Characterizing the Genetic Architecture of Parkinson's Disease in Latinos

Abstract

Objective: This work was undertaken in order to identify Parkinson's disease (PD) risk variants in a Latino cohort, to describe the overlap in the genetic architecture of PD in Latinos compared to European-ancestry subjects, and to increase the diversity in PD genome-wide association (GWAS) data.

Methods: We genotyped and imputed 1,497 PD cases and controls recruited from nine clinical sites across South America. We performed a GWAS using logistic mixed models; variants with a p-value <1 × 10^-5 were tested in a replication cohort of 1,234 self-reported Latino PD cases and 439,522 Latino controls from 23andMe, Inc. We also performed an admixture mapping analysis where local ancestry blocks were tested for association with PD status.

Results: One locus, SNCA, achieved genome-wide significance (p-value <5 × 10^-8 ); rs356182 achieved genome-wide significance in both the discovery and the replication cohorts (discovery, G allele: 1.58 OR, 95% CI 1.35-1.86, p-value 2.48 × 10^-8 ; 23andMe, G allele: 1.26 OR, 95% CI 1.16-1.37, p-value 4.55 × 10^-8 ). In our admixture mapping analysis, a locus on chromosome 14, containing the gene STXBP6, achieved significance in a joint test of ancestries and in the Native American single-ancestry test (p-value <5 × 10^-5 ). A second locus on chromosome 6, containing the gene RPS6KA2, achieved significance in the African single-ancestry test (p-value <5 × 10^-5 ).

Interpretation: This study demonstrated the importance of the SNCA locus for the etiology of PD in Latinos. By leveraging the demographic history of our cohort via admixture mapping, we identified two potential PD risk loci that merit further study. ANN NEUROL 2021;90:353-365.

Figure 1:. LARGE-PD ancestry.

A: Mean ancestry proportions by LARGE-PD site. We estimated ancestry proportions using ADMIXTURE and a K of 5 in a joint dataset that included LARGE-PD and 1000 Genomes Project samples. Using 1000 Genomes super-population codes to infer the ancestry underlying each cluster, C1 represents East Asian, C2 represents South Asian, C3 represents Native American, C4 represents African, and C5 represents European ancestry B: PCA plot of LARGE-PD subjects. We conducted a principal components analysis using PC-AiR in the merged 1000 Genomes-LARGE-PD dataset. Note the preponderance of individuals with high Amerindian and European ancestries. Principal components were calculated using the PC-AiR algorithm from the GENESIS package in R.

Figure 2:. LARGE-PD GWAS results.

A. Manhattan plot of log-transformed p-values by chromosome. P-values were obtained via a logistic mixed model adjusting for age, sex, and the first five principal components using the GENESIS package in R. The significant peak is located within SNCA on chromosome 4. The suggestive peak one near chromosome 3 is near NRROS. B. QQ-plot of GWAS p-values. GC lambda was 1.017. C. Locuszoom-style plot of the chromosome 4 locus using LARGE-PD linkage disequilibrium data.

Figure 3:. Replication of GWAS significant results from Nalls et al. 2019.

A:Beta-beta plot of study beta coefficients. On the scatterplot of beta coefficients, the x-axis corresponds to betas obtained from LARGE-PD and the y-axis corresponds to beta coefficients from Nalls et al. 2019 for 76 of the 90 GWAS significant variants. In LARGE-PD, we successfully imputed 84 of the 90 variants; this figure excludes three variants with a MAC less than 10 and five strand ambiguous (CG/AT) sites that did not pass our filters (see Methods). The color scheme represents p-values obtained from LARGE-PD. Significant (p-value < 5.9×10⁻⁴) and nominally significant (p-value < 0.05) variants are labeled by their respective nearest genes. B: Difference in study beta coefficients by MAF. Six variants demonstrated a difference in beta coefficients from LARGE-PD and Nalls et al. greater than one standard deviation of the mean. All of the variants have a MAF lower than 4.52%; three have a MAC lower than 10. This was partially due to the larger effect sizes of rare variants, but also could be attributed to inaccurate beta estimates due to insufficient sample size.

Figure 4:. LARGE-PD Admixture mapping results.

A: Admixture mapping result of a joint test as implemented by the GENESIS package in R. The significance level of 5×10–5 is indicated by the dashed line. B and C: The admixture mapping results are fit using spline interpolation (solid line) and overlaid on the GWAS results in that region. The gene co-localized with the admixture mapping peak is labeled and highlighted.