Identification of genetic risk loci associated with aquaporin 4-positive neuromyelitis optica spectrum disorder: a genome-wide association study

Background Little is known about the causes of serum aquaporin 4 (AQP4) antibody-positive neuromyelitis optica spectrum disorder (AQP4-positive NMOSD) or how its pathophysiology differs from other demyelinating autoimmune diseases, which limits therapeutic and preventative opportunities despite available diagnostic biomarkers. By performing a pan-ancestry genome-wide association study (GWAS), we aimed to deepen our understanding of the genetic architecture of AQP4-positive NMOSD and to explore shared heritability with other autoimmune diseases. Methods We performed a pan-ancestry GWAS in 2833 individuals, including 1573 AQP4-positive NMOSD cases and 1260 controls (comprising non-affected relatives, healthy non-relatives, and those with other autoimmune diseases). Samples were collected from the International NMOSD Genetics Consortium, comprising 36 hospitals and research facilities across the world. All individuals with AQP4-positive NMOSD fulfilled the 2015 international consensus diagnostic criteria for NMOSD, including positive AQP4-IgG status. We also performed a second GWAS with only the 1857 samples of European ancestry (803 cases and 1054 controls). We then compared our GWAS findings to those from other autoimmune diseases. Findings We found three independent associations with AQP4-positive NMOSD that reached genome-wide significance: two within the MHC and a third within an intron of the STAT4 gene. In the pan-ancestry study, we identified a complement C4A-associatied variant, rs1150753 (chr6:32092090:A>G, p=1·61 × 10 –29 , odds ratio [OR] 2·95, 95% CI 2·44–3·56), rs607929 (chr6:32619221:C>G, p=2·87 × 10 –24 , OR 1·93, 95% CI 1·70–2·20), and a STAT4-associated variant, rs35593987 (chr2:191051800:AC>A, p=8·49 × 10 –14 , OR 1·75, 95% CI 1·51–2·03). In Europeans, we identified variants that were either in linkage disequilibrium with or the same as those from the pan-ancestry study: rs1270942 (chr6:31951083:A>G, p=2·52 × 10 –28 , OR 3·01, 95% CI 2·47–3·66), rs607929 (p=1·12 × 10 –20 , OR 1·99, 95% CI 1·72–2·30), and rs3821236 (chr2:191038032:G>A, p=1·20 × 10 –10 , OR 1·74, 95% CI 1·47–2·06). rs1270942 is in linkage disequilibrium with rs1150753 (r 2 =0·96) as well as two AQP4-positive NMOSD-associated HLA alleles, HLA-DRB1*03:01 (p=2·80 × 10 –26 , OR 2·79, 95% CI 2·30–3·37) and HLA-B*08:01 (p=1·02 × 10 –24 , OR 2·68, 95% CI 2·22–3·24). A priori testing of the P1104A variant (rs34536443, chr19:10352442:G>C) within the TYK2 gene, which acts upstream in the STAT4 pathway, found it to be protective (p=0·0008, OR 0·52, 95% CI 0·35–0·76 in the pan-ancestry study). Genetic sharing was observed with several comorbid autoimmune diseases for both the complement C4A-associated and STAT4-associated variants, including Sjögren's syndrome and systemic lupus erythematosus. Interpretation AQP4-positive NMOSD is more genetically similar to systemic autoimmune diseases than to multiple sclerosis, despite sharing overlapping clinical phenotypes. Specifically, a polymorphism associated with reduced complement C4 was identified as the biggest disease genetic risk factor, which has been shown to facilitate the development of autoantibody-producing B cells. Our findings also support a pathogenic role of HLA-restricted CD4 + T cells, owing to both a genome-wide significant association of HLA-DRB1*03:01 as well as heritable risk within the TYK2-STAT4 signalling pathway. Having already been shown to be a successful target for treating psoriatic arthritis and, potentially, systemic lupus erythematosus, we propose the TYK2-STAT4 pathway as a possible therapeutic target in AQP4-positive NMOSD. Funding The OAK Foundation, The Guthy Jackson Charitable Trust Foundation, and the UK Medical Research Council.