Effectiveness of metformin as an adjunctive therapy in rheumatoid arthritis: a systematic review and meta-analysis of randomized controlled trials Authors: Manar M. Alshammari, Noor A. Abood, Sundus Kahwaji, Fatemeh Sharif-Askari, Hala Alobaidi, Zelal Kharaba, Feras Jirjees Published: 2026 ABSTRACT BackgroundRheumatoid arthritis (RA) is a chronic inflammatory disease that impacts patients’ quality of life. Evidence regarding the effectiveness of metformin use as an adjunct therapy in RA is limited and fragmented. This systematic review and meta-analysis aimed to evaluate the effectiveness of metformin as an adjunct therapy in RA and to quantitatively synthesize available evidence.MethodsThis systematic review and meta-analysis were conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, focusing on randomised controlled trials (RCTs) that assessed the effect of metformin as an adjunct therapy for patients with RA receiving conventional synthetic disease-modifying antirheumatic drugs (csDMARDs). C-reactive protein (CRP), Disease Activity Score in 28 joints (DAS-2)8, and Health Assessment Questionnaire Disability Index (HAQ-DI) were included as an outcome. A search was conducted across four databases up to 31 December 2025, including PubMed, Embase, Cochrane Central, and Scopus. The methodological quality of the included RCTs was evaluated using risk of bias 2 (RoB 2) tool. Statistical analysis was conducted with StataNow version 19.5. For each outcome, pooled mean differences (MDs) with 95% confidence intervals (CIs) were estimated using random-effects models.ResultsA total of four RCTs with 350 participants were included in this review. The use of metformin was associated with significant decreases in several indicators after three months, including a reduction in the CRP level (MD = -5.55, 95% CI = -7.46 to -4.52, p-value = 0.00), DAS-28 (MD = -0.8, %CI = -0.87 to 0.73, p-value < 0.001), and HAQ-DI (MD =-0.12, %CI = -0.19 to -0.05, p-value < 0.001). Additionally, a significant reduction in CRP after six months (MD = -6.96, %95 CI = -7.79 to -6.13, p-value < 0.001) compared to the placebo group.ConclusionThis systematic review provides moderate-to-high certainty evidence supporting the use of metformin as an adjunctive therapy in RA. However, larger, longer-duration randomized trials are required to confirm these findings and improve generalizability.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s41927-026-00659-3. Introduction Rheumatoid arthritis (RA) is defined as an autoimmune disease mediated by the activity of immune cells. These cells lose the ability to distinguish between synovial and foreign tissue, resulting in the production of cytokines within synovial tissue, which drives inflammation, the hallmark characteristic of the disease [1]. The most common indicator of RA is the symmetrical polyarthritis that mainly affects the small joints [2]. The disease can progress, leading to extraarticular manifestations in the skin, eyes, heart, lungs, kidneys, nervous system, and gastrointestinal system [3]. In advanced cases, rheumatoid nodules can develop, which are benign granulomatous foci with central necrosis that are the most common extraarticular feature in advanced cases of RA [4]. Additionally, RA is associated with elevated cardiovascular risk that is not fully explained by traditional risk factors; therefore, suppressing inflammation by using RA therapies may also reduce the risk of heart disease [5]. According to the World Health Organization (WHO), approximately 18 million people worldwide were living with RA in 2019 [6], with a global prevalence exceeding 0.5% [7]. In addition, the estimated global incidence reached nearly half a million new cases in 2021 [8]. RA affects individuals of all ages and both sexes; however, its incidence increases after the age of 40, and the condition is significantly more common in women, who are affected approximately two to three times more frequently than men [7, 9]. It significantly affects the quality of life (QoL) of patients, as patients with RA suffer from persistent pain that interferes with their daily lives [10]. In fact, the effect on daily life is so profound that it aligns closely with the WHO’s International Classification of Functioning, Disability and Health (ICF), which categorizes these consequences as impairments in body structure and functions, limitations in daily activities, and restrictions in social participation [11]. Despite advances in treatment and management improvement during the past decades, the QoL of many patients does not improve with current therapies. This highlights the need for adjunct low-cost strategies to control RA disease activity and improve patients’ QoL, without serious side effects [12]. Metformin is a Biguanide derivative derived from Galegine and is widely used to treat type 2 diabetes mellitus. It regulates metabolism by acting on multiple intracellular signaling pathways, including adenosine monophosphate (AMP)-activated protein kinase (AMPK) activation [13]. The AMPK and its downstream intracellular signaling control the activation and differentiation of immune cells and anti-inflammatory responses. These properties suggest that metformin may have immunomodulatory and anti-inflammatory effects [14]. Several studies have explored the potential role of metformin in treating RA. Evidence includes a population-based cohort study showing an association between metformin use and a reduced incidence of RA [15]. Additionally, a two-sample Mendelian randomization study provided genetic evidence supporting the potential effect of metformin against RA [16]. Understanding the efficacy of metformin in RA involves evaluating clinical outcomes, including C-reactive protein (CRP), the Disease Activity Score in 28 joints (DAS-28), and the Health Assessment Questionnaire–Disability Index (HAQ-DI), which respectively reflect biological inflammation, composite disease activity, and patient-reported functional disability. Although several clinical trials have been conducted, to the best of our knowledge, no systematic review or meta-analysis has yet comprehensively synthesized the available evidence to assess the effectiveness of metformin as an adjunct therapy in the treatment of RA. Therefore, this systematic review aimed to address this gap by systematically evaluating and integrating the existing research. Methodology This systematic review and meta-analysis were conducted according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) [17], and the recommendations of the Cochrane Handbook [18]. The study protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) with registration number CRD420251271612. Search strategy A systematic search was conducted across four databases, including PubMed, Embase, Cochrane Central, and Scopus. Central Medical Subject Headings (MeSH) terms were used for keywords such as “rheumatoid arthritis”, RA, metformin, biguanide, “randomized clinical trial”, and RCT. There were no restrictions on language or country. No time limits were applied during the database search; all records from database inception to 31 December 2025 were considered. The detailed search strategy is provided in the supplementary (Table 1). Table 1Characteristics of the included randomized clinical trials (n = 4)Author / yearStudy DesignCountryTotal ParticipantsTrial ArmMetformin DoseInclusion CriteriaOutcome MeasuresMain ConclusionInterventionControlGharib et al. 2021 [24]Prospective, randomized, single-blinded controlled studyEgypt60Metformin + stable csDMARDsPlacebo + stable csDMARDs850 mg twice dailyAdults > 18 years with RA diagnosed by ACR/EULAR 2010, moderate to high disease activity (DAS-28-CRP > 3.2), receiving stable csDMARDs ≥ 3 monthsCRP, DAS-28-CRP & HAQ-DI.Metformin significantly improved inflammation, disease activity, and QoL with a favorable safety profile.Abdallah et al. 2021 [23]Randomized, double-blind, placebo-controlled trialEgypt120Metformin + methotrexate (7.5 mg/week)Placebo + methotrexate(7.5 mg/week)XR 1000 mg once dailyAdults > 18 years with active RA fulfilling ACR/EULAR 2010 criteria, ≥ 4 tender and swollen joints, elevated ESR (> 28 mm/h) or CRP (> 7 mg/L)CRP, DAS-28-CRP & HAQ-DIMetformin enhanced methotrexate efficacy and significantly improved disease activity and inflammatory markers without serious adverse effects.Kumar et al. 2023 [25]Prospective, randomized, single-blinded controlled studyIndia70Metformin + csDMARDsPlacebo + csDMARDs500–1500 mg daily based on weight in divided dosesAdults > 18 years with RA based on elevated CRP, receiving stable csDMARDs ≥ 3 monthsSerum CRP, QoL and tolerabilityMetformin significantly reduced CRP after 6 months and improved QoL, supporting its role as an adjuvant therapy.Kumar et al. 2024 [22]Prospective, randomized, single-blinded controlled studyIndia100Metformin + csDMARDsPlacebo + csDMARDs850 mg twice dailyAdults > 18 years with RA diagnosed by ACR/EULAR 2010, DAS-28-CRP > 3.2, receiving stable csDMARDs ≥ 3 monthsCRP, DAS-28-CRP, HAQ-DIMetformin reduced disease activity and inflammation and improved QoL with minimal adverse effects.* Abbreviations: ACR/EULAR: American College of Rheumatology/European Alliance of Associations for Rheumatology, CRP: C-Reactive Protein; csDMARDS, Conventional synthetic Disease-Modifying Antirheumatic Drugs, DAS-28-CRP: Disease Activity Score in 28 joints using C-Reactive Protein, ESR: Erythrocyte Sedimentation Rate, HAQ-DI: Health Assessment Questionnaire–Disability Index, QOL: Quality of Life, RA: Rheumatoid Arthritis, XR: Extended Release Table 1Characteristics of the included randomized clinical trials (n = 4)Author / yearStudy DesignCountryTotal ParticipantsTrial ArmMetformin DoseInclusion CriteriaOutcome MeasuresMain ConclusionInterventionControlGharib et al. 2021 [24]Prospective, randomized, single-blinded controlled studyEgypt60Metformin + stable csDMARDsPlacebo + stable csDMARDs850 mg twice dailyAdults > 18 years with RA diagnosed by ACR/EULAR 2010, moderate to high disease activity (DAS-28-CRP > 3.2), receiving stable csDMARDs ≥ 3 monthsCRP, DAS-28-CRP & HAQ-DI.Metformin significantly improved inflammation, disease activity, and QoL with a favorable safety profile.Abdallah et al. 2021 [23]Randomized, double-blind, placebo-controlled trialEgypt120Metformin + methotrexate (7.5 mg/week)Placebo + methotrexate(7.5 mg/week)XR 1000 mg once dailyAdults > 18 years with active RA fulfilling ACR/EULAR 2010 criteria, ≥ 4 tender and swollen joints, elevated ESR (> 28 mm/h) or CRP (> 7 mg/L)CRP, DAS-28-CRP & HAQ-DIMetformin enhanced methotrexate efficacy and significantly improved disease activity and inflammatory markers without serious adverse effects.Kumar et al. 2023 [25]Prospective, randomized, single-blinded controlled studyIndia70Metformin + csDMARDsPlacebo + csDMARDs500–1500 mg daily based on weight in divided dosesAdults > 18 years with RA based on elevated CRP, receiving stable csDMARDs ≥ 3 monthsSerum CRP, QoL and tolerabilityMetformin significantly reduced CRP after 6 months and improved QoL, supporting its role as an adjuvant therapy.Kumar et al. 2024 [22]Prospective, randomized, single-blinded controlled studyIndia100Metformin + csDMARDsPlacebo + csDMARDs850 mg twice dailyAdults > 18 years with RA diagnosed by ACR/EULAR 2010, DAS-28-CRP > 3.2, receiving stable csDMARDs ≥ 3 monthsCRP, DAS-28-CRP, HAQ-DIMetformin reduced disease activity and inflammation and improved QoL with minimal adverse effects.* Abbreviations: ACR/EULAR: American College of Rheumatology/European Alliance of Associations for Rheumatology, CRP: C-Reactive Protein; csDMARDS, Conventional synthetic Disease-Modifying Antirheumatic Drugs, DAS-28-CRP: Disease Activity Score in 28 joints using C-Reactive Protein, ESR: Erythrocyte Sedimentation Rate, HAQ-DI: Health Assessment Questionnaire–Disability Index, QOL: Quality of Life, RA: Rheumatoid Arthritis, XR: Extended Release Characteristics of the included randomized clinical trials (n = 4) Placebo + methotrexate (7.5 mg/week) * Abbreviations: ACR/EULAR: American College of Rheumatology/European Alliance of Associations for Rheumatology, CRP: C-Reactive Protein; csDMARDS, Conventional synthetic Disease-Modifying Antirheumatic Drugs, DAS-28-CRP: Disease Activity Score in 28 joints using C-Reactive Protein, ESR: Erythrocyte Sedimentation Rate, HAQ-DI: Health Assessment Questionnaire–Disability Index, QOL: Quality of Life, RA: Rheumatoid Arthritis, XR: Extended Release Eligibility criteria The eligibility criteria were developed according to the Population, Intervention, Comparison, Outcomes, and Study design (PICOS) framework [19]. Thus, the inclusion criteria were as follows: patients over 18 years old with RA receiving conventional synthetic disease-modifying antirheumatic drugs (csDMARDs) with no restriction on the duration of the condition or gender. The main intervention of the included studies was using metformin as an adjunctive to csDMARDs. The intervention groups were compared to the placebo group. Finally, with respect to study design, only randomized controlled trials (RCTs) published in peer-reviewed journals were included to maximize methodological rigor and reduce bias in the evidence synthesis. Exclusion criteria included patients receiving metformin for indications other than the study condition, pregnant patients, and individuals with a prior diagnosis of heart failure or diabetes mellitus. Patients treated with biologic or targeted synthetic DMARDs were also excluded to ensure a more homogeneous study population and to better isolate the adjunctive effect of metformin, as these agents exert potent and independent immunomodulatory effects that could confound disease activity and inflammatory outcomes. Additionally, this restriction aligns with the study objective of evaluating metformin as a potentially cost-effective adjunct within the standard csDMARD treatment setting. Study selection The search results from each database were exported to an Excel spreadsheet (Microsoft Corp., Redmond, WA), where duplicates were removed. The remaining records were screened independently by two authors in a two-step process: first by title and abstract, and then by full-text review. Exclusions were made based on the predefined eligibility criteria. In addition, a manual search was conducted by reviewing the reference lists of the included studies. Any discrepancies between the two authors were resolved by a third independent author. Data extraction Data extraction was conducted using Microsoft Excel to collect relevant information from the included studies by the first author and subsequently verified by all authors. Extracted data included: (1) study characteristics, (2) participant characteristics for each study, and (3) outcome data. For outcome data, a separate Excel file was created containing all numerical data for each outcome, organized into individual worksheets by outcome, which was subsequently used for the meta-analysis. Quality and evidence assessment The methodological quality of the included studies was evaluated using Cochrane Risk of Bias 2 (RoB 2) tool for RCTs [20]. The evaluation was based on five domains, assessing bias related to randomization, intervention adherence, missing data, outcome measurement, and selective reporting. Each RCT was classified as having a high risk of bias, some concerns, or a low risk of bias based on an overall judgment after evaluating all five domains by two independent authors, and any disagreement in the overall judgment was resolved by a third independent author. The quality of evidence for each outcome was assessed using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach. In the GRADE assessment, initially, all included studies were classified as having high quality, and the quality could be downgraded based on predefined criteria. Downgrading of the initial level of evidence was based on five domains, including risk of bias, inconsistency, indirectness, imprecision, and publication bias. The first domain was judged based on the RoB 2 results. The second domain was determined according to heterogeneity, and if there was a clear explanation for the heterogeneity, such as differences in dose or study population, the evidence was not downgraded. For the third domain, the quality was downgraded if the evidence was not directly applicable. The imprecision domain led to downgrading when the results for each outcome were uncertain or unstable. The publication bias was assessed as the final domain. Based on the overall assessment, the certainty of evidence for each outcome was classified as high, moderate, low, or very low. Any discrepancies in the GRADE assessments were resolved by a third independent author. Statistical and sensitivity analysis The meta-analysis was performed using StataNow version 19.5 (StataCorp LLC, College Station, TX, USA). The data extracted for continuous outcomes included sample size, mean, and standard deviation (SD) to calculate the mean difference (MD) with 95% confidence intervals (CIs). Pooled data were analyzed using a random-effects model to account for between-study variability. Heterogeneity was assessed using the inconsistency index (I²) and tau-squared (τ²). For I², values above 50% indicate substantial heterogeneity, whereas values greater than 90% indicate considerable heterogeneity among studies [21]. Given the small number of included studies (n = 4), formal statistical tests for publication bias, such as Egger’s test and funnel plot, were not conducted to avoid misleading or unreliable results. Additionally, formal inter-reviewer agreement statistics were not calculated, as all disagreements between the two authors were resolved through consensus discussion, with involvement of a third independent author when required. A leave-one-out sensitivity analysis was performed to evaluate the robustness of the pooled estimates for outcomes exhibiting substantial heterogeneity by sequentially omitting each study to assess its influence on the overall effect size and heterogeneity. Results Search results A total of 67 articles were retrieved from four databases, including PubMed, Cochrane Central, Embase, and Scopus, and only 55 articles remained after duplicate removal. These articles were screened by title and abstract against predefined PICOS criteria, resulting in four articles that met the eligibility criteria. Following full-text screening, four articles [22–25] were included in the review, as illustrated in the PRISMA flow diagram (Fig. 1). Fig. 1PRISMA flowchart of the randomized clinical trials PRISMA flowchart of the randomized clinical trials Studies characteristics The included four RCTs collectively enrolled 350 participants with active RA conducted in two countries, Egypt and India, with two RCTs in each country. Patient populations varied in size, with sample sizes ranging from 60 to 120 participants per study. The intervention arms consistently involved the administration of metformin as an adjunct to stable conventional synthetic disease-modifying antirheumatic drug (csDMARD) regimens, either methotrexate monotherapy or combination therapy. Metformin dosing varied across trials, ranging from 500 to 1500 mg daily, administered once daily or in divided doses. Treatment durations and follow-up periods spanned 12 weeks to 6 months. Participants were adults aged 18 years and older, and primary outcomes were assessed using standardized measures of disease activity (DAS-28-CRP), systemic inflammation (CRP), and patient-reported quality of life (HAQ-DI) (Table 1). Participants characteristics The average age of participants ranged from 44 to 51 years, with gender distribution showing a higher proportion of female participants in most groups. Body mass index (BMI) was reported in three studies, with mean values ranging from 24.18 to 32.7. Across three of the included studies, disease duration overall ranged from 6.8 to 13 years (Table 2). Table 2Characteristics of the included patients in the randomized clinical trials (n = 4)StudyGroupNumber of participants in each groupAge (years) (mean ± SD)Gender (male / female)BMI (mean ± SD)diseaseduration(years)Clinical outcomes at baselineCRP (mg/L)DAS-28-CRPHAQ-DIGharib et al. 2021 [24]Metformin Group3051.3 ± 9.92/2832.7 ± 5.113 (6–20)*16(3–30)*5.47(4.32–6.87)*2.13(0.63–2.5)*Control Group3050.9 ± 7.00/3031.4 ± 3.112 (4–17)*12(3–24)*5.65(3.68–6.85)*2(1.25–2.63)*Abdallah et al. 2021 [23]Metformingroup6044.3 ± 10.423/3724.18 ± 1.576.9 ± 3.127.6 ± 8.75.1 ± 1.51.2δControl Group6046.1 ± 1025/3524.74 ± 1.677.5 ± 3.125.3 ± 7.84.9 ± 1.41.3δKumar et al. 2023 [25]Metformin Group3549.7 ± 8.63/3232.50 ± 5.86Not reported15.37 ± 4.62Not reportedNot reportedControl Group3548.9 ± 7.20/3531.69 ± 3.82Not reported11.43 ± 1.91Not reportedNot reportedKumar et al. 2024 [22]Metformin Group5046.4 ± 4.420/30Not reported6.7 ± 1.325.56 ± 4.275.39 ± 1.321.94 ± 0.55Control Group5045.2 ± 5.222/28Not reported7.2 ± 1.426.32 ± 4.645.65 ± 1.211.98 ± 0.23Abbreviations: BMI: Body mass index, CRP: C-Reactive Protein; DAS-28-CRP: Disease Activity Score in 28 joints using C-Reactive Protein; HAQ-DI: Health Assessment Questionnaire-Disability Index; SD: standard deviation*Median (range)δThe Mean was reported without SD Table 2Characteristics of the included patients in the randomized clinical trials (n = 4)StudyGroupNumber of participants in each groupAge (years) (mean ± SD)Gender (male / female)BMI (mean ± SD)diseaseduration(years)Clinical outcomes at baselineCRP (mg/L)DAS-28-CRPHAQ-DIGharib et al. 2021 [24]Metformin Group3051.3 ± 9.92/2832.7 ± 5.113 (6–20)*16(3–30)*5.47(4.32–6.87)*2.13(0.63–2.5)*Control Group3050.9 ± 7.00/3031.4 ± 3.112 (4–17)*12(3–24)*5.65(3.68–6.85)*2(1.25–2.63)*Abdallah et al. 2021 [23]Metformingroup6044.3 ± 10.423/3724.18 ± 1.576.9 ± 3.127.6 ± 8.75.1 ± 1.51.2δControl Group6046.1 ± 1025/3524.74 ± 1.677.5 ± 3.125.3 ± 7.84.9 ± 1.41.3δKumar et al. 2023 [25]Metformin Group3549.7 ± 8.63/3232.50 ± 5.86Not reported15.37 ± 4.62Not reportedNot reportedControl Group3548.9 ± 7.20/3531.69 ± 3.82Not reported11.43 ± 1.91Not reportedNot reportedKumar et al. 2024 [22]Metformin Group5046.4 ± 4.420/30Not reported6.7 ± 1.325.56 ± 4.275.39 ± 1.321.94 ± 0.55Control Group5045.2 ± 5.222/28Not reported7.2 ± 1.426.32 ± 4.645.65 ± 1.211.98 ± 0.23Abbreviations: BMI: Body mass index, CRP: C-Reactive Protein; DAS-28-CRP: Disease Activity Score in 28 joints using C-Reactive Protein; HAQ-DI: Health Assessment Questionnaire-Disability Index; SD: standard deviation*Median (range)δThe Mean was reported without SD Characteristics of the included patients in the randomized clinical trials (n = 4) 16 (3–30)* 5.47 (4.32–6.87)* 2.13 (0.63–2.5)* 12 (3–24)* 5.65 (3.68–6.85)* 2 (1.25–2.63)* Metformin group Abbreviations: BMI: Body mass index, CRP: C-Reactive Protein; DAS-28-CRP: Disease Activity Score in 28 joints using C-Reactive Protein; HAQ-DI: Health Assessment Questionnaire-Disability Index; SD: standard deviation *Median (range) δThe Mean was reported without SD Quality and evidence assessment The methodological quality of the included RCTs was assessed using the RoB 2 tool. Two RCTs Gharib et al. 2021 [24] and Abdalla et al. 2021 [23] were judged to have an overall low risk of bias, while the other two RCTs Kumar et al. 2023 [25] and Kumar et al. 2024 [22] were judged to have an overall high risk of bias. The high risk of bias was primarily due to concerns related to deviations from intended interventions, as both studies were single-blinded, and no clinical trial registration numbers were reported. Additionally, Kumar et al. 2023 raised concerns regarding missing outcome data, as the study did not report how the withdrawal of three participants was handled in the analysis [25]. Additionally, in Kumar et al. 2023 [25], the methods of random sequence generation and allocation concealment were not described, resulting in concerns regarding the randomization process (Fig. 2). The GRADE assessment indicated that the quality of each outcome ranged from moderate to high, as shown in the Summary of Findings table (Table 3). For two outcomes, the certainty of evidence was downgraded due to the high risk of bias associated with the trials by Kumar et al. 2023 [25] and Kumar et al. 2024 [22]. For the remaining outcomes, the presence of high risk of bias did not affect the overall certainty of evidence as only one included trial was at high risk. Fig. 2Risk of bias 2 tool for RCTs Risk of bias 2 tool for RCTs Table 3Summary of findings from the randomized clinical trials (n = 4)OutcomeStudy designTotal number of patientsMale/femaleMean difference (95% CI)Overallcertainty ofevidenceMetforminControlThe CRP levels (mg/L) after three months4 RCTs48/12747/128-5.99 [-7.46, -4.52]⨁⨁⨁◯Moderatea, bThe CRP levels (mg/L) after six months3 RCTs25/9022/93-6.96 [-7.79, -6.13]⨁⨁⨁◯Moderatea, bThe change in DAS-28 after three months3 RCTs45/9547/93-0.80 [-0.87, -0.73]⨁⨁⨁⨁HighThe change in HAQ-DI after three months3 RCTs45/9547/93-0.12 [-0.19, -0.05]⨁⨁⨁⨁High*Abbreviations: CRP: C-reactive protein, RCT: Randomized control trials, DAS-28: Disease Activity Score in 28 joints, HAQ-DI: Health Assessment Questionnaire-Disability IndexaThe certainty of evidence was downgraded in the risk of bias domain because two included studies had a high risk of biasbAlthough some visual variability was observed in the forest plot (I² = 53.22%), leave-one-out sensitivity analyses demonstrated that the pooled effect remained stable and statistically significant after sequential omission of each study, indicating that no single study dominated the overall effect and that the inconsistency was not seriousHigh certainty: We have high confidence that the true effect is close to the estimated effectModerate certainty: We have moderate confidence in the estimated effect; the true effect is likely to be close to the estimate, although it may differ substantially Table 3Summary of findings from the randomized clinical trials (n = 4)OutcomeStudy designTotal number of patientsMale/femaleMean difference (95% CI)Overallcertainty ofevidenceMetforminControlThe CRP levels (mg/L) after three months4 RCTs48/12747/128-5.99 [-7.46, -4.52]⨁⨁⨁◯Moderatea, bThe CRP levels (mg/L) after six months3 RCTs25/9022/93-6.96 [-7.79, -6.13]⨁⨁⨁◯Moderatea, bThe change in DAS-28 after three months3 RCTs45/9547/93-0.80 [-0.87, -0.73]⨁⨁⨁⨁HighThe change in HAQ-DI after three months3 RCTs45/9547/93-0.12 [-0.19, -0.05]⨁⨁⨁⨁High*Abbreviations: CRP: C-reactive protein, RCT: Randomized control trials, DAS-28: Disease Activity Score in 28 joints, HAQ-DI: Health Assessment Questionnaire-Disability IndexaThe certainty of evidence was downgraded in the risk of bias domain because two included studies had a high risk of biasbAlthough some visual variability was observed in the forest plot (I² = 53.22%), leave-one-out sensitivity analyses demonstrated that the pooled effect remained stable and statistically significant after sequential omission of each study, indicating that no single study dominated the overall effect and that the inconsistency was not seriousHigh certainty: We have high confidence that the true effect is close to the estimated effectModerate certainty: We have moderate confidence in the estimated effect; the true effect is likely to be close to the estimate, although it may differ substantially Summary of findings from the randomized clinical trials (n = 4) ⨁⨁⨁◯ Moderatea, b ⨁⨁⨁◯ Moderatea, b ⨁⨁⨁⨁ High ⨁⨁⨁⨁ High *Abbreviations: CRP: C-reactive protein, RCT: Randomized control trials, DAS-28: Disease Activity Score in 28 joints, HAQ-DI: Health Assessment Questionnaire-Disability Index aThe certainty of evidence was downgraded in the risk of bias domain because two included studies had a high risk of bias bAlthough some visual variability was observed in the forest plot (I² = 53.22%), leave-one-out sensitivity analyses demonstrated that the pooled effect remained stable and statistically significant after sequential omission of each study, indicating that no single study dominated the overall effect and that the inconsistency was not serious High certainty: We have high confidence that the true effect is close to the estimated effect Moderate certainty: We have moderate confidence in the estimated effect; the true effect is likely to be close to the estimate, although it may differ substantially Meta-analysis C-reactive protein (CRP) levels after three months The CRP outcome was reported in all included studies [22–25]. The meta-analysis demonstrated a significant reduction in CRP levels after three months, with an overall mean difference of − 5.99 (95% CI: -7.46 to -4.52) and a p-value < 0.001, as shown in Fig. 3. However, substantial heterogeneity was observed (I² = 53.22%), which was explored using a leave-one-out sensitivity analysis. Fig. 3The change in CRP levels after three months The change in CRP levels after three months CRP levels after six months Three studies reported CRP levels after six months [22, 24, 25]. The meta-analysis demonstrated a significant reduction in CRP, with a mean difference of -6.96 (95% CI: -7.79 to -6.13) and a p-value < 0.001, with no observed heterogeneity (I² = 0%) as shown in Fig. 4. Fig. 4The CRP levels after six months The CRP levels after six months The change in DAS-28 after three months Three studies reported the change in DAS-28 after three months [22–24]. The meta-analysis showed a significant reduction with a mean difference of -0.8 (95% CI: -0.87 to -0.73) and p-value < 0.001. The heterogeneity across the studies was zero (I² = 0%) (Fig. 5). Fig. 5The change in DAS-28 after three months The change in DAS-28 after three months The change in HAQ-DI after three months The quality of life (QoL) for patients is assessed by using the Health Assessment Questionnaire Disability Index (HAQ-DI) in three studies [22–24]. The meta-analysis showed a significant improvement in QoL in the intervention group compared with the control group, with a mean difference of -0.12 (95% CI: -0.19 to -0.05; p < 0.001), and no heterogeneity was observed among the studies (I² = 0%). (Fig. 6) Fig. 6The change in HAQ-DI after three months The change in HAQ-DI after three months Sensitivity analysis A leave-one-out sensitivity analysis was conducted for the three-month CRP outcome. After sequentially omitting each study, the pooled effect estimate for CRP reduction remained statistically significant, indicating that no single study disproportionately influenced the overall result (Fig. 7). Fig. 7Leave one out (sensitivity analysis) CRP after three months effect Leave one out (sensitivity analysis) CRP after three months effect Discussion This systematic review and meta-analysis integrated the existing evidence to evaluate the effectiveness of metformin as an adjunctive therapy in the management of RA. The findings suggest that adjunctive metformin therapy is associated with statistically significant reductions in disease activity and systemic inflammation in patients with RA receiving csDMARDs. Metformin was associated with clinically meaningful improvements in DAS-28-CRP, reductions in C-reactive protein (CRP), and enhanced health-related quality of life (QoL) as measured by the Health Assessment Questionnaire–Disability Index (HAQ-DI). This meta-analysis consolidates findings from four independent RCTs collectively suggesting the efficacy of metformin as an adjunct to csDMARDs in patients with RA [22–25]. Evidence from the trial by Gharib et al. 2021 indicates that metformin (850 mg twice daily) may contribute to reductions in serum CRP levels and DAS-28-CRP at six months compared with placebo, alongside improvements in health-related QoL as measured by HAQ-DI and a decrease in serum adiponectin, supporting its proposed anti-inflammatory properties [24]. Similarly, findings from Abdallah et al. 2021 demonstrated that adding metformin XR (1000 mg daily) to methotrexate significantly enhanced clinical response, with 80.8% of patients achieving American College of Rheumatology 20% (ACR20) response [23], which is an improvement criterion response rate [26], and attaining DAS-28 remission compared with placebo. Complementing these results, two trials by Kumar et al. 2023 and Kumar et al. 2024 conducted in India indicate a consistent, time-dependent pattern of improvement in CRP, DAS-28, and HAQ-DI scores, together with favorable tolerability over six months of therapy [22, 25]. These findings address a critical unmet need in RA management, a condition increasingly recognized not only for chronic synovial inflammation but also for its systemic cardiometabolic consequences, including elevated risks of cardiovascular diseases (CVDs) and type 2 diabetes mellitus (T2DM) [5, 27]. Chronic inflammation and traditional risk factors predispose RA patients to accelerated atherosclerosis, while long-term corticosteroid exposure further increases the risk of diabetes and cardiovascular events in this population [28]. These burdens are among the most prevalent diseases worldwide. The prevalence of prediabetes, T2DM, and CVDs among middle-aged adults is high, contributing to a substantial overlap between metabolic disease and inflammatory arthritis [5, 27, 29–32]. In this context, metformin emerges as a rational adjunctive therapy, offering clinically relevant advantages beyond its glucose-lowering effects. Its well-established cardioprotective properties, favorable safety profile, low cost, and wide availability align closely with the complex comorbidity profile of RA [14]. Beyond its role as an adjunctive therapy in active RA, emerging evidence suggests metformin may modify the underlying disease process itself. In the trial by Abdallah et al., metformin significantly upregulated AMP-activated protein kinase (AMPK) and rebalanced key immune pathways, suppressing pro-inflammatory mediators (TNF-α, IL-6, IL-17 A, NF-κB, TGF-β1) while enhancing the anti-inflammatory cytokine IL-10 [23], changes that closely correlated with meaningful improvements in DAS-28. The biological plausibility of these findings is further supported by preclinical animal studies demonstrating that metformin reduces arthritis severity through immunomodulatory mechanisms, particularly by restoring the balance between Th17 and regulatory T (Treg) cells [33]. This mechanistic profile aligns with preclinical data showing the ability of metformin to inhibit fibroblast-like synoviocyte proliferation, osteoclastogenesis, and adiponectin-driven joint inflammation [34]. At the cellular level, preclinical in vitro studies demonstrate that metformin inhibits fibroblast-like synoviocyte proliferation and migration, which are key contributors to pannus formation, suppresses osteoclast differentiation through modulation of RANKL signaling, and attenuates adiponectin-driven joint inflammation via AMPK-dependent pathways [35–37]. These findings indicate a direct effect on synovial inflammation and bone remodeling by reducing osteoclast and bone resorption. Additionally, at the organism level, animal models of arthritis, particularly collagen-induced arthritis models, further support these observations by demonstrating reductions in synovitis, inflammatory cytokine expression, cartilage destruction, and bone erosion following metformin treatment [38–41]. Collectively, these mechanistic insights provide a strong biological foundation for the clinical findings observed in this meta-analysis and support the role of metformin as a potential immunometabolic modulator with disease-modifying properties and implications for long-term structural and systemic outcomes in RA. Critically, this therapeutic rationale is now supported by human epidemiological and causal evidence. Naffaa et al.’s population-based cohort study 2020 reported that adherence to metformin was associated with a 38% lower risk of incident RA (HR = 0.62), an effect notably pronounced in women [15]. Even more compelling, Liang et al.’s two-sample Mendelian randomization study 2025 (MR) provided genetic evidence of a causal protective relationship, reporting an odds ratio of 0.023 (p = 0.006) for RA incidence among individuals with genetic variants predisposing them to metformin use [16]. The absence of heterogeneity or horizontal pleiotropy in this MR analysis strengthens the inference that metformin exposure directly attenuates RA risk, likely through the same anti-inflammatory and immunometabolic pathways observed in clinical trials. Together, these findings suggest that metformin may not merely suppress symptoms but modify the underlying trajectory of RA from preclinical autoimmunity to established disease. All four included RCTs consistently demonstrate that metformin is well-tolerated in non-diabetic patients with RA [22–25]. The most reported adverse effects were mild to moderate gastrointestinal symptoms, including nausea, abdominal discomfort, and diarrhea, with only a small proportion of patients discontinuing treatment due to intolerance. For instance, Gharib et al. (2021) reported that 3 out of 33 participants (9.1%) in the metformin group withdrew from the study due to GI-related side effects [24], while the other three RCTs noted similarly low discontinuation rates and no serious adverse events [22, 23, 25]. Beyond its favorable safety profile, metformin offers distinct practical advantages: it is low-cost, widely available, and possesses cardioprotective properties, particularly relevant given the elevated cardiovascular disease risk in RA populations, and its weight-neutral to modest weight-reducing effect [5, 42, 43]. As highlighted by Kumar et al. 2024, these attributes make metformin a highly promising adjunctive therapy, especially in resource-limited settings where access to expensive biologic DMARDs remains constrained [22]. Although the findings of this meta-analysis are consistent and statistically significant, several important limitations warrant careful consideration. Methodological limitations inherent to the included trials constrain the strength of our conclusions. First, the duration of all trials was short-term, no longer than six months, precluding conclusions about the long-term efficacy of metformin, durability of the response, its impact on radiographic joint damage, or the risk of rare or delayed adverse events. Second, all included trials had relatively small sample sizes, ranging from 60 to 120 participants per study, yielding a total pooled population of only 350 individuals. This limited statistical power restricts the generalizability of the results, particularly for underrepresented subgroups such as elderly patients, those with comorbidities, or individuals with long-standing or refractory disease. Third, while all trials were randomized, only one was double-blinded [23]; the remaining three were single-blinded or open-label in design, raising the potential for performance and detection bias, especially for subjective outcomes like HAQ-DI and global disease activity. Finally, as noted, a substantial heterogeneity was present across trials for CRP after three months, which may include variations in background csDMARD regimens (methotrexate monotherapy versus combination therapy with leflunomide, sulfasalazine, or hydroxychloroquine) and metformin dosing (850–1000 mg twice daily). However, a leave-one-out sensitivity analysis confirmed the robustness of the pooled effect, and heterogeneity was negligible (I² = 0%) for all other outcomes, including CRP at six months, DAS-28, and HAQ-DI. Beyond these methodological constraints, critical evidence gaps remain. None of the trials assessed structural joint damage via validated radiographic scoring (e.g., Sharp-van der Heijde), despite this being a cornerstone of RA disease modification. Similarly, cardiovascular endpoints, highly relevant given the 1.5- to 2-fold increased CVD risk in RA and metformin’s known cardiometabolic benefits, were entirely absent. Additionally, the studies were conducted exclusively in Egypt and India, limiting generalizability to other ethnic, genetic, and socioeconomic populations. Finally, key patient subgroups, such as the elderly, those with long-standing or refractory disease, or individuals with significant comorbidities, were underrepresented, preventing meaningful subgroup analyses beyond sex, BMI, or disease duration. These gaps underscore the need for larger, longer-term follow-up duration, multicenter trials with standardized protocols and clinically meaningful endpoints, with a consideration of cardiovascular outcomes as future research directions to rigorously evaluate the structural and cardiovascular effects of metformin in RA. Specifically, trials should incorporate radiographic endpoints, such as the Sharp-van der Heijde score, to determine whether metformin slows or halts joint damage progression, a critical unmet outcome in current studies. Moreover, studies should aim to identify the optimal dosing regimen of metformin in non-diabetic RA patients and explore predictive biomarkers of therapeutic response like the baseline serum adiponectin levels, AMPK gene expression, or high-sensitivity CRP, as these may help stratify patients most likely to benefit from adjunctive therapy. Further investigation is also warranted into the potential synergistic effects of metformin when combined with specific conventional synthetic DMARDs (e.g., methotrexate, leflunomide, or hydroxychloroquine), which could inform personalized treatment algorithms. Finally, pre-specified subgroup analyses should be integrated into trial design to examine differential treatment effects based on sex, particularly in light of Naffaa et al. study finding of a stronger protective association in women, body mass index, and disease duration (early vs. established RA) [15]. Such analyses will enhance our understanding of metformin’s therapeutic scope and support its targeted use in diverse clinical contexts. Conclusion This systematic review suggests that the addition of metformin as an adjunctive therapy to conventional synthetic disease-modifying antirheumatic drugs (csDMARDs) may enhance anti-rheumatic effects through its anti-inflammatory and immunomodulatory effects. These benefits were demonstrated by reductions in serum C-reactive protein (CRP) levels and Disease Activity Score-28 (DAS-28), along with improvements in patients’ quality of life, as assessed by the Health Assessment Questionnaire Disability Index (HAQ-DI). However, larger, well-designed clinical trials with greater sample sizes and longer follow-up periods are required to validate these findings and improve their generalizability. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1