|

 

Methylprednisolone in patients undergoing cardiopulmonary bypass (sirs): a randomised, double-blind, placebo-controlled trial

Methylprednisolone in patients undergoing
cardiopulmonary bypass (SIRS): a randomised, double-blind,
placebo-controlled trial
Richard P Whitlock, P J Devereaux, Kevin H Teoh, Andre Lamy, Jessica Vincent, Janice Pogue, Domenico Paparella, Daniel I Sessler, Ganesan Karthikeyan, Juan Carlos Villar, Yunxia Zuo, Álvaro Avezum, Mackenzie Quantz, Georgios I Tagarakis, Pallav J Shah, Seyed Hesameddin Abbasi, Hong Zheng, Shirley Pettit, Susan Chrolavicius, Salim Yusuf, for the SIRS Investigators* Summary
Background Cardiopulmonary bypass initiates a systemic infl ammatory response syndrome that is associated with Lancet 2015; 386: 1243–53
postoperative morbidity and mortality. Steroids suppress infl ammatory responses and might improve outcomes in See Editorial page 1212
patients at high risk of morbidity and mortality undergoing cardiopulmonary bypass. We aimed to assess the eff ects See Comment page 1215
of steroids in patients at high risk of morbidity and mortality undergoing cardiopulmonary bypass.
*Members listed in the appendix Population Health Research
Methods The Steroids In caRdiac Surgery (SIRS) study is a double-blind, randomised, controlled trial. We used a Institute, Hamilton Health
central computerised phone or interactive web system to randomly assign (1:1) patients at high risk of morbidity and Sciences and McMaster
mortality from 80 hospital or cardiac surgery centres in 18 countries undergoing cardiac surgery with the use of University, Hamilton, ON,
Canada (R P Whitlock MD,
cardiopulmonary bypass to receive either methylprednisolone (250 mg at anaesthetic induction and 250 mg at Prof P J Devereaux MD,
initiation of cardiopulmonary bypass) or placebo. Patients were assigned with block randomisation with random A Lamy MD, J Vincent MSc,
block sizes of 2, 4, or 6 and stratifi ed by centre. Patients aged 18 years or older were eligible if they had a European J Pogue PhD, S Pettit RN,
System for Cardiac Operative Risk Evaluation of at least 6. Patients were excluded if they were taking or expected to S Chrolavicius BA,
Prof S Yusuf DPhil); Department
receive systemic steroids in the immediate postoperative period or had a history of bacterial or fungal infection in the of Surgery (R P Whitlock,
preceding 30 days. Patients, caregivers, and those assessing outcomes were masked to allocation. The primary A Lamy), Department of Clinical
outcomes were 30-day mortality and a composite of death and major morbidity (ie, myocardial injury, stroke, renal Epidemiology and Biostatistics
failure, or respiratory failure) within 30 days, both analysed by intention to treat. Safety outcomes were also analysed (R P Whitlock, Prof P J Devereaux,
A Lamy), and Department of
by intention to treat. This study is registered with ClinicalTrials.gov, number NCT00427388.
Medicine (Prof P J Devereaux,
Prof S Yusuf), McMaster
Findings Patients were recruited between June 21, 2007, and Dec 19, 2013. Complete 30-day data was available for all University, Hamilton, ON,
Canada; Department of
7507 patients randomly assigned to methylprednisolone (n=3755) and to placebo (n=3752). Methylprednisolone, Surgery, Southlake Regional
compared with placebo, did not reduce the risk of death at 30 days (154 [4%] vs 177 [5%] patients; relative risk [RR] Health Centre, Newmarket,
0·87, 95% CI 0·70–1·07, p=0·19) or the risk of death or major morbidity (909 [24%] vs 885 [24%]; RR 1·03, 95% CI ON, Canada (K H Teoh MD);
0·95–1·11, p=0·52). The most common safety outcomes in the methylprednisolone and placebo group were infection University of Bari Aldo Moro,
Bari, Italy (D Paparella MD);
(465 [12%] vs 493 [13%]), surgical site infection (151 [4%] vs 151 [4%]), and delirium (295 [8%] vs 289 [8%]).
Department of Outcomes
Research, Cleveland Clinic,
Interpretation Methylprednisolone did not have a signifi cant eff ect on mortality or major morbidity after cardiac Cleveland, OH, USA
surgery with cardiopulmonary bypass. The SIRS trial does not support the routine use of methylprednisolone for (Prof D I Sessler MD); All India
Institute of Medical Sciences,
patients undergoing cardiopulmonary bypass.
New Delhi, India
(G Karthikeyan MD); Fundación
Funding Canadian Institutes of Health Research.
Cardio Infantil—Instituto de
Cardiología, Bogotá, Colombia
(J C Villar MD); Department of
Steroids attenuate the infl ammatory response to Anesthesiology, West China
Cardiac surgery is a common surgical procedure and cardiopulmonary bypass,16 but their eff ect on clinical Hospital, Sichuan University,
cardiopulmonary bypass is used in most procedures.1
outcomes is uncertain. Meta-analyses17 of small trials Chengdu, China
Cardiopulmonary bypass initiates a systemic infl am- suggest that steroids decrease perioperative atrial (Prof Y Zuo MD); Divisão de
Pesquisa, Instituto Dante
matory response syndrome, which is associated with fi brillation and possibly mortality, but defi nitive evidence Pazzanese de Cardiologia,
adverse clinical outcomes.2 Infl ammatory responses is not available. After this meta-analysis was published,17 São Paulo, Brazil
include activation of platelets, neutrophils, monocytes, the Dexamethasone for Cardiac Surgery (DECS) trial18 (Prof Á Avezum MD); London
macrophages, cascades (coagulation, fi brinolytic, and did not show a benefi t of dexamethasone for patients Health Sciences Centre,
London, ON, Canada
kallikrein),3–5 which results in increased endothelial undergoing cardiopulmonary bypass, but suggested that (M Quantz MD); Department of
permeability and vascular and parenchymal damage.6–9
steroids might benefi t patients at high risk of morbidity Cardiovascular and Thoracic
ammatory responses are associated with and mortality undergoing cardiopulmonary bypass.
Surgery, University of Thessaly,
the development of postoperative complications including We aimed to assess whether prophylactic steroids Larissa, Greece
(G I Tagarakis MD); Princess
myocardial injury and infarction, respiratory failure, renal benefi t patients at high risk of morbidity and mortality Alexandra Hospital,
and neurological dysfunction, excessive bleeding, altered undergoing cardiac surgery with cardiopulmonary Woolloongabba, Brisbane,
liver function, multiple organ failure, and death.10–15
QLD, Australia (P J Shah FRACS);
www.thelancet.com Vol 386 September 26, 2015
Tehran Heart Centre, Tehran
University of Medical Sciences,
Research in context
Tehran, Iran (S H Abbasi MD);
and First Teaching Hospital of
Evidence before this study
or respiratory failure within 30 days (RR 0·83, 0·67–1·01). Xinjiang Medical University,
In 2008, we published a systematic review and meta-analysis A subgroup analysis suggested the possibility that the steroid Urumqi, China
of 44 randomised trials assessing the eff ect of a steroid in was benefi cial in patients at higher risk of morbidity and (Prof H Zheng MD) patients undergoing cardiopulmonary bypass. These trials were mortality (EuroSCORE ≥5; RR 0·77, 0·61–0·98).
Correspondence to: identifi ed through a search of Embase, MEDLINE, Cochrane, Dr Richard Whitlock, Population Added value of this study
Health Research Institute, CINAHL, and OVID between 1977 and October, 2007, using the The SIRS trial included 7500 patients with a EuroSCORE of at Hamilton Health Sciences and search terms "cardiac surgery", "cardiac surgical procedure", least 6. In the SIRS trial, methylprednisolone compared with McMaster University, Hamilton, "open heart surgery", "coronary artery bypass", "mitral valve", ON L8L 2X2, Canada placebo had no eff ect on mortality (154 deaths vs 177; RR 0·87, "aortic valve", "heart valve", "cardiopulmonary bypass", 0·70–1·07). An updated meta-analysis that included "extracorporeal circulation", and "preoperative" and 14 027 patients showed no eff ect of steroids on mortality See Online for appendix
"prophylactic" in combination with generic and trade names of (RR 0·85, 0·71–1·02). SIRS identifi ed a signifi cant increase in steroid preparations. We hand searched the reference lists from myocardial injury based on raised cardiac enzymes.
eligible trials. Trials were eligible irrespective of their primary objective or language of publication. This meta-analysis Implications of all the available evidence
showed a non-signifi cant reduction in mortality with the use of The collective data from all trials suggests no benefi t to steroids (relative risk [RR] 0·73, 95% CI 0·45–1·18) that, if real, perioperative steroids but an increased risk of myocardial would be clinically important. This meta-analysis result was injury with routine use of steroids in patients undergoing based on few events (n=65 deaths) and the results were cardiac surgery; therefore, the routine use of steroids for cardiopulmonary bypass is cautioned. Future studies should elucidate the mechanism of myocardial injury associated with Subsequently, the Dexamethasone for Cardiac Surgery (DECS) the administration of steroids at the time of cardiac surgery trial of 4494 patients did not show a reduction in mortality with cardiopulmonary bypass.
(RR 0·92, 0·57–1·49) or a signifi cant reduction in the primary outcome of death, myocardial infarction, stroke, renal failure, had an allergy or intolerance to steroids, were expected Study design and participants
to receive aprotinin, or had previously participated The Steroids In caRdiac Surgery (SIRS) trial was an in SIRS. The protocol has already been published.23 international, multicentre, parallel-group, double- All participating sites obtained institutional ethics blind, randomised, placebo-controlled trial of adult approval.
patients at high risk of morbidity and mortality
undergoing cardiopulmonary bypass. Patients were Randomisation and masking
recruited from 80 hospital-based cardiac surgery Patients were randomly assigned (1:1) by the local study
practices in 18 countries by dedicated local research team using a central computerised phone or interactive
teams. Patients aged 18 years or older were eligible if web system to receive either intravenous methyl-
they had a European System for Cardiac Operative Risk
prednisolone 250 mg at anaesthetic induction and Evaluation (EuroSCORE) of at least 6 and provided 250 mg at initiation of cardiopulmonary bypass or written informed consent. At the time of initiation of placebo. Patients were assigned by block randomisation SIRS, EuroSCORE was a widely used and validated risk with random block sizes of 2, 4, or 6, stratifi ed by centre.
prediction index.19 We needed patients to have a Methylprednisolone was obtained from the centre's EuroSCORE of at least 6 because previous research20 local pharmacy, and the study drug was prepared and identifi ed that patients with this score were at high risk masked by the local pharmacy following procedures of mortality and would have an expected mortality rate described in a provided study manual. Patients, of at least 6%. From July 5, 2011, in China and India, we health-care providers, data collectors, and outcome allowed inclusion of patients with a EuroSCORE of at adjudicators were masked to treatment allocation.
least 4 if the patient was undergoing valvular surgery
because research21,22 showed that patients from China Procedures
and India with these lower EuroSCOREs had higher In previous trials24 of patients undergoing cardio
than expected mortality rates (ie, observed mortality monary bypass, the most common methylprednisolone rates of 4·9% with EuroSCORE predicted mortality dose was 30 mg/kg (ie, >2 g for a 70 kg patient). We rates of 3·0%). Patients were excluded if they were decided, however, to use a cumulative dose of 500 mg of taking or expected to receive systemic steroids in the methylprednisolone given intraoperatively for the immediate postoperative period, had a history of following reasons: data for surrogate endpoints from the bacterial or fungal infection in the preceding 30 days, SIRS pilot study16 showed that this lower dose was www.thelancet.com Vol 386 September 26, 2015
eff ective in abolishing the infl ammatory response to measure CK-MB. Second, at the time of starting the cardiopulmonary bypass across a broad array of measured SIRS trial, substantially more data were available for mediators; a meta-analysis by Ho and Tan24 suggested CK-MB, suggesting CK-MB was a prognostically that low-dose and moderate-dose corticosteroids are as important measure of myocardial injury after cardio-eff ective as high-dose cortico steroids in improving pulmonary bypass compared with troponin measure- clinical outcomes; and data from this meta-analysis24 also ments. Third, although we recognised that we would suggested that higher doses of steroids might be have to assess two CK-MB assays (ie, mass and activity), associated with adverse outcomes such as prolonged determination of the optimum thresholds for troponin ventilation. Therefore, the dose of 500 mg of would be more challenging and we would probably methylprednisolone seemed to strike the optimum have had insuffi cient power because in addition to balance between potential effi cacy and safety.
troponin T and several troponin I assays, we would also Electrocardiograms were done preoperatively, at 24 h encounter both non-high-sensitive and high-sensitive postoperatively, and at hospital discharge or on troponin assays for both troponin T and I assays. postoperative day 4, whichever was earlier. Creatine Therefore, from a masked analysis of the fi rst kinase myocardial band (CK-MB) was measured 7000 patients included in SIRS, we used a modifi cation preoperatively, and at 8 h and 24 h after surgery. of the method developed by Mazumdar and Creatinine was measured preoperatively. Study colleagues25,26 to identify the lowest CK-MB threshold personnel collected all in-hospital creatinine measure- that had an independent hazard ratio (HR) of more ments until 14 days after surgery, peak blood glucose than 2 for 30-day mortality after adjustment for patients' measurement until 24 h after surgery, and the EuroSCORE values. Thresholds were established for Confusion Assessment Method (CAM) delirium score CK-MB measured by mass assay and by activity assay, on postoperative day 3. After randomisation, patients as well as separately for patients who had isolated were followed up for 30 days for all outcomes and for coronary artery bypass and for those having other 6 months for vital status.
cardiac surgeries. On the basis of these analyses, the SIRS protocol diagnostic criterion of early myocardial injury was any one of the following: for the mass assay, Primary outcomes were mortality at 30 days after a CK-MB measurement at least six times the upper randomisation and a composite of death, myocardial limit of normal in patients who had isolated coronary injury, stroke, renal failure (stage 3 acute kidney injury, artery bypass or at least 15 times the upper limit of 2012 Kidney Disease Improving Global Outcomes normal in patients who had other cardiac surgery; for [KDIGO] guidelines), or respiratory failure (un interrupted the activity assay, a CK-MB activity measurement of at postoperative mechanical ventilation for more than 48 h) least 40 U/L in patients who had isolated coronary at 30 days after randomisation. 30-day secondary artery bypass or at least 120 U/L in patients who had outcomes included individual components of the primary other cardiac surgery; angiographic graft occlusion or composite outcome, myocardial injury or mortality, new new native coronary artery occlusion; or imaging atrial fi brillation, chest drain output during the fi rst 24 h evidence of new loss of viable myocardium.
after surgery, the number of patients with transfusions In patients with a CK-MB greater than the upper during the fi rst 24 h after surgery, the duration of limit of normal before randomisation, an absolute mechanical ventilation, duration of intensive care unit increase in a patient's CK-MB measurement based on stay, and length of hospital stay. Safety outcomes included our stated defi nition was needed. For example, if a infection, stroke, wound complications (superfi cial or patient undergoing isolated coronary artery bypass deep surgical site infection, or sterile wound dehiscence), surgery had a myocardial injury before cardiac surgery gastrointestinal haemorrhage, gastro intestinal perfor- and had a peak CK-MB activity assay value of 60 U/L ation within 30 days, delirium on postoperative day 3, and 2 days before surgery, this patient would need a CK-MB postoperative insulin use and peak blood glucose during activity assay measurement of at least 100 U/L during the fi rst 24 h after surgery. We also assessed mortality at the fi rst 72 h after surgery to fulfi l the diagnostic 6 months.
criterion of myocardial injury.
Few data inform the diagnostic criteria for early Outcome defi nitions are described in more detail in myocardial injury after cardiac surgery (ie, within 72 h the appendix. An adjudication committee whose of surgery), particularly after non-coronary-artery- members were unaware of study group assignments bypass-graft surgery. In this study, we established a assessed all deaths and myocardial injuries and their
prognostically relevant postoperative CK-MB threshold decisions were used in the analyses.
for the protocol diagnostic criteria of early myocardial
injury. We decided to use CK-MB instead of troponin Statistical analysis
for the following reasons. First, some centres We planned to enrol 7500 patients to have more than
participating in SIRS did not have access to troponin 80% power to detect a 25% relative risk (RR) reduction
measurements, whereas all centres were able to for the fi rst primary outcome of death at 30 days with an
www.thelancet.com Vol 386 September 26, 2015
α of 0·041 (two-sided), anticipating a 6% mortality rate secondary outcomes with a χ² test or a t test (or a in the control group. The study had more than 99·9% non-parametric test if data were not normally distributed) power to detect a 20% RR reduction for the second where appropriate.
primary outcome of death, myocardial injury, stroke, We did subgroup analyses based on sex, diabetes, renal failure, or respiratory failure at 30 days with an α EuroSCORE, cardiopulmonary bypass duration, age, of 0·01 (two-sided), anticipating a 25% rate in the and surgery type by stratifi ed analyses through logistic control group.
regression or Cox proportional hazards models, as For the analysis of the two primary effi cacy outcomes, appropriate. A priori, we stated the expected direction of the overall type I error was partitioned. We tested the fi rst eff ects in the subgroups. The test of interaction between primary outcome of death at a 0·041 level of signifi cance each subgroup factor and the treatment group was done and the second primary outcome at a level of 0·01. This by including a product term in the model already maintained the overall type I error rate for both primary containing treatment and the subgroup factor, comparisons at 5%, under the assumption that the designated as signifi cant at p<0·05. An independent overlap between the two outcomes was at least 15%. data and safety monitoring board reviewed the interim The non-additivity of the type I error rates shows the analyses when 50% and 75% of the 30-day follow-up data correlation between these two outcomes, estimated from were available. SAS version 9.1 was used for all analyses.
10 million simulations with the Mantel-Haenszel statistic.
This study is registered with ClinicalTrials.gov, number All analyses were based on the intention-to-treat NCT00427388.
principle. We compared the proportions of patients developing the primary outcomes using the Pearson χ² Role of the funding source
test. The RR was calculated along with its 95% CIs The Population Health Research Institute designed and associated with methylprednisolone. We compared all coordinated the trial and was responsible for the 7507 randomly assigned 3755 allocated to receive methylprednisolone 3752 allocated to placebo 146 did not receive allocated intervention* 154 did not receive allocated intervention* 40 did not undergo surgery 50 did not undergo surgery 15 given non-study steroids 19 given non-study steroids 13 surgery cancelled or rescheduled and study personnel unaware of this 6 study personnel did not administer 5 study personnel did not administer 5 surgery cancelled or rescheduled and 4 were ineligible after randomisation 4 were ineligible after randomisation 2 surgeon refused study drug 2 surgeon refused study drug 3755 included in 30-day intention-to treat 3752 included in 30-day intention-to treat 3 lost to follow-up because they 5 lost to follow-up because they could not be contacted and vital could not be contacted and vital status could not be obtained status could not be obtained through a third party through a third party 3755 included in 6-month intention-to treat 3752 included in 6-month intention-to treat Figure 1: Trial profi le
*Data were not collected for the fi rst 490 patients in the pilot study.
www.thelancet.com Vol 386 September 26, 2015
Methylprednisolone Placebo
Methylprednisolone Placebo
(Continued from previous column) Vitamin K antagonists H2 antagonist or proton-pump inhibitor Oral hypoglycaemic Repeat cardiac surgery Coexisting medical conditions Any cardiac valve Any coronary artery Isolated cardiac valve Isolated coronary artery Previous myocardial Bypass time (min) Congestive heart failure Cross-clamp time (min) Hypothermic arrest Peripheral arterial disease† Hypothermic arrest time Preoperative atrial Chronic renal failure Antifi brinolytic drugs Preoperative inotropes, Preoperative creatinine in vasopressors, intra-aortic patients not on dialysis balloon pump, or ventricular Chronic obstructive pulmonary disease Received at least one dose of Peptic ulcer disease study drug or placebo ¶ Previous gastrointestinal Non-study postoperative Data are mean (SD), n (%), or median (IQR). CK-MB=creatine kinase myocardial band. ACE=angiotensin converting enzyme inhibitor. ARB=angiotensin II receptor blocker. *The body-mass index is the weight (kg) divided by the square of the height (m). †Diagnosis of any one of intermittent claudication, vascular surgery for atherosclerotic disease, ankle to arm systolic ratio of at least 0·90 in either leg at rest, angiographic, or Doppler study showing at least 70% stenosis in a non-cardiac artery. ‡From ejection fraction or reported. §Angiotensin converting enzyme inhibitor or angiotensin II receptor blocker. ¶Data for these variables were not collected in the pilot phase of the fi rst 490 patients, thus the denominator is slightly lower. **Thienopyridine, ticagrelor, glycoprotein IIb, or glycoprotein IIIa inhibitor. Left ventricular grade I‡ Coronary stenosis >50% Table 1: Baseline characteristics
Left anterior descending randomisation scheme, database, data validation, and analyses. The Operations Committee designed the trial, prespecifi ed the statistical analysis plan, and were Valvular heart disease responsible for the data and analyses. The funders of Preoperative medications the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had fi nal responsibility for the decision to submit for publication.
Other antiplatelet drugs** (Table 1 continues in next column) Between June 21, 2007, and Dec 19, 2013, we recruited
7507 patients to this study (fi gure 1). The 30-day
www.thelancet.com Vol 386 September 26, 2015
Methylprednisolone Placebo (n=3752)
Relative risk (95% CI)
0·87 (0·70–1·07) Death, myocardial injury, stroke, new renal failure, or respiratory 1·03 (0·95–1·11) Components of composite primary outcome
Myocardial injury 1·22 (1·07–1·38) 0·90 (0·65–1·23) New renal failure 0·87 (0·69–1·08) Respiratory failure 0·91 (0·79–1·05) Death or myocardial injury 1·14 (1·02–1·27) New atrial fi brillation 0·98 (0·94–1·03) Chest drain output (mL) Length of ICU stay (h) 46·0 (23·0–90·0) 47·0 (24·0–91·0) Length of hospital stay (days) 9·0 (7·0–13·0) 9·0 (7·0–13·0) 0·94 (0·84–1·06) Surgical site infection 1·00 (0·80–1·25) 1·02 (0·87–1·19) Gastrointestinal perforation or haemorrhage 1·19 (0·81–1·76) Peak blood glucose (mmol/L) Postoperative insulin (units) Data are n (%), median (IQR), or mean (SD), unless otherwise specifi ed. ICU=intensive care unit.
Table 2: Eff ects of methylprednisolone on outcomes
patients received at least one dose of the study drug The primary outcome of death at 30 days occurred in 909 (24%)
885 (24%)
154 patients (4%) allocated to methylprednisolone and in 177 patients (5%) allocated to placebo (RR for the Myocardial injury methylprednisolone group 0·87, 95% CI 0·70–1·07, p=0·19; table 2). The primary outcome of death, myocardial injury, stroke, new renal failure, or respiratory Respiratory failure failure at 30 days occurred in 909 patients (24%) allocated to methyl prednisolone and in 885 patients (24%) allocated to placebo (RR 1·03, 95% CI 0·95–1·11; fi gure 2).
Secondary outcomes of stage 3 renal failure, stroke, Figure 2: Relative risk of the composite primary outcome and its components
brillation, transfusion requirements, The horizontal lines represent the 95% CI. infection, gastrointestinal complications, delirium, respiratory failure, length of intensive care unit stay, or length of hospital stay did not diff er between the follow-up was complete in 7507 patients (100%) and the methylprednisolone and placebo groups. The most 6-month follow-up was complete in 7499 patients common adverse events in both groups were infection,
(99·9%). All 7507 patients were included in the 30-day surgical site infection, and delirium (table 2).
and 6-month intention-to-treat analyses.
Methylprednisolone reduced the chest drain output The baseline characteristics were similar in the during the fi rst 24 h after surgery (table 2). The peak methylprednisolone and placebo groups (table 1). Of postoperative serum glucose and the postoperative 7507 patients, the mean age was 67·4 years (SD 13·7), insulin requirements in the fi rst 24 h after surgery were 1587 (21%) underwent isolated coronary artery bypass both increased by methylprednisolone (table 2).
surgery, 2437 (32%) underwent isolated valvular surgery, The eff ects of methylprednisolone on the primary 3379 (45%) underwent combined or other surgery, and outcomes did not vary signifi cantly across the the average EuroSCORE was 7·1 (SD 2·0). 96% of prespecifi ed subgroups (fi gure 3). The Kaplan-Meier www.thelancet.com Vol 386 September 26, 2015
Figure 3: Primary outcome of
death (A) and primary
composite outcome (B) by
prespecifi ed subgroups
The size of the black box is proportional to the weight and the horizontal lines represent the 95% CI. CABG=coronary artery bypass grafting.
www.thelancet.com Vol 386 September 26, 2015
Furthermore, myocardial injury increased mortality in both the methylprednisolone and placebo treatment =0·07; appendix). With the high use of statin therapy in this population, we assessed whether an interaction between methylprednisolone and myocardial injury existed because both can induce myopathy. The excess in enzymatic evidence of myocardial injury was similar in those receiving and not receiving a statin (p =0·75). Finally, we did a post-hoc analysis to assess the eff ect of methyl- prednisolone on the primary outcomes by geographical Cumulative event rate region and identifi ed no signifi cant interaction Discussion
In patients at high risk of morbidity and mortality
undergoing cardiac surgery with the use of cardio- Number at risk
pulmonary bypass, administration of perioperative Methylprednisolone 3755 methylprednisolone did not decrease the risk of death, or the composite risk of death, myocardial injury, stroke, renal failure, and respiratory failure at 30 days. Perioperative methylprednisolone did, however, signifi cantly increase the risk of myocardial injury (as defi ned by the SIRS protocol; RR 1·22, 95% CI Our previously published systematic review17 that included 44 randomised controlled trials of 3205 patients suggested that perioperative steroids reduced the risk of new postoperative atrial fi brillation (424 total events; RR 0·71, 95% CI 0·59–0·87), postoperative bleeding Cumulative event rate (weighted mean diff erence –100 mL, 95% CI –150 to –59), length of intensive care unit stay (weighted mean diff erence –0·23 days, –0·40 to –0·07), and length of hospital stay (weighted mean diff erence, –0·59 days; –1·17 to –0·02). The RR for the eff ect of steroids on mortality was 0·73 (95% CI 0·45–1·18), but Follow-up (months) Number at risk
this was based on only 65 deaths, indicating substantial Methylprednisolone 3755 The DECS trial18 randomly assigned 4494 patients to Figure 4: Eff ects of methylprednisolone versus placebo on mortality at 30 days and 6 months shown on a
receive one intraoperative dose of dexamethasone scale from 0 to 1·0 (A) and expanded scale from 0 to 0·1(B)
1 mg/kg or placebo. The primary outcome was a composite of death, myocardial infarction, stroke, renal curve for 6-month mortality by treatment group failure, or respiratory failure within 30 days, which was shows no eff ect of methylprednisolone on mortality one of the primary outcomes in our study. In the DECS (fi gure 4).
trial, 7% of patients had the primary outcome in the Methylprednisolone increased the risk of myocardial dexamethasone group versus 8% in the placebo group injury (as defi ned in our protocol) compared with (RR 0·83, 95% CI 0·67–1·01, p=0·07). A statistically placebo (table 2). A post-hoc analysis showed that the signifi cant benefi t of dexamethasone was reported in a risk of Q-wave myocardial infarction did not diff er prespecifi ed subgroup of patients with a EuroSCORE of between patients allocated to methylprednisolone at least 5 (RR 0·77, 95% CI 0·61–0·98); however, the (21 patients [<1%]) and those allocated to placebo test for interaction was not signifi cant (p (24 patients [<1%]; RR 0·87, 95% CI 0·49–1·57). The Dexamethasone also signifi cantly reduced infection increase in myocardial injury with methylprednisolone (212 patients [9%] vs 333 patients [15%]; RR 0·64, was reported in early events that were non Q-wave 0·54–0·75), delirium (RR 0·79, 0·66–0·94), and (RR 1·24, 1·09–1·41). This early myocardial injury respiratory failure (RR 0·69, 0·51–0·94).
increased the RR of death at 30 days compared with Our study did not show a benefi t of perioperative those without myocardial injury (RR 2·3, 1·8–2·9). steroids on death or major morbidity in patients at high www.thelancet.com Vol 386 September 26, 2015
RR for all-cause mortality
RR (95% CI)
n (events)
n (events)
3·00 (0·12–96·71) 0·50 (0·04–4·30) 0·50 (0·04–4·30) 0·17 (0·01–3·66) 0·37 (0·07–1·63) Liakopolous (2007) 0·67 (0·11–3·61) 0·20 (0·01–3·45) 0·55 (0·20–1·31) 0·69 (0·12–3·85) 0·92 (0·56–1·48) 0·87 (0·70–1·07) 7039 (213)
6988 (248)
0·85 (0·71–1·02)
Figure 5: 2008 meta-analysis updated with data from SIRS and DECS trials
RR=relative risk.
risk of morbidity and mortality as suggested in the incidence of 2%), whereas in SIRS, we used systematic DECS trial. We have updated our 2008 meta-analysis17 monitoring of CK-MB and identifi ed 885 myocardial with the SIRS and DECS mortality data (fi gure 5). This injuries (12%). Another trial in cardiac surgery that meta-analysis includes data for 14 027 patients and does mandated collection of CK-MB reported an incidence of not show a signifi cant reduction in mortality with myocardial injury of 11% with a similar defi nition to the steroids (RR 0·84, 95% CI 0·71–1·02, p=0·08). This SIRS defi nition.20 Perioperative myocardial injury based fi nding suggests that the potential benefi ts, if any, are on biomarker increase alone is associated with an likely to be much smaller than previously postulated. increase in short-term and long-term mortality and Both the DECS trial and SIRS identifi ed no reduction therefore the excess myocardial injury reported in our in new postoperative atrial fi brillation that was trial is probably clinically important.28 suggested by our previous meta-analysis.17 SIRS did Although the mechanism of the increase in CK-MB show a statistically signifi cant reduction in bleeding in with steroids is not clear, we have generated several the fi rst 24 h after surgery; however, we noted no eff ect hypotheses. First, steroids might impair metabolic on transfusion needs, suggesting that the small modulation of the ischaemic insult. After ischaemia, diff erence between the median chest drain output glucose entry into the myocyte is essential for its (40 mL) was not clinically important.
recovery of contractile function. Steroids induce The increase in myocardial injury noted with insulin resistance, as shown by the higher peak serum methylprednisolone in this study was not reported in glucose measurements and the increased need for DECS or previous trials. The DECS trial did not, insulin use in the methylprednisolone group. This however, mandate the measurement of postoperative insulin resistance might block glucose from entering biomarkers, and probably did not detect some the myocyte and thereby worsen ischaemic injury. prognostically important asymptomatic myocardial Second, steroids might alter CK-MB release or its injuries that were masked by analgesic medications clearance; for example, through a mechanism of early after surgery. Furthermore, the DECS trial used steroid-induced myositis in which an increased release the Universal Defi nition of Myocardial Infarction, of skeletal CK-MB might occur. However, we could not which the Joint Task Force indicated was based fi nd any supportive evidence for this possibility. on arbitrary biomarker thresholds.27 Therefore, a Roberts and colleagues29 showed that methyl-signifi cant number of clinically important myocardial prednisolone had no direct eff injuries were probably missed in the DECS trial (ie, only phosphokinase concentrations or the kinetics of its 74 myocardial infarctions were reported, with an release. Furthermore, the increase in CK-MB www.thelancet.com Vol 386 September 26, 2015
concentrations in our study was associated with a 2 Paparella D, Yau TM, Young E. Cardiopulmonary bypass induced doubling of risk of death at 30 days in the infl ammation: pathophysiology and treatment. An update.
Eur J Cardiothorac Surg 2002; 21: 232–44.
methylprednisolone group, suggesting that its increase Asimakopoulos G, Smith PL, Ratnatunga CP, Taylor KM. probably has adverse prognostic consequences. Third, Lung injury and acute respiratory distress syndrome after infl ammation after myocardial injury has been shown cardiopulmonary bypass. Ann Thorac Surg 1999; 68: 1107–15.
to be a prerequisite of healing, and steroids might 4 Boyle EM, Pohlman TH Jr, Johnson MC, Verrier ED. Endothelial cell injury in cardiovascular surgery: the systemic infl ammatory interfere with this process.30 Early studies of steroids response. Ann Thorac Surg 1997; 63: 277–84.
after acute myocardial infarction increased infarct 5 Mojcik CF, Levy JH. Aprotinin and the systemic infl ammatory size.29 Therefore, the suppression of infl ammation response after cardiopulmonary bypass. Ann Thorac Surg 2001;
71: 745–54.
after cardiac injury might inhibit necessary processes 6 Chaney MA. Corticosteroids and cardiopulmonary bypass: a review of myocardial repair, specifi cally, neutrophil and of clinical investigations. Chest 2002; 121: 921–31.
macrophage recruit ment.31,32 Although the exact mech- Kirklin JK, McGiffi n DC. Early complications following cardiac surgery. Cardiovasc Clin 1987; anism is not clear, the highly signifi cant association Teoh KH, Bradley CA, Gauldie J, Burrows H. Steroid inhibition of with increased mortality in such patients should urge cytokine-mediated vasodilation after warm heart surgery. Circulation substantial caution in using steroids in patients 1995; 92 (suppl 9): II347–53.
undergoing cardiac surgery.
Vincent JL, Wan S, Yim AP. Steroids in cardiopulmonary bypass.
Crit Care Med 2000; 28: 3373–74.
The defi nition of myocardial injury after cardiac 10 Kirklin JK, Westaby S, Blackstone EH, Kirklin JW, Chenoweth DE, surgery is challenging in that all patients have cardiac Pacifi co AD. Complement and the damaging eff ects of biomarker release by the very nature of the surgery. cardiopulmonary bypass. J Thorac Cardiovasc Surg 1983;
86: 845–57.
Robust evidence to determine thresholds of biomarkers 11 Levi M, Cromheecke ME, de Jonge E, et al. Pharmacological to defi ne clinically important myocardial injury is not strategies to decrease excessive blood loss in cardiac surgery: available, particularly for non-coronary artery bypass a meta-analysis of clinically relevant endpoints. Lancet 1999;
354: 1940–47.
cardiac surgery. However, our approach to defi ning 12 Royston D. Preventing the infl ammatory response to open-heart myocardial injury in SIRS was objective, defi ned a surgery: the role of aprotinin and other protease inhibitors. priori, and was associated with increased mortality, Int J Cardiol 1996; 53 (suppl): S11–37.
suggesting that the approach we used is clinically 13 Rubens FD, Mesana T. The infl ammatory response to cardiopulmonary bypass: a therapeutic overview. Perfusion 2004; relevant. However, if this fi nding is an epiphenomenon 19 (suppl 1): S5–12.
rather than a clinically relevant eff ect, it does not alter 14 Wan S, LeClerc JL, Schmartz D, et al. Hepatic release of the fact that we did not show a benefi t in important interleukin-10 during cardiopulmonary bypass in steroid-pretreated
patients. Am Heart J 1997; 133: 335–39.
outcomes with methylprednisolone.
15 Whitlock RP, Rubens FD, Young E, Teoh KH. Pro: steroids should We did not measure infl ammatory markers in this be used for cardiopulmonary bypass. J Cardiothorac Vasc Anesth study. However, we used the same dosing regimen in 2005; 19: 250–54.
this study that we used in the SIRS pilot study that 16 Whitlock RP, Young E, Noora J, Farrokhyar F, Blackall M, Teoh KH. Pulse low dose steroids attenuate post-cardiopulmonary bypass showed our dosing regimen was eff ective to prevent the SIRS; SIRS I. J Surg Res 2006; 132: 188–94.
infl ammatory response to cardiopulmonary bypass.16 17 Whitlock RP, Chan S, Devereaux PJ, et al. Clinical benefi t of The SIRS trial does not support the routine use of steroid use in patients undergoing cardiopulmonary bypass: a meta-analysis of randomized trials. Eur Heart J 2008; methylprednisolone for patients undergoing cardio- pulmonary bypass, but does suggest an increased risk 18 Dieleman JM, Nierich AP, Rosseel PM, et al. Intraoperative of myocardial injury. The routine use of steroids high-dose dexamethasone for cardiac surgery: a randomized
controlled trial. JAMA 2012; 308: 1761–67.
for cardiopulmonary bypass is cautioned. Further 19 Nashef SA, Roques F, Michel P, Gauducheau E, Lemeshow S, research should elucidate the mechanism by which Salamon R. European system for cardiac operative risk evaluation methylprednisolone led to an increased risk of (EuroSCORE). Eur J Cardiothorac Surg 1999; 16: 9–13.
myocardial injury.
20 Verrier ED, Shernan SK, Taylor KM, et al. Terminal complement blockade with pexelizumab during coronary artery bypass graft surgery requiring cardiopulmonary bypass: a randomized trial. RPW, KHT, and SY initiated the study. RPW and SY wrote the fi rst and JAMA 2004; 291: 2319–27.
fi nal drafts of the report. JP analysed the data. PJD, KHT, AL, JV, DP, 21 Wang C, Yao F, Han L, Zhu J, Xu ZY. Validation of the European DIS, GK, JCV, YZ, AA, MQ, GIT, PJS, SHA, HZ, SP, and SC system for cardiac operative risk evaluation (EuroSCORE) in contributed to each draft of the report. All authors were international Chinese heart valve surgery patients. J Heart Valve Dis 2010;
19: 21–27.
leads, members of the International Steering Committee of the SIRS trial, or members of the central coordinating offi 22 Malik M, Chauhan S, Malik V, Gharde P, Kiran U, Pandey RM. Is EuroSCORE applicable to Indian patients undergoing cardiac Declaration of interests
surgery? Ann Card Anaesth 2010; 13: 241–45.
We declare no competing interests.
23 Whitlock R, Teoh K, Vincent J, et al. Rationale and design of the steroids in cardiac surgery trial. Am Heart J 2014; SIRS was funded by the Canadian Institutes of Health Research and the 24 Ho KM, Tan JA. Benefi ts and risks of corticosteroid prophylaxis in Canadian Network and Centre for Trials Internationally.
adult cardiac surgery: a dose-response meta-analysis. Circulation 2009; 119: 1853–66.
Roger VL, Go AS, Lloyd-Jones DM, et al. Heart disease and stroke 25 Mazumdar M, Smith A, Bacik J. Methods for categorizing a statistics—2012 update: a report from the American Heart prognostic variable in a multivariable setting. Stat Med 2003; Association. Circulation 2012; 125: e2–220.
www.thelancet.com Vol 386 September 26, 2015
26 Mazumdar M, Glassman JR. Categorizing a prognostic variable: 30 Frangogiannis NG, Smith CW, Entman ML. The review of methods, code for easy implementation and applications infl ammatory response in myocardial infarction. Cardiovasc Res to decision-making about cancer treatments. Stat Med 2000; 2002; 53: 31–47.
31 Hammerman H, Kloner RA, Hale S, Schoen FJ, Braunwald E. 27 Thygesen K, Alpert JS, White HD, et al. Universal Dose-dependent eff ects of short-term methylprednisolone on defi nition of myocardial infarction. Circulation 2007; 116: 2634–53.
myocardial infarct extent, scar formation, and ventricular function. 28 Domanski MJ, Mahaff ey K, Hasselblad V, et al. Circulation 1983; 68: 446–52.
Association of myocardial enzyme elevation and survival 32 Bulkley BH, Roberts WC. Steroid therapy during acute myocardial following coronary artery bypass graft surgery. JAMA 2011; infarction. A cause of delayed healing and of ventricular aneurysm. Am J Med 1974; 56: 244–50.
29 Roberts R, DeMello V, Sobel BE. Deleterious eff ects of methylprednisolone in patients with myocardial infarction.
Circulation 1976; 53: 1204–06.
www.thelancet.com Vol 386 September 26, 2015

Source: https://events.vanderbilt.edu/uploads/Methylprednisolone%20for%20CPB%20-%20SIRS%20study.pdf