The etiology of Type I diabetes (T1D), an immune condition that leads to the destruction of the pancreatic beta-cell, remains elusive. Previous studies suggest that genetic predisposition plays a role- (GWAS)- reporting a dominant genetic risk for T1D by (MHC) class II region. Otherwise, GWAS studies have identified over 60 risk regions across the human genome, marked by single nucleotide polymorphisms, which confer a genetic predisposition to T1D.

We studied a well-defined patient cohort diagnosed with T1D (n=25), Systemic Lupus Erythematosus (SLE, n=25), and Juvenile Idiopathic Arthritis (JIA, n=25) settled in the city of Barranquilla, localized up to the north Colombian Caribbean coast. Whole exome sequences of 75 patients were contrasted against a manually curated list of previously reported genes predisposing to autoimmune diseases worldwide. Single- and multi-locus linear mixed-effects models were used to identify T1D-associated genomic variants when contrasted to other phenotypes.

Significant associations to T1D were harbored in genes related to the biosynthesis of glycoprotein oligosaccharides, phospholipid binding, pancreatic adenocarcinoma, systolic blood pressure, and fasting insulin, among others, including MGAT5 (P_FDR=1.64x10^-22), RUNX1 (P_FDR=1.8x10^-12), PSD3 (P_FDR=8.1x10^-12) and HLA-DBP2 (P_FDR=2.18x10^-9). Other significantly associated variants were harbored in the GRB2, GABRA2, CD86, KSR2, and EDEM3 genes. Interestingly, in our data, some of these variants associated with T1D were also present in individuals with SLE and AIJ.

Our study shows that sharing genomic variants by two or more autoimmune diseases provides supporting evidence of autoimmune tautology in this understudied population. Future studies are needed to evaluate their implication in the etiology of T1D and their feasibility for developing Machine Learning to enable precision medicine tools for diagnosis and follow-up.