Abstract
Background Genome wide association studies (GWAS) have identified >100 SNPs associated with systemic lupus erythematosus (SLE) risk. There may be additional loci impacting the age of diagnosis. The purpose of this study was to identify genetic variants associated with age of SLE diagnosis.
Methods Our cohort included patients from tertiary care centres who met ACR and/or SLICC SLE classification criteria. We censored patients missing age at diagnosis. Patients were genotyped on the Illumina Multiethnic Array (MEGA) or Global Screen Array (GSA). Ungenotyped SNPs were imputed using the TopMed reference. We restricted to SNPs with a minor allele frequency (MAF) ≥ 0.01 and imputation quality R2≥ 0.3. Ancestry was genetically inferred using principal components (PCs) and ADMIXTURE in reference to 1000 Genome Project. We completed genome-wide linear regression of log-transformed age at SLE diagnosis with GENESIS (significance P<5x10-8). Multivariate models adjusted for sex and 5 PCs. We also conducted a GWAS of childhood-onset SLE (cSLE) patients, defined as diagnosis <18 years of age, vs. adult-onset SLE (aSLE), using logistic regression, and adjusted for the same covariates. We conducted sensitivity analyses where we stratified GWAS by cSLE and aSLE, then meta-analyzed results using inverse variance weighting, as well as ancestry-stratified analyses (Europeans, East Asians, Africans, Amerindians and Admixed).
Results Our cohort included 1489 patients, 761 (51%) cSLE, 1306 (88%) female. Median age at diagnosis was 17.7 years (IQR 14, 31) in the total cohort, 14.1 years (IQR 11.8, 15.8) in cSLE, and 31.2 years (IQR 24.7, 42) in aSLE. In the total cohort, 576 (39%) were of European ancestry, 278 (19%) East Asian, and 253 (17%) Admixed. We included 11.7M SNPs in GWAS. In the age of SLE diagnosis GWAS, 2 loci on chr16 were genome-wide significantly associated with younger age at diagnosis (top SNP rs11641349, Beta -0.03y, SE 0.15y, P=4.33x10-8, MAF 0.2). Both SNPs were intronic to CCDC113, a component of centriolar satellites. These loci were also the most significant in the GWAS of cSLE (top SNP rs16959933, OR 1.75 [95% CI: 1.43, 2.14, P=5.45 x10-8]). Sensitivity analyses showed similar results, yet they did not reach genome-wide significance with top SNPs rs11641349 (P=3.84x10-7) and rs16959933 (P=4.50x10-7) in the age group model, and rs11641349 (P=3.52x10-7) and rs16959933 (P=4.22x10-7) in the ancestry model.
Conclusion In our multiethnic cSLE and aSLE cohort, we identified genome-wide significant loci associated with age at diagnosis and cSLE risk, intronic to CCDC113. Our study requires independent validation.
