Abstract
The objective of this study was to evaluate the relation of metformin with change in cognition and brain pathology. During a mean of 8 years (SD = 5.5) of annual follow-up visits, 262/3029 participants were using metformin at any time during the study. Using a linear-mixed effect model adjusted for age, sex, and education, metformin users had slower decline on a score of global cognition compared to non-users (estimate = 0.017, SE = 0.007, p = 0.027). Analyses of cognitive domains showed a slower decline in episodic memory and semantic memory specifically. In sensitivity analysis, when examining any diabetes medication use vs none, no association was observed of any diabetes medication use with cognitive function. In the autopsy subset of 1584 participants, there was no difference in the level of Alzheimer's disease (AD) pathology or the presence of infarcts (of any size or location) between groups of metformin users vs non-users. However, in additional analyses, metformin users had higher odds of subcortical infarcts, and lower odds of atherosclerosis and arteriosclerosis.