Genome-wide association study of eosinophilic granulomatosis with polyangiitis reveals genomic loci stratified by ANCA status

Abstract

Eosinophilic granulomatosis with polyangiitis (EGPA) is a rare inflammatory disease of unknown cause. 30% of patients have anti-neutrophil cytoplasmic antibodies (ANCA) specific for myeloperoxidase (MPO). Here, we describe a genome-wide association study in 676 EGPA cases and 6809 controls, that identifies 4 EGPA-associated loci through conventional case-control analysis, and 4 additional associations through a conditional false discovery rate approach. Many variants are also associated with asthma and six are associated with eosinophil count in the general population. Through Mendelian randomisation, we show that a primary tendency to eosinophilia contributes to EGPA susceptibility. Stratification by ANCA reveals that EGPA comprises two genetically and clinically distinct syndromes. MPO+ ANCA EGPA is an eosinophilic autoimmune disease sharing certain clinical features and an HLA-DQ association with MPO+ ANCA-associated vasculitis, while ANCA-negative EGPA may instead have a mucosal/barrier dysfunction origin. Four candidate genes are targets of therapies in development, supporting their exploration in EGPA.

Fig. 1

Manhattan plot of genetic associations with EGPA. Manhattan plots showing the association between genetic variants for (a) all EGPA cases (n = 534) vs. controls (n = 6688), (b) the subset of cases with MPO+ EGPA (n = 159) vs. controls, and (c) ANCA−negative EGPA cases (n = 352) vs. controls. Genetic variants at loci reaching genome-wide significance are highlighted in red. The red horizontal lines indicate the threshold for declaring genome-wide significance (p = 5 × 10⁻⁸). P-values for genetic association are from a linear mixed model (BOLT-LMM)

Fig. 2

Genomic features at four EGPA-associated loci. Genomic positions from the hg19 genome build, representative RefSeq genes, long-range DNA interactions, genetic variant associations with EGPA, causal variant mapping (expressed as posterior probabilities, PP) and H3K4 mono-methylation data are shown for (a) the BCL2L11 region, (b) the LPP region, (c) the C5orf56-IRF1-IL5 region and (d) the 10p14 intergenic region. Arrows indicate direction of transcription. P-values for genetic association are from a linear mixed model (BOLT-LMM). For further details regarding the promoter enhancer interaction mapping, including cell types analysed at each locus, see the ‘Methods’ section and Supplementary Fig. 7

Fig. 3

Clinically and genetically distinct subsets within EGPA, and their relation to MPO+ AAV. Above: schematic showing relationship between MPO+ AAV, MPO+ EGPA and ANCA-negative EGPA, and putative genes underlying this classification. Below: Unshaded cells in the table show a comparison of the clinical features of MPO+ and ANCA-negative EGPA from this study as % (*p < 0.0002 compared to other EGPA subset: see Table 1), but also see refs. ^, . Shaded cells show data from external sources: MPO+ AAV clinical data was derived from the EVGC AAV GWAS, and rituximab response rates for MPO+ AAV from the RAVE study and for EGPA from ref. . n.d. not determined

Fig. 4

Relationship between genetic control of eosinophil count and risk of EGPA. a Correlation between the effect on EGPA risk and eosinophil count for the lead EGPA-associated genetic variants outside the HLA region. Blue points indicate variants discovered through standard case-control analysis, red points indicate variants discovered through cFDR. Horizontal and vertical lines indicate 95% confidence intervals. b Mendelian randomisation analysis supports a causal role for eosinophil abundance in EGPA aetiology. Points represent genome-wide significant conditionally independent variants associated with blood eosinophil count in the GWAS by Astle et al. (where typed or reliably imputed in the EGPA dataset). Coloured lines represent estimated causal effect of eosinophil count on risk of EGPA from Mendelian randomisation (MR) methods. IVW inverse-variance weighted