Intensive Glycemic Control and the Prevention of Cardiovascular Events: Implications of the ACCORD, ADVANCE, and VA Diabetes Trials

What Did the ACCORD, ADVANCE, and VA Diabetes Trials Show?

The Table provides a summary of baseline characteristics, glycemic treatment strategies and goals, concomitant risk factor control, achieved glycemic control, and primary results of each of the 3 studies. The ACCORD study randomized 10 251 participants with either history of a CVD event (age, 40 to 79 years) or significant CVD risk (age, 55 to 79 years with anatomical CVD, albuminuria, left ventricular hypertrophy, or at least 2 other CVD risk factors) to a strategy of intensive glycemic control (target A_1c <6.0%) or standard glycemic control (target A_1c, 7.0% to 7.9%). Investigators used multiple glycemic medications in both arms. ACCORD participants were on average 62 years of age and had a mean duration of diabetes of 10 years, with 35% already treated with insulin at baseline. From a baseline median A_1c of 8.1%, the intensive arm reached a median A_1c of 6.4% within 12 months of randomization, whereas the standard group reached a median A_1c of 7.5%. Other risk factors were treated aggressively and equally in both groups. The intensive glycemic control group had more use of insulin in combination with multiple oral agents, significantly more weight gain, and more episodes of severe hypoglycemia than the standard group.

In February 2008, the glycemic control study of ACCORD was halted (embedded blood pressure and lipid studies are ongoing) on the recommendation of the study’s data safety monitoring board as a result of the finding of an increased rate of mortality in the intensive arm compared with the standard arm (1.41% versus 1.14% per year; 257 versus 203 deaths over a mean 3.5 years of follow-up; hazard ratio [HR], 1.22; 95% CI, 1.01 to 1.46); there was a similar increase in cardiovascular deaths. The primary outcome of ACCORD (MI, stroke, or cardiovascular death) was reduced in the intensive glycemic control group because of a reduction in nonfatal MI, although this finding was not statistically significant when the study was terminated (0.90; 95% CI, 0.78 to 1.04; P=0.16).

Exploratory analyses of the mortality findings of ACCORD (evaluating variables including weight gain, use of any specific drug or drug combination, and hypoglycemia) were unable to identify an explanation for the excess mortality in the intensive arm.¹⁰ In both study arms, participants with severe hypoglycemia had higher mortality than those without severe hypoglycemia. However, there was a complex interaction between hypoglycemia, study arm, and mortality: among participants with at least 1 episode of severe hypoglycemia, mortality was higher in those in the standard treatment arm, whereas among participants with no history of severe hypoglycemia, mortality was higher in those in the intensive treatment arm. Other prespecified subset analyses showed that participants with no previous CVD event and those who had a baseline A_1c <8% had a statistically significant reduction in the primary CVD outcome.

The ADVANCE study randomized 11 140 participants at sites in Europe, Australia/New Zealand, Canada, and Asia to a strategy of intensive glycemic control (with primary therapy being the sulfonylurea gliclizide and additional medications as needed to achieve a target A_1c of ≤6.5%) or to standard therapy (in which any medication but gliclizide could be used, with the glycemic target set according to “local guidelines”). ADVANCE participants (required to be at least 55 years of age with either known vascular disease or at least 1 other vascular risk factor) were slightly older and of a high CVD risk similar to that in ACCORD participants. However, they had an average duration of diabetes that was 2 years shorter, lower baseline A_1c (median, 7.2%), and almost no use of insulin at enrollment. The median A_1c levels achieved in the intensive and standard arms were 6.3% and 7.0%, respectively, and maximal separation between the arms took several years to achieve. Use of other drugs that favorably affect CVD risk (aspirin, statins, ACE inhibitors) was lower in ADVANCE than in ACCORD or VADT.

The primary outcome of ADVANCE was a combination of microvascular events (nephropathy and retinopathy) and major adverse cardiovascular events (MI, stroke, and cardiovascular death). Intensive glycemic control significantly reduced the primary end point (HR, 0.90; 95% CI, 0.82 to 0.98; P=0.01), although this was due to a significant reduction in the microvascular outcome (0.86; 95% CI, 0.77 to 0.97; P=0.01), primarily development of macroalbuminuria, with no significant reduction in the macrovascular outcome (0.94; 95% CI, 0.84 to 1.06; P=0.32). There was no increase in overall or cardiovascular mortality in the intensive compared with the standard glycemic control arms.¹¹

VADT randomized 1791 participants with type 2 diabetes uncontrolled on insulin or maximal-dose oral agents (median entry A_1c, 9.4%) to a strategy of intensive glycemic control (goal A_1c, <6.0%) or standard glycemic control, with a planned A_1c separation of at least 1.5%. Medication treatment algorithms were used to achieve the specified glycemic goals, with a goal of using similar medications in both groups. Median A_1c levels of 6.9% and 8.5% were achieved in the intensive and standard arms, respectively, within the first year of the study. Other CVD risk factors were treated aggressively and equally in both groups, with the trial achieving excellent blood pressure control, high levels of aspirin and statin use, and a high degree of smoking cessation.¹²

The primary outcome of VADT was a composite of CVD events (MI, stroke, cardiovascular death, revascularization, hospitalization for heart failure, and amputation for ischemia). During a median 5.6-year follow-up period, the cumulative incidence of the primary outcome was not significantly lower in the intensive arm (HR, 0.88; 95% CI, 0.74 to 1.05; P=0.12). There were more CVD deaths in the intensive arm than in the standard arm (38 versus 29; sudden deaths, 11 versus 4), but the difference was not statistically significant. Post hoc subgroup analyses suggested that duration of diabetes interacted with randomization such that participants with duration of diabetes less than ≈12 years appeared to have a CVD benefit of intensive glycemic control, whereas those with longer duration of disease before study entry had a neutral or even adverse effect of intensive glycemic control. Other exploratory analyses suggested that severe hypoglycemia within the past 90 days was a strong predictor of the primary outcome and of CVD mortality, with an association of severe hypoglycemia with all-cause mortality apparent only for participants in the standard arm. An embedded ancillary study within the main VADT showed that baseline coronary or aortic calcium scores predicted future CVD events and that intensive glycemic control significantly reduced the primary CVD end point in those with low baseline coronary artery calcium scores but not in those with high baseline scores.

What Are Potential Explanations for the Increased CVD Deaths With Intensive Glycemic Control in ACCORD?

Numerous post hoc analyses have been unable to prove or disprove causes; in fact, the design of the study renders such “proof” elusive. Randomization to the intensive arm was associated with or led to many downstream effects such as higher rates of severe hypoglycemia; more frequent use of insulin, thiazolidinediones, other drugs, and drug combinations; and greater weight gain. Such factors may be associated statistically with the higher mortality rate in the intensive arm but may not be causative. It is biologically plausible that severe hypoglycemia could increase the risk of cardiovascular death in participants with high underlying CVD risk. This might be further confounded by the development of hypoglycemia unawareness, particularly in patients with coexisting cardiovascular autonomic neuropathy (a strong risk factor for sudden death). Death resulting from a hypoglycemic event may be mistakenly ascribed to coronary artery disease, because there may not have been a blood glucose measurement and because there are no anatomic features of hypoglycemia detected postmortem. Other plausible mechanisms for the increase in mortality in ACCORD include weight gain, unmeasured drug effects or interactions, or the “intensity” of the ACCORD intervention (use of multiple oral glucose-lowering drugs along with multiple doses of insulin, frequent therapy adjustments to push A_1c and self-monitored blood glucose to very low targets, and an intense effort to rapidly reduce A_1c by ≈2% in participants entering the trial with advanced diabetes and multiple comorbidities).

Because the ADVANCE trial did not show any increase in mortality in the intensive glycemic control arm, examining the differences between ADVANCE and ACCORD supports additional hypotheses. ADVANCE participants on average appeared to have earlier or less advanced diabetes, with shorter duration by 2 to 3 years and lower A_1c at entry despite very little use of insulin at baseline. A_1c was also lowered, even more gradually, in the ADVANCE trial, and there was no significant weight gain with intensive glycemic therapy. Although severe hypoglycemia was defined somewhat differently in the 3 trials, it appears that this occurred in <3% of intensively treated ADVANCE participants for the entire study duration (median, 5 years) compared with ≈16% of intensively treated subjects in ACCORD and 21% in VADT.

It is likely that the increase in mortality in ACCORD was related to the overall treatment strategies for intensifying glycemic control in the study population—not the achieved A_1c per se. The ADVANCE study achieved a median A_1c in its intensive arm similar to that in the ACCORD study, with no increased mortality hazard. Thus, the ACCORD mortality findings do not imply that patients with type 2 diabetes who can easily achieve or maintain low A_1c levels with lifestyle modifications with or without pharmacotherapy are at risk and need to “raise” their A_1c.

Why Did None of the Trials Show a Significant Benefit of Intensive Glycemic Control on CVD in Type 2 Diabetes—in Contrast to Many Epidemiological Studies and the DCCT Follow-Up Study?

Although randomized controlled trials often confirm hypotheses grounded in observational evidence or physiological studies of surrogate end points, this is certainly not the first time that such trials have failed to do so. The results of ACCORD, ADVANCE, and VADT highlight the critical need for randomized controlled trials with meaningful clinical outcomes such as in these trials, to help answer major clinical questions.

In the 3 glucose-lowering trials, other CVD risk factors were treated to a moderate or high degree, and likely as a result of this, all had lower rates of CVD in the standard arm than originally predicted. The evidence for CVD prevention by statin therapy, blood pressure treatment, aspirin therapy in high-risk participants, and other interventions is robust. In type 2 diabetes, in which other CVD risk factors are highly prevalent, the additive benefits of intensive glycemic control might be difficult to demonstrate except in even larger or longer trials. It is likely that a real benefit of glucose lowering on CVD in type 2 diabetes, even if it could be proven, is modest compared with and incremental to treatment of other CVD risk factors.

Additionally, the 3 trials compared treatments to A_1c levels in the “flatter” part of the observational glycemia–CVD risk curves (median A_1c, 6.4% to 6.9% in the intensive arms compared with 7.0% to 8.4% in the standard arms). Their results should not be extrapolated to imply that there would be no cardiovascular benefit of glucose lowering from very poor control (eg, A_1c >9%) to good control (eg, A_1c <7%).

All 3 trials were carried out in participants with established diabetes (mean duration, 8 to 11 years) and either known CVD or multiple risk factors, suggesting the presence of established atherosclerosis. Subset analyses of the 3 trials suggested a significant benefit of intensive glycemic control on CVD in participants with shorter duration of diabetes, lower A_1c at entry, and/or absence of known CVD. The finding of the DCCT follow-up study, that intensive glycemic control initiated in relatively young participants free of CVD risk factors was associated with a 57% reduction in major CVD outcomes, supports the above hypothesis. Of note, the benefit on CVD in the DCCT–Epidemiology of Diabetes Interventions and Complications (EDIC) required 9 years of follow-up beyond the end of the DCCT to become statistically significant.

A recent report¹³ of 10 years of follow-up of the UKPDS cohort describes, for the participants originally randomized to intensive glycemic control compared with those randomized to conventional glycemic control, long-term reductions in MI (15% with sulfonylurea or insulin as initial pharmacotherapy and 33% with metformin as initial pharmacotherapy; both statistically significant) and in all-cause mortality (13% and 27%, respectively; both statistically significant). These findings support the hypothesis that glycemic control early in the course of type 2 diabetes may have CVD benefit. As is the case with microvascular complications, it may be that glycemic control plays a greater role before macrovascular disease is well developed and a minimal or no role when it is advanced.

People with type 1 diabetes, in whom insulin resistance does not predominate, tend to have lower rates of coexisting obesity, hypertension, and dyslipidemia than those with type 2 diabetes, and yet are also at high lifetime risk of CVD.¹⁴ It is possible that CVD is more strongly glycemia mediated in type 1 diabetes and that intervening on glycemia would ameliorate CVD to a greater extent in type 1 than in type 2 diabetes.

Finally, the inability of ACCORD, ADVANCE, and VADT to demonstrate a significant reduction in CVD with intensive glycemic control could also suggest that current strategies for treating hyperglycemia in patients with more advanced type 2 diabetes may have counterbalancing consequences for CVD (such as hypoglycemia, weight gain, or other metabolic changes). Results of long-term CVD outcome trials using specific antihyperglycemic drugs, intensive lifestyle therapy (such as the Action for Health in Diabetes [Look AHEAD] study), bariatric surgery, or other emerging therapies may shed light on this issue.

What Are the Implications of These Findings for Clinical Care?

The benefits of intensive glycemic control on microvascular and neuropathic complications are well established for both type 1 and type 2 diabetes. The ADVANCE trial has added to that evidence base by demonstrating a significant reduction in the risk of new or worsening albuminuria when median A_1c was lowered to 6.3% compared with standard glycemic control achieving an A_1c of 7.0%. The lack of significant reduction in CVD events with intensive glycemic control in ACCORD, ADVANCE, and VADT should not lead clinicians to abandon the general target of an A_1c <7.0% and thereby discount the benefit of good control on serious and debilitating microvascular complications.

The ADA’s Standards of Medical Care in Diabetes⁶ and the AHA and ADA’s scientific statement on prevention¹⁵ advocate controlling nonglycemic risk factors (through blood pressure control, lipid lowering with statin therapy, aspirin therapy, and lifestyle modifications) as the primary strategies for reducing the burden of CVD in people with diabetes. The lower-than-predicted CVD rates in ACCORD, ADVANCE, and VADT, as well as the recent long-term follow-up of the Steno-2 multiple risk factor intervention,¹⁶ provide strong confirmation of the concept that comprehensive care for diabetes involves treatment of all vascular risk factors—not just hyperglycemia.

The evidence for a cardiovascular benefit of intensive glycemic control remains strongest for those with type 1 diabetes. However, subset analyses of ACCORD, ADVANCE, and VADT suggest the hypothesis that patients with shorter duration of type 2 diabetes and without established atherosclerosis might reap cardiovascular benefit from intensive glycemic control. Conversely, it is possible that potential risks of intensive glycemic control may outweigh its benefits in other patients such as those with a very long duration of diabetes, known history of severe hypoglycemia, advanced atherosclerosis, and advanced age/frailty. Certainly, providers should be vigilant in preventing severe hypoglycemia in patients with advanced disease and should not aggressively attempt to achieve near-normal A_1c levels in patients in whom such a target cannot be achieved reasonably easily and safely.

The evidence obtained from ACCORD, ADVANCE, and VADT does not suggest the need for major changes in glycemic control targets but, rather, additional clarification of the language that has consistently stressed individualization:

• Microvascular disease: lowering A_1c to below or approximately 7% has been shown to reduce microvascular and neuropathic complications of type 1 and type 2 diabetes. Therefore, the A_1c goal for nonpregnant adults in general is <7%. ADA, A-level recommendation; ACC/AHA, Class I recommendation (Level of Evidence: A).*

• Macrovascular disease: in type 1 and type 2 diabetes, randomized controlled trials of intensive versus standard glycemic control have not shown a significant reduction in CVD outcomes during the randomized portion of the trials. However, long-term follow-up of the DCCT and UKPDS cohorts suggests that treatment to A_1c targets below or near 7% in the years soon after the diagnosis of diabetes is associated with long-term reduction in risk of macrovascular disease. Until more evidence becomes available, the general goal of <7% appears reasonable. ADA, B-level recommendation; ACC/AHA, Class IIb recommendation (Level of Evidence: A).*

For some patients, individualized glycemic targets other than the above general goal may be appropriate:

• Subgroup analyses of clinical trials such as the DCCT and UKPDS and the microvascular evidence from the ADVANCE trial suggest a small but incremental benefit in microvascular outcomes with A_1c values closer to normal. Therefore, for selected individual patients, providers might reasonably suggest even lower A_1c goals than the general goal of <7% if it can be achieved without significant hypoglycemia or other adverse effects of treatment. Such patients might include those with short duration of diabetes, long life expectancy, and no significant cardiovascular disease. ADA, B-level recommendation; ACC/AHA, Class IIa recommendation (Level of Evidence: C).*

• Conversely, less stringent A_1c goals than the general goal of <7% may be appropriate for patients with a history of severe hypoglycemia, limited life expectancy, advanced microvascular or macrovascular complications, or extensive comorbid conditions or those with long-standing diabetes in whom the general goal is difficult to attain despite diabetes self-management education, appropriate glucose monitoring, and effective doses of multiple glucose-lowering agents, including insulin. ADA, C-level recommendation; ACC/AHA, Class IIa recommendation (Level of Evidence: C).*

For primary and secondary CVD risk reduction in patients with diabetes, providers should continue to follow the evidence-based recommendations for blood pressure treatment, including lipid lowering with statins, aspirin prophylaxis, smoking cessation, and healthy lifestyle behaviors delineated in the ADA Standards of Medical Care in Diabetes⁶ and the AHA/ADA guidelines for primary CVD prevention.¹⁵

Appendix. Description of ACC/AHA Evidence Grading Schema.

The recommendations in this statement were originally developed by the American Diabetes Association using ADA’s evidence grading schema. The American College of Cardiology Foundation and the American Heart Association applied the ACC/AHA practice guideline evidence-grading schema as defined below to these recommendations for the convenience of their readership. Some but not all of these recommendations—though approved by the ACCF and AHA in this scientific statement—are incorporated into the formal ACC/AHA guidelines to date. Future updates of ACC/AHA guidelines may include these recommendations as deemed appropriate by the relevant ACC/AHA writing committees.

ACC/AHA Classification of Recommendations

• Class I: Conditions for which there is evidence for and/or general agreement that a given procedure or treatment is beneficial, useful, and effective.

• Class II: Conditions for which there is conflicting evidence and/or a divergence of opinion about the usefulness/efficacy of a procedure or treatment.

• Class IIa: Weight of evidence/opinion is in favor of usefulness/efficacy.

• Class IIb: Usefulness/efficacy is less well established by evidence/opinion.

Class III: Conditions for which there is evidence and/or general agreement that a procedure/treatment is not useful/effective and in some cases may be harmful.

ACC/AHA Level of Evidence

• Level of Evidence A: Data derived from multiple randomized clinical trials or meta-analyses.

• Level of Evidence B: Data derived from a single randomized trial or nonrandomized studies.

• Level of Evidence C: Only consensus opinion of experts, case studies, or standard-of-care.