The relationship of glucose levels to cardiovascular disease (CVD) risk, especially coronary heart disease (CHD), in observational data sets has been the subject of several studies. These studies have shown that the relationship between fasting (and postprandial glucose) and CHD risk is continuous and graded, and that this relationship extends below the currently defined threshold for diagnosing diabetes mellitus. The assumption has been that glycemic control in patients with diabetes mellitus should favorably affect CVD outcomes in randomized clinical trials; however, the results of several large trials have not consistently confirmed this hypothesis. In fact, ACCORD (Action to Control Cardiovascular Risk in Diabetes) data suggest a small increased risk in mortality for patients at high risk for CHD events.
The relationship of glucose levels to cardiovascular disease (CVD) risk, especially coronary heart disease (CHD), in observational data sets has been the subject of several studies. These studies have shown that the relationship between fasting (and postprandial glucose) and CHD risk is continuous and graded,1-10 and that this relationship extends below the currently defined threshold for diagnosing diabetes mellitus. The assumption has been that glycemic control in patients with diabetes mellitus should favorably affect CVD outcomes in randomized clinical trials; however, the results of several large trials have not consistently confirmed this hypothesis.11-17 In fact, ACCORD (Action to Control Cardiovascular Risk in Diabetes) data suggest a small increased risk in mortality for patients at high risk for CHD events.13
This article reviews key data from the UKPDS (United Kingdom Prospective Diabetes Study), ACCORD, ADVANCE (Action in Diabetes and Vascular Disease), and VADT (Veterans Affairs Diabetes Trial) studies, each of which were performed in subjects with type 2 diabetes, used 2 glucose-lowering strategies, had atherosclerotic vascular disease as a primary11,13,14 or key outcome,15-18 and provided data on all-cause mortality. Comparisons of the mortality rates from these and other trials have been recently published.19 Implications of these trials for glucose-lowering strategies in the clinical setting will also be provided.
Clinical trials of glycemic control and CVD outcomes
UKPDS included patients with newly diagnosed type 2 diabetes mellitus (n = 3867) with a mean age of 54 years. The study intervention used insulin or sulfonylureas in an intensive policy compared with a conventional policy in the main trial. In a parallel study in obese patients (n = 1704), there was an intensive policy arm using metformin. This trial did not have a glycated hemoglobin (HgbA1c) target. During the trial, HgbA1c increased over time in both the intensive and the conventional policy groups, with a mean in-trial HgbA1c of 7.0% and 7.9%, respectively. At the end of the study, there was a 16% reduction in myocardial infarctions (MIs) in the insulin/sulfonylureas group (P = .052).17 A 30% reduction in macrovascular disease, including MI, sudden death, angina, stroke, and peripheral vascular disease, was observed in the metformin group (P = .020 vs conventional policy).16 Post hoc epidemiological analyses reported a relationship of in-trial HgbA1c and CVD events in which a 10% (relative) change in HgbA1c was associated with a 14% reduction in MI (P <.001).15 Hypoglycemia events were very low in this study. Recently, the UKPDS long-term follow-up reported statistically significant reductions in MI and mortality in each of the intensive policy groups, suggesting glucose lowering has a “legacy” effect (Figure 1).18
The ACCORD trial included 10,251 patients with CVD (35%) or at very high risk for developing CHD (Figure 2).12,13,20 ACCORD participants had a mean age of 62 years, mean HgbA1c of 8.3%, and a mean duration of diabetes mellitus of 10 years. The intensive glucose arm had an HgbA1c target of 5.9% or less, and the standard arm had an HgbA1c target range of 7.0% to 7.9%.12,20 The actual achieved mean HgbA1c values were 6.4% and 7.5%, respectively, during most of the trial, and the HgbA1c delta was achieved within 8 months.13 This trial closed early because of a small increased risk in all-cause mortality; the intensive group had 14 deaths per 1000 patient-years versus 11 deaths per 1000 patient-years in the standard group (hazard ratio [HR], 1.22; 95% confidence interval [CI], 1.01-1.46; P = .04). Most of these deaths were related to cardiovascular causes. There was a modest reduction in the primary end point, but this did not achieve statistical significance (HR, 0.90; 95% CI, 0.78-1.04). The increased mortality was associated with the treatment regimen, but not with any particular medication or combinations of medication. Not surprisingly, there was more hypoglycemia in the intensively treated group and somewhat greater weight gain.
The ADVANCE trial included 11,140 patients who could be characterized as being at high risk for CVD (Figure 3). ADVANCE participants had a mean age of 66 years, a mean duration of diabetes mellitus of 8 years, and 32% had a history of CVD. The intensive-control intervention used gliclazide, a sulfonylurea, with add-on therapy as needed to achieve a target HgbA1c of 6.5%. The standard group used a usual-care approach. The mean HgbA1c of 6.53% was achieved by year 3 of the trial and the standard group had a mean HgbA1c of 7.3%. ADVANCE also had a slight, but statistically insignificant, reduction in major cardiovascular outcomes (HR, 0.94; 95% CI, 0.84-1.06). There was also no statistically significant difference between the 2 treatment groups in all-cause mortality (HR, 0.93; 95% CI, 0.83-1.06), and severe hypoglycemia was more common in the intensive-control group.
In June 2008, results from VADT were presented at the American Diabetes Association (ADA) 68th Scientific Sessions. The study included 1791 type 2 diabetes patients, who had a mean age of 60 years and a mean entry HgbA1c of 9.4%. A history of CHD was present in 40%. The treatment strategy included an intensive HgbA1c target of 6.5% and an HgbA1c delta of 1.5% or higher between groups. The actual HgbA1c values were 8.4% versus 6.9%. This study also showed no difference in major cardiovascular outcomes with intensive therapy (HR, 0.868; P = .11) or CV mortality (HR, 1.258; P = .36).
In the presentation at the ADA, the investigators reported that hypoglycemia was associated with increased death risk only in the standard group. The investigators conducted a post hoc analysis of clinical outcomes based on an analysis of coronary calcium scores in a subgroup of participants that showed possible benefits with more intensive glucose lowering in those with shorter duration of diabetes. A similar analysis of clinical outcomes in the VADT also suggested that intensive glucose control in subjects with shorter duration of diabetes may have CVD benefit, whereas a subject with longer duration of diabetes may have increased CVD risk. The cutoff of no benefit or risk was at approximately 12 to 15 years of diabetes mellitus.
Trial implications for glucose management in patients at risk for CHD
Although there will be a number of ongoing analyses and reports from each of the aforementioned trials, some general conclusions can be drawn from the currently available data. First, when it comes to mitigating the risk for CHD events, there is no clear benefit in reducing HgbA1c values lower than 6.5% compared with values in the 7.0% to 9.0% range, at least over periods of 3 to 6 years. Nevertheless, there is a trend toward a benefit in each of the studies, and the long-term follow-up of the UKPDS trial suggests that the modest reductions in CHD events could become statistically significant over time. UKPDS, which included participants with newly diagnosed diabetes mellitus, and the post hoc analysis of VADT in patients with shorter duration of diabetes, suggest that any benefits to reduce CHD risk may occur earlier in the disease process. Whether these observations will be confirmed by ACCORD and ADVANCE analyses has yet to be determined.
Second, ACCORD has provided some evidence of increased mortality from an intensive treatment strategy. This observation is not a function of the HgbA1c level, but some function of the treatment strategy. Currently, hypoglycemia may be associated with increased mortality (ACCORD) and CHD events (VADT). More detailed analyses will hopefully clarify these observations.
Weight gain is commonly seen with more intensive glycemic control and may contribute to mortality risk, possibly attenuating any benefits of glycemic control. Differences were observed between ACCORD and ADVANCE in terms of how rapidly nadir mean intensive HgbA1c levels were achieved (Figures 2 and 3).The rate at which glucose levels are decreased may also be a risk for adverse outcomes, including mortality. In general, these studies suggest that glucose-lowering therapies associated with low risk for hypoglycemia and weight gain should be considered preferentially in patients with CHD or at high risk for developing it.
Finally, the death rates in these trials, especially ACCORD, are quite low. The patients in these studies were usually aggressively treated with other strategies (low-density lipoprotein reduction, blood pressure control, use of aspirin), which are generally associated with reduced CHD and mortality risk. These studies indirectly affirm the importance of ensuring that CHD risk factors are aggressively treated in patients with type 2 diabetes.
Four large trials do not show benefits of glycemic control on CHD risk over periods of less than a decade. Although point estimates with lower HgbA1c values are less than 1.0, 3 recent trials do not show statistically significant reductions in atherosclerotic vascular disease events. Very aggressive glycemic intervention may be associated with increased risk of mortality. At present, hypoglycemia seems to be the most likely explanation for the observation of increased mortality.