Key Points
Question
Does metformin, compared with placebo, reduce knee pain in patients with symptomatic knee osteoarthritis and overweight or obesity?
Findings
In this randomized clinical trial that included 107 participants, compared with placebo, metformin, 2000 mg/d, reduced knee pain measured on a 0- to 100-mm visual analog scale (score of 100 = worst): −31.3 mm in the metformin group and −18.9 mm in the placebo group (between-group difference, −11.4 mm; P = .01), corresponding to an effect size of 0.43.
Meaning
In patients with knee osteoarthritis and overweight or obesity, metformin, 2000 mg/d, for 6 months had a moderate and statistically significant effect on knee pain reduction compared with placebo.
Abstract
Importance
Preclinical and preliminary human evidence suggests that metformin, a first-line treatment for type 2 diabetes, reduces inflammation, preserves cartilage, and improves knee pain in knee osteoarthritis.
Objective
To evaluate the effects of metformin on knee pain at 6 months in participants with symptomatic knee osteoarthritis and overweight or obesity.
Design, Setting, and Participants
Community-based randomized, parallel-group, double-blind, placebo-controlled clinical trial that used telemedicine to recruit and follow up participants remotely. Individuals with knee pain for 6 months or longer, a pain score greater than 40 mm on a 100-mm visual analog scale (VAS), and body mass index of 25 or higher were recruited from the community through local and social media advertisements in Victoria, Australia, between June 16, 2021, and August 1, 2023. Final follow-up occurred on February 8, 2024.
Interventions
Participants were randomly assigned to receive either oral metformin, 2000 mg/d (n = 54), or identical placebo (n = 53) for 6 months.
Main Outcomes and Measures
The primary outcome was change in knee pain, measured using a 100-mm VAS (score range, 0-100; 100 = worst; minimum clinically important difference = 15) at 6 months.
Results
Of 225 participants assessed for eligibility, 107 (48%) were randomized (mean age, 58.8 [SD, 9.5] years; 68% female) and assigned to receive metformin or placebo. Eighty-eight participants (82%) completed the trial. At 6 months, the mean change in VAS pain was −31.3 mm in the metformin group and −18.9 mm in the placebo group (between-group difference, −11.4 mm; 95% CI, −20.1 to −2.6 mm; P = .01), corresponding to an effect size (standardized mean difference) of 0.43 (95% CI, 0.02-0.83). The most common adverse events were diarrhea (8 [15%] in the metformin group and 4 [8%] in the placebo group) and abdominal discomfort (7 [13%] in the metformin group and 5 [9%] in the placebo group).
Conclusions and Relevance
These results support use of metformin for treatment of symptomatic knee osteoarthritis in people with overweight or obesity. Because of the modest sample size, confirmation in a larger clinical trial is warranted.
Trial Registration
ANZCTR Identifier: ACTRN12621000710820
This randomized clinical trial assesses the effect of metformin on knee pain among patients with symptomatic knee osteoarthritis and overweight or obesity.
Introduction
Few effective therapies exist for knee osteoarthritis (OA), which affects 365 million people worldwide.1 More than 50% of patients with knee OA have overweight or obesity.2,3,4 Obesity-related knee OA is mediated by excess weight-bearing on joints, inflammation, and impaired glucose and lipid metabolism, which promote systemic inflammation, oxidative stress, and metabolic dysfunction in joint tissues, contributing to cartilage degradation and disease progression.5,6,7 Drugs that treat obesity and associated inflammatory and glucose and lipid abnormalities may improve outcomes in knee OA.
Metformin is a safe, inexpensive, well-tolerated oral biguanide that has been first-line therapy for type 2 diabetes for more than 60 years. Metformin reduces hepatic glucose production, insulin resistance, and endogenous hyperinsulinemia; causes modest weight loss; and reduces inflammation in people with and without diabetes.8,9 Pleiotropic effects of metformin, such as anti-inflammatory properties, and improved glucose and lipid metabolism, such as increased adenosine monophosphate–activated protein kinase stimulation or reduced insulin resistance, may reduce knee pain in OA.10,11 Therefore, this randomized clinical trial was conducted to determine whether 6 months of metformin, compared with placebo, reduced knee pain at 6 months in individuals with symptomatic knee OA and overweight or obesity.
Methods
Trial Design
The study was a community-based, randomized, double-blind, placebo-controlled clinical trial conducted in Victoria, Australia, from June 16, 2021, to February 8, 2024. Because of COVID-19 restrictions, the clinical trial was conducted via telemedicine. The trial protocol and statistical analysis plan (SAP) are available in Supplement 1 and Supplement 2, respectively. The study was conducted through the School of Public Health and Preventive Medicine, Monash University. Ethics approval was obtained from the Alfred Hospital Ethics Committee (708/20) and the Monash University Human Research Ethics Committee (28498). Informed consent was collected electronically using REDCap. The clinical trial was originally designed as 2 studies: study 1 was designed to compare the change in visual analog scale (VAS)–assessed knee pain from baseline to 6 months between metformin and placebo; study 2 was designed to assess the annual percentage change in tibial cartilage volume and the change in VAS knee pain from baseline to 24 months. As funding was only available for study 1, only study 1 was conducted, and study 2 was not initiated. The study protocol for study 1 has been published.12 The trial reporting followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.13
Participants
Participants were recruited from the community within 100 km of Melbourne, Victoria, Australia, using social and local media advertisements. Potential participants contacted the research center through a website, email, or phone number and were provided with general information about the study.
For each potential participant, the knee with symptomatic OA was the index knee. If both knees were symptomatic and eligible based on VAS pain score, the knee with the higher VAS pain score was designated the index knee; if pain levels were equal, the knee with less severe radiographic OA (joint space narrowing) was selected as the index knee; if both were identical, the dominant knee, defined as the knee corresponding to the participant’s preferred leg to kick a ball, was the index knee.
Inclusion criteria were as follows: age older than 40 years; body mass index (BMI) of 25 or higher (calculated as weight in kilograms divided by height in meters squared); knee pain for 6 months or longer; pain score greater than 40 mm on a 100-mm VAS; and presence of knee OA according to criteria established by the American College of Rheumatology14 that were adapted to telemedicine. Because physical examination was not performed, during the telemedicine visit participants were instructed about how to examine their knee and asked to report knee crepitus, tenderness, and warmth. Bony enlargement was not evaluated because assessment of bony enlargement can be subjective.
Exclusion criteria were severe radiographic knee OA (Kellgren-Lawrence grade 4 [range: grade 0, no OA, to grade 4, severe OA]), severe knee pain (>80 mm on VAS), inflammatory arthritis, significant knee injury, diabetes requiring glucose-lowering therapy, knee surgery on the index knee in the past year or planned in the next 6 months, intra-articular hyaluronic acid injection during the past 6 months, corticosteroid injection during the past 3 months, use of investigational drug or device within 30 days prior to randomization, and conditions affecting lower limb function, such as neurological conditions (eg, stroke). Patients with kidney impairment, defined as any reduction in estimated glomerular filtration rate, or liver impairment, defined as 1.5-fold increase in any liver enzymes, were excluded due to the potential for harm from metformin treatment.
Assessment for Eligibility With Telehealth
Participants were screened by a research assistant and a rheumatology-trained physician using questionnaires administered by telemedicine. Potentially eligible participants received referral forms and instructions for knee radiography and blood tests via email. Radiography and blood tests were conducted at community health facilities, with results electronically transmitted to study investigators via online platforms to assess eligibility and safety. Radiography results were used to identify patients with severe radiographic knee OA (Kellgren-Lawrence grade 4), who were excluded from the clinical trial. Blood tests were performed at the community health facilities to measure kidney and liver function, plasma glucose, and vitamin B12 levels.
Randomization and Blinding
Participants were allocated (1:1) to receive metformin or placebo using computer-generated random numbers prepared by a statistician not otherwise involved in the trial. Randomization was performed using a central automated allocation procedure with permuted blocks of sizes 4 and 6. Outcome measures were obtained by research assistants unaware of group allocation. Participants, investigators, and statisticians were blinded to group allocation.
Intervention
The intervention was metformin extended release (up to 2000 mg) or placebo of identical appearance (Syntro) taken once daily. Medications were dispensed by the Syntro pharmacy. Study medication (metformin or placebo) was initiated at 500 mg once daily with the evening meal and increased to 2000 mg once daily over 6 weeks to minimize gastrointestinal adverse effects. Titration was performed by research staff following a prespecified protocol under clinical supervision (Y.Z.L., F.M.C.). Participants were asked to keep other medications unchanged if possible and to use acetaminophen (paracetamol) for significant knee pain. Adherence was assessed using pill counts of returned medication and monitoring through telephone calls at 2 weeks, at 4 weeks, and monthly, during which research staff recorded participant-reported medication use, dosage, and any missed study medication.
Primary Outcome
The primary outcome was change in pain in the index knee at 6 months relative to baseline. This outcome was assessed at baseline, 3 months, and 6 months using a 100-mm VAS. Participants rated their knee pain over the last 7 days on a 100-mm VAS with anchors of “no pain” (score of 0) and “worst imaginable pain” (score of 100). The minimum clinically important difference (MCID) was 15.15
Secondary Outcomes
Secondary outcomes were changes at 6 months from baseline in pain, stiffness, and function, assessed using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain (0-500), stiffness (0-200), and function (0-1700) subscales (MCID = 12% change from baseline value for each score16)17; health-related quality of life, measured using the Assessment of Quality of Life 8 Dimensions (AQoL-8D; MCID = 0.06)18; and percentage of participants meeting treatment response criteria by the Outcome Measures in Rheumatology–Osteoarthritis Research Society International (OMERACT-OARSI).19 A validated, modified OMERACT-OARSI responder definition not requiring a patient global assessment20 was used, as this measure was omitted due to an oversight in the study’s data collection process. Participants were categorized as having a response if their OMERACT-OARSI score met either of the following criteria: (1) at least a 50% improvement and at least a 20-point change in mean WOMAC pain or WOMAC function scores or (2) at least a 20% improvement and at least a 10-point change in both WOMAC pain and WOMAC function scores. Because the study ended after 6 months, none of the secondary outcomes prespecified at 12 or 24 months were obtained. Magnetic resonance imaging (MRI) was not performed, as it was only planned to be performed as part of study 2. Therefore, the prespecified secondary outcomes related to MRI-assessed cartilage defects, bone marrow lesions, and effusion synovitis were not obtained.
Adverse Events
Adverse events were ascertained at 2 weeks and 4 weeks after randomization, and monthly to 6 months follow-up by telemedicine. At other times, participants contacted the study center, and a telemedicine visit was conducted by the research assistant and/or clinician. Serious adverse events were determined by study physicians. Details of adverse events and the relationship to the intervention were documented and reported to the ethics committees.
Other Assessments
Age, sex, education level, employment status, and comorbidities were collected using a self-administered questionnaire at baseline. Height and weight were self-reported at screening and at 6-month follow-up. Radiography of the index knee (weight-bearing, posterior-anterior, semiflexed) was performed at screening to assess joint space narrowing (grade 0-3) and presence or absence of osteophytes, scored by an observer with training in radiology and more than 20 years of experience in OA imaging assessment (Y.W.; intrarater reliability, 0.97), according to the published atlas.21
Sample Size Calculation
The mean VAS pain score at baseline was 55 mm in previous clinical trials of knee OA using similar eligibility criteria, with an SD of 24 mm for change from baseline.22,23 To detect a 15-mm between-group difference in VAS pain scores at 6 months, 41 participants per group were required for 80% power and a 2-sided P = .05. Accounting for 20% loss to follow-up,22,24 102 participants (51 per group) were required.
Statistical Analysis
The original SAP included study 1 and study 2. Because only study 1 was undertaken, the SAP was modified on March 26, 2021, and the current SAP was finalized in September 2023, before any data were reviewed.12 The primary analyses were performed according to the original randomization group for each participant. Continuous variables are expressed as means and standard deviations or medians and interquartile ranges, and categorical variables are expressed as counts and percentages. t Tests and χ2 tests were used to compare characteristics between participants who completed the study and those who withdrew. Primary and secondary outcomes (except the OMERACT-OARSI responder outcomes) were analyzed using linear mixed-effects models with treatment, time, and treatment × time interaction as covariates, as well as the corresponding outcome measures at baseline. A random intercept was included for individual participants, and also a random slope with time. Results were reported as between-group differences in mean change from baseline to 6 months, including 95% confidence intervals and P values. Logistic regression was used to analyze OMERACT-OARSI response. To account for missing data, multiple imputation by chained equations was performed for each treatment group and each outcome, using age, sex, BMI, and nonmissing outcome measures at baseline. Estimates from 20 imputed datasets were averaged following the Rubin rules. The standardized mean difference (effect size) was calculated for the primary outcome by modeling standardized changes from baseline (standardized by their standard deviation), adjusted for the standardized baseline VAS score, using imputed data. The Hochberg adjustment for multiple co–primary end points, specified in the original SAP for study 2, was removed in the final SAP, as study 1 had 1 primary outcome. Secondary outcomes should be considered exploratory due to lack of adjustment for multiple comparisons.
Prespecified stratified and post hoc analyses were performed to evaluate subgroup responses (female vs male; VAS pain severity at baseline [scores of <70 mm vs ≥70 mm25,26]; and joint space narrowing [grade 0-1 vs 2]) for the primary outcome. These analyses were not specified in the final protocol but were specified in the published methods article.12
Post Hoc Analyses
Interaction testing for the subgroup analyses was not prespecified and was conducted post hoc. A post hoc analysis of primary and secondary outcomes was conducted using complete cases (defined as those who provided the primary outcome, ie, VAS pain score, at 6 months) with baseline and 6-month assessments.
Statistical analyses were conducted using Stata version 17.0 (StataCorp). A 2-sided P < .05 was considered statistically significant.
Results
Of 225 participants screened for eligibility between June 16, 2021, and August 1, 2023, 107 were randomly assigned to receive metformin (n = 54) or placebo (n = 53) (Figure 1). The mean age of participants was 58.8 (SD, 9.5) years, 73 (68%) were female, and the mean BMI was 32.7 (SD, 6.2) (Table 1). Eighty-eight participants (82%) completed the primary outcome measure at 6 months, 45 (83%) in the metformin group and 43 (81%) in the placebo group. There were no statistically significant differences between participants who completed the primary outcome at 6 months and those who did not (eTable 1 in Supplement 3).
Figure 1. Participant Flow in a Trial of Metformin for Symptomatic Knee Osteoarthritis in People With Overweight or Obesity.
aAbnormal blood test results included fasting glucose >6.0 mmol/L (>108 mg/dL), high-sensitivity C-reactive protein >5 mg/L, vitamin B12 <200 pmol/L, or cholesterol <3.5 or >5.5 mmol/L (<135 or >212 mg/dL).
Table 1. Baseline Participant Characteristics.
| Characteristics | Metformin (n = 54) | Placebo (n = 53) |
|---|---|---|
| Age, mean (SD), y | 57.8 (9.5) | 59.7 (9.4) |
| Sex, No. (%) | ||
| Female | 38 (70) | 35 (66) |
| Male | 16 (30) | 18 (34) |
| Body mass indexa | ||
| Mean (SD) | 33.1 (6.3) | 32.2 (6.1) |
| Median (IQR) | 32.3 (27.7-36.4) | 30.6 (28.1-34.9) |
| Education level, No. (%) | ||
| Completed tertiary degree or diploma | 40 (74) | 34 (64) |
| Certificate or work toward a tertiary degree or diploma | 10 (19) | 9 (17) |
| Partial or completed high school/technical school | 4 (7) | 10 (19) |
| Employment, No. (%) | ||
| Full-time | 24 (44) | 22 (42) |
| Part-time/casual | 13 (24) | 15 (28) |
| Unemployed | 17 (31) | 16 (30) |
| Pain medication use, No. (%) | ||
| Acetaminophen | 33 (61) | 27 (51) |
| Nonsteroidal anti-inflammatory drugs | 23 (43) | 27 (51) |
| Antidepressants | 10 (19) | 9 (17) |
| Turmeric | 6 (11) | 4 (8) |
| Glucosamine | 5 (9) | 6 (11) |
| Other analgesicsb | 6 (11) | 5 (9) |
| Comorbidities, No. (%) | ||
| Dyslipidemia | 15 (28) | 15 (28) |
| Hypertension | 13 (24) | 14 (26) |
| Cardiovascular diseasec | 2 (4) | 0 |
| Radiographic joint space narrowing, No. (%) | ||
| Grade 0 (normal) | 20 (37) | 12 (23) |
| Grade 1 (mild) | 15 (28) | 25 (47) |
| Grade 2 (moderate) | 19 (35) | 16 (30) |
| Radiographic osteophyte, No. (%) | 35 (65) | 37 (70) |
| Fasting glucose, mmol/L | ||
| Mean (SD) | 5.3 (0.6) | 5.3 (0.6) |
| Median (IQR) | 5.3 (4.9-5.8) | 5.2 (5.0-5.7) |
| Vitamin B12, pg/mL | ||
| Mean (SD) | 386.7 (148.6) | 355.4 (125.9) |
| Median (IQR) | 357 (285-471) | 348 (267-422) |
| Knee pain VAS scored | ||
| Mean (SD) | 60.2 (18.5) | 58.5 (20.9) |
| Median (IQR) | 63.5 (55.0-70.0) | 60.0 (50.0-73.0) |
| WOMAC score, mean (SD)e | ||
| Pain | 226.2 (106.3) | 220.8 (109.3) |
| Stiffness | 104.6 (45.7) | 92.9 (48.6) |
| Function | 829.0 (365.8) | 773.3 (395.8) |
| Quality-of-life score, mean (SD)f | 0.7 (0.2) | 0.7 (0.2) |
SI conversion: To convert glucose to milligrams per deciliter, divide by 0.0555.
Calculated as weight in kilograms divided by height in meters squared.
Any of opioids, steroids, and compound analgesics.
Any of ischemic heart disease, transient ischemic attack, stroke, and carotid artery stenosis.
Participants self-reported knee pain over the past 7 days on a 100-mm visual analog scale (VAS; score range, 0-100; 100 = worst; minimum clinically important difference = 15).
The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) is self-administered with 5 questions on pain, 2 questions on stiffness, and 17 questions on physical function (scale range for each question, 0 [none] to 100 [unbearable]; total scores for pain, 0-500; for stiffness, 0-200; and for function, 0-1700).
Quality of life was measured using the self-administered Assessment of Quality of Life 8 Dimensions, which includes 35 questions evaluating various quality-of-life dimensions. The resulting utility score ranges from 1.00 (full health) to 0.00 (health states equivalent to death) and −0.04 (health states worse than death).
Study Drug Adherence
Only 38.3% of participants returned medication in this telemedicine trial (41% in the metformin group and 36% in the placebo group). Adherence based on returned medication was 82% in the metformin group and 79% in the placebo group. Ninety-seven participants (91%) reported adherence via telemedicine (90% in the metformin group and 81% in the placebo group). Over 6 months, 81% of participants in the metformin group and 74% in the placebo group had more than 80% adherence as assessed by telemedicine. Forty-three metformin group participants (80%) achieved the 2000-mg/d dose of metformin, and 39 placebo group participants (74%) reached the full placebo dose after 6 weeks. Mean weight change over 6 months was −1.8 kg in the metformin group and −1.2 kg in the placebo group.
Primary Outcome
At 6 months, the metformin group experienced a greater reduction in VAS pain scores compared with the placebo group (mean, –31.3 mm vs –18.9 mm) (Table 2 and Figure 2). The adjusted between-group difference in VAS score change was –11.4 mm (95% CI, –20.1 to –2.6 mm; P = .01) (Table 2), consistent with a standardized mean difference (effect size) of 0.43 (95% CI, 0.02-0.83).
Table 2. Change in Study End Points Over 6 Monthsa.
| Outcomes | Metformin (n = 54) | Placebo (n = 53) | Between-group difference in changeb | P value | ||||
|---|---|---|---|---|---|---|---|---|
| Baseline | 6 mo | Change | Baseline | 6 mo | Change | |||
| Primary outcome | ||||||||
| Knee pain VAS score, mean (95% CI), mmc | 60.2 (55.2-65.1) | 28.8 (21.6-36.1) | −31.3 (−38.4 to −24.2) | 58.5 (52.8-64.2) | 39.6 (29.5-49.6) | −18.9 (−29.7 to −8.2) | −11.4 (−20.1 to −2.6) | .01 |
| Secondary outcomes | ||||||||
| WOMAC score, mean (95% CI)d | ||||||||
| Pain | 226.2 (197.5-254.9) | 112.3 (80.7-143.8) | −113.9 (−148.6 to −79.3) | 220.8 (191.0-250.6) | 152.6 (109.5-195.7) | −68.2 (−113.2 to −23.2) | −42.4 (−83.9 to −1.0) | .045 |
| Stiffness | 104.6 (92.3-117.0) | 47.8 (33.8-61.8) | −56.9 (−70.9 to −42.9) | 92.9 (79.7-106.2) | 66.3 (48.3-84.4) | −26.7 (−46.7 to −6.7) | −23.0 (−40.4 to −5.7) | .01 |
| Function | 829.0 (730.3-927.7) | 402.9 (290.9-514.8) | −426.1 (−524.1 to −328.2) | 773.3 (665.5-881.1) | 551.5 (407.8-695.3) | −221.7 (−367.2 to −76.2) | −179.8 (−313.0 to −46.6) | .009 |
| Quality-of-life score, mean (95% CI)e | 0.67 (0.63-0.72) | 0.70 (0.65-0.75) | 0.02 (−0.002 to 0.05) | 0.69 (0.64-0.74) | 0.70 (0.64-0.76) | 0.01 (−0.03 to 0.05) | 0.01 (−0.02 to 0.05) | .47 |
| OMERACT-OARSI response, No./total (%)f | 35/54 (65) | 24/53 (45) | 19 (−1 to 40)g | .07 | ||||
The analysis was based on imputed data to address missing primary and secondary outcomes. Imputation refers to a statistical method used to estimate missing values in a dataset. In this study, multiple imputation by chained equations was applied separately for each treatment group and outcome. Predictors used for the imputation included baseline age, sex, body mass index, and nonmissing baseline values of the outcomes. The imputed datasets were then combined using Rubin rules to generate the reported estimates.
Adjusted for baseline value of corresponding outcome.
Participants self-reported knee pain over the past 7 days on a 100-mm visual analog scale (VAS; score range, 0-100; 100 = worst; minimum clinically important difference = 15).
The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) is self-administered with 5 questions on pain, 2 questions on stiffness, and 17 questions on physical function (scale range for each question, 0 [none] to 100 [unbearable]; total scores for pain, 0-500; for stiffness, 0-200; and for function, 0-1700).
Quality of life was measured using the self-administered Assessment of Quality of Life 8 Dimensions, which includes 35 questions evaluating various quality-of-life dimensions. The resulting utility score ranges from 1.00 (full health) to 0.00 (health states equivalent to death) and −0.04 (health states worse than death).
Treatment response was assessed at 6 months using the Outcome Measures in Rheumatology–Osteoarthritis Research Society International (OMERACT-OARSI) criteria to measure change from baseline. For details on response criteria, please refer to the Methods section of the text.
Between-group absolute difference in percentage change (95% CI). The odds ratio for this measure is 2.21 (95% CI, 0.92-5.31; P = .07).
Figure 2. Changes in 6-Month Knee Pain VAS Scores in Patients Treated With Metformin vs Placebo.
Baseline knee pain scores on a 100-mm visual analog scale (VAS; score range, 0-100; 100 = worst; minimum clinically important difference = 15) are shown in ascending order for the metformin group and descending order for the placebo group. Lines connect each participant’s baseline score to their 6-month follow-up knee pain VAS score. An x indicates participants with a baseline score but no 6-month follow-up data. The box plots show the distribution of baseline and 6-month knee pain scores for each group. In the right panel, box plots for each treatment group show the distribution of changes in knee pain from baseline to 6 months, excluding participants with missing follow-up data. Each box spans the IQR, with the median shown as a solid line and the mean shown as a dashed line. Whiskers extend to the most extreme values within 1.5 times the IQR, and open circles represent individual data points outside this range.
Secondary Outcomes
Compared with placebo, metformin significantly reduced WOMAC scores for pain (−113.9 with metformin vs −68.2 with placebo; adjusted between-group difference in change, −42.4; 95% CI, −83.9 to −1.0; P = .045), stiffness (−56.9 vs −26.7; adjusted difference, −23.0; 95% CI, −40.4 to −5.7; P = .01), and function (−426.1 vs −221.7; adjusted difference, −179.8; 95% CI, −313.0 to −46.6; P = .009) (Table 2; eFigures 1 and 2 in Supplement 3). There was no significant difference between the 2 groups in the AQoL-8D score change (0.01; 95% CI, −0.02 to 0.05; P = .47). There was no significant difference in the proportion of participants who met criteria for OMERACT-OARSI response (35/54 [65%] in the metformin group and 24/53 [45%] in the placebo group; odds ratio, 2.21; 95% CI, 0.92-5.31; P = .07) or in the difference in VAS pain score change at 3 months (−2.5 mm; 95% CI, −11.7 to 6.6 mm; P = .58) (Figure 3).
Figure 3. Knee Pain VAS Scores Over 6 Months in Patients Treated With Metformin vs Placebo Using Imputed Data.
Participants self-reported knee pain over the past 7 days on a 100-mm visual analog scale (VAS; score range, 0-100; 100 = worst; minimum clinically important difference = 15).
Subgroup Analyses
Adjusted between-group differences in VAS pain score changes over 6 months by sex were as follows: for females, −12.9 mm (95% CI, −23.7 to −2.1 mm; P = .02) and for males, −5.4 mm (95% CI, −21.4 to 10.6 mm; P = .51). Adjusted between-group differences in VAS pain score changes over 6 months by baseline VAS pain severity were as follows: for mild to moderate pain (<70 mm), −15.8 mm (95% CI, −25.2 to −6.4 mm; P = .001) and for severe pain (≥70 mm), −4.3 mm (95% CI, −19.8 to 11.2 mm; P = .59). Adjusted between-group differences in VAS pain score changes over 6 months by joint space narrowing grade were as follows: for grade 0 to 1, −11.0 mm (95% CI, −21.7 to −0.2 mm; P = .045) and for grade 2, −14.2 (95% CI, −26.8 to −1.5; P = .03) (eTable 2 in Supplement 3).
Post Hoc Analyses
No statistically significant interactions were found for subgroup analyses by sex (P = .85), VAS pain severity score (P = .74), or joint space narrowing grade (P = .94). Post hoc analyses using complete-case data showed significant between-group differences in change in VAS score (−12.5 mm; 95% CI, −22.1 to −2.9 mm; P = .01) and in WOMAC scores for pain (−39.9; 95% CI, −70.8 to −9.0; P = .01), stiffness (−22.1; 95% CI, −35.5 to −8.6; P = .001), and function (−151.6; 95% CI, −283.2 to −20.0; P = .02) (eTable 3 in Supplement 3).
Adverse Events
Forty-one adverse events were reported, including 25 in the metformin group among 16 participants (30%) and 16 in the placebo group among 10 participants (19%) (Table 3). No serious adverse events were reported. The most frequent adverse events were diarrhea (8 participants [15%] in the metformin group and 4 [8%] in the placebo group) and abdominal discomfort (7 [13%] in the metformin group and 5 [9%] in the placebo group). These gastrointestinal adverse events were generally mild or moderate in severity and were considered either possibly or probably related to the study medication, resulting in 8 participants (5 in the metformin and 3 in the placebo group) temporarily discontinuing study medication and 3 participants (2 in the metformin and 1 in the placebo group) permanently discontinuing study medication.
Table 3. Summary of Adverse Eventsa.
| Adverse events | No. (%) | |
|---|---|---|
| Metformin (n = 54) | Placebo (n = 53) | |
| Total No. of adverse events | 25 | 16 |
| Participants with any adverse events | 16 (30) | 10 (19) |
| Gastrointestinal adverse events | ||
| Diarrhea | 8 (15) | 4 (8) |
| Abdominal discomfort | 7 (13) | 5 (9) |
| Nausea or vomiting | 4 (7) | 4 (8) |
| Indigestion | 2 (4) | 1 (2) |
| Respiratory tract infection (COVID-19) | 2 (4) | 0 |
| Headache | 2 (4) | 0 |
| Knee swelling | 0 | 1 (2) |
| Bad fall | 0 | 1 (2) |
No serious adverse events were reported in either group. A serious adverse event is defined as any event that results in death, is life-threatening, requires hospitalization or prolongation of existing hospitalization, results in persistent or significant disability, or is a congenital anomaly or birth defect.
Discussion
In this community-based, randomized, placebo-controlled trial of participants with symptomatic knee OA and overweight or obesity, treatment with metformin, 2000 mg/d, for 6 months resulted in a moderate reduction in knee pain and stiffness and improvement in knee function. These findings support a potential role of metformin in improving symptoms in individuals with knee OA and overweight or obesity. However, given the relatively small sample size, a confirmatory study is needed.
Animal studies consistently showed a benefit of metformin for pain and joint structural outcomes in OA, with supporting data from human studies.27,28,29 Among people with obesity in the Osteoarthritis Initiative followed up for 4 years, the 56 people treated with metformin had less knee cartilage loss compared with the 762 people not treated with metformin (0.71% vs 1.57%; difference, −0.86%; 95% CI, −1.58% to −0.15%). Over 6 years of follow-up in that study, there were no significant differences in rates of knee replacements (among those treated with metformin, 5.4% [3/56]; among those not treated, 11.6% [88/762]).27 A retrospective cohort study in Taiwan (matched on gender, age, and index year) reported that those with OA and type 2 diabetes treated with a combination of a cyclooxygenase 2 inhibitor and metformin (n = 968) had a lower rate of joint replacements over 10 years compared with those treated with a cyclooxygenase 2 inhibitor alone (n = 1936) (with metformin, 12.8% [124/968]; without metformin, 16.2% [314/1936]; P = .003).28 In a retrospective cohort study of 20 937 individuals with diabetes treated with metformin who were matched to individuals taking sulfonylurea drugs, in analyses using propensity scores and time-conditional matching, those taking metformin had a lower risk of developing OA (2.7% [568/20 937] vs 3.9% [817/20 937]; P < .001).29 An electronic health record study in primary care did not show a significant association between a new diagnosis of OA in patients prescribed metformin (n = 1838) compared with those not prescribed metformin (n = 1379) (18.9% [347/1838] vs 17.7% [244/1379]).30 To date, only 1 trial has examined metformin in patients with knee OA and overweight.31
In the current study, at 6-month follow-up, metformin had a moderate effect on pain reduction (effect size, 0.43; 95% CI, 0.02-0.83), which is higher than the effect size reported for nonsteroidal anti-inflammatory drugs over 6 to 12 weeks in a study of patients with hip and knee OA (effect size, 0.32; 95% CI, 0.24-0.39).32 For the primary outcome in this trial, the 15-mm MCID used to calculate statistical power was not obtained. Thus, the results may not be clinically meaningful. The current study did not identify a benefit of metformin at 3-month follow-up but identified a significant benefit at 6-month follow-up. The time course for any potential benefit of metformin in knee OA is unknown. To date, prospective human studies investigating metformin in OA have examined outcomes after at least 1 year of treatment,27,28,29 except for a 4-month trial.31 Similarly, animal studies in murine and rat models have assessed outcomes after treatment durations equivalent to more than 4 years in humans.33 The pleiotropic mechanisms of metformin, including its effects on inflammation and glucose and lipid metabolism, suggest that the benefits may take longer than 3 months in humans to manifest. This may explain why an effect was not observed at 3 months but was observed at 6 months.11,33
Limitations
This study has several limitations. First, 19 participants (18%) were lost to follow-up and did not provide primary outcome data at 6 months. Second, adherence was assessed through telemedicine monitoring by research staff because only 38.3% of participants returned their medication for the adherence measure consisting of pill counts. For this reason, adherence may have been overestimated. Third, due to the use of remote assessments in this study, weight and the documentation of arthritis features were self-reported and remotely evaluated, which may have introduced inaccuracy and bias. Fourth, race and ethnicity data were not reported for this study. This affects the ability to generalize study results. Fifth, it is possible that metformin’s adverse effects may affect blinding and bias self-reporting. Sixth, this study did not include objective measures of functioning, which would have provided a more objective and less biased assessment of the effects of metformin.
Conclusions
These results support use of metformin for treatment of symptomatic knee OA in people with overweight or obesity. Because of the modest sample size, confirmation in a larger clinical trial is warranted.
Trial Protocol
Statistical Analysis Plan
eTable 1. Baseline Participants’ Characteristics Between Those Completed and Withdrew
eTable 2. Subgroup Analysis on Change in Primary Outcome (VAS Pain) Over 6 Months
eTable 3. Change in Study End Points Over 6 Months in Complete Data
eFigure 1. Changes in WOMAC Pain (A), Stiffness (B) and Function (C) in Patients Treated With Metformin vs. Placebo
eFigure 2. WOMAC Pain (A), Stiffness (B) and Function (C) in 6 Months Using Imputed Data
Data Sharing Statement
References
- 1.Long H, Liu Q, Yin H, et al. Prevalence trends of site-specific osteoarthritis from 1990 to 2019: findings from the Global Burden of Disease Study 2019. Arthritis Rheumatol. 2022;74(7):1172-1183. doi: 10.1002/art.42089 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Knoop J, van der Leeden M, Thorstensson CA, et al. Identification of phenotypes with different clinical outcomes in knee osteoarthritis: data from the Osteoarthritis Initiative. Arthritis Care Res (Hoboken). 2011;63(11):1535-1542. doi: 10.1002/acr.20571 [DOI] [PubMed] [Google Scholar]
- 3.Dell’Isola A, Allan R, Smith SL, Marreiros SS, Steultjens M. Identification of clinical phenotypes in knee osteoarthritis: a systematic review of the literature. BMC Musculoskelet Disord. 2016;17(1):425. doi: 10.1186/s12891-016-1286-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Deveza LA, Melo L, Yamato TP, Mills K, Ravi V, Hunter DJ. Knee osteoarthritis phenotypes and their relevance for outcomes: a systematic review. Osteoarthritis Cartilage. 2017;25(12):1926-1941. doi: 10.1016/j.joca.2017.08.009 [DOI] [PubMed] [Google Scholar]
- 5.Thijssen E, van Caam A, van der Kraan PM. Obesity and osteoarthritis, more than just wear and tear: pivotal roles for inflamed adipose tissue and dyslipidaemia in obesity-induced osteoarthritis. Rheumatology (Oxford). 2015;54(4):588-600. doi: 10.1093/rheumatology/keu464 [DOI] [PubMed] [Google Scholar]
- 6.Batushansky A, Zhu S, Komaravolu RK, South S, Mehta-D’souza P, Griffin TM. Fundamentals of OA: an initiative of Osteoarthritis and Cartilage: obesity and metabolic factors in OA. Osteoarthritis Cartilage. 2022;30(4):501-515. doi: 10.1016/j.joca.2021.06.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wei G, Lu K, Umar M, et al. Risk of metabolic abnormalities in osteoarthritis: a new perspective to understand its pathological mechanisms. Bone Res. 2023;11(1):63. doi: 10.1038/s41413-023-00301-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Saisho Y. Metformin and inflammation: its potential beyond glucose-lowering effect. Endocr Metab Immune Disord Drug Targets. 2015;15(3):196-205. doi: 10.2174/1871530315666150316124019 [DOI] [PubMed] [Google Scholar]
- 9.Rojas LB, Gomes MB. Metformin: an old but still the best treatment for type 2 diabetes. Diabetol Metab Syndr. 2013;5(1):6. doi: 10.1186/1758-5996-5-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Yerevanian A, Soukas AA. Metformin: mechanisms in human obesity and weight loss. Curr Obes Rep. 2019;8(2):156-164. doi: 10.1007/s13679-019-00335-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Song Y, Wu Z, Zhao P. The effects of metformin in the treatment of osteoarthritis: current perspectives. Front Pharmacol. 2022;13:952560. doi: 10.3389/fphar.2022.952560 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Lim YZ, Wang Y, Urquhart DM, et al. Metformin for knee osteoarthritis with obesity: study protocol for a randomised, double-blind, placebo-controlled trial. BMJ Open. 2023;13(12):e079489. doi: 10.1136/bmjopen-2023-079489 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Moher D, Hopewell S, Schulz KF, et al. CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340:c869. doi: 10.1136/bmj.c869 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Altman R, Asch E, Bloch D, et al. ; Diagnostic and Therapeutic Criteria Committee of the American Rheumatism Association . Development of criteria for the classification and reporting of osteoarthritis: classification of osteoarthritis of the knee. Arthritis Rheum. 1986;29(8):1039-1049. doi: 10.1002/art.1780290816 [DOI] [PubMed] [Google Scholar]
- 15.Tubach F, Ravaud P, Martin-Mola E, et al. Minimum clinically important improvement and patient acceptable symptom state in pain and function in rheumatoid arthritis, ankylosing spondylitis, chronic back pain, hand osteoarthritis, and hip and knee osteoarthritis: results from a prospective multinational study. Arthritis Care Res (Hoboken). 2012;64(11):1699-1707. doi: 10.1002/acr.21747 [DOI] [PubMed] [Google Scholar]
- 16.Angst F, Aeschlimann A, Stucki G. Smallest detectable and minimal clinically important differences of rehabilitation intervention with their implications for required sample sizes using WOMAC and SF-36 quality of life measurement instruments in patients with osteoarthritis of the lower extremities. Arthritis Rheum. 2001;45(4):384-391. doi: [DOI] [PubMed] [Google Scholar]
- 17.Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol. 1988;15(12):1833-1840. [PubMed] [Google Scholar]
- 18.Osborne RH, Hawthorne G, Lew EA, Gray LC. Quality of life assessment in the community-dwelling elderly: validation of the Assessment of Quality of Life (AQoL) instrument and comparison with the SF-36. J Clin Epidemiol. 2003;56(2):138-147. doi: 10.1016/S0895-4356(02)00601-7 [DOI] [PubMed] [Google Scholar]
- 19.Pham T, van der Heijde D, Altman RD, et al. OMERACT-OARSI initiative: Osteoarthritis Research Society International set of responder criteria for osteoarthritis clinical trials revisited. Osteoarthritis Cartilage. 2004;12(5):389-399. doi: 10.1016/j.joca.2004.02.001 [DOI] [PubMed] [Google Scholar]
- 20.Jin X, Jones G, Cicuttini F, et al. Effect of vitamin D supplementation on tibial cartilage volume and knee pain among patients with symptomatic knee osteoarthritis: a randomized clinical trial. JAMA. 2016;315(10):1005-1013. doi: 10.1001/jama.2016.1961 [DOI] [PubMed] [Google Scholar]
- 21.Altman RD, Gold GE. Atlas of individual radiographic features in osteoarthritis, revised. Osteoarthritis Cartilage. 2007;15(suppl A):A1-A56. doi: 10.1016/j.joca.2006.11.009 [DOI] [PubMed] [Google Scholar]
- 22.Bennell KL, Paterson KL, Metcalf BR, et al. Effect of intra-articular platelet-rich plasma vs placebo injection on pain and medial tibial cartilage volume in patients with knee osteoarthritis: the RESTORE randomized clinical trial. JAMA. 2021;326(20):2021-2030. doi: 10.1001/jama.2021.19415 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Cai G, Laslett LL, Aitken D, et al. Zoledronic acid plus methylprednisolone versus zoledronic acid or placebo in symptomatic knee osteoarthritis: a randomized controlled trial. Ther Adv Musculoskelet Dis. 2019;11:X19880054. doi: 10.1177/1759720X19880054 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Cai G, Aitken D, Laslett LL, et al. Effect of intravenous zoledronic acid on tibiofemoral cartilage volume among patients with knee osteoarthritis with bone marrow lesions: a randomized clinical trial. JAMA. 2020;323(15):1456-1466. doi: 10.1001/jama.2020.2938 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kelly AM. The minimum clinically significant difference in visual analogue scale pain score does not differ with severity of pain. Emerg Med J. 2001;18(3):205-207. doi: 10.1136/emj.18.3.205 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Baamer RM, Iqbal A, Lobo DN, Knaggs RD, Levy NA, Toh LS. Utility of unidimensional and functional pain assessment tools in adult postoperative patients: a systematic review. Br J Anaesth. 2022;128(5):874-888. doi: 10.1016/j.bja.2021.11.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wang Y, Hussain SM, Wluka AE, et al. Association between metformin use and disease progression in obese people with knee osteoarthritis: data from the Osteoarthritis Initiative—a prospective cohort study. Arthritis Res Ther. 2019;21(1):127. doi: 10.1186/s13075-019-1915-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Lu CH, Chung CH, Lee CH, et al. Combination COX-2 inhibitor and metformin attenuate rate of joint replacement in osteoarthritis with diabetes: a nationwide, retrospective, matched-cohort study in Taiwan. PLoS One. 2018;13(1):e0191242. doi: 10.1371/journal.pone.0191242 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Baker MC, Sheth K, Liu Y, Lu D, Lu R, Robinson WH. Development of osteoarthritis in adults with type 2 diabetes treated with metformin vs a sulfonylurea. JAMA Netw Open. 2023;6(3):e233646. doi: 10.1001/jamanetworkopen.2023.3646 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Barnett LA, Jordan KP, Edwards JJ, van der Windt DA. Does metformin protect against osteoarthritis? an electronic health record cohort study. Prim Health Care Res Dev. 2017;18(6):623-628. doi: 10.1017/S1463423617000287 [DOI] [PubMed] [Google Scholar]
- 31.Alimoradi N, Tahami M, Firouzabadi N, Haem E, Ramezani A. Metformin attenuates symptoms of osteoarthritis: role of genetic diversity of Bcl2 and CXCL16 in OA. Arthritis Res Ther. 2023;25(1):35. doi: 10.1186/s13075-023-03025-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Zhang W, Moskowitz RW, Nuki G, et al. OARSI recommendations for the management of hip and knee osteoarthritis, I: critical appraisal of existing treatment guidelines and systematic review of current research evidence. Osteoarthritis Cartilage. 2007;15(9):981-1000. doi: 10.1016/j.joca.2007.06.014 [DOI] [PubMed] [Google Scholar]
- 33.Lim YZ, Wang Y, Estee M, et al. Metformin as a potential disease-modifying drug in osteoarthritis: a systematic review of pre-clinical and human studies. Osteoarthritis Cartilage. 2022;30(11):1434-1442. doi: 10.1016/j.joca.2022.05.005 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial Protocol
Statistical Analysis Plan
eTable 1. Baseline Participants’ Characteristics Between Those Completed and Withdrew
eTable 2. Subgroup Analysis on Change in Primary Outcome (VAS Pain) Over 6 Months
eTable 3. Change in Study End Points Over 6 Months in Complete Data
eFigure 1. Changes in WOMAC Pain (A), Stiffness (B) and Function (C) in Patients Treated With Metformin vs. Placebo
eFigure 2. WOMAC Pain (A), Stiffness (B) and Function (C) in 6 Months Using Imputed Data
Data Sharing Statement