Genetic sharing and heritability of paediatric age of onset autoimmune diseases

Abstract

Autoimmune diseases (AIDs) are polygenic diseases affecting 7-10% of the population in the Western Hemisphere with few effective therapies. Here, we quantify the heritability of paediatric AIDs (pAIDs), including JIA, SLE, CEL, T1D, UC, CD, PS, SPA and CVID, attributable to common genomic variations (SNP-h(2)). SNP-h(2) estimates are most significant for T1D (0.863±s.e. 0.07) and JIA (0.727±s.e. 0.037), more modest for UC (0.386±s.e. 0.04) and CD (0.454±0.025), largely consistent with population estimates and are generally greater than that previously reported by adult GWAS. On pairwise analysis, we observed that the diseases UC-CD (0.69±s.e. 0.07) and JIA-CVID (0.343±s.e. 0.13) are the most strongly correlated. Variations across the MHC strongly contribute to SNP-h(2) in T1D and JIA, but does not significantly contribute to the pairwise rG. Together, our results partition contributions of shared versus disease-specific genomic variations to pAID heritability, identifying pAIDs with unexpected risk sharing, while recapitulating known associations between autoimmune diseases previously reported in adult cohorts.

Figure 1. Autoimmune disease prevalence and heritability estimates.

(a) Mean population-based AI disease prevalence (orange) and heritability (blue) estimates (mean±s.d.). Data are curated from epidemiological surveys among Caucasian populations in Europe or North America based on studies indexed in PubMed between 1975 and 2015. Where multiple sources of data are available for a given trait, we reported a simple non-weighted arithmetic mean and provided as error bars the standard deviation. Most heritability estimates were based on twin concordance rates. Raw data used and references can be found in Supplementary Tables 1 and 2. (b) Univariate SNP-heritability (SNP-h², orange) compared with estimates reported by prior studies. (SNP-h² (lit), blue) based on variations across the autosomes compared with population-based estimates (POP-h², red) as reported in the literature (lit). Raw data used from prior GWAS SNP-h² estimates are provided in Supplementary Table 3. Error bars denote standard error. (c) Univariate SNP-heritability (autosomal) estimates with (Light green, wide) and without the extended MHC (orange, narrow). Results are compared with corresponding heritability estimates reported using population-based (red, narrow) versus other published SNP-heritability estimates (blue, narrow), when available for a given disease. Literature data used and references can be found in Supplementary Table 2 and Supplementary Tables 6 and 7. Error bars denote standard error. (d) Partitioning phenotypic variance to genetic and non-genetic (ENV, green) components in the four largest pAID cohorts. Genetic components include contributions from the entire autosomal regions excluding the MHC (exMHC, orange), the extended MHC (MHC, blue) alone as well as from the X-chromosome (ChrX, red).

Figure 2. Prevalence of AI disease co-morbidities and estimates of genetic correlation (co-heritability) across pAIDs.

(a) Observed prevalence of pAID comorbidity observed in Caucasian populations in Europe and North America as curated from large-scale cohort studies. For each pairwise combination (for example, Disease 1–Disease 2), the rate (y axis) indicates the percentage of patients with Disease 2 who have also been diagnosed with Disease 1. Literature data used and references can be found in Supplementary Table 9. (b) Bivariate estimates of genetic correlation (pairwise co-heritability) across pAIDs. The heritability (SNP-h²) for the first and second disease are shown for each pAID pair (blue and green bars, respectively) along with the genetic correlation (rG) for the pair estimated based on total autosomal common genetic variants (orange) and based on autosomal variants excluding the MHC (red). Displayed are those pairs for which the rG estimates reached nominal significance (P<0.05). P-values are based on restricted maximum likelihood ratio test. Error bars represent standard error. (c) Genetic sharing using the genome-wide pairwise sharing statistic (GPS). Correlation plot of the P-values obtained from the genome-wide pairwise shared analysis. Significant P-values support evidence of genetic sharing based on the correlation of significant association findings reported by GWAS for each pair of diseases.

Figure 3. Disease prediction using a support vector machine model.

Shown are the mean (orange) and maximum (blue) areas under the curve (AUC) achieved in the validation set as obtained for each disease in the ten-fold cross-validation analysis. The mean and maxima refer to the best AUC's when testing a range of P-value thresholds from which to pick SNPs in training the linear SVM. SPA, spondyloarthropathy.