Birthweight, BMI in adulthood and latent autoimmune diabetes in adults: a Mendelian randomisation study

Aims/hypothesis Observational studies have found an increased risk of latent autoimmune diabetes in adults (LADA) associated with low birthweight and adult overweight/obese status. We aimed to investigate whether these associations are causal, using a two-sample Mendelian randomisation (MR) design. In addition, we compared results for LADA and type 2 diabetes. Methods We identified 43 SNPs acting through the fetal genome as instrumental variables (IVs) for own birthweight from a gcnomc-wide association study (GWAS) of the Early Growth Genetics Consortium (EGG) and the UK Biobank. We identified 820 SNPs as IVs for adult BMI from a GWAS of the UK Biobank and the Genetic Investigation of ANthropometric Traits consortium (GIANT). Summary statistics for the associations between IVs and LADA were extracted from the only GWAS involving 2634 cases and 5947 population controls. We used the inverse-variance weighted (IVW) estimator as our primary analysis, supplemented by a series of sensitivity analyses. Results Genetically determined own birthweight was inversely associated with LADA (OR per SD [similar to 500 g] decrease in birthweight 1.68 [95% CI 1.01, 2.82]). In contrast, genetically predicted BMI in adulthood was positively associated with LADA (OR per SD [similar to 4.8 kg/m(2)] increase in BMI 1.40 [95% CI 1.14, 1.71]). Robust results were obtained in a range of sensitivity analyses using other MR estimators or excluding some IVs. With respect to type 2 diabetes, the association with birthweight was not stronger than in LADA while the association with adult BMI was stronger than in LADA. Conclusions/ interpretation This study provides genetic support for a causal link between low birthweight, adult overweight/obese status and LADA.

Automated summary

This study investigates the causal relationship between low birthweight, adult overweight/obese status, and latent autoimmune diabetes in adults (LADA) using a two-sample Mendelian randomisation design. The results suggest a negative association between low birthweight and LADA, as well as a positive association between adult BMI and LADA.