Association Between Metformin and Alzheimer’s Disease: A Systematic Review and Meta-Analysis of Clinical Observational Studies

Abstract

Background:

As one of the widely used drugs for the management of type 2 diabetes mellites (T2DM), metformin is increasingly believed to delay cognitive deterioration and therapeutically for Alzheimer’s disease (AD) patients especially those with T2DM. However, studies of the potential neuroprotective effects of metformin in AD patients have reported contradictory results.

Objective:

This study aimed to evaluate the association between metformin and the risk of developing AD.

Methods:

We systematically searched the PubMed, EMBASE, Web of Science, Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov databases to identify clinical observational studies on the relationship between AD risk and metformin use published before December 20, 2021. Two investigators independently screened records, extracted data, and assessed the quality of the studies. Pooled odds ratios (ORs) and corresponding 95% confidence intervals (CIs) were calculated using random-effect models.

Results:

After screening a total of 1,670 records, we included 10 studies involving 229,110 participants. The meta-analysis showed no significant association between AD incidence and metformin exposure (OR 1.17, 95% CI 0.88–1.56, p = 0.291). However, subgroup analysis showed that among Asians, the risk of AD was significantly higher among metformin users than those who did not (OR 1.71, 95% CI 1.24–2.37, p = 0.001).

Conclusion:

The available evidence does not support the idea that metformin reduces risk of AD, and it may, in fact, increase the risk in Asians. Further well-designed randomized controlled trials are required to understand the role played by metformin and other antidiabetic drugs in the prevention of AD and other neurodegenerative diseases.

Keywords

Alzheimer’s disease cognitive dysfunction meta-analysis metformin

INTRODUCTION

Alzheimer’s disease (AD) is one of the most common neurodegenerative diseases worldwide: in 2016 it was reported to affect >44 million people and caused annual economic losses amounting to > $600 billion [1]. As the global population ages, it is estimated that, by 2050, the number of AD patients worldwide will exceed 100 million, resulting in a heavy social and economic burden [2 –4]. Currently, there are no effective treatments for AD, and the few drugs available, such as donepezil, memantine, and galantamine, are just symptomatic treatment [5]. Despite considerable investment from the pharmaceutical industry, most clinical trials of drugs to target AD have been unsuccessful [6, 7]. Therefore, it is critical to identify novel therapeutic approaches to prevent the development of AD, especially in the aging population.

Diabetes and AD are both age-related diseases that may share similarities in their pathogenesis. Previous studies have shown that diabetes is one of the risk factors for AD [8, 9]. Metformin is a safe, widely used drug to treat type 2 diabetes mellitus (T2DM): it has a wide range of pharmacological effects, including increasing the uptake and utilization of glucose in skeletal muscles and other tissues, improving insulin sensitivity in the peripheral tissues, and delaying the absorption of glucose from the gastrointestinal tract [10]. Emerging evidence, including animal and clinical trials, emphasized the neuroprotective potential of metformin for AD or AD animal models [9 , 11–13]. It was proved to improve spatial memory, learning, sociability, and coordination while reduces cognitive deficits in experimental models of AD [11, 14]. Clinically, some studies have suggested that metformin can reduce the risk of developing AD [15, 16], indicating that metformin might be a promising new therapeutic strategy against AD. On the other hand, several studies have reported that metformin may increase risk of AD [17 –19]. A study conducted in Finland involving 29,412 individuals with diabetes reported that although metformin use was not associated with an increased risk of AD, long-term (≥10 years) or high-dose (average 2 g per day) metformin use was associated with lower risk of AD [20]. Another study conducted in Korea reported that metformin use increased the risk of AD among individuals who were newly diagnosed with T2DM, and this association increases with cumulative defined daily doses per day [17].

Some preclinical studies have also drawn conflicting conclusions. It was proved that metformin can mitigate AD by reducing amyloid-β (Aβ) plaque deposition, normalizing tau protein phosphorylation, enhancing autophagy, and restoring synaptic deficits [11 , 22]. On the other hand, another study involving SH-SY5Y cells and Tg6799 AD model mice reported that metformin promoted the production of Aβ, aggravating AD-like pathology [23]. Chen et al. came to a similar conclusion that metformin might increase intra- and extracellular Aβ levels by upregulating β-secretase [24]. In the P301S mutant mouse, metformin reduced tau phosphorylation in the cortex and hippocampus through the AMPK/mTOR and the PP2A pathways, but it increased the levels of insoluble tau species and resulted in severe motor and behavioral defects [25]. It was confirmed that metformin can exert antiglycemic effects by activating AMPK [26], while abnormal upregulation of AMPK has been linked to loss of synapses, dendritic spines, and cognitive impairment [27, 28]. In high-fat diet mice, McNeilly et al. showed that metformin only helped mitigate metabolic dysfunction but not the cognitive impairment [29]. These inconsistent findings demonstrated that the role of metformin in AD remains highly debated.

Since most individuals with T2DM eventually receive antidiabetic drugs, it is important to investigate the relationship between hypoglycemic drugs such as metformin and AD risk [30]. Therefore, we examined the available literature on whether metformin use influences risk of AD in humans. Our findings can provide further insight in the field of T2DM and AD management, shedding lights on other age-related neurodegenerative diseases.

METHODS

Identification of eligible studies

Two reviewers (ALL and PPN) independently and systematically searched PubMed, EMBASE (via Ovid), Web of Science, the Cochrane Central Registry of Controlled Trials (CENTRAL) databases, Chinese National Knowledge Infrastructure, Wanfang, and SinoMed from inception to till December 20, 2021. The following terms were used: Alzheimer’s, dementia, AD, cognition disorders, metformin, antidiabetic drugs, antidiabetic agents, and antidiabetics. All of them were combined using the Boolean operator “OR” or “AND”. There are no language restrictions. The reference lists of included paper were reviewed for additional studies. This study was conducted in accordance with the Meta-analysis of Observational Studies in Epidemiology and the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) [31].

Studies would be included if they met the following conditions: 1) Clearly define or evaluate metformin exposure; 2) Be observational studies in humans with a cross-sectional, case-control or cohort design investigating the relationship between metformin treatment and AD; 3) The diagnosis of AD needs to be clearly stated or be standardized using accepted clinical criteria (e.g., the National Institute of Neurological and Communicative Disorders and Stroke and AD and Related Disorders Association criteria, NINCDS-ADRDA criteria; International Classification of Diseases, ICD; Diagnostic and Statistical Manual of Mental Disorders, DSM); or by a record of anti-dementia drug use; 4) Report either relative risks, hazard ratios, and odds ratios (ORs) or provided quantitative measures of their calculation, and their 95% confidence intervals (CIs). If necessary, only monotherapy, the longest medication time, or the largest cumulative dose are included; 5) If there were more than one study evaluated the same population, only larger studies were included. Exclusion criteria were as follows: 1) Letters, meeting abstracts, meta-analyses, reviews, experimental studies, case reports or editorials; 2) Insufficient data unsuitable for quantitative analysis even after contacting the authors; 3) Cognitive impairment or AD secondary to other reasons such as severe infections, electrolyte disturbances, patients with severe psychiatric disorders, organic brain disorders, drug-related cognitive impairment and so on; 4) Other types of dementia (e.g., vascular, Lewy body, or frontotemporal dementia) or all-cause dementia (There was no subgroup analysis or specified for AD after reviewing the full text); 5) Duplicated data. The flow diagram summarizing study identification and selections is shown in Fig. 1.

Fig. 1

Flow diagram of study selection.

Two authors (ALL and PPN) used the same standardized method to independently screen all titles and abstracts to determine their applicability, including research quality, population characteristics, publication years, and results. If a study meets the initial selection criteria but cannot be decided to be included or not, the full text of the article will be evaluated. The differences included in the study were resolved through discussion. When no consensus was reached, the third expert (YMX) made a decision. The following information was extracted from each article of the 10 included studies and cross-checked by a second researcher: 1) name of first author, 2) study population, 3) country of origin, 4) publication year, 5) study design, 6) length of follow-up, 7) number of the total participants in the study, 8) average age of participants, 9) diagnostic method of the outcomes, 10) OR, relative risk, or valid data for their calculation, and 95% CIs for statistical analysis and so on. If necessary, we would further contact the author to obtain missing information.

Quality assessment

The methodological quality of case–control and cohort studies was assessed by two authors independently using the 9-star Newcastle-Ottawa Scale (NOS) [32]. It is used to assess the quality of non-randomized studies which scored 4, 2, and 3 respectively for selection, comparability, and outcome/exposure respectively, while cross-sectional studies were evaluated by the Agency for Healthcare Research and Quality (AHRQ) methodology checklist with 11 items answering each question with “yes”, “no”, or “don’t know” [33]. Any discrepancies were resolved by a joint re-evaluation of the original article. We consider the studies with a cumulative NOS score of ≥7 or a AHRQ score of ≥8 as high quality.

Statistical analysis

Heterogeneity among studies was evaluated using the Q test and was quantified using I². An I² value below 25% was considered to indicate homogeneity; 25% to just under 50%, low heterogeneity; 50% to just under 75%, moderate heterogeneity; and at least 75%, substantial heterogeneity [34]. When the combined data were classified as homogeneous or low heterogeneity, the fixed effect model was used for meta-analysis, while the random effect model was used for meta-analysis to classify the data as showing moderate or substantial heterogeneity.

Sensitivity analysis is performed by removing each study one by one and repeating the analysis. Publication bias was assessed using Begg’s rank correlation test and Egger’s linear regression test, which are quantitative methods to evaluate publication bias in meta-analysis; Subgroup analysis was used to investigate the source of heterogeneity by several major covariates based on cumulative usage of metformin, region, and study design. Stata/MP 14.0 (Stata Corp LP, Texas, USA) was used for statistical analysis.

RESULTS

Literature screening

A total of 1,670 potentially relevant observational studies were identified from the databases examined. After a detailed assessment based on the eligibility criteria, 10 studies involving a total of 229,110 participants were included in the meta-analysis [9 , 35–39] (Table 1 and Fig. 1).

Table 1

Characteristics and quality of included studies describing the risk of neurodegenerative diseases with metformin use

Study	Country of origin	Study design	Time period	Demographical characteristics between groups	AD diagnosis	Exposure identification	Control identification	Number (exposure vs. control)	Risk estimate (95% CI)	Adjusted variables	Study quality (NOS)
Orkaby et al. [26]	Caucasian (USA)	CS	2001–2012	Not significant after using the IPTW method	AD/ICD-9	DM with Met	DM with SU	Age <75 y 14,562 (10,437 vs. 4,125); Age ≥75 y 14,078 (6,763 vs. 7,315)	Age <75 y, HR 0.93 (0.76–1.14); Age ≥75 y, HR 0.95 (0.82–1.10)	Adjusted for race, sex, BMI, HbA1c, renal function, and comorbidities associated with both the exposure and outcome such as coronary artery disease, stroke, atrial fibrillation, heart failure, hypertension, and cancer	8/9
Huang et al. [9]	Asian (Taiwan)	CS	1997–2007	NR	AD/ICD-9	DM with Met	DM without Met	71,433 (4,978 vs. 66,455	Met as monotherapy 0.69 (0.28–1.71); Met as combination therapy 0.57 (0.26–1.26)	Adjusted for age, sex, comorbidities (including hypertension, hyperlipidemia, stroke, coronary artery disease, arrhythmia, heart failure, and depression), geographic area, and urbanization status	8/9
Weinstein et al. [28]	Caucasian (UK)	CS	1998–2011	NR	AD/NINCDS-ADRDA	DM with Met	DM without Met	NR	1.6 (0.87–2.93)	Adjusted for age, sex and education, Physical activity, hypertension, CVD, stroke, total cholesterol, smoking, depression, BMI, HbA1C, ApoE4, and eGFR.	8/9
Imfeld et al. [20]	Caucasian (UK)	CCS	1998–2008	There were more underweight (BMI ≤18.4 kg/m2) AD cases than controls, and more overweight (BMI 25–29.9 kg/m2) or obese (BMI ≥30 kg/m2) controls than cases	AD/specific dementia tests or AD records	Newly onset AD cases with long-term Met (≥60 prescriptions)	individual without evidence of any type of dementia	14,172 (7,086 vs. 7,086)	1.73 (1.11–2.68	Adjusted for sulfonylurea prescriptions, insulin prescriptions, thiazolidinedione prescriptions, smoking, BMI, hypertension, dyslipidemia, and use of angiotensin-converting enzyme inhibitors and statins	8/9
Sluggett et al. [18]	Caucasian (Finland)	CCS	2005–2011	Cases were more likely to have atrial fibrillation, psychiatric disorders, and coronary artery disease, and less likely to have received antihypertensive therapy than controls;	AD/NINCDS-ADRDA and the DSM-IV	DM (diagnosed ≥3 years before AD) with Met	DM without Met	29,412 (9,862 vs. 19,550)	0.85 (0.76–0.95)	Adjusted for region of residence, occupational social class, cardiovascular disease (stroke, hypertension, coronary artery disease, chronic heart failure, atrial fibrillation, peripheral arterial disease), psychiatric disorders (bipolar, schizophrenia, depression), renal disease, statin use, antihypertensive use, use of sulfonylureas, insulin. and other diabetes medications.	8/9
Wu et al. [17]	Caucasian (Canada)	CCS	2005–2019	NR	AD/NINCDS-ADRDA	DM with Met	DM without Met	2,670 (807 vs. 1,863)	0.67 (0.56–0.78)	No	8/9
Akimoto et al. [29]	Asian (Japan)	CCS	2004–2018	NR	AD/take AD therapy drugs	DM with Met	DM without Met	66,085 (1,250 vs. 64,835)	1.51 (1.35–1.69)	No	7/9
Shi et al. [27]	USA	CS	2004–2010	NR	AD/ICD-9-CM	DM with Met	DM without Met	5,530 (2,774 vs. 2,756)	0.39 (0.18–0.85)	PSW-adjusted variables	6/9
Kuan et al. [19]	Asian (Taiwan)	CS	2000–2010	Not significant	ICD-9-CM	DM with Met for at least 90 days	DM without Met	9,302 (4,651 vs. 4,651)	2.13 (1.20–3.79) for Met use ≥400 days 2.84 (2.12–3.82)	Adjusted for age; sex; Charlson Comorbidity Index; Adapted Diabetes Complications Severity Index; comorbidities of hypertension, chronic kidney disease, hyperlipidemia, heart failure, arrhythmia, stroke, head injury, and coronary artery disease; and medications of antidiabetes mellitus drug, anti-hypertensive drug, and statin.	9/9
Ha et al. [30]	Asian (Korea)	CCS	2002–2017	AD cases were more likely to have depression, to use antiplatelet agents and insulin and less likely to physically active. The proportion of individuals with current smoking status and heavy alcohol intake was higher among cases than controls.	ICD-10	DM with Met	DM without Met	10,050 (1,675 vs. 8,375)	1.50 (1.23–1.83)	Analysis was adjusted for the following covariates: hypertension, ischemic heart disease, dyslipidemia, Charlson Comorbidity Index, Diabetes Complications Severity Index, depression, statin use, aspirin use, antiplatelet use, anticoagulant use, antihypertensive drug use, antiarrhythmic drug use, use of antidiabetic medications, fasting blood glucose levels, systolic blood pressure, diastolic blood pressure, total cholesterol levels, creatinine levels, BMI, smoking status, alcohol consumption, and physical activity.	8/9

AD, Alzheimer’s disease; CI, confidence interval; USA, The United States of America; CS, cohort study; Met, metformin; SU, sulfonylurea; IPTW, inverse probability of treatment weighting; ICD-9, International Classification of Diseases, Ninth Revision; DM, diabetes mellitus; HR, hazard ratio; BMI, body mass index; HbA1c, hemoglobin A1c; NOS, Newcastle-Ottawa Quality Assessment Scale; NR, not reported; UK, The United Kingdom; NINCDS-ADRDA criteria, the National Institute of Neurological and Communicative Disorders and Stroke and AD and Related Disorders Association criteria; CVD, Cardiovascular disease; ApoE4, apolipoprotein E; eGFR, estimated glomerular filtration rate; CCS case–control study; DSM-IV, Diagnostic Statistical Manual Mental Disorders, fourth Revision; DDDs, defined daily doses; aOR, adjusted odds ratio; PSW, propensity score weights; ICD-9-CM, the International Classification of Diseases, Ninth Revision, Clinical Modification; ICD-10, the International Classification of Diseases, Tenth Revision.

Characteristics of included studies

We included five population-based cohort studies [9 , 35–37] and five case-control studies [17 , 39] that explored the risk of AD in patients who received metformin. One longitudinal study [37] provided data on five community-based cohorts (Table 1). All studies had relatively large samples ranging from 2,670 to 66,085 participants [38]. Most studies included comparable numbers of men and women, while two [35, 36] included nearly all men. Four studies were conducted in Asia [9 , 39] and six in Europe or North America [18 , 35–38].

AD was diagnosed based on the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria in three studies [20 , 38], the International Classification of Diseases, Ninth Revision or Diagnostic Statistical Manual Mental Disorders, fifth Revision (ICD-9/DSM-5) criteria in four studies [9 , 36], and the International Classification of Diseases, Tenth Revision (ICD-10) criteria in one study [17], Other studies diagnosed AD based on use of anti-AD drugs [39] or based on medical records [18].

Four of the observational studies [9 , 18–20] addressed the relationship between the cumulative usage of metformin and the risk of AD. These studies used different methods to quantify metformin usage: three studies used total metformin administration time (in days or years) [9 , 20], while one study used prescriptions [18].

Quality of included studies

Most studies adjusted for age, sex, study region body mass index, comorbidities, and other medications when calculating ORs. One study [38] also adjusted for genotype at the apolipoprotein E (APOE) gene since people with the ɛ4 allele have been reported to be at 3.7 times higher risk of AD than those with the ɛ3/ɛ3 genotype [40]. The median quality score for all cohort and case-control studies was 8 (range 6–9, full score of 9), indicating that the studies were of high quality and had large samples (Table 2).

Table 2

Quality assessment of the included studies using Newcastle-Ottawa Scale Assessment Scale

Study	Selection				Comparability	Outcome			Exposure			Total
	1	2	3	4		1	2	3	1	2	3
Cohort study
Shi et al. 2019 [237]	★	★	★	★	★ ★	★	★	★				6
Huang et al. 2019 [9]	★	★	★	★	★ ★	★	★	★				8
Weinstein et al. 2017 [28]	★	★	★	★	★ ★	★	★	★				8
Kuan et al. 2017 [19]	★	★	★	★	★ ★	★	★	★				9
Orkaby et al. 2014 [26]	★	★	★	★	★ ★	★	★	★				8
Case-control study
Ha et al. 2021 [30]	★	★	★	★	★ ★				★	★	★	8
Sluggett et al. 2020 [18]	★	★	★	★	★ ★				★	★	★	8
Wu et al. 2020 [17]	★	★	★	★	★ ★				★	★	★	8
Akimoto et al. 2020 [29]	★	★	★	★	★ ★				★	★	★	7
Imfeld et al. 2012 [20]	★	★	★	★	★ ★				★	★	★	8

Association between metformin use and risk of AD

Based on the meta-analysis, we found that metformin did not significantly affect risk of AD (OR 1.17, 95% CI 0.88–1.56, p = 0.291; Fig. 2). However, there was considerable heterogeneity among the studies (I² = 94.2%, Cochran’s Q test p < 0.001). When the meta-analysis was repeated after removing studies individually, the pooled OR ranged from 0.89 to 1.43 (Fig. 4). There was no indication of publication bias based on Begg’s test (Z = –0.09, p = 0.929; Fig. 5) and Egger’s test (t = 0.33, p = 0.752; Fig. 6).

Fig. 2

Forest plot of all studies in the meta-analysis of metformin exposure on AD risk.

Fig. 3

Adjusted OR for metformin exposure on AD subgrouped by regions.

Fig. 4

Sensitivity analysis in which all studies except the one indicated in each row were meta-analyzed for AD risk.

Fig. 5

Detection of publication bias using Begg’s funnel plot.

Fig. 6

Egger test for publication bias.

Subgroup analyses

To determine the effects of metformin treatment on risk of AD, we performed subgroup analysis based on the cumulative use of metformin, study design, and region. Long-term use of metformin did not significantly alter risk of AD (OR 1.29, 95% CI 0.75–2.22, p = 0.359; Supplementary Figure 1). For the study design, 5 cohort studies yielded an OR of 1.09 (95% CI 0.56–2.11, p = 0.798; Supplementary Figure 2). The pooled OR of 5 case-control studies was 1.18 (95% CI 0.81–1.70, p = 0.389; Supplementary Figure 2). It is consistent with the pooled OR of all studies, suggesting that the outcomes of this meta-analysis were reliable. For study region, among Asian participants, those who received metformin were at significantly higher risk of AD than those who did not (OR 1.71, 95% CI 1.24–2.37, p = 0.001; Fig. 3). In contrast, among non-Asian participants, metformin did not significantly alter AD risk (OR 0.92, 95% CI 0.72–1.16, p = 0.462; Fig. 3).

DISCUSSION

In this meta-analysis, we systematically analyzed clinical observational studies to determine the effect of metformin, a widely prescribed antidiabetic drug [41], on the risk of developing AD. We performed a comprehensive meta-analysis of data collected from 229,110 participants and found that metformin did not appear to alter risk of AD (pooled adjusted OR 1.17, 95% CI 0.88–1.56, p = 0.291). The same was observed for long-term use of metformin (OR 1.29, 95% CI 0.75–2.22, p = 0.359). However, we found that metformin used by Asian participants increased their risk of AD (OR 1.71, 95% CI 1.24–2.37, p = 0.001).

Consistent with the present work, previous studies failed to detect a significant effect of metformin on risk of AD [9] or cognitive function [37] in individuals with diabetes. We are aware of only two randomized controlled clinical trials that assessed the effect of metformin on cognitive deficits in individuals with amnestic mild cognitive impairment or mild dementia [42, 43]. Both studies, which had small samples, found minimal effects of metformin. Ping et al. recently conducted a meta-analysis on the correlation between metformin exposure and risk of neurodegenerative diseases. Specifically, they conducted a subgroup analysis on AD. Though only three studies were included, it found that metformin treatment neither increase nor reduce the risk of AD [44]. Another meta-analysis concluded that metformin sensitizers including metformin or thiazolidinediones had a marginal significant trend for dementia or AD, and in the subgroup analysis of non-diabetic subjects, metformin did not decrease the incidence of dementia [45]. Our results may be even more reliable than those from previous work because we meta-analyzed 10 large studies. In contrast, two meta-analyses reported that metformin could reduce risk of cognitive impairment or dementia in individuals with diabetes [46, 47]. However, the network meta-analysis failed to take into account confounding due to the use of combinations of anti-glycemic drugs and due to other types of heterogeneity among studies [47]. In any case, further research is required in order to clarify whether metformin has beneficial effects on cognitive function and whether it increases or decreases risk of dementia and AD.

Other molecular crosstalk may be involved. Metformin may show a sex bias in its effects. It is reported that female mice had higher Aβ loads than males [48, 49]. However, a study by DiTacchio et al. showed that metformin was only cognitively protective in female mice [50]. The efficacy of metformin may also depend on APOE genotype. For example, metformin may be less effective than dipeptidyl peptidase 4 inhibitors in AD patients carrying the APOE4 allele [38]. Since it might be multifactorial, large, well-designed clinical trials with rigorous subgroup analyses are needed in order to clarify how the effects of metformin on AD risk depend on sex, APOE status, and diabetic and non-diabetic populations. Besides, the animal (or cellular) models of AD are not perfect, and these results are not completely applicable to humans [51]. Therefore, more faithful animal models that are representative to human AD cases are also needed in future studies.

Based on our subgroup analyses, long-term use of metformin did not significantly affect risk of AD. The literature is unclear on whether greater cumulative dose or longer-term use of metformin affects risk of AD. One study reported that metformin users who had received ≥60 prescriptions were at greater risk of AD [18]. Conversely, another study reported that individuals with T2DM who had been receiving metformin for >4 years were at reduced risk of AD [36], whereas those on metformin treatment for <1 year were at increased risk. Similarly, a nested case-control study showed that 1–3 years of metformin treatment could increase the risk of AD, while long-term (≥10 years) and high-dose (average 2 g per day) metformin use could have a protective effect [20]. However, these results should be interpreted carefully, since very few subjects in that study received high-dose metformin [20]. In addition, metformin is not applicable to patients with type 1 diabetes, as well as those who suffer from renal insufficiency or insufficient insulin secretion. Other confounding factors such as prodromal symptoms of AD (increased number of visits to the doctor) can help in the early identification of T2DM, thus increasing the likelihood of using metformin. Future studies must include long-term follow up in order to gain a better understanding of the effects of metformin treatment on risk of AD onset or progression.

Our study is the first to report a significantly higher risk of AD among Asian people who received metformin treatment than those who did not (OR 1.71, 95% CI 1.24–2.37, p = 0.001). Therefore, clinicians treating Asians with metformin must consider the possibility of an increased risk of developing AD. This association may have multiple explanations. First, metformin can increase levels of amyloid-β protein precursor (AβPP) and presenilin, which may exacerbate Aβ aggregation [24, 52]. Second, metformin treatment has been associated with cognitive impairment in individuals with diabetes [53, 54], which has been linked to reduced levels of vitamin B12 [55 –57]. In fact, reduced B12 levels have been associated with AD and Parkinson’s disease [58]. Third, previous research has shown that metformin can exert antiglycemic effects by activating AMPK [26], However, if AMPK was abnormally upregulated, it would lead to loss of synapses, dendritic spines and cognitive impairment [27, 28].

Even in Asian populations, clinical studies have reported inconclusive results on the relationship between metformin and AD. Among the four Asian studies included, metformin has a neutral effect on AD [9] or increases the risk of AD [17 , 39]. In the cohort study of Huang et al., they reported that the use of metformin, either monotherapy or combination therapy, does not alter the risk of AD even after adjusting for relevant influencing factors [9]. Another 12-year follow-up cohort study showed that metformin was found to be associated with an increased risk of AD. Besides, long-term exposure to metformin would contribute to neurodegenerative diseases, including PD and dementia [19]. Similarly, two case-control studies found that metformin increased the risk of AD [17] or that metformin use was associated with a higher risk of AD compared with rosiglitazone and glucagon-like peptide 1 receptor agonists [39], respectively. Both studies had comparatively large sample sizes. In addition, Cheng et al. [59] conducted a cohort study in Taiwan involved 67,731 participants aged 65 or older, they found that patients taking metformin a lower risk of dementia than thiazolidinediones (TZDs). However, conclusions should be extrapolated carefully, for it included relatively small number of TZDs users. In addition, the outcome is ‘all-cause dementia’, containing dementias other than AD. Furthermore, the effect of monotherapy or combination, duration of antidiabetic medication and severity of diabetes was not discussed. A cross-sectional study with conducted in Singapore showed that long-term (>6 years) metformin use could protect diabetic patients against cognitive impairment [16]. However, with a relatively small sample size (365 individuals involved), the outcome of this study was not AD, but Mini-Mental State Examination scores to determine whether patients have cognitive impairment. Differences in study designs, methods, sample sizes, and so on among studies may explain these conflicting results. Therefore, large-scale, multicenter, and well-designed studies are warranted in the future to further clarify the relationship between metformin exposure and AD in different regions or populations.

In fact, Asian countries, such as China, India, Japan, and Indonesia, are four of the seven countries with the largest number of dementia patients in the world [2]. 60% of people with dementia live in developing countries, which is expected to increase to 71% by 2040 [4]. Previous research has confirmed higher prevalence of cardiovascular disease in Singapore and other Asian countries [60, 61], and such disease has been linked to greater Aβ deposition and faster neurodegeneration, which can translate to cognitive decline [62]. However, our results should be interpreted with caution, as there may be confounding factors. There were only four studies included (two in Taiwan, one in Japan, and one in South Korea) and might be hardly representative. Further work should explore to what extent the greater AD risk among Asian samples in our meta-analysis apply to the entire region.

Strengths and limitations of the study

The main advantage of the present study is that it used a sufficiently large sample to provide an updated summary of the latest research on the association between metformin use and risk of AD. However, given the limited number of studies that have explored this association, the results of our study should be interpreted conservatively. Our analysis was limited to clinical observational studies with different study designs and baseline characteristics, thus increasing the heterogeneity among studies. Nevertheless, our sensitivity analysis suggests that the pooled results are reliable. For lack of relevant data, we were also unable to meta-analyze the effects of APOE genotype, exact threshold of metformin exposure time, and serum levels of vitamin B12 on AD risk. Further research is required to understand the effects of metformin on AD risk in individuals with or without T2DM, as well as the effect of hypoglycemic combination therapy on risk of AD.

Conclusion

Based on a total of 10 clinical observational studies, this systematic review and meta-analysis showed that metformin exposure does not significantly affect risk of AD, but increased risk of AD in Asian people. Large prospective studies with adequate follow-up should be conducted in order to validate and extend these results.

Footnotes

ACKNOWLEDGMENTS

The authors thank Dr Chapin for his advice on the English revision and valuable statistical suggestions.

This work was supported by a grant from Sichuan Provincial Health Commission (21PJ031).

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

GBD 2016 Dementia Collaborators (2019) Global, regional, and national burden of Alzheimer’s disease and other dementias, 1990-2016: A systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol 18, 88–106.

(2021) 2021 Alzheimer’s disease facts and figures. Alzheimers Dement 17, 327–406.

Cahill

(2020) WHO’s global action plan on the public health response to dementia: Some challenges and opportunities. Aging Ment Health 24, 197–199.

Ferri

, Prince

, Brayne

, Brodaty

, Fratiglioni

, Ganguli

, Hall

, Hasegawa

, Hendrie

, Huang

, Jorm

, Mathers

, Menezes

, Rimmer

, Scazufca

, Alzheimer’s Disease International (2005) Global prevalence of dementia: A Delphi consensus study. Lancet 366, 2112–2117.

Berk

, Paul

, Sabbagh

(2014) Investigational drugs in Alzheimer’s disease: Current progress. Expert Opin Investig Drugs 23, 837–846.

Siemers

, Sundell

, Carlson

, Case

, Sethuraman

, Liu-Seifert

, Dowsett

, Pontecorvo

, Dean

, Demattos

(2016) Phase 3 solanezumab trials: Secondary outcomes in mild Alzheimer’s disease patients. Alzheimers Dement 12, 110–120.

Salloway

, Sperling

, Fox

, Blennow

, Klunk

, Raskind

, Sabbagh

, Honig

, Porsteinsson

, Ferris

, Reichert

, Ketter

, Nejadnik

, Guenzler

, Miloslavsky

, Wang

, Lu

, Lull

, Tudor

, Liu

, Grundman

, Yuen

, Black

, Brashear

, Bapineuzumab 301 and 302 Clinical Trial Investigators (2014) Two phase 3 trials of bapineuzumab in mild-to-moderate Alzheimer’s disease. N Engl J Med 370, 322–333.

Barnes

, Yaffe

(2011) The projected effect of risk factor reduction on Alzheimer’s disease prevalence. Lancet Neurol 10, 819–828.

Huang

, Chung

, Leu

, Lin

, Chiu

, Hsu

, Chiang

, Huang

, Chen

, Lin

, Chen

, Chan

(2014) Diabetes mellitus and the risk of Alzheimer’s disease: A nationwide population-based study. PLoS One 9, e87095.

10.

Chung

, Erion

, Florez

, Hattersley

, Hivert

, Lee

, McCarthy

, Nolan

, Norris

, Pearson

, Philipson

, McElvaine

, Cefalu

, Rich

, Franks

(2020) Precision medicine in diabetes: A Consensus Report from the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetologia 63, 1671–1693.

11.

, Kong

, Sun

, He

, Zhang

, Gong

, Huang

, Xu

, Long

, Li

, Xu

, Xuan

(2018) Metformin treatment prevents amyloid plaque deposition and memory impairment in APP/PS1 mice. Brain Behav Immun 69, 351–363.

12.

Cassano

, Leo

, Tallarico

, Nesci

, Cimellaro

, Fiorentino

, Citraro

, Hribal

, De Sarro

, Perticone

, Sesti

, Russo

, Sciacqua

(2020) Metabolic and cognitive effects of ranolazine in type 2 diabetes mellitus: Data from an in vivo model. Nutrients 12, 382.

13.

Farr

, Roesler

, Niehoff

, Roby

, McKee

, Morley

(2019) Metformin improves learning and memory in the SAMP8 mouse model of Alzheimer’s disease. J Alzheimers Dis 68, 1699–1710.

14.

Syal

, Kosaraju

, Hamilton

, Aumont

, Chu

, Sarma

, Thomas

, Seegobin

, Dilworth

, He

, Wondisford

, Zimmermann

, Parent

, Fernandes

, Wang

(2020) Dysregulated expression of monoacylglycerol lipase is a marker for anti-diabetic drug metformin-targeted therapy to correct impaired neurogenesis and spatial memory in Alzheimer’s disease. Theranostics 10, 6337–6360.

15.

Chin-Hsiao

(2019) Metformin and the risk of dementia in type 2 diabetes patients. Aging Dis 10, 37–48.

16.

, Feng

, Yap

, Lee

, Tan

, Winblad

(2014) Long-term metformin usage and cognitive function among older adults with diabetes. J Alzheimers Dis 41, 61–68.

17.

, Choi

, Kim

, Cho

, Kim

, Koh

, Nam

, Kim

(2021) Association of metformin use with Alzheimer’s disease in patients with newly diagnosed type 2 diabetes: A population-based nested case-control study. Sci Rep 11, 24069.

18.

Imfeld

, Bodmer

, Jick

, Meier

(2012) Metformin, other antidiabetic drugs, and risk of Alzheimer’s disease: A population-based case-control study. J Am Geriatr Soc 60, 916–921.

19.

Kuan

, Huang

, Lin

, Hu

, Kao

(2017) Effects of metformin exposure on neurodegenerative diseases in elderly patients with type 2 diabetes mellitus. Prog Neuropsychopharmacol Biol Psychiatry 79, 77–83.

20.

Sluggett

, Koponen

, Bell

, Taipale

, Tanskanen

, Tiihonen

, Uusitupa

, Tolppanen

, Hartikainen

(2020) Metformin and risk of Alzheimer’s disease among community-dwelling people with diabetes: A national case-control study. J Clin Endocrinol Metab 105, dgz234.

21.

Niccoli

, Cabecinha

, Tillmann

, Kerr

, Wong

, Cardenes

, Vincent

, Bettedi

, Li

, Gronke

, Dols

, Partridge

(2016) Increased glucose transport into neurons rescues Abeta toxicity in drosophila. Curr Biol 26, 2550.

22.

Chen

, Zhao

, Fan

, Li

, Zhu

, Shen

, Li

, Yan

, Tian

, Liu

, Zhang

(2021) Metformin attenuates plaque-associated tau pathology and reduces amyloid-beta burden in APP/PS1 mice. Alzheimers Res Ther 13, 40.

23.

Son

, Shin

, Byun

, Kook

, Moon

, Chang

, Mook-Jung

(2016) Metformin facilitates amyloid-beta generation by beta- and gamma-secretases via autophagy activation. J Alzheimers Dis 51, 1197–1208.

24.

Chen

, Zhou

, Wang

, Liu

, Kwak

, Ma

, Thompson

, Zhao

, Smith

, Gasparini

, Luo

, Xu

, Liao

(2009) Antidiabetic drug metformin (GlucophageR) increases biogenesis of Alzheimer’s amyloid peptides via up-regulating BACE1 transcription. Proc Natl Acad Sci U S A 106, 3907–3912.

25.

Barini

, Antico

, Zhao

, Asta

, Tucci

, Catelani

, Marotta

, Xu

, Gasparini

(2016) Metformin promotes tau aggregation and exacerbates abnormal behavior in a mouse model of tauopathy. Mol Neurodegener 11, 16.

26.

Hawley

, Gadalla

, Olsen

, Hardie

(2002) The antidiabetic drug metformin activates the AMP-activated protein kinase cascade via an adenine nucleotide-independent mechanism. Diabetes 51, 2420–2425.

27.

, Chen

, Vingtdeux

, Zhao

, Viollet

, Marambaud

, Klann

(2014) Inhibition of AMP-activated protein kinase signaling alleviates impairments in hippocampal synaptic plasticity induced by amyloid beta. J Neurosci 34, 12230–12238.

28.

Mairet-Coello

, Courchet

, Pieraut

, Courchet

, Maximov

, Polleux

(2013) The CAMKK2-AMPK kinase pathway mediates the synaptotoxic effects of Abeta oligomers through Tau phosphorylation. Neuron 78, 94–108.

29.

McNeilly

, Williamson

, Balfour

, Stewart

, Sutherland

(2012) A high-fat-diet-induced cognitive deficit in rats that is not prevented by improving insulin sensitivity with metformin. Diabetologia 55, 3061–3070.

30.

Oishi

, Yamazaki

, Okuguchi

, Sugimoto

, Kanatsuka

, Kashiwagi

, Japan Diabetes Clinical Data Management Study Group (2014) Changes in oral antidiabetic prescriptions and improved glycemic control during the years 2002-2011 in Japan (JDDM32). J Diabetes Investig 5, 581–587.

31.

Page

, McKenzie

, Bossuyt

, Boutron

, Hoffmann

, Mulrow

, Shamseer

, Tetzlaff

, Akl

, Brennan

, Chou

, Glanville

, Grimshaw

, Hrobjartsson

, Lalu

, Li

, Loder

, Mayo-Wilson

, McDonald

, McGuinness

, Stewart

, Thomas

, Tricco

, Welch

, Whiting

, Moher

(2021) The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ 372, n71.

32.

Stang

(2010) Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in meta-analyses. Eur J Epidemiol 25, 603–605.

33.

Shekelle

, Ortiz

, Rhodes

, Morton

, Eccles

, Grimshaw

, Woolf

(2001) Validity of the Agency for Healthcare Research and Quality clinical practice guidelines: How quickly do guidelines become outdated? JAMA 286, 1461–1467.

34.

Higgins

, Altman

, Gotzsche

, Juni

, Moher

, Oxman

, Savovic

, Schulz

, Weeks

, Sterne

, Cochrane Bias Methods Group; Cochrane Statistical Methods Group (2011) The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 343, d5928.

35.

Orkaby

, Cho

, Cormack

, Gagnon

, Driver

(2017) Metformin vs sulfonylurea use and risk of dementia in US veterans aged > /=65 years with diabetes. Neurology 89, 1877–1885.

36.

Shi

, Liu

, Fonseca

, Thethi

, Shi

(2019) Effect of metformin on neurodegenerative disease among elderly adult US veterans with type 2 diabetes mellitus. BMJ Open 9, e024954.

37.

Weinstein

, Davis-Plourde

, Conner

, Himali

, Beiser

, Lee

, Rawlings

, Sedaghat

, Ding

, Moshier

, van Duijn

, Beeri

, Selvin

, Ikram

, Launer

, Haan

, Seshadri

(2019) Association of metformin, sulfonylurea and insulin use with brain structure and function and risk of dementia and Alzheimer’s disease: Pooled analysis from 5 cohorts. PLoS One 14, e0212293.

38.

, Ouk

, Wong

, Anita

, Edwards

, Yang

, Shah

, Herrmann

, Lanctot

, Kapral

, MacIntosh

, Rabin

, Black

, Swardfager

(2020) Relationships between memory decline and the use of metformin or DPP4 inhibitors in people with type 2 diabetes with normal cognition or Alzheimer’s disease, and the role APOE carrier status. Alzheimers Dement 16, 1663–1673.

39.

Akimoto

, Negishi

, Oshima

, Wakiyama

, Okita

, Horii

, Inoue

, Ohshima

, Kobayashi

(2020) Antidiabetic drugs for the risk of Alzheimer disease in patients with type 2 DM using FAERS. Am J Alzheimers Dis Other Demen 35, 1533317519899546.

40.

Huynh

, Liao

, Francis

, Robinson

, Serrano

, Jiang

, Roh

, Finn

, Sullivan

, Esparza

, Stewart

, Mahan

, Ulrich

, Cole

, Holtzman

(2017) Age-dependent effects of apoE reduction using antisense oligonucleotides in a model of beta-amyloidosis. Neuron 96, 1013–1023e1014.

41.

Chung

, Erion

, Florez

, Hattersley

, Hivert

, Lee

, McCarthy

, Nolan

, Norris

, Pearson

, Philipson

, McElvaine

, Cefalu

, Rich

, Franks

(2020) Precision medicine in diabetes: A consensus report from the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care 43, 1617–1635.

42.

Luchsinger

, Perez

, Chang

, Mehta

, Steffener

, Pradabhan

, Ichise

, Manly

, Devanand

, Bagiella

(2016) Metformin in amnestic mild cognitive impairment: Results of a pilot randomized placebo controlled clinical trial. J Alzheimers Dis 51, 501–514.

43.

Koenig

, Mechanic-Hamilton

, Xie

, Combs

, Cappola

, Xie

, Detre

, Wolk

, Arnold

(2017) Effects of the insulin sensitizer metformin in Alzheimer disease: Pilot data from a randomized placebo-controlled crossover study. Alzheimer Dis Assoc Disord 31, 107–113.

44.

Ping

, Jiang

, Li

(2020) Association between metformin and neurodegenerative diseases of observational studies: Systematic review and meta-analysis. BMJ Open Diabetes Res Care 8, e001370.

45.

, Luo

, Xiao

, Yu

, Yi

(2016) Impact of insulin sensitizers on the incidence of dementia: A meta-analysis. Dement Geriatr Cogn Disord 41, 251–260.

46.

Campbell

, Stephenson

, de Courten

, Chapman

, Bellman

, Aromataris

(2018) Metformin use associated with reduced risk of dementia in patients with diabetes: A systematic review and meta-analysis. J Alzheimers Dis 65, 1225–1236.

47.

Zhou

, Tang

, Han

, Yang

, Simo

(2020) Impact of antidiabetic agents on dementia risk: A Bayesian network meta-analysis. Metabolism 109, 154265.

48.

Matthes

, Hettich

, Schilling

, Flores-Dominguez

, Blank

, Wiglenda

, Buntru

, Wolf

, Weber

, Vorberg

, Dagane

, Dittmar

, Wanker

, Ehninger

, Krauss

(2018) Inhibition of the MID1 protein complex: A novel approach targeting APP protein synthesis. Cell Death Discov 4, 4.

49.

Placanica

, Zhu

, Li

(2009) Gender- and age-dependent gamma-secretase activity in mouse brain and its implication in sporadic Alzheimer disease. PLoS One 4, e5088.

50.

DiTacchio

, Heinemann

, Dziewczapolski

(2015) Metformin treatment alters memory function in a mouse model of Alzheimer’s disease. J Alzheimers Dis 44, 43–48.

51.

Puzzo

, Gulisano

, Palmeri

, Arancio

(2015) Rodent models for Alzheimer’s disease drug discovery. Expert Opin Drug Discov 10, 703–711.

52.

Picone

, Nuzzo

, Caruana

, Messina

, Barera

, Vasto

, Di Carlo

(2015) Metformin increases APP expression and processing via oxidative stress, mitochondrial dysfunction and NF-kappaB activation: Use of insulin to attenuate metformin’s effect. Biochim Biophys Acta 1853, 1046–1059.

53.

Moore

, Mander

, Ames

, Kotowicz

, Carne

, Brodaty

, Woodward

, Boundy

, Ellis

, Bush

, Faux

, Martins

, Szoeke

, Rowe

, Watters

, AIBL

Investigators

(2013) Increased risk of cognitive impairment in patients with diabetes is associated with metformin. Diabetes Care 36, 2981–2987.

54.

Koo

, Kim

, Lee

, Moon

(2019) Taking metformin and cognitive function change in older patients with diabetes. Geriatr Gerontol Int 19, 755–761.

55.

Khattar

, Khaliq

, Vaney

, Madhu

(2016) Is metformin-induced vitamin B12 deficiency responsible for cognitive decline in type 2 diabetes? Indian J Psychol Med 38, 285–290.

56.

Chapman

, Darling

, Brown

(2016) Association between metformin and vitamin B12 deficiency in patients with type 2 diabetes: A systematic review and meta-analysis. Diabetes Metab 42, 316–327.

57.

Kancherla

, Elliott

Jr , Patel

, Holland

, Johnson

2nd , Khakharia

, Phillips

, Oakley

Jr. , Vaughan

(2017) Long-term metformin therapy and monitoring for vitamin B12 deficiency among older veterans. J Am Geriatr Soc 65, 1061–1066.

58.

Moore

, Mander

, Ames

, Carne

, Sanders

, Watters

(2012) Cognitive impairment and vitamin B12: A review. Int Psychogeriatr 24, 541–556.

59.

Cheng

, Lin

, Tsai

, Chou

, Lan

(2014) Type 2 diabetes and antidiabetic medications in relation to dementia diagnosis. J Gerontol A Biol Sci Med Sci 69, 1299–1305.

60.

Hilal

, Mok

, Youn

, Wong

, Ikram

, Chen

(2017) Prevalence, risk factors and consequences of cerebral small vessel diseases: Data from three Asian countries. J Neurol Neurosurg Psychiatry 88, 669–674.

61.

Lam

BYK

, Yiu

, Ampil

, Chen

, Dikot

, Dominguez

, Ganeshbhai

, Hilal

, Kandiah

, Kim

, Lee

, Ong

, Senanarong

, Leung

, Wang

, Yang

, Yong

, Arshad

, Alladi

, Wong

, Ko

, Lau

AYL

, Mok

VCT

(2021) High burden of cerebral white matter lesion in 9 Asian cities. Sci Rep 11, 11587.

62.

Petrovitch

, Ross

, Steinhorn

, Abbott

, Markesbery

, Davis

, Nelson

, Hardman

, Masaki

, Vogt

, Launer

, White

(2005) AD lesions and infarcts in demented and non-demented Japanese-American men. Ann Neurol 57, 98–103.