Metformin: a 60-year odyssey with the journey still continuing – a personal commentary from Professor Ian Campbell

As Current Medical Research and Opinion (CMRO) looks forward to celebrating its 50th anniversary, I was invited by the Editor to write an article on my own research over the past 50 years following the publication of my first paper in CMRO in 1973. Although I have written over 200 original articles and reviews, as well as textbooks and book chapters over the past 50 years, dealing with diabetic autonomic neuropathy, hypoglycaemia, ketoacidosis and various oral hypoglycaemic agents, I decided I would write a personal commentary on metformin reflecting my involvement with this drug in each of the past five decades.

The metformin journey is one of discovery, dismissal and re-discovery and its very survival due to some defining moments over the past 60 years. Professor Jean Sterne introduced metformin into clinical practice in Hospital Laennec in Paris in 1957. This led to licence approval in France in 1959 and then globally over the next 30 years with the only exception being the United States. The initial lack of well controlled clinical studies led to the drug being considered as less effective than the sulphonylureas (SUs) which entered clinical practice in 1955. Metformin was fated to be the second-line drug in combination with a SU. The situation improved for metformin when three studies from Edinburgh were published between 1965 and 1977, and I was one of the clinical research team led by Dr. Basil Clarke and Dr. Leslie Duncan. The first, published in The Lancet in 1965, showed that the combination of metformin and chlorpropamide in SU failures had a synergistic effect in lowering blood glucose. The second, also published in The Lancet in 1968, compared chlorpropamide and metformin therapy on weight and blood glucose responses in uncontrolled obese subjects with type 2 diabetes mellitus (DM). Metformin was equivalent to the SU in lowering blood glucose but had better weight reduction. The third paper, published in Brit Med J in 1977, described for the first time that metformin was as effective as a SU (chlorpropamide) in blood glucose control in non-obese type 2 DM subjects¹. In 1995, in collaboration with Dr. Harry Howlett, a meta-analysis was done to look at the worldwide experience of metformin as an effective blood glucose-lowering agent, comparing metformin with SUs in published controlled, randomized prospective trials between 1957 and 1994. Metformin was as efficacious as the SUs with a fall in glycated HbA1c of 1.2% from a baseline of 9.6% for both drugs (a fall of 12.5% for both drugs). However, as most type 2 DM patients are obese, a net 5% reduction in body weight in favour of metformin was seen².

The withdrawal of phenformin in 1978 from the USA raised issues of safety (lactic acidosis) for the biguanide class of agents. Metformin, never licensed in the USA, continued to be prescribed in the neighbouring countries of Canada and Mexico and in fact was never suspended in any country worldwide. Compared to phenformin, metformin has a distinct molecular and stereochemical structure with minimal risk of lactic acidosis if prescribed correctly. It was often stated that SUs were safer than metformin because the risk of developing or possibly dying from hypoglycaemia is much less than that of developing lactic acidosis associated with metformin. I obtained from the Swedish Adverse Drugs Advisory Committee accurate data on metformin-associated lactic acidosis (MALA) and glibenclamide (a sulphonylurea) induced hypoglycaemia (SIH), over the period 1972–1981. The mortality rates were similar, 0.024 per 1000 patient-years for MALA and 0.033 for SIH³. I followed this up by doing a review of the world literature until 1982 on MALA and SIH and showed in terms of mortality SUs were not safer than metformin, with mortality rates for MALA and SIH of 2–3 cases per 100,000 patient years⁴. A Cochrane review in 2006 showed that MALA was 6.3 per 100,000 patient-years compared to 7.8 per 100.00 patient-years in subjects without metformin therapy. The risk of MALA was zero if prescribing information was followed especially avoiding the use of metformin in advanced chronic kidney disease (CKD). With ever-increasing evidence of metformin safety, regulatory authorities have licensed metformin use for type 2 DM patients with CKD with estimated glomerular filtration rate (eGFR) values of greater than 30 ml/min/1.73 m², so that since 2016 patients with CKD stages 1, 2 and 3A/3B are no longer denied the agent. Other historic contraindications dating back to the 1960s have also been relaxed for metformin such as chronic stable heart failure, allowing more patients to be treated with the drug.

If the Edinburgh studies (1965–1977) were a defining moment from the clinical efficacy and safety viewpoint in helping metformin to continue to be prescribed, the next defining moment came in 1986 when the US Food and Drug Administration approved for the first time research trials to assess efficacy, safety and the drug’s mode of action. This led to licence approval in 1994 and the drug’s use in clinical practice as both a monotherapy and in combination with sulphonylureas. Looking back, the research outcomes of these studies confirmed and expanded the comparative trials done in Edinburgh some 20 years earlier⁵. Metformin now received wider international recognition outside Europe, especially in those countries that looked to the USA for evidence-based medicine.

United Kingdom Prospective Diabetes Study

The results of the United Kingdom Prospective Diabetes Study (UKPDS) in 1998 were a seminal moment for metformin. I co-chaired the session at the EASD meeting in Barcelona when the findings were first presented and the packed audience were amazed to learn that metformin therapy, when compared to conventional therapy (diet alone), was associated with a reduction in heart attacks and all-cause mortality in overweight, newly diagnosed type 2 DM subjects. The same comparative analysis failed to show significant benefits for patients randomized to insulin or the sulphonylureas. The improvement in outcome was greater than predicted from the glucose-lowering effect of metformin, implying a pleiotropic action of the drug on the cardiovascular (CV) system. The 10-year post-trial follow-up results were presented at the EASD meeting in Rome in 2008 which I again co-chaired, and once more the audience learnt new data regarding CV protection with metformin. Early intensive management with metformin led to continuing beneficial vascular outcomes during the 10-year follow-up. The significance of this finding for clinical practice is that it showed that good early control of diabetes from diagnosis with metformin had long-term CV benefits, and this has been termed the “legacy effect” of the drug. Both the initial and the 10-year follow-up studies were published in The Lancet⁶ and N Engl J Med⁷. The vascular benefits of metformin are summarized in Box 1. These vascular effects are reviewed in detail in the scientific handbook celebrating 50 years of metformin in the chapters reviewing the mechanism and clinical outcomes of metformin and CV protection⁸.

Guidelines

Since 2002, metformin has become established as the foundation therapy in national and international guidelines for type 2 DM (IDF, ADA, EASD, NICE and more). Metformin over the past 50–60 years has become today the most widely prescribed glucose-lowering therapy for type 2 DM. In recent years, new drugs have appeared for the treatment of type 2 DM and these too have pleiotropic effects which benefit type 2 DM patients. In particular, the GLP-1 receptor agonists (GLP-1RAs) and the sodium glucose transporter 2 (SGLT2) inhibitors have been shown in outcome trials to reduce the incidence of major CV events as well as improve renal outcomes. Between 70–80% of patients recruited to the studies were on background medication with metformin. This has been recognized in recent guideline updates, resulting in continued advice to use metformin as first-line therapy along with lifestyle advice on weight management and physical activity. Today’s guidelines stratify patients into those with and without atherosclerotic vascular disease and in particular renal disease. Whilst outcome trials show benefits for the newer agents, there is evidence that metformin per se impacts beneficially on cardiorenal outcomes. Whilst often forgotten, these have been reviewed in a book published to acknowledge 60 years of clinical experience of metformin⁹. This topic was further discussed at a symposium at the EASD meeting in Barcelona in 2019 in which I participated with colleagues from Scotland and Denmark. This led to an updated review article on metformin re-appraising its benefits on cardiorenal outcomes in diabetes¹⁰. The authors concluded that, after six decades of continuous therapeutic use of metformin, further trials will better define its cardio- and reno-protective effects. Such studies are underway in both patients with prediabetes and those with type 2 DM.

The VA impact trial¹¹ in the USA will recruit almost 8000 patients with prediabetes who have atherosclerotic cardiovascular disease. Metformin will be compared with placebo over 5 years with the primary end-point being death, non-fatal myocardial infarction or stroke, hospitalization for unstable angina, or revascularization. The SMARTEST trial¹², started in 2019 and continuing until 2024, will compare metformin with a SGLT2 inhibitor (dapagliflozin) as standard treatment of early-stage type 2 DM. A total of 4300 subjects have been recruited and this open-label trial will address the efficacy of both drugs with respect to clinically important macro- and microvascular events. There is potential from real-world observational studies in subjects with CKD (stages 3A, 3B) that metformin will not only show CV benefit but also slow the progression of CKD. The RENOMET trial¹³ will assess metformin as a reno-protector of progressive renal disease. It will compare metformin versus placebo in non-diabetic subjects with CKD (stages 2–3B) with a history of progressive renal disease, with a history of proteinuria of more than 2 g/24 h. The primary end-point is the time to a 30% decline in eGFR.

Gestational diabetes mellitus and prediabetes

Gestational diabetes mellitus

A memorable moment was listening to Janet Rowan from New Zealand presenting the results from the Australasian Study, Metformin in Gestational Diabetes (MiG) at the American Diabetes Association meeting in San Francisco in 2008. This seminal trial, published in The New England Journal of Medicine¹⁴, randomized 751 females with gestational diabetes mellitus (GDM) to either metformin or insulin. Of the women randomized to metformin supplementary insulin was required in 46%. There was no evidence of perinatal complications with metformin but the latter showed less maternal weight gain. No minor or major congenital abnormalities were reported with metformin treatment. Long-term follow-up for up to 10 years of motor and social development has been reassuring. Longer-term follow-up results are awaited later in 2021 of the CLUE study¹⁵ from Finland which has assessed the consequences for the life of children with in-utero exposure to metformin, studying children from 1996 onwards (including a cohort of children born to mothers with GDM from 2004). Despite not being licensed for use in pregnancy, including GDM, since the MiG study there has been a rapid uptake of metformin use in GDM. Metformin has been endorsed internationally by various associations, organizations and societies. I have recently reviewed metformin in a book chapter published in 2020¹⁶.

Prediabetes (non-diabetic hyperglycaemia)

Prediabetes is associated with an increased risk of cardiovascular death, especially in people with existing CV disease, compared to subjects with normal glucose tolerance. The Diabetes Prevention Program (DPP) and its follow-on Outcomes Study (DPPOS) with over 20 years of follow-up showing significant benefits of both intensive lifestyle intervention (ILI) and metformin in preventing overt type 2 DM. At the American Diabetes Association meeting in June 2020, the latest 22-year average follow-up data showed a persistent reduction of type 2 DM with prior ILI of 25%, and of 18% for metformin. Metformin now has a therapeutic indication for the prevention or delay of onset of new type 2 diabetes in at-risk subjects in 67 countries. Metformin is most effective in younger and heavier subjects, and also in women with a previous history of GDM. A recent 2021 review article by Dr. Hostalek and myself in this journal¹⁷ gives an update of the evidence base of metformin for diabetes prevention with a detailed literature review, including the DPP and DPPOS findings from 2002, 2009, 2015 and 2020.

Metformin and the future

The metformin journey is not over and we have much to learn from its mode of action possibly distinct from glucose-lowering. Research is actively being pursued in the following areas.

Metformin and cancer

In 2005 Scottish and Canadian registry studies showed that metformin reduced the risk of various cancers, and further clinical observational studies worldwide with metformin reported reduced risk and improved survival in diabetic subjects with cancer¹⁸. Multiple experimental and mechanistic laboratory studies showed that metformin targets cellular pathways (AMP-kinase, mTor), of interest to cancer experts. Metformin is not licensed for treating cancer but prospective trials are underway in different solid tumours in those with and without the disease. Results are awaited to see if metformin could be beneficial as an anti-cancer drug to be used as an adjuvant to standard oncology management.

Metformin in Alzheimer’s disease

The Metformin in Alzheimer’s dementia Prevention (MAP) study¹⁹ is a randomised trial assessing if metformin can help prevent a decline in memory and thinking abilities for people with mild cognitive impairment.

Metformin and old age

The TAME (Targeting Aging with Metformin) trial²⁰ is an FDA-approved trial in the USA. Studies have shown that metformin can delay aging in animals. TAME is a multi-centre trial recruiting 3000 subjects, aged 65–79 years. Metformin will be assessed to see if it can delay the development or progression of age-related chronic diseases such as heart disease, cancer and dementia.

Conclusion

Medical history will no doubt tell us that metformin has risen like the proverbial “phoenix from the ashes”, and it can be argued that it continues to fly in unexpected ways that Professor Jean Sterne would never have imagined. My interest in diabetes has taken several directions but my involvement with metformin has given me much personal satisfaction over the years 1971–2021. I remain intrigued over the potential future developments with this drug. In the preparation of this short commentary, I have kept short the references and these reflect primarily my involvement with metformin. However, key references are included, namely the USA trial⁵, the UKPDS studies^6,7, the MiG trial¹⁴ and selected ongoing trials, all of which have influenced significantly worldwide prescribing, the awareness of the CV benefits and the successful use in GDM of metformin. The two books which I co-edited to celebrate the 50th and 60th birthdays of metformin^8,9 cover all the principal scientific and clinical literature from 1957 to 2017, and I have updated with more recent publications till 2021^10,16,17. For a broader historical overview of metformin, I have added a final reference²¹ which is an excellent comprehensive review by Professor Cliff Bailey and undoubtedly the best I have read and would recommend to the reader.

Transparency

Declaration of funding

This commentary was not funded.

Declaration of financial/other relationships

I.C. has disclosed that he has received consultancy and/or speaking fees from Johnson and Johnson, Lily, Merck and Sanofi. CMRO peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Box 1 Vascular effects of metformin

Anti-atherogenic actions

• Decreased cholesterol deposition

• Decreased lipid peroxidation

• Decreased oxidative stress

• Increased endothelial function

Anti-thrombotic actions

• Decreased platelet activation

• Increased blood flow

• Decreased PAI-1

• Increased fibrin breakdown

Anti-inflammatory actions

• Decreased C-reactive protein

Acknowledgements

I thank colleagues Professor Cliff Bailey, Dr. Caroline Day and Dr. Harry Howlett for their collaboration and support over the past 40 years. I also thank Professor Rury Holman for inviting me to chair the presentation of key results of UKPDS at four international meetings: three at EASD, one at IDF, 1998–2013. I also thank Dr. Kerstin Brand and Dr. Ulrike Hostalek of Merck KGaA for their support in recent years of my interest and writings in GDM and prediabetes. Finally, I am extremely grateful to my three mentors during my training years in diabetes in Edinburgh: Drs Leslie Duncan, Basil Clarke and John Munro. It is said that “to mentor is to invest in the future”. I thank them for investing in mine.