5×10⁻⁸) beyond known APOE/KLHDC4, with eQTL colocalization PP>0.85 in brain tissue 5. **PRS predictive power:** NfL PRS explains R²=9.2% (SE=1.1%) of plasma NfL variance in independent validation cohort (n=800), p=3.4×10⁻¹⁸ 6. **Mendelian randomization:** Causal effect of genetically predicted NfL on AD risk OR=1.34 per SD (95% CI: 1.18-1.52), p=4.7×10⁻⁶, no horizontal pleiotropy (MR-Egger intercept p=0.32) 7. **Conditional analysis:** Two independent signals at APOE locus (rs429358 and rs卫图); KLHDC4 signal remains after conditioning on APOE (residual β=0.07, p=2.3×10⁻⁷) ---

SUCCESS CRITERIA

- **Genomic inflation:** λGC < 1.05 confirming proper control of population stratification - **Replication:** At least 2 of top 5 SNPs replicated in independent East Asian cohorts with p<0.05/5 = 0.01 and consistent direction of effect - **Effect size consistency:** β correlation between discovery and replication >0.7 (meta-telemetry analysis) - **Colocalization:** At least one novel locus shows eQTL colocalization with brain expression (PP>0.8) using COLOC algorithm - **Heritability:** h²SNP significantly >0 (p<0.05) and explains ≥5% of NfLi phenotypic variance - **PRS validation:** PRS built on discovery cohort explains ≥7% variance in plasma NfL in held-out validation cohort (n=400) - **APOE signal:** Robust dose-response relationship confirmed (β per ε4 allele 0.08-0.15, p<1×10⁻⁸) regardless of statistical adjustments

PROTOCOL

### GWAS of Plasma Neurofilament Light Chain (NfL) in East Asian Cohort (K-ROAD) **Objective:** Identify genetic variants associated with plasma NfL levels in East Asian patients with or at risk for Alzheimer's disease, focusing on KLHDC4 and APOE loci. --- ### Phase 1: Cohort Recruitment & Sample Collection (Months 1-6) **Timepoints:** - Screening (T0) - Baseline plasma collection (T0) - 12-month follow-up (T12) **Methods:** - **Study population:** 3,000 East Asian ancestry participants from K-ROAD (Korean Research on Aging and Neurodegeneration) cohort - 1,500 AD cases (NINCDS-ADRDA criteria) - 1,000 MCI due to AD (Janzer criteria) - 500 cognitively normal controls - **Inclusion criteria:** Age 55-85, East Asian ethnicity (self-reported + PCA confirmation), valid plasma sample - **Sample collection:** 30 mL fasting venous blood in EDTA tubes (Becton Dickinson, #366643), plasma isolated within 2 hours by centrifugation (2,000×g, 15 min, 4°C) - **NfL measurement:** Simoa NF-light Advantage Kit (Quanterix, #103187), measured on HD-X analyzer - **DNA extraction:** QIAamp DNA Blood Kit (Qiagen, #51104), ≥1.5 μg DNA per sample, 260/280 ratio >1.8 **Controls:** - Standard pooled plasma control (n=20) run every 96 samples - NIST SRM 1955 (Metrology Institute reference) --- ### Phase 2: Genotyping & QC (Months 4-12) **Methods:** - **Platform:** Illumina Asian Screening Array (ASA) + custom imputation panel - **Quality control thresholds:** - Sample call rate >98% - SNP call rate >99% (autosomes), >95% (sex chromosomes) - MAF >1% (for discovery), >5% (for replication) - HWE p >1×10⁻⁶ - Identity-by-descent Pi-Hat <0.2 - **Ancestry estimation:** PCA with 1000G East Asian reference (EAS superpopulation) - **Relatedness check:** KING algorithm for cryptic relatedness removal **Controls:** - HapMap CEU/CHB/JPT reference samples on each 96-well plate - Duplicates (n=20) for genotype concordance (>99.9% required) - 1% blind duplicates for QC --- ### Phase 3: GWAS Analysis (Months 10-18) **Methods:** - **Software:** Regenie v3.2 (Python 3.9+) - **Association model:** Linear mixed effects model (LMM) - Outcome: log-transformed plasma NfL (continuous) - Covariates: age, sex, APOE ε4 status (binary), PC1-5 ancestry, BMI, eGFR - First-degree relatives excluded or modeled as random effect - **Genomic inflation factor (λGC):** Calculated to assess population stratification; λGC <1.05 acceptable - **Significance threshold:** - Genome-wide: 5×10⁻⁸ (Bonferroni for ~1M independent tests) - Suggestive: 1×10⁻⁵ - **Conditional analysis:** GCTA-COJO for independent signals within ±500 kb of top loci **Specific focus on KLHDC4 region (chr16:85,000,000-87,000,000) and APOE (chr19:44,400,000-45,200,000):** - eQTL colocalization with brain NfL expression (GTEx v8) - Fine-mapping with SuSiE (PIP >0.8) - Rare variant burden test (SKAT-O) in K-ROAD whole-exome sequences (n=1,200) --- ### Phase 4: Replication & Functional Follow-up (Months 16-24) **Replication cohorts:** - Korean AD cohort (n=800) - Japanese ADNI (n=400) - Chinese CHARGE consortium subset (n=1,200) **Methods:** - **Genotyping confirmation:** TaqMan SNP assays (Applied Biosystems) for top 20 signals - **eQTL analysis:** cis-eQTL mapping in brain tissue (n=500, ROS/MAP dataset) - **Proteomic validation:** SomaScan v4 platform for NfL-associated proteins **Controls:** - European ancestry samples (n=500) to test跨-ancestry effect size correlation (β correlation >0.7) --- ### Phase 5: Bioinformatics Integration (Months 20-28) **Methods:** - **Pathway analysis:** MAGMA v1.08, gene-set enrichment from KEGG/Reactome - **Mendelian randomization:** Two-sample MR using IEU-Bristol GWAS catalog (NfL: GCST90239039) - **Proteome-wide:** PRS for NfL using pruned SNPs (r²<0.1, p<0.05) - **Cell-type specificity:** LDSC-SEG with single-cell RNA-seq from human brain **Cross-validation:** - Polygenic risk score (PRS) computed at P-thresholds 0.001, 0.05, 0.1, 0.5 - R² explained by PRS >8% for top model --- ### Phase 6: Validation & Reporting (Months 26-30) **Measurement endpoints:** - Primary: β (effect size) per copies of risk allele - Secondary: OR for AD risk per NfL-associated SNP - Exploratory: Heritability h²SNP from GREML **Outcome measures:** - Plasma NfL: pg/mL (expected range 5-80 pg/mL in controls, 20-200 pg/mL in AD) - Genetics: allele frequencies, OR, β, SE, p-value - Bioinformatics: colocalization posterior probability, MR β, PIP ---

LINKED HYPOTHESES

HTML iframe

<iframe src="http://scidex.ai/artifact/exp-12e37301-cace-448f-a5e5-d03822faf9ef?embed=1" width="100%" height="600" style="border:0;border-radius:8px"></iframe>

Markdown link

[GWAS of plasma NfL in East Asian cohort](http://scidex.ai/artifact/exp-12e37301-cace-448f-a5e5-d03822faf9ef)

Direct URL

http://scidex.ai/artifact/exp-12e37301-cace-448f-a5e5-d03822faf9ef

Preview