Statistics from Altmetric.com
Editor,—We write in response to the editorial “Diabetes and coronary artery disease: time to stop taking the tablets”.1 The authors highlight previous studies where diabetic patients treated with sulphonylureas have an excess cardiovascular mortality during myocardial infarction compared with diabetic patients treated by other means. As Connaughton and Webber point out ischaemic preconditioning has refocused our attention on these trials.
The profound protective effects of ischaemic preconditioning are thought to be mediated by opening of a KATP channel, while the hypoglycaemic action of sulphonylureas is mediated by closure of these channels within the membrane of β cells in the islets of Langerhan. The authors suggest that it may be simply a case of “adding a potassium channel opener along with insulin during [myocardial infarction] MI” to improve outcome in diabetics presenting with infarction. However, recent findings suggest this is an oversimplification and probably incorrect.
One problem is that, there are at least two different KATP channels within cardiac myocytes. Evidence is emerging that it is the mitochondrial and not the cell membrane KATPchannels that initiates the cardioprotective effects of preconditioning. This conclusion is based on recent work from Marban’s group.2-4 These investigators show that diazoxide, an agonist that opens mitoKATP channels > 1000-fold more potently than their surface counterparts in heart cells, cardioprotects at concentrations that only open the mitoKATP channels. In addition, at this low concentration of diazoxide a specific mitoKATP channel blocker abolishes myocyte protection.
Although nicorandil opens membrane KATP channels, to our knowledge it is not known whether it activates the mitochondrial KATP channel. Indeed its efficacy in treating patients with symptomatic coronary artery disease may equally relate to the fact that nicorandil is a nitrate.
Coronary angioplasty is thought to be a surrogate model of ischaemic preconditioning in man. It has been shown by several groups that the first balloon inflation can protect the myocardium against ST depression in subsequent inflations. The author cites Tomaiet al’s paper5 as evidence that the KATP channel is pivotal in protection in this model. They demonstrated that pretreatment with glibenclamide abolished the protection afforded during angioplasty. However, this model has its limitations; first, the preconditioning may not be caused by endogenous adaptation but rather opening of myocardial collateral vessels during the initial ischaemia. Glibenclamide is also known to inhibit vasodilatation in vascular smooth muscle and could therefore be preventing coronary collateral recruitment. Glibenclamide also has direct electrophysiological effects, as opening of membrane KATP channels shortens the action potential causing ST segment shift, the index of depth of ischaemia in this study. As Connaughton and Webber point out, the concentrations of sulphonylureas required to activate cardiac KATP channels is between 100 and 1000 times higher than those required to induce pancreatic insulin release. These observations raise serious doubts as to whether glibenclamide used to treat diabetes will block ischaemic preconditioning.
In conclusion, Connaughton and Webber suggest the need for clinical trials to support the theoretical superiority of insulin. Such a trial has now been published—3867 newly diagnosed diabetics were randomly assigned to sulphonylurea, insulin or diet alone.6 Over a 10 year follow up the outcome of these treatments were compared and no difference was found in the rate of myocardial infarction or diabetes related death between participants assigned sulphonylurea or insulin treatment.
We agree with Connaughton and Webber that the time to stop taking the tablets is therefore “not yet”.
This letter was shown to the authors, who reply as follows:
We are grateful to Edwards et al for drawing attention to studies pertinent to our editorial that have been published since it was written. The report of the UK prospective diabetes study is especially valuable and answers one arm of a hypothesis we considered, finding no difference in long term cardiovascular outcome between diabetic patients treated with insulin or sulphonylureas. This provides strong evidence against our speculation that diabetic patients with coronary disease should not be treated with sulphonylureas.
We also suggested that adding a potassium channel opener to insulin treatment in the setting of acute myocardial infarction might be beneficial. Edwards and colleagues suggest this rationale is based on an oversimplification, and is therefore probably incorrect. This conclusion does not necessarily follow from their premise. It is virtually a truism than any conjecture in science proves sooner or later to be an oversimplification. We would certainly acknowledge that current understanding of the biology of the KATP channel is incomplete, as signalled by the very recent reporting of a mitochondrial KATP channel, to which Edwardset al allude. They are likewise justifiably cautious about drawing inferences from models of preconditioning such as human coronary angioplasty, and we made it plain that our assessment of such evidence carried significant qualifications. It would nonetheless be a mistake to confuse the limitations of current understanding with attempts to find improved treatment strategies. Nicorandil is effective in both stable and unstable angina,1-1 and its underlying mode of action may be important in endogenous myocardial protection against ischaemia. Its antagonists can block such protection in animal and human models, and have been shown to be inferior to insulin when considering prognosis after myocardial infarction. It therefore continues to seem reasonable to us to investigate nicorandil’s effect in diabetic patients who have a high incidence of coronary disease and worse than average consequences from this.
It may indeed be oversimplistic—or even wrong—to suggest that opening potassium channels in myocardial ischaemia or infarction is a good thing, and closing them is a bad thing. This does not mean that such a hypothesis cannot stimulate a useful clinical study, and it was this for which we argued. As Edwards and colleagues are no doubt aware, the way to support or refute a clinical hypothesis is to do the study rather than to predict its outcome from other data. A main purpose of our speculations was to stimulate debate, and we find it gratifying that Marber’s group and ourselves have come to similar conclusions from rather different directions.