Genome-Wide Association Analysis Reveals Genetic Heterogeneity of Sjögren's Syndrome According to Ancestry

Abstract

Objective: The Sjögren's International Collaborative Clinical Alliance (SICCA) is an international data registry and biorepository derived from a multisite observational study of participants in whom genotyping was performed on the Omni2.5M platform and who had undergone deep phenotyping using common protocol-directed methods. The aim of this study was to examine the genetic etiology of Sjögren's syndrome (SS) across ancestry and disease subsets.

Methods: We performed genome-wide association study analyses using SICCA subjects and external controls obtained from dbGaP data sets, one using all participants (1,405 cases, 1,622 SICCA controls, and 3,125 external controls), one using European participants (585, 966, and 580, respectively), and one using Asian participants (460, 224, and 901, respectively) with ancestry adjustments via principal components analyses. We also investigated whether subphenotype distributions differ by ethnicity, and whether this contributes to the heterogeneity of genetic associations.

Results: We observed significant associations in established regions of the major histocompatibility complex (MHC), IRF5, and STAT4 (P = 3 × 10^-42 , P = 3 × 10^-14 , and P = 9 × 10^-10 , respectively), and several novel suggestive regions (those with 2 or more associations at P < 1 × 10^-5 ). Two regions have been previously implicated in autoimmune disease: KLRG1 (P = 6 × 10^-7 [Asian cluster]) and SH2D2A (P = 2 × 10^-6 [all participants]). We observed striking differences between the associations in Europeans and Asians, with high heterogeneity especially in the MHC; representative single-nucleotide polymorphisms from established and suggestive regions had highly significant differences in the allele frequencies in the study populations. We showed that SSA/SSB autoantibody production and the labial salivary gland focus score criteria were associated with the first worldwide principal component, indicative of higher non-European ancestry (P = 4 × 10^-15 and P = 4 × 10^-5 , respectively), but that subphenotype differences did not explain most of the ancestry differences in genetic associations.

Conclusion: Genetic associations with SS differ markedly according to ancestry; however, this is not explained by differences in subphenotypes.

Figure 1

Manhattan plots of the results of the genome‐wide association studies (GWAS) with logistic regression in patients with Sjögren's syndrome. A, Global GWAS (1,405 cases, 1,622 Sjögren's International Collaborative Clinical Alliance [SICCA] controls, and 3,125 external controls) using 1,444,854 single‐nucleotide polymorphisms (SNPs), with adjustment for 9 intercontinental principal components (PCs) and smoking (λ = 1.02). B, European subgroup GWAS (585 cases, 966 SICCA controls, and 580 external controls) using the same set of SNPs described in A, with adjustment for 1 intra‐European PC and smoking (λ = 1.002). C, Meta‐analysis of the Chinese (λ = 1.03) and Japanese (λ = 1.03) subgroups of the Asian GWAS (460 cases, 224 SICCA controls, and 901 external controls) using 302,688 SNPs, with adjustment for 2 intra‐Asian PCs. Insets, Q–Q plots. MHC = major histocompatibility complex; Chr. = chromosome.

Figure 2

Major histocompatibility complex association plots. A, Asian subgroup. B, European subgroup. Inset. Relative positions of top single‐nucleotide polymorphisms and HLA genes from dbSNP. C, European subgroup (blue) with sample sizes (460 cases and 1,125 controls) equal to those in the Asian subgroup (red).

Figure 3

Associations in Europeans versus associations in Asians. Cases designated 3‐positive are positive for the SSA/SSB, focus score, and ocular staining score criteria. The single‐nucleotide polymorphisms are ordered according to decreasing heterogeneity (het) based on higher Q values. 95% CI = 95% confidence interval.