Across 3,295,564 Taiwanese adults with type 2 diabetes followed for 8.5 years, metformin (aHR 0.78), thiazolidinediones (0.71), sulfonylureas/meglitinides (0.82) and α-glucosidase inhibitors (0.83) were each associated with fewer incident aortic aneurysms. GLP-1 receptor agonists (0.69), SGLT2 inhibitors (0.90) and DPP-4 inhibitors (0.95) pointed the same direction, while insulin carried higher risk (1.14). The newer-drug signals vanished when analysis was limited to patients starting from 2011 onward.
The finding helps explain the diabetes paradox, where people with diabetes develop aneurysms less often despite worse vascular health. Plausible mechanisms include metformin's suppression of matrix metalloproteinases and vascular inflammation, and glycation-driven aortic wall stiffening in diabetes itself. Earlier cohort data and small animal studies have hinted at a metformin benefit, so this is confirmatory rather than paradigm-shifting. The caveats are substantial. Observational drug-class comparisons suffer from confounding by indication: insulin is a marker of long-standing, severe disease, so its higher hazard probably reflects that rather than harm. Residual confounding from smoking, family history and imaging frequency is likely, since an aneurysm is only found if someone looks for it. The GLP-1RA estimate rests on few events, with a wide interval (0.50 to 0.95), and its fragility across time periods suggests immortal-time or channeling bias. Practically, no one should choose a diabetes drug to prevent aneurysm. Smoking cessation and blood-pressure control remain the proven levers, and an ultrasound screen for men over 65 who have ever smoked is worth raising with a doctor. Randomised confirmation is still needed.