Advertisement

Lower A1c Targets in Type 2 Diabetes and Increased Mortality: Causal or Casual?

Published:September 12, 2013DOI:https://doi.org/10.1016/j.amjmed.2013.04.014
      Nobody would disagree that intensive glycemic control reduces microvascular complications of diabetes mellitus, but its role in macrovascular complications is controversial. The United Kingdom Prospective Diabetes Study (UKPDS) was a randomized trial, which showed that a 0.9% reduction in hemoglobin A1c produced a statistically significant 25% reduction in microvascular complications. In contrast, the effect size for macrovascular complications was a smaller and nonsignificant 14% reduction.
      • UK Prospective Diabetes Group
      Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33).
      The UKPDS was not designed a priori to test this hypothesis and could not detect reliably the most plausible small- to moderate-effect sizes.
      • Hennekens C.H.
      • DeMets D.
      The need for large scale randomized evidence without undue emphasis on small trials, meta-analyses or subgroup analyses.
      Several larger-scale trials were undertaken, including the Action to Control Cardiovascular Risk in Diabetics (ACCORD) trial.
      • Gerstein H.C.
      • Miller M.E.
      • Byington R.P.
      • et al.
      The Action to Control Cardiovascular Risk in Diabetes Study Group
      Effects of intensive glucose lowering in type 2 diabetes.
      In ACCORD, 10,251 patients with type 2 diabetes were randomized to intensive therapy to attain hemoglobin A1c (A1c) levels of <6.0% compared with conventional therapy to attain levels of 7.0%-7.9%. The trial was terminated early, after 3.5 of the scheduled 5 years, due to no decreases in coronary heart disease events and a 22% increase in total mortality. For coronary heart disease, there were 352 events in the intensive-therapy group, as compared with 371 in the standard-therapy group (hazard ratio 0.90; 95% confidence interval [CI], 0.78-1.04; P = .16). Of 460 patients who died, there were 257 in the intensive and 203 in conventional therapy, a 22% increase (hazard ratio 1.22; 95% CI, 1.01-1.46; P = .04). This level of statistical significance would have been sufficient to declare a reliable result at the scheduled termination, but is far from the threshold established in any well-described statistical stopping guidelines for early termination. Thus, this result may not have achieved a P < .05 level of significance at the scheduled termination, so the early termination based on total mortality may have been a false-positive result.
      • Hennekens C.H.
      • DeMets D.
      Data and safety monitoring boards of randomized trials: emerging principles and practical suggestions.
      Further, there was inconsistency in various subgroup analyses, useful to formulate but not test hypotheses. For example, hypoglycemia increased mortality to a far greater degree in the conventional than the intensive group, with no temporal relationship with either the time or date of death.
      • Bonds D.E.
      • Miller M.E.
      • Bergenstal R.M.
      • et al.
      The association between severe symptomatic hypoglycemia and mortality in type 2 diabetes: retrospective epidemiological analysis of ACCORD Study.
      Further, there was no association between those with the lowest A1c levels and hypoglycemia. Instead, those with the highest A1c levels had the highest rates of hypoglycemia in both the intensive and conventional arms. Indeed, hypoglycemia was highest in patients with A1c levels of >8.0% and substantially less in those with levels <7%.
      • Miller M.E.
      • Bonds D.E.
      • Gerstein H.C.
      • et al.
      Effect of baseline characteristics glycemia treatment approach and glycated hemoglobin concentration on the risks of severe hypoglycemia: post hoc epidemiological analysis of the ACCORD Study.
      Further, in the intensive group, the highest rates of hypoglycemia occurred in subgroups whose A1c levels did not decrease. Finally, patients with the lowest A1c levels had the lowest total mortality, with a linear relationship of increases in A1c with mortality. In the conventional arm there was no association between A1c and mortality.
      • Riddle M.C.
      • Ambrosius W.T.
      • Brillon D.J.
      • et al.
      For the ACCORD Investigators
      Epidemiologic relationships between A1c and all cause mortality during a median 3.4 year followup of glycemic treatment in the ACCORD trial.
      Despite these limitations in ACCORD, a major consequence of its reported finding on mortality has been recommendations of higher glycemic goals in patients with diabetes. Specifically, the 2013 American Diabetes Association Standards of Care refer to the ACCORD finding as “concerning” and recommend higher A1c targets in patients with complications or long-standing diabetes.
      • American Diabetes Association
      Standards of medical care in diabetes – 2013.
      The existing totality of evidence, however, suggests that such recommendations are unwarranted. For example, other large-scale trials did not find an association between A1c levels and mortality. The Action in Diabetes and Vascular Disease: Preterax and Diamicron MR Controlled Evaluation (ADVANCE) trial randomized 11,140 patients with type 2 diabetes to intensive versus conventional therapy. The intensive group achieved an A1c goal of 6.5%, while the conventional arm achieved 7.3%, with no significant evidence of increases in total mortality or coronary heart disease events. In fact, those assigned to intensive treatment had significant reductions in microvascular complications (hazard ratio, 0.90; P < .01).
      • American Diabetes Association
      Standards of medical care in diabetes – 2013.
      In the Veterans Affairs Diabetes Trial, 1791 patients were randomized to either intensive (A1c = 6.9%) or conventional therapy (A1c = 8.4%). In this trial, there was a possible but nonsignificant association between intensive glycemic control and a reduction in the primary end point (hazard ratio 0.88; P = .14).
      • Holman R.R.
      • Paul S.K.
      • Bethel M.A.
      • et al.
      10 year follow-up of intensive glucose control in diabetes.
      In addition, there was no significant association between intensive control of A1c and increased total mortality (hazard ratio 1.07; 95% CI, 0.8-1.42; P = .62). Thus, in the context of the totality of evidence, it seems more plausible that improving A1c to 6.5%-7.0% by intensive treatment is not associated with increased total mortality. With respect to safety, there is no reliable randomized evidence to support the conclusion that higher A1c levels are safer for any group of patients with type 2 diabetes.
      In long-term observational follow-up of UKPDS and the Diabetes Control and Complications Trial (DCCT),
      • Miller M.E.
      • Bonds D.E.
      • Gerstein H.C.
      • et al.
      Effect of baseline characteristics glycemia treatment approach and glycated hemoglobin concentration on the risks of severe hypoglycemia: post hoc epidemiological analysis of the ACCORD Study.
      • Riddle M.C.
      • Ambrosius W.T.
      • Brillon D.J.
      • et al.
      For the ACCORD Investigators
      Epidemiologic relationships between A1c and all cause mortality during a median 3.4 year followup of glycemic treatment in the ACCORD trial.
      reduction of hyperglycemia produced significant benefits on microvascular and macrovascular complications. The totality of evidence suggests that the optimal A1c to reduce microvascular complications is 6.5%. It is plausible that little further benefit may accrue below that level, although there is wide variation in rates of protein glycation between individuals, so A1c targets should be individualized.
      • Miller M.E.
      • Bonds D.E.
      • Gerstein H.C.
      • et al.
      Effect of baseline characteristics glycemia treatment approach and glycated hemoglobin concentration on the risks of severe hypoglycemia: post hoc epidemiological analysis of the ACCORD Study.
      Of greater importance, however, is the selection of antihyperglycemic agents. Hypoglycemia seems to be associated with substantially increased morbidity and mortality in multiple randomized trials of patients with diabetes. Thus, agents should be selected alone or in combination based on having the least intrinsic risk of hypoglycemia. Such a strategy would minimize many disease-related complications of diabetes without producing iatrogenic complications, which include several life-threatening conditions. In this regard, more than 85% of type 2 diabetic patients in ACCORD were taking agents from each of the 3 major oral agent classes, including sulfonylureas. In addition, 77% were taking insulin.
      • Miller M.E.
      • Bonds D.E.
      • Gerstein H.C.
      • et al.
      Effect of baseline characteristics glycemia treatment approach and glycated hemoglobin concentration on the risks of severe hypoglycemia: post hoc epidemiological analysis of the ACCORD Study.
      Based on the totality of evidence, the serious and life-threatening risks of hypoglycemia are definite, but mortality hazards of lowering A1c below 7.0 to 6.5 are not.

      References

        • UK Prospective Diabetes Group
        Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33).
        Lancet. 1998; 352: 837-853
        • Hennekens C.H.
        • DeMets D.
        The need for large scale randomized evidence without undue emphasis on small trials, meta-analyses or subgroup analyses.
        JAMA. 2009; 302: 2361-2362
        • Gerstein H.C.
        • Miller M.E.
        • Byington R.P.
        • et al.
        • The Action to Control Cardiovascular Risk in Diabetes Study Group
        Effects of intensive glucose lowering in type 2 diabetes.
        N Engl J Med. 2008; 358: 2545-2559
        • Hennekens C.H.
        • DeMets D.
        Data and safety monitoring boards of randomized trials: emerging principles and practical suggestions.
        Clin Invest. 2011; 1: 1-5
        • Bonds D.E.
        • Miller M.E.
        • Bergenstal R.M.
        • et al.
        The association between severe symptomatic hypoglycemia and mortality in type 2 diabetes: retrospective epidemiological analysis of ACCORD Study.
        BMJ. 2010; 340: b4909
        • Miller M.E.
        • Bonds D.E.
        • Gerstein H.C.
        • et al.
        Effect of baseline characteristics glycemia treatment approach and glycated hemoglobin concentration on the risks of severe hypoglycemia: post hoc epidemiological analysis of the ACCORD Study.
        BMJ. 2010; 340: b5444
        • Riddle M.C.
        • Ambrosius W.T.
        • Brillon D.J.
        • et al.
        • For the ACCORD Investigators
        Epidemiologic relationships between A1c and all cause mortality during a median 3.4 year followup of glycemic treatment in the ACCORD trial.
        Diabetes Care. 2010; 33: 983-990
        • American Diabetes Association
        Standards of medical care in diabetes – 2013.
        Diabetes Care. 2013; 36: S11-S66
        • Holman R.R.
        • Paul S.K.
        • Bethel M.A.
        • et al.
        10 year follow-up of intensive glucose control in diabetes.
        N Engl J Med. 2008; 359: 1577-1589

      Linked Article

      • Lowering Blood Glucose Without Increasing Mortality
        The American Journal of MedicineVol. 127Issue 5
        • Preview
          A recent editorial in the Journal discussed the question of increased mortality among patients whose blood glucose levels were intensively decreased to an A1c goal of 6.5% or less.1 This risk was attributed partly to extensive use of sulfonylureas and insulin, which tend to cause harmful hypoglycemic episodes. The authors advised treating diabetic patients with medications having “the least intrinsic risk of hypoglycemia.” Such medications exist but often are insufficient to achieve intensive blood glucose lowering, even when they are combined.
        • Full-Text
        • PDF