Genetic and polygenic risk score analysis for Alzheimer's disease in the Chinese population

Abstract

Introduction: Dozens of Alzheimer's disease (AD)-associated loci have been identified in European-descent populations, but their effects have not been thoroughly investigated in the Hong Kong Chinese population.

Methods: TaqMan array genotyping was performed for known AD-associated variants in a Hong Kong Chinese cohort. Regression analysis was conducted to study the associations of variants with AD-associated traits and biomarkers. Lasso regression was applied to establish a polygenic risk score (PRS) model for AD risk prediction.

Results: SORL1 is associated with AD in the Hong Kong Chinese population. Meta-analysis corroborates the AD-protective effect of the SORL1 rs11218343 C allele. The PRS is developed and associated with AD risk, cognitive status, and AD-related endophenotypes. TREM2 H157Y might influence the amyloid beta 42/40 ratio and levels of immune-associated proteins in plasma.

Discussion: SORL1 is associated with AD in the Hong Kong Chinese population. The PRS model can predict AD risk and cognitive status in this population.

Keywords: Alzheimer's disease; SORL1; disease risk; polygenic risk score; population genetics.

FIGURE 1

Study design schematic diagram. AD, Alzheimer's disease; AUC, area under the receiver operating characteristic curve; MRI, magnetic resonance imaging; PEA, proximity extension assay; PRS, polygenic risk score; PWH, Prince of Wales Hospital; SNP, single nucleotide polymorphism; WGS, whole‐genome sequencing.

FIGURE 2

Meta‐analysis of the association of SORL1 rs11218343 with Alzheimer's disease. Forest plot showing the meta‐analysis results of SORL1 rs11218343 (C allele) including data from the literature and array genotyping data. Effect sizes obtained from independent datasets and the meta‐analysis are denoted by black rectangles and the black diamond, respectively. Each row represents an independent dataset; lines indicate 95% confidence intervals, and rectangle size is proportional to the weight in the meta‐analysis (RE2 meta‐P = 1.20E−19, random effect β = −0.217, SE = 0.024). RE2, Han and Eskin's random effects model; SE, standard error. Data sources: rows 1‐15, ¹ , ² 16‐29, ³ 30‐32, ² 33, ⁴ 34, ⁵ 35, ⁶ 36. ⁷

FIGURE 3

Polygenic risk score model for the prediction of cognitive performance and Alzheimer's disease risk. Performance of the PRS model to distinguish individuals with AD or worsening cognitive performance from healthy controls (NCs) in the array cohort and WGS replication cohort. A, Summary plot of lasso regression analysis of the array dataset. The x‐axis denotes the iterations in which different λ‐values were applied. The left (red) and right (blue) y‐axes denote the prediction accuracy of AD (measured according to the MSE) and the number of surviving covariates in each iteration, respectively. The dashed line marks the selected parameters used to construct the PRS model. B–D, Results from the array dataset: (B) ROC plot of AD prediction accuracy; red and blue denote the results from the PRS model and APOE genotypes, respectively; (C, D) association between PRS and normalized MoCA score. C, Total participants; low versus high: T =  −3.34, ***P < 0.001. D, APOE‐ε3 homozygous participants; low versus high: T =  −1.99, *P < 0.05. E–G, WGS replication dataset results: (E) ROC plot of AD prediction accuracy; red and blue denote the results from the PRS model and APOE genotypes, respectively; (F, G) Association between PRS and normalized MoCA score. F, Total participants; low versus high: T  =  −2.37, *P < 0.05. G, APOE‐ε3 homozygous participants; low versus high: T  =  −2.65, **P < 0.01. C, D, F, G, Boxes indicate the 25th to 75th percentiles, and whiskers indicate the 10th and 90th percentiles; numbers indicate the numbers of individuals in the corresponding group. AD, Alzheimer's disease; AUC, area under the receiver operating characteristic curve; ROC, receiver operating characteristic; MoCA, Montreal Cognitive Assessment; MSE, mean squared error; PRS, polygenic risk score; WGS, whole‐genome sequencing.

FIGURE 4

Associations between polygenic risk score and disease‐related endophenotypes. A, Association between PRS and plasma Aβ₄₂/Aβ₄₀ ratio; T  =  −3.14, P = 1.89E−03 (n = 266). Green, yellow, and red (along with vertical lines for clarity) specify the ranges of PRSs (i.e., tertiles) in the low, medium, and high PRS groups, respectively. B, C, Association between PRS and brain volume. B, Association between PRS and hippocampal volume; low versus high: T  =  −2.57, *P < 0.05. C, Association between PRS and amygdala volume; low versus high: T  =  −2.70, **P < 0.01. D, E, Associations between PRS and levels of plasma proteins. D, Association between PRS and plasma OPN protein level; AD versus NC: T  =  1.15, P = 0.25; low versus medium: T  =  1.68, P = 0.10; low versus high: T = 3.31, **P < 0.01. E, Association between PRS and plasma NCAN protein level; AD versus NC: T  =  0.29, P = 0.77; low vs high: T = 2.36, *P < 0.05; medium vs high: T  =  −3.49, ***P < 0.001. F–H, Protein–protein interaction network and enrichment analysis of plasma proteins associated with PRS (P < 0.05). Protein–protein interaction enrichment; P = 7.78E−11. B–E, Boxes indicate the 25th to 75th percentiles, and whiskers indicate the 10th and 90th percentiles; numbers indicate the numbers of individuals in the corresponding group. F, Protein–protein interaction network of PRS‐associated plasma proteins. Node color specifies the protein class, and edge color and thickness are proportional to the pairwise interaction score. G, H, Enrichment analysis of the plasma protein interaction network. Results (FDR < 0.05) for Gene Ontology (G) and pathway analysis (H) are shown. Aβ, amyloid‐beta; AD, Alzheimer's disease; FDR, false discovery rate; MCI, mild cognitive impairment; MoCA, Montreal Cognitive Assessment; NC, normal control; NCAN, neurocan; OPN, osteopontin; PRS, polygenic risk score; WGS, whole‐genome sequencing.

FIGURE 5

Identification of TREM2 H157Y and its possible associations with cognitive performance and disease‐associated endophenotypes. A, Sanger validation of the TREM2 H157Y variant (rs2234255). Upper and lower panels show the sequencing results obtained from TREM2 H157Y noncarriers (GG) and carriers (GA), respectively. Arrows indicate the TREM2 H157Y site. B, C, Associations between TREM2 H157Y and normalized MoCA score in (B) all participants (GA versus GG: T = −2.16, *P< 0.05) and (C) AD patients (GA versus GG: T = −1.41, P = 0.15). D, Plasma Aβ₄₂/Aβ₄₀ ratio in TREM2 H157Y noncarriers and carriers; GA versus GG: T = 2.12, *P < 0.05. E–G, Plasma biomarker levels in AD patients with or without TREM2 H157Y variants; GA versus GG: T = −4.53, 2.80, and 3.99 for (E) NOTCH 3, (F) IL‐2, and (G) CASP3, respectively (**P  <  0.01, ***P < 0.001). B–G, Boxes indicate the 25th to 75th percentiles, and whiskers indicate the 10th and 90th percentiles; numbers indicate the numbers of individuals in the corresponding group. Aβ, amyloid beta; AD, Alzheimer's disease; MoCA, Montreal Cognitive Assessment.