Metformin in Diabetes: Evidence Overwhelming, or Unclear? 

Becky McCall

September 26, 2014

VIENNA — The extent of evidence supporting the most widely used, first-line drug for type 2 diabetes, metformin, was the topic under discussion during the Michael Berger debate on the closing day of this year's European Association for the Study of Diabetes (EASD) 2014 Meeting.

Synthesized in 1922, with its first trial in 1957, metformin has been available in Europe since 1960 but made its debut in the United States only in 1995. Arguing for the motion, Metformin: Where is the evidence? Rury Holman, MD, director of the University of Oxford Diabetes Trials Unit, United Kingdom, said it isn't clear that there is sufficient proof to support metformin use: "If it were clear, we wouldn't be having this debate."

Against the motion, claiming that the existing evidence supports first-line use of metformin, was Harold Lebovitz, MD, professor of medicine at the division of endocrinology and metabolism/diabetes at the State University of New York Health Center, Brooklyn, NY.

The Evidence for...

Dr. Lebovitz took the audience through an extensive tour of trials supporting the use of metformin, beginning with a 1995 study that showed, "It's very effective at improving HbA1c" (N Engl J Med 1995;333:541-549).

Dr. Lebovitz then referred to another paper showing that metformin is second only to thiazolidinediones in its durability of action and a lot better than sulfonylureas in this regard (N Engl J Med 2006;355:2427-2443).

Regarding the evidence for metformin's use as first-line therapy, Dr. Lebovitz cited a paper by the UK Prospective Diabetes Study (UKPDS) group that concluded: "Since intensive glucose control with metformin appears to decrease the risk of diabetes-related end points in overweight diabetic patients and is associated with less weight gain and fewer hypoglycemic attacks than insulin and sulfonylureas, it may be the first-line pharmacological therapy of choice in these patients" (Lancet. 1998;352:854-865).

This conclusion was based on a study that compared intensive therapy with metformin with that using insulin and sulfonylureas. "It was found to be no more effective at lowering HbA1c than the other agents, but in reducing diabetes-related deaths, metformin was statistically significant," he noted.

Dr. Lebovitz added that this same UKPDS study showed that cardiovascular disease accounted for 62% of the mortality in overweight patients assigned to conventional treatment, which was primarily diet alone.

And patients assigned to metformin had a significant 42% reduction in cardiovascular mortality and a 39% lower risk of myocardial infarction (MI) compared with patients assigned to diet alone. "Metformin is also weight neutral or may even cause small weight loss. It works by mechanisms that may be independent of lowering of HbA1c."

Lack of Good RCTs, Inconclusive Observational and Meta-analyses

Despite the evidence from the UKPDS, Dr. Lebovitz conceded that since this study there had not been any well-designed randomized controlled trials (RCTs) that measured outcomes with metformin.

"One study, called HOME, investigated all vascular outcomes, and no difference was found between metformin and placebo. But when incidence of macrovascular events, a secondary end point, was considered, metformin caused a highly statistically significant decrease. This confirms the UKPDS."

All other studies Dr. Lebovitz could locate were observational. "These generate hypotheses, and you really need to do randomized controlled trials to determine" whether the findings are true, he said.

Acknowledging this caveat, he shared some of the findings from observational studies, which include:

•  The hazard ratio for all-cause mortality is around 0.6 in metformin patients compared with sulfonylureas. Sulfonylureas caused a 50% to 60% increase in deaths.

•  The Reduction of Atherothrombosis for Continued Health (REACH) registry, which looked at metformin use among patients with diabetes and atherothrombosis, showed all-cause mortality at 6.3% with metformin and 9.3% without metformin.

•  A 2014 paper, a retrospective analysis of the UK Clinical Practice Research Datalink 2000–2012, looked at first-line monotherapy with a sulfonylurea vs metformin and risk of all-cause mortality and found that sulfonylureas were associated with a hazard ratio of 1.58 for mortality compared with metformin (Diabetes Obes Metab. 2014;16:957–962).

Dr. Lebovitz went on to critique meta-analyses, saying he was skeptical of the findings of a number of these that did not find in favor of metformin: "You can't compare apples to oranges to pears," he observed.

He then moved on to present evidence supporting metformin use with regard to delaying development of type 2 diabetes. The US Diabetes Prevention Program (DPP) study showed that metformin was not as good as lifestyle modification for this outcome, but half as good — particularly in young or middle-aged obese patients (Lancet. 2009;374:1677–1686). "Most data from the DPP show that it is weight loss with metformin that is the factor that delays development of diabetes," he noted.

Metformin Compared With Other Agents

Next he compared metformin with other agents for type 2 diabetes.

Thiazolidinediones have a durable effect, preserve beta-cells well, and do not cause hypoglycemia but can cause weight gain and are associated with edema, congestive heart failure, and bone fracture, he noted.

Sulfonylureas cause hypoglycemia; alpha-glucosidase inhibitors cause gastrointestinal side effects and are associated with poor compliance; dipeptidyl peptidase-4 (DPP-4) inhibitors appear to be associated with increased hospitalization due to congestive heart failure; glucagonlike peptide-1 (GLP-1) receptor agonists can cause nausea and vomiting and may be associated with acute pancreatitis; and the sodium-glucose cotransporter 2 (SGLT-2) inhibitors can cause fungal infections and have insufficient long-term data.

"I conclude the other agents do not have data to suggest they replace metformin as a first-line therapy."

Speaking particularly to the issue of hypoglycemia, Dr. Lebovitz noted that if a patient has 1 or more episodes of severe hypoglycemia they are more likely to develop a cardiovascular complication and die in the next 5 years. "I use agents that cause hypoglycemia only if absolutely necessary."

To illustrate his point, he focused on the ORIGIN/ORIGINALE trial: "This is a magnificent trial with patients treated with insulin or sulfonylurea for around 6 years, and if a patient had an episode of severe hypoglycemia, they were found to have an increased mortality in either group."

Dr. Lebovitz also mentioned a series of studies that suggest metformin might have a favorable effect on cancers.

Finally and "most important, every patient can afford metformin; it's not an expensive drug," he concluded.

The Evidence is Unclear...

Taking the podium to outline his stance, Dr. Holman first spoke to the positives, acknowledging that metformin reduces HbA1c to a similar extent as sulfonylureas and even outperforms some of the newer drugs. "There's a low risk of hypoglycemia, no weight gain, and it can reduce LDL cholesterol and has few adverse effects," he observed.

He also conceded that it is "pretty much a done deal" that metformin prevents onset of type 2 diabetes, with a study by the DPP "showing a 31% reduction in new-onset diabetes over 2.8 years" (N Engl J Med. 2002;346:393-403).

"There's a lot to like about metformin, and it's inexpensive," he commented.

However, there remain many unanswered questions, he said.

One sore point is the issue of metformin-associated lactic acidosis (MALA).

"Lactic acidosis is extremely rare (0.02 to 0.09 per 1000 patient-years), but it is fatal in 30% to 50% of cases, so use of metformin is restricted in patients with impaired renal function. It…does frighten people," he observed.

And he noted that guidance on this differs around the world.

The Food and Drug Administration (FDA) has a boxed warning on the label to signify that metformin can cause serious or life-threatening side effects. "This was driven by the increased risk of lactic acidosis as shown with phenformin, another biguanide, in the [University Group Diabetes Program] UGDP study. Because lactate is eliminated through the kidneys, it is wise to limit exposure in renal impairment, where risk of undue accumulation is high," he observed.

The FDA advises stopping metformin if the estimated glomerular filtration rate (eGFR) is less than 60 mL/min/1.73m2, but the UK National Institute for Health and Care Excellence (NICE) guidance is more liberal, indicating the dose of metformin should be reviewed if eGFR is less than 45 mL/min/1.73m2 and stopped if eGFR is less than 30 mL/minute/1.73m2.

There have been two Citizen Petitions in the United States asking the FDA to reconsider its recommendations, but the agency responded by effectively saying the issues were "unclear and 'complex,' " Dr. Holman said.

He conceded that it is unclear from existing evidence whether metformin "is causative, a passive bystander, or in some way augmenting a metabolic problem" with regard to lactic acidosis. But "nearly all cases are due to inappropriate coprescribing in patients with impaired renal function or other comorbidities."

Cardiovascular and Cancer Risk: New Trial to Start in Oxford

Dr. Holman then went on to address whether metformin reduces cardiovascular risk. There appears to be a 33% to 39% reduction in MI and a 27% to 36% reduction in deaths from UKPDS, but this study was small, involving only 753 patients, he said.

"To date, there have been numerous observational analyses, but no further randomized controlled trials. I was taught you need 2 pivotal RCTs ideally on 2 continents to change practice. UKPDS has stood the test of time, but we need another trial."

He then moved to the increased risk for cancer in type 2 diabetes, pointing out that there is an approximate doubling of the risk for pancreatic, liver, and endometrial cancer.

"Epidemiological analyses have suggested that metformin therapy can reduce this risk, but a meta-analysis of currently available short-term metformin RCTs showed no difference in cancer rates.

"What is clear is that we need more robust evidence [for metformin] — in particular with respect to reducing cardiovascular and cancer outcomes. Where there is genuine equipoise, then randomized controlled trials are needed."

Indeed, he announced, 1 such study is beginning in Oxford: a new UK multicenter cardiovascular-outcomes RCT called Glucose Lowering in Nondiabetic Hyperglycaemia Trial (GLINT).

This will enroll around 13,000 patients with elevated but not diabetic HbA1c levels who also have an estimated 10-year cardiovascular disease risk of 20% or greater; they will be randomized to receive metformin or placebo.

The primary end point is time to cardiovascular death, nonfatal MI, or nonfatal stroke, with prespecified secondary end points that include incident cancer and new-onset diabetes.

No Money to Be Made

At this point further discussion was invited from the audience and both speakers.

Dr. Lebovitz said: "We've had metformin for so long, but we haven't had another trial. The reason is because nobody can make money marketing it. A major problem in clinical diabetes is that critical studies that have no profit linked to them are unlikely to be done. We need to encourage industry and government to collaborate and answer important questions that might not make someone rich.

"I also see a problem with long-term clinical trials. If a study comes out next year to say metformin decreases cancer risk and mortality, it is unethical to continue a study to reach the end point," he continued. "This is a real problem in a 10-year study. We're going to need to take the best clinical and basic-science data and put them together to make a judgment."

One audience member from Argentina asked about the use of metformin in pregnancy.

"Data from South Africa, where metformin has been used in pregnant women for a very long time, show no evidence of abnormalities," replied Dr. Lebovitz. "The doctors who treat infertility with metformin find that you can reestablish ovulation and pregnancy, but if you stop the metformin you have a high incidence of abortion. So I would say metformin is widely used in pregnancy," he added.

Another attendee, from Denmark, noted that neither speaker had mentioned a second UKPDS metformin trial that showed an increase in all-cause and cardiovascular mortality vs standard care.

In response, Dr. Holman said: "There was no second trial, but [there was] a post hoc subgroup analysis of a group of patients in whom metformin was added to those previously randomized to sulfonylurea. At the time, it showed a statistical doubling of risk of cardiovascular death and myocardial infarction that was unexplained. The 10-year follow-up data showed this separation in the curves had in fact disappeared. This will be published shortly.

"The fact that metformin is added to nearly every [diabetes] drug licensed to improve efficacy inexpensively means that people have not accepted this [the CV findings above]," Dr. Holman added.

"I think people have assessed these data and decided this is not a real risk. But it would be good to do a study," he concluded.

Dr. Lebovitz declares he has shares in Merck, Abbott, Abbvie, and Mannkind; is on the scientific advisory boards of Biocon, Intarcia, MetaCure, and Poxel Pharma; performs occasional consultancy to AstraZeneca, Janssen, and Sanofi; he does not participate in any speaker's bureaus. Dr. Holman has declared that he receives research funding from Bayer, Bristol-Myers Squibb, and MSD and honoraria from Amgen, Bayer, Elcelyx, Janssen, MSD, Novartis, and Novo Nordisk.

European Association for the Study of Diabetes. September 19, 2014; Vienna, Austria.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.