Bethesda, MD - Experts within and outside of the ACCORD trial are equally flummoxed by the finding that the group receiving intensive glucose lowering showed a higher mortality rate than those receiving standard care.

The announcement that the glucose-lowering part of the trial was being stopped because of this mortality difference was made yesterday by the National Heart, Lung, and Blood Institute (NHLBI), the organization coordinating the study.

Chair of the ACCORD steering committee, Dr William Friedewald (Columbia University, New York), commented to heartwire: "The simple and honest answer is that we have done extensive analyses and not identified a cause for the increased mortality. We will now do even more extensive analyses with all of our investigators, who are now unblinded to the results, and prepare a paper with the data and our best impressions of the possible causes."

In an interview with heartwire, Dr John Buse (University of North Carolina, Chapel Hill), ACCORD steering committee member and president of medicine and science at the American Diabetes Association, suggested three basic possibilities that would explain the higher mortality rate in the intensively treated group: it could have been a spurious observation, and it might have disappeared with further follow-up; it could have been due to adverse effects of a particular drug or drug combination that has not yet been teased out; or it could be that the observation is true, that lowering blood-sugar levels too much in older diabetics with heart disease is a bad thing.

He added that it would be difficult to detect an adverse effect of one or two drugs, given the large array of therapies used in the study and that the power would be very much reduced when the population is "sliced and diced" so much.

Buse's personal belief is that the increased mortality may not have been due to the level of A_1C but more to do with the intensity of the intervention. "The patients enrolled in this study were quite vulnerable in that they were relatively old (average age 62) and had heart disease or at least two or more other risk factors for heart disease. Maybe we just flogged them too hard to get their sugar levels down. The intensive group had extremely rigorous treatment, with some patients taking four shots of insulin and three pills and checking their blood-sugar levels four times a day. Perhaps this was just too many drugs at too high a dosage, and the effort required just stressed them out too much. I think our conclusion is therefore that we should not be zealots about lowering blood sugar at all costs. We must understand that there are risks and benefits and one size probably does not fit all patients," he said.

But he pointed out that both groups in ACCORD did very well compared with similar patients in real life, who have a mortality rate two to three times that seen in either group in this study. "Taking that into account, the difference in mortality between the two arms seems like a pretty small signal, but the [data safety and monitoring board] DSMB were in a very difficult position. They could have decided to carry on for longer but then would have faced criticism if the mortality difference had increased. I am not uncomfortable with the decision they made, but I do feel we need more data, and I'm glad there are other studies under way that may shed more light on this issue," he added.

And more information should become available fairly soon, as there are at least three more trials under way looking at intensive vs standard lowering of blood-sugar levels in diabetics (ORIGIN, ADVANCE, and the VA Diabetes trial), two of which are thought to be reporting later this year.

"My key message is that the ACCORD result is not a reason to panic. The kind of patients enrolled in this study are a minority in terms of diabetes patients, and the treatment given was exceptionally intensive, so the average diabetic patient need not worry about these results," Buse reiterated.

He noted that previous trials have shown mixed results on the benefits of lowering blood-sugar levels in diabetics. The DCCT trial in type 1 diabetics showed no major differences in the actual six-year study follow-up, but after 13 years of follow-up, there was a reduction in cardiovascular events in the original group whose blood sugar was lowered most. But in that study, A_1C levels went from around 9 to around 7 and so were higher than in ACCORD, and this was in a different patient population from ACCORD, he explained. And the UKPDS trial showed benefits of intensive blood-sugar lowering in terms of a reduction in retinopathy and kidney damage in patients with type 2 diabetes, and there was some suggestion of a benefit on MI but no effect on mortality. A smaller VA cooperative study had suggested an increase in mortality with more intensive glucose lowering, but this was underpowered and generally dismissed as "bad luck," Buse added.

Observers from outside the trial also had little to offer in the way of possible explanations, all saying they needed to see more data from the study in the form of a published paper. But all were surprised by the result.

Dr Steve Nissen (Cleveland Clinic) commented to heartwire: "This result really does defy conventional wisdom." Noting that benefit has been seen in lowering blood sugar in terms of diabetic complications but the effect on major cardiovascular events and mortality is not known, he added: "I suppose it wouldn't have been a major surprise if there was no effect, but to show harm is really a big surprise. This effect could have been due to some of the drugs being used to lower glucose levels, which may have other effects that cause harm. We know that rosiglitazone increases MI risk, so others may do this too." He said not enough data on rosiglitazone use in ACCORD had been released to establish whether it could have played a role in the adverse outcome. "We don't even know what percentage of people in each group were on rosiglitazone. So we can't answer that question yet. All in all, this trial has raised a lot more questions than it has answered."

Dr Rury Holman (Churchill Hospital, Oxford, UK) who led the UKPDS trial, told heartwire: "We need to see what medications the patients who died were taking and the causes of death to come up with explanations, but I suspect that because so many different treatments were used in both groups, it will be hard to come up with any clear-cut answers."

He said that the main question was whether the low sugar levels or the multiple treatments were the cause of the excess deaths and that in this regard it would be important to establish whether the patients who died had the lowest blood-sugar levels. "The intensive arm was aiming for an A_1c level of 6, and they got to an average of 6.4, so some patients would have gone quite a way below 6. It would be interesting to know if more deaths occurred in these patients who went very low. If so, then it could be the low A_1c. But they haven't released that information yet."

Holman also suggested that the observation that patients in the intensive arm of ACCORD were less likely to have an MI, but if they did have an MI it was more likely to be fatal, might provide some clues, adding that "if you have an MI, the chances of surviving depend on your background metabolic status."

Holman also emphasized the low mortality rates in the study. "These mortality rates are lower than we saw in the UKPDS trial in newly diagnosed and younger diabetic patients, so the standard of care was very good in ACCORD. For me, the message to patients remains that near-normal glucose levels are still better than high levels." He said that the UKPDS trial (in which A_1c levels were reduced from 9 to 7) had shown that lowering sugar levels did reduce microvascular complications, and that was important to remember.

Dr Robert Califf (Duke Clinical Research Institute, Durham, NC), who is cochairing (with Holman) a large diabetes-prevention trial testing nateglinide, repeated his mantra that surrogate end points are never a replacement for clinical outcomes and that ACCORD is another example of that. "I'm afraid I have to make the same points I've made over and over. An epidemiologic relationship between glucose level and risk of cardiovascular events does not mean lowering glucose levels with an intervention will have the desired effect. This is the same story as [premature ventricular complexes] PVCs and sudden death, low cardiac output and inotropes, HDL cholesterol and torcetrapib, LDL cholesterol and HRT; you name it. All this is reaffirmation that there are no shortcuts. We just don't know what we're doing until an adequate randomized trial is done."

Califf agreed with many others polled that the two main explanations for the excess mortality were probably hypoglycemic spells or other drug toxicities. "Either is possible, but from the discussion so far, I don't think we'll be able to tell," he said.

Dr Darren McGuire (University of Texas Southwestern Medical Center at Dallas) also expressed disappointment at the result. "Once again, the most sound of scientific and biologic plausibility has been refuted by a large clinical-outcomes trial, and for both patients with type 2 diabetes and for those of us who care for them, this is a very disappointing result," he told heartwire. And he made similar points to Califf's on the importance of appraising therapies with regard to overall clinical-outcome effects, regardless of how promising the science on which they are based. "We cardiologists have learned similar lessons to this time and again about the disconnect between intermediate markers of disease risk and clinical outcomes, and this has led to the critical clinical-outcomes trial appraisal of virtually every domain of cardiology. Hopefully, observations such as these, added now to the recent rosiglitazone observations and the failure of the first CETP antagonist, will continue to drive the endocrinology/metabolism clinical, pharmaceutical, and regulatory community toward a mandate for the generation and application of true clinical-outcomes data for existing and certainly for emerging metabolic therapies," he added.