Intraoperative High-Dose Dexamethasone for Cardiac Surgery:  A Randomized Controlled Trial FREE

Cardiac surgery is among the most commonly performed surgical procedures.¹ Despite important improvements in surgical technique, anesthesia management, and postoperative care, cardiac surgery is still associated with a substantial risk of major adverse events.^1- 3

Cardiopulmonary bypass (CPB) may play a role in the development of many of these adverse outcomes.^4- 5 Cardiopulmonary bypass induces a complex acute phase response, characterized by both cell and protein activation, which may further be intensified by the surgical trauma, ischemia-reperfusion injury, and endotoxemia. In a significant number of patients, a postoperative systemic inflammatory response syndrome develops, characterized by fever, organ dysfunction, and even multisystem organ failure. It therefore seems reasonable to try to attenuate the inflammatory response, which is part of routine care in many cardiac surgical centers. Often, this is accomplished with long-acting corticosteroids, typically by administering a high dose of intravenous methylprednisolone or dexamethasone during the surgery. These drugs are low-cost, potent anti-inflammatory agents and therefore represent an appealing treatment option in this scenario.^6- 7 However, concerns about potential adverse effects, including inadequate serum glucose control, infectious complications, poor wound healing, and gastrointestinal bleeding, have precluded their widespread adoption in cardiac surgical practice.^8- 9 In contrast with several European countries, corticosteroids are not routinely used during cardiac surgery in most centers in the United States.¹⁰

Previous studies have shown that corticosteroids attenuate the increase of serum inflammatory markers and may improve pulmonary gas exchange and reduce the need for postoperative inotropic support.⁷ However, appropriately sized studies on important clinical outcomes are lacking. Also, recent meta-analyses did not generate sufficient statistical power to allow conclusions on the effectiveness of corticosteroids in the reduction of major adverse events.^11- 13 As a result, corticosteroid administration during cardiac surgery is still controversial—in many European hospitals it is part of routine care, whereas this is not the case in most North American cardiac surgical centers.¹⁴

We conducted a large randomized clinical trial to quantify the effect of a single intraoperative dose of dexamethasone on the incidence of major adverse events in patients undergoing cardiac surgery.

The Dexamethasone for Cardiac Surgery (DECS) study is a multicenter, randomized, double-blind, placebo-controlled study comparing high-dose intravenous dexamethasone with placebo treatment in patients undergoing cardiac surgery. Between April 13, 2006, and November 23, 2011, we recruited patients in 8 cardiac surgical centers in the Netherlands. Patients aged 18 years or older who were scheduled for any type of elective or urgent cardiac surgical procedure requiring CPB were considered eligible. Exclusion criteria included an emergent or off-pump procedure and a life expectancy of less than 6 months.

The study protocol was designed by the academic authors, in collaboration with the members of the DECS Study Group. The study was conducted in accordance with Good Clinical Practice principles and applicable national regulations. The research ethics committee at each participating center approved the protocol. All patients were required to provide written informed consent before randomization.

After providing written informed consent, patients were randomized to receive either dexamethasone or placebo treatment. Dexamethasone (1 mg/kg of body weight, with a 100 mg maximum) or placebo was administered as a single intravenous injection after induction of anesthesia, but before initiation of CPB. The study drug was supplied in packaged ampoules, each assigned to a unique study number. Packages and ampoules of dexamethasone and placebo were identical and contained an equal volume (5 mL) of a 20 mg/mL dexamethasone solution or normal saline, respectively. An independent statistician created a computer-generated 1:1 randomization scheme, which was stratified to participating center and in blocks of 40. The research pharmacist of the University Medical Center Utrecht, Utrecht, the Netherlands, prepared and delivered batches of 40 ampoules to each center. When a consenting patient arrived in the operating department, a packaged ampoule was taken from the batch. When the ampoule had been opened and the study drug was administered, the patient was considered randomized and the corresponding study number was assigned to that patient. Patients, caregivers, and researchers were unaware of study group assignment.

Anesthesia and surgical treatment were performed according to the standard procedures of each participating center. Surgical access to the heart was achieved via midline sternotomy. The anesthetic technique was based on either total intravenous anesthesia or a combination of intravenous opioids and muscle relaxants in combination with volatile anesthetics. Techniques for cardioplegia, myocardial protection, and CPB, as well as use of inotropic support, antifibrinolytic therapy, and cell saving techniques, were left at the discretion of the attending team. Use of corticosteroid-containing solutions for cardioplegia or bypass circuit prime was not allowed. When indicated, patients taking preoperative systemic corticosteroids received a perioperative corticosteroid stress regimen.

After surgery, patients were transferred to the intensive care unit (ICU) and weaned from mechanical ventilation when there was no excessive ongoing blood loss and patients were cooperative and hemodynamically stable. Perioperative serum glucose was regulated according to local sliding scale protocols.

The primary study end point of major adverse events was a composite of death, myocardial infarction (MI), stroke, renal failure, or respiratory failure, occurring within 30 days of randomization. Perioperative MI was defined as the presence of new Q waves or a new left bundle branch block on the electrocardiogram, combined with a biomarker (creatine kinase–MB or troponin) elevation of more than 5 times the upper reference limit. Data from routine cardiac biomarker surveillance were used to detect possible perioperative MI. The specific type of biomarker used was dictated by the local protocol in each center, rather than by the study protocol. Postdischarge MI was defined according to the criteria of the Universal Definition of Myocardial Infarction.¹⁵ Stroke was defined as a neurologic deficit lasting more than 24 hours, with increased invalidity (increase on Rankin scale¹⁶ of ≥1 point) and signs of a new ischemic cerebral infarction on computed tomography or magnetic resonance imaging scan. Renal failure in patients not previously receiving dialysis was defined according to the RIFLE criteria as an increase in postoperative serum creatinine of at least 3 times the preoperative value, or a serum creatinine level of more than 4 mg/dL (to convert to micromoles per liter, multiply by 88.4) associated with an acute increase of serum creatinine of at least 0.5 mg/dL.¹⁷ Respiratory failure was defined as postoperative mechanical ventilation or reinstitution of mechanical respiratory support via an orotracheal tube or tracheostomy for an uninterrupted period of at least 48 hours.

An exploratory analysis of prospectively defined secondary outcomes included each separate component of the primary end point (ie, death, MI, stroke, renal failure, or respiratory failure, within the first 30 days); postoperative infections; postoperative atrial fibrillation; highest serum glucose concentration in the ICU; highest body temperature in the ICU; postoperative delirium (defined as the postoperative indication for treatment with neuroleptic drugs); time to weaning from postoperative mechanical ventilation; and time to discharge from the ICU and from the hospital.

An independent, blinded critical event adjudication committee reviewed all cases of death, possible MI, and possible stroke. Cases of possible MI or stroke were either confirmed or revoked according to the study definitions of these events.

We hypothesized that dexamethasone administration would reduce the incidence of the primary study outcome. Based on the Society of Thoracic Surgeons database,¹⁸ the incidence of the primary outcome in the placebo group was estimated to be 6%. To detect an absolute difference of 2% (from 6% to 4%) with a power of 80% at a 2-sided .05 significance level, 1962 patients would be required in each study group. To compensate for possible losses to follow-up, we planned to include 2250 patients per study group.

During the study, 3 preplanned interim analyses on the primary study end point were performed when 1000, 2000, and 3250 patients, respectively, had been enrolled. Interim analyses were performed by the independent data and safety monitoring board, which consisted of an epidemiologist, a cardiac surgeon, and a cardiac anesthesiologist not involved in the study. These blinded analyses were adjusted according to an O’Brien and Fleming type I error spending function,¹⁹ using an overall .05 significance level. As a result of these interim analyses, the threshold for significance of the primary study end point in the final analysis was .044.

Patient follow-up for secondary outcomes was until 1 year from randomization by study protocol. Herein, we report the primary study end point together with exploratory analyses of other outcomes in the first 30 days after randomization. Analyses were conducted according to randomization (intention-to-treat analyses). Baseline characteristics in the 2 study groups were evaluated using frequency distributions.

We also performed preplanned subgroup analyses for the primary outcome and its separate components, which included 4 age groups (<65, 65-74, 75-79, and ≥80 years), sex, diabetes, chronic obstructive pulmonary disease, higher (≥5) vs low (≤4) EuroScore preoperative risk estimate²⁰ (cutoff value based on the median EuroScore of the study population), and prolonged CPB duration (defined as >150 minutes).

For the comparison of the proportions of patients with primary and secondary outcomes, we used the χ² test. Absolute risk reduction or relative risk (RR) with 95% CIs was calculated for each dichotomous outcome measure. Logistic regression was used for assessing heterogeneity in the subgroup analyses, with a .10 threshold for significance. For comparison of mean and median values of the continuous secondary outcome measures, we used Student t test or Mann-Whitney U test, as appropriate. IBM SPSS version 19 (SPSS Inc) was used for all analyses.

An estimated 25 085 patients scheduled to undergo elective or urgent cardiac surgery were screened, of whom 21 581 were eligible for inclusion. Of the 4827 patients who provided written informed consent, 4494 eventually underwent randomization (Figure 1). Two patients who were unintentionally randomized without having provided informed consent were excluded from the analysis. Of the 4494 randomized patients, 2239 (49.8%) were randomized to dexamethasone treatment and 2255 (50.2%) to placebo treatment. Two patients (0.1%) in the dexamethasone group and 5 patients (0.2%) in the placebo group withdrew consent; therefore, their primary outcome could not be assessed. We were unable to obtain 30-day outcome information in 2 patients (0.1%) in the dexamethasone group and 3 patients (0.1%) in the placebo group after hospital discharge. Thus, the analyzed population consisted of 4482 patients (99.7%). Patients in the study groups were similar with respect to baseline demographic, clinical, and surgical characteristics (Table 1).

In total, 348 patients (7.8%) reached the composite primary study end point of death, MI, stroke, renal failure, or respiratory failure, within 30 days after randomization (Table 2). The primary study end point occurred in 157 of the 2235 patients (7.0%) randomized to dexamethasone and in 191 of the 2247 patients (8.5%) randomized to placebo (RR, 0.83; 95% CI, 0.67-1.01; absolute risk reduction, −1.5%; 95% CI, −3.0% to 0.1%; P = .07).

The rate of death, MI, stroke, and renal failure was similar in both groups. In the dexamethasone group, 67 patients (3.0%) experienced respiratory failure compared with 97 patients (4.3%) in the placebo group (RR, 0.69; 95% CI, 0.51-0.94; P = .02) (Table 2). The RR for a composite end point without the respiratory failure component, consisting of only mortality, MI, stroke, and renal failure, was 0.84 (95% CI, 0.66-1.08; P = .18).

The median (interquartile range [IQR]) time to weaning from mechanical ventilation was 7.0 (4.7-10.0) hours (mean, 11.0 hours) in the dexamethasone group and 7.0 (5.0-11.0) hours (mean, 14.3 hours) in the placebo group (P < .001) (Table 3). The median (IQR) time to discharge from the ICU was 22.0 (19.0-24.0) hours (mean, 34.2 hours) in the dexamethasone group and 22.0 (19.0-25.0) hours (mean, 43.6 hours) in the placebo group (P < .001). The low P values for both comparisons despite similar median values are the result of a higher proportion of patients requiring prolonged ventilation times and prolonged ICU stay in the placebo group. For example, the proportion of patients requiring more than 24 hours of postoperative mechanical ventilation was 3.4% in the dexamethasone group compared with 4.9% in the placebo group. Similarly, the proportion of patients requiring more than 48 hours of ICU stay was lower in the dexamethasone group than in the placebo group (10.2% vs 14.0%, respectively). The median (IQR) time to discharge from the hospital in the dexamethasone group was 8 (7-13) days (mean, 11.3 days) vs 9 (7-13) days (mean, 11.7 days) in the placebo group (P = .009).

The risk of developing a postoperative infection was lower in the dexamethasone group than in the placebo group (9.5% vs 14.8%, respectively; RR, 0.64; 95% CI, 0.54-0.75; P < .001) (Table 3). This protective effect was primarily related to a decreased incidence of pneumonia in the dexamethasone group (6.0% vs 10.6% in the placebo group; RR, 0.56; 95% CI, 0.46-0.69; P < .001). In contrast, the mean highest serum glucose concentration was higher in the dexamethasone group and the incidence of postoperative fever was higher in the placebo group (Table 3).

The preplanned subgroup analyses suggested an age-dependent effect of dexamethasone on the primary study end point (P for heterogeneity = .08) (Figure 2). In patients younger than 65 years, dexamethasone was associated with lower likelihood for the primary end point (RR, 0.65; 95% CI, 0.44-0.96; P = .03), whereas in patients aged 80 years or older, the RR was 1.69 (95% CI, 0.92-3.10; P = .09). There was no differential treatment effect in the subgroup analyses on sex, diabetes, chronic obstructive pulmonary disease, EuroScore, or prolonged CPB duration.

The bidirectional effect of dexamethasone across the 4 age groups appeared to be predominantly caused by the mortality component of the primary study end point, which showed significant heterogeneity (P = .05). In patients younger than 65 years, the RR for mortality was 0.42 (95% CI, 0.13-1.34; P = .13), but it gradually increased with age to 3.87 (95% CI, 1.10-13.6; P = .02) in patients aged 80 years or older.

Our randomized study of 4494 patients undergoing cardiac surgery failed to show a statistically significant benefit of intraoperative administration of dexamethasone on the incidence of the composite primary study end point of major adverse events (P = .07). In an exploratory analysis of secondary end points, a reduced incidence of respiratory failure was found, which was accompanied by an overall reduced time to weaning from mechanical ventilation, a lower risk of pneumonia, and a reduction in ICU and hospital stay.

The DECS trial is the first large randomized controlled trial to our knowledge on the controversial topic of routine corticosteroid use during cardiac surgery in adults. The numerous small randomized studies published in the last decades had conflicting results¹³ and provided only limited guidance for selection of components for the composite primary end point of our study. No significant benefit from dexamethasone treatment was observed on the composite primary end point, which was largely cardiovascular. However, further exploration of the study data suggested a consistent pattern of improved pulmonary condition, manifested as a lower risk of postoperative respiratory failure, shorter times to weaning from the ventilator, and reduced risk of pneumonia during postoperative hospitalization in the dexamethasone group. This improved respiratory condition was accompanied by earlier discharge from both the ICU and the hospital. It might be argued that the beneficial effect of dexamethasone on these multiple secondary outcomes was only a coincident finding. However, also at a more conservative cutoff value for statistical significance of .0025—to correct for testing a total of 19 secondary outcomes using a Bonferroni approach—most of the effects found remained significant. These effects may actually be interrelated and explained by attenuation of the perioperative systemic inflammatory response syndrome, which is the pathophysiologic goal of the single high-dose corticosteroid treatment.^11- 14 Thus, although the effect of dexamethasone on the primary end point was negative, there is the possibility that a clinically significant effect was missed. Therefore, a new prospective study focusing on pulmonary outcomes seems a logical next step to further explore the secondary findings of our trial. Such a study should also consider patient selection for the therapy as we found a larger beneficial effect in younger patients and no apparent benefit in those aged 80 years or older.

Our study failed to confirm that corticosteroids reduce the incidence of postoperative atrial fibrillation, as demonstrated in a previous study.²¹ The 241 patients in that particular trial received moderate-dose hydrocortisone 3 days postoperatively, instead of the single intraoperative high dose of dexamethasone in our study.

The reduced risk of postoperative infections, in particular pulmonary infections, in the dexamethasone group was unexpected, and contrary to existing knowledge that corticosteroids increase the risk of infections.⁷ However, this adverse effect is mainly related to chronic corticosteroid use, rather than to a single prophylactic pulse dose in circumstances wherein the activation of the immune system could be detrimental.^11- 14 The reduced infection risk persisted when patients with respiratory failure were excluded from the analysis, suggesting that the observation is not the result of shorter exposure to mechanical ventilation.

A pooled analysis of this study with the results of our previous meta-analysis of 56 studies on prophylactic corticosteroids in cardiac surgery, which was recently published,¹³ showed no effect of corticosteroids for mortality or cardiac complications. However, corticosteroid treatment reduced respiratory complications (Peto odds ratio, 0.59; 95% CI, 0.49-0.71; P < .001) (eMethods and eFigure).

A limitation of our trial is that we studied a high-dose dexamethasone regimen, which is often used during cardiac surgery in several European countries. In North America, however, methylprednisolone is usually preferred when corticosteroids are administered during cardiac surgery. The effect of high-dose methylprednisolone for cardiac surgery is being studied in an ongoing large study (Steroids In cardiac Surgery [SIRS] trial, NCT00427388). A strength of our study is that blinding for treatment was well maintained during the perioperative period. The small differences in serum glucose and postoperative body temperature are unlikely to have caused awareness of randomization. It is thus unlikely that such awareness, if any, could have influenced clinical management to a degree that could have produced the present effects on pulmonary outcome and duration of ICU and hospital stay.

In conclusion, in our trial of adults undergoing cardiac surgery, the use of intraoperative dexamethasone did not reduce the 30-day incidence of major adverse events compared with placebo.