This article is part of this issue's CME feature

Diabetes mellitus is a significant risk factor for cardiovascular morbidity and mortality, and its prevalence has reached epidemic proportions in the United States.^1,2 Dyslipidemia, high blood pressure, a family history of diabetes, obesity, insulin resistance, polycystic ovary syndrome, and physical inactivity are risk factors for diabetes.³ The metabolic syndrome, which is characterized by obesity, hypertension, insulin resistance, and dyslipidemia, carries significant risk.⁴ The 2 most important methods of decreasing new-onset diabetes are lifestyle modifications, such as diet and exercise, and treatment with metformin (Fortamet, Glucophage, Riomet), which decreases insulin resistance and weight.⁵ A recent meta-analysis evaluated the metabolic effects of metformin treatment in individuals at risk for diabetes.⁶

Subjects and methods

Electronic databases and selected journals were reviewed for randomized controlled trials that compared metformin treatment with placebo or no treatment in individuals at risk for diabetes; trials were included if they lasted 8 weeks or longer and provided information on the development of new-onset diabetes or supplied metabolic parameters. Net treatment effects were pooled for each variable measured to determine weighted mean differences with 95% confidence intervals (CIs). We obtained a summary odds ratio and absolute risk difference by pooling the proportion of subjects with new-onset diabetes to subjects without diabetes from each trial. Subgroup analyses evaluated the differences in results for subjects with and without obesity (body mass index [BMI] >30 kg/m²) and for those with and without polycystic ovary syndrome.

Results

Of the 2000 studies identified by the search, the inclusion criteria were met by 31 studies, with a total of 4570 participants followed for 8267 patient-years.⁶ The average length of the trials was 1.8 years, with an average metformin dose of 1.6 g/day (range, 500-2550 mg/day).

Among subjects receiving metformin, BMI decreased by 5.3% (95% CI, 4.0%-6.7%) compared with placebo and 5.9% from baseline. Statistically significant reductions in BMI were observed in individuals with and without polycystic ovary syndrome, with and without obesity, and in those taking a higher-than-average or lower-than-average daily dose of metformin.

Low-density lipoprotein (LDL) cholesterol levels decreased by 5.6% (95% CI, 3.0%-8.3%) and high-density lipoprotein (HDL) cholesterol levels increased by 5.0% (95% CI, 1.6%-8.3%) among those taking metformin. Furthermore, triglyceride levels decreased by 5.3% (95% CI, 0.03%-10.5%), and the LDL/HDL ratio decreased by 8.5% (95% CI, 2.6%-14.0%).

Analyses among subgroups showed no significant differences.

Figure. The effect of metformin on the incidence of diabetes in persons at risk for diabetes.

As shown in the Figure, compared with no treatment or placebo, metformin treatment reduced new-onset diabetes by 40%, with a pooled odds ratio of 0.6 (95% CI, 0.5-0.8). Metformin was shown to decrease the absolute risk of diabetes by 6% (95% CI, 4%-8%) over an average period of 1.8 years, when all trials with and without new cases of diabetes were pooled together.

Discussion

Based on data from 31 randomized controlled trials, HDL cholesterol levels were significantly increased, and triglyceride levels, LDL cholesterol levels, BMI, fasting glucose levels, and insulin resistance were significantly decreased among subjects at risk for diabetes who received metformin treatment. Furthermore, the risk of developing new-onset diabetes was decreased by 40% with metformin, with an absolute risk reduction of 6% over the trial duration. This indicates that only 17 subjects would need to be treated for 1.8 years to prevent 1 case of diabetes mellitus.

The trials in this meta-analysis assessed different populations of persons with risk factors for diabetes, including dyslipidemia, high blood pressure, peripheral vascular disease, polycystic ovary syndrome, obesity, abdominal obesity, glucose intolerance or insulin resistance, a family history of diabetes, and metabolic syndrome. Subjects with and without obesity as well as those with and without poly-cystic ovary syndrome experienced similar improvements in metabolic parameters; these results were observed in adults, adolescents, and children alike.

Other favorable cardiovascular benefits were shown with met-formin treatment in the trials we evaluated, including decreases in tissue-type plasminogen activator antigen, urinary albumin excretion, the incidence of metabolic syndrome, left ventricular mass, systolic and diastolic blood pressures, waist-to-hip ratio, waist circumference, and visceral fat mass.⁶ Metformin has been extensively studied in patients with polycystic ovary syndrome. Patients with this condition generally exhibit insulin resistance, obesity, dyslipidemia, hyperandrogenism, infertility, and anovulation. Metformin has been shown to stimulate ovulation, increase pregnancy rates, and markedly decrease systolic and diastolic blood pressures among these patients.

Because most of the trials included in the meta-analysis recommended lifestyle interventions, such as diet and exercise, in the treatment and control groups, the metabolic effects shown in our analysis were a result of treatment plus lifestyle modification, although the extent to which lifestyle modification was followed is not known. Two trials specifically compared the effects of metformin treatment with lifestyle modification on the development of new-onset diabetes.^5,10 The pooled results show that lifestyle modification was significantly more effective than treatment with metformin. Compared with either metformin treatment alone or lifestyle modification alone, the greatest decreases in weight occurred with the combination of metformin treatment and intensive lifestyle modifications.¹²

It is not known whether metformin reduces diabetes strictly through its effect on weight or whether other mechanisms are at work, such as a direct action on insulin resistance. Subgroup analysis of this meta-analysis showed that metformin produced equivalent weight loss in nonobese and obese patients. Obesity and insulin resistance usually precede the development of diabetes, and weight loss through lifestyle modifications decreases insulin resistance and new-onset diabetes, such that a 1% decrease in weight results in a 5% reduction in insulin resistance and a 5% to 15% decrease in diabetes.⁵ In this meta-analysis, for each 1% decrease in BMI, metformin decreased insulin resistance by 5% and new-onset diabetes by 8%, indicating that the main mechanism of action of metformin is probably achieved through weight loss.

Metformin has a direct effect on reducing fasting blood glucose levels, so it is possible that the diagnosis of diabetes in the trials was affected by a direct pharmacologic action of metformin on glucose. One study evaluated that effect by measuring glucose tolerance tests 2 weeks after the discontinuation of the trial. Compared with placebo, metformin still significantly decreased the risk of diabetes, even after the direct pharmacologic effects had resolved.¹³

There is some evidence that, over time, the metabolic changes that occur in individuals at risk for diabetes and taking metformin decrease cardiovascular risk.¹⁴ This effect would not be unexpected, as obesity, dyslipidemia, insulin resistance, hyperglycemia, and diabetes increase the risk of cardiac events and death. Metformin treatment among individuals at risk for diabetes might possibly increase life expectancy, at a cost of $1800 per quality-adjusted life-year gained, according to a cost-effectiveness analysis.¹⁵

Metformin has been shown to be safe since its introduction 50 years ago.¹⁶ The chief side effects are mild nausea, abdominal discomfort, and diarrhea. Some patients treated with metformin, as well as those treated with sulfonylureas and insulin, have been reported to develop lactic acidosis, which is a rare metabolic disorder associated with such serious metabolic conditions as renal failure and shock.¹⁷ A meta-analysis assessing the association between the risk of lactic acidosis and metformin showed no cases of lactic acidosis in almost 50,000 patient-years of metformin use.¹⁷ Approximately 96% of the trials had allowed for the inclusion of at least 1 of the standard contraindications to metformin. This indicates that there is no evidence that metformin actually increases the risk of lactic acidosis under the present conditions of use.

Another class of drugs used for the treatment of diabetes is thiazo-lidinediones (rosiglitazone [Avan-dia] and pioglitazone [Actos]). Like metformin, these agents decrease insulin resistance and can decrease the risk of new-onset diabetes. One study showed a 60% decrease in the development of diabetes over 3 years among individuals at risk for diabetes who took rosiglitazone.¹⁸ Compared with placebo, however, treatment was associated with a 7-fold increase in congestive heart failure and a 3% increase in weight. Treatment with rosiglita-zone in patients with diabetes was shown to be associated with a nonsignificant trend toward increased mortality from all cardiovascular causes and a significant 45% increase in myocardial infarction in a recent meta-analysis.¹⁹

Conclusion

Significant improvements in weight, lipid profiles, fasting glucose levels, and insulin resistance, as well as a 40% decrease in the progression to new-onset diabetes over time, have been shown with metformin treatment among individuals at risk for diabetes. Studies have shown that the most successful strategy to decrease weight and the incidence of diabetes is to implement intensive lifestyle modifications in persons at risk for diabetes, although long-term sustained effects are difficult to achieve. The available evidence indicates that metformin now can be added to lifestyle interventions as an approach to the prevention of diabetes. The results of new long-term studies will show whether the metabolic improvements achieved with metformin treatment will ultimately result in a decrease in cardiovascular morbidity and mortality.

DISCLOSURE: The author has no relationship with any commercial entity that might represent a conflict of interest with this paper.

References