Article Text
PDF
Abstract
Background There is conflicting evidence regarding two different insulin regimens for acute myocardial infarction (AMI), one focusing on delivering insulin (‘insulin focus’, glucose-insulin-potassium (GIK)) and one focusing on tight glycaemic control (‘glycaemia focus’, insulin-glucose). A longstanding controversy has focused on which strategy provides the greatest reduction in mortality. The aim of this study was to perform a meta-analysis of randomised controlled trials (RCTs) comparing GIK or insulin-glucose therapy versus standard therapy for AMI in the reperfusion era.
Methods A MEDLINE/EMBASE/CENTRAL search was conducted of RCTs evaluating GIK or insulin-glucose as adjunctive therapy for AMI. The primary endpoint was all-cause mortality. The data were analysed with a random effect model.
Results A total of 11 studies (including 23 864 patients) were identified, eight evaluating insulin focus with GIK and three evaluating glycaemia focus with insulin-glucose. Overall, insulin focus with GIK was not associated with a statistically significant effect on mortality (RR 1.07, 95% CI 0.89 to 1.29, p=0.487). Before the use of reperfusion, GIK also had no clear impact on mortality (RR 0.92, 95% CI 0.70 to 1.20, p=0.522). Pooled data from the three studies evaluating glycaemia focus showed that insulin-glucose did not reduce mortality in the absence of glycaemia control in patients with AMI with diabetes (RR 1.07, 95% CI 0.85 to 1.36, p=0.547).
Conclusions Current evidence suggests that GIK with insulin does not reduce mortality in patients with AMI. However, studies of glycaemia are inconclusive and it remains possible that glycaemic control is beneficial.
- Glucose-insulin-potassium
- acute myocardial infarction
- insulin focus
- glycaemia focus
- reperfusion
- stemi
Statistics from Altmetric.com
Introduction
Glucose-insulin-potassium (GIK) infusion has a long and controversial history since the first report by Sodi-Pallares et al,1 both before and since the introduction of reperfusion. Over the past decades, numerous clinical studies using GIK or glucose-insulin for the treatment of acute myocardial infarction (AMI) have reported remarkably varying results, in contrast to the positive results from early preclinical studies.
In 1997 Fath-Ordoubadi et al2 published a meta-analysis of studies of GIK treatment for AMI that included nine trials with 1932 patients published from 1966 through 1996. This meta-analysis showed a reduction in AMI mortality of 28–48% with GIK.2 However, it is important to realise that eight of the nine studies were performed in the pre-reperfusion era, and the in-hospital mortality rate of 21% would be considered quite high by today's standards.
Clinical trials of insulin therapy for myocardial infarction can be divided into those with a primary aim of delivering insulin (‘insulin focus’)3–10 and those with a primary aim of achieving tight glycaemic control (‘glycaemia focus’).11–13 Additional therapeutic aims could be direct protective effects on ischaemic tissue by insulin-mediated glucose uptake and decreased reperfusion-induced myocardial injury.14 The primary objective of an insulin focus regimen is to deliver large doses of insulin. The initial rationale was based on the ability of insulin to suppress free fatty acid (FFA) levels, stimulate potassium reuptake (and thus stabilise the sarcolemmal membrane) and enhance glycolysis in the ischaemic myocardium. The blood glucose level is not a primary concern, except for the avoidance of hypoglycaemia. In contrast, the primary goal of glycaemia focus regimens is the use of insulin to control hyperglycaemia.15
The comparative efficacy of these two regimens in patients with AMI remains unknown.16 17 The purpose of the present meta-analysis is therefore to examine two apparently discordant metabolically-based therapeutic strategies for insulin administration for AMI in the reperfusion era.
Methods
We carried out this meta-analysis in accordance with standards set out in the Quality of Reporting of Meta-Analyses of Randomised Controlled Trials (QUOROM) statement.18
Search strategy
A comprehensive search was conducted to identify all human randomised controlled trials (RCTs) of GIK or insulin-glucose that recorded all-cause mortality as the outcome. MEDLINE, CENTRAL and EMBASE were searched for any relevant human RCTs or reviews published from January 1980 to November 2009 using the terms ‘glucose-insulin-potassium’, ‘insulin’, ‘glucose’, ‘coronary heart disease’, ‘myocardial infarction’ and ‘acute coronary syndrome’. The search was not limited to English language publications. Foreign language papers were translated when required. Two reviewers independently evaluated identified titles, and manuscripts were retrieved for any publication that either reviewer felt was potentially relevant. Additional publications were sought using the reference lists of identified papers; published reviews on the topic were also evaluated. For each study, the results sections and tables were examined to see if data on all-cause mortality were reported. Attempts were made to contact authors of GIK trials that did not report the sequence of GIK and reperfusion therapy.
Clinical trial selection
Two blinded reviewers re-evaluated all of the abstracts and manuscripts identified as potentially relevant, and publications were selected for this review if both reviewers felt that they met the following criteria: (1) RCTs with parallel design; (2) comparing GIK or insulin-glucose as an adjunctive therapy; and (3) collecting data on all-cause mortality during follow-up. Studies were included in this review if both authors felt they were relevant. Any discordance between reviewers was resolved by consensus.
Assessment of quality
A quality review of each RCT was done to include details of randomisation, generation of random numbers, details of double-blinding procedure, information on withdrawals and allocation concealment. One point was awarded for the specification of each criterion, with a maximum score of 5. High-quality RCTs scored ≥3 points whereas low-quality RCTs scored ≤2 points, according to a modified Jadad score.19 No summary score was used to identify low- or high-quality trials; we did not perform weighting by quality scores as this practice has not been recommended by some.20 21
Statistical analysis
Two reviewers independently extracted data from eah study using a standardised form. Inconsistencies between reviewers' data were resolved through discussion until a consensus was reached. Data were managed according to the intention-to-treat principle. The pooled relative risks (RRs) were estimated using a random effects model with the Mantel–Haensel method that considers between-study heterogeneity. Subgroup analyses of the time of GIK administration (before or after reperfusion) were done on the basis of a priori protocol of RCTs. The I2 statistic was used to assess the heterogeneity of summary estimates, where a value >50% was considered as evidence of significant heterogeneity.22 Publication bias was assessed using a funnel plot and the Begg rank correlation method (p<0.05 indicated significant bias). Stata Version 11.0 (Stata Corp, College Station, Texas, USA) was used for all calculations.
Results
Study selection and description
Our electronic searches identified 964 studies of which 11 fulfilled the inclusion criteria and were subjected to meta-analysis3–13; eight3–10 were based on insulin focus and three11–13 were based on glycaemia focus (figure 1). There were two reports from the Organization to Assess Strategies for Ischaemic Syndromes (OASIS)-610 and Estudios Cardiologicos Latinoamerica (CREATE-ECLA)7 study regarding GIK; data from letters by Opie23 used for this meta-analysis reported the sequence of GIK and reperfusion.
Flow diagram of included studies. RCT, randomised controlled trials.
The characteristics of the included studies are summarised in table 1. There were 23 864 patients in the identified studies, 22 215 in eight GIK RCTs based on insulin focus3–10 and 1649 in three insulin-glucose RCTs based on glycaemia focus.11–13 The mean age of the study populations ranged between 56.6 and 68.6 years; the percentage of men was 62–100%. The baseline mean blood glucose concentration for both treatment strategies was in the range 6.9–15.7 mmol/l in eight trials; this information was not reported in the remaining three trials. The mean blood glucose concentration 24 h after treatment ranged between 4.25 and 9.6 mmol/l in eight trials; in three trials this information was not reported.
Characteristics of included studies
Effect of GIK on mortality based on insulin focus
In eight RCTs studying GIK in patients with AMI,3–10 meta-analysis did not demonstrate a significant reduction in mortality (RR 1.07, 95% CI 0.89 to 1.29, p=0.487), although there was still considerable heterogeneity between trials (I2=48.0%, p=0.062, figure 2).
Comparison of glucose-insulin-potassium (GIK) versus controls based on an insulin focus strategy for all-cause mortality.
Because the effect of GIK on AMI might vary depending upon the time of GIK treatment (before and after reperfusion),14 23 we performed a subgroup analysis to examine the heterogeneity of the results. Four RCTs3 5 8 9 24 and combined CREATE-ECLA and OASIS-6 subgroup analysis23 evaluated the effect of receiving GIK before reperfusion and found no significant reduction in mortality (RR 0.92, 95% CI 0.70 to 1.20, p=0.522). The trial by Krljanac et al8 showed the largest effect of treatment but it was also the smallest trial, accounting for 0.7% of patients in this analysis. Exclusion of this trial removed the statistical heterogeneity (I2=0%, p=0.418) and did not affect the lack of effect on mortality (RR 0.82; 95% CI 0.49 to 1.37, p=0.447). Two trials4 6 and combined CREATE-ECLA and OASIS-6 subgroup analysis23 studied the effect of GIK after reperfusion. Overall, there was no significant reduction in mortality (RR 1.34, 95% CI 0.92 to 1.94, p=0.127).
Effect of insulin-glucose on mortality based on glycaemia focus
Three trials compared insulin-glucose with standard therapy11–13 in the treatment of AMI with diabetes (figure 3). Overall, there was no reduction in mortality (RR 1.07, 95% CI 0.85 to 1.36, p=0.552). The result from trials on glycaemia focus were not statistically heterogeneous (I2=0, p=0.432); however, differences in 24 h blood glucose control and a small difference in the mean 24 h blood glucose level were achieved (table 1). To resolve the above clinical heterogeneity, we performed subgroup analysis according to whether glycaemic control was achieved (figure 3). Only the Diabetes Mellitus, Insulin-Glucose Infusion in Acute Myocardial Infarction (DIGAMI) trial demonstrated a non-significant reduction in mortality with RR of 18%.11 The DIGAMI trial was also the only one of the three studies to show lower blood glucose levels at 24 h in those who received insulin (9.6 mmol/l) compared with conventionally-treated subjects (11.7 mmol/l). The DIGAMI-212 and Hyperglycemia: Intensive Insulin Infusion in Infarction (HI-5)13 trials did not achieve the glucose target of the study protocol; there was no reduction in mortality with insulin-glucose compared with conventional therapy (RR 1.18, 95% CI 0.90 to 1.55, p=0.22). As shown in table 1, the control subjects in the DIGAMI-2 and HI-5 studies had similar or even better glycaemia control than intervention subjects in the DIGAMI study. By achieving only a small difference in glucose levels between the study arms, it can therefore be argued that the DIGAMI-2 and HI-5 studies failed to adequately test their primary hypothesis.
Comparison of insulin-glucose (GI) versus controls based on a glycaemia focus strategy for all-cause mortality.
Sensitivity analyses and publication bias
In view of the fact that the CREATE-ECLA and OASIS-6 subgroups receiving GIK before and after reperfusion have greater weight in the pooled results, analysis was performed with exclusion of these two subgroups. This did not change the pooled result of receiving GIK before reperfusion (RR 0.82, 95% CI 0.49 to 1.37, p=0.447). However, the pooled result in two trials receiving GIK after reperfusion showed that GIK increased mortality (RR 1.78, 95% CI 1.13 to 2.80, p=0.013).
Funnel plot and Begg rank correlation did not indicate publication bias (p=0.458; figure 4).
Assessment of publication bias.
Discussion
This meta-analysis, based on eight RCTs with GIK regimens, indicates that the insulin focus strategy is not effective in reducing the mortality of AMI by inhibiting FFA and reperfusion injury. Limited evidence, based on three RCTs, suggests that the glycaemia focus strategy is also not effective in reducing mortality in the absence of tight glycaemia control in AMI.
Insulin focus strategy
Our meta-analysis did not show any benefit on mortality from GIK regimens based on the insulin focus strategy; this is different from results reported in a meta-analysis of studies performed before the use of reperfusion which found a 28% reduction in mortality.2 An earlier theory held that the benefits of GIK were due to both increased availability and facilitated entry of glucose into the cardiomyocyte and decreased FFA levels by insulin.25 However, this ‘oversupply’ of glucose (depending on the GIK formulation used) can lead to acute elevation of glucose compared with the pretreatment level. Hyperglycaemia 24 h after admission is now recognised to be a predictor of mortality during AMI.26 The CARDINAL study27 showed that both higher baseline glucose and hyperglycaemia in the first 24 h after AMI predict higher mortality in non-diabetic patients. GIK-induced hyperglycaemia may have counterbalanced or neutralised any benefit of insulin in the GIK administration, particularly when given in a high-dose formulation.
Further subgroup analysis of trials refuted the possibility that GIK would only benefit patients when initiated before reperfusion therapy,14 23 although preclinical studies suggested that GIK could suppress ischaemia reperfusion injury.28 29 There are several possible explanations for the discordance between the positive findings in preclinical studies and the predominantly negative RCTs. First, comorbid conditions leading to ischaemic heart disease (eg, atherosclerosis, hypercholesterolaemia, hypertension and diabetes), advanced age and impaired nutritional status are typically absent in preclinical studies and may blunt the efficacy of cardioprotective therapies in clinical trials.30 Second, the concomitant medications used during current clinical practice, which are typically absent in experimental studies, may influence the effectiveness of agents directed against reperfusion injury. Third, in experimental studies, ischaemia-reperfusion is typically produced by mechanically-induced abrupt and total coronary occlusion and reperfusion. Importantly, the repetitive intermittent ischaemia caused by both thrombolysis (incomplete reperfusion) and balloon dilation and/or stent implantation may induce postconditioning, a feature that is typically avoided in experimental studies.31
Glycaemia focus strategy
Our result from subgroup meta-analysis indicates that insulin-glucose does not reduce mortality if the target for glycaemia control was not achieved in the trials with glycaemia focus, although this review may have limitations because only three studies were available.
Several major studies link hyperglycaemia in AMI to a poor prognosis15 32 33 because hyperglycaemia could have harmful pro-oxidative and proinflammatory effects.34 The AHA scientific statement15 states that normalising the blood glucose should become the major aim of intensive insulin therapy. The study by Marfella et al,35 which provides for the first time data on the histological features of the human myocardium, indicated that tight glycaemic control by reducing oxidative stress and inflammation might reduce apoptosis in peri-infarcted areas and remodelling in patients with AMI. In our meta-analysis, however, only one (DIGAMI11) of the three RCTs11–13 based on glycaemia focus achieved improved glycaemic control. Another likely major contributory reason for the negative result of the HI-513 study is that a significant difference in glycaemia between treatment and control groups was not achieved. Similarly, glucose targets were not achieved in DIGAMI-2,12 although insulin-treated subjects had a lower fasting glycaemia after 24 h (9.1 vs 10 mmol/l). It is evident that, despite a glycaemia focus, strict control of glycaemia in patients with AMI is difficult to attain. Furthermore, the advent of rapid angioplasty/stent and widespread use of aspirin, β blockers and statins have dramatically reduced mortality, overshadowing much of the effect of glycaemic control.
Some limitations inherent the meta-analytical method must be acknowledged, such as problems in the randomisation processes used in controlled trials, publication biases, variation of standard treatments over time and heterogeneity of studies.36 In addition, a potential limitation of our meta-analysis is that it was based on trial-specific rather than patient-specific data. Its most important limitation, on the other hand, was the fact that only three trials on glycaemia focus were included.
In conclusion, the results of our meta-analysis suggest that treatment with an insulin focus strategy only—where the aim is to deliver large doses of insulin without regard to the glucose level—does not improve mortality when other modern therapies for AMI are used in the reperfusion era. Nevertheless, it remains possible that control of hyperglycaemia by insulin infusion with a glycaemia focus improves mortality in patients with AMI.
References
Footnotes
Y-T Z and C-L W are co-authors and contributed equally to the manuscript.
Competing interests None.
Ethics approval This study was a meta-analysis which did not require ethics approval.
Provenance and peer review Not commissioned; externally peer reviewed.