Background: Severe insulin resistance diabetes (SIRD) is a subgroup of type 2 diabetes (T2D) with particularly high risk of cardiovascular complications. A subset of T2D genome-wide association study (GWAS) associations are suggested to act through pathways underlying SIRD, but the number of known loci are limited (<50), and the unavailability of large SIRD patient cohorts hampers directly conducting SIRD GWAS. We developed a novel model to computationally infer a full GWAS for SIRD, aiming to expand our understanding of its causal genes and relevant biological pathways.

Methods and Results: We developed a Bayesian model of latent SIRD association statistics conditional upon observed association statistics for 6 insulin resistance (IR)-related traits in up to 1 million individuals. We trained this model using 43 genetic variants within T2D GWAS loci suggested to be associated with SIRD, observing the expected “lipodystrophy” pattern such as positive effect sizes for insulin, TG, waist-hip-ratio, and negative for HDL and adiponectin. We then used the trained model to infer SIRD associations for over 2 million additional variants that were present in 6 IR-related GWAS. The resulting SIRD GWAS identified 429 independent variants at P < 5×10-8, which were mapped to 316 putatively causal genes, including known genes IRS1, COBLL1, PPARG, as well as novel ones such as TGFB2. Notably, TGFB2 encodes TGF-β2, an adipokine linked to IR in mice, and our data suggest its potential as a therapeutic target for SIRD in humans. Pathway-enrichment analysis showed enrichment in multiple lipid-related pathways. Furthermore, a polygenic risk score built with the GWAS was strongly associated with coronary artery disease risk (P < 1 × 10-10).

Conclusions: Using a novel Bayesian method, we inferred the first GWAS of SIRD, providing insights into molecular mechanisms of SIRD. The method is also applicable to other diabetes subtypes for which a training set of variants is available.

Disclosure

S. Yoshiji: None. P. Smadbeck: None. A. Sayici: Employee; Evotec International GmbH. J. Flannick: None.

© 2024 by the American Diabetes Association
2024
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.