The ACCORD Trial

Christine Choy, Pharm.D., BCPS
Stuart Haines, Pharm.D., BCPS, BC-ADM
Katie Kiser, Pharm.D., BCPS
Gerstein H. Effects of Intensive Glucose Lowering in Type 2 Diabetes. N Engl J Med 2008. 358:2545-59

The primary study (glycemia) in the ACCORD trial was terminated early after concerns emerged regarding the safety of a very aggressive treatment strategy.  This landmark trial has led many to re-examine the risks and benefits of intensive glycemic control in patients with type 2 diabetes. 

Get Adobe Flash player



There has been a metanalysis (and there are always caveats about them...) that combining the data from these studies did show a benefit vis-a-vis cardiovascular disease. Note that these three studies were very short so that macrovascular benefit may not have had time to manifest such as that found in UKPDS and EDIC. Clearly, the types of patients in ACCORD and ADVANCE and VADT were similar (longer duration, other co-morbid conditions) however the increased mortality was only seen in ACCORD. The differences between ACCORD patients and the others need to be examined more closely. For example, when the ACCORD study group looked at medication differences and said that there was no difference between the patients who died and those who did not, it is my understanding that they only looked at diabetes medications. Clearly many other medications taken by patients with diabetes may have an adverse cardiovascular profile under certain circumstances.

Also, while the comorbid conditions of hypertension and dyslipidemia were evaluated, others were not. Finally, one always wants to examine the baseline characteristics of any study population. An examination of the baseline characteristics of ACCORD, ADVANCE and VADT populations finds a significant difference in weight in the ACCORD patients and the VADT patients from the ADVANCE patients. However in ACCORD, there was a lower A1C than in VADT. One might postulate a variety of things that could have happened that were not necessarily related to the actual A1C or FPG reached.

For example, if an ACCORD patient has pre-existing cardiovascular disease (seen as a high calcium point count, previous subendocardial MI that went unnoticed, etc) and, because of significantly higher weight has a higher chance of, or existence of Obstructive Sleep Apnea, then perhaps in the middle of the night an apnea episode triggered catecholamine release, the hypoxemia associated with OSA rendered the heart more sensitive as well as releasing its own cytokines and other mediators and the result was a fatal arrhythmia. Perhaps the patient was taking an SSRI that made this even worse...

Clearly there are MANY things other than the hypoglycemic agents used or the targets chosen that could have resulted in a higher mortality in the ACCORD trial. I agree with Stuart that the takeaway should be to more deeply investigate cardiovascular risk measures before determining an individual target for a given patient. Unfortunately, the generalizations that appear in the press will likely push clinicians to be more conservative and thus (perhaps) deny appropriate patients the lower glucose levels that over a LONGER period (such as UKPDS or EDIC) would result in significant cardiovascular benefit...

Excellent review of the ACCORD study. I would agree that this study was flawed in several ways. The short duration was unfortunate as I think more long term data might have found different results. I think more detail needs to be given to specific antidiabetic therapies. The points raised about the high use of insulin and TZDs in this study are an example of this. Further research on specific therapies and their impact on CVD risk might be more helpful than simply looking for the optimal A1c goal that will lower risk. In my own practice, patients who reach A1c goals of <6.5 often expereince hypoglycemia far too often or in many cases they are not aware of it making it difficult to achieve such low A1c levels. It is hard to talk about patient specific goals from a public health standpoint, as Stuart mentioned, but I understand there needs to be a general rule of thumb for providers.

However, it is not possible to only focus on individual components because the metabolic picture in these patients is often very complex with elevated A1c''s, lipids, BP, weight, and the factors we can’t change (family history, age). Just lowering the A1c to 6.2 and leaving their BP and cholesterol uncontrolled is useless (except for lowering the risk of microvascular complications) and is unlikely to reduce CVD risk. The good thing about ACCORD is it has us all talking about it and hopefully all of this hypothesis generating will lead to research that can truly shed more light on how to decrease CVD risk in this population.