Fig. 4: Single-ancestry Mendelian randomization results.

Mendelian randomization (MR) results for different global genetic ancestry groups using cluster-stratified genetic predisposition for T2D and risk for T2D non-cardiovascular comorbidities (EUR=individuals genetically similar to Europeans, EAS=individuals genetically similar to East Asians) for the causal relationships (q-value < 0.05). Causal estimates are expressed as the odds ratios (ORs) of each comorbidity per doubling (2-fold increase) in genetically determined dichotomous T2D risk. Points represent MR causal estimates derived from summary statistics (ORs, measure of center), and error bars denote 95% confidence intervals (CI). Filled circles mark robust estimates that passed all sensitivity analyses. Sample size of the T2D GWAS meta-analysis used as exposure datasets: AMR = 29,375 cases, n = 59,368 controls; EAS = 88,109 cases, n = 339,395 controls; EUR = 242,283 cases, n = 1,569,734 controls; SAS = 16,832 cases, n = 33,767 controls. Sample sizes of the GWAS used as outcome datasets: COPD (AMR: n = 1503 cases, n = 13,583 controls; EUR: n = 58,559 cases, n = 937,358 controls; meta-analysis: n = 81,084 cases, n = 1,289,334 controls), osteoporosis (EAS: n = 9794 cases, n = 168,932 controls; EUR: n = 8520 cases, n = 479,981 controls; meta-analysis: n = 18,314 cases, n = 648,913 controls), rheumatoid arthritis (EAS: n = 11,025 cases, n = 162,608 controls; meta-analysis: n = 33,375 cases, n = 237,431 controls), asthma (EAS: n = 18,549 cases, n = 322,655 controls; EUR: n = 121,940 cases, n = 1,254,131 controls; SAS: n = 4015 cases, n = 23,076 controls; meta-analysis: n = 153,624 cases, n = 1,641,573 controls), and depression (EAS: n = 21,980 cases, n = 360,956 controls; EUR: n = 170,756 cases, n = 233,773 controls). Abbreviations: T2DGGI Type 2 Diabetes Global Genomics Initiative, CI confidence interval, PI proinsulin, COPD chronic obstructive pulmonary disease.