Metformin and depression:A protocol for systematic review and meta-analysis

Jin Lin,Yue Ji,Jin-Hua Si,Xu-Ran Liu,Zhi-Wang Feng,Hong-Tao Yang*,Li Shen*

1Department of Psychosomatic,First Teaching Hospital of Tianjin University of Traditional Chinese Medicine,National Clinical Research Center for Chinese Medicine Acupuncture and Moxibustion,Tianjin,300381,China.

2Graduate School,Tianjin University of Traditional Chinese Medicine,Tianjin,301617,China.

3Library,Tianjin University of Traditional Chinese Medicine,Tianjin,301617,China.

4Department of Pediatrics,First Teaching Hospital of Tianjin University of Traditional Chinese Medicine,Tianjin,300381,China.

5Department of Nephrology,First Teaching Hospital of Tianjin University of Traditional Chinese Medicine,National Clinical Research Center for Chinese Medicine Acupuncture and Moxibustion,Tianjin,300381,China.

#Jin Lin and Yue Ji contributed equally to this work.

Abstract Introduction: Metformin (MET) has been shown in animal models and diabetic patients with depression to have antidepressant effects. MET is gaining attraction as a treatment for depression;nevertheless, a comprehensive review and evaluation are essential. The primary objective of this review article is to evaluate metformin's effectiveness and safety in the treatment of depression.Methods and analysis:From inception, a wide-ranging survey of the literature will be conducted to categorize possibly suitable studies from online databanks such as the Wan-Fang Database,Chinese Scientific Journal Database (VIP database), Chinese Biomedical Literature Database(CBM), China National Knowledge Infrastructure (CNKI), Traditional Chinese Medicine databases,the Web of Science, EMBASE, PubMed, and the Cochrane Central Register of Controlled Trials(CENTRAL). For this search, no linguistic constraints will be applied. Only randomized controlled trials will be considered,and each study's bias risk will be assessed using the Cochrane Risk of Bias tool. The primary outcome measures will be self-rating scales or clinician-rated scales. Depression remission and quality-of-life indices will be secondary outcomes. The evidence quality for each study will be assessed by grading the review process, development, and recommendations.Ethics and dissemination: The primary data will not be collected in this review; therefore, no ethical consent is required. The results of this methodical review will be available in a peer-reviewed scientific publication and will be made freely accessible.PROSPERO registration number CRD42021286394 Strengths and limitations of this study: This protocol will ensures that the methodologies and processes employed are transparent, allows for peer review, and avoids the risk of bias and duplication. Meta-analyses will not be used as a method of quantitative synthesis. This review study will map the literature on randomized controlled trials of metformin for depression treatment, assess the scope and significance of future systematic reviews on the topic,and identify research gaps and limitations of this study.

Keywords:Metformin, Depression, Antidepressant effects

Introduction

Depressive disorder is the most prevalent type of mental illness.It is a broad term referring to a group of disorders characterized by low mood and inconstant degree of cognitive and behavioral abnormalities. Psychotic symptoms such as hallucinations and delusions may accompany it. Certain patients have engaged in self-injurious and suicidal conduct and have even died as a result [1,2]. Over 300 million people worldwide are believed to suffer from depression,identified by World Health Organization as the single most crucial cause contributing to global disability. Depression is a mental disorder that is most closely related to suicide. Approximately 1 million people commit suicide every year.Around 50%of suicides can be diagnosed as depression. Depression can increase the mortality of other physical diseases.Depressive disorders are most common among patients with cardiovascular disease [3-5]. Although in recent years,researchers have shed light on the underlying mechanism of depression from various angles. Including the association between depression's onset and the incidence of inflammatory reactions in the body [6, 7]; the relationship between depression and monoamine neurotransmitters and their receptors [8, 9].

Recent years, substantial advancements have been made in our understanding of the pathophysiology of depression and specific treatment targets. Antidepressant medications can be broadly classified into the following categories based on their mechanism of action: selective serotonin reuptake inhibitors (SSRIs), serotonin and norepinephrine reuptake inhibitors (SNRIs), tricyclic antidepressants,and monoamine oxidase inhibitors [10]. However, these antidepressants have some limitations in clinical treatment, such as developing drug resistance and triggering metabolic abnormalities[11]

Diabetes mellitus type 2 (T2DM) is a common chronic condition.Numerous studies have established an intrinsic link between T2DM and depression. Patients with T2DM are at higher risk of depression than healthy individuals. Similarly, patients with depression are at a higher risk of developing T2DM [12, 13]. From a pathophysiological point of view, there is a partial overlap in the biological pathways between T2DM and depression. This is mainly manifested in the increased circulatory inflammatory markers in T2DM patients with depressive symptoms compared with T2DM patients alone[14–16].

Metformin (MET) has numerous pharmacological effects. It is used to treat T2DM, considerably improve blood sugar levels, and is safer than first-line T2DM medications [17]. Through various mechanisms,MET has been shown to improve cognitive function in patients (such as lowering blood sugar, repairing damaged hippocampus,anti-inflammation). MET is also used to treat certain neurological diseases such as Alzheimer's, autoimmune disease, autism spectrum disorder, and cerebral ischemia [18-20]. Therefore, compared with other hypoglycemic drugs, MET has the potential to be used in the treatment of depression. According to a recent large cohort study of the Danish population, MET reduces the incidence rate of depression[21]. Many experimental animal studies have shown that MET may exert an antidepressant effect through anti-diabetic,anti-inflammatory, antioxidant, and direct repair of the damaged hippocampal structure [22].

Currently, several randomized controlled studies demonstrated that MET could reduce the risk and incidence rate of depression [21, 23].However, there are discrepancies in various studies' research protocol and efficacy of different clinical trials, resulting in uneven research results and affecting the promotion of this therapy to a certain extent.Subsequently,this research plans to assess the effectiveness and safety of MET in treating depression.

Methods

This systematic review adheres to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA,http://www.prisma-statement.org/) guidelines. Studies that meet review criteria and RCTs with sufficient data were analyzed [24].

Study selection and inclusion criteria

Studies types.Only RCTs (randomized controlled trials) will be selected for this study.In case of insufficient evidence,the results from quasi RCTs will be addressed but will not considered part of the analysis. The funding source will be documented while retrospective data, non-randomized case-control studies, case series, and case reports will not be included in this review.

Participants types.We included studies with individuals of any gender or ethnic origin (age ≥ 16 years) diagnosed clinically, with depression as their primary or comorbid condition.

Diagnosis.Depression was diagnosed based on the Diagnostic and Statistical Manual of Mental Disorders (DSM-III, DSM-IV, or DSM-5;APA 2015) [25]. Or the RDC (Research Diagnostic Criteria) [26], or ICD-10 (International Classification of Diseases and Related Health Problems) [27], Chinese Classification of Mental Disorders, 2nd Edition (CCMD-2) or 3rd Edition (CCMD-3).

Comorbidities.Studies that included patients with common mental or concomitant physical conditions were acceptable as long as depression was the central target of the investigation.

Settings for treatment.All settings, including community, tertiary,secondary, and primary, were considered suitable for inclusion.

Interventions types

Intervention.MET will be included regardless of route of administration, dose, frequency, or duration, and whether used alone or in combination.

Interventions in comparison.We looked at trials that compared metformin to placebo or no treatment and those that compared metformin to conventional medical treatment (typically SSRIs) or any other active intervention. Cointervention trials (e.g., Metformin plus SSRIs vs SSRIs) will be identified and evaluated independently.

Outcome measure types

Primary outcome.Reduced severity of depression, as measured post-intervention primarily as a continuous variable on self-rating scales such as the Beck Depression Inventory (BDI) [28], or clinician-rated scales such as the Hamilton Depression Rating Scale(HAMD) [29].

Secondary outcomes.Depression remission was identified in study reports and based on the HAMD or other clinician-rated scale of depression severity and presented as a binary result. Indices of life quality(SF-36,Short-Form Health Survey) [30],for example, physical and emotional well-being, were recorded for each particular domain of life).

Dropouts from therapy, defined as failure to finish the trial,included the number of participants who left early and the reasons for early dropout.

Methods for identifying studies in searches

Electronic searches.The online databases such as Wan-Fang Database,Chinese Scientific Journal Database(VIP database),Chinese Biomedical Literature Database (CBM), China National Knowledge Infrastructure (CNKI), Traditional Chinese Medicine databases, the Web of Science, EMBASE, PubMed, and the Cochrane Central Register of Controlled Trials (CENTRAL) will be searched regardless of the language and status of the publication.

Looking for additional resources

We will also look for related un-published or current studies on the WHO International Clinical Trials Registry Platform and ClinicalTrials.gov databases. Google Scholar was explored using keywords for grey literature identification [31]. Reference lists for all extracted papers and previous review articles were analyzed, as well as conference proceedings on this subject.

Collection and analysis of data

Studies selection.Two review authors (JL and YJ) will conduct all literature searches independently. The EndNote software will be utilized to import references through electronic database searches and other sources [32]. The duplicate records will be removed from the analysis. The title and abstract will be evaluated when a study will potentially eligible, followed by a full-text evaluation. A 3rd reviewer(JS) will be consulted in case of any disagreements. Using duplicated data in a meta-analysis can overstate treatment efficacy [33]. If there will be any remaining doubts about the validity of several publications based on a single data set, the original reporting author will be contacted for clarification [33]. The process of study selection will be illustrated using a PRISMA flow diagram.

Acquisition and handling of dataTwo review authors (JL and YJ) will independently collect information from selected papers using a standardized form for data extraction.The format will be created specifically for this purpose and will be tested by a responsible referee. Any discrepancies will be resolved through discussions and consensus, and in case, the agreement could not be established. A 3rd reviewer (JS) will be contacted for the final decision.

Characteristics of the study

The selected study's features include title, author, year of publication,design, sample size, blinding, funding source, allocation concealment,randomization, and control, depression diagnostic criteria, number of each group, sex, and age. The statistical analysis included the route of administration, dose, comparator treatment, intervention treatment,and the number of cases.The efficacy and safety-related results will be reported in the study.

Evaluation of the risks of partiality in selected studies

Through the Cochrane Risk of Bias Tool in RevMan (V.5.3), two review writers will autonomously evaluate the bias risk for every included study [33]. Seven methodological quality areas; (i) selective reporting, (ii) incomplete result data, (iii) blinding of outcome assessment, (iv) participants and personnel, (v) concealment of allocation, (vi) random sequence generation, and (vii) other possible causes of bias, of the selected studies will be considered [33].

The reviewers will analyze each of the selected seven potential sources of bias one by one and provided a risk rating of high, low, or unclear to each item. Any disagreements will be resolved through discussions. When necessary, a 3rd reviewer will seek to reach an agreement. If the information provided is insufficient to make a decision, the original study's authors should be contacted for additional information. The assessment's findings will be summarized in a risk of bias summary figure and graph [33].

Measurements of treatment effect

The risk ratio (RR) with 95% confidence intervals (CIs) will be generated for dichotomous results. Constantly, the standardized mean difference (SMD) or mean difference (MD) will be calculated with 95% of confidence intervals (CIs). The MD will be applied if the same scale is used across different studies to measure the results. The SMD will be used if several scales are applied to evaluate similar results. If the included studies provided continuous results using a combination of endpoint and alteration from standard data, in that situation,subgroup analyses will be conducted independently, and effect size values for distinct subgroups will be retrieved [34]. A descriptive summary of the results will be provided if less than two studies are included for a single result measure [34].

This study will assess data only from the first part of a crossover trial [35]. The authors conducted numerous pair-wise comparisons among pairings of interest to the review question for multiple treatment groups studies[36]to avoid including the same study group more than once.

Managing incomplete data

In case the results of interest data are insufficient or missing. When needed,the authors communicated with the original study's authors to obtain further information. The missing data from results will not be added in the primary analysis;instead,the impact of missing data will be evaluated through sensitivity analysis. When the standard deviation for a result will not provide, it will be computed using the provided data or additional data obtained from the authors.

Heterogeneity assessment

Visual inspection of the forest plot and the χ2test will be applied to calculate statistical heterogeneity. Additionally, to determine heterogeneity, the I2statistical value will be produced. Heterogeneity will be classified using the following criteria.I2of 75%–100% as significant heterogeneity.I2of 50%–90% as may represent considerable heterogeneity.I2of 30%–60% as may represent moderate heterogeneity.I2of 0%–40% as might not be significant.

When the p-value from an χ2test is ＜0.10 or the I2value is greater than 50%, this indicates discernible between-study variation[36, 37],the subgroup analysis will be conducted using predefined parameters to determine a reasonable explanation for statistical heterogeneity.

Reporting bias assessment

If a meta-analysis contains more than ten papers,the pattern of results will be addressed by inspecting a funnel plot for evidence of asymmetries [35]. After identifying the asymmetry, possible explanations will be attempted.

Analysis of data

A composite estimate of the intervention impact will be derived by merging the data of many studies in a meta-analysis, assuming that the precise conclusion of the same involvement will be assessed using similar approaches in a homogeneous population through studies.The Mantel-Haenszel method will be used to estimate the overall RR for dichotomous data [35]. The MD or SMD will be computed for continuous variables in different cases.The fixed-effects model will be used when statistical heterogeneity is minimal. Otherwise, the difference will be estimated using the random-effects model, which is more cautious. A descriptive summary of individual study results will be provided when a meta-analysis was impossible.

Analysis of the subgroup and heterogeneity evaluation

Different pairs of comparisons will be subjected to subgroup analysis.Forest plots will be used to display the results of subgroup scrutiny. If one or two trials accounted for more than 80% of participants in a meta-analysis,the fixed-effects technique will be adopted to produce a more reliable estimation [38]. The studies will be analyzed in one subgroup as well [38]. Meta-regression analyses will be performed to identify heterogeneity's statistically significant causes (P＜0.10).

Analysis of sensitivity

A sensitivity study will be performed to verify if our findings are reliable,taking into account the impact of the below factors.To switch between the random-effects and fixed-effects models, the model will be chosen for data pulling. Included studies' methodological quality:research having a higher or uncertain risk of bias will be excepted.The use of a crossover design in studies will be prohibited. Studies with more than 50%of their patients lost to follow-up will be omitted.

Table of findings summary

The two review authors will be autonomously calculated the quality of evidence for each result through the GRADE approach (Grading of Recommendations Assessment, Development, and Evaluation). The evidence quality might be reduced by five factors: publication bias,indirectness, inconsistency, imprecision, and research constraints. The evidence's quality will be classified into four groups, ranging from very low to high[39].Through the GRADEpro guideline development tool,a descriptive summary of the outcomes table will be constructed,containing all of the resulting trials [40].

Patients and public participation

There will be no involvement of the patients in the research question development or outcome measures, nor in the design or administration of this review. Therefore, the study's findings will not be shared with the participants or any other relevant patient population.

Corrections

In the event of protocol changes, we will provide the rationale,description, and date for any changes.

Discussion

People pay more and more attention to the treatment of depression as society evolves and people's consciousness improves. Even though there are numerous antidepressant medicines on the market, there are still several drawbacks. MET has an antidepressant impact, according to several clinical and animal investigations.A recent meta-analysis of metformin's cognitive and antidepressant effects suggests that MET may be used to treat cognitive deficits in specific clinical settings[41].Another meta-analysis of the relationship between diabetes and depression found that pioglitazone was associated with an improvement in depressive symptoms, particularly in women, the effects on blood sugar and insulin resistance are harder to explain.There is no consistent benefit of metformin to depressive symptoms[42]. However, due to the lack of a large crowd cohort study, the authors believe that more clinical physicians will focus on the benefits of metformin in treating depression soon. The increase in research must result from an integration analysis to obtain the relevant evidence to help more clinicians focus on this. MET's efficacy and safety for depression patients will be evaluated in this systematic review. The findings will be crucial to developing reliable clinical guidelines for treating depressive disorders. A peer-reviewed scientific journal will publish our systematic review findings, and the data sets will be made publically available.