Genetic basis of lacunar stroke: a pooled analysis of individual patient data and genome-wide association studies

Abstract

Background: The genetic basis of lacunar stroke is poorly understood, with a single locus on 16q24 identified to date. We sought to identify novel associations and provide mechanistic insights into the disease.

Methods: We did a pooled analysis of data from newly recruited patients with an MRI-confirmed diagnosis of lacunar stroke and existing genome-wide association studies (GWAS). Patients were recruited from hospitals in the UK as part of the UK DNA Lacunar Stroke studies 1 and 2 and from collaborators within the International Stroke Genetics Consortium. Cases and controls were stratified by ancestry and two meta-analyses were done: a European ancestry analysis, and a transethnic analysis that included all ancestry groups. We also did a multi-trait analysis of GWAS, in a joint analysis with a study of cerebral white matter hyperintensities (an aetiologically related radiological trait), to find additional genetic associations. We did a transcriptome-wide association study (TWAS) to detect genes for which expression is associated with lacunar stroke; identified significantly enriched pathways using multi-marker analysis of genomic annotation; and evaluated cardiovascular risk factors causally associated with the disease using mendelian randomisation.

Findings: Our meta-analysis comprised studies from Europe, the USA, and Australia, including 7338 cases and 254 798 controls, of which 2987 cases (matched with 29 540 controls) were confirmed using MRI. Five loci (ICA1L-WDR12-CARF-NBEAL1, ULK4, SPI1-SLC39A13-PSMC3-RAPSN, ZCCHC14, ZBTB14-EPB41L3) were found to be associated with lacunar stroke in the European or transethnic meta-analyses. A further seven loci (SLC25A44-PMF1-BGLAP, LOX-ZNF474-LOC100505841, FOXF2-FOXQ1, VTA1-GPR126, SH3PXD2A, HTRA1-ARMS2, COL4A2) were found to be associated in the multi-trait analysis with cerebral white matter hyperintensities (n=42 310). Two of the identified loci contain genes (COL4A2 and HTRA1) that are involved in monogenic lacunar stroke. The TWAS identified associations between the expression of six genes (SCL25A44, ULK4, CARF, FAM117B, ICA1L, NBEAL1) and lacunar stroke. Pathway analyses implicated disruption of the extracellular matrix, phosphatidylinositol 5 phosphate binding, and roundabout binding (false discovery rate <0·05). Mendelian randomisation analyses identified positive associations of elevated blood pressure, history of smoking, and type 2 diabetes with lacunar stroke.

Interpretation: Lacunar stroke has a substantial heritable component, with 12 loci now identified that could represent future treatment targets. These loci provide insights into lacunar stroke pathogenesis, highlighting disruption of the vascular extracellular matrix (COL4A2, LOX, SH3PXD2A, GPR126, HTRA1), pericyte differentiation (FOXF2, GPR126), TGF-β signalling (HTRA1), and myelination (ULK4, GPR126) in disease risk.

Funding: British Heart Foundation.

Figure 1

Analysis pipeline GWAS=genome-wide association study.

Figure 2

Manhattan plot of genome-wide SNP associations by genomic position (A) Lacunar stroke transethnic analysis. (B) Lacunar stroke multi-trait analysis. SNP=single nucleotide polymorphism. Dashed lines indicate the genome-wide significance threshold of 5 × 10⁻⁸.

Figure 3

Genes for which expression is associated with lacunar stroke in six tissues from transcriptome-wide association analysis Evidence of colocalisation of gene expression and lacunar stroke signals is shown by triangle size, with larger triangles indicating stronger evidence of colocalisation. CMC=CommonMind Consortium. COLOC.PP4=posterior probability of hypothesis 4 in colocalisation analysis (that there is a consistent association between lacunar stroke and expression of the given gene). GTEx=Genotype-Tissue Expression portal. WMH=white matter hyperintensities. YFS=Young Finns Study.

Figure 4

Associations between genetically proxied cardiovascular risk factors and lacunar stroke from mendelian randomisation analysis using the inverse variance weighted method Estimates are presented as ORs per genetically proxied increase in each risk factor (original scale). OR=odds ratio.