European Ancestry Is Associated with Acute Chest Syndrome in Sickle Cell Disease

Background: Acute chest syndrome (ACS) is a major cause of morbidity and mortality in patients with sickle cell disease (SCD). The elucidation of genetic modifiers and mechanisms responsible for the phenotypic variability in ACS and SCD would have important clinical implications such as the identification of individuals at high risk for ACS complications and the potential for use of individualized preventive and therapeutic strategies that could ultimately improve outcomes for patients with SCD.

We hypothesize that the development of ACS is significantly influenced by ancestry.

Methods: Autosomal SNPs were genotyped with the Affymetrix PanAFR array in a University of Illinois cohort and with the Illumina 6.0 platform in the NIH Walk-PHaSST cohort. We merged the two GWAS datasets and there were 100,079 SNPs available after removing SNPs with > 2 alleles and Hardy-Weinberg Equilibrium (HWE) P<0.001. From HapMap Phase 3 genotype data, we selected 5,000 SNPs that exhibit large frequency differences between YRI and CEU (P < 1.57 x 10^-29). Five hundred and fifteen of these SNPs were identified in the 100,079 SNPs genotyped in both GWAS arrays. After removing 152 SNPs with F_ST < 0.3 and 9 SNPs linked to other SNPs with r²>0.2, the total number of SNPs available for ancestry estimates was 354. Genetic ancestry was estimated using HapMap YRI and CEU genotypes as ancestral population (K=2) using STRUCTUR 2.1. We ran STRUCTURE under the admixture model using prior population information and assuming allele frequencies among populations, with a burn-in length of 30,000 for 70,000 repetitions. Logistic regression was used to examine the association between European ancestry (EA) and Acute Chest Syndrome. Age, gender, hemoglobin genotype (HbSS/HbSβ⁰ vs. others), and use of hydroxyurea were additionally considered during modeling.

Results: Seven hundred and thirty patients (UIC= 233, Walk-PHaSST = 497) were included in the analysis (Table 1). In binary logistic regression analysis adjusting for age, gender, Hb genotypes (severe vs. mild) and hydroxyurea use, EA was associated with and increase in ACS risk in both cohorts and when the cohorts where combined (UIC P=0.02, Walk-PHaSST P=0.001, combined P<0.001).

A similar trend was observed when degree of EA was stratified (Table 2).

Summary: Population structure and admixture influence the risk of ACS. European ancestry is associated with an increased ACS risk.

^aP<0.05 comparing ACS to non-ACS within site

^bP<0.002 comparing ACS to non-ACS within site