2. Glucose Lowering
2.1 Epidemiologic Data on Glucose and Micro- and Macrovascular Risk
Observational data and analyses from the UKPDS (United Kingdom Prospective Diabetes Study) confirmed independent positive associations between glycosylated hemoglobin (HbA1c) and vascular complications, including cardiovascular (CV) complications. Similarly, findings from large population-based observational studies have also confirmed this continuous positive association between glycemia and CV disease (CVD) risk. Subsequently, data from the HOPE study also demonstrated an independent progressive relationship between indices of glycemia and incident CV events, renal disease, and mortality. Data from 698 782 individuals from 102 prospective studies in the Emerging Risk Factors Collaboration showed that serum glucose levels were associated with an increased risk of coronary heart disease (hazard ratio [HR] 2.00; 95% confidence interval [CI] 1.83, 2.19), ischemic stroke (HR 2.27; 95% CI 1.95, 2.65), and aggregate of other vascular deaths (HR 1.73; 95% CI 1.51, 1.98). These hazard ratios did not change after adjustment for lipid, renal, and inflammatory markers. Fasting blood glucose level was non-linearly associated with vascular risk, with a relationship emerging at fasting blood glucose values >5.59 mmol/L. This relationship between glucose levels and vascular risk, more so because of the absence of any demonstrated threshold in most studies, supported the need for planning clinical trials to assess the impact of glucose lowering to normoglycemic/near-normoglycemic levels.
It is equally important to understand the precise contribution of glycemia to CV mortality in order to correctly predict the anticipated benefit of glucose lowering. Interestingly, recent findings from a Finnish observational study suggest that there are significant differences in the effect of glucose control on CV mortality between patients with type 1 and type 2 diabetes. While a 1% increase in HbA1c increased the risk of CV mortality by more than 50% in patients with type 1 diabetes, it contributed to only a 7.5% increment in CV mortality in patients with type 2 diabetes. These data suggest that a significant, and perhaps dominant, role is played by non-glycemic factors in the risk of CV disease in people with type 2 diabetes, and therefore the anticipated benefit of glucose lowering alone in these patients is unlikely to be significant.
2.2 Lessons From Three Major Recent Trials: ACCORD, ADVANCE, and VADT
Three major randomized, controlled trials designed to evaluate the impact of tight glucose control in patients with type 2 diabetes and high CV risk have been completed. This section summarizes the findings, specifically those with relevance to clinical decision making. A comparative tabulation of characteristics and outcomes of trials with glycemia as the primary target is provided in Table I .
ACCORD (The Action to Control Cardiovascular Risk in Diabetes): This study, which was exclusively conducted in the US and Canada, randomly assigned 10 251 patients (35% had previous macrovascular disease) with a mean age of 62.2 years, baseline median HbA1c of 8.1%, and median diabetes duration of 10 years, to receive either intensive therapy targeting an HbA1c of <6% or standard therapy with a target between 7% and 7.9%. This study also included BP-lowering and lipid-lowering components, details of which are provided in sections 3 and 4, respectively. The primary outcome was a composite of non-fatal myocardial infarction (MI) The study was designed to have a power of 89% to detect a 15% reduction in the rate of the primary outcome, pre-supposing a 2.9% annual primary outcome rate in the standard therapy group. Patients received instructional material, behavioral counseling, glucose-lowering medication (including rosiglitazone for all intensive arm patients), and glucose-monitoring supplies. All patients in the intensive therapy arm were seen on a monthly basis till 4 months, and then every 2 months thereafter, punctuated by at least one interim phone call. This contributed to an exceptionally rapid improvement in glucose control in the intensive therapy arm, with median HbA1c declining from 8.1% to 6.7% within 4 months. Stable median HbA1c levels of 6.4% and 7.5% were achieved by 1 year in the intensive and standard therapy groups, respectively.
Intensive therapy was associated with increased all-cause mortality (HR 1.22; 95% CI 1.01, 1.46; p = 0.04). Rates of death in the two study groups showed separation as early as after 1 year of follow-up, and this difference persisted thereafter. Epidemiologic analyses of the data suggested that factors associated with persistently high (i.e. non-responders) glucose levels, rather than lowering of these levels in the trial, may have been responsible for the higher mortality rate in the intensive therapy group. Importantly, however, the primary outcome rates were lower than predicted, with the standard therapy arm showing an annual rate of 2.29%, which was higher, although non-significantly, than that reported from the intensive therapy arm (2.11% annually) [HR 0.9; 95% CI 0.78, 1.04; p = 0.16]. The difference in rates of primary outcome between treatment groups started to emerge after about 3 years of follow-up, which may indicate that any benefit of glucose lowering may take several years to manifest. In the intensive therapy group, non-fatal MI rates were lower (3.6% vs 4.6% in the standard arm; HR 0.76; 95% CI 0.62, 0.92; p = 0.004) while rates of death from CV causes were higher than the standard therapy arm. Heterogeneity was observed for the primary outcome, so that individuals with no CV event prior to randomization, or whose baseline HbA1c was ≤8%, had fewer CV events (fatal and non-fatal) compared with patients in the standard therapy arm. Furthermore, there was no benefit on health-related quality of life from the intensive intervention, but patient satisfaction and acceptability with treatment were not compromised. In addition, body weight gain greater than 10 kg was more frequent in the intensively treated group while hypoglycemia requiring medical assistance was reported in 10.5% subjects compared with only 3.5% in the standard therapy arm (p < 0.001).
Analysis of pre-specified microvascular outcomes, including nephropathy, retinopathy, and neuropathy, showed no difference between intensive and standard therapy arms. However, intensive therapy had a beneficial effect on some measures of microvascular complications such as onset of albuminuria, prevalence of macroalbuminuria, frequency of cataract extraction, loss of ankle jerk, and light touch.
After premature discontinuation of the intensive glucose control component of ACCORD, all subjects in the intensive glucose control arm were switched to standard glycemic therapy, with an HbA1c target between 7% and 7.9%. The investigators have now released fresh data that compares outcomes in the original intensive glucose control arm (exposed to 3.7 years of tight control and 1.2 years of standard control) with subjects randomized to standard glucose control throughout the study. Subjects randomized to intensive glucose control, despite de-escalation of glycemic targets (HbA1c increasing from 6.4% to 7.2%), continued to show trends for more frequent primary outcomes including all-cause mortality (HR 1.19; 95% CI 1.03, 1.35), and lower non-fatal MI rate (HR 0.82; 95% CI 0.70, 0.96). The authors conclude that these results strongly argue against use of strategies to achieve near-normoglycemia for high CV risk type 2 diabetes subjects, although no conclusions can be drawn on the possible impact of this strategy in newly diagnosed subjects with diabetes.
Rates of hypoglycemia in the two glucose control arms were equivalent in the post-transition period, suggesting that severe hypoglycemia cannot be causally implicated in the excess mortality rates observed in subjects initially randomized to intensive glucose control. The study also reports an interaction between intensive glucose control and intensive BP control with respect to death from any cause, such that subjects randomized to both intensive glucose and BP control had marginally higher death rates compared with the standard glucose-lowering arm subjects subjected to tight BP control.
ADVANCE (Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation): This global, factorial, randomized study included 11 140 patients (32% having previous macrovascular disease), with baseline HbA1c of 7.5%, and median disease duration of 8 years. Patients were randomized to either intensive (HbA1c target ≤6.5%) or standard glucose level control, and a fixed-dose combination of perindopril and indapamide or matching placebo (details of the BP-lowering component of the study are discussed below in section 3). Unlike ACCORD, the primary endpoints were composites of major macrovascular events (non-fatal MI, non-fatal stroke, death from any cause) and major microvascular events (new or worsening retinopathy or nephropathy), assessed both jointly and separately. The study was designed to have 90% power to detect a 16% relative risk reduction for each of the primary outcomes between groups.
All intensive therapy arm subjects received gliclazide modified release, while patients randomized to the standard therapy arm were permitted to receive any sulfonylurea other than this agent. For the initial 2 years, study subjects were not provided any support for glucose monitoring. Intensive therapy arm patients were seen monthly for the first 4 months, and then 3-monthly after the first 6 months. There was no mandatory requirement for additional contact (personal or telephonic) between designated study visits, although this was encouraged for patients randomized to the intensive therapy arm. Consequently, the rate of decline in HbA1c among study participants was gradual, with a stable difference in HbA1c between the two intervention groups (6.5% in intensive therapy arm; 7.3% in the standard therapy arm) being achieved only after 3 years.
Intensive glycemic control resulted in a 10% reduction in the incidence of combined major macrovascular and microvascular events (primary outcome) [HR 0.90; 95% CI 0.82, 0.98; p = 0.01], as well as that of major microvascular events (HR 0.86; 95% CI 0.77, 0.97; p = 0.01). The difference in major microvascular outcomes was primarily consequent to a difference in new/worsening nephropathy between arms (HR 0.79; 95% CI 0.66, 0.93; p = 0.006), while no significant difference was observed in retinopathy rates. Even the substudy evaluating retinopathy using more objective criteria did not demonstrate any effect of intensive glucose lowering on new or worsening retinopathy (odds ratio [OR] 0.84; 95% CI 0.61, 1.15; p = 0.27). Intensive control did not reduce the risk of major macrovascular events or all-cause mortality. There was no heterogeneity of effects observed when analyzing different patient subgroups. A 0.7% difference in HbA1c would be expected to reduce the rate of macrovascular events by one-sixth, which is consistent with the CIs for the estimate of the effect of treatment on such events in ADVANCE. However, the study did not have adequate statistical power to detect this difference reliably, partly because the annual rate of macrovascular events (2.2%) was again lower than the anticipated rate of 3%. Severe hypoglycemia was also significantly more frequent in the intensive therapy arm (HR 1.86; 95% CI 1.42, 2.40; p < 0.001).
VADT (Veterans Administration Diabetes Trial): This is the smallest of the three major trials, involving 1791 patients (nearly all men; 40% with a history of prior CV event), with a mean age of 60.4 years, baseline HbA1c of 9.4%, and median disease duration of 11.5 years. The goal for individuals randomized was an absolute difference of 1.5% in HbA1c between study arms. The primary outcome was time to the first occurrence of a major CV event, which was a composite of MI, stroke, death from CV causes, congestive heart failure, surgery for vascular disease, inoperable coronary artery disease, and amputation for ischemic gangrene. Based on the results of the VADT pilot study, this study had 86% power to detect a 21% difference in the rate of the composite CV outcome (an expected 40% event rate over 6 years in the standard arm).
There was a rapid separation in median HbA1c levels between intervention arms, with the desired difference of 1.5% (6.9% in the intensive control arm and 8.4% in the standard therapy arm) being attained at 6 months and maintained thereafter. There was no difference between groups with regard to either the primary outcome (HR 0.88; 95% CI 0.74, 1.05; p = 0.14), any of its components, or all-cause mortality (HR 1.07; 95% CI 0.81, 1.42; p = 0.62). While there was no overall difference in the frequency of microvascular complications between groups, there was a non-significant trend toward a beneficial effect in the intensive therapy group with respect to worsening of retinopathy, worsening of albumin excretion, and progression to macroalbuminuria. There was a 3-fold, albeit non-significant, excess of sudden deaths in the intensive control arm. Similar to ACCORD and ADVANCE, the event rate observed was significantly less than that predicted – with 33.5% of standard arm patients reporting an event over 6 years compared with an expected 40%. There was no heterogeneity in the effects when evaluating pre-specified patient subgroups. Hypoglycemia was significantly more prevalent in the intensive control arm.
2.3 UKPDS and its Extension: Implications for Interpreting Recent Trials
While discussing the three 'newer' trials, it is pertinent to review the results of the UKPDS, which involved newly diagnosed patients with type 2 diabetes. At trial completion patients receiving sulfonylureas/insulin for intensive therapy showed a significant reduction in microvascular complications and a non-significant reduction (16%) in the relative risk of MI; and patients receiving metformin had a statistically significant reduction in risk of MI (39%) and all-cause mortality (36%) [Table I].
After trial completion, all surviving participants were requested to attend annual UKPDS clinics for the first 5 years and were assessed using annual questionnaires from years 6–10. As anticipated, between-group differences in HbA1c were lost after 1 year of trial cessation. In the sulfonylurea/insulin group, risk reductions for MI (15%) and all-cause mortality (13%) emerged over time, while the earlier reported risk reduction in microvascular complications persisted. In the metformin group, the risk reduction in MI and death from any cause reported at the end of the trial remained unaltered even 10 years after trial completion. The authors speculated that a legacy of good glycemic control was responsible for the benefits observed even 10 years after the intensive glucose intervention was terminated. Similar results have been reported from the extension of the DCCT (Diabetes Control and Complications Trial), conducted in patients with type 1 diabetes, wherein intensive treatment reduced the risk of CVD by 42%, and risk of non-fatal MI, stroke, or death from CV cause by 57%, once again suggesting a continued benefit of previous good glycemic control.
Consistent with other epidemiologic associations between glucose levels and microvascular disease, the UKPDS demonstrated that in new-onset type 2 diabetes patients, tight glucose control resulted in a 25% risk reduction in microvascular endpoints, predominantly retinopathy. As mentioned above, both ACCORD and ADVANCE have also evaluated the impact of intensive glucose level lowering on microvascular complications, although in patients with 8–10 years of disease. In ACCORD, while there was no difference between the glucose-lowering arms in pre-specified composite microvascular outcomes, intensive glucose control was associated individually with reduced risk for microalbuminuria, macroalbuminuria, and components of peripheral neuropathy, including loss of ankle reflexes and loss of light touch. However, a significantly higher proportion of patients in the intensive glucose control arm had doubling of serum creatinine compared with those in the standard arm (HR 1.07; 95% CI 1.01, 1.13; p = 0.0160). A beneficial effect of intensive glucose control on new or worsening nephropathy has also been reported in ADVANCE and VADT.
Data on the impact of intensive glucose control on the incidence and progression of retinopathy appears to be less consistent. Intensive glucose control was not associated with any reduction in incidence or progression of retinopathy in either ADVANCE or VADT. A subgroup of patients in ACCORD (ACCORD Eye Study, n = 2856) were evaluated for the effect of intensive glucose control on retinopathy progression or the development of retinopathy requiring laser photocoagulation or vitrectomy. Intensive glucose control resulted in a 33% reduction in likelihood of progression of retinopathy (95% CI 0.51, 0.87; p = 003), but the risk of moderate visual loss was similar in the two glucose-lowering arms. However, a similar retinopathy substudy of ADVANCE, called AdRem, did not demonstrate any benefit of intensive glucose control on incidence and progression of retinopathy. Further analysis of AdRem indicates that this study was hampered by a smaller than planned sample size, and also by a much lower event rate in the standard glucose-lowering arm (12.2% [actual] vs 27.8% [expected]) than anticipated. Intensive glucose control in AdRem resulted in a 16% reduction in incidence/progression of retinopathy (95% CI 0.61, 1.15; p = 0.27) after approximately 4 years of follow-up. Similarly, the UKPDS also showed a non-significant 17% reduction in retinopathy (incidence/progression) [95% CI 0.67, 1.01] after 6 and 9 years of follow-up. However, it was only after 12 years of follow-up that intensive glucose control in the UKPDS resulted in a marginally significant 21% reduction (0.63–1.00) in the incidence or progression of retinopathy. This comparison suggests that the results of AdRem, with respect to the effect of tight glucose control on retinopathy, are consistent with those reported in UKPDS. Even in DCCT, reduction in retinopathy risk with intensive glucose control was found only after 6 years. These data indicate that longer duration of exposure with glucose lowering is required to result in clinically relevant risk reduction in diabetic retinopathy. Furthermore, once established, these differences persist, as has been demonstrated in the 10-year post-trial follow-up of the UKPDS.
2.4 Differences Between Studies and Their Clinical Relevance
The observed excess CV deaths in ACCORD and the absence of this phenomenon in ADVANCE has been a subject of much discussion and speculation. The following factors have been considered to contribute to the excess mortality in the intensive control arm of ACCORD: magnitude and speed of reduction of HbA1c, changes in drug regimens and probable interaction between drugs, and rates of hypoglycemia. The probable role of hypoglycemia in the causal pathway for excess mortality is discussed in detail in section 2.5.
The comparison of the various therapeutic agents used in the intensive therapy arms of ACCORD and ADVANCE reveals that there was a striking difference in the use of certain agents that influence CV outcomes: HMG-CoA reductase inhibitors (statins) were used in 46% of patients in ADVANCE versus 88% in ACCORD, while aspirin (acetylsalicylic acid) was prescribed in 57% patients in ADVANCE compared with 76% of patients in ACCORD. In terms of glucose-lowering agents, rosiglitazone use was five times more frequent in the intensive therapy arm of ACCORD compared with ADVANCE, while gliclazide MR was used in nearly all intensive therapy arm subjects in ADVANCE but not at all in ACCORD. While glimepiride was prescribed in nearly 80% of intensive therapy arm patients in ACCORD, it was not used at all in subjects randomized to the intensive glucose-lowering arm of ADVANCE. The baseline HbA1c, which was lower in ADVANCE, and the pace of glucose lowering, which was significantly faster in ACCORD, could also be contributory to the differential effect of intensive glucose lowering on CV outcomes observed in these two trials. Ray et al., in a meta-analysis of these target-driven, glucose-lowering trials, opined that the optimum mechanism, speed, and extent of HbA1c reduction may contribute to the differential results seen in differing populations.
The discordance between UKPDS and the newer studies with regard to microvascular complications, especially retinopathy, has also generated interest. It is pertinent to point out that patients with newly diagnosed diabetes were recruited in UKPDS, while the mean disease duration in ACCORD, ADVANCE, and VADT ranged from 9 to 11.5 years. As already discussed above, a significant difference in retinopathy incidence and progression became apparent only after 12 years of follow-up in UKPDS, and 6 years of follow-up in DCCT among patients with type 1 diabetes. The relatively shorter duration of follow-up in all three of the more recent trials could also in part be responsible for the difference in the magnitude of effect of intensive glucose lowering seen between these trials and the UKPDS.
2.5 Hypoglycemia and Weight Gain: Impediments in Attaining Targets
One of the main impediments in achieving tight glucose control has been the heightened risk of hypoglycemia, which can mechanistically be related to CV death. Putative pathways through which hypoglycemia can exert an influence on CV death include sympatho-adrenal activation, abnormal cardiac repolarization, and vasoconstriction. In ACCORD, 10.5% of intensively treated subjects suffered from hypoglycemic episodes requiring medical assistance, raising the concern that such episodes could have contributed to the excess mortality in intensive therapy arm patients. Individuals in both intensive and standard glucose-lowering arms who suffered from one or more episodes of hypoglycemia requiring assistance had higher annual mortality than those without such episodes. Interestingly, however, among those with at least one such hypoglycemic episode, subjects randomized to the intensive therapy arm actually had lower death rates than those in the standard glucose-lowering arm. Furthermore, subjects who had poorer glycemic control at baseline had a higher risk of developing severe hypoglycemia. This analysis suggested that severe hypoglycemia did not seem to account for the difference in death rates between the two glucose-lowering arms, although a limitation of such an inference is that undetected or unrecorded hypoglycemic episodes may have been responsible for some deaths. In a similar analysis in ADVANCE participants, wherein severe hypoglycemia was documented in 2.7% of those randomized to the intensive therapy arm, it was demonstrated that while severe hypoglycemia was significantly associated with a range of adverse outcomes including major macrovascular and microvascular outcomes, it was likely to be a surrogate for vulnerability to such events and not causally related. Factors independently associated with severe hypoglycemia included older age, longer disease duration, poor renal function, lower body mass index (BMI), lower cognitive function, use of two or more glucose-lowering drugs, and assignment to intensive glucose control. Prior identification of such individuals should guide clinicians not to subject them to aggressive glucose control regimens. In the VADT, 8.5% of individuals randomized to the intensive arm reported a hypoglycemia serious adverse event, as opposed to only 3.1% of those in the standard control arm. However, there is no detailed analysis available regarding the relationship between mortality and hypoglycemia in this study.
Intensive glucose control regimens have been consistently associated with mean weight gain in excess of the standard intervention arms of up to 4 kg. This also has to be factored in while planning strategies for stringent control, because increase in body weight can result in an enhanced dose of agents used for glucose lowering and poorer quality of life and/or satisfaction reported by patients.
2.6 Absence of Significant Benefit of Intensive Glucose Control on Cardiovascular Outcomes in Target Trials (ACCORD, ADVANCE, and VADT)
In type 2 diabetes, with optimal or near optimal treatment of hyperglycemia and other CV risk factors, the additive benefits of intensive glucose control may be difficult to demonstrate. The trials really compared treating to different HbA1c levels (6.4–6.9% vs 7.0–8.4%) in the relatively flat part of the HbA1c-CV risk curve generated from observational studies. As a result, the event rates in standard glucose control populations were already lower than expected and given the small incremental difference in glycemic control achieved, observing a difference in occurrence of the primary outcomes would require massive samples and/or duration of follow-up. The absence of benefit in these comparisons does not preclude a CV benefit when reducing HbA1c from higher values (>9%) to less than 7%. Also, long duration of pre-existing diabetes and the existence of known CV risk factors and/or disease suggest a high prevalence of established atherosclerosis and a higher risk of first or repeat events in these study subjects. Subgroup analysis from ACCORD also showed that individuals with no prior CV event benefited from intensive glucose control. Interestingly, a recent 5-year longitudinal observational study from diabetes clinics in Italy has also demonstrated that patients with high levels of co-morbidity that are common in type 2 diabetes may receive diminished CV benefit from intensive blood glucose control. Notably, the choice of composite outcomes in each of these trials, while directed at ensuring adequate power, also created murkiness regarding the effects on individual outcomes.
Furthermore, extensions of earlier glucose-lowering trials (UKPDS and DCCT) that had recruited new-onset patients showed unmasking of CV benefits only after prolonged follow-up, in excess of 10 years. This prompts the speculation that intensive glucose control introduced early in the course of disease may result in CV risk reduction. Finally, the adverse effects of currently available anti-hyperglycemic strategies may counterbalance the benefits of glucose lowering, and thereby result in the lack of any CV benefit.
2.7 Meta-Analyses of Recent Trials
Several recent meta-analyses have pooled data from studies including ACCORD, ADVANCE, and VADT to analyze the cumulative impact of intensive glucose control on CV outcomes and death (Table II). Ray et al. evaluated a pooled population of 33 040 patients with 163 000 person-years of follow-up. The mean difference in HbA1c between standard and intensive control arms was 0.9% and was associated with 17% reduction in non-fatal MI and 15% reduction in coronary heart disease, but no effect on either stroke or overall mortality.
A similar meta-analysis from the Collaborators on Trials of Lowering Glucose (CONTROL) also demonstrated 9% reduction of major CV events, primarily because of a 15% reduction in risk of MI in subjects exposed to intensive glucose control. This meta-analysis also did not report any difference in mortality rates between glucose-lowering arms. Akin to what was reported in ACCORD, participants without a prior history of macrovascular disease appeared to benefit from intensive glucose control, while those with a history of macrovascular disease did not benefit from such an intervention.
Table II contains data from six meta-analyses, each of which compiled and analyzed data from between four to eight treat-to-target glycemia control trials. The data show that aggressive glycemia lowering in very high-risk type 2 diabetes patients (e.g. elderly patients, those with co-morbidities, and possibly those with pre-existing CVD events) was associated with reduced occurrence of non-fatal MIs, but no reduction of strokes or CV deaths.
2.8 Conclusions
A review of recent trials indicates that, at least in older patients with co-morbidities, intervention with intensive glucose control several years after diagnosis of diabetes does not result in any mortality benefit. Such an intervention does result in a marginal benefit in prevention of non-fatal MI, but no reduction in the likelihood of occurrence of strokes. Similarly, while intensive glucose control retarded development of new-onset nephropathy and progression to macroalbuminuria, no retinopathy benefits were identifiable. Intensive glucose control brought with it a significant increase in risk of episodes of hypoglycemia, including those requiring medical intervention for recovery. However, subsequent analysis suggested that such episodes did not support the contention that these were causal in excess mortality.
In contrast, an extension of the UKPDS, a study that was initiated in younger patients with new-onset diabetes, reported benefits with intensive glucose control on prolonged follow-up in mortality and micro- and macrovascular outcomes.