Association of HbA1c trajectory and variability with cognitive function

Abstract Background Type 2 diabetes (T2D) is an important modifiable risk factor for cognitive decline, and increasing evidence suggests that worse glucose control, in addition to a diagnosis of T2D, may play a role in cognitive function in later life. Despite this increasing concern, evidence regarding the relative contributions of poor glucose control and higher glucose variability to cognitive dysfunction among individuals with overweight or obesity is limited, mostly due to lack of repeated measures in many studies. Method Look AHEAD MIND Study participants, aged 45‐76 years at baseline (2001‐04) with T2D and overweight or obesity, were randomized to an intensive lifestyle intervention focused on weight loss or a diabetes support and education condition, which were discontinued after ∼10 years. HbA1c levels were evaluated at years 1‐4, 6, 8, and 10. Cognitive function was evaluated up to four times between years 8‐18 with a cognitive battery that assessed memory, attention, and executive function domains. Cognitive scores were binned into four groups based on relative timing of cognitive assessments (8‐9; 10‐11; 12‐14; and 15‐18 years). We classified HbA1c variability according to: Variability Independent of the Mean (VIM), Average Real Variability (ARV), and data‐driven latent trajectories. Mixed linear regression models for standardized composite cognitive scores were adjusted for: repeated measures, order of cognitive test, years from randomization, randomization arm, age, sex, race/ethnicity, years of education, CVD history, and APOE ε4. Result Four HbA1c trajectories were identified: increasing, decreasing, stable‐high, and stable‐low. Composite cognitive scores were significantly lower for increasing, decreasing, and stable‐high HbA1c trajectories ‐compared to stable‐low HbA1c ( Table ). In multivariable analysis, higher baseline HbA1c, ARV, and VIM were associated with lower composite cognitive scores, and differences between groups were similar by years since randomization. Estimates for ARV and VIM were attenuated when adjusted for baseline HbA1c (not shown). Estimates were similar for tests from all cognitive domains, with little heterogeneity by randomization arm. Conclusion Among individuals with T2D, higher HbA1c at baseline and greater HbA1c variability over 10 years were significantly associated with lower cognitive scores, suggesting that maintaining glucose control may be useful for the prevention of cognitive decline.