Effect of Remote Ischaemic preconditioning on Clinical outcomes in patients undergoing Coronary Artery bypass graft surgery (ERICCA study): a multicentre double-blind randomised controlled clinical trial

Primary research objective

The primary research objective was to determine whether or not RIPC improves 1-year clinical outcomes in patients undergoing CABG with or without valve surgery.

Secondary research objectives

1. To determine if RIPC improves 30-day clinical outcomes in patients undergoing CABG with or without valve surgery.

2. To determine if RIPC has an effect on all-cause death at 12 months.

3. To determine if RIPC reduces PMI in higher-risk patients undergoing CABG with or without valve surgery (assessed by measuring serum troponin T over the 72-hour perioperative period).

4. To determine if RIPC reduces AKI and preserves renal function post CABG with or without valve surgery [assessed by measuring serum neutrophil gelatinase-associated lipocalin (NGAL) and serum creatinine and the AKI score].

5. To determine if RIPC improves patient morbidity as measured by duration of intensive therapy unit (ITU) stay, inotrope score, the 6-minute walk test (6MWT) and quality-of-life assessment.

6. In a substudy of patients recruited through selected hospitals to determine the effect of RIPC on left ventricular ejection fraction (LVEF) measured by echocardiography.

Recruitment

The 30 centres included in the study were the Royal Brompton Hospital, London; Harefield Hospital, Middlesex; UCLH Heart Hospital, London; King’s College Hospital, London; Papworth Hospital, Cambridge; Hammersmith Hospital, London; St George’s Hospital, London; St Thomas’ Hospital, London; London Chest Hospital; Essex Cardiothoracic Centre, Basildon; Royal Sussex County Hospital, Brighton; Royal Wolverhampton Hospitals NHS Trust; Bristol Royal Infirmary; Golden Jubilee National Hospital, Clydebank; Edinburgh Royal Infirmary; Morriston Hospital, Swansea; Cardiff and Vale University Health Board; Manchester Royal Infirmary; Derriford Hospital, Plymouth; Northern General Hospital, Sheffield; Trent Cardiac Centre, Nottingham; Blackpool Victoria Hospital; Wythenshawe Hospital, Manchester; Glenfield Hospital, Leicester; Southampton General Hospital; Leeds General Infirmary; James Cook University Hospital, Middlesbrough; North Staffordshire University Hospital, Stoke-on-Trent; Castle Hill Hospital, Hull; and University Hospital Coventry.

Patients were recruited from two groups: outpatients on the waiting list for CABG surgery or inpatients awaiting CABG surgery.

Randomisation, concealment and blinding

Patients were randomly assigned in a 1 : 1 ratio to receive sham RIPC or RIPC with randomly permuted block sizes of four and six stratified by recruiting centre. Randomisation was conducted via a secure website (Sealed Envelope™; Clekenwell Workshops, London, UK) through the Clinical Trials Unit (CTU) at the London School of Hygiene & Tropical Medicine (LSHTM).

The enrolment and preconditioning procedures were performed by an unblinded research nurse or clinician who was not involved in sample or data collection. Cardiac anaesthetists, cardiac surgeons, ITU staff, ward staff and all other research personnel at each study site were blinded to the treatment allocation. The patients were also blinded to the allocation of their randomised intervention as they were anaesthetised at the time of the RIPC intervention. A limited number of staff at the CTU were unblinded to the allocations to enable them to enter data onto the electronic case report form (eCRF), which was a web-based data capture system provided by Sealed Envelope. The unblinded trial statistician supporting the Data Monitoring Committee (DMC) also had access to the randomisation codes during the study. However, no member of staff at the CTU who was involved in the collection of outcome data had access to the treatment allocation of the patients.

Interventions

The trial intervention was a physiological procedure and was performed on patients after induction of anaesthesia but prior to CABG surgery. There were no other interventions. The active RIPC procedure consisted of four 5-minute inflations of a standard blood pressure cuff on the upper arm to 200 mmHg separated by 5-minute periods when the cuff was deflated. For patients with systolic blood pressure of > 185 mmHg the cuff was inflated to at least 15 mmHg above their systolic blood pressure. The sham RIPC procedure, which also used a standard blood pressure cuff, consisted of four 5-minute simulated inflations of the blood pressure cuff on the upper arm. The simulated inflations involved opening the air valve on the blood pressure cuff such that the cuff did not inflate on squeezing the bulb. The bulb was then squeezed for a duration of 15 seconds to give the impression that the cuff was being inflated. After 5 minutes the air valve was closed again to give the impression that the cuff was being deflated. After 5 minutes the air valve was opened again and the bulb squeezed as before. This cycle was undertaken four times in total.

Monitoring and site visits

The first site visit was a pre-recruitment visit for training (demonstration of the trial intervention and procedures) and to ensure that all relevant documentation was in place prior to recruitment commencing. After the first patient was recruited in each site, the senior data manager provided training on the eCRF. This training either took place as part of a visit to the site or was performed remotely using the standard operating procedure document.

On-site monitoring visits were not conducted routinely as this was a low-risk trial; they took place only if problems were identified (e.g. poor data collection or under-reporting of primary end point data) or when a site requested a visit to discuss specific issues (including data collection, screening patients, recruitment and staff training).

After the first five patients had been recruited and completed the 6-week follow-up, the CTU reviewed data from each site. If no problems were identified the sites were informed that there was no need for a site monitoring visit. Data validation was then conducted centrally by statistical monitoring across all sites. If problems were identified, the CTU would contact the relevant site to discuss the situation and arrange a site monitoring visit.

Central statistical monitoring was performed by the unblinded statistician by reviewing event rates, unusual trends and data anomalies.

Baseline assessment

At baseline all patients had a medical history taken and a clinical examination as part of usual care. Baseline information collected included weight, height, sex, ethnicity, blood pressure, heart rate, electrocardiogram (ECG), date of birth, euroSCORE, medication, levels of creatinine, troponin T, NGAL and biomarkers, 6MWT distance and European Quality of Life-5 Dimensions (EQ-5D) questionnaire; patients were also given a patient diary to record health resource use and an echocardiogram was obtained if they were part of the substudy.

Follow-up

All patients were followed up approximately 6 weeks post CABG. They were reviewed in clinic as per normal standard of care. Investigations performed at this visit were the 6MWT, creatinine level, ECG and EQ-5D questionnaire; the patient diary was also checked for health resource use.

At 3, 6 and 9 months post CABG all patients were requested to complete an EQ-5D questionnaire, which was carried out by post, telephone or e-mail.

At 1 year post CABG all patients were reviewed, when possible, in a research outpatient clinic. If they were unable to attend in person the follow-up was completed by telephone. General practitioner and hospital notes were checked for any events. At this visit, when possible, the following information was collected: weight, heart rate, blood pressure, recording of primary end points, ECG, creatinine level, echocardiogram (if part of the substudy), 6MWT distance and EQ-5D questionnaire; the patient diary was also checked for resource use.

Safety assessments

All untoward occurrences were termed adverse events rather than adverse reactions. Adverse events were assessed for relatedness to the intervention and were reported to the sponsor and the DMC. Safety assessments were collected from the time of randomisation to the completion of follow-up. There was a list of expected adverse events (both serious and non-serious) for which information was not collected. A detailed listing of individual adverse events was supplied as part of the reports to the DMC.

Primary outcome

The primary study end point was the rate of MACCE consisting of cardiovascular death, MI, coronary revascularisation and stroke within 12 months of surgery.

Prespecified secondary outcomes were:

1. 30-day MACCE

2. components of 30-day and 12-month MACCE

3. 12-month MACCE including definite MIs and strokes only

4. death from all causes

5. 72-hour area under the curve (AUC) troponin T

6. AKI injury score after 72 hours

7. serum creatinine level at baseline, 6 weeks and 12 months

8. 24-hour AUC NGAL

9. length of ITU stay

10. inotrope score after 72 hours

11. length of hospital stay

12. 6MWT at 6 weeks and 12 months

13. quality of life (EQ-5D) at 6 weeks and 3, 6, 9 and 12 months

14. substudy: ejection fraction at 12 months.

An additional secondary outcome was postoperative atrial fibrillation (AF), which was selected for analysis after finalising the statistical analysis plan and unblinding the trial data.

Collection of outcome data

Cardiovascular death included death resulting from an acute MI, sudden cardiac death, death from heart failure, death from stroke and death from other cardiovascular causes. This also included those deaths in which there was no clear non-cardiovascular cause. MI included perioperative MI (occurring within 72 hours of surgery) and postoperative MI (occurring >72 hours following surgery). Perioperative MI was defined as (1) troponin T level > 10 times the 99th percentile of the normal reference range (≥ 14 ng/ml) during the first 72 hours following surgery associated with the appearance of new pathological Q waves or new left bundle branch block or angiographically documented new graft or native coronary artery occlusion or imaging evidence of new loss of viable myocardium or (2) a troponin T level > 100 times the 99th percentile of the normal reference range during the first 72 hours following surgery (≥ 140 ng/ml) or (3) ECG evidence of perioperative MI in the absence of postsurgery troponin T results. Postoperative MI was defined as a MI occurring > 72 hours following the operation (definition of MI according to recent guidelines41).

Repeat coronary revascularisation was defined as repeat PCI or CABG of any segment of the coronary arteries. Stroke was defined as a focal neurological deficit of cerebrovascular cause that persisted beyond 24 hours or was interrupted by death within 24 hours.

Collection of major adverse cardiac and cerebrovascular events

Data were collected on the eCRF in relation to any events. An initial evaluation was made by the principal investigator (PI), or someone delegated by the PI, at each of the recruiting centres.

An independent Event Validation Committee (EVC) was set up consisting of a cardiologist (chairperson), surgeon and neurologist. All members were blinded to treatment allocation. Every effort was made to collect as much information as possible for assessing the events. A LSHTM-hosted secure file-sharing system was used to load the event data. Access to this was password protected and could be accessed at a time that suited each of the committee members. Quarterly conference calls were arranged to discuss adjudication of events and agree a final classification. In cases in which a unanimous decision could not be made the majority classification was used.

The following was used to classify each category.

Primary outcome

The primary analysis was a comparison of the MACCE rate 1 year after CABG with or without valve surgery between the RIPC arm and the sham arm of the trial. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated using Cox proportional hazards modelling and Kaplan–Meier curves were produced. The time scale used for survival analysis was time since intervention. This date was taken as the date of RIPC/sham RIPC if this was attempted, the date of surgery if the RIPC/sham RIPC was not carried out and the date of randomisation if neither RIPC/sham RIPC nor surgery was carried out. Participants who did not experience MACCE were censored on the date of non-cardiovascular death, loss to follow-up or withdrawal from the study or at 12 months. The proportional hazard assumption underlying the Cox model was assessed graphically and through global test of the scaled Schoenfeld residuals. 44

Secondary outcomes

Missing data

Initial models used complete case analysis and so included only participants with full data, which assumes that outcomes are missing at random given the treatment group. For analysis of AUC troponin T and AUC NGAL this required participants to have data available for all time points, and for analysis of serum creatinine this required participants to have data available at baseline and at the relevant follow-up visit. As a large number of participants were excluded from the complete case analysis of these outcomes, further imputation analysis was conducted to examine whether or not these missing data had an impact on the findings.

Multiple imputation was used to replace any missing values of troponin T, NGAL and serum creatinine. Data were imputed on the natural logarithm scale for each time point using Gaussian normal regression multiple imputation by chained equations. Twenty imputed data sets were generated using a separate imputation model for troponin T, NGAL and serum creatinine. Variables collected at baseline or during follow-up were included in the imputation model in which they were independent predictors of a participant having missing data (assessed using logistic regression models) or were predictive of the values of the outcome (assessed using linear regression).

The imputation model for natural logarithm of troponin T included log of troponin T at each time point (baseline and 6, 12, 24, 48 and 72 hours); treatment group; sex; baseline euroSCORE, New York Heart Association (NYHA) class, Canadian Cardiovascular Society angina class, LVEF class, smoking status, age, natural logarithm of creatinine, diastolic blood pressure and heart rate; previous use of aspirin, beta-blockers, nitrates, diuretics, clopidogrel and metformin; previous diagnosis of MI, peripheral arterial disease and hypercholesterolaemia; bypass duration; use of cardioplegia during surgery; use of nitrates during surgery; number of grafts; post-surgical outcomes of number of days in ITU, requirement for cardiac pacing, renal failure and AKI; and death from cardiovascular causes within 12 months of surgery.

The imputation model for natural logarithm of NGAL included log of NGAL at each time point (baseline and 6, 12 and 24 hours); treatment group; sex; baseline euroSCORE, natural logarithm of creatinine, age and diastolic blood pressure; previous use of aspirin, nitrates, cholesterol-lowering medication, angiotensin-converting enzyme inhibitors/angiotensin receptor antagonists; previous diagnosis of diabetes; previous PCI; bypass duration; use of cardioplegia during surgery; valve repair/replacement; post-surgical outcomes of number of days in ITU, AF, renal failure and AKI; and death from cardiovascular causes within 12 months of surgery.

The imputation model for natural logarithm of serum creatinine included log of creatinine at each time point (baseline, 6 weeks and 12 months); treatment group; sex; baseline euroSCORE, NYHA class, LVEF class, age, heart rate, diastolic blood pressure and body mass index; previous use of aspirin, nitrates, diuretics, sulphonylureas, metformin and insulin; previous diagnosis of stroke; and post-surgical outcomes of AF and AKI.

The multiple imputation analysis for all outcomes excluded 15 participants who did not have any baseline data and one patient who was randomised but following enrolment was found not to meet the eligibility criteria for the trial. Analysis additionally excluded any patients who died before the outcome could have been collected: 16 patients for 72-hour AUC hsTnT; 11 patients for 24-hour AUC NGAL; 57 patients for creatinine at 6 weeks; and 125 patients for creatinine at 12 months. All other patients were included in the imputation analyses.

Subgroup analyses

Subgroup analysis were conducted to assess if the effect of RIPC on MACCE differed according to age, baseline euroSCORE, baseline LVEF, aortic cross-clamp time, cardiac bypass time, method of cardioplegia and presence/absence of diabetes. Each subgroup was assessed through a Wald test of the interaction between the subgroup variable and treatment group in the Cox proportional hazards model. The subgroup variable was entered as a continuous variable for age, euroSCORE, clamp time and bypass time and as a categorical variable for LVEF class (normal/good vs. moderate vs. poor) and diabetes (binary: yes vs. no). Two post hoc subgroup analyses were carried out for the time interval between the start of RIPC and the end of bypass (continuous) and the type of anaesthetic used during surgery (binary: volatiles with or without propofol vs. propofol with no use of volatiles). When subgroup variables were continuous we provide the numbers with events and HRs and 95% CIs for RIPC treatment among those above and those below the median level of the subgroup variable. Otherwise, results are given for each category of the subgroup.

Subgroup analysis for major adverse cardiac and cerebral events

There was no evidence that the effect of the RIPC intervention differed between any of the prespecified subgroup analyses for the incidence of MACCE at 12 months (Figure 3). Although a subgroup analysis was planned by type of cardioplegia, only 19 participants had retrograde cardioplegia and so this subgroup analysis was not conducted. The other post hoc subgroup analyses found no evidence that the effect of RIPC on MACCE differed by the type of anaesthetic used during surgery (p = 0.17 interaction test) or by the duration between the start of RIPC and the initiation of bypass (p = 0.66 interaction test).

FIGURE 3.

Prespecified subgroup analyses for the incidence of MACCE at 12 months. HRs are shown for subgroups of participants above/below the median value for each factor.

Remote ischaemic preconditioning in coronary artery bypass graft surgery

Our research group was the first to demonstrate that RIPC had the potential to reduce PMI in adult patients undergoing elective CABG surgery. 13 In a pioneering single-blinded RCT13 involving 57 patients undergoing elective CABG surgery with either cardioplegia or intermittent cross clamp fibrillation and randomised to RIPC (three 5-minute cycles of inflation and deflation of a blood pressure cuff placed on the upper arm) or a control treatment (an uninflated blood pressure cuff placed on the upper arm for 30 minutes) we found that preconditioned subjects had a 43% reduction in troponin T release over the 72-hour perioperative period compared with control subjects. These findings were confirmed in a further study involving 45 non-diabetic patients undergoing elective CABG with or without valve surgery and receiving cold-blood cardioplegia alone,14 with RIPC (three 5-minute cycles of upper-arm ischaemia–reperfusion) significantly reducing the 72-hour AUC of troponin T by 42.4%. The same preconditioning stimulus was applied by Ali et al. 49 in a study including 100 patients undergoing elective CABG for two- or three-vessel CAD and similarly led to a significant reduction in postoperative CKMB levels.

The concept of RIPC in the context of elective CABG surgery was then extended to patients receiving antegrade cold crystalloid cardioplegia in two seminal studies by Thielmann et al. 50,51 In one of the studies50 non-diabetic patients with triple-vessel CAD subjected to three cycles of 5-minute transient upper-arm ischaemia sustained a significantly lessened PMI than control subjects, with a 44.5% reduction in the total 72-hour AUC of cardiac troponin I. In the other study,51 the largest proof-of-concept RCT on RIPC in cardiac surgery so far, the same preconditioning stimulus improved myocardial protection (ratio of RIPC to control for cardiac troponin I AUC was 0.83) and reduced the combined end point of all-cause mortality, major adverse cardiac and cerebrovascular events and repeat revascularisation.

However, recently, a number of studies have failed to demonstrate significant RIPC-induced cardioprotection. Within the context of crystalloid cardioplegia, Wagner et al. 52 showed only a small beneficial effect of RIPC applied 18 hours prior to elective CABG surgery with or without aortic valve replacement. Similarly, Lomivorotov et al. 53 did not find any statistically significant benefit of RIPC for PMI in patients undergoing CABG surgery with cold crystalloid cardioplegia. In a trial involving 162 patients undergoing CABG surgery, Rahman et al. 54 found no statistically significant difference in troponin T release, ECG changes, cardiac index, inotrope and vasoconstrictor requirements, renal impairment and lung injury between patients receiving a sham protocol and patients receiving a RIPC protocol (three 5-minute cycles of upper-limb ischaemia–reperfusion). However, importantly, the study included patients undergoing elective or urgent (post-acute coronary syndrome) CABG surgery and it is therefore possible that the beneficial effects of RIPC might have been attenuated by the previous acute event, which could have already ‘preconditioned’ the patients. Moreover, in this double-blinded study, patients were prepared and draped to obscure the visibility of both the blood pressure cuff placed around the upper arm and the one placed around a ‘dummy arm’ and, although the correct inflation was verified though the disappearance of a pulsatile signal on a pulse oximeter, it is still possible that during the inflation and deflation phases the cuff might have moved and the RIPC stimulus might not have been delivered correctly. Third, and importantly, a significant proportion of these patients received glyceryl trinitrate (GTN) intraoperatively, which might have interfered with the cardioprotection provided by RIPC. Subsequently, Young et al. 55 failed to demonstrate that a standard preconditioning stimulus could improve PMI, AKI incidence or inotrope requirement in a study enrolling 96 patients undergoing high-risk cardiac surgery, including combined CABG and valve surgery, CABG surgery with a LVEF of < 50%, ‘redo-operation’, mitral valve surgery and double or triple valve surgery.

Additionally, Karuppasamy et al. 56 showed no beneficial effects of standard RIPC in patients undergoing elective CABG surgery and receiving the volatile anaesthetic isoflurane before cardiopulmonary bypass and the intravenous anaesthetic propofol thereafter until the completion of surgery. Two other major clinical studies used a strict anaesthetic regime with similarly no significant impact of RIPC on PMI. 57,58 Furthermore, in a proof-of-concept study involving off-pump CABG surgery,59 RIPC induced by four cycles of 5-minute upper-limb ischaemia–reperfusion reduced the total cardiac troponin I AUC by 26%, which did not reach statistical significance. However, the same group found that the combination of RIPC with remote ischaemic postconditioning (RIPostC) in an analogous surgical setting [the same stimulus was applied twice, immediately after anaesthesia induction (RIPC) and just after completion of anastomoses (RIPostC)] could lead to a significant cardiac troponin I AUC reduction in the preconditioned group. 60 Crucially, our research group has more recently been able to demonstrate that an enhanced preconditioning stimulus, consisting of simultaneous multilimb RIPC, was able to reduce PMI and improve short-term clinical outcomes in patients undergoing elective cardiac surgery. Additionally, a number of systematic reviews investigating the effects of RIPC on PMI with or without clinical outcomes in patients undergoing cardiac or vascular surgery or elective PCI have been conducted,61,62 concluding that RIPC reduces post-procedure myocardial damage in these subjects but does not impact on their clinical outcomes.

It is also crucial to note that a significant part of the studies investigating the effects of RIPC on clinical outcomes in patients undergoing elective cardiac surgery included surrogate end points such as biochemical assessments of PMI through serial evaluation of serum levels of cardiac enzymes, yet very few data are available in the literature with regard to the potential beneficial effects of RIPC on clinical outcomes. The very first study to describe the impact of RIPC on patient morbidity and mortality in the context of cardiac surgery reported no postoperative deaths in either the preconditioned group or the control group 30 days after elective aortic valve replacement, mitral valve surgery or double valve replacement. 63 Similarly, no significant difference in MACCE was found 30 days postoperatively in two small studies involving patients undergoing elective CABG surgery with crystalloid cardioplegia. 50,53 Interestingly, in a study including high-risk patients,55 again no difference in mortality rate was found at 30 days post surgery; however, an improvement in NYHA functional status and mean LVEF at 3 months postoperatively was found in preconditioned patients undergoing elective valve replacement. 64 Additionally, in patients undergoing CABG surgery under a strict anaesthetic regime,58 RIPC was associated with a higher perioperative composite end point of new arrhythmias and new MI, yet no significant difference was found at 6 months’ follow-up.

Crucially, in the largest proof-of-concept clinical trial to date investigating the effects of RIPC in the context of elective CABG surgery, unpublished at the time of the initiation of the ERICCA trial, Thielmann et al. 65 reported a statistically significant improvement in all-cause mortality and MACCE rate in preconditioned patients at 1 year and at completion of follow-up (1.54 years ± 1.22 years), which was mainly driven by the reduced incidence of new MIs, whereas no significant difference was found in the occurrence of cardiac death, stroke and repeat revascularisation. Interestingly, of the 329 patients randomised and included in the ITT analysis, 71 were excluded in the final PP analysis, of whom 61 had known diabetes and, therefore, a total of 258 subjects were included in the PP analysis. However, the study was a single-centre trial and adequately powered for the primary end point of PMI but not for secondary end points, including clinical outcomes, and, crucially, despite still lower, the rate of all-cause mortality became non-statistically significant when deaths from sepsis were excluded. In our single-centre RCT, we could not demonstrate that simultaneous multilimb preconditioning reduces the rate of death, MI, revascularisation and stroke at 6 weeks post cardiac surgery, a finding that was also confirmed in the subsequent subgroup analyses; however, once again the study was not powered for this type of evaluation. 66 In addition, a significant number of systematic reviews and meta-analyses in patients undergoing cardiac or vascular surgery or elective PCI have been conducted;61,62,67–76 the overall conclusions confirmed the beneficial effects of RIPC on PMI reduction, but no statistically significant improvements in clinical outcomes were observed, including rate of death, perioperative MI, renal failure, stroke, mesenteric ischaemia or hospital or ITU stay.

Two recently published and adequately powered clinical trials in cardiac surgery failed to demonstrate any beneficial effect of RIPC on inpatient clinical outcomes in 299 paediatric patients (primary end point of postoperative hospital stay)77 and 1280 adult patients (combined primary end point of death, MI, arrhythmia, stroke, coma, renal failure, respiratory failure, cardiogenic shock, gastrointestinal complication and multiorgan failure),78 although the latter study used a combined remote pre- and post-conditioning stimulus.

In our ERICCA trial we demonstrated that four 5-minute cycles of upper-arm ischaemia–reperfusion were unable to improve clinical outcomes at 1 year in higher-risk patients undergoing elective CABG with or without valve surgery using blood cardioplegia. Multiple potential factors can be identified in this regard, which can be divided into patient characteristics, clinical settings, anaesthetic regimes and other agents administered in the perioperative period.

Patient characteristics involve baseline factors such as age and presence of comorbidities. More recently, the risk profile of patients undergoing cardiac surgery has substantially changed because of the ageing population. 79 The response of ageing myocardium to cardioprotection provided by IPC, RIPC and RIPostC, as well as by pharmacological agents including opioids, remains controversial. 80 Moreover, it has been established that the presence of comorbidities such as diabetes, hypertension and dyslipidaemia may interfere with cardioprotection. Experimental studies have shown that the diabetic myocardium may have an increased resistance to ischaemia–reperfusion injury compared with the non-diabetic heart, although significant differences in results have been obtained from different animal models and with different techniques of diabetes induction (reviewed by Galagudza et al. 81). Intriguingly, cardioprotection provided by IPC may be lost in ageing hypertensive hearts and discordant results have been obtained from experimental and human studies evaluating the effects of dyslipidaemia on myocardial ischaemia–reperfusion injury and more importantly on its impact on IPC, RIPC and RIPostC. 82

In addition, pharmacological agents administered concomitantly with cardiac surgery may have a significant impact on cardioprotection achieved with RIPC, in particular anaesthetic drugs and intravenous nitrates. GTN, a nitric oxide donor widely used in the context of cardiac surgery to achieve rapid blood pressure control and coronary vasodilatation,83 has been demonstrated to interfere with IPC and RIPC in experimental studies; however, its role in the clinical setting is yet to be clarified. 84–95 Inhaled anaesthetics have been shown to provide myocardial protection in patients undergoing cardiac surgery,58,96–98 either used alone or in combination with propofol. 58,59 However, the use of propofol alone does not lead to cardioprotection57 and this is probably because of the lack of action on adenosine triphosphate (ATP)-sensitive potassium channels and its interference with reactive oxygen species, which have both been implicated in mechanisms underlying RIPC. 57 It is therefore possible that inhaled anaesthetics, either alone or in combination with propofol, are capable of reaching the necessary threshold to induce cardioprotection and that the addition of RIPC may not provide any further benefit. In our trial, > 90% of patients received propofol at some point during surgery.

The clinical setting is another crucial aspect potentially able to interfere with cardioprotection. Patients undergoing elective CABG surgery sustain an overall small magnitude of PMI compared with that observed in patients presenting with STEMI. 99 Moreover, preclinical investigations of RIPC have been largely confined to experimental models of acute coronary artery occlusion/reflow (which represent the in vitro model of STEMI), rather than more clinically relevant animal models of cardiopulmonary bypass surgery, highlighting the discordancy between experimental and clinical cardioprotection studies. 57,80,100

Crucially, given the recent developments of treatment options in CAD and, more importantly, the advances in operative methods of myocardial preservation, surgical techniques and anaesthetic agents, it is possible that the additional benefit provided to these patients by RIPC might not be significant or identifiable with the current strategies.

It is also important to note that in our trial there was a suggestion of an increased risk of cardiovascular death with RIPC. However, our study was not powered to detect a difference in this individual end point and this finding should therefore be interpreted with caution. Crucially, the overwhelming majority of previously published studies51,77 and meta-analyses75,101,102 have failed to demonstrate any harmful effects of RIPC during cardiac bypass surgery in terms of mortality and, in fact, one recent study showed a lower all-cause mortality rate with RIPC. 51

Our multicentre study had several potential limitations.

1. We did not standardise pre-/perioperative anaesthesia and medication, although this was because we wanted to reflect current clinical practice in cardiac bypass surgery as much as possible.

2. The number of missing and incomplete data for some of the secondary end points.

3. Although an echocardiography substudy had been planned, because of logistical issues very few patients were included in the study and so no meaningful data on left ventricular function were available for analysis.

Note on version numbers

Protocol version 5 (6 September 2010) was the first ethically approved version of the protocol and so is considered the original protocol; versions 1–4 were drafts and have not been included in this document.

The following changes were made to version 6 (28 October 2010):

1. Further information has been added with regard to the definition of the primary objective, and the study end point has been clarified as cardiovascular death.

2. Treatment period changed – the RIPC protocol has been changed from three 5-minute cycles to four cycles of ischaemia and reperfusion applied to the upper arm. This is because of a recent publication19 which has shown that RIPC using four 5-minute cycles reduces infarct size in patients presenting with an acute MI.

3. Changed from peripheral vascular disease to peripheral arterial disease, as the former is vague and includes venous disease such as deep-vein thrombosis. Peripheral arterial disease is more accurate and specific for the patients we would like to include.

4. The laboratory where we plan to undertake the troponin T assays for all of the sites has switched to hsTnT, which has different cut-off values.

5. Further information included on calculating inotrope score.

6. Further details included with regard to ejection fraction protocol.

7. Clinical outcome confirmed as death and not cardiovascular death. Information will be collected on all deaths. The definition of cardiovascular death was improved.

The following changes were made to version 7 (14 January 2011):

1. The laboratory where the samples are to be collected has been changed and, therefore, the size of the tube in which blood is collected has changed.

The following changes were made to version 8 (1 March 2011):

1. Further centres have been added. It was originally felt that 11 centres would be adequate for recruitment for this trial. However, following an investigator meeting it was decided to increase the number of centres to ensure that the trial is recruiting on target and on time.

2. The total time for the intervention has been changed from 40 minutes to 35 minutes. This is because the last 5 minutes is when the cuff is deflated and so it can be removed after the final inflation.

3. Recruiting centres will be advised to avoid the routine use of intravenous GTN during surgery, as this agent may interfere with RIPC. The use of any GTN will be recorded on the patient’s CRF.

4. The following exclusion criterion has been removed: positive troponin T or I at baseline. This has been removed as this status may not always be known as troponin is not measured routinely in these patients.

5. A definition of cardiogenic shock has been included as requested by investigators. This is a standard definition.

6. The Doctors Laboratory will now analyse all troponin samples.

7. Information will be collected on the day of surgery on any angiograms that have been performed during the last 5 days. During this procedure, dye is used and this may affect some of the blood tests.

8. As we have removed a positive troponin T or I at baseline as an exclusion criterion but are still measuring troponin T and troponin I as part of the trial procedures pre and post surgery, they will be reviewed further as part of the statistical analysis plan.

9. Richard Evans has now been added as the assistant trial manager.

The following changes were made to version 9 (4 May 2011):

1. An earlier time point for assessment of MACCE was agreed to be of benefit. It has therefore been added as a secondary end point at 6 weeks (which coincides with the first follow-up appointment of the patients). In addition, all-cause death will be noted as a secondary end point.

2. The original plan was for the control group not to have the blood pressure cuff inflated. However, the reviewers were concerned that this may unblind more staff. In light of this we have changed the control group to a sham control group. When patients are randomised to the sham control group they will undergo 5 minutes of simulated inflations of the blood pressure cuff followed by deflation.

3. As centres use different types of blood bottles, EDTA has been removed to avoid confusion at sites. Blood samples can be stored at –20 °C and not –70 °C as originally thought.

4. Three-dimensional techniques have been added to measurement of the LVEF.

5. Samples for genetic analysis have now changed slightly following on from the advice of a biochemist.

6. Luciano Candilio has been added as a clinical fellow.

7. It has been decided that an independent EVC will be convened to adjudicate all of the events that make up the primary end point.

8. The membership for the TSC has been updated.

9. Changes to inclusion criteria – recruitment to the trial has now started and all patients who are undergoing CABG are screened. Many patients are not meeting the inclusion criteria and so we are modifying these to maximise recruitment to the trial. There are two main changes: the euroSCORE is being lowered from ≥ 6 to ≥ 5 and the word ‘cold’ has been removed from point 1.

10. Changes to exclusion criteria – bilirubin of > 20 mmol/l has been deleted as this will also maximise our recruitment and centres have been updated: Mr Aprim Youhana, Swansea Morriston Hospital, has been added and Bristol Royal Infirmary has been removed.

11. In relation to the inotrope score some centres use milrinone and so this has been added.

The following changes were made to version 10 (12 January 2012):

1. The PI at Harefield Hospital has been changed from Professor John Pepper to Dr Andre Simon. Nine other sites have been added to increase and accelerate recruitment.

2. The number of sites has increased from 16 to 28. The recruitment rates have been updated to reflect the involvement of extra hospitals and the timetable has been updated to reflect the training of 28 sites.

3. St Thomas’ Hospital has now also agreed to take part in the echo substudy.

4. Cardiogenic shock and cardiac arrest are listed on separate lines on the CRF. Hepatic dysfunction has been changed from using the prothrombin ratio definition to using the international normalised ratio to reflect current practice.

5. The inotrope score has been adjusted slightly to incorporate alternative inotropes.

6. Dry ice is now being provided by the courier company and not locally to simplify transfer procedures.

7. The study timetable has been corrected to remove creatinine and urine volume at baseline as these need to be recorded preoperatively only.

8. The reporting of events has been adjusted as many centres are submitting SAEs relating to complications of CABG. Listed are the most common complications of surgery, which we hope will lead to the reporting of fewer SAEs.

9. Wording changed in the consent procedure for inpatients recommending that they are given 24 hours to consider the trial.

10. Trial unblinded statistician updated from Cono Ariti to Jennifer Nicholas.

The following changes were made to version 11 (7 December 2012):

1. The number of recruiting sites has been increased from 28 to 30 to expedite recruitment.

2. The echo substudy has been opened to other sites to increase recruitment.

3. The wording for randomisation timing has been amended as some sites need to randomise earlier than the morning before surgery for logistical reasons.

4. The recruitment targets have been amended to reflect expected recruitment numbers.

5. The requirement to record a screenshot of the euroSCORE has been removed to reflect the different methods that sites use to calculate it. The screenshot of the euroSCORE calculator has been removed from the appendix.

6. Wording has been changed to reflect the fact that the EQ-5D questionnaire can be completed over the telephone.

7. Wording has been changed to reflect the fact that if a patient is unable to attend the 1-year follow-up it may be conducted over the telephone.

8. Infection of donor site has been added to the list of expected SAEs.

9. The job descriptions of Richard Evans, Steven Robertson and Rosemary Knight have been updated to reflect their current roles.

10. The details for the EVC have been added.

11. The primary end point definitions have been clarified after input from the EVC.

12. The list of local investigators has been finalised and updated.

The following changes were made to version 12 (27 November 2014):

1. Update to Derek Hausenloy’s title (Professor).

2. Update to remove troponin I from the list of blood samples as only hsTnT was collected for the trial.

3. Update to perioperative MI definition to match the current universal definition of MI. Updated to include probable MI as a category to allow for cases in which sufficient evidence to consider definite MI is not available.

4. Update to make clear that both definite and probable MIs will be included in the MACCE end point.

5. Update to stroke definition to allow probable stroke to be recorded when there is insufficient evidence to classify as definite stroke.

6. Update to make clear that both definite and probable strokes will be included in the MACCE end point.

7. EVC updated to reflect membership changes.

8. Reference added supporting change to MI definition.