Effect of Metformin on Plasma and Cerebrospinal Fluid Biomarkers in Non-Diabetic Older Adults with Mild Cognitive Impairment Related to Alzheimer’s Disease

Abstract

Background:

Alzheimer’s disease (AD) is a complicated condition involving multiple metabolic and immunologic pathophysiological processes that can occur with the hallmark pathologies of amyloid-β, tau, and neurodegeneration. Metformin, an anti-diabetes drug, targets several of these disease processes in in vitro and animal studies. However, the effects of metformin on human cerebrospinal fluid (CSF) and plasma proteins as potential biomarkers of treatment remain unexplored.

Objective:

Using proteomics data from a metformin clinical trial, identify the impact of metformin on plasma and CSF proteins.

Methods:

We analyzed plasma and CSF proteomics data collected previously (ClinicalTrials.gov identifier: NCT01965756, conducted between 2013 and 2015), and conduced bioinformatics analyses to compare the plasma and CSF protein levels after 8 weeks of metformin or placebo use to their baseline levels in 20 non-diabetic patients with mild cognitive impairment (MCI) and positive AD biomarkers participants.

Results:

50 proteins were significantly (unadjusted p < 0.05) altered in plasma and 26 in CSF after 8 weeks of metformin use, with 7 proteins in common (AZU1, CASP-3, CCL11, CCL20, IL32, PRTN3, and REG1A). The correlation between changes in plasma and CSF levels of these 7 proteins after metformin use relative to baseline levels was high (r = 0.98). The proteins also demonstrated temporal stability.

Conclusions:

Our pilot study is the first to investigate the effect of metformin on plasma and CSF proteins in non-diabetic patients with MCI and positive AD biomarkers and identifies several candidate plasma biomarkers for future clinical trials after confirmatory studies.

Keywords

Alzheimer’s disease biomarkers cerebrospinal fluid clinical trial metformin plasma

INTRODUCTION

Alzheimer’s disease (AD) is a neurodegenerative disease of older age, affecting approximately 1 in 9 individuals 65 and older [1]. While defined by its signature amyloid-β (Aβ) and tau pathologies, it is a complex and biologically heterogeneous disorder, with multiple pathophysiological processes driving its emergence, clinical expression, and progression. These include metabolic derangements, inflammation and immune dysregulation, oxidative stress, neurovascular injury, and other factors. Indeed, the clinical benefits of recently approved disease-modifying anti-amyloid monoclonal antibodies are modest, and halting or preventing AD might only be achieved by targeting concurrent pathophysiological processes [2].

Metformin, a biguanide derivative, is commonly used as first-line treatment for insulin resistance in type 2 diabetes mellitus (T2DM). However, it has garnered growing interest in aging and dementia due to its potential effects on extending lifespan and improving cognitive function. Metformin activates 5’ adenosine monophosphate-activated protein kinase (AMPK), which may lead to various positive effects on cellular metabolism, including improved mitochondrial function and increased stress resistance, potentially slowing down neurodegeneration. Preclinical studies show that metformin decreases accumulation of senescent cells, which are thought to contribute to aging and various age-related diseases. Insulin resistance and glucose dysregulation are shared features of T2DM and AD [3]. Several observational studies note a beneficial association between metformin use and population-based risk of AD [4 –7], and a recent causal inference-based analysis by our colleagues associated metformin with lower cause-specific hazard of dementia onset [8]. Aside from insulin sensitization effects, metformin may also target other aspects of AD pathophysiology [9] by decreasing Aβ plaque formation [10], reducing oxidative stress, which can reduce neuroinflammation and support neuroprotection [11, 12], and through restoration of hippocampal neurogenesis [13], which can alleviate cognitive impairment in several in vivo and in vitro models of AD [14, 15].

We and others have previously reported positive outcomes from randomized controlled trials of metformin in mild cognitive impairment (MCI). Luchsinger et al. [16], in a study of 80 overweight individuals with amnestic MCI, found that twelve months of metformin treatment significantly improved total recall of the Selective Reminding Test, without improving scores on the Alzheimer’s Disease Assessment Scale-cognitive subscale (ADAS-Cog). We too conducted a previous placebo-controlled crossover study of metformin on 20 non-diabetic subjects with MCI or mild dementia due to AD [17]. Participants were randomized to receive either metformin or a placebo for 8 weeks and vice versa for 8 weeks, with plasma and cerebrospinal fluid (CSF) collected for biomarker analyses, as well as neuroimaging and cognitive data. Metformin was safe, well-tolerated, and measurable in CSF at an average steady-state concentration of 95.6 ng/ml. Notably, metformin use was associated with improved executive functioning, and there were indications of potential improvements in learning/memory and attention. Overall, trial findings suggest that metformin may hold promise as a therapeutic option for AD, with positive effects on cognitive function, warranting further investigation in more extensive and longer-term studies. Planned clinical trials, such as ‘Targeting Aging with Metformin (TAME)’ [18] will benefit from using more sensitive, specific, and temporally stable biomarkers of the metformin effect; of particular value may be the identification of a subset of plasma proteins that are both modulated by metformin and show similar metformin effects in CSF, suggestive of a possible relationship of these biomarkers with the central nervous system. Here we investigate the effects of metformin on plasma and CSF proteins with bioinformatics analysis of proteomics data collected from our placebo-controlled study [17]. To understand how plasma protein levels were related to central nervous system changes, we compared changes in CSF and plasma proteins. Additionally, we explored the relationship between AD biomarkers and biomarkers related to the action of metformin, to see if AD biomarkers are affected by metformin use.

MATERIALS AND METHODS

Trial design, demographics, and biofluid collection

Details of the original clinical trial (conducted between 2013 and 2015, ClinicalTrials.gov NCT01965756) can be found elsewhere [17]. The clinical study was approved by the University of Pennsylvania’s Institutional Review Board. In brief, enrollment criteria allowed for subjects aged 55–80 with no known history of diabetes or pre-diabetes, and with a diagnosis of mild cognitive impairment or early dementia due to AD (Clinical Dementia Rating scale (CDR)-Global ≤1, screening Mini-Mental State Examination (MMSE)>19), and with at least one biomarker consistent with AD (e.g., CSF analysis, fluorodeoxyglucose-positron emission tomography (FDG-PET), amyloid-PET). 19 of 20 individuals had CDR global scores of 0.5, and one individual with a CDR global of 1. Participants were randomized 1 : 1 to receive metformin (2000 mg/d) for 8 weeks followed by placebo for 8 weeks (Group A; n = 10), or placebo for 8 weeks followed by metformin for 8 weeks (Group B; n = 10). Metformin dosing began with 500 mg daily, followed by an increase of 500 mg/d in divided doses each week until 2000 mg/d was reached (week 4). Plasma was collected at baseline, week 8, and week 16, and CSF was collected at baseline and week 8. All subjects completed baseline CSF collection. Nine of 10 subjects in Group A, and 8 of 10 subjects in Group B agreed to repeat CSF collection at 8 weeks. See Fig. 1 for a diagram of the study design and sample collection. Blood (collected in EDTA-containing tubes) and CSF (collected in uncoated polypropylene tubes) were collected in the mornings, and plasma and CSF aliquots were aliquoted in polypropylene tubes, immediately frozen, and stored at –80°C. Subject demographics included 9 women and 11 men, all Caucasian with 1 Hispanic individual. The mean age was 70.1 years (SD 6.89).

Fig. 1

Study design and fluid sample collection timing. Individuals were randomized to groups A or B. Baseline and 8-week blood and cerebrospinal fluid were collected. Blood only was collected at 16 weeks. Group A was treated with metformin for the first 8 weeks followed by a cross-over to placebo treatment (washout period). In contrast, group B was treated with a placebo drug for 8 weeks (biotemporal stability period), followed by 8-week treatment with metformin.

Biofluid assays

Plasma and CSF samples were analyzed in duplicate using O-link proximity extension assays, performed by O-link (Boston, MA). A total of 60 plasma and 37 CSF samples were analyzed on the following panels: 1) Cardiometabolic (v.3602), 2) Cardiovascular III (v.6113), 3) Immuno-oncology (v.3111), 4) Inflammation (v.3021), and 5) Neuro Exploratory (v.3911), for a total of 454 proteins that passed QC (374 unique after removing duplicates). Relative protein levels were expressed as Normalized Protein eXpression (NPX), in log2 scale, based on quantitative polymerase chain reaction (qPCR) differences in cycle threshold (Ct) values [19]. Sample quality was assessed based on evaluating the deviation of each sample from the median value of the controls for each individual sample. Samples that deviate less than 0.3 NPX from the median pass quality control. All but one plasma sample passed quality control. The intra-assay coefficient of variance (% CV) was determined based on control samples (pooled plasma) on each plate. Most plasma proteins of each grouped assay had <5 % CV (average 67.0% of proteins per assay, standard deviation 8.0%), with very few having a >15 % CV (average 0.86% of proteins per assay, standard deviation 0.90%). 91.1% of plasma proteins were detectable (standard deviation 8.3%). All CSF samples passed quality control. The intra-assay CV varied by assay, with most proteins per assay having <5 % CV in the Cardiometabolic, Immuno-oncology, and Neuro Exploratory panels. Most proteins from the Cardiovascular III and Inflammation panels’ % CV fell into the 10–15% range. At least 50% of proteins from all panels were detectable in CSF (91, 91, 72, 66, and 50% detected proteins in panels 1–5, respectively). All QC-passing NPX values were included in analyses, regardless of % CV, and missing values corresponded to the samples that failed QC or missing data from individuals who did not undergo a repeat lumbar puncture.

Data analyses and statistics

Metformin effects refer to the within-subject fold-change in protein levels of plasma or CSF collected at 8 weeks after the start of metformin treatment relative to the baseline level (Group A –plasma and CSF: 8 weeks versus baseline, Group B –plasma only: week 16 versus 8 weeks). Washout effects refer to within-subject plasma changes from week 8 to week 16 in Group A. Biotemporal stability is captured in Group B and is based on within-subject differences of plasma between baseline and week 8, during which only a placebo was taken. Results are reported as unadjusted p-values≤0.05 and adjusted p-values (by False Discovery Rate).

Data was analyzed using R version 4.2.1. Differential protein levels were identified using the limma package in R, adjusting for within-subject correlations between baseline and post-treatment sample levels. The cor.test function was used to compute Pearson correlation. The ggplot2 package was used to create the volcano and scatter plots. Heatmaps were generated using the ComplexHeatmap package in R.

RESULTS

Metformin effects

We identified 19 unique upregulated and 31 downregulated plasma proteins (unadjusted p < 0.05) in comparing baseline to 8 weeks after metformin initiation (Fig. 2A). GDF-15 had the largest increase (log2FC = 0.75) and Ep-CAM had the largest decrease (log2FC = –1.13). In CSF, there were 7 upregulated and 19 downregulated proteins (unadjusted p < 0.05) (Fig. 2B). Notably, 7 proteins with a significant metformin effect in plasma also showed a significant metformin effect in CSF (AZU1, CASP-3, CCL11, CCL20, IL32, PRTN3, and REG1A). All metformin effect proteins with unadjusted p-value≤0.05 are listed in Table 1, with the 7 overlapping proteins from plasma and CSF highlighted in a darker shade. The correlation between plasma and CSF metformin effect (i.e., fold-change in protein level at 8 weeks after the start of metformin treatment relative to the baseline level) of the 7 overlapping proteins was high (r = 0.98, p = 9.9e-05, see Fig. 2 C), suggesting that the plasma proteins may be reflective of changes in the central nervous system and can serve as good proxies of metformin effects on the brain. The baseline levels of these plasma and CSF for these 7 proteins had a correlation of 0.65 (Fig. 2D).

Table 1

Metformin effect proteins in plasma and cerebrospinal fluid. Proteins for which a significant (p-unadjusted value < 0.05) difference between pre- and 8 weeks post-metformin treatment in plasma and cerebrospinal fluid are listed by fold change (FC), in order of highest to lowest plasma FC. Cells listing upregulated proteins are shaded pink/red and cells listing downregulated proteins are shaded blue. Proteins with overlapping significant metformin effects in cerebrospinal fluid and plasma are highlighted in a darker shade of red/blue.

Fig. 2

Metformin effects on protein biomarkers in plasma and cerebrospinal fluid. A) Up- and downregulated proteins in plasma after 8 weeks of metformin treatment. B) Up- and downregulated proteins in cerebrospinal fluid after 8 weeks of metformin treatment. C) Correlation between the 7 overlapping up- and downregulated proteins in plasma and cerebrospinal fluid after 8 weeks of metformin treatment. D) Correlation between the 7 overlapping up- and downregulated proteins in plasma and cerebrospinal fluid at baseline.

Longitudinal stability of metformin effects

We examined the extent to which each of the 19 upregulated, and 31 downregulated plasma proteins remained altered after cessation of metformin (i.e., within-subject plasma changes from week 8 to week 16 in Group A). Figure 3A shows a heatmap of log2FC of the 19 upregulated plasma proteins in the 10 Group A participants at 3-time points: 1) Week 0: Baseline as reference (denoted as BL, log2FC = 0); 2) Week 8: metformin use versus baseline (i.e., metformin effect, denoted as METF); and 3) Week 16 : 8 weeks post-cessation of metformin relative to baseline (denoted as Washout). Figure 3B uses line graphs to represent the same results, showing NPX values by individual before, 8 weeks after initiation of metformin, and 8 weeks post-cessation of metformin. As illustrated, some metformin-upregulated proteins, like KLK6, continue to increase in the 8 weeks post-cessation of metformin, some, like MMP-1 remain generally stable, and others like GDF-15 and REG1A are strongly responsive to metformin but washout almost entirely by 8 weeks post-cessation. Figure 3C and 3D show downregulated proteins in a similar presentation as Fig. 3A and 3B. Some proteins, such as Ep-CAM, show a short-term decrease in expression during the metformin treatment period, returning to baseline levels by 8 weeks post-cessation of the drug. Other proteins, such as AZU1, NPM1, and EGF continue to decrease in overall expression after 8 weeks of drug washout.

Fig. 3

Longitudinal stability of metformin effects. A) Heatmap of log2FC of the 19 upregulated plasma proteins of Group A participants as compared to baseline protein levels. B) Line graphs depicting a selection of the upregulated plasma proteins from panel A. C) Heatmap of log2FC of downregulated plasma proteins of Group A participants as compared to baseline protein levels. D) Line graphs depicting a selection of the downregulated plasma proteins from panel C. Each connected line represents a different trial participant.

Biological stability and variability

Next, we investigated the temporal stability of the plasma proteins affected by metformin in Group B. Figure 4A shows a heatmap of log2FC of the 19 upregulated plasma proteins in the 10 Group B participants at 3 time points: 1) Week 0 : 8 weeks on placebo versus baseline (denoted as BL); 2) Week 8: Placebo samples as reference (denoted as PBO); and 3) Week 16 : 8 weeks of metformin use relative to baseline (denoted as METF). Figure 4B shows the same results for two of the selected proteins. For both REG1A and GDF-15, the variability between baseline levels and 8 weeks on placebo was much lower than the increase after 8 weeks of metformin treatment. Similarly, Fig. 4C and 4D show the heatmap and line graphs of the 31 downregulated proteins, respectively. Again, the decrease in plasma levels after 8 weeks of metformin use is much larger than the changes observed with placebo use.

Fig. 4

Biological stability of up- and downregulated proteins. A) Heatmap of log2FC of the 19 upregulated plasma proteins from Group B participants at baseline (BL), after 8 weeks of placebo drug (PBO), and after 8 weeks of metformin (METF). B) Line graphs depicting a selection of the upregulated plasma proteins from panel A. C) Heatmap of log2FC of downregulated plasma proteins of Group B participants as compared to baseline protein levels. D) Line graphs depicting a selection of the downregulated plasma proteins from panel C. Each connected line represents a different trial participant.

DISCUSSION

In this plasma and CSF biomarker analysis of a crossover trial of older individuals with MCI and early AD receiving metformin or placebo for 8 weeks, followed by placebo or metformin for 8 weeks, respectively, with plasma collected at baseline, 8 weeks, and 16 weeks, and CSF collected at baseline and 8 weeks, we sought to identify patterns of circulating protein levels that informed a metformin effect, a washout effect, and biotemporal stability and biological variability. Samples were subjected to O-link proximity extension assay, and results were passed through quality control metrics before further data analysis.

Metformin, a drug prescribed for managing T2DM, or used off-label for weight control, acts on several known pathways to modulate energy homeostasis. Metformin activates AMP-activated protein kinase (AMPK) and adenosine monophosphate deaminase (AMPD), lowers insulin resistance by inhibiting the activation of insulin and insulin-like growth factor 1 receptor (IGF-1 R) pathways, inhibits the electron transport chain and ATP production in the mitochondria, promotes insulin secretion through increasing glucagon-like peptide-1 receptor (GLP-1 R) expression in pancreatic beta cells, and reduces liver lipid synthesis by inhibiting sterol regulatory element-binding protein (SREBP-1) [20]. Given the increased risk of AD in individuals with insulin resistance, it has been thought that metformin effects on insulin resistance may confer indirect protection from AD through the same mechanisms by which it has been FDA-approved for T2DM. Other mechanisms by which metformin may serve as an anti-AD treatment include anti-inflammatory properties [21], anti-oxidative and neuroprotective properties, actions in suppressing brain Aβ burden [22], and enhancement of autophagy [21], the impairment of which has been implicated in the pathogenesis of AD [23].

Of the major proteins upregulated by metformin in our data, GDF-15, or growth differentiation factor-15, has recently been established as a novel biomarker for metformin treatment [24]. That study further observed that adjustment for glucose, HbA1C, insulin, or proinsulin did not attenuate the metformin effect on this protein, suggesting that this protein’s regulation is independent of glycemic influence. GDF-15 is associated with cardiovascular, endocrine, and kidney function. In further support of the protein’s direct relationship with metformin, our data supports near complete washout of GDF-15 to near-baseline levels by 8 weeks post-cessation of metformin. Moreover, within individuals, GDF-15 levels have high temporal stability, remaining relatively unchanged after 8 weeks of placebo treatment. Thus, the demonstrated biotemporal responsivity of GDF-15 further marks this protein as a metformin-sensitive biomarker. Notably, we did not detect any metformin effect on change in CSF GDF-15 (p = 0.98, data not shown), despite robust metformin-induced plasma GDF-15 upregulation, suggesting that this biomarker of metformin function is largely or wholly peripherally produced. Another upregulated protein, REG1A, promotes the regeneration and proliferation of pancreatic beta cells and possesses anti-inflammatory and cell protective properties supportive of maintaining normal glucose metabolism [25]. We found significant increases in both CSF and plasma levels of REG1A after 8 weeks of metformin treatment, which washed out after 8 weeks. Additional studies will be interesting to gain a deeper understanding of the relationship of REG1A to the pathogenesis or biomarkers of AD. Some of our other results also reaffirm those from previous metformin biomarker studies. For instance, a study of 8,401 participants, 2,317 of whom received metformin, identified 26 independent plasma biomarkers of metformin use [24]. While our array of protein assays did not entirely overlap theirs, we did identify two other matching factors: Gal-3 (upregulated in plasma in both studies), and myoglobin (downregulated in plasma in Gerstein et al. [24] and downregulated in CSF in our present study).

Plasma proteins such as KLK6, and several others from Fig. 3A show partial resistance to washout after 8 wks of metformin cessation. These factors support the possibility of lasting effects of metformin effect beyond its metabolism (elimination half-life in plasma of 20 h [26]). This possibility is supported by diabetes-related clinical outcomes: one study found that 6 months of metformin treatment led to sustained improved glucose tolerance for 6 months after stopping treatment [27]. Another study observed a continued partial reduction in the incidence of diabetes two weeks following metformin washout [28].

Those proteins impacted by metformin use in both CSF and plasma: AZU1, CASP-3, CCL11, CCL20, IL32, and PRTN3, have been studied to varying extents in the context of AD. Several of these proteins are immune-related: AZU1 is a protein that belongs to the azurophil granule family and is involved in the immune response. Studies have indicated that AZU1 may play a role in neuroinflammation [29], which is a prominent feature of AD. CCL11 is a chemokine that plays a role in immune cell migration. Elevated levels of CCL11 have been reported in the brains of individuals with AD, and it has been associated with neuroinflammation and cognitive decline [30]. CCL20 is another chemokine involved in immune responses. It is upregulated in AD [31], and it may contribute to neuroinflammatory processes in the brain [32]. IL32 is a pro-inflammatory cytokine that can affect amyloidogenesis in AD [33]. PRTN3 has been identified as a potential protective factor against cognitive decline in a longitudinal cohort [34]. Lastly, CASP-3 is a protein that plays a central role in apoptosis, the process of programmed cell death. In the brains of patients with AD, there is evidence of increased activation of caspase-3 in neuronal post-synaptic densities [35], suggesting its involvement in the neuronal death observed in the disease.

Our study has several limitations. The sample sizes were limited and lacked diversity in race/ethnicity, which could impact the generalizability of the findings to broader populations. We report significantly changed proteins without multiple comparison corrections, and thus future studies with larger sample sizes are required to confirm these findings. A limited set of 420 proteins was assayed using the O-link technology; unbiased proteomics experiments are required to fully characterize changes with metformin use. The intervention period of 8 weeks might not be sufficient to capture all potential changes in protein levels caused by metformin. Longer treatment durations could provide a more comprehensive understanding of the drug’s effects. Finally, further mechanistic are required to understand the underlying mechanisms of how metformin affects the identified proteins and elucidate the pathways involved.

Despite these limitations, this is the first study to investigate the effect of metformin use simultaneously on plasma and CSF proteins within a clinical trial involving non-diabetic participants with AD. The innovative cross-over design enables the exploration of washout effects and biological variability over time. In conclusion, along with GDF-15, we present 7 novel plasma biomarkers of metformin (AZU1, CASP-3, CCL11, CCL20, IL32, PRTN3, and REG1A) with potential relevance to AD pathophysiology that show consistent changes in CSF and remain stable over time. These biomarkers hold promise for utilization in future clinical trials of metformin for AD.

Footnotes

ACKNOWLEDGMENTS

Dr. Arnold served as a principal investigator on the original clinical trial. Members of the Alzheimer’s Clinical and Translational Research Unit (Massachusetts General Hospital) prepared and compiled the results of the O-link biomarker analyses. Drs. Weinberg, Arnold, and Das designed the study and wrote the manuscript. Dr. Das and Ms. He performed data analyses.

FUNDING

This study was supported by the following grants: NIH T32-MH112485, NIH U13AG067696, Alzheimer’s Association AACSF-22-970716, and Harvard Catalyst NIH UL1 TR002541 (Dr Weinberg), Cure Alzheimer’s Foundation (Drs Weinberg and Arnold), and NIH NIA-P30AG062421 to (Drs Weinberg, Arnold, and Das).

CONFLICT OF INTEREST

Sudeshna Das served as a guest editor for the Journal of Alzheimer’s Disease but was not involved in the peer-review process nor had access to any information regarding its peer review.

SEA, MSW, PK, and YH have no conflicts of interest to report.

DATA AVAILABILITY

The data supporting the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

References

(2022) 2022 Alzheimer’s disease facts and figures. Alzheimers Dement 18, 700–789.

Gong

, Dai

, Liu

, Iqbal

(2022) Multi-targets: An unconventional drug development strategy for Alzheimer’s disease. Front Aging Neurosci 14, 837649.

Burillo

, Marqués

, Jiménez

, González-Blanco

, Benito

, Guillén

(2021) Insulin resistance and diabetes mellitus in Alzheimer’s disease. Cells 10, 1236.

Sluggett

, Koponen

, Simon 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, E963–E972.

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.

Munõz-Jiménez

, Zaarkti

, Garciá-Arnés

, Garciá-Casares

(2021) Antidiabetic drugs in Alzheimer’s disease and mild cognitive impairment: A systematic review. Dement Geriatr Cogn Disord 49, 423–434.

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.

Charpignon

, Vakulenko-Lagun

, Zheng

, Magdamo

, Su

, Evans

, Rodriguez

, Sokolov

, Boswell

, Sheu

, Somai

, Middleton

, Hyman

, Betensky

, Finkelstein

, Welsch

, Tzoulaki

, Blacker

, Das

, Albers

(2022) Causal inference in medical records and complementary systems pharmacology for metformin drug repurposing towards dementia. Nat Commun 13, 7652.

El Massry

, Alaeddine

, Ali

, Saad

, Eid

(2020) Metformin:Agrowing journey from glycemic control to the treatment of Alzheimer’s disease and depression. Curr Med Chem 28, 2328–2345.

10.

Oliveira

, Braga

, Lós

, Araújo

SMR

, França

MER

, Duarte-Silva

, Rodrigues

, Rocha

SWS

, Peixoto

(2021) Metformin prevents p-tau and amyloid plaque deposition and memory impairment in diabetic mice. Exp Brain Res 239, 2821–2839.

11.

Baradaran

, Vakilian

, Zare

, Hashemzehi

, Hosseini

, Dinpanah

, Beheshti

(2021) Metformin improved memory impairment caused by chronic ethanol consumption during adolescent to adult period of rats: Role of oxidative stress and neuroinflammation. Behav Brain Res 411, 113399.

12.

Harding

, Chen

H-JC

, Schwiening

, Aggarwal

, Rowley

, Swinden

, Merkle

Metformin may reduce dementia risk through neuroprotection not mitigation of diabetes. BioRxiv. 2023.07.18.549549.

13.

, Xiao

, Li

, Pang

, Xue

, Wan

, Pei

, Yan

(2021) Metformin restores hippocampal neurogenesis and learning and memory via regulating gut microbiota in the obese mouse model. Brain Behav Immun 95, 68–83.

14.

Sanati

, Aminyavari

, Afshari

, Sahebkar

(2022) Mechanistic insight into the role of metformin in Alzheimer’s disease. Life Sci 291, 120299.

15.

Markowicz-Piasecka

, Sikora

, Szydłowska

, Skupień

, Mikiciuk-Olasik

, Huttunen

(2017) Metformin –a future therapy for neurodegenerative diseases: Theme: Drug discovery, development and delivery in Alzheimer’s disease Guest Editor: Davide Brambilla. Pharm Res 34, 2614–2627.

16.

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.

17.

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.

18.

Kulkarni

, Gubbi

, Barzilai

(2020) Benefits of metformin in attenuating the hallmarks of aging. Cell Metab 32, 15–30.

19.

Olink. What is NPX? –Olink. https://olink.com/faq/whatis-npx/. Accessed August 2, 2023.

20.

Demaré

, Kothari

, Calcutt

, Fernyhough

(2021) Metformin as a potential therapeutic for neurological disease: Mobilizing AMPK to repair the nervous system. Expert Rev Neurother 21, 45–63.

21.

Kodali

, Attaluri

, Madhu

, Shuai

, Upadhya

, Gonzalez

, Rao

, Shetty

(2021) Metformin treatment in late middle age improves cognitive function with alleviation of microglial activation and enhancement of autophagy in the hippocampus. Aging Cell 20, e13277.

22.

Chen

, Zhao

, Fan

, Li

, Zhu

, Shen

, Li

, Yan

, Tian

, Liu

, Zhang

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

23.

, Liu

, Sun

(2017) Autophagy and Alzheimer’s disease. Cell Mol Neurobiol 37, 377–388.

24.

Gerstein

, Pare

, Hess

, Ford

, Sjaarda

, Raman

, McQueen

, Lee

, Haenel

, Steinberg

(2017) Growth differentiation factor 15 as a novel biomarker for metformin. Diabetes Care 40, 280–283.

25.

Yuan

, Jeng

, Chen

, Hsieh

, Yang

, Lee

, Hsu

(2005) Opposite roles of human pancreatitis-associated protein and REG1A expression in hepatocellular carcinoma: Association of pancreatitis-associated protein expression with low-stage hepatocellular carcinoma, β-catenin mutation, and favorable prognosis. Clin Cancer Res 11, 2568–2575.

26.

Graham

, Punt

, Arora

, Day

, Doogue

, Duong

, Furlong

, Greenfield

, Greenup

, Kirkpatrick

, Ray

, Timmins

, Williams

(2011) Clinical pharmacokinetics of metformin. Clin Pharmacokinet 50, 81–98.

27.

Lehtovirta

, Forsén

, Gullström

, Häggblom

, Eriksson

, Taskinen

, Groop

(2001) Metabolic effects of metformin in patients with impaired glucose tolerance. Diabet Med 18, 578–583.

28.

Molitch

(2003) Effects of withdrawal from metformin on the development of diabetes in the diabetes prevention program. Diabetes Care 26, 977–980.

29.

Stock

, Kasus-Jacobi

, Pereira

(2018) The role of neutrophil granule proteins in neuroinflammation and Alzheimer’s disease. J Neuroinflammation 15, 240.

30.

Ivanovska

, Abdi

, Murdjeva

, Macedo

, Maes

(2020) Ccl-11 or eotaxin-1: An immune marker for ageing and accelerated ageing in neuro-psychiatric disorders. Pharmaceuticals (Basel) 13, 230.

31.

Nielsen

, Pedersen

, Vestergård

, Maltesen

, Christiansen

, Lundbye-Christensen

, Moos

, Kristensen

, Pedersen

(2020) Novel blood-derived extracellular vesicle-based biomarkers in alzheimer’s disease identified by proximity extension assay. Biomedicines 8, 199.

32.

Liao

, Zhang

, Gu

, Sun

(2020) Targeting CCL20 inhibits subarachnoid hemorrhage-related neuroinflammation in mice. Aging (Albany NY) 12, 14849–14862.

33.

Yun

, Kim

, Hwang

, Jin

, Baek

, Lee

, Hong

, Lee

, Han

, Oh

, Choi

, Yoon

, Hong

(2015) Neuroinflammatory and amyloidogenic activities of IL-32β in Alzheimer’s disease. Mol Neurobiol 52, 341–352.

34.

McCorkindale

, Patrick

, Duce

, Guennewig

, Sutherland

(2022) The key factors predicting dementia in individuals with Alzheimer’s disease-type pathology. Front Aging Neurosci 14, 831967.

35.

Louneva

, Cohen

, Han

, Talbot

, Wilson

, Bennett

, Trojanowski

, Arnold

(2008) Caspase-3 is enriched in postsynaptic densities and increased in Alzheimer’s disease. Am J Pathol 173, 1488–1495.