Cardioprotective Effects of Glucose-Insulin-Potassium Infusion in Patients Undergoing Cardiac Surgery: A Systematic Review and Meta-Analysis

The infusion of glucose-insulin-potassium (GIK) has yielded con ﬂ icting results in terms of cardioprotective effects. We conducted a meta-analysis to examine the impact of perioperative GIK infusion in early outcome after cardiac surgery. Randomized controlled trials (RCTs) were eligible if they examined the ef ﬁ cacy of GIK infusion in adults undergoing cardiac surgery. The main study endpoint was postoperative myocardial infarction (MI) and secondary outcomes were hemodynamics, any complications and hospital resources utilization. Subgroup analyses explored the impact of the type of surgery, GIK composition and timing of administration. Odds ratio (OR) or mean difference (MD) with 95% con ﬁ dence interval (CI) were calculated with a random-effects model. Fifty-three studies (n=6129) met the inclusion criteria. Perioperative GIK infusion was effective in reducing MI (k=32 OR 0.66[0.48, 0.89] P =0.0069), acute kidney injury (k=7 OR 0.57[0.4, 0.82] P =0.0023) and hospital length of stay (k=19 MD -0.89[-1.63, -0.16] days P =0.0175). Postoperatively, the GIK-treated group presented higher cardiac index (k=14 MD 0.43[0.29, 0.57] L/min P <0.0001) and lesser hyperglycemia (k=20 MD -30[-47, -13] mg/dL P =0.0005) than in the usual care group. The GIK-associated protection for MI was effective when insulin infusion rate exceeded 2 mUI/kg/min and after coronary artery bypass surgery. Certainty of evidence was low given imprecision of the effect estimate, heterogeneity in outcome de ﬁ nition and risk of bias. Perioperative GIK infusion is associated with improved early outcome and reduced hospital resource utilization after cardiac surgery. Supporting evidence is heterogenous and further research is needed to standardize the optimal timing and composition of GIK solutions


INTRODUCTION
Each year, cardiac surgery is performed worldwide in »1.5 million individuals with ischemic, congenital and valvular disorders. 1 Over time, outcomes after cardiac surgery have improved along with better preoperative patient preparation, progress in surgical and anesthetic management as well as cardioprotective protocols. [2][3][4][5] Perioperative ischemia-reperfusion injuries and the release of free radicals and inflammatory mediators are incriminated in causing ventricular dysfunction that either resolves spontaneously or requires cardiovascular drug support and occasionally circulatory assistance. [6][7][8] Importantly, cardiac complications such as postoperative myocardial infarction (MI) and heart failure are known predictors of increasing medical costs, poor survival and decreased quality of life. 9,10 Among various cardioprotective protocols, the infusion of glucose-insulin-potassium (GIK) has been studied extensively. In animal models, GIK has been shown effective in reducing the extent of MI and the occurrence of ventricular arrhythmias while preserving ventricular function. 11 These cardioprotective effects are mediated by pleiotropic glucose-dependent and -independent mechanisms of insulin involving preferential high-energy substrate production from glucose metabolism as well as upregulation of the reperfusion injury salvage kinase pathway. 12 Since its introduction in 1962, 13 GIK has failed to show conclusive clinical cardioprotective effects following percutaneous coronary intervention whereas favorable results have been reported after cardiac surgery. 12,14 In previous systematic reviews, [15][16][17][18] the interactions between GIK therapy and confounding factors (eg, diabetes mellitus, type of surgery, glycemia or timing and composition of GIK infusion) have not been examined. Hence, our meta-analysis addresses these issues and provides an up-to-date review of the impact of GIK on early postoperative outcome.

Search Strategy
This review was conducted according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines and the Cochrane methodology as well as in agreement with a preregistered protocol (PROSPERO CRD 42022120746). 19,20 Ethical review board approval was waived due to the absence of new data collection. Minor deviations from the protocol are reported in a Supplemental file (S1). Three investigators (R.S., A.H. and A.P.) independently searched MEDLINE, EMBASE and the Cochrane Central Register of Clinical Trials from inception to September 19th, 2022. The search strategy aimed to select RCTs with the following terms: glucose-insulin-potassium, GIK, cardiac surgery, cardiopulmonary bypass, CPB, coronary artery bypass surgery, CABG, valve (S2). Additional articles were identified by manual review of the references of included studies.

Study Selection
Search results were examined at the abstract level and the fulltext version was retrieved if relevant. Eligibility criteria were defined following the PICOS approach: (P) adult patients scheduled for elective or emergent cardiac surgery with or without cardio-pulmonary bypass (CPB); (I) use of GIK in the perioperative period; (C) usual care or placebo, (O) MI and (S) RCT. Exclusion criteria were inclusion of pediatric cases, studies with overlapping population or irrelevant study endpoints. Four authors (A.H, R.S, A.P., and G.K-B.) independently made the final assessment for inclusion into the analysis and disagreements were resolved through consensus or by third party adjudication (M.L.). If documents did not contain MI data or were unavailable as full-texts, the corresponding authors were contacted for further information. No language restriction was imposed.

Data Abstraction
The relevant information was extracted from each selected study by a single author (R.S.) and checked by 2 others (A.P. and G.K-B.), disagreements being resolved by a fourth author (A.H.). Sources of clinical heterogeneity were also extracted according to the same process (ie, study design, clinical setting, inclusion/exclusion criteria. Study characteristics were collected regarding demographic data, the type of surgery, the duration of surgery as well as GIK composition (dose of insulin and glucose) and timing of administration (before, during or/ and after CPB). The primary outcome was postoperative MI and secondary outcomes were in-hospital mortality, the postoperative occurrence of stroke, acute kidney injury (AKI), atrial fibrillation (AF), ventricular fibrillation (VF), any infections, postoperative gylcemia, cardiac index, the need for pharmacological or mechanical circulatory support as well as the duration of mechanical ventilation, intensive care unit (ICU) and hospital stay.

Quality Assessment
Two authors (R.S. and A.P.) independently assessed the internal validity of included trials according to the Cochrane Collaboration methodology (risk of bias 1 tool), namely: risk of bias associated to the random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, completeness of outcome data, selective reporting and other biases. 20 Studies were rated as low, unclear, or high risk of bias. Included trials were rated as low risk of bias when 5 or more evaluation domains were judged as low risk of bias. 21 Studies that did not detail allocation concealment, blinding of participants and personnel or random sequence generation were graded as unclear.
The certainty of evidence was assessed using GRADE: the grading of recommendations assessment, development, and evaluation framework. 22

Statistical Analysis
Odds ratio (OR) or mean difference (MD) with 95% confidence intervals (95%CI) were reported. Random effects models were used in all cases. Between-study variance for binary analyses was assessed using the Paule-Mandel estimator since the DerSimonian-Laird estimator is known to be unreliable with sparse data. 23 Continuous models used the DerSimonian-Laird estimator. Prediction intervals were computed for all models. Heterogeneity was assessed using Cochrane's Q and the I 2 value. All models used a continuity correction of 0.5 at each step, except for Peto models. The analysis was performed using R 4.0.4 with package "meta". 24,25 Analysis of the primary outcome was stratified by GIK timing, composition, insulin infusion rate (cutoff 2 mUI/kg/min), 11,26,27 presence of diabetes mellitus and type of surgical procedure (coronary artery bypass, valve or combined surgery). Sensitivity assessments were performed using both fixed and random effects models for continuous meta-analyses, while Peto models were used for binary meta-analyses. Small-study effect for the primary outcome was investigated by the trim-and-fill method. 28
The proportion of participants with a MI was 5.3% and 8.2% in the GIK and control groups, respectively. As illustrated in Figure 3 Meta-analyses of secondary endpoints are summarized in Table 2. Perioperative GIK infusion was associated with a reduction in postoperative AF (19.8% vs 24.8% in control groups) and in AKI (3.3% vs 5.7% in control groups), along with higher CI (3.16 vs 2.77 L/min/m2 in control groups), faster ventilatory weaning as well as shorter ICU and hospital length of stay. There was no evidence of an association between GIK and in-hospital mortality, cardiovascular drug support, ventricular arrhythmias, infection or stroke.
Postoperative glycemia was higher in control than in GIKtreated patients (mean[SD] 185 [49]  A summary of the GRADE assessment is reported in supplement S6. The certainty of evidence was graded as low for MI and secondary outcomes due to imprecision of the effect estimate and indirectness related to heterogeneity in outcome definition or GIK regimen.
A video summarizes the key features of this systematic review and is available on the journal website.

DISCUSSION
The previous systematic reviews had focused more on CABG surgery and had examined a restricted time frame [15][16][17][18] whereas this updated meta-analysis of 53 RCTs (N=6129) included the last well-powered trials and covers the full spectrum of cardiac surgical procedures over more than 4 decades (Supplemental file S7).
In this meta-analysis, we found that, compared with usual care, perioperative GIK infusion was associated with fewer MI, AF and AKI, increased cardiac index, better postoperative glycemic control as well as earlier weaning from the ventilator, shorter length of stay in ICU and faster discharge from the

ADULT À Original Submission
Seminars in Thoracic and Cardiovascular Surgery Volume 00, Number 00

ADULT À Original Submission
Seminars in Thoracic and Cardiovascular Surgery Volume 00, Number 00

ADULT À Original Submission
hospital. GIK-induced cardioprotection effects was effective in both diabetic and non-diabetic patients, regardless of the perioperative timing of administration and at insulin dosages higher than 2 mUI/kg/min. Over the 4 decades spanned by this review, perioperative care and the study population risk profile have considerably evolved. Indeed, the infusion of blood or buffered crystalloid solutions directly into the aortic root or selectively into coronary arteries to minimize myocardial injuries became standard since the late 90s. 92 Furthermore, myocardial preconditioning with inhaled anesthetics has been introduced that could also contribute to improve clinical outcomes. 3 Meanwhile, the recent trials have included older participants and new criteria have been adopted to diagnose MI after cardiac surgery. 93,94 The stable proportion of MI over the review period likely resulted from the (opposite) effects of improved cardioprotective strategies and higher risk profiles of cardiac surgical patients with more sensitive criteria for MI. Altogether, despite improvements in surgical techniques and the higher risk profile of surgical patients, we found similar GIK-induced cardioprotective effects in the early and most recent periods (1977-2005 vs 2006-2021).
Interestingly, we found that perioperative reduction in myocardial injuries was associated with insulin given at rates higher than 2 mU/kg/min. 11,12,26,95 In the context of fasting and surgical stress, this insulin dose regimen causes a shift in ATP production from free fatty acid to glucose oxidation 26,27 and the increased myocardial ATP store has been associated with improved left ventricular function after experimental ischemiareperfusion and in patients with heart failure. 96,97 In this meta-analysis, the usual care group more frequently exhibited hyperglycemia that has been incriminated in blunting anesthetic preconditioning 98-100 and increasing cardiomyocyte apoptosis. 101 Likewise, the presence of hyperglycemia after major trauma/ surgery and myocardial infarct has been associated with a higher risk of cardiovascular complications and mortality. [102][103][104] In line with our findings, the negative results reported in GIK trials involving patients with acute coronary syndrome could be attributed to insufficient insulin dosages (1 to 1.25 mU/kg/min) and the consequent prolonged period of hyperglycemia. 35,105 In most RCTs (39 out of 53), GIK infusion was started before CPB, continued over a median duration of 6 hours and was associated with higher cardiac index (+15%). In contrast with the "oxygen-wasting" effects of adrenergic inotropes owing to fatty acid oxidation, GIK infusion has been shown to provide more efficient cardiac mechanical work by promoting the "oxygen-sparing" pathway of glucose oxidation. 106 In cardiac surgical patients, better preservation of the left ventricular function has been reported following CPB in GIK-treated patients compared with controls. 37,67,107 Recent case reports confirm that GIK administration may enhance cardioprotection in non-cardiac surgery and facilitate weaning from mechanical circulatory support in patients with acute cardiogenic shock. 108,109 In this meta-analysis, faster extubation time, lesser AKI as well as shorter ICU and hospital stay could all be attributed to the improved hemodynamic conditions and in turn, enhanced oxygen transport to the respiratory muscles and the kidneys in GIK treated patients.

Limitations
Our findings are to be interpreted cautiously due to several limitations. Firstly, definitions and reporting of the primary outcome (MI) varied over time and across the different trials with MI criteria being specifically validated in the context of CABG (type V MI) but not for other cardiac procedures according to the Fourth Universal Definition of Myocardial Infarction. 93 However, the various treatments and diagnostic criteria for MI (also AKI) reported in this meta-analysis were equally distributed in the 2 treatment arms and therefore equally influenced the occurrence of MI and other adverse events. Secondly, our study population mainly included middle-aged patients with moderately reduced or preserved left ventricular function undergoing on-pump elective cardiac procedures with AXC < 120 min. Ischemic and anesthetic conditioning are less effective in the senescent heart and in patients treated with beta-blockers. 110 Our data did not allow to discriminate the effects of GIK among elderly patients, those undergoing  complex surgical procedures, as well as the interactions with cardiovascular treatments. Nevertheless, blunting of the postoperative hyperglycemic in GIK-treated patients could contribute to mitigate the perioperative myocardial injuries in addition to the direct insulin mediated anti-apoptotic and metabolic effects on the cardiomyocytes. Thirdly, since all RCTs investigated early outcomes, we ignore whether short-term functional and clinical improvements could translate into better quality of life and prolonged survival. Fourthly, many RCTs included small populations while some large, recent RCTs held little weight in the analysis. 90 Moreover, the analysis of the impact of diabetes mellitus, combined surgery or valve replacement, emergency procedures and off-pump CABG yielded small subgroups of trials, precluding clinical interpretation. Finally, a majority of RCTs were graded as unclear or high risk of bias and the GRADE certainty of evidence was fairly low, indicating a potential for modification or even reversal of effect estimates in future studies. 22 CONCLUSION Perioperative GIK infusion is associated with improved early outcome and reduced hospital resource utilization after cardiac

ADULT À Original Submission
Seminars in Thoracic and Cardiovascular Surgery Volume 00, Number 00 surgery ( Figure 4, Video abstract). Current supporting evidence is heterogenous and further research should examine the dose-response and timing of GIK regimens to protect the heart also in higher risk patients such as those with failing heart or ongoing myocardial ischemia.

SUPPLEMENTARY MATERIAL
Scanning this QR code will take you to the article title page to access supplementary material.