Metformin

Serious harms

Metformin is prescribed to lower blood glucose in type 2 diabetes ¹.

Condition	How often	Who is at risk	Source
Lactic acidosis; deaths reported	None in 70,490 patient-years of trials; upper limit 4.3 per 100,000	eGFR below 30: hazard ratio 2.07 among 75,413 patients	¹²³
Lactic acidosis in mitochondrial disease	Not quantified	MELAS and maternally inherited diabetes with deafness	⁴
Vitamin B12 deficiency	Up to 1 in 10 users	Higher dose, longer use	⁵
Hypoglycaemia	14% of 431 against 5% of 144 on glyburide alone	Users of insulin or an insulin secretagogue	⁴
Liver injury, three patterns	Not estimable from reports	Avoided in hepatic disease	⁴
Small-for-gestational-age infant	30 (13%) against 15 (7%), 502 pregnancies	Type 2 diabetes in pregnancy, added to insulin	⁶
Overweight or obesity at 4 years	26 (32%) against 14 (18%) of 182 children; 1 extra per 7	Mothers with polycystic ovary syndrome	⁷
Major birth defects	5.2% against 3.3% of 1451 offspring; genital defects 0.90% against 0.24%, all in boys	Father’s use during sperm development	⁸⁹

Metformin decreases liver uptake of lactate, so blood lactate rises; onset is subtle, with malaise, myalgias, somnolence and abdominal pain, and death, hypothermia and resistant bradyarrhythmias have followed ¹. The labelled risk factors are renal impairment, interacting drugs, age 65 or more, contrast imaging, surgery, hypoxic states, excess alcohol and hepatic impairment ⁴. Anaemia or neuropathy from B12 deficiency can become permanent if treatment comes late ⁵.

Other pathways it affects

Its proposed targets are housekeeping enzymes of every respiring cell ¹⁵.

Site	Finding	Shown in
Liver, kidney, bile	OCT1, OCT2, MATE1 and MATE2-K carry metformin; MATE2-K also carries creatinine and thiamine ¹⁶	Reference annotation
Every respiring tissue	Complex I inhibited only at millimolar concentrations, against 50 to 100 µM in plasma ¹⁷	Laboratory work, reviewed
Liver mitochondria	Complex IV inhibited at clinical concentrations, cutting glucose synthesis; a complex I inhibitor had no such effect ¹⁸¹⁹	Rats
Islets, brown fat, muscle, brain, testis	Mitochondrial glycerol-3-phosphate dehydrogenase is 11 to 250 times more active than in liver; knockout mice have reduced fertility ²⁰	Ten rat tissues; mice
Gut microbes and bile acids	Escherichia rose and Intestinibacter fell; three days lowered serum FGF19 ²¹²²	784 metagenomes; 22 people
Skeletal muscle	1700 mg daily during 14 weeks of resistance training blocked the gains: lean mass rose 1.95% against 0.41% ²³	94 adults over 65

What it is for and how it works

Metformin is given with diet and exercise in type 2 diabetes. It reduces the glucose the liver releases and raises uptake in tissues, leaving insulin secretion unchanged ¹. Both the 2018 and the 2026 label state that no study has established conclusive evidence of a reduction in large-vessel disease ¹⁴.

In the body

The kidneys clear metformin unchanged, at about 3.5 times the rate of creatinine, so kidney function sets both the dose and the risk: a single 850 mg dose peaks at 1.86 µg/mL at a creatinine clearance of 61 to 90 mL/min and at 4.12 µg/mL at 31 to 60 ¹. Cimetidine, ranolazine, vandetanib and dolutegravir block the same renal transporters and raise exposure ¹.

Genetic variation

Variants of the transporter OCT1 predict who tolerates the drug, not who responds. Among 2166 new users, carriers of two reduced-function alleles were more often intolerant (odds ratio 2.41), and the odds reached 4.13 when they also took a drug blocking the same transporter, such as a proton-pump inhibitor or verapamil ²⁶. Those alleles did not predict the fall in HbA1c ²⁷.

Benefit against harm

Group	Benefit	Serious harm
Type 2 diabetes, hard outcomes	None: death 0.99 (0.75 to 1.31) across 13 trials, 13,110 patients ²⁸	Lactic acidosis, rare; B12 deficiency in up to 1 in 10 ⁵
Type 2 diabetes, blood test	HbA1c -1.4% against +0.4% over 29 weeks ¹	Diarrhoea in 53% against 12%; 6% stopped ¹
Added to a sulphonylurea	None	Death 1.53 (1.02 to 2.31); cardiac death 2.20 (1.20 to 4.03) ²⁸
Coronary disease, against glipizide	Events in 25.0% against 35.1% at equal HbA1c ²⁹	As above
Breast cancer, no diabetes	None: 2.78 against 2.74 events per 100 patient-years ³⁰	Grade 3 events in 21.5% against 17.5% ³⁰

Without the medicine

Weight loss can clear the diagnosis. A programme that began by withdrawing glucose-lowering drugs put 46% of 149 adults into remission at one year against 4% on usual care, and 86% of those who lost 15 kg or more ³⁶. Thirty years after a six-year programme in Da Qing, cardiovascular deaths were a third lower ³⁷.

How the recommendations were set

Aiming lower is not always better. Targeting an HbA1c below 6.0% in 10,251 people at high cardiovascular risk raised deaths from 203 to 257 over a mean 3.5 years, one extra death per 95 treated, and that arm was stopped early ⁴⁰. Of 288 members of 14 North American diabetes and lipid guideline panels, 150 had financial ties to the makers of the drugs ⁴¹.

Stopping

The label names the moments for stopping rather than reducing the dose: an eGFR below 30, iodinated contrast, withheld food and fluid around surgery, a hypoxic state such as acute heart failure or sepsis, and any suspicion of lactic acidosis ¹⁴. European product information adds short interruptions during dehydration from vomiting, diarrhoea or fever ⁴³.

A prescription is changed together with the prescriber.

References

Regulator
US Food and Drug Administration. GLUCOPHAGE (metformin hydrochloride) tablets and GLUCOPHAGE XR extended-release tablets: full prescribing information US Food and Drug Administration, Drugs@FDA, NDA 020357 and NDA 021202, label revised May 2018, 2018.
https://www.accessdata.fda.gov/drugsatfda_docs/label/2018/020357s034,021202s018lbl.pdf
Funding and finding
Funding
US Food and Drug Administration
Finding
Indicated as an adjunct to diet and exercise to improve glycaemic control in adults and, for the immediate-release tablet, children of 10 years and older with type 2 diabetes. Mechanism of action: decreases hepatic glucose production, decreases intestinal absorption of glucose and improves insulin sensitivity by increasing peripheral glucose uptake and utilisation; insulin secretion remains unchanged. Pharmacokinetics: absolute bioavailability of a 500 mg tablet under fasting conditions about 50 to 60%; steady-state plasma concentrations generally below 1 µg/mL; apparent volume of distribution 654 ± 358 L after 850 mg; negligible plasma protein binding; excreted unchanged in urine with no hepatic metabolism, no identified human metabolites and no biliary excretion; about 90% of absorbed drug eliminated renally within 24 hours; plasma elimination half-life about 6.2 hours and blood half-life about 17.6 hours; renal clearance about 3.5 times creatinine clearance, so tubular secretion is the major route. Single 850 mg dose: peak concentration 1.86 ± 0.52 µg/mL and renal clearance 384 ± 122 mL/min with creatinine clearance 61 to 90 mL/min, against 4.12 ± 1.83 µg/mL and 108 ± 57 mL/min at 31 to 60 mL/min and 3.93 ± 0.92 µg/mL and 130 ± 90 mL/min at 10 to 30 mL/min. Contraindicated at eGFR below 30 mL/min/1.73 m2; initiation not recommended at 30 to 45; benefit and risk assessed if eGFR falls below 45; discontinue below 30. Discontinue at or before iodinated contrast imaging when eGFR is 30 to 60 mL/min/1.73 m2, in liver disease, alcoholism or heart failure, or with intra-arterial contrast, and re-evaluate eGFR after 48 hours. Boxed warning: postmarketing metformin-associated lactic acidosis, characterised by blood lactate above 5 mmol/L, anion gap acidosis without ketonuria or ketonaemia, raised lactate to pyruvate ratio and plasma metformin generally above 5 µg/mL; metformin decreases liver uptake of lactate, raising blood lactate; prompt haemodialysis recommended, metformin being dialysable with a clearance up to 170 mL/min. Lactic acidosis reported in about 32% of overdose cases. Vitamin B12: in 29-week trials a decrease to subnormal levels of previously normal serum vitamin B12 occurred in about 7% of patients, possibly through interference with B12 absorption from the B12-intrinsic factor complex, may be associated with anaemia, and appears rapidly reversible on discontinuation or B12 supplementation; measure haematological parameters annually and B12 every 2 to 3 years. Drug interactions: inhibitors of organic cation transporter 2 (OCT2) and multidrug and toxin extrusion (MATE), among them ranolazine, vandetanib, dolutegravir and cimetidine, may increase systemic exposure; single-dose cimetidine 400 mg with metformin 850 mg gave geometric mean ratios of 1.40 for AUC and 1.61 for peak concentration; carbonic anhydrase inhibitors such as topiramate may increase the risk of lactic acidosis. No clinical study has established conclusive evidence of macrovascular risk reduction. In a 29-week double-blind placebo-controlled US trial in obese patients with type 2 diabetes, fasting plasma glucose changed by −53.0 mg/dL with metformin up to 2550 mg daily (n = 141) and +6.3 mg/dL with placebo (n = 145), and HbA1c by −1.4% and +0.4% (both P = 0.001); mean weight change was −1.4 lb and −2.4 lb. Adverse reactions with metformin against placebo: diarrhoea 53% against 12%, nausea or vomiting 26% against 8%, flatulence 12% against 6%; diarrhoea led to discontinuation in 6%.
Independent research
Salpeter SR, Greyber E, Pasternak GA, Salpeter EE. Risk of fatal and nonfatal lactic acidosis with metformin use in type 2 diabetes mellitus Cochrane Database of Systematic Reviews, 2010.
doi:10.1002/14651858.CD002967.pub4 PubMed 20393934
Funding and finding
Participants
96,295
Follow-up
125,941 patient-years in all; mean study duration 1.3 years, range 1 month to 10.7 years
Funding
Internal sources: Santa Clara Valley Medical Center, USA. External sources: no sources of support supplied. Declarations of interest: none known.
Finding
No case of fatal or nonfatal lactic acidosis occurred in 69,642 participants and 70,490 patient-years of metformin use, or in 26,653 participants and 55,451 patient-years without metformin. By Poisson statistics the upper limit for the true incidence per 100,000 patient-years was 4.3 cases with metformin and 5.4 without. Of the 334 prospective studies, 143 (53%) admitted patients with renal insufficiency, defined as creatinine above 1.5 mg/dl, covering 37,360 patient-years of metformin use, and 324 (97%) admitted patients with at least one standard contraindication; an estimated 26% of participants were older than 65 years. Baseline lactate before metformin was 1.13 ± 0.25 mmol/L, and lactate did not differ between metformin and comparison treatments as mean values or as net change from baseline. Bristol-Myers Squibb, the manufacturer, reported no unpublished trials. The authors state that the number of participants with each contraindication is unclear, so safety in their presence cannot be established from these data.
Independent research
Lazarus B, Wu A, Shin JI, Sang Y, Alexander GC, Secora A, et al. Association of Metformin Use With Risk of Lactic Acidosis Across the Range of Kidney Function: A Community-Based Cohort Study JAMA Internal Medicine, 2018.
doi:10.1001/jamainternmed.2018.0292 PubMed 29868840
Funding and finding
Participants
75,413
Follow-up
median 5.7 years (interquartile range 2.5 to 9.9)
Funding
Ms Sang and Dr Coresh receive funding from the National Institutes of Health via grant R01DK100446; Dr Grams via R01DK100446 and K08DK092287. The National Institutes of Health had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Finding
Mean age 60.4 years (SD 15.5), 51% women. There were 2335 hospitalisations with acidosis over a median 5.7 years. Against alternative diabetes management, time-dependent metformin use was not associated with incident acidosis overall (adjusted hazard ratio 0.98, 95% CI 0.89 to 1.08), at eGFR 45 to 59 mL/min/1.73 m2 (1.16, 0.95 to 1.41) or at eGFR 30 to 44 (1.09, 0.83 to 1.44), but was associated with an increased risk at eGFR below 30 (2.07, 1.33 to 3.22). Results were consistent against new sulfonylurea users (hazard ratio at eGFR 30 to 44, 0.77, 0.29 to 2.05), in a propensity-matched cohort (0.71, 0.45 to 1.12), after excluding baseline insulin users (1.16, 0.87 to 1.57) and in the replication cohort (0.86, 0.37 to 2.01). The authors state that about 1 million patients in the United States with type 2 diabetes and mild-to-moderate kidney disease do not receive metformin, and conclude that their results support cautious use at an eGFR of at least 30 mL/min/1.73 m2.
Regulator
US Food and Drug Administration. GLUMETZA (metformin hydrochloride) extended-release tablets: full prescribing information US Food and Drug Administration, Drugs@FDA, NDA 021748 supplement 31, label revised August 2026, 2026.
https://www.accessdata.fda.gov/drugsatfda_docs/label/2026/021748s031lbl.pdf
Funding and finding
Funding
US Food and Drug Administration
Finding
Label revised 08/2026; Recent Major Changes list the Boxed Warning and Warnings and Precautions 5.1 (Lactic Acidosis). The approval letter of 11 August 2026 states that FDA notified the applicant on 31 March 2026, under section 505(o)(4) of the Federal Food, Drug, and Cosmetic Act, of new safety information that should be included in the labeling of metformin-containing products, concerning the risk of lactic acidosis in patients with mitochondrial diseases such as mitochondrial encephalopathy with lactic acidosis and stroke-like episodes (MELAS) syndrome and maternally inherited diabetes and deafness (MIDD) who were taking metformin products. Boxed warning: postmarketing cases of metformin-associated lactic acidosis have resulted in death, hypothermia, hypotension and resistant bradyarrhythmias, with blood lactate above 5 mmol/L, anion gap acidosis without ketonuria or ketonaemia, raised lactate to pyruvate ratio and plasma metformin generally above 5 mcg/mL; risk factors are renal impairment, concomitant use of certain drugs such as carbonic anhydrase inhibitors, age 65 years or greater, radiological studies with contrast, surgery and other procedures, hypoxic states such as acute congestive heart failure, excessive alcohol intake, hepatic impairment and mitochondrial diseases. Section 5.1 adds: metformin may increase the risk of lactic acidosis in patients with mitochondrial diseases such as MELAS and MIDD and is not recommended in them; if a patient develops elevated lactic acid levels, with or without neurologic signs and symptoms, discontinue the drug and consider evaluating the patient for MELAS, MIDD or other mitochondrial disease. Section 5.2: in 29-week trials of metformin tablets a decrease to subnormal levels of previously normal serum vitamin B12 occurred in about 7% of patients, possibly through interference with B12 absorption from the B12-intrinsic factor complex, may be associated with anaemia, appears rapidly reversible on discontinuation or supplementation; measure haematologic parameters annually and vitamin B12 every 2 to 3 years. Section 5.3: increased risk of hypoglycaemia with insulin or an insulin secretagogue. Section 5.4: there have been no clinical studies establishing conclusive evidence of macrovascular risk reduction. Section 6.1, add-on to glyburide: hypoglycaemia in 14% of 431 patients on GLUMETZA plus glyburide against 5% of 144 on placebo plus glyburide, diarrhoea 13% against 6%, nausea 7% against 4%; diarrhoea caused discontinuation in 0.7% against none. Section 6.2 postmarketing: cholestatic, hepatocellular and mixed hepatocellular liver injury.
Regulator
Medicines and Healthcare products Regulatory Agency. Metformin and reduced vitamin B12 levels: new advice for monitoring patients at risk Drug Safety Update volume 15, issue 11, June 2022: 1, published 20 June 2022, 2022.
https://www.gov.uk/drug-safety-update/metformin-and-reduced-vitamin-b12-levels-new-advice-for-monitoring-patients-at-risk
Funding and finding
Funding
Medicines and Healthcare products Regulatory Agency
Finding
Decreased vitamin B12 levels, or vitamin B12 deficiency, is now considered a common adverse drug reaction of metformin. After a European review of the brand leader Glucophage with MHRA input, the product information was updated to state that vitamin B12 deficiency is common and may affect up to 1 in 10 people who take it, and that the risk increases with higher dose, longer treatment duration and pre-existing risk factors. The agency advises testing serum vitamin B12 when deficiency is suspected, for example in patients presenting with megaloblastic anaemia or new-onset neuropathy, considering periodic monitoring in patients with risk factors, and continuing metformin while the deficiency is corrected. The mechanism is described as multifactorial, comprising altered intestinal motility, bacterial overgrowth and reduced uptake of vitamin B12 in the small intestine. Patients with deficiency can be asymptomatic or present with megaloblastic anaemia, neuropathy or both; other features include depression, irritability, cognitive impairment, glossitis, mouth ulcers and visual and motor disturbances, and the statement notes that anaemia or neuropathy caused by deficiency should be diagnosed and treated as soon as possible to avoid permanent symptoms. Listed risk factors: baseline B12 at the lower end of the normal range; conditions reducing absorption, including older age, total or partial gastrectomy, Crohn's disease and other inflammatory bowel disorders and autoimmune conditions; diets low in B12 such as strict vegan and some vegetarian diets; concomitant medicines impairing B12 absorption, including proton pump inhibitors and colchicine; and genetic predisposition such as Imerslund-Grasbeck syndrome and transcobalamin II deficiency. The four cited references are Aroda 2016, Beulens 2015, de Jager 2010 and Miller 2018.
Independent research
Feig DS, Donovan LE, Zinman B, Sanchez JJ, Asztalos E, Ryan EA, et al. Metformin in women with type 2 diabetes in pregnancy (MiTy): a multicentre, international, randomised, placebo-controlled trial Lancet Diabetes and Endocrinology, 2020.
doi:10.1016/S2213-8587(20)30310-7 PubMed 32946820
Funding and finding
Registration
NCT01353391
Participants
502
Follow-up
from randomisation between 6 and 22 weeks' gestation to delivery
Funding
Canadian Institutes of Health Research, Lunenfeld-Tanenbaum Research Institute, University of Toronto. The ClinicalTrials.gov record for NCT01353391 names Mount Sinai Hospital, Canada as lead sponsor, with University Health Network Toronto, Sunnybrook Research Institute and the Canadian Institutes of Health Research as collaborators.
Finding
Metformin 1000 mg twice daily or placebo was added to insulin. The primary composite of fetal and neonatal outcomes did not differ (40% against 40%, relative risk 1.02, 95% CI 0.83 to 1.26, p = 0.86). Metformin-treated women had lower HbA1c at 34 weeks (41.0 mmol/mol, SD 8.5, against 43.2, SD 10; 5.90% against 6.10%, p = 0.015), needed less insulin (1.1 against 1.5 units per kg per day, difference -0.4, -0.5 to -0.2, p < 0.0001), gained less weight (7.2 kg against 9.0 kg, difference -1.8, -2.7 to -0.9, p < 0.0001) and had fewer caesarean births (125 of 234, 53%, against 148 of 236, 63%; relative risk 0.85, 0.73 to 0.99, p = 0.031); hypertensive disorders did not differ (23% against 23%). Metformin-exposed infants weighed less (mean birthweight 3156 g, SD 742, against 3375 g, SD 742; difference -218 g, -353 to -82, p = 0.002), fewer were above the 97th centile for birthweight (20, 9%, against 34, 15%; relative risk 0.58, 0.34 to 0.97, p = 0.041) and fewer weighed 4000 g or more (28, 12%, against 44, 19%; 0.65, 0.43 to 0.99, p = 0.046), with lower adiposity measures. Small-for-gestational-age infants numbered 30 (13%) with metformin against 15 (7%) with placebo, relative risk 1.96 (1.10 to 3.64, p = 0.026). Cord C-peptide did not differ. The most common adverse events were gastrointestinal, 38 in each group. The authors conclude that the lower adiposity and infant size measurements produced fewer large infants but a higher proportion of small-for-gestational-age infants, and that understanding the implications of these effects on infants will be important in advising patients considering metformin in pregnancy.
Independent research
Hanem LGE, Stridsklev S, Juliusson PB, Salvesen O, Roelants M, Carlsen SM, et al. Metformin Use in PCOS Pregnancies Increases the Risk of Offspring Overweight at 4 Years of Age: Follow-Up of Two RCTs Journal of Clinical Endocrinology and Metabolism, 2018.
doi:10.1210/jc.2017-02419 PubMed 29490031
Funding and finding
Participants
182
Follow-up
anthropometry at birth, 6 months, 1, 1.5, 2 and 4 years of age
Funding
This work was supported by the Research Council of Norway, which funded the follow-up study (registration number 239987), the NovoNordisk Foundation and Felles Forskningsutvalg.
Finding
Mothers had received metformin 1700 or 2000 mg daily or placebo from the first trimester to delivery in the original trials; there was no intervention in the follow-up. At 4 years of age the difference in height z score between groups was not significant (0.07, 95% CI -0.22 to 0.36, p = 0.651), while the metformin group had a higher weight z score (difference in means 0.38, 0.07 to 0.69, p = 0.017) and a higher body-mass index z score (0.45, 0.11 to 0.78, p = 0.010). More children were overweight or obese in the metformin group, 26 (32%) against 14 (18%) in the placebo group, odds ratio 2.17 (1.04 to 4.61, p = 0.038); the number needed to harm for an additional case of overweight or obesity at 4 years was 7. The difference in mean head circumference z score at 1 year was 0.27 (-0.04 to 0.58, p = 0.093). Data from all 182 children entered the mixed linear models, though the main outcomes at 4 years were missing for 21 children and head circumference at 1 year for 28. The authors note that participants were representative of women with a known diagnosis of polycystic ovary syndrome and that the results might not translate to mothers without the syndrome.
Independent research
Wensink MJ, Lu Y, Tian L, Shaw GM, Rizzi S, Jensen TK, et al. Preconception Antidiabetic Drugs in Men and Birth Defects in Offspring: A Nationwide Cohort Study Annals of Internal Medicine, 2022.
doi:10.7326/M21-4389 PubMed 35344380
Funding and finding
Participants
1,116,779
Follow-up
one year from birth for the diagnosis of birth defects
Funding
Primary funding source: National Institutes of Health. Grant support from NIH grants HD096468 and 1UL1TR003142 and from Centers for Disease Control and Prevention grant U01DD001226. The funders had no role in the analysis, decision to publish, or writing of the manuscript.
Finding
Major birth defects, classified by EUROCAT guidelines, affected 36,585 of 1,116,779 offspring (3.3%), and 51.4% of offspring were male. Insulin-exposed offspring had the reference frequency of defects (3.3%; adjusted odds ratio 0.98, 95% CI 0.85 to 1.14). Metformin-exposed offspring had 5.2% defects (adjusted odds ratio 1.40, 1.08 to 1.82, p = 0.012) and sulfonylurea-exposed offspring 5.1% (1.34, 0.94 to 1.92, p = 0.107). Offspring whose fathers filled a metformin prescription in the year before sperm development (n = 1751) or in the year after (n = 2484) had reference frequencies (0.88, 0.59 to 1.31 and 0.92, 0.68 to 1.26), and unexposed siblings of exposed offspring had 3.2% defects, with a within-father odds ratio of 1.54 (0.94 to 2.53) and a within-mother odds ratio of 1.66 (1.00 to 2.75). Among metformin-exposed offspring, genital birth defects were raised, 0.90% against 0.24% (adjusted odds ratio 3.39, 1.82 to 6.30); all occurred in boys, each from a different father, and genital defects were not raised with sulfonylureas (0.96). The proportion of male offspring was 49.4% against 51.4% (p = 0.073). The E-value was 2.15 to move the point estimate to the null and 1.37 to move the lower confidence limit to the null, so the authors state that unmeasured confounding of that strength would be needed to explain the association away, that the statistical significance may disappear after adjustment for unobserved confounding, and that a documented preconception diagnosis of type 2 diabetes gave the baseline defect frequency (3.1%) when the father was not receiving metformin and an elevated frequency (4.6%) when he was. They further report an elevated odds ratio for birth defects with paternal beta-blockers (1.25, 1.06 to 1.47).
Independent research
Meng LC, van Gelder MMHJ, Chuang HM, Chen LK, Hsiao FY, Nordeng HME. Paternal metformin use and risk of congenital malformations in offspring in Norway and Taiwan: population based, cross national cohort study BMJ, 2024.
doi:10.1136/bmj-2024-080127 PubMed 39414354
Funding and finding
Participants
3,183,201
Follow-up
congenital malformations diagnosed up to one year after birth
Funding
This study was supported by the International Alliance for PharmacoGenetic Epidemiology Excellence project funded by the Norwegian Research Council (grant 322176) and the National Science and Technology Council of Taiwan (NSTC 113-2314-B-002-167-MY3). The funders had no role in the study design; the collection, analysis, and interpretation of data; the writing of the report; or the decision to submit the article for publication.
Finding
Congenital malformations, classified by EUROCAT guidelines, affected 104 of 2075 offspring of metformin-using fathers in Norway (5.0%) against 24,041 of the remainder (3.9%), and 512 of 15,276 in Taiwan (3.4%) against 79,278 (3.1%). Unadjusted relative risks were 1.29 (95% CI 1.07 to 1.55) in Norway and 1.08 (0.99 to 1.17) in Taiwan. Restricting the cohort to fathers with diagnosed type 2 diabetes gave 1.20 (0.94 to 1.53) and 0.93 (0.80 to 1.07); adding overlap propensity score weighting for diabetes severity and other confounders gave 0.98 (0.72 to 1.33) and 0.87 (0.74 to 1.02), pooled 0.89 (0.77 to 1.03). No association was observed with any organ-specific malformation. Sibling-matched comparisons, from 76 pairs in Norway and 581 in Taiwan, gave adjusted odds ratios of 0.83 (0.43 to 1.59) and 0.84 (0.68 to 1.04). The authors conclude that paternal metformin use during sperm development is not associated with congenital malformations in offspring and that metformin can continue to be considered a suitable initial oral agent for men with type 2 diabetes who plan on having children.
Industry-funded
de Jager J, Kooy A, Lehert P, Wulffele MG, van der Kolk J, Bets D, et al. Long term treatment with metformin in patients with type 2 diabetes and risk of vitamin B-12 deficiency: randomised placebo controlled trial BMJ, 2010.
doi:10.1136/bmj.c2181 PubMed 20488910
Funding and finding
Registration
NCT00375388
Participants
390
Follow-up
4.3 years (52 months); samples at baseline and 4, 17, 30, 43 and 52 months
Funding
The Hyperinsulinaemia: the Outcome of its Metabolic Effects (HOME) trial was supported by grants from Altana, Lifescan, Merck Sante, Merck Sharp & Dohme and Novo Nordisk. The sponsors had no role in the design and conduct of the study; in the collection, analysis and interpretation of the data; or in the preparation, review or approval of the manuscript. All authors declare financial support for the submitted work from Merck Sharp & Dohme.
Finding
Randomisation to metformin 850 mg three times daily (n = 196) or identical placebo (n = 194) added to insulin; mean achieved dose 2050 mg daily; 277 of 390 (72%) were still taking study drug at the end, and 46 patients (30 on metformin, 16 on placebo) discontinued because of adverse effects. Against placebo, metformin lowered vitamin B12 by 19% (95% CI -24% to -14%, p < 0.001) and folate by 5% (-10% to -0.4%, p = 0.033; not significant after adjustment for body-mass index and smoking) and raised homocysteine by 5% (-1% to 11%, p = 0.091). Vitamin B12 fell by 89.8 pmol/L from baseline on metformin and by 0.2 pmol/L on placebo, and the interaction between treatment and time was significant (p = 0.023), the fall increasing with time. At study end, deficiency (below 150 pmol/L) affected 19 patients (9.9%) on metformin against 5 (2.7%) on placebo, an absolute risk difference of 7.2 percentage points (2.3 to 12.1, p = 0.004) and a number needed to harm of 13.8 per 4.3 years (43.5 to 8.3); low concentrations (150 to 220 pmol/L) affected 35 (18.2%) against 13 (7.0%), a difference of 11.2 percentage points (4.6 to 17.9, p = 0.001) and a number needed to harm of 8.9 per 4.3 years (21.7 to 5.6). The hazard ratio for developing deficiency was 5.5 (1.6 to 19.1, p = 0.01) and for a low concentration 3.0 (1.3 to 6.6, p = 0.007). Patients deficient at study end had a mean homocysteine of 23.7 micromol/L (18.8 to 30.0) against 18.1 (16.7 to 19.6) with low B12 and 14.9 (14.3 to 15.5) with normal B12. The authors state that they measured only total vitamin B12, not holotranscobalamin or methylmalonic acid; that all participants received repeated dietary counselling, which may have attenuated the effect; that eight patients deficient at baseline or interim analysis were supplemented and excluded; and that no additional data are available.
Industry-funded
Out M, Kooy A, Lehert P, Schalkwijk CA, Stehouwer CDA. Long-term treatment with metformin in type 2 diabetes and methylmalonic acid: Post hoc analysis of a randomized controlled 4.3 year trial Journal of Diabetes and its Complications, 2018.
doi:10.1016/j.jdiacomp.2017.11.001 PubMed 29174300
Funding and finding
Registration
NCT00375388
Participants
390
Follow-up
52 months
Funding
Funders recorded by the publisher in the Crossref metadata for this article are Merck Sharp & Dohme and Novo Nordisk; the parent HOME trial was supported by grants from Altana, Lifescan, Merck Sante, Merck Sharp & Dohme and Novo Nordisk. The article's own text is behind a subscription and absent from PubMed Central, so the role of the funders in this analysis could not be read.
Finding
Results read from the published abstract. Methylmalonic acid, a biomarker of tissue vitamin B12 deficiency, was higher with metformin than placebo at the end of the study (95% CI of the difference 0.019 to 0.055, p = 0.001), and the rise was progressive. In a mediation analysis the effect of metformin on a validated Neuropathy Score comprised a beneficial component through lower HbA1c (-0.020 per gram year) and an adverse component through higher methylmalonic acid (0.042 per gram year), giving a non-significant net effect (0.032 per gram year, 95% CI -0.121 to 0.182, p = 0.34); the rise in methylmalonic acid in metformin users was associated with significant worsening of the Neuropathy Score. The authors conclude that metformin lowers serum B12 and progressively raises serum methylmalonic acid, that this supports the clinical relevance of metformin-related B12 deficiency, and that monitoring of B12 in metformin users should be considered.
Industry-funded
Eppenga WL, Lalmohamed A, Geerts AF, Derijks HJ, Wensing M, Egberts A, et al. Risk of lactic acidosis or elevated lactate concentrations in metformin users with renal impairment: a population-based cohort study Diabetes Care, 2014.
doi:10.2337/dc13-3023 PubMed 24842984
Funding and finding
Participants
258,539
Follow-up
mean 4.3 years in metformin users and 4.9 years in non-users
Funding
Funding and Duality of Interest: the Division of Pharmacoepidemiology and Clinical Pharmacology, Utrecht Institute for Pharmaceutical Sciences, employing five of the authors, has received unrestricted research funding from the Netherlands Organisation for Health Research and Development, the Dutch Health Care Insurance Board, the Royal Dutch Pharmacists Association, the private-public funded Top Institute Pharma, the EU Innovative Medicines Initiative, EU 7th Framework Program, the Dutch Medicines Evaluation Board, the Dutch Ministry of Health, and industry (including GlaxoSmithKline, Pfizer, and others). No study-specific funding is stated and no separation of this study from the industry funding of the employing division is given, so the record is classed industry. No other potential conflicts of interest were reported.
Finding
The outcome was lactic acidosis by clinical code or a recorded plasma lactate above 5 mmol/L. Of 68 events, 50 (74%) came from a code for lactic acidosis and 18 (26%) from a laboratory value. The crude incidence was 7.4 events per 100,000 person-years in current metformin users against 2.2 per 100,000 person-years in non-users; the adjusted hazard ratio for current use overall was 4.03 (95% CI 0.97 to 16.8). Among current users, risk was significantly raised at a most recent renal function below 60 mL/min/1.73 m2 (adjusted hazard ratio 6.37, 1.48 to 27.5) but not at 60 or above (2.87, 0.67 to 12.3); using a cut-point of 45 the ratios were 6.74 (1.34 to 33.8) below and 3.16 (0.75 to 13.3) above. Risk in impaired renal function rose further with a cumulative dose of 730 g or more in the preceding year (11.8, 2.27 to 61.5) and with a recent daily dose above 2 g (13.0, 2.36 to 72.0). Renal function data were missing for 27.4% of current users and 34.6% of non-users; the proportion with eGFR below 30 was 0.3% of metformin users against 2.8% of non-users, and heart failure was less prevalent in metformin users (3.9% against 8.3%). The authors state that including patients with lactate above 5 mmol/L may overestimate lactic acidosis and that acute falls in GFR could not be analysed.
Funding unknown
Inzucchi SE, Lipska KJ, Mayo H, Bailey CJ, McGuire DK. Metformin in patients with type 2 diabetes and kidney disease: a systematic review JAMA, 2014.
doi:10.1001/jama.2014.15298 PubMed 25536258
Funding and finding
Funding
No funding statement appears in the PubMed Central author manuscript, and the publisher page is behind a subscription, so study funding could not be read and the class is undetermined.
Finding
Narrative synthesis. The overall incidence of lactic acidosis in metformin users varied across studies from about 3 to 10 per 100,000 person-years and was generally indistinguishable from the background rate in diabetes; estimates cited are about 1 per 23,000 to 30,000 person-years with metformin against 1 per 18,000 to 21,000 person-years with other agents. Metformin clearance falls by 23% to 33% in mild chronic kidney disease (creatinine clearance 60 to 90 mL/min) and by 74% to 78% in moderate disease (30 to 60 mL/min), yet median trough concentrations remained within the therapeutic range of about 4 to 20 micromol/L: 4.50 micromol/L above eGFR 60, 7.71 at eGFR 30 to 60 and 8.88 below 30. In a UK general-practice database, incidence among metformin users was 10.37 per 100,000 patient-years overall and, by kidney function, 7.6 above eGFR 90, 4.6 at 60 to 90, 17 at 30 to 60 and 39 below 30, with incidence rate ratios against normal function of 0.61, 2.27 and 5.26, none significant. Unlike metformin, phenformin is lipophilic, partly hepatically metabolised, has a half-life of 7 to 15 hours, increases muscle lactate release and inhibits lactate oxidation; lactic acidosis fell in Swedish patients from 1.5 cases per 10,000 patient-years when phenformin was available to 0.24 per 10,000 patient-years when metformin was the only biguanide. The review states that no randomised controlled trial has tested the safety of metformin in significantly impaired kidney function, that observational studies may underestimate risk through confounding by indication or overestimate it through ascertainment bias, that prevailing US guidelines prohibited metformin for at least 2.5 million people, and that up to 1 in 4 patients received metformin counter to those guidelines without higher rates of lactic acidosis.
Industry-funded
Yang W, Cai X, Wu H, Ji L. Associations between metformin use and vitamin B12 levels, anemia, and neuropathy in patients with diabetes: a meta-analysis Journal of Diabetes, 2019.
doi:10.1111/1753-0407.12900 PubMed 30615306
Funding and finding
Funding
No funding statement is given in the record indexed by PubMed or in the Crossref metadata, and the article is behind a subscription and absent from PubMed Central, so study funding could not be read. One of the four authors is affiliated with Eisai China Inc., Shanghai, so an industry role in the work and its publication is present and the record is classed industry.
Finding
Results read from the published abstract. Compared with patients with diabetes not taking metformin, those taking it had a higher risk of vitamin B12 deficiency (risk ratio 2.09, 95% CI 1.49 to 2.93, p < 0.0001, I2 = 64%), lower serum vitamin B12 (mean difference -63.70 pmol/L, -74.35 to -53.05, p < 0.00001, I2 = 87%) and a greater percentage fall from baseline (-14.68%, -17.98% to -11.39%, p < 0.00001, I2 = 33%), with the size of the effect depending on dose and duration of treatment. The analyses showed no significant association between metformin use and the prevalence of anaemia or of neuropathy. The authors recommend annual assessment of vitamin B12 in patients with diabetes taking metformin and state that more quality studies are needed on anaemia and neuropathy.
Textbook or database
UniProt Consortium. UniProt: the Universal Protein Knowledgebase in 2025 Nucleic Acids Research, 2025.
doi:10.1093/nar/gkae1010 PubMed 39552041
Funding and finding
Funding
National Human Genome Research Institute (NHGRI) [HG002273]; Office of the Director, NIH OD [U24HG007822]; National Institute of Allergy and Infectious Diseases (NIAID); National Institute on Aging (NIA); National Institute of General Medical Sciences (NIGMS) [R35GM141873]; National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK); National Eye Institute (NEI); National Cancer Institute (NCI); National Heart, Lung, and Blood Institute (NHLBI); Biotechnology and Biological Sciences Research Council (BBSRC) [BB/T015608/1]; National Science Foundation's Directorate for Biological Sciences [BB/X002179/1]; Open Targets; Horizon 2020 - Research and Innovation Framework Programme [825575]; State Secretariat for Education, Research and Innovation (SERI); European Molecular Biology Laboratory (EMBL) Australia. Funding for open access charge: National Institutes of Health.
Finding
Reviewed human UniProtKB entries, release 2026_03, read through rest.uniprot.org on 2026-10-21. NAMPT (P43490) condenses nicotinamide with 5-phosphoribosyl-1-pyrophosphate to nicotinamide mononucleotide, the rate-limiting step of mammalian NAD biosynthesis, and NMNAT1 (Q9HAN9) forms NAD+ from nicotinamide mononucleotide and ATP. RFK (Q969G6) phosphorylates riboflavin (vitamin B2) to FMN, the rate-limiting step of FAD synthesis, and FLAD1 (Q8NFF5) adenylates FMN to FAD. SDHA (P31040), the flavoprotein of succinate dehydrogenase (complex II), binds FAD through tele-8alpha-FAD histidine 99 and transfers electrons from succinate to ubiquinone. NDUFV1 (P49821) of complex I accepts electrons from NADH on FMN and passes them along iron-sulfur clusters to ubiquinone; complex I reaction: ubiquinone + NADH + 5 H+(in) = ubiquinol + NAD+ + 4 H+(out). In the Q cycle of complex III, which contains the two haem b groups of MT-CYB (P00156), the [2Fe-2S] cluster of UQCRFS1 (P47985) and the covalent haem c of CYC1 (P08574), 2 protons are consumed from the matrix, 4 are released into the intermembrane space and 2 electrons pass to cytochrome c. CYCS (P99999), in the intermembrane space, carries electrons from cytochrome c1 to cytochrome oxidase. In complex IV, electrons pass through CuA of MT-CO2 (P00403) and haem a of MT-CO1 (P00395) to the haem a3-CuB centre; 4 Fe(II)-cytochrome c + O2 + 8 H+(in) = 4 Fe(III)-cytochrome c + 2 H2O + 4 H+(out). The respiratory chain creates an electrochemical gradient over the inner membrane that drives the ATP synthase, which couples ATP synthesis by rotation of its central stalk to proton translocation (ATP5F1B, P06576); its c-ring consists of eight c subunits (ATP5MC1, P05496).
Textbook or database
UniProt Consortium. UniProt: the Universal Protein Knowledgebase in 2025 Nucleic Acids Research, 2025.
doi:10.1093/nar/gkae1010 PubMed 39552041
Funding and finding
Funding
National Human Genome Research Institute (NHGRI) [HG002273]; Office of the Director, NIH OD [U24HG007822]; NIAID; NIA; NIGMS [R35GM141873]; NIDDK; NEI; NCI; NHLBI; Biotechnology and Biological Sciences Research Council (BBSRC) [BB/T015608/1]; National Science Foundation's Directorate for Biological Sciences [BB/X002179/1]; Open Targets; Horizon 2020 [825575]; State Secretariat for Education, Research and Innovation (SERI); EMBL Australia. Funding for open access charge: National Institutes of Health. Conflict of interest statement: none declared.
Finding
Reviewed human entries read 2026-10-21 through rest.uniprot.org. SLC22A1 (O15245, OCT1) is an electrogenic voltage-dependent uniporter of organic cations, including cationic drugs, that is primarily expressed at the basolateral membrane of hepatocytes and proximal tubules and takes up and disposes of cationic compounds by hepatic and renal clearance from the blood. SLC22A2 (O15244, OCT2) is a related Na+-independent bidirectional uniporter, mainly expressed in the kidney cortex and medulla; its isoform 2, unlike isoform 1, cannot transport guanidine, creatinine, cimetidine or metformin. SLC47A1 (Q96FL8, MATE1) is a multidrug efflux pump acting as an H+/organic cation antiporter at the apical cell membrane, which excretes cationic compounds through kidney and liver into urine and bile respectively. SLC47A2 (Q86VL8, MATE2-K) is an apical H+/organic cation antiporter expressed mainly in kidney that effluxes cationic compounds, cimetidine, cisplatin, oxaliplatin and the antidiabetic drug metformin among them, as well as creatinine, thiamine and estrone-3-sulfate. GPD1 (P21695) is the cytoplasmic NAD-dependent glycerol-3-phosphate dehydrogenase, EC 1.1.1.8, catalysing sn-glycerol 3-phosphate + NAD+ = dihydroxyacetone phosphate + NADH + H+. GPD2 (P43304) is the mitochondrial glycerol-3-phosphate dehydrogenase, EC 1.1.5.3, catalysing a quinone + sn-glycerol 3-phosphate = dihydroxyacetone phosphate + a quinol, and is calcium-responsive.
Independent research
Petersen MC, Vatner DF, Shulman GI. Regulation of hepatic glucose metabolism in health and disease Nature Reviews Endocrinology, 2017.
doi:10.1038/nrendo.2017.80 PubMed 28731034
Funding and finding
Funding
M.C.P. acknowledges grant support from the US National Institutes of Health (NIH; grants F30 DK-104596 and T32 GM-007205). D.F.V. acknowledges grant support from the NIH (grant K23 DK-102874). G.I.S. acknowledges grant support from the NIH (grants R01 DK40936, R01 DK113984 and P30 DK045735).
Finding
Narrative review. Glucagon receptor (GCGR) activation raises cAMP and activates PKA, which inhibits L-type pyruvate kinase, phosphorylates PFK-2/FBPase-2 to favour its phosphatase activity and lower fructose 2,6-bisphosphate, an allosteric inhibitor of FBPase-1 (printed as fructose-2,6-bisphosphatase 1), phosphorylates and inactivates glycogen synthase, and activates phosphorylase kinase, which phosphorylates glycogen phosphorylase at Ser15; PKA also activates the CREB-CBP-CRTC2 complex for gluconeogenic gene transcription. FOXO proteins with PGC1-alpha raise PCK1 and G6PC expression and are excluded from the nucleus after AKT phosphorylation. Glucose enters hepatocytes by facilitated diffusion through GLUT2; glucose allosterically inhibits glycogen phosphorylase; glucose 6-phosphate is the key allosteric activator of GYS2. In overnight-fasted humans under somatostatin, hyperinsulinaemia was required to stimulate hepatic glycogen synthesis and hyperglycaemia to suppress glycogenolysis. Hepatic glucose production and storage are dysregulated in type 1 and type 2 diabetes mellitus, and this imbalance contributes to hyperglycaemia in the fasted and postprandial states. Metformin is used as first-line therapy for type 2 diabetes and acts primarily through the suppression of hepatic gluconeogenesis, by a molecular mechanism that remains under investigation. Several lines of evidence cast doubt on the necessity of AMPK: metformin continues to inhibit glucose production in Ampk-knockout primary mouse hepatocytes and improves glucose tolerance in Ampk-knockout mice to a similar extent as in wild-type mice. Inhibition of mitochondrial complex I is observed only at suprapharmacological, millimolar concentrations, in contrast to the clinically relevant plasma concentration range of 50 to 100 µM in patients who take 2 g of metformin daily. Metformin non-competitively inhibits mitochondrial glycerol-3-phosphate dehydrogenase in isolated hepatocytes with a clinically relevant Ki of about 50 µM, which predicts impaired gluconeogenesis only from the redox-dependent substrates lactate and glycerol and not from pyruvate, alanine or dihydroxyacetone phosphate; the importance of the glycerol-3-phosphate shuttle in gluconeogenesis in the human liver is unclear, so studies in humans are required.
Independent research
LaMoia TE, Butrico GM, Kalpage HA, Goedeke L, Hubbard BT, Vatner DF, et al. Metformin, phenformin, and galegine inhibit complex IV activity and reduce glycerol-derived gluconeogenesis Proceedings of the National Academy of Sciences, 2022.
doi:10.1073/pnas.2122287119 PubMed 35238637
Funding and finding
Funding
This study was supported by US Public Health Service Grants R01 DK116774, R01 DK119968, R01 DK113984, R01 DK114793, P30 DK045735, F31 DK126362, T32 GM007324 and K99 HL150234, recorded by the publisher as awards of the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute of General Medical Sciences and the National Heart, Lung, and Blood Institute.
Finding
Acute portal infusion of piericidin A, a potent and specific inhibitor of mitochondrial complex I, did not reduce hepatic gluconeogenesis in vivo, so inhibition of complex I was shown not to be sufficient to lower plasma glucose. Metformin, phenformin and galegine instead interacted with complex IV and reduced its enzymatic activity at clinically relevant concentrations, leading indirectly to inhibition of mitochondrial glycerol-3-phosphate dehydrogenase (GPD2), an increased cytosolic redox state and selective inhibition of glycerol-derived hepatic gluconeogenesis in vitro and in vivo. Inhibition of complex IV with potassium cyanide reproduced these effects in vitro, selectively reducing glycerol-derived gluconeogenesis through an increased cytosolic redox state, and complex IV inhibition was sufficient to inhibit glycerol-3-phosphate-mediated respiration and gluconeogenesis.
Independent research
Madiraju AK, Qiu Y, Perry RJ, Rahimi Y, Zhang XM, Zhang D, et al. Metformin inhibits gluconeogenesis via a redox-dependent mechanism in vivo Nature Medicine, 2018.
doi:10.1038/s41591-018-0125-4 PubMed 30038219
Funding and finding
Funding
This publication was supported by grants from the US Department of Health and Human Services: R01 DK113984, P30 DK45735, P30 DK034989, K99 CA215315 and R01 DK114793, with further support from a Canada Research Chair in Metabolism and Obesity, an endowed chair at McMaster University, Diabetes Canada, the National Health and Medical Research Council of Australia and the Victorian Government Operational Infrastructure Support Scheme.
Finding
An acute intravenous dose of 50 mg/kg achieved plasma concentrations of 25 to 50 micromolar, which the authors state approximate plasma concentrations in people with type 2 diabetes after an acute 1 g oral dose or during chronic treatment with 1 g twice daily; parenteral doses of 200 mg/kg or more, used in many earlier rodent studies, give peak plasma concentrations of about 1200 micromolar, which the authors call supra-pharmacological. At clinically relevant concentrations metformin inhibited gluconeogenesis from lactate and glycerol but not from pyruvate or alanine, raised the hepatic lactate to pyruvate ratio about threefold and lowered the mitochondrial beta-hydroxybutyrate to acetoacetate ratio, with unaltered hepatic energy charge and citrate synthase flux, indicating no complex I inhibition. Infusion of pyruvate or dihydroxyacetone abolished the fall in plasma glucose, while lactate and glycerol infusion did not; methylene blue, which lowers the cytosolic NADH to NAD+ ratio, abrogated the effects of metformin on plasma glucose, redox ratios and endogenous glucose production. Infusion of 50 mg/kg into the portal vein gave peak plasma concentrations of about 130 micromolar and hepatic concentrations of about 25 micromolar at two hours and inhibited endogenous glucose production by about 25%, showing an effect independent of the gut and its microbiota. Hepatic AMPK was activated by acute treatment without a detectable effect on its substrate acetyl-CoA carboxylase.
Independent research
MacDonald MJ, Ansari IH, Longacre MJ, Stoker SW. Metformin's Therapeutic Efficacy in the Treatment of Diabetes Does Not Involve Inhibition of Mitochondrial Glycerol Phosphate Dehydrogenase Diabetes, 2021.
doi:10.2337/db20-1143 PubMed 33849997
Funding and finding
Funding
This work was supported by the Nowlin Family Trust of the InFaith Community Foundation and the estate of the late Ramona Garde.
Finding
Four different enzyme assays were used to retest the claim that metformin inhibits mitochondrial glycerol phosphate dehydrogenase (mGPD) by 30% to 50%. Metformin at 10 micromol/L to 1 mmol/L and phenformin at 250 micromol/L to 10 mmol/L produced no inhibition of mGPD in homogenates or mitochondria from insulin-secreting or liver cells, in a timed-and-stopped assay at a saturating glycerol phosphate concentration of 25 mmol/L, in three continuous spectrophotometric assays, or at 200 micromol/L glycerol phosphate, near the enzyme's Km of 400 micromol/L, where competitive inhibition would have been detectable. The authors set out the tissue distribution of mGPD: liver has the lowest activity and protein level of ten tissues, while the pancreatic islet level is 30- to 60-fold higher than liver, brown adipose tissue up to 250-fold, skeletal muscle up to 67-fold, brain up to 23-fold and testis 11- to 25-fold. Liver and beta cells also possess the malate-aspartate redox shuttle, which is far more active in liver and redundant with the glycerol phosphate shuttle; in their total-body mGPD knockout mouse the hepatic lactate to pyruvate and glycerol phosphate to dihydroxyacetone phosphate ratios are normal with no or minimal hepatic metabolic changes, insulin secretion is normal unless the malate-aspartate shuttle is also inhibited, and male fertility is markedly decreased. In another knockout, glycerol phosphate and the lactate to pyruvate ratio are normal in liver but threefold higher in skeletal muscle. The authors argue that if metformin did inhibit mGPD, adverse effects would be expected in tissues where mGPD is far more abundant than in liver.
Independent research
Forslund K, Hildebrand F, Nielsen T, Falony G, Le Chatelier E, Sunagawa S, et al. Disentangling type 2 diabetes and metformin treatment signatures in the human gut microbiota Nature, 2015.
doi:10.1038/nature15766 PubMed 26633628
Funding and finding
Participants
784
Funding
Funding from the European Community's Seventh Framework Programme (MetaHIT HEALTH-F4-2007-201052, MetaCardis HEALTH-2012-305312, International Human Microbiome Standards HEALTH-2010-261376), the Metagenopolis grant ANR-11-DPBS-0001, the European Research Council CancerBiome project, the European Union Horizon 2020 programme, the Lundbeck Foundation, the European Molecular Biology Laboratory and the Innovation Fund Denmark. The Novo Nordisk Foundation Center for Basic Metabolic Research is stated to be an independent research centre at the University of Copenhagen partially funded by an unrestricted donation from the Novo Nordisk Foundation.
Finding
Antidiabetic medication confounds reported gut microbial signatures of type 2 diabetes. Comparing metformin-treated with untreated patients, metformin was associated with a significant increase in Escherichia species and a reduced abundance of Intestinibacter, the latter consistent across country datasets; fasting serum metformin concentrations correlated significantly with the abundance of both genera, and an independent amplicon-based cohort reproduced both changes, Intestinibacter being detected only in the 9 of 30 untreated samples. Functional analysis showed enrichment of virulence factors and gas metabolism genes, attributable to the increase in Escherichia, which the authors describe as consistent with the known gastrointestinal side effects of metformin; they note that at least 30% of patients report adverse effects including diarrhoea, nausea, vomiting and bloating, with mechanisms poorly understood. Multivariate contrasts between treated and untreated patients were significant only at bacterial family level.
Independent research
Sun L, Xie C, Wang G, Wu Y, Wu Q, Wang X, et al. Gut microbiota and intestinal FXR mediate the clinical benefits of metformin Nature Medicine, 2018.
doi:10.1038/s41591-018-0222-4 PubMed 30397356
Funding and finding
Participants
22
Follow-up
3 days of treatment
Funding
This work was supported by the National Key Research and Development Program of China, the National Natural Science Foundation of China, the National Program for Support of Top-notch Young Professionals, the Fundamental Research Funds for the Central Universities of Peking University and the National Cancer Institute Intramural Research Program.
Finding
Metagenomic and metabolomic analysis of stool and serum before and after three days of metformin showed that the gut microbiota composition was substantially reshaped, with the genus Bacteroides most reduced and the species Bacteroides fragilis among the top discriminating species, and the bile acid glycoursodeoxycholic acid (GUDCA) increased. Serum FGF19 fell markedly and 7-alpha-hydroxy-4-cholesten-3-one rose, indicating suppression of intestinal farnesoid X receptor (FXR) signalling and increased hepatic CYP7A1 activity. In mice, colonisation with B. fragilis worsened glucose intolerance on a high-fat diet and abolished the metabolic benefit of metformin, and GUDCA was identified as an antagonist of intestinal FXR that improved several metabolic endpoints in established obesity. The authors note that metformin reaches high concentrations in the small intestine with much lower serum concentrations, so part of its benefit may arise from actions in the gut.
Independent research
Walton RG, Dungan CM, Long DE, Tuggle SC, Kosmac K, Peck BD, et al. Metformin blunts muscle hypertrophy in response to progressive resistance exercise training in older adults: A randomized, double-blind, placebo-controlled, multicenter trial: The MASTERS trial Aging Cell, 2019.
doi:10.1111/acel.13039 PubMed 31557380
Funding and finding
Registration
NCT02308228
Participants
94
Follow-up
14 weeks of progressive resistance training, with assessments 3 days after the final exercise bout
Funding
The study was funded by the National Institute on Aging, grant R01AG046920, and the National Center for Advancing Translational Sciences, grants UL1TR001998 and UL1TR003096.
Finding
Participants received 1700 mg metformin daily or placebo throughout, after a two-week drug wash-in and a two-week training ramp-up, and all performed 14 weeks of supervised progressive resistance training; 15 participants discontinued, 6 of them because of adverse events. Training reduced percentage body fat equally in both groups (-1.86 against -2.02 percentage points, p = 0.714). Placebo gained more lean mass than metformin (1.95%, SD 2.69, against 0.41%, SD 2.25; between-group p = 0.003) and more thigh muscle mass by dual-energy X-ray absorptiometry (3.90%, SD 5.54, against 0.45%, SD 3.95; p < 0.001). By computed tomography, placebo gained more total thigh muscle area (6.43%, SD 5.45, against 2.27%, SD 6.91; p = 0.005), more normal-density muscle area (10.5% against 4.16%, p = 0.001) and more muscle density (p = 0.020). Strength gains tended to be lower with metformin without reaching significance: knee extension one-repetition maximum 23.1% against 15.3% (p = 0.055), isometric strength 11.8% against 6.7% (p = 0.082), peak power 29.4% against 14.3% (p = 0.064); strength per unit thigh muscle mass did not differ (19.7% against 14.5%, p = 0.188). Metformin did not alter fibre cross-sectional area, satellite cell or macrophage abundance, but prevented the fall in type I fibre percentage seen with placebo (p = 0.007), raised AMPK signalling and tended to blunt mTORC1 signalling; in cultured human myotubes metformin inhibited p70S6K1 phosphorylation after electrical pulse stimulation. The authors conclude that these results do not support the use of metformin to enhance the benefits of physical activity in healthy older people.
Industry-funded
Konopka AR, Laurin JL, Schoenberg HM, Reid JJ, Castor WM, Wolff CA, et al. Metformin inhibits mitochondrial adaptations to aerobic exercise training in older adults Aging Cell, 2019.
doi:10.1111/acel.12880 PubMed 30548390
Funding and finding
Participants
53
Follow-up
12 weeks of aerobic exercise training
Funding
This work was supported by the Dairy Management, Inc and Dexcom, Inc. Analysis of plasma insulin was performed at a centre supported in part by NIH/NCATS grant UL1TR001082.
Finding
Double-blind randomisation to placebo (n = 26) or metformin (n = 27) during 12 weeks of supervised aerobic exercise training; dose titrated by 500 mg weekly to 2000 mg daily, reduced to 1500 mg daily in 11 participants who reported gastrointestinal discomfort (3 on placebo, 8 on metformin) and capped at 1500 mg for 5 participants weighing under 75 kg; pill compliance 98%. Independently of treatment, training reduced body weight, whole-body, trunk and leg fat mass, HbA1c, fasting insulin, HOMA-IR, 24-hour mean glucose and glycaemic variability. Maximal oxygen uptake rose significantly within the placebo group (p < 0.01) but not within the metformin group, metformin attenuating the increase by about 50% (p = 0.08). Whole-body insulin sensitivity, measured by the Matsuda index during a 75 g oral glucose tolerance test, rose after training with placebo but not metformin (time by treatment interaction p < 0.05), and metformin inhibited the increase (p = 0.02); the change with metformin was highly variable, with 58% of participants responding positively and 42% negatively. Metformin blunted the training-induced increase in skeletal muscle mitochondrial respiration and in the ratio of maximal coupled oxidative phosphorylation to uncoupled electron transport capacity, without affecting mitochondrial protein synthesis; the change in insulin sensitivity correlated with the change in complex I-linked respiration in the placebo group but not the metformin group. The authors state that these data raise concerns about broad recommendations for metformin as a treatment to target ageing, and that earlier studies similarly found metformin diminished the gain in cardiorespiratory fitness by 40% to 60% after 10 and 22 weeks of training.
Industry-funded
Madiraju AK, Erion DM, Rahimi Y, Zhang XM, Braddock DT, Albright RA, et al. Metformin suppresses gluconeogenesis by inhibiting mitochondrial glycerophosphate dehydrogenase Nature, 2014.
doi:10.1038/nature13270 PubMed 24847880
Funding and finding
Funding
This publication was supported by grants from the National Institutes of Health: R24 DK-085638, R01 DK-40936, P30 DK-45735, P30 DK-034989 and the Novo Nordisk Foundation Center for Basic Metabolic Research.
In-kind support
Antisense oligonucleotides carrying ISIS compound identifiers, selected from about 80 sequences pre-screened in vitro and in vivo; whole-body mGPD knockout mice from M. J. MacDonald, University of Wisconsin.
Finding
Metformin inhibited isolated rat mitochondrial glycerol-3-phosphate dehydrogenase non-competitively with a Ki of about 38 µM and purified recombinant human mGPD non-competitively with a Ki of about 55 µM; guanides and biguanides inhibited mGPD activity by 30 to 50% in rat mitochondrial lysates and did not alter the activity of purified recombinant cytosolic GPD. In awake 24-hour-fasted rats, intravenous metformin at 20 and 50 mg/kg lowered fasting plasma glucose and endogenous glucose production as early as 30 minutes after administration, raised the plasma lactate to pyruvate ratio and lowered the plasma beta-hydroxybutyrate to acetoacetate ratio; 30 minutes after 50 mg/kg, peak metformin averaged 74 µM in plasma and 100 µM in liver. Antisense knockdown of hepatic mGPD in rats reproduced the metformin phenotype and abolished the metformin-induced rise in cytosolic redox state, fall in plasma glucose and inhibition of endogenous glucose production; whole-body mGPD knockout mice reproduced these findings. The paper states that plasma metformin in patients taking 1 g twice daily is 10 to 40 µM, and that reported in vivo doses in animal studies range from 50 to 500 mg/kg.
Independent research
Dujic T, Zhou K, Donnelly LA, Tavendale R, Palmer CNA, Pearson ER. Association of Organic Cation Transporter 1 With Intolerance to Metformin in Type 2 Diabetes: A GoDARTS Study Diabetes, 2015.
doi:10.2337/db14-1388 PubMed 25510240
Funding and finding
Participants
2,166
Follow-up
prescribing records of incident metformin users from 1 January 1994 to 1 June 2011
Funding
The Wellcome Trust United Kingdom Type 2 Diabetes Case Control Collection (GoDARTS) cohort collection was funded by the Wellcome Trust and informatics support is provided by the Chief Scientist Office, Scotland. This work was specifically supported by a European Foundation for the Study of Diabetes Albert Renold Travel Fellowship; the senior author holds a Wellcome Trust New Investigator Award (102820/Z/13/Z).
Finding
Women and older people were more likely to be intolerant of metformin. Concomitant use of medicines known to inhibit OCT1 transport was associated with intolerance (odds ratio 1.63, 95% CI 1.22 to 2.17, p = 0.001), as was carriage of two reduced-function OCT1 alleles against one or none (2.41, 1.48 to 3.93, p < 0.001); individuals with two reduced-function alleles who also took OCT1 inhibitors were over four times more likely to be intolerant (4.13, 2.09 to 8.16, p < 0.001). Five reduced-function variants were studied: R61C, C88R, G401S, M420del and G465R. Drugs shown in vitro to inhibit OCT1 include tricyclic antidepressants, proton pump inhibitors, alpha-adrenoceptor antagonists and the calcium-channel blockers verapamil and diltiazem. The paper states as background, citing earlier literature, that gastrointestinal side effects occur in 20% to 30% of patients taking metformin and that about 5% develop severe symptoms and discontinue the drug, and that the intolerance phenotype used here is a proxy derived from prescribing patterns.
Independent research
Zhou K, Donnelly LA, Kimber CH, Donnan PT, Doney AS, Leese G, et al. Reduced-function SLC22A1 polymorphisms encoding organic cation transporter 1 and glycemic response to metformin: a GoDARTS study Diabetes, 2009.
doi:10.2337/db08-0896 PubMed 19336679
Funding and finding
Participants
3,450
Follow-up
maximum HbA1c reduction within 18 months; average HbA1c on monotherapy from 6 to 42 months
Funding
This research was specifically funded by Diabetes UK (grant 07/0003525). The Wellcome Trust provided support for the Wellcome Trust U.K. Type 2 Diabetes Case Control Collection and funded the Scottish Health Informatics Programme; informatics support was provided by the Chief Scientist Office, which also funds a Clinician Scientist award held by one author.
Finding
Metformin is actively transported into the liver by organic cation transporter 1, encoded by SLC22A1. R61C (rs12208357) and 420del were genotyped in 3450 incident metformin users; minor allele frequencies were 6.7% and 19.8%, and the R61C variant occurred only on the wild-type allele at codon 420. Among 1531 patients with a definable metformin response, neither variant affected the initial HbA1c reduction (P = 0.47 and P = 0.92), the chance of reaching HbA1c below 7% (P = 0.83 and P = 0.36), the average HbA1c on monotherapy up to 42 months (P = 0.44 and P = 0.75) or the hazard of monotherapy failure (P = 0.85 and P = 0.56); covariates were genotype, body-mass index, average dose, adherence and creatinine clearance. With 1500 subjects the analysis had 90% power to detect a difference in HbA1c reduction of 0.15% per copy of 420del and 80% power to detect 0.2% per copy of R61C. The authors note that the design is observational, since the decision to start metformin and the rate of titration rested with patients and physicians, and that intolerance was not measured directly.
Independent research
Boussageon R, Supper I, Bejan-Angoulvant T, Kellou N, Cucherat M, Boissel JP, et al. Reappraisal of metformin efficacy in the treatment of type 2 diabetes: a meta-analysis of randomised controlled trials PLoS Medicine, 2012.
doi:10.1371/journal.pmed.1001204 PubMed 22509138
Funding and finding
Participants
13,110
Follow-up
4 months to 10.7 years
Funding
No direct funding was received for this study. The authors were personally salaried by their institutions during the period of writing, though no specific salary was set aside or given for the writing of this paper.
Finding
Metformin did not significantly affect the primary endpoints all-cause mortality (risk ratio 0.99, 95% CI 0.75 to 1.31) or cardiovascular death (1.05, 0.67 to 1.64), with significant heterogeneity when the UK Prospective Diabetes Study subgroups were included (I2 41% and 59%); excluding UKPDS 34 left the estimates unchanged (0.98, 0.77 to 1.24 and 0.95, 0.72 to 1.26) and removed the heterogeneity. Secondary endpoints: all myocardial infarctions 0.90 (0.74 to 1.09), all strokes 0.76 (0.51 to 1.14), heart failure 1.03 (0.67 to 1.59), peripheral vascular disease 0.90 (0.46 to 1.78), leg amputations 1.04 (0.44 to 2.44) and microvascular complications 0.83 (0.59 to 1.17). In the comparison of metformin plus sulphonylurea against sulphonylurea alone, all-cause mortality rose (risk ratio 1.53, 1.02 to 2.31) and cardiovascular death rose (2.20, 1.20 to 4.03); the discussion gives 1.55 (1.03 to 2.33) for deaths across the five add-on trials, with UKPDS 34 carrying 35.1% of the weight. The authors state that a 25% reduction or a 31% increase in all-cause mortality, and a 33% reduction or a 64% increase in cardiovascular mortality, cannot be excluded; that only four of 25 eligible trials had a clinical event as the primary outcome; that 12 trials could not be included for lack of outcome data; that UKPDS 34 changed its primary endpoint and study length during the study and lacked a placebo group and blinding; that the favourable UKPDS 34 overweight-group result is cited while the unfavourable sulphonylurea add-on result is not, although both arms were randomised; and that HbA1c cannot be considered a valid surrogate endpoint for the clinical efficacy of glucose-lowering drugs.
Independent research
Hong J, Zhang Y, Lai S, Lv A, Su Q, Dong Y, et al. Effects of metformin versus glipizide on cardiovascular outcomes in patients with type 2 diabetes and coronary artery disease Diabetes Care, 2013.
doi:10.2337/dc12-0719 PubMed 23230096
Funding and finding
Registration
NCT00513630
Participants
304
Follow-up
3 years of study drug administration; median 5.0 years (range 3.7 to 5.7) for the primary endpoint
Funding
Grants from the 863 Project, the National Natural Science Foundation of China, the Shanghai Shenkang Hospital Development Center, the Key Laboratory for Endocrine and Metabolic Diseases of the Ministry of Chinese Public Health, the National Key New Drug Creation and Manufacturing Program, the Shanghai Committee on Science and Technology, the Program for Innovative Research Team of Shanghai Municipal Education Commission and the Sector Funds of the Ministry of Health. All funding sources had no role in study design, data collection, data analysis, data interpretation or writing of the manuscript. No other potential conflicts of interest relevant to the article were reported.
In-kind support
Both glipizide and metformin were provided by Xinyi Pharmaceutical Co., Shanghai.
Finding
Double-blind randomisation after a two-week run-in during which all previous antidiabetic agents were withdrawn: glipizide 30 mg daily with metformin placebo, or metformin 1.5 g daily with glipizide placebo, for 3 years; mean achieved doses 28.3 mg and 1.4 g daily. HbA1c fell from 7.6% to 7.1% with glipizide and 7.6% to 7.0% with metformin within 6 months, with no between-group difference. Over the whole study, 103 composite primary endpoints occurred in 91 patients: 52 patients (35.1%) in the glipizide group and 39 (25.0%) in the metformin group. Events were 60 with glipizide (14 deaths from any cause, 6 non-fatal myocardial infarctions, 15 non-fatal strokes, 25 arterial revascularisations) against 43 with metformin (7 deaths, 5 non-fatal myocardial infarctions, 10 non-fatal strokes, 21 revascularisations). The adjusted hazard ratio for the composite was 0.54 (95% CI 0.30 to 0.90, p = 0.026); mortality did not differ (p = 0.55). Secondary endpoints, recorded only during the three years of study drug, did not differ: new or worsening heart failure 10 (6.8%) against 9 (5.8%), new critical arrhythmia 27 (18.2%) against 30 (19.2%), new or worsening angina 71 (48%) against 77 (49.4%), peripheral vascular events 6 (4.1%) against 1 (0.6%), hypoglycaemic attacks 4 against 3. Statin use at the end of the study was lower in the metformin group (59.5% against 73.7%, p = 0.013). The authors state that they did not monitor diet or exercise, had no washout period, and advise caution in interpretation.
Independent research
Goodwin PJ, Chen BE, Gelmon KA, Whelan TJ, Ennis M, Lemieux J, et al. Effect of Metformin vs Placebo on Invasive Disease-Free Survival in Patients With Breast Cancer: The MA.32 Randomized Clinical Trial JAMA, 2022.
doi:10.1001/jama.2022.6147 PubMed 35608580
Funding and finding
Registration
NCT01101438
Participants
3,649
Follow-up
median 96.2 months in the hormone receptor-positive group and 94.1 months in the receptor-negative group
Funding
Supported by grant 021039 from the Canadian Cancer Society Research Institute; CA077202, CA180868 and CA180822 from the US National Cancer Institute; the Breast Cancer Research Foundation; Canadian Breast Cancer Foundation-Ontario Region; 10NOV-467 from the Ontario Institute for Cancer Research; 02689-08-2010 from the Stiftung Krebsforschung Schweiz; Cancer Research UK; Apotex Canada (in-kind donation of placebo and metformin); and the Hold'em for Life Charity. The Canadian Cancer Trials Group was responsible for protocol development and trial conduct but had no role in analysis, interpretation and approval of the manuscript or the decision to submit it.
In-kind support
In-kind donation of placebo and metformin from Apotex Canada.
Finding
Patients were randomised to metformin or placebo in a phase 3 double-blind trial. Futility was declared at a second interim analysis for hormone receptor-negative patients, so the primary analysis covered the 2533 oestrogen or progesterone receptor-positive patients, among whom 465 invasive disease-free survival events occurred. Event rates were 2.78 per 100 patient-years with metformin against 2.74 with placebo (hazard ratio 1.01, 95% CI 0.84 to 1.21, p = 0.93) and death rates 1.46 against 1.32 per 100 patient-years (1.10, 0.86 to 1.41, p = 0.47). Among receptor-negative patients the rates were 3.58 against 3.60 per 100 patient-years (1.01, 0.79 to 1.30, p = 0.92). None of the three secondary outcomes in the receptor-positive group differed significantly. Grade 3 non-haematological adverse events were more frequent with metformin than placebo (21.5% against 17.5%, p = 0.003), the most common being hypertension (2.4% against 1.9%) and irregular menses (1.5% against 1.4%).
Industry-funded
UK Prospective Diabetes Study (UKPDS) Group. Effect of intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34) The Lancet, 1998.
doi:10.1016/S0140-6736(98)07037-8 PubMed 9742977
Funding and finding
Registration
ISRCTN75451837
Participants
753
Follow-up
median 10.7 years
Funding
Funding read from the UKPDS study website and the ISRCTN entry, not from the paper, whose study organisation and funding are printed at the end of the full text. Primary funders: UK Medical Research Council, British Diabetic Association, UK Department of Health, British Heart Foundation, US National Institutes of Health (NEI, NIDDK), Wellcome Trust; also Novo-Nordisk, Bayer, Bristol-Myers Squibb, Hoechst, Lilly, Lipha, Glaxo Wellcome, SmithKline Beecham, Pfizer, Zeneca, Pharmacia & Upjohn, Roche, Farmitalia Carlo Erba; Charles Wolfson Charitable Trust, Clothworkers' Foundation, Health Promotion Research Trust, Alan and Babette Sainsbury Trust, Oxford University Medical Research Fund Committee. The ISRCTN entry names the MRC as funder and the University of Oxford Diabetes Trials Unit as sponsor, with 'several pharmaceutical companies' as a further funder.
Finding
Results read from the published abstract; the publisher's full text returned HTTP 403 to automated access and is absent from PubMed Central, so absolute event rates and the paper's own funding statement were not read. In the main randomised comparison, 342 patients were allocated metformin (target fasting plasma glucose below 6 mmol/L) and 411 conventional policy, primarily diet alone. Median glycated haemoglobin was 7.4% with metformin against 8.0% with conventional policy. Risk reductions with metformin against conventional policy: 32% for any diabetes-related endpoint (95% CI 13 to 47, p = 0.002), 42% for diabetes-related death (9 to 63, p = 0.017), 36% for all-cause mortality (9 to 55, p = 0.011). Against 951 overweight patients allocated intensive control with chlorpropamide, glibenclamide or insulin, metformin showed a greater effect for any diabetes-related endpoint (p = 0.0034), all-cause mortality (p = 0.021) and stroke (p = 0.032). In a supplementary trial, 537 patients already on maximum sulphonylurea were allocated continued sulphonylurea (n = 269) or added metformin (n = 268); added metformin was associated with a 96% increased risk of diabetes-related death (95% CI 2 to 275, p = 0.039). A combined analysis of the main and supplementary trials gave a 19% reduction in diabetes-related endpoints (2 to 33, p = 0.033). Epidemiological assessment in 4416 patients showed no increased diabetes-related death with combined sulphonylurea and metformin (risk reduction 5%, -33 to 32, p = 0.78).
Industry-funded
Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year follow-up of intensive glucose control in type 2 diabetes New England Journal of Medicine, 2008.
doi:10.1056/NEJMoa0806470 PubMed 18784090
Funding and finding
Registration
ISRCTN75451837
Participants
4,209
Follow-up
median 17.7 years in the metformin comparison, equivalent to 12,431 person-years, including 8.8 years of post-trial monitoring
Funding
Post-trial monitoring was initiated by the investigators and supported for the first 5 years by the UK Medical Research Council, UK Department of Health, Diabetes UK, the British Heart Foundation and the UK National Institute for Health, and for the final 5 years by Bristol-Myers Squibb, GlaxoSmithKline, Merck Serono, Novartis, Novo Nordisk and Pfizer. The original interventional trial was supported by UK and US public bodies, the Wellcome Trust and charities together with Novo Nordisk, Bayer, Bristol-Myers Squibb, Hoechst, Lilly, Lipha and Farmitalia Carlo Erba, with consumables or logistical support from further companies. The paper states that the investigators designed and conducted the study, analysed the data and prepared the manuscript independently of any funding bodies.
Finding
Between-group differences in glycated haemoglobin were lost within the first year, with similar improvements thereafter in all groups. In the metformin group against overweight patients on conventional therapy, risk reductions seen during the trial were maintained at 10 years of post-trial monitoring: any diabetes-related endpoint 21% (p = 0.01), diabetes-related death 30% (p = 0.01), myocardial infarction 33% (p = 0.005) and death from any cause 27% (p = 0.002). No significant risk reduction was observed during or after the trial for microvascular disease, stroke or peripheral vascular disease; microvascular risk reductions of 29% during the trial and 26% afterwards were not significant. During the original trial the metformin group showed reductions in myocardial infarction of 39% (p = 0.01) and in all-cause death of 36% (p = 0.01). Plasma creatinine in the metformin group was on average 15% higher than in the conventional-therapy group (p < 0.04). Overall mortality during follow-up was 44.0%; 3.5% of patients were lost to follow-up; questionnaires replaced clinic visits in years 6 to 10 and may not have captured all non-fatal outcomes. Counts of events underlying these percentages are given only in the article's Table 2, which could not be read.
Industry-funded
Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, Walker EA, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin New England Journal of Medicine, 2002.
doi:10.1056/NEJMoa012512 PubMed 11832527
Funding and finding
Participants
3,234
Follow-up
average 2.8 years (range 1.8 to 4.6)
Funding
Supported by the National Institutes of Health through the National Institute of Diabetes and Digestive and Kidney Diseases, the Office of Research on Minority Health, the National Institute of Child Health and Human Development and the National Institute on Aging; the Indian Health Service; the Centers for Disease Control and Prevention; the General Clinical Research Center Program, National Center for Research Resources; the American Diabetes Association; Bristol-Myers Squibb; and Parke-Davis.
In-kind support
Metformin and placebo from Lipha Pharmaceuticals; materials, equipment and medicines for concomitant conditions from LifeScan, Health-O-Meter, Hoechst Marion Roussel, Merck-Medco Managed Care, Merck, Nike, Slim-Fast Foods and Quaker Oats.
Finding
Participants were assigned placebo (n = 1082), metformin 850 mg twice daily (n = 1073) or intensive lifestyle modification (n = 1079); a troglitazone arm was discontinued in 1998 and is not reported. Crude incidence of diabetes was 11.0, 7.8 and 4.8 cases per 100 person-years. Metformin reduced incidence by 31% against placebo (95% CI 17 to 43) and lifestyle by 58% (48 to 66). Estimated cumulative incidence at 3 years was 28.9% with placebo, 21.7% with metformin and 14.4% with lifestyle; the number needed to treat for 3 years to prevent one case was 13.9 (8.7 to 33.9) for metformin and 6.9 (5.4 to 9.5) for lifestyle. Average weight loss was 0.1 kg with placebo, 2.1 kg with metformin and 5.6 kg with lifestyle (P < 0.001). Gastrointestinal symptoms occurred at 77.8 events per 100 person-years with metformin, 30.7 with placebo and 12.9 with lifestyle; musculoskeletal symptoms at 20.0, 21.1 and 24.1. Hospitalisation and mortality rates did not differ. Metformin was less effective at lower body-mass index or lower fasting glucose.
Funding unknown
Gnesin F, Thuesen ACB, Kahler LKA, Madsbad S, Hemmingsen B. Metformin monotherapy for adults with type 2 diabetes mellitus Cochrane Database of Systematic Reviews, 2020.
doi:10.1002/14651858.CD012906.pub2 PubMed 32501595
Funding and finding
Participants
10,680
Follow-up
one year to 10.7 years; last search 2 December 2019
Funding
The review's own Sources of support section could not be read: the Cochrane Library and Wiley pages return HTTP 403 to automated access and the PubMed Central record contains the abstract and plain-language summary only. The Europe PMC core record for this review lists grants from Steno Diabetes Center Copenhagen (SDCC 3.F CMP) and the NNF Center for Basic Metabolic Research (Hansen Group). The class is therefore undetermined.
Finding
About 58% of participants finished the trials in all groups. No trial was judged at low risk of bias on all risk-of-bias domains. No trial compared metformin with placebo or with a behaviour-changing intervention. Against sulphonylureas, four trials reported all-cause mortality: in three no participant died, and in the fourth 31 of 1454 (2.1%) died on metformin against 31 of 1441 (2.2%); serious adverse events occurred in 331 of 1454 (22.8%) against 308 of 1441 (21.4%); cardiovascular death in 4 of 1441 (0.3%) against 8 of 1447 (0.6%); non-fatal myocardial infarction in 21 of 1454 (1.4%) against 15 of 1441 (1.0%); all very low-certainty evidence. Against thiazolidinediones, all-cause mortality risk ratio 0.88 (95% CI 0.55 to 1.39, 5 trials, 4402 participants), serious adverse events 0.95 (0.84 to 1.09, 3208 participants) and cardiovascular death 0.71 (0.21 to 2.39, 3211 participants); very low-certainty evidence for all outcomes. Against dipeptidyl peptidase-4 inhibitors, a glucagon-like peptide-1 analogue, a meglitinide or no intervention, no substantial differences were found, all at very low certainty. Health-related quality of life and end-stage renal disease were essentially unreported. The authors conclude that there is no clear evidence whether metformin monotherapy, compared with no intervention, behaviour-changing interventions or other glucose-lowering drugs, influences patient-important outcomes, and note that almost all studies investigated laboratory measures such as fasting blood glucose.
Independent research
Griffin SJ, Leaver JK, Irving GJ. Impact of metformin on cardiovascular disease: a meta-analysis of randomised trials among people with type 2 diabetes Diabetologia, 2017.
doi:10.1007/s00125-017-4337-9 PubMed 28770324
Funding and finding
Participants
4,158
Follow-up
6 to 212 months; three studies followed patients for more than 4 years
Funding
One author is funded by a National Institute for Health Research (NIHR) Clinical Lectureship; the other is supported by MRC Epidemiology Unit programme funding (MC_UU_12015/4) and is an NIHR senior investigator. All researchers were independent of funding bodies, which had no role in study design, data collection, analysis and interpretation of data, in the writing of the report, or the decision to submit the article for publication. This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Finding
Only four trials, including 417 patients allocated metformin, simply compared metformin with placebo and collected cardiovascular outcome data. Summary estimates rested on small numbers of events: 347 cardiovascular deaths in five studies, 416 myocardial infarctions or ischaemic heart disease events in seven studies and 111 strokes in four studies. Mantel-Haenszel risk ratios: all-cause mortality 0.96 (95% CI 0.84 to 1.09), cardiovascular death 0.97 (0.80 to 1.16), myocardial infarction 0.89 (0.75 to 1.06), stroke 1.04 (0.73 to 1.48) and peripheral vascular disease 0.81 (0.50 to 1.31); all outcomes except stroke favoured metformin, none significantly. The UK Prospective Diabetes Study contributed the majority of the data, with weights from 52.3% for myocardial infarction to 70.5% for stroke. No included trial was assessed as having a low risk of bias. The authors state that metformin may reduce all-cause mortality by up to 16% but could increase the risk of stroke by up to 48%; that trial evidence does not support its use for cardiovascular prevention in older people, in those with HbA1c below 8%, in non-white ethnic groups or outside North America and northern Europe; that limitations of UKPDS include only 342 patients allocated metformin, absence of placebo and blinding, the subgroup nature of the analysis with updated significance thresholds, possible between-group differences in other risk factor management and attrition over nearly 18 years; that concern remains about increased mortality when metformin is added to a sulfonylurea; and that a definitive placebo-controlled cardiovascular endpoint trial in diabetes is unlikely to be conducted.
Independent research
Lean ME, Leslie WS, Barnes AC, Brosnahan N, Thom G, McCombie L, et al. Primary care-led weight management for remission of type 2 diabetes (DiRECT): an open-label, cluster-randomised trial Lancet, 2018.
doi:10.1016/S0140-6736(17)33102-1 PubMed 29221645
Funding and finding
Registration
ISRCTN03267836
Participants
306
Follow-up
12 months
Funding
Diabetes UK. The ISRCTN registry entry for ISRCTN03267836 names Diabetes UK as funder and Cambridge Weight Plan (UK) as a provider of product and training support, with NHS Greater Glasgow and Clyde as sponsor.
In-kind support
Product and training support from Cambridge Weight Plan (UK), as recorded in the ISRCTN registry entry.
Finding
Practices were randomised to a weight management programme comprising withdrawal of antidiabetic and antihypertensive drugs, total diet replacement at 825 to 853 kcal/day for 3 to 5 months, stepped food reintroduction over 2 to 8 weeks and structured support for weight loss maintenance, or to best-practice care by guidelines. At 12 months, weight loss of 15 kg or more was recorded in 36 of 149 intervention participants (24%) and none of the control participants (p < 0.0001). Remission of diabetes, defined as HbA1c below 6.5% after at least 2 months off all antidiabetic medication, was achieved by 68 of 149 (46%) against 6 of 149 (4%), odds ratio 19.7 (95% CI 7.8 to 49.8, p < 0.0001). Remission varied with weight loss across the whole study population: none of 76 who gained weight, 6 of 89 (7%) who stayed within 5 kg of baseline, 19 of 56 (34%) losing 5 to 10 kg, 16 of 28 (57%) losing 10 to 15 kg and 31 of 36 (86%) losing 15 kg or more. Mean weight fell by 10.0 kg (SD 8.0) against 1.0 kg (3.7), an adjusted difference of -8.8 kg (-10.3 to -7.3, p < 0.0001). Quality of life on the EuroQol 5 Dimensions visual analogue scale improved by 7.2 points (SD 21.3) in the intervention group and fell by 2.9 (15.5) in the control group (adjusted difference 6.4 points, 2.5 to 10.3, p = 0.0012). Nine serious adverse events were reported by 7 of 157 intervention participants (4%) and two by two control participants (1%); two events in one participant, biliary colic and abdominal pain, were deemed potentially related to the intervention, and no serious adverse event led to withdrawal.
Independent research
Gong Q, Zhang P, Wang J, Ma J, An Y, Chen Y, et al. Morbidity and mortality after lifestyle intervention for people with impaired glucose tolerance: 30-year results of the Da Qing Diabetes Prevention Outcome Study Lancet Diabetes and Endocrinology, 2019.
doi:10.1016/S2213-8587(19)30093-2 PubMed 31036503
Funding and finding
Participants
577
Follow-up
up to 30 years; 540 of 576 participants (94%) assessed for outcomes
Funding
The study was supported from 1986 to 1992 by the World Bank, the Ministry of Public Health of the People's Republic of China and Da Qing First Hospital; from 2004 to 2009 by a US Centers for Disease Control and Prevention and WHO cooperative agreement, China-Japan Friendship Hospital and Da Qing First Hospital; and from 2015 to 2018 by a US CDC and Chinese CDC cooperative agreement, the National Center for Cardiovascular Diseases and Fuwai Hospital, China-Japan Friendship Hospital and Da Qing First Hospital. Investigators from the funding organisations were involved in study design, data collection, analysis, interpretation and writing.
Finding
Over 30 years the combined intervention group had a median delay in diabetes onset of 3.96 years (95% CI 1.25 to 6.67, p = 0.0042). Cardiovascular disease events occurred in 195 of 405 intervention participants (48%, cumulative incidence 52.9%, 47.5 to 57.9) and 80 of 135 control participants (59%, cumulative incidence 66.5%, 57.0 to 74.4), 26% fewer events (8 to 41, p = 0.0060), with nine people needing treatment over 30 years to prevent one cardiovascular event (5 to 36). Composite microvascular disease was 35% lower (5 to 55, p = 0.025), cumulative incidence 25.1% against 34.0%, with retinopathy 40% lower (5 to 62, p = 0.032) and nephropathy and neuropathy numerically but not significantly lower. Cardiovascular death fell 33% (6 to 52, p = 0.022), cumulative incidence 25.6% against 35.2%, and all-cause death 26% (11 to 39, p = 0.0015), with ten people needing treatment over 30 years to prevent one death (6 to 25); average life expectancy increased by 1.44 years (0.20 to 2.68, p = 0.023). Reductions in mortality and vascular complications were greater in women than men, and adjusting for the time of diabetes onset nullified the significance of the intervention effect for each primary outcome, which the authors read as showing that much of the benefit follows from the delay in diabetes onset.
Industry-funded
Look AHEAD Research Group, Wing RR, Bolin P, Brancati FL, Bray GA, Clark JM, et al. Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes New England Journal of Medicine, 2013.
doi:10.1056/NEJMoa1212914 PubMed 23796131
Funding and finding
Registration
NCT00017953
Participants
5,145
Follow-up
median 9.6 years (interquartile range 8.9 to 10.3); planned maximum 13.5 years, stopped early for futility on 14 September 2012
Funding
The study was sponsored by the National Institutes of Health with additional support from other federal partners and the clinical research centres of participating institutions. Major contributions were provided by the FedEx Corporation, Health Management Resources, LifeScan Inc (a Johnson & Johnson company), Nestle HealthCare Nutrition Inc, Hoffmann-La Roche Inc, Abbott Nutrition and Unilever North America; the paper states that none of the corporate sponsors had any role in the trial design, data analysis or reporting of results.
Finding
Participants were randomised to an intensive lifestyle intervention aiming at 7% weight loss through a calorie goal of 1200 to 1800 kcal/day, meal replacements and at least 175 minutes of moderate activity weekly, or to diabetes support and education. Weight loss was greater in the intervention group throughout: 8.6% against 0.7% at one year and 6.0% against 3.5% at the end of the study. The intervention also produced greater reductions in HbA1c and greater initial improvements in fitness and all cardiovascular risk factors except LDL cholesterol. The primary composite outcome of cardiovascular death, non-fatal myocardial infarction, non-fatal stroke or hospitalised angina occurred in 403 of 2570 intervention participants and 418 of 2575 control participants, 1.83 against 1.92 events per 100 person-years (hazard ratio 0.95, 95% CI 0.83 to 1.09, p = 0.505), with no significant differences in the prespecified secondary composite outcomes, in any individual component or across prespecified subgroups. The trial was designed with more than 80% power to detect an 18% difference assuming a control event rate of 2% per year; at termination the probability of a significant positive result at the planned end of follow-up, assuming a true hazard ratio of 0.82, was estimated at 1%. Fewer than 4% of participants were lost to follow-up.
Industry-funded
Goldberg RB, Orchard TJ, Crandall JP, Boyko EJ, Budoff M, Dabelea D, et al. Effects of Long-term Metformin and Lifestyle Interventions on Cardiovascular Events in the Diabetes Prevention Program and Its Outcome Study Circulation, 2022.
doi:10.1161/CIRCULATIONAHA.121.056756 PubMed 35603600
Funding and finding
Participants
3,234
Follow-up
median 21 years
Funding
Supported by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health with 22 named awards, and by other National Institutes of Health institutes, the Centers for Disease Control and Prevention and the American Diabetes Association. Merck KGaA provided medication for the Outcomes Study; Bristol-Myers Squibb, Parke-Davis, LifeScan, Health O Meter, Hoechst Marion Roussel, Merck-Medco Managed Care, Merck and Co, Nike Sports Marketing, Slim Fast Foods and Quaker Oats donated materials, equipment or medicines for concomitant conditions. The sponsor of the study was represented on the steering committee and played a part in study design, conduct and publication.
In-kind support
Medication from Merck KGaA; materials, equipment and medicines for concomitant conditions from Bristol-Myers Squibb, Parke-Davis, LifeScan and others.
Finding
Over a median 21 years, 310 participants had a first major cardiovascular event, defined as non-fatal myocardial infarction, non-fatal stroke or cardiovascular death and adjudicated by a masked committee. Event rates were 5.51 per 1000 person-years with metformin (101 events), 6.10 with lifestyle (111 events) and 5.28 with placebo (98 events): metformin against placebo hazard ratio 1.03 (95% CI 0.78 to 1.37, p = 0.81), lifestyle against placebo 1.14 (0.87 to 1.50, p = 0.34). Non-fatal myocardial infarction occurred at 2.49, 1.90 and 2.30 per 1000 person-years; non-fatal stroke at 0.86, 2.12 and 1.48 (metformin against placebo 0.57, 0.31 to 1.06, p = 0.07); cardiovascular death at 2.07, 1.99 and 1.42 (metformin against placebo 1.46, 0.90 to 2.39, p = 0.13; lifestyle 1.40, 0.85 to 2.29, p = 0.19). The extended cardiovascular outcome occurred at 8.86, 9.93 and 8.73 per 1000 person-years (metformin against placebo 1.00, 0.80 to 1.25). Risk factor adjustment and adjustment for out-of-study metformin use did not change these results, and prespecified subgroups by age, sex, race or ethnicity and diabetes development showed no significant heterogeneity. A prespecified futility analysis in March 2019 gave a hazard ratio of 0.91 (0.68 to 1.23) with conditional power of 4.1%, and the data safety monitoring board declared futility. Study metformin use fell from 77% at the end of the Diabetes Prevention Program to 41% at the most recent assessment while out-of-study metformin use rose in all three groups; 68% to 74% of participants were taking antihypertensive drugs and 53% to 62% statins at 21 years, with mean blood pressure 122/73 mm Hg, mean LDL cholesterol 2.8 mmol/L and mean HbA1c 6.0% to 6.1%.
Independent research
Action to Control Cardiovascular Risk in Diabetes Study Group, Gerstein HC, Miller ME, Byington RP, Goff DC Jr, Bigger JT, et al. Effects of intensive glucose lowering in type 2 diabetes New England Journal of Medicine, 2008.
doi:10.1056/NEJMoa0802743 PubMed 18539917
Funding and finding
Registration
NCT00000620
Participants
10,251
Follow-up
mean 3.5 years; the intensive regimen was stopped early for excess mortality
Funding
Supported by grants from the National Heart, Lung, and Blood Institute, by other components of the National Institutes of Health including the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute on Aging and the National Eye Institute, by the Centers for Disease Control and Prevention and by General Clinical Research Centers. Abbott Laboratories, Amylin Pharmaceutical, AstraZeneca, Bayer HealthCare, Closer Healthcare, GlaxoSmithKline, King Pharmaceuticals, Merck, Novartis, Novo Nordisk, Omron Healthcare, Sanofi-Aventis and Schering-Plough provided study medications, equipment or supplies.
In-kind support
Study medications, equipment or supplies donated by thirteen companies.
Finding
Patients were randomised to intensive therapy targeting an HbA1c below 6.0% or standard therapy targeting 7.0% to 7.9%; stable median HbA1c values of 6.4% and 7.5% were achieved at one year and an absolute difference of 1.1 percentage points maintained for 3.5 years. The primary composite outcome of non-fatal myocardial infarction, non-fatal stroke or cardiovascular death occurred in 352 patients on intensive therapy against 371 on standard therapy (hazard ratio 0.90, 95% CI 0.78 to 1.04, p = 0.16). Deaths numbered 257 against 203 (hazard ratio 1.22, 1.01 to 1.46, p = 0.04), a relative increase of 22% and an absolute increase of 1.0 percentage point over the follow-up period, equivalent to one extra death for every 95 patients treated for 3.5 years, with similar differences in cardiovascular death. The annualised rate of hypoglycaemia requiring medical assistance was 3.1% on intensive therapy against 1.0% on standard therapy and mean weight gain at three years was 3.5 kg against 0.4 kg; one death in each group was adjudicated as probably related to hypoglycaemia. Glucose-lowering regimens were altered a mean 4.4 times per year on intensive therapy against 2.0 times on standard therapy, with greater exposure to every drug class. On 8 January 2008 the data and safety monitoring committee concluded that the harm from the increased rate of death outweighed any potential benefit and recommended that the intensive regimen be discontinued; patients were informed on 5 February 2008. The authors state that these findings identify a previously unrecognised harm of intensive glucose lowering in high-risk patients.
Independent research
Neuman J, Korenstein D, Ross JS, Keyhani S. Prevalence of financial conflicts of interest among panel members producing clinical practice guidelines in Canada and United States: cross sectional study BMJ, 2011.
doi:10.1136/bmj.d5621 PubMed 21990257
Funding and finding
Participants
288
Funding
This project was not directly supported by any research funds. One author is funded by a Veterans' Administration HSR&D career development award; one is supported by the National Institute on Aging (K08 AG032886) and by the American Federation of Aging Research through the Paul B Beeson Career Development Award Program.
Finding
Five of the 14 guidelines had no accompanying declaration of conflicts of interest. Of 288 panel members, 138 (48%) declared conflicts at publication; among 73 who declared none, 8 (11%) were found to have one or more; overall 150 (52%) had conflicts, 12 undeclared. Of members declaring conflicts, 93% reported honoraria, speakers' fees, consultancy or stock ownership from manufacturers of relevant drugs. Six of 12 identified chairs had conflicts. Government-sponsored panels had fewer conflicts than others (15 of 92, 16%, against 135 of 196, 69%). Among hyperlipidaemia panels, 44 of 99 members (44%) had conflicts, and 40 of 61 (66%) in the hyperlipidaemia guidelines with declarations.
Expert consensus
Nathan DM, Buse JB, Davidson MB, Heine RJ, Holman RR, Sherwin R, Zinman B. Management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy. A consensus statement from the American Diabetes Association and the European Association for the Study of Diabetes Diabetes Care, 2006.
doi:10.2337/dc06-9912 PubMed 16873813
Funding and finding
Funding
Funding of the consensus process is not stated in the article, which records that it was published simultaneously in Diabetes Care and Diabetologia by the American Diabetes Association and the European Association for the Study of Diabetes. Author disclosures are published in full.
Finding
This statement set the algorithm that made metformin the first drug of type 2 diabetes: the panel's consensus was that metformin therapy should be initiated concurrently with lifestyle intervention at diagnosis, recommended as initial pharmacologic therapy in the absence of specific contraindications for its effect on glycaemia, absence of weight gain or hypoglycaemia, generally low level of side effects, high acceptance and relatively low cost, and titrated to its maximally effective dose over one to two months. The authors state that the guidelines and algorithm are based on clinical trials and on the authors' clinical experience and judgment; that the paucity of high-quality evidence directly comparing treatment regimens remains a major impediment to recommending one class over another; that apart from effects on glycaemia there are insufficient data to recommend one class of glucose-lowering agent over others with regard to complications; that the UK Prospective Diabetes Study could not demonstrate clear superiority of sulphonylurea, metformin or insulin for complications; and that the beneficial effect of metformin on cardiovascular outcomes in that study needs to be confirmed. Metformin monotherapy is given as lowering HbA1c by about 1.5 percentage points, with lactic acidosis quite rare at fewer than 1 case per 100,000 treated patients. On targets, the statement records that the most recent American Diabetes Association goal, selected on the basis of practicality and the projected reduction in complications over time, is in general an HbA1c below 7%, with as close to normal as possible without significant hypoglycaemia for the individual patient, and that the European Union and International Diabetes Federation goal is below 6.5%; the panel's consensus was that an HbA1c of 7% or more should serve as a call to action, while noting that this goal is not appropriate or practical for some patients and that life expectancy and hypoglycaemia risk must be considered. The most appropriate target levels for blood glucose and HbA1c are stated not to have been systematically studied.
Regulator
European Medicines Agency. Metformin and metformin-containing medicines: Article 31 referral. Use of metformin to treat diabetes now expanded to patients with moderately reduced kidney function European Medicines Agency, reference EMA/603690/2016, first published 14 October 2016, 2016.
https://www.ema.europa.eu/en/medicines/human/referrals/metformin-metformin-containing-medicines
Funding and finding
Funding
European Medicines Agency; review initiated at the request of the Netherlands
Finding
The review was initiated on 28 January 2016 at the request of the Netherlands under Article 31 of Directive 2001/83/EC. On 13 October 2016 the Agency concluded that metformin-containing medicines can be used in patients with moderately reduced kidney function (glomerular filtration rate 30 to 59 ml/min), revising the previous contraindication; use remains contraindicated below 30 ml/min. GFR is to be assessed before starting treatment and at least annually thereafter, and reduced doses considered in moderate reduction. The product information was also to detail risk factors for lactic acidosis to be reviewed before and during treatment, and to advise patients to stop metformin for a short time and seek advice when dehydrated through severe vomiting, diarrhoea, fever, heat exposure or reduced fluid intake. Companies marketing metformin-containing medicines were requested to monitor and analyse future lactic acidosis cases and report them in periodic safety reviews. Several fixed-dose combinations remain not recommended in moderately reduced kidney function because of the second active substance. The Agency lists among the evidence considered Eppenga et al., Diabetes Care 2014;37(8):2218; Richy et al., Diabetes Care 2014;37(8):2291; and Salpeter et al., Cochrane Database Syst Rev 2010, CD002967.
Funding unknown
Rachmani R, Slavachevski I, Levi Z, Zadok B, Kedar Y, Ravid M. Metformin in patients with type 2 diabetes mellitus: reconsideration of traditional contraindications European Journal of Internal Medicine, 2002.
doi:10.1016/s0953-6205(02)00131-0 PubMed 12384131
Funding and finding
Participants
393
Follow-up
4 years
Funding
Not stated in the published abstract. The full text is behind a subscription, the article is absent from PubMed Central and no repository copy is indexed, so the funding statement could not be read and the class is undetermined.
Finding
Results read from the published abstract. Patients were randomised to continue or to stop metformin and followed for 4 years, with analysis by intention to treat. Those who stopped metformin showed a significantly greater rise in body-mass index and in HbA1c than those who continued. No case of lactic acidosis occurred in either group; lactate values did not differ between groups and correlated only with serum creatinine and body-mass index. Microvascular diabetic complications, cardiovascular events, and cardiovascular and total mortality were identical in the two groups. The authors conclude that patients who tolerate metformin may continue it as mild renal impairment develops, possibly up to a serum creatinine of 220 micromol/L, and that they see no apparent reason for patients with coronary heart disease, congestive heart failure or chronic obstructive pulmonary disease to discontinue it. They note that the strict limiting criteria for metformin stem largely from reports in the 1970s of mortality and lactic acidosis with phenformin and that data on metformin rest mainly on case reports.

Updated 2026-10-22