All patients with symptoms consistent with an acute MI of >30 min duration, presenting within 24 hour after the onset of symptoms and with a ST-segment elevation of more than 1 mm (0.1 mV) in two or more contiguous leads on the electrocardiogram and treated with primary percutaneous coronary intervention (PCI) were included in this study. At baseline age, gender, previous cardiovascular disease defined as a history of coronary artery bypass grafting (CABG), previous PCI, stroke and MI, existence of DM, smoking status, Killip class, electrocardiographic site of infarction, time of onset of symptoms, and time of hospital admission were recorded. Patients were defined as diabetic when treated with a diet, oral hypoglycaemic drugs and/or insulin. The research protocol was reviewed and approved by the medical ethics committee, and patients were included after informed consent.

Glucose levels were based on measurements of whole-blood glucose (Modular System, Roche/Hitachi, Basel, Switzerland). The first glucose available after admission was defined as admission glucose. To determine the time-averaged glucose level for an individual patient, a dedicated computer algorithm interpolated all glucose measurements into a curve, after which the area under this glucose curve was calculated for the first 48 hours (persistent hyperglycemia). The area under the curve was then divided by 48 hours.

Enzymatic infarct size was estimated by serial measurements of creatine kinase (CK) fraction. CK was determined enzymatically on a Hitachi 717 automatic analyzer according to the International Federation of Clinical Chemistry (IFCC) recommendation at 30 degrees Celsius. Frequent CK determinations were performed according to a schedule that called for 4 to 8 measurements in the first 96 hours to calculate the area under the CK curves.

Left ventricular ejection fraction (LVEF) was measured before discharge by radionuclide ventriculography or by echocardiography. Radionuclide ventriculography was performed with the multiple-gated equilibrium method following the labeling of red blood cells of the patient with technetium (^99mTc-pertechnate). A General Electric 300 gamma-camera with a low-energy all-purpose parallel-hole collimator was used. Global ejection fraction was calculated by a Star View computer (General Electric, Wisconsin, USA) using the fully automatic PAGE program. LVEF as assessed by two-dimensional transthoracicechocardiography was reported as a descriptive grade of function, using subjective visual assessment by two independent observers. This approach is less time-consuming than other methods, such as Simpson's rule or the wall motion index score. Nevertheless, studies of subjective visual assessment of LVEF suggest that this approach can be at least as accurate as other methods 15. A LVEF ≤ 30% was defined as a poor left ventricular function before the analysis 16.

In all patients data were obtained with respect to mortality, recurrent myocardial infarction or repeat PCI during the first 30 days after admission. The primary endpoint of the study was the presence of a major adverse cardiac event (MACE) during the first 30 days after admission for acute MI. MACE was defined as the composite incidence of death, recurrent infarction, repeat PCI or a LVEF ≤ 30%. The combination of death and non-fatal major events as recurrent infarction, repeat intervention, and heart failure has been shown to be a valid predictor of 1-year mortality 1718. If in one patient more than one event occurred, these events accounted for one MACE. Recurrent myocardial infarction was defined as the occurrence of symptoms consistent with an acute MI of >30 min duration, signs of infarction on the electrocardiogram, and a second increase in serum CK level to more than twice the upper limit of normal. If the CK level had not decreased to normal levels, a second increase of more than 200 IU per litre over the previous value was regarded as indicative of a recurrent infarction 19. Repeat PCI was defined as angioplasty performed within 30 days due to repeat signs and symptoms of myocardial ischemia. If an event occurred within 48 hours after admission persistent hyperglycemia was calculated based on the values available before the event.

Differences between groups were assessed with the Student's t-test or the Mann-Whitney U-test. The Chi-square test and the Chi-square test for trend were used to test differences between proportions. Receiver operator characteristic (ROC) curves were computed to assess the ability of glucose-derived parameters to predict MACE or mortality. We also performed a Cox proportional-hazards regression model with factors at admission related to MACE at least with a single-sided level of significance of 0.2 or less. The Statistical Package for the Social Sciences (SPSS Inc., Chicago, IL, USA) version 11.5 was used for all statistical analysis.

Several mechanisms can be postulated to explain the relation between acute and persistent hyperglycaemia and an unfavourable outcome after reperfusion therapy for MI. First, the relation between persistent hyperglycaemia and outcome may be related to the presence of ongoing stress in more severely ill patients. During myocardial ischemia an increase in glucose levels is observed 21. It has been reported that glucose was related to adrenalin and cortisol in patients with ST-segment elevation MI 22. In the current study, Killip class ≥ 2 on admission was more frequent and enzymatic infarct size was larger, suggesting that hyperglycaemia reflects extensive myocardial damage.

The second possible explanation is that patients with hyperglycaemia are likely to have diabetes, even if it has not been diagnosed 78923. It has been described that patients with DM have an impaired outcome after MI. Potentially, patients with persistent hyperglycaemia are more likely to have DM either known or unknown 911. In our study, patients with DM had indeed an elevated glucose at admission and a persistent hyperglycemia. The number of patients without DM in the highest quartile of admission glucose and persistent hyperglycemia was approximately 2 out of 3 patients. Diabetes is associated with extensive coronary artery disease and adverse outcomes in MI patients. In the current study, elevated admission glucose and persistent hyperglycemia were related to a higher prevalence of DM. Persistent hyperglycemia may reflect insulin resistance. Insulin resistance is more often present in non-diabetic patients with an acute MI compared to matched controls 24. When 181 patients admitted with acute MI and no history of diabetes mellitus were compared to 180 matched controls without previously known diabetes or cardiovascular disease, glucose, HbA1c, proinsulin, proinsulin/insulin ratio, triglycerides, insulin resistance and fibrinogen were all consistently higher in patients than controls (P < 0.01). Recently, in a retrospective study, it was shown that non-diabetic patients with signs of insulin resistance had impaired left ventricular function recovery after coronary angioplasty compared to patients without insulin resistance 25.

Finally, it is possible that hyperglycemia and concomitant metabolic abnormalities may exacerbate myocardial damage in MI. A recent study in 4102 non-diabetic patients admitted for heart failure found a strong association between admission glucose and short-term mortality 26. As in our study, the heart failure study does not allow conclusions whether hyperglycemia was an indicator or cause of adverse outcome.

The obvious clinical difference between admission glucose and persistent hyperglycemia is that hyperglycemia during hospital stay is amenable to therapy, whereas hyperglycemia already present on admission cannot be modified. Therefore, in the clinical setting, the ultimate proof that the hyperglycemic state may have harmful effects on the ischemic myocardium is to demonstrate that cardiac outcomes can be improved by strict glucose regulation. The beneficial effect of an infusion of insulin during myocardial ischemia has been suggested by a small number of clinical studies, although they lacked a randomized design and were not conclusive 272829. The only large trials to investigate treatment focussed on hyperglycemia in MI patients were the Diabetes Insulin-Glucose in Acute Myocardial Infarction (DIGAMI) studies. The DIGAMI included 620 patients with known diabetes mellitus or serum-glucose concentrations of over 11.0 mmol/L 3031. Patients were randomized for an insulin-glucose infusion for 24 hours, followed by a minimum of 3 months of intensive insulin therapy or conventional treatment. After 1 year, patients that had no previous insulin treatment gained most by glucose-metabolism intervention, with a mortality of 8.6 versus 18.0% in the control group. For the study population as a whole, the authors found an absolute reduction in mortality of 7.5%. The recently published DIGAMI 2 randomized 1253 MI patients with hyperglycemia to 24 hour insulin-glucose infusion to obtain glucose levels between 7.0 and 10.0 mmol/L and subcutaneous long-term insulin treatment thereafter, to 24 hours glucose control followed by standard glucose control thereafter, or to standard glucose control started after admission 32. After a follow-up of 2 years no difference in mortality could be observed: 23.4% versus 21.2% and 17.9% (NS). Glucose levels at admission averaged 12.7 mmol/L and fall to 9.1 mmol/L in the intensive treatment group compared to 10.0 mmol/L after standard treatment. The drop in glucose level of 3.4 mmol/L obtained in DIGAMI 2 was thereby smaller than in their first trial (5.8 mmol/L). Moreover, in the smaller Hyperglycemia: Intensive Insulin Infusion In Infarction (HI-5) study with 240 patients, the mean blood glucose level after 24 hours was 8.3 mmol/L in the treatment group compared to 9.0 mmol/L in the conventional group (NS) 33. A mean insulin dose administered over this time in the treatment group was (only) 1.9 units/hours. An effect on mortality could not be observed.

The relation between hyperglycemia and outcome could be one of the explanations for the lack of benefit of trials using insulin in combination with glucose-potassium, i.e. GIK 2934353637. Patients randomized to GIK suffered more often from hyperglycemia after 24 hours. Future trials have to determine whether or not to aim for strict glucose regulation during acute MI, since in critically ill patients it has been studied and proven to be effective in critically ill patients 3839. The level of glucose levels obtained (4.4–6.1 mmol/L) were much lower than in both DIGAMI, HI-5 and GIK trials.

Although in this analysis persistent hyperglycemia depicted 30-day MACE over admission glucose the area under the ROC curves for both measures were small. Other factors than glucose deregulation must account for unfavourable outcome after MI. On the other hand admission glucose is seen as one among other measures that help to predict outcome and possibly persistent hyperglycemia could serve as a better prognostic value. The number of patients in our analysis is not large, although the level of significance for observed relations was strong. The number of patients with diabetes was too small to draw conclusions on the relation of persistent hyperglycemia and outcome in the subgroup of patients with diabetes mellitus. It would have been useful if we had been informed about the degree of chronic hyperglycemia before the myocardial infarction, as reflected by glycated hemoglobin (HbA_1c). Unfortunately, we did not systematically record HbA_1c. Diabetes was diagnosed after admission in two patients, although the true number of diabetics might still have been higher. Finally, in a recent study, fasting glucose was superior to admission glucose with regard to 30-day mortality 12. We did collect glucose measures according to a time-scheme but had no information whether the measures collected during the morning were in the fasting state. Therefore, we could not perform an analysis on the relation of fasting glucose.