Bioinformatic and rare‐variant collapsing analyses for type 1 and type 2 diabetes in the UK Biobank reveal novel pleiotropic susceptibility loci

Type 1 diabetes (T1D) is a chronic condition caused by the autoimmune destruction of pancreatic β-cells.1 In contrast, type 2 diabetes (T2D) is characterized by impaired glucose metabolism arising from defects in insulin resistance and secretion.2 More than 75 genetic loci influencing T1D risk have been identified.1 Genome-wide association studies (GWAS) of T2D have identified over 700 risk loci.2 Whole exome sequencing (WES) studies may reveal the association of rare variants to common diseases such as T1D and T2D. However, only a few large-scale WES studies have been published until Wang et al reported the relationships between rare protein-coding variants and 17 361 binary phenotypes using WES data from 269 171 UK Biobank participants (https://azphewas.com/).3 Recently, Karczewski et al determined gene-based association investigating 4529 phenotypes in 394 841 UK Biobank exomes (https://app.genebass.org/).4 We used the two published UK Biobank portals (https://azphewas.com/ and https://app.genebass.org/)3, 4 to access gene collapsing analyses of rare variation for T1D and T2D (Table 1). Ethical statements are not required for the study as no human or animal studies are involved. In order not to discard potential candidate genes we present genes with p values <.05/20000 genes = 2.5 × 10−6 commonly used for WES studies. Identified T1D and T2D genes were bioinformatically analyzed using the GWAS catalog (https://www.ebi.ac.uk/gwas/), OMIM (https://www.omim.org/), and Genecards (https://www.genecards.org/).5-8 The literature was searched for identified genes using https://pubmed.ncbi.nlm.nih.gov/. We compared the union of the same three-digit ICD-10 codes (International Classification of Diseases, Tenth Revision).3, 4 In Table 1 only the genes with genome wide significant results are shown with p values for the most significant model. One previously T1D linked gene (HLA-DRB5) and four novel T1D genes (PSMB9, NELFE, SLC44A4, and VWA7) were identified. For T2D four previously linked genes (GCK, HNF1A, HNF4A, and ANKH) were confirmed. In addition, GIGYF1 has recently already been linked to T2D in UK Biobank.9 Two novel associations were identified, the DENND6A and RPS3A genes. The identified genes were specific for each of T1D and T2D (Table 1). Phenome-wide association studies (PheWAS) data (Table 1) could link all five identified T1D genes to other immune-mediated diseases: ankylosing spondylitis, iridocyclitis, hypothyroidism, asthma, celiac disease, sarcoidosis, psoriasis, and rheumatoid arthritis (Table 1). Thus, the five identified pleiotropic T1D genes may all contribute to the previously observed epidemiological associations between T1D and other immune-mediated diseases.10 Only the GIGYF1 gene among the T2D linked genes was associated with a potential immune-mediated disorder (hypothyroidism) (Table 1). However, even more interesting is the association between the T2D linked GIGYF1 gene and chronic obstructive pulmonary disease (COPD) in the PheWAS analysis (Table 1). COPD and T2D are recognized to be associated conditions with shared environmental exposures.11 Treatment with the novel antihyperglycemic drugs glucagon-like peptide 1 (receptor agonists and sodium glucose transporter 2 inhibitors have recently been associated with a reduced risk of severe exacerbations in COPD among patients with T2D.12 Thus, the GIGYF1 gene might contribute to the observed epidemiological association between COPD and T2D and may open novel treatments for T2D and COPD. A limitation is that the validity of T1D is not perfect in UK Biobank. However, a diagnosis of T1D in UK Biobank may still be useful for research in large studies. Two papers about T1D in UK Biobank have been published suggesting that the UK Biobank might be useful for T1D research: one in Lancet Diabetes & Endocrinology and one in Diabetes Medicine.13, 14 Moreover, an article by Thomas et al has shown that the accuracy of T1D and T2D tested with two different methods range from 71% to 88%.15 These articles are in line with the findings in the present study. For instance, we confirmed one previously recognized T1D gene (HLA-DRB5) and four previously linked T2D genes (GCK, HNF1A, HNF4A, and ANKH). Thus, the definition used in UK Biobank for T1D and T2D could differentiate between known T1D and T2D genes, which is reassuring. Moreover, all the T1D linked genes (one old and four novel T1D genes) could in the PheWAS and the bioinformatic analysis be linked to immune-mediated disorders (Tables 1 and 2). It is well known that genetic links between many different immune-mediated disorders exist.10 Moreover, no T1D gene was linked to T2D. Thus, there was no overlap between T1D and T2D genes. Moreover, only one T2D gene could be linked in the PheWAS and bioinformatic analysis to a potential immune-mediated condition (ie, hypothyroidism). Thus, we believe the accuracy is acceptable for large studies of T1D and T2D genetics in UK Biobank. In conclusion, rare variations in 12 genes (six novel) were associated with diabetes in the UK Biobank, T1D (five genes) and T2D (seven genes). Thus, rare variation contributes to T1D and T2D in the general population. Rare variation in all five T1D linked genes is also linked to other immune-mediated diseases in UK Biobank, whereas the T2D gene GIGYF1 is associated with COPD. We thank the free access to the Genebass and the AstraZeneca PheWAS portals that made this work possible (https://azphewas.com/ and https://app.genebass.org).3, 4 This work was supported by a grant awarded to Dr Bengt Zöller by ALF-funding from Region Skåne, Sparbanken Skåne, and by the Swedish Research Council. The funders had no role in the study.