By Makoto Ishii, MD, PhD
Assistant Professor of Neurology, University of Texas Southwestern Medical Center
SYNOPSIS: In two recent cohort studies, metformin use was found to decrease the risk of developing dementia in older adults with type 2 diabetes mellitus.
SOURCES: Zimmerman SC, Ferguson EL, Choudhary V, et al. Metformin cessation and dementia incidence. JAMA Netw Open 2023;6:e2339723.
Sun M, Chen WM, Wu SY, Zhang J. Metformin in elderly type 2 diabetes mellitus: Dose-dependent dementia risk reduction. Brain 2023; Oct 25. doi: 10.1093/brain/awad366. [Online ahead of print].
In epidemiological studies, diabetes mellitus type 2 (DM2) consistently has been found to be associated with increased dementia risk. Therefore, effective DM2 therapies, such as metformin, may be beneficial in preventing dementia. However, prior observational studies investigating the effects of metformin on preventing diabetes-associated dementia have resulted in mixed results that may be the result of heterogenous study designs, small sample sizes, short follow-up periods, biases, and confounding factors, such as severity and duration of diabetes. Furthermore, withholding metformin, a first-line agent, from patients with DM2 would be impractical, therefore making a randomized placebo-controlled study investigating metformin and dementia prevention challenging to implement. Recently, two cohort studies sought to address these challenges and examine whether metformin treatment could reduce dementia risk in patients with DM2.
The first cohort study by Zimmerman and colleagues examined if termination of metformin treatment for reasons unrelated to kidney dysfunction in older DM2 patients affected dementia incidence using an emulated trial design. Electronic health records data were obtained from older adults in the Kaiser Permanente Northern California health system who had used metformin, were dementia-free at baseline, and completed a harmonized health survey.
Individuals who stopped using metformin without prior history of abnormal kidney function were defined as early terminators. Each early terminator was matched with up to four routine users or individuals who remained on metformin at the age when the matched early terminator stopped using metformin. The final analysis consisted of 12,220 early terminators (46.2% women; mean age at start of metformin, 59.4 years; 13.4% Asian, 8.2% Black, 14.9% Hispanic, and 62.7% white) and 29,126 routine users (46.6% women; mean age at start of metformin, 61.1 years; 16.0% Asian, 7.3% Black, 13.2% Hispanic, and 62.8% white). Using Cox proportional hazard models, early terminators had 1.21 times the hazard of dementia diagnosis compared with routine users in a fully adjusted model (95% confidence interval [CI], 1.15-1.28).
To compare with metformin initiation studies, the reciprocal of the fully adjusted hazard ratio (HR) was 0.83 (95% CI, 0.76-0.90). Using accelerated failure time models, accelerated time to dementia was 0.89 (95% CI, 0.82-0.97) in the fully adjusted model. Only a small fraction of the acceleration of dementia diagnosis could be attributed to measured changes in hemoglobin A1c level or insulin use.
The second cohort study by Sun and colleagues used Taiwan’s National Health Insurance Database to examine the association between metformin use and diabetes-associated dementia in adults age 60 years and older with DM2 on at least one type of anti-diabetic drug. The case group consisted of DM2 patients who received a minimum of 28 cumulative defined daily doses of metformin, while the control group consisted of DM2 patients who did not receive metformin throughout the entire follow-up period. Propensity score matching was used to balance metformin users and nonusers on a variety of covariates, resulting in 14,558 individuals included in the final analysis. Using a Cox proportional model with a fully adjusted model, metformin users had a significant reduction in dementia risk compared to nonusers (HR, 0.34; 95% CI, 0.33-0.36), with metformin use having a dose-dependent relationship, as higher daily and cumulative dosages of metformin had a greater dementia risk reduction.
COMMENTARY
Despite the different study designs (one investigating the effects of terminating metformin use while the other compared metformin users to nonusers) and study populations (Northern California and Taiwan), the two cohort studies reaching a similar conclusion provides additional strong support for the use of metformin to prevent diabetes-associated dementia. Strengths of the studies include using medical records and careful considerations of potential biases and confounding factors to investigate a relatively large number of DM2 patients that would have been impractical to conduct as a placebo-controlled study.
Limitations of the studies include difficulty in controlling every potential source of bias or confounder because of the nature of the data. Importantly, the precise reason for metformin cessation or never initiating metformin use was not known, which could serve as a potential bias or confounding factor. Additionally, the use of dementia diagnosis from medical records does not provide information such as the onset of cognitive decline and neurocognitive measures. Furthermore, there was no information regarding apolipoprotein-E (APOE) genotyping or Alzheimer’s disease biomarkers.
Collectively, these studies support not only initiating but maintaining metformin treatment for the prevention of diabetes-associated dementia, which may require finding more effective ways to mitigate or manage any adverse effects from metformin use. Although the results from the two studies support the use of metformin to prevent diabetes-associated dementia, there are several outstanding questions. Which DM2 patient would benefit the most and at what dose? Are the beneficial effects of metformin related to APOE genotype? Furthermore, the positive effects of metformin on dementia risk reduction appear to be independent of its beneficial effects on insulin signaling and blood glucose levels. If so, what are the underlying mechanisms? Could metformin be beneficial to those without DM2?
While clinical trials such as the Metformin in Alzheimer’s Dementia Prevention study will help address the latter question, additional well-designed studies clearly are needed. Nonetheless, the findings from these two cohort studies build toward the goal of one day being able to personalize one’s risk factor for developing dementia and prescribing effective dementia prevention strategies on an individual basis.