Ischemic stroke is a major complication of sickle cell disease (SCD) and without preventive care affects 11% of children with SCD before the age of 20. Ischemic stroke is more common in children, whereas hemorrhagic stroke is more common in SCD patients who are 20-30 year-olds. Previous genetic studies of stroke in SCD have been either un-replicated or conflicting. To address this gap, we have conducted a genome-wide association study (GWAS) of ischemic stroke within SCD with three novel features: 1) the largest cohort to date at N=1,333 HbSS homozygotes; 2) early-life risk association with a cox proportional hazards model; 3) whole genome sequencing data.
Medical records were abstracted including a history of stroke using standardized definitions across 6 Brazilian sites in Brazil. Cox Proportional Hazards model was run with stroke as the outcome on 1,333 homozygous HbSS individuals (178 Ischemic strokes) using whole genome sequence (MAF > 1%; minimum depth 10x) within the NHLBI Trans-Omics for Precision Medicine (TOPMed), and the top 10 principal components were used as covariates to control for population stratification. Alleles were coded as dosages (0, 1, or 2) across 14M sites.
14 genomic regions were associated with early ischemic stroke at genome wide significance (P < 5x10-8). This included variants near two genes which have been previously linked to pediatric or early onset stroke in non-SCD cohorts: ADAMTS2 (rs147625068, P = 3.70 x 10-9) and CDK18 (rs12144136, P= 2.38 x 10-9). Individuals in our study harboring multiple risk alleles exhibited increasing rates of stroke at earlier timepoints (P < 0.001, Gehan-Wilcoxon). Significant gene enrichment was observed in multiple tissue classes, including upregulated genes in the coronary (8 genes, P = 0.0005, FDR), tibial (9, P = 0.03, FDR) and aorta arteries (7, P = 0.03, FDR), as well as downregulated genes in the spleen (7, P = 0.005), pancreas (13 genes, P = 0.02), esophagus mucosa (8, P = 0.03), hypothalamus (10, P = 0.03), and substantia nigra (11, P = 0.03). We also observed enrichment within the GWAS Catalog disease category white matter lesion progression (CNNM2, NT5C2, P = 0.01) and the combined phenotype of coronary artery disease or large artery stroke (CNNM2, NT5C2, P = 0.04).
In conclusion, we have conducted the largest GWAS of ischemic stroke in SCD to date. Results have replicated known associations with pediatric and early stroke (<65 years) for genes ADAMTS2 and CDK18, respectively. Pathway analysis suggests gene dysregulation in the arteries (coronary, tibial, and aorta), spleen, pancreas, esophagus mucosa, hypothalamus, and substantia nigra.