The genomic landscape of pediatric autoimmune diseases

Genome-wide associations studies (GWAS) have identified hundreds of susceptibility genes, including shared associations across clinically-distinct disease groups, associated with autoimmune diseases. I performed an inverse chi-square metanalysis across ten pediatric or early-onset autoimmune diseases (pAIDs) in a case-control study including over 6,000 cases and 10,000 shared controls. Twenty-seven loci, associated with one or more pAIDs, reached genome-wide significance (P\u3c5×10-8), mapping to previously-reported autoimmune disease-associated genes (e.g. IL2RA: P\u3c2.41×10-10), as well as novel candidate loci with established immunoregulatory functions (e.g. CD40LG: P\u3c5.37×10-9). The pAID-associated SNPs are functionally-enriched for DNase-hypersensitivity sites (P\u3c7.2x10-3), eQTLs (P\u3c1.0x10-4), micro-RNA binding sites (P\u3c1.8x10-3) and coding variants ( P\u3c1.0×10-4). I also identified biologically-correlated, pAID-associated gene-sets based on immune cell expression profiling, which show strong evidence of genetic sharing among the pAIDs. Network and protein-interaction analyses demonstrated converging roles for the JAK-STAT, IFN, and TH1-TH2/TH17-Interleukin signaling pathways, which are attractive pharmacological targets for downstream functional studies. To quantify the contribution of common genomic variations to the risk of pAIDs, including Juvenile Idiopathic Arthritis (JIA), Systemic Lupus Erythematosus (SLE), Celiac Disease (CEL), Type 1 Diabetes (T1D), Ulcerative Colitis (UC), Crohn’s Disease (CD), Psoriasis, Ankylosing Spondylitis (AS), and Common Variable Immunodeficiency (CVID), I calculated the heritability attributable to genetic differences (SNP-h2) using genome-wide SNP-genotypes for each pAID and pair of pAIDs. SNP-h2 estimates were significant across multiple pAIDs, most prominently with T1D (0.863 +/- SE 0.07) and JIA (0.727 +/- SE 0.037), but were more modest in UC (0.386 +/- SE 0.04) and CD (0.454 +/- 0.025). The MHC strongly contributed to SNP-h2 in T1D and JIA, but did not contribute strongly to SNP-h2 in other pAIDs (e.g., UC or CD). These GWAS results demonstrate that the MHC is robustly associated with pAID, and are consistent with population reported h2 from adult studies and, in some cases, greater than SNP-h2 calculated based adult GWAS (i.e., T1D, UC, JIA/RA). Pair-wise genetic correlation (rG) between UC and CD and between JIA and CVID were the highest, consistent with estimated pAID genetic sharing obtained through independent statistical measures. The MHC did not play a significant role in rG. Future work utilizing these data may help validate shared drug targets across pAIDs, as well as develop clinical tools that can identify both populations at high risk for developing pAIDs, given family history of autoimmunity, and patients with a secondary pAID