Effects of Long-term Metformin and Lifestyle Interventions on Cardiovascular Events in the Diabetes Prevention Program and Its Outcome Study

Originally published2 3 May 2022


Circulation. 2022;145:1632–1641



Lifestyle intervention and metformin have been shown to prevent diabetes; however, their efficacy in preventing cardiovascular disease associated with the development of diabetes is unclear.

We examined whether these interventions reduced the incidence of major cardiovascular events over a 21-year median follow-up of participants in the DPP trial (Diabetes Prevention Program) and DPPOS (Diabetes Prevention Program Outcomes Study).


During DPP, 3234 participants with impaired glucose tolerance were randomly assigned to metformin 850 mg twice daily, intensive lifestyle or placebo, and followed for 3 years. During the next 18-year average follow-up in DPPOS, all participants were offered a less intensive group lifestyle intervention, and unmasked metformin was continued in the metformin group.

The primary outcome was the first occurrence of nonfatal myocardial infarction, stroke, or cardiovascular death adjudicated by standard criteria. An extended cardiovascular outcome included the primary outcome or hospitalization for heart failure or unstable angina, coronary or peripheral revascularization, coronary heart disease diagnosed by angiography, or silent myocardial infarction by ECG. ECGs and cardiovascular risk factors were measured annually.


Neither metformin nor lifestyle intervention reduced the primary outcome: metformin versus placebo hazard ratio 1.03 (95% CI, 0.78–1.37; P = 0.81) and lifestyle versus placebo hazard ratio 1.14 (95% CI, 0.87–1.50; P = 0.34).

Risk factor adjustment did not change these results. No effect of either intervention was seen on the extended cardiovascular outcome.


Neither metformin nor lifestyle reduced major cardiovascular events in DPPOS over 21 years despite long-term prevention of diabetes. Provision of group lifestyle intervention to all, extensive out-of-study use of statin and antihypertensive agents, and reduction in the use of study metformin together with out-of-study metformin use over time may have diluted the effects of the interventions.


URL: https://www.clinicaltrials.gov; Unique identifiers: DPP (NCT00004992) and DPPOS (NCT00038727).

Clinical Perspective
What Is New?
• During the 21-year follow-up of the 3234 DPP (Diabetes Prevention Program) participants who began with impaired glucose tolerance and were followed in the DPPOS (Diabetes Prevention Program Outcomes Study), neither metformin nor the lifestyle interventions reduced major adverse cardiovascular events compared with placebo, despite decreasing diabetes development.
• Results should be viewed in the context of the modest progression of hyperglycemia, extensive out-of-study use of lipid-lowering and antihypertensive medications, provision of a less intensive lifestyle intervention to all DPPOS participants, and increased out-of-study metformin use over time, which may both have limited the apparent effects of the interventions and have been valuable preventive strategies.
What Are the Clinical Implications?
• Despite significant long-term reduction in diabetes development, metformin and lifestyle interventions may not have additional effects on cardiovascular disease prevention in the setting of impaired glucose tolerance or early type 2 diabetes with minimal cardiovascular disease and modern glucose-lowering, lipid-lowering, and antihypertensive treatment strategies.
• Metformin and lifestyle intervention reduce the risk of type 2 diabetes, but may not provide additional protection against cardiovascular disease when glycemia, lipids, and blood pressure are well controlled.
Läs hela texten som pdf utan lösenord free
From the Discussion


In this 21-year-long follow-up study of the effects of metformin and lifestyle interventions, each of which reduced the development of diabetes compared with placebo, on cardiovascular events in an ethnically diverse cohort with prediabetes, we found no overall beneficial or unfavorable effects associated with either intervention on time to first major cardiovascular event, or to the extended cardiovascular outcome.

These metformin findings are consistent with a meta-analysis of clinical trials of metformin in type 2 diabetes35 that found no significant benefit for metformin on CVD, although there were favorable, nonsignificant, trends.

However, the original UKPDS study performed in the prestatin era, and which was not included in the meta-analysis, found that metformin significantly reduced myocardial infarction by 39% and stroke by 41% in overweight subjects with newly diagnosed diabetes.11 In addition, metformin treatment was associated with a 53% reduction in stroke over a 4-year period in an observational Taiwanese database study.36We found no significant difference in stroke events between the metformin and placebo groups.

Our earlier observation that coronary calcification, measured after 14 years of follow-up, was lower in men in the metformin than in the placebo group,25 raised the possibility that metformin would have beneficial effects on cardiovascular events. The absence of such an effect may be related to the fact that coronary calcification is a subclinical precursor of cardiovascular events, and more time may be needed to determine whether this beneficial effect of metformin on coronary calcification translates into a reduced number of events.

Overall rates for major cardiovascular events were slightly lower than national cross-sectional estimates for prediabetes and considerably lower than those for diagnosed diabetes.37 

Our cohort with prediabetes is unique because of the selection criteria used for inclusion in the study and the semiannual fasting glucose and annual oral glucose tolerance testing protocol used to diagnose diabetes at the earliest time point.

Furthermore, even after 21 years of follow-up, the degree of hyperglycemia in the cohort was mild, with mean glycohemoglobin values 6.0% to 6.1% despite the fact that 53% to 60% of participants had developed diabetes on the basis of fasting glucose and oral glucose tolerance testing in the 3 treatment groups.

Furthermore, although only 16% and 4% of participants were taking antihypertensive and statin agents at randomization, between 68% to 74% and 53% to 62%, respectively, were receiving these medications from their physicians, and the mean blood pressure was 122/73 and low-density lipoprotein cholesterol was 2.8 mmol/L (108 mg/dL) in the 3 groups at 21 years of follow-up. Effective control of major cardiovascular risk factors in cohorts with diabetes has been shown to reduce the risk of CVD.38,39 

This was therefore a relatively low-risk cohort from the standpoint of the prevention of CVD. It is thus possible that the accelerating effect of the duration of clinically diagnosed diabetes on cardiovascular risk40,41 has still not had enough time to manifest, making it more difficult to identify a beneficial effect of metformin.

The finding that lifestyle intervention had no beneficial effect on cardiovascular events, despite favorable influences on cardiovascular risk factors, may in part be because the lifestyle intervention was intensive only during DPP, after which group lifestyle was offered to all participants.

Nevertheless, our findings are consistent with the lack of effect on cardiovascular events in 2 other studies of lifestyle change on diabetes prevention, namely the 10-year Finnish Diabetes Prevention Study14 and the 20-year follow-up of the Da Qing study.15 

Lifestyle intervention also failed to show benefit despite favorable risk factor changes in the Look AHEAD trial (Action for Health in Diabetes) cohort with type 2 diabetes.42 However, a longer, 30-year follow-up of the Da Qing study did demonstrate a benefit with lifestyle for major cardiovascular events (hazard ratio 0.74 [95% CI, 0.59–0.92]).20 It should also be noted that the Da Qing cohort was a higher-risk population with a greater proportion of smokers, a higher prevalence of hypertension, and diabetes with more severe hyperglycemia and a higher overall CVD event rate.19 Thus, further follow-up of DPPOS may clarify our findings in terms of lifestyle intervention.

Unlike the studies just referred to, the current report raises the possibility of a sex difference in the effect of lifestyle on major cardiovascular events where there was borderline heterogeneity (P=0.053), with lifestyle being potentially harmful in women yet somewhat protective in men.

We have also previously reported a less favorable effect of metformin in DPP on the development of the metabolic syndrome in women compared with men.23 These sex differences merit further investigation. Similarly, the finding of a reduced number of events in men aged <45 years compared with placebo that was not significant, given our earlier findings of a beneficial effect of metformin on coronary calcification in this group,25 should be further investigated to determine whether it too is simply the result of a play of chance.

A further limitation of these analyses is the possibility that the impact of our interventions was lessened by the reduced intensity of the lifestyle intervention after the DPP phase of the study as mentioned earlier, and by the gradual reduction in adherence to study metformin over time. Another limitation is the expanded use of out-of-study metformin that may have diluted the differences between study groups, especially in those with diabetes, although sensitivity analysis failed to show such an effect. Increased use of statin and antihypertensive treatment prescribed by participants’ primary care providers overall and significantly lower use of statins and nominally lower use of antihypertensive medications in the lifestyle group may also have influenced the results, although adjustment for medication use did not alter the results. It should also be recognized that these results cannot be generalized to all people with prediabetes because we selected a subgroup with both impaired glucose tolerance and an elevated fasting glucose >5.3 mmol/L (95 mg/dL) who were at particularly high risk for diabetes development. Last, there may also have been some underestimation of nonfatal events resulting from loss to follow-up.

These findings ultimately need to be evaluated in the context of the role of metformin and lifestyle intervention in diabetes prevention. Both interventions have demonstrated long-term reduction in diabetes development in DPP/DPPOS.

Although it is reassuring that metformin was not associated with any overall unfavorable effects on CVD, it is surprising that neither intervention yielded benefit for CVD through their effect on diabetes prevention. It may be that a beneficial effect related to diabetes prevention was not apparent in our study because the development of diabetes in its very early stages may not, per se, have increased cardiovascular risk above the effect of known risk factors.


www red DiabetologNytt