In a recently published cohort study JAMA Network Open, Researchers evaluated whether metformin treatment was associated with an increased risk of all-cause dementia in people with type 2 diabetes (T2D).
The researchers examined whether other mechanisms besides improved glucose control, such as hemoglobin A1c (HbA1c) levels, mediate this association.
Background
Metformin (dimethylbiguanide) has been the preferred first-line treatment for T2D since 2006. Research suggests that starting metformin treatment may reduce the risk of dementia in type 2 diabetes patients.
However, patients often discontinue metformin treatment for reasons such as gastrointestinal adverse effects and renal dysfunction. In fact, one-fifth of primary users use alternative antidiabetic drugs instead of metformin.
According to current recommendations, the benefits should be carefully weighed against the risks of metformin therapy when the estimated glomerular filtration rate (eGFR) falls below a certain threshold.
About the study
In the current study, researchers investigated whether stopping metformin treatment for reasons unrelated to kidney dysfunction (eGFR rates falling below a safe threshold) was associated with an increased incidence of dementia.
They examined the electronic health records (EHRs) of a cohort of metformin users enrolled at Kaiser Permanente Northern California (KPNC), one of the largest integrated health care delivery systems in the United States (USA) with more than 4.6 million members.
These individuals were born before January 1, 1955 (older adults), used metformin, did not suffer from dementia, and completed one of two health surveys before enrollment in the study.
Follow-up for dementia cases began in 1996 with the implementation of their EHRs and continued through 2020. They censored participants at age 90, death, dementia diagnosis, or the start of the 90-day membership interval.
They used the KPNC mortality database to obtain dates of death, while race and ethnicity data were self-reported. Self-identified as Asian, Black, Hispanic/Latino, White, and Other or Uncertain participants. KPNC health plan databases confirmed the race and ethnicity of individuals who did not support a category.
This cohort study used a simulated trial design, which replicated critical features of an RCT. Accordingly, the team estimated associations similar to intention-to-treat (ITT) estimates when comparing early metformin terminators with a group of routine metformin users.
The cohort was matched with early terminators and routine metformin users in a 1:4 ratio. They were of the same age, sex, and had diabetes for the same duration.
all accounts p Values were two-tailed, and the statistical threshold of significance for the proportional hazards estimate was set at α = .05. Data was analyzed between November 2021 and September 2023.
Causal mediation analyzes were performed to determine whether changes in HbA1c levels or insulin prescription status mediated the association between early metformin discontinuation and all-cause dementia.
Mean changes in HbA1c levels were measured 8–16 months after metformin discontinuation and in age-matched routine users. For additional analyses, they used changes in insulin prescription status and HbA1c levels measured five years after initial metformin cessation as mediators.
The researchers conducted mediation analyzes using the accelerated failure time (AFT) model, which provided a more straightforward interpretation of the observed associations.
Further, they performed Cox proportional regression modeling to determine hazard ratios (HRs) for dementia diagnosis, using age at initiation of metformin treatment and age at the end of metformin as a scale.
The team also conducted sensitivity analyzes using creatinine instead of eGFR with sex-specific cutoffs. They also conducted data analysis restricted to participants with greater than 80% medication adherence.
In a further sensitivity analysis, the researchers limited the differences between initial and routine users, where they excluded differences in HbA1c levels of more than 0.1% and differences in diabetes duration of more than one year.
They also analyzed data on early terminators who started metformin within the past two years; Hence, they did not show any diabetes progression.
Findings and Conclusions
In total, 12,220 early terminators (46.2% female) and 29,126 (46.6% female) routine users were analyzed, with a mean age of 59.4 and 61.1 years at the start of the first metformin prescription, respectively.
Early terminators had a 1.21 times greater risk of a diagnosis of dementia than matched routine users.
Mediation analysis showed that changes in HbA1c level or insulin use contributed to the association between early metformin cessation and global dementia, with no contribution from insulin use five years after metformin cessation and 0.07 for HbA1c level one year after cessation, they suggested. Minimal association mediation. The results of the sensitivity analysis were similar to the primary analysis.
Overall, this cohort study of metformin in older adults largely supports previous observational studies showing that initiation of metformin was associated with a reduced risk of dementia.
Its large sample size and long follow-up period helped the researchers to better estimate the association between early metformin discontinuation and overall dementia. Furthermore, the study design mitigated the potential for cohort effects and confounding by indication.
This study has important implications for the clinical management of diabetes, particularly in older adults.
The risk of dementia is higher for people with diabetes, as are its carriers Apolipoprotein E (APOE) ε4 allele, mitigating adverse effects or using alternative medications rather than stopping metformin use may be more effective.
These patients may consider switching to other metformin formulations (eg, slow-release metformin) to overcome gastrointestinal adverse effects.
Future studies could evaluate the variation in the observed association of metformin with known risk factors for dementia. Studies can also extrapolate study results to prediabetic groups.