Eur J Anaesthesiol 2014; 31:685–694
ORIGINAL ARTICLE
The effect of perioperative administration of glucocorticoids on pulmonary complications after transthoracic oesophagectomy A systematic review and meta-analysis Teus J. Weijs, Jan M. Dieleman, Jelle P. Ruurda, A. Christiaan Kroese, Hans J.T.A. Knape and Richard van Hillegersberg BACKGROUND Severe pulmonary complications occur frequently following transthoracic oesophagectomy. An exaggerated immunological response is probably a main driving factor, and this might be prevented by perioperative administration of a glucocorticoid. OBJECTIVE To determine the clinical benefits and harms of perioperative glucocorticoid during transthoracic oesophagectomy, using pulmonary complications as the primary outcome. Mortality, anastomotic leakage rate and infection were secondary outcomes. METHODS A systematic review of interventional trials with a meta-analysis of randomised controlled trials (RCTs). RESULTS The search retrieved seven RCTs and four interventional nonrandomised studies. In total, 367 patients received perioperative glucocorticoid and 415 patients did not. A meta-analysis of the RCTs showed no significant effect
of glucocorticoid. For pulmonary complications, the pooled risk ratio was 0.69 [95% confidence interval (CI) 0.26 to 1.79], for anastomotic leakage 0.61 (95% CI 0.23 to 1.61) and for infections 1.09 (95% CI 0.41 to 2.93). A subgroup analysis of RCTs that used weight-dependent dosing within 30 min preoperatively showed a pooled risk ratio of 0.28 (95% CI 0.10 to 0.77) for pulmonary complications compared with placebo. CONCLUSION In this meta-analysis, perioperative administration of glucocorticoid did not affect the risk of pulmonary complications after transthoracic oesophagectomy, nor did it cause adverse effects. A subgroup analysis showed that a weight-dependent dose of methylprednisolone 10 to 30 mg kg1 within 30 min preoperatively might be the most promising dosing regimen for further research. Published online 16 May 2014
Background Oesophageal cancer is the eighth most common cancer and the sixth most common cause of cancer-related death globally.1 A transthoracic surgical approach offers the best chance for long-term survival. It allows meticulous mediastinal dissection resulting in extended lymph node retrieval and radical resection.2 However, this approach is accompanied by severe pulmonary complications, usually pneumonia and acute lung injury, which lead to increased perioperative mortality.3 The intense inflammatory response induced by surgery, in combination with local inflammation due to one-lung ventilation, is thought to play an
important role in the development of these complications.4–6 Perioperative glucocorticoid administration reduces the postoperative inflammatory response after transthoracic oesophagectomy.7–13 In addition, it has been suggested that administration of a glucocorticoid can reduce the incidenceofpulmonarycomplications.7 However,thepotential of glucocorticoid to increase the risks of infection, tumour recurrence and wound healing abnormalities has limited widespread use during transthoracic oesophagectomy.9,14
From the Department of Surgery (TJW, JPR, RVH), and Department of Anaesthesiology, Intensive Care and Emergency Medicine, University Medical Center Utrecht, Utrecht, The Netherlands (JMD, ACK, HJTAK) Correspondence to Jelle P. Ruurda, Department of Surgery, University Medical Center Utrecht, Heidelberglaan 100, 3508 GA Utrecht, The Netherlands Tel: +088 7558074; e-mail: [email protected] 0265-0215 ß 2014 Copyright European Society of Anaesthesiology
DOI:10.1097/EJA.0000000000000093
Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
686 Weijs et al.
Two previous meta-analyses have investigated the outcomes of perioperative glucocorticoid administration for oesophagectomy.15,16 However, the most recent metaanalysis included several studies with a high risk of bias. In the other meta-analysis three randomised controlled trials were not included. This makes the results difficult to interpret. Consequently, we undertook an updated systematic review and meta-analysis of RCTs on perioperative glucocorticoid administration for transthoracic oesophagectomy. The primary objective of this meta-analysis was to determine the effect of perioperative glucocorticoid administration for transthoracic oesophagectomy on pulmonary complications. Secondary objectives were to determine the risks for adverse outcomes: mortality, anastomotic leakage and infections.
Materials and methods Eligibility criteria
All articles that investigated perioperative administration of glucocorticoids for transthoracic oesophagectomy were sought. In order to present a systematic review of available evidence, we considered all randomised and nonrandomised comparative interventional studies for inclusion. Studies not reporting on the primary outcome were excluded. Information sources and search strategy
We conducted a systematic search of Medline (via Pubmed), Embase and Cochrane databases. The search strategy combined all synonyms regarding the intervention, ‘glucocorticoids’ and the domain, ‘oesophagectomy’ (e-supplement 1, http://links.lww.com/EJA/A48). No limitations were applied. The sensitivity of the search strategy was improved by screening all references of included articles for relevant publications that were not retrieved in the initial search. Also, all citing articles and related articles were screened using Web of Science version 5.7. Authors were contacted by e-mail when articles were not available in full text. When additional eligible articles were identified, the search strategy was evaluated and improved if possible. Study selection
Two independent investigators (T.W., S.D.) screened all articles obtained from the database search on title and abstract. Studies were excluded when their population consisted of nonhumans or patients not undergoing oesophagectomy, and if the intervention investigated was not glucocorticoid administration. The remaining articles were fully reviewed. We included all articles fulfilling the eligibility criteria for the systematic review, but only RCTs were included for meta-analysis. No language restrictions were applied to the selection process.
Risk of bias in individual studies
Two investigators (T.W., S.D.) independently assessed the risk of bias at the study level in the included studies. For the studies published in Japanese only, the necessary data were extracted by a Japanese medical student (N.P.) on forms designed by the lead investigator (T.W.). The Cochrane collaboration’s tool for assessing risk of bias was used for RCTs and the Ottawa quality assessment scale for nonrandomised studies.17,18 In addition, the risk of bias on outcome level was assessed by extracting the outcome definitions of individual studies. Disagreements were resolved by discussion. Data collection process
All data were extracted by the leading reviewer (T.W.) and checked for errors by the second reviewer (S.D.). Only published data were extracted; however, authors were contacted by e-mail when verification of data was necessary. Data items
We extracted data on study location and period to prevent inclusion of duplicate publications. Data on study design that were considered important were type, dosing regimen and timing of glucocorticoid administration; oesophagectomy approach and type of intestinal tract reconstruction; use of postoperative epidural analgesia; location of extubation; outcome definitions and duration of follow-up. Finally, the number of events in the intervention and placebo group was extracted for primary and secondary outcomes. The outcome ‘pulmonary complications’ was defined as respiratory failure and/or pneumonia because these are most probably linked to the inflammatory response. For ‘mortality’, primarily in-hospital mortality rates were sought, and secondarily, 30-day mortality rates. For the outcome ‘infections’, all events due to pneumonia were excluded because these were already included in the primary outcome. Data analysis
Absolute risks and risk ratios were calculated and compared using the Mantel–Haenszel method for dichotomous outcomes. A P value of less than 0.05 was considered significant. For individual studies with no events in one or more groups, a continuity correction of 0.5 was added to each cell for each effect measure.19 Publication bias was evaluated for the primary outcome using funnel plot analysis of all RCTs. A meta-analysis of all RCTs was performed. Statistical heterogeneity was investigated using the x2 test, wherein a P value of less than 0.1 was considered to indicate significant heterogeneity. Pooled weighted risk ratios with 95% confidence intervals (CIs) were calculated for all investigated outcomes except mortality because reported mortality rates were too low for reliable calculations. A sensitivity analysis was used to determine the
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
Glucocorticoids for oesophagectomy
impact of single studies on the pooled effect for the primary outcome. Some studies used a fixed-dose, and other a weightdependent dose, without presenting the mean weight of their patients. Because in general, a weight-dependent dosing regimen (10 or 30 mg kg1) would result in higher dosage than a fixed dose (125, 250 or 500 mg), a weightdependent dosing regimen was considered as a surrogate for ‘high dose’, and a fixed dose as a surrogate for ‘low dose’. Studies were grouped by this variable and pooled risk ratios were calculated. No other subgroup analyses were planned or performed. In all analyses without significant statistical and methodological heterogeneity, a fixed effects model was used; in all other analyses, a random effects model was used. All analyses were performed with Comprehensive Metaanalysis software, version 2.20
Results Search results
The results of the search strategy are shown in Fig. 1. The final search was performed on 17 September 2013. From the 30 articles screened in full text, 16 articles investigated irrelevant populations, outcomes or interventions, and two were nonsystematic reviews. Screening references from a meta-analysis retrieved one additional article.13 Finally, 11 articles were included: seven RCTs and four nonrandomised comparative studies.7–9,11–14,21–24
687
Characteristics of included studies
The baseline characteristics of the included studies are summarised in Table 1. This review includes 367 patients receiving perioperative glucocorticoid and 415 patients who did not. In all studies, the glucocorticoid used was methylprednisolone. All seven RCTs were conducted more than a decade ago in Japan.7–9,11–13,21 Methylprednisolone was administered as a fixed dose of 250 or 500 mg, or as a weightdependent dose of 10 or 30 mg kg1. The dosage most frequently used was 10 mg kg1.7–9,12 Timing of methylprednisolone administration ranged from 3 h preoperatively to just before induction of anaesthesia. In all RCTs, a thoracotomy was performed and the stomach was the preferred organ for reconstruction. Data on postoperative epidural analgesia were missing in most RCTs. All RCTs reported on pulmonary complications, but only two provided clear definitions for pulmonary complications (e-supplement 2, http://links.lww.com/EJA/ A48).7,13 Mortality rates were reported in three RCTs.7,9,21 None of the five RCTs reporting anastomotic leakage rates provided clear-cut outcome definitions, and for infection, only one of the five did so.7,9,11,18,21 The length of follow-up was only 3 days in one RCT; in all other RCTs, follow-up was for between 7 days and 5 years. The four nonrandomised studies14,22–24 were conducted more recently, three in Japan, and one in Korea (Table 1). The dosage of methylprednisolone ranged from 125 to
Fig. 1
. Pubmed N = 175
Embase N = 486
Cochrane N = 15
N = 676 Excluding duplicates: N = 183 N = 493 Exclusion based on the title and abstract: N = 462 N = 31
Screening references, citing and related publications: 1 additional articles retrieved
N = 10
Exclusion based on FT: Different outcomes: 4 Different population: 7 Different intervention: 6 Review: 2 Meta-analysis: 2
Included: 11 primary studies
Flowchart showing results of study selection process. FT, full text.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
Location
2003 to 2004 vs. 2002 1995 to 1996 vs. 1997 to 1998 107 (57/50)
36 (21/15)
184 (100/84)
234 (83/151)
17 (8/9)
30 (15/15)
33 (14/19)
18 (9/9)
66 (33/33)
17 (7/10)
40 (20/20)
N (I/C)
Physiological saline Physiological saline ? Physiological saline Physiological saline ?
10 mg kg1
10 mg kg1
250 mg
n.a.
n.a.
n.a.
2 125 mg
250 mg
n.a.
125 mg
250 mg
30 mg kg1
250 mg
10 mg kg1
Physiological saline
Placebo
500 mg
MP Dose
After construction of anastomosis At start of thoracotomy and at closure 60 min preoperatively 60 min preoperatively and 125 mg on POD 1 and 2
Before anaesthesia 30 min preoperatively 30 min preoperatively At induction of anaesthesia Before anaesthesia 120 to 180 min preoperatively
120 min preoperatively
Time of administration
Thoracotomy
Thoracotomy
Thoracotomy or VATS
Thoracotomy
Thoracotomy
Thoracotomy
Thoracotomy
Thoracotomy
Thoracotomy
Thoracotomy
Thoracotomy
Approach
2 or 3-field
3-field
2 or 3-field
2 or 3-field
2 or 3-field
3-field
3-field
2 or 3-field
3-field
3-field
2 or 3-field
Lymph node dissection
?
?
Gastric tube or colon Gastric tube or colon
Stomach or small intestine
Gastric tube
Stomach
Stomach
Stomach
Stomach, small intestine or colon Stomach
Reconstruction
?
?
Cervical or intrathoracic Cervical
Cervical or intrathoracic Cervical or intrathoracic
?
Intrathoracic
Cervical or intrathoracic Cervical
?
Level of anastomosis
?
?
Yes
?
?
Yes
?
?
?
?
?
Epidural analgesia
ICU
ICU
ICU
OR
ICU
ICU
?
ICU
ICU
ICU
?
Location of extubation
?, unknown; C, control group; I, intervention group; i.v., intravenously; MP, methylprednisolone; N, total number of included patients; n.a., not applicable; OR, operation room; POD, postoperative day; VATS, video-assisted thoracoscopy.
Japan
Japan
?
Japan Sayama et al.13 Nonrandomised studies Park et al.24 Korea
Tsukada et al.22 Shimada et al.14
?
Japan
Japan
?
Japan
Nakamura et al.23
1996 to 1997
Japan
Takemura et al.11 Matsutani et al.12 Takeda et al.21
2007 to 2008 vs. 2005 to 2006 2001 to 2005 vs. 1991 to 2000
1996 to 1999
Japan
Sato et al.7
?
Japan
Takeda et al.8
1997 to 1999
Period
Characteristics of included interventional studies
Randomised studies Japan Yano et al.9
Study
Table 1
688 Weijs et al.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
Glucocorticoids for oesophagectomy
250 mg, and the timing of administration ranged from 1 h preoperatively to directly after construction of the anastomosis. Additional doses of methylprednisolone were administered after surgery in two nonrandomised studies.14,23 In general, the oesophagectomy was conducted through a thoracotomy, but in one study,23 a videoassisted thoracoscopic approach was also used. All four nonrandomised studies provided clear definitions for pulmonary complications, but all lacked definitions for anastomotic leakage and infections (e-supplement 2, http://links.lww.com/EJA/A48).14,22–24 The duration of follow-up was 30 days in two of the nonrandomised studies, and the entire admission time in the other two studies. Risk of bias in individual studies
The results of the critical appraisal on study level are summarised in Table 2. Among the seven RCTs, five did not describe whether they concealed allocation and blinded participants, personnel and outcome assessment.8,11–13,21 Thus, the risk for bias was high in these studies.25 The study by Yano et al.9 was deemed at risk of bias because the authors did not describe allocation concealment. The only study considered to be at low risk of bias was conducted by Sato et al.7 Among the nonrandomised studies, the study of Shimada et al.14 did not investigate nor correct for confounding factors. All nonrandomised studies14,22–24 compared patients from different time periods, and therefore, the risk for bias was considered to be very high. Results of individual studies
The individual study results are shown in Figs 2–5. Table 2
689
Seven RCTs, which included 221 patients, reported on the primary outcome, pulmonary complications (Fig. 2). The risk ratios for pulmonary complications after perioperative administration of methylprednisolone compared with placebo ranged from 0.09 (95% CI 0.01 to 1.51) to 3.00 (95% CI 0.14 to 65.16).7–9,11–13,21 The two large and high-quality RCTs obtained somewhat contradictory results. In the study by Sato et al.,7 a risk ratio of 0.30 (95% CI 0.09 to 0.99) favouring methylprednisolone was found. In contrast, Yano et al.9 found a nonsignificant increase in risk in the methylprednisolone group to 2.0 (95% CI 0.58 to 6.91). Regarding the secondary outcomes, mortality rates were reported in three RCTs, which included 136 patients (Fig. 3). Takeda et al.21 reported the only two cases of postoperative mortality, both in the placebo group. In five RCTs that included 171 patients, anastomotic leakage was reported (Fig. 4). The risk ratios for anastomotic leakage ranged from 0.20 (95% CI 0.01 to 3.85) to 3.00 (95% CI 0.14 to 65.16).7,9,11,13,21 Infection rates were reported in the same five RCTs (Fig. 5). The risk ratios for infections after perioperative administration of glucocorticoid compared with placebo ranged from 0.16 (95% CI 0.01 to 2.67) to 2.33 (95% CI 0.70 to 7.76). All four nonrandomised studies,14,22–24 which included 561 patients, reported on the primary outcome, pulmonary complications (Fig. 2). The risk ratios for pulmonary complications ranged from 0.24 (95% CI 0.12 to 0.50) to 0.70 (95% CI 0.41 to 1.20). Regarding the secondary outcomes, all reported on mortality (Fig. 3). The risk ratios for postoperative mortality ranged from 0.42 (95% CI 0.08 to 2.24) to 0.88 (95% CI 0.06 to 13.66). On anastomotic leakage, three nonrandomised studies,14,22,24 which included 377 patients, found that the risk ratios for
Critical appraisal of studies
Randomised studies Item Randomisation? Allocation concealment? Blinding of participants and personnel? Blinding of outcome assessment? Outcome data complete? No selective outcome reporting? Risk of bias Nonrandomised studies Item Representative exposed cohort? Nonexposed selected from same population? Ascertainment of exposure reliable? Outcome of interest not present from start? Adequately dealt with confounding factors? Reliable outcome assessment? Adequately dealt with loss to follow-up? No selective outcome reporting? Risk of bias
Sayama et al.13
Takeda et al.21
Matsutani et al.12
Takemura et al.11
Sato et al.7
Takeda et al.8
Yano et al.9
þ ? ? ? þ ? High
þ ? ? ? þ þ High
þ ? ? ? þ High
þ ? ? ? þ ? High
þ þ þ ? þ þ Low
þ ? ? ? þ High
þ ? þ ? þ þ High
Shimada et al.14
Tsukada et al.22
Nakamura et al.23
Park et al.24
þ þ þ ? þ High
þ þ þ þ ? þ High
þ þ þ þ ? þ High
þ þ þ þ ? þ High
þ, yes; , no; ?, unclear.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
690 Weijs et al.
Fig. 2
Study
Year
Randomised studies Yano9 2005 Takeda8a 2003 2002 Sato7 1999 Takemura11a Matsutani12a 1998 1997 Takeda21a Sayama13 1994 Pooled result
Control-group
MP-group
Risk ratio
Events
No events
Events
No events
6 0.5 3 1.5 0.5 0.5 1
14 7.5 30 8.5 14.5 15.5 7
3 0.5 10 0.5 2.5 5.5 1
17 10.5 23 9.5 17.5 10.5 8
Non-randomised studies Park24 2012 2008 Nakmura23 2006 Tsukada22 2000 Shimada14
Forest plot
2.00 (0.58 – 6.91) 1.38 (0.03 – 62.25) 0.30 (0.09 – 0.99)b 3.00 (0.14 – 65.16) 0.27 (0.01 – 5.15) 0.09 (0.01 – 1.51) 1.13 (0.08 – 15.19) 0.69 (0.26 – 1.79)
2 8 3 16
81 92
16 28 4 20
18 41
135 56 11 30
0.23 (0.05 – 0.96)b,c 0.24 (0.12 – 0.50)b,c 0.54 (0.14 – 2.05) 0.70 (0.14 – 1.20) 0.01 0.1 Favours methylprednisolone
1
10 Favours control
100
Table and forest plot summarising study results concerning pulmonary complications. Numbers between brackets are lower and upper limits of 95% confidence intervals. MP, methylprednisolone. aContinuity correction applied. bSignificant result. cSignificant result in multivariate analysis.
postoperative anastomotic leakage ranged from 0.22 (95% CI 0.05 to 0.99) to 1.36 (95% CI 0.49 to 3.80) (Fig. 4). Infection rates were reported in one nonrandomised study,22 which included 36 patients (Fig. 5); the risk ratio for infections was 1.07 (95% CI 0.20 to 5.65).
Secondary outcomes
The pooled risk ratio for anastomotic leakage in five RCTs7,9,11,13,21 was 0.61 (95% CI 0.23 to 1.61; P ¼ 0.32) (Fig. 4). Pooling the results for infections from the same five RCTs resulted in a risk ratio of 1.09 (95% CI 0.41 to 2.93; P ¼ 0.86) (Fig. 5).
Meta-analysis
The results of the meta-analysis of RCTs are shown in Figs 2–5. Primary outcome
The results of the seven RCTs that reported on pulmonary complications7–9,11–13,21 were pooled using a random effects model (Fig. 2). This resulted in a pooled risk ratio of 0.69 (95% CI 0.26 to 1.79; P ¼ 0.44) for pulmonary complications when glucocorticoid was administered perioperatively compared with placebo.
Ancillary analyses
Statistically, there was no significant interstudy heterogeneity. However, the studies were heterogeneous in their study methods; therefore, a random effects model was used for pooling of the results. No indication for publication bias regarding the primary outcome was found in the funnel plot (e-supplement 3, http://links. lww.com/EJA/A48). The studies of Yano et al.9 and Sato et al.7 had the greatest impact on the pooled result, in opposite directions
Fig. 3
Study
Year
Randomised studies 2005 Yano9 2002 Sato7 21 Takeda 1997 Non-randomised studies Park24 2012 Nakmura23 2008 2006 Tsukada22 Shimada14 2000
MP-group
Control group
Risk ratio
Events
No events
Events
No events
0 0 0
20 33 15
0 0 2
20 33 15
n.a. n.a. n.a.
2 2 0 1
83 100 20 56
8 4 0 1
151 84 15 49
0.45 (0.10 – 2.09) 0.42 (0.08 – 2.24) n.a. 0.88 (0.06 – 13.66)
Forest plot
0.01
0.1
Favours methylprednisolone
1
10
100
Favours placebo
Table and forest plot summarising study results concerning mortality. Numbers between brackets are lower and upper limits of 95% confidence intervals. MP, methylprednisolone; n.a., not applicable.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
Glucocorticoids for oesophagectomy
691
Fig. 4
Study
MP-group
Year
Randomised studies 2005 Yano9 Sato7 2002 11a Takemura 1999 21a Takeda 1997 13a 1994 Sayama Pooled result Non-randomised studies Park24 2012 Tsukada22 2006 2000 Shimada14
Control-group
Risk ratio
Events
No events
Events
No events
3 1 1.5 0.5 0.5
17 32 9.5 15.5 8.5
4 2 0.5 2.5 2.5
16 31 9.5 15.5 9.5
0.75 (0.19 – 2.93) 0.50 (0.05 – 5.25) 3.00 (0.14 – 65.16) 0.20 (0.01 – 3.85) 0.22 (0.01 – 4.04) 0.61 (0.23 – 1.61)
6 2 2
77 19 55
8 2 8
143 13 42
0.36 (0.49 – 3.80) 0.71 (0.11 – 4.52) 0.22 (0.05 – 0.99)b
Forest plot
0.01 0.1 1 Favours methylprednisolone
10 Favours placebo
100
Table and forest plot summarising study results concerning anastomotic leakage. Numbers between brackets are lower and upper limits of 95% confidence intervals. MP, methylprednisolone. aContinuity correction applied. bsignificant result.
(e-supplement 4, http://links.lww.com/EJA/A48). The pooled risk ratio excluding the results of Yano et al.9 was 0.45 (95% CI 0.15 to 1.40); the pooled risk ratio excluding the results of Sato et al.7 was 0.96 (95% CI, 0.31 to 3.00). Subgroup analysis
The impact of using a weight-dependent dosage of glucocorticoid rather than a fixed dose on the risk for pulmonary complications is shown in Fig. 6. A fixed dose was used in three studies,9,11,13 which included 75 patients. Pooling their results resulted in a risk ratio for pulmonary complications after a fixed dose of glucocorticoid of 1.91 (95% CI 0.67 to 5.46; P ¼ 0.23) compared with placebo. A weight-dependent dosage was used in four studies,7,8,12,21 which included 146 patients. The pooled risk ratio for pulmonary complications was 0.28 (95% CI 0.10 to 0.77; P ¼ 0.01) after a weight-dependent dosage of glucocorticoid compared with placebo. A random effects model was used for the subgroup analysis.
Discussion This systematic review and meta-analysis summarises the evidence for the benefits and harms of perioperative glucocorticoid administration for transthoracic oesophagectomy. In total, 782 patients were included in 11 studies, which were heterogeneous in both design and results. In a pooled analysis of RCTs, we were unable to demonstrate either clinical benefits or adverse effects of perioperative administration of glucocorticoid. A subgroup analysis of RCTs using a weight-dependant dose of 10 to 30 mg kg1 of glucocorticoid within 30 min preoperatively suggests that this regimen might be effective in reducing pulmonary complications compared with a fixed dose of 250 or 500 mg given 120 to 180 min preoperatively. Recently, Engelman and Maeyens15 published a metaanalysis on this subject. The authors found significantly decreased risks of pulmonary complications and infections in their pooled analyses, and their analysis showed a trend towards decreased risks of anastomotic leakage and
Fig. 5
Study
Year
Randomised studies 2005 Yano9 2002 Sato7 11 1999 Takemura 21a 1997 Takeda 13a 1994 Sayama Pooled result Non-randomised studies Tsukada22 2006
MP-group Events
Control-group
No events Events
Risk ratio
Forest plot
No events
7 3 1 0.5 0.5
13 30 8 15.5 8.5
3 2 1 2.5 3.5
17 31 8 13.5 6.5
2.33 (0.70 – 7.76) 1.50 (0.27 – 8.40) 1.00 (0.07 – 13.84) 0.20 (0.01 – 3.85) 0.16 (0.01 – 2.67) 1.09 (0.41 – 2.93)
3
18
2
13
1.07 (0.20 – 5.65) 0.01 0.1 1 Favours methylprednisolone
10 Favours placebo
100
Table and forest plot summarising study results concerning infections. Numbers between brackets are lower and upper limits of 95% confidence intervals. MP, methylprednisolone. aContinuity correction applied.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
692 Weijs et al.
Fig. 6 Study
Year
MP-group Events
Fixed dose Yano9 2005 1999 Takemura11a Sayama13 1994 Pooled result
Control-group
No events Events
Risk ratio
Forest plot
No events
5 1.5 1
14 0.5 7
3 0.5 1
17 9.5 8
2.00 (0.58 – 6.91) 3.00 (0.14 – 65.16) 1.13 (0.06 – 15.19) 1.91 (0.67 – 5.46)
0.5 3 0.5 0.5
7.5 30 14.4 15.5
0.5 10 2.5 5.6
10.5 23 17.5 10.5
1.38 (0.03 – 62.25) 0.30 (0.09 – 0.99)b 0.27 (0.01 – 5.15) 0.09 (0.01 – 1.51)b 0.28 (0.10 – 0.77)
Weight dependent dose Takeda8a Sato7 Matsutsni10a Takeda21a Pooled result
2003 2002 1998 1997
0.01 0.1 Favours methylprednisolone
1
10 Favours control
100
Table and forest plot in which the effect of a fixed dose of methylprednisolone is compared with a weight-dependent dose. Numbers between brackets are lower and upper limits of 95% confidence intervals. MP, methylprednisolone. aContinuity correction applied. bSignificant result.
early mortality. However, their results should be interpreted with caution. In their meta-analysis, two nonrandomised studies,14,22 highly at a risk for bias, were included. In addition, they did not identify one other relevant study.11 Furthermore, studies with zero events in the study group were not included in this metaanalysis. In these cases, it has been shown that including such studies by performing continuity correction (of 0.5 events to each arm) for the meta-analysis leads to less bias.19 Finally, the results of the RCT by Yano et al.9 were presented, and probably analysed, the wrong way round, making their results difficult to interpret. Consequently, we decided to conduct the present review and metaanalysis, avoiding these pitfalls. Our new approach, excluding studies highly at risk of bias, may explain the different results obtained in this meta-analysis, compared with the results obtained by Engelman and Maeyens.15 The strengths of our review are its systematic search, thorough critical appraisal and restriction of the metaanalysis to RCTs only. The use of two independent reviewers minimised the risk of mistakes. We further decreased the risk of introducing bias in the analyses by performing a continuity correction. A weakness of our review is that the outcomes ‘pneumonia’ and ‘respiratory failure’, which we combined in the outcome ‘pulmonary complications’, were poorly defined by the authors of the RCTs. This might, for example, explain the opposite results obtained by Sato et al.,7 who carefully defined their outcomes, and Yano et al.,9 who did not provide outcome definitions. However, at the present time, these are the highest quality data available. In addition, the included RCTs were somewhat heterogeneous in terms of timing and dosage of glucocorticoid administration. However, all RCTs administered the glucocorticoid preoperatively, when the greatest effect is expected, and this did not lead to statistical significant heterogeneity. Because these variables are important determinants of the effect of
glucocorticoids on the immune response, we conducted the meta-analysis using a random effects model.26–28 We hypothesised that the perioperative administration of glucocorticoids would reduce pulmonary complications after transthoracic oesophagectomy through dampening of the inflammatory response. This hypothesis was not supported by the results of our meta-analysis of RCTs. However, we did observe variation in effects among studies. Our hypothesis was that there might be interstudy differences concealing a significant effect when all studies were pooled.29,30 The RCTs differed in dosing and timing of methylprednisolone. Weight-dependent doses investigated were a single dosage of 10 or 30 mg kg1 within 30 min of surgery in four RCTs, and fixed doses were a single dose of 250 or 500 mg within 120 to 180 min preoperatively in three RCTs. Mean weight of the patients was not provided, preventing a formal assessment of a dose–response relationship. However, we used fixed dosing as a surrogate for a low dosage, and weightdependent dosing as a surrogate for a high dosage, because, in general, weight-dependent dosing would result in higher dosages. Although this is a surrogate variable, it is interesting that none of the studies that used fixed dosing found a decreased risk of pulmonary complications. Only one RCT using weight-dependent dosing with a very short follow-up (3 days) did not find a decreased risk of pulmonary complications.8 The two subgroups did not differ in other variables. However, because we were not able to perform a formal dose– response assessment and had very small subgroups, we can only suggest that a weight-dependent dosing regimen of 10 mg kg1 within 30 min before surgery seems the most promising regimen for future studies. There is only one high-quality RCT, conducted by Sato et al.7 They found a significantly reduced risk ratio of 0.30 (95% CI 0.09 to 0.99) of pulmonary complications after a dosage of methylprednisolone 10 mg kg.1 In our institution, we base our policy on these results and treat all our
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
Glucocorticoids for oesophagectomy
patients undergoing transthoracic oesophagectomy with this dosage 30 min preoperatively. The mortality rates were provided in only three RCTs. The overall mortality was very low. Takeda et al.21 reported the only two cases, and therefore, a reliable meta-analysis could not be performed. With such low mortality rates, much larger trials are needed for a reliable estimation of the effect on mortality of perioperative glucocorticoid administration. Concerning anastomotic leakage, the meta-analysis points to a potential beneficial effect of glucocorticoids. Although not statistically significant, this effect is the opposite of the hypothesis that perioperative glucocorticoid administration might increase anastomotic leak rates.9,14 However, the CI around the pooled effect was broad (0.23 to 1.61), and no hard conclusion can be drawn. The results regarding infections were somewhat conflicting. The trials by Takeda et al.21 and Sayama et al.13 reported statistically nonsignificant decreased risks for infections after glucocorticoid administration. Sato et al.7 and Yano et al.9 reported statistically nonsignificant increased risks. The reason for these somewhat conflicting results might be lack of blinding in the trials by Takeda et al.21 and Sayama et al.,13 different definitions or chance. These results caused the interval of the pooled result to be broad (0.41 to 2.93), thus preventing drawing of hard conclusions. The results of this review and meta-analysis cannot be directly extrapolated to all patients undergoing transthoracic oesophagectomy in 2014 without some comments. All RCTs were conducted more than a decade ago. We do not know why no RCTs have been conducted more recently. However, a recent large RCT on this subject in cardiothoracic surgery31 showed promising effects of glucocorticoids on the incidence of postoperative pneumonia. This trial may reopen this research field for oesophageal surgery, in which pneumonia is encountered much more frequently. Nowadays, the perioperative inflammatory response is reduced by the use of minimally invasive surgery and protective ventilation, which could lead to reduced pulmonary complications.32,33 Transthoracic minimally invasive oesophagectomy in the prone position is a new procedure that aims to further reduce pulmonary inflammation. This innovative approach does not require one-lung ventilation.34 In addition, postoperative epidural analgesia is now administered more routinely.35 In contrast, the recent introduction of neoadjuvant therapy might negatively affect the inflammatory response.36 Finally, all the RCTs were conducted in Japan. A possible explanation is the locally high incidence of oesophageal cancer, resulting in more research on this issue by Japanese surgeons. The same trend can be observed for neutrophil elastase inhibitors, which are currently under investigation for the same patient group,
693
again mainly in Japan. Recently, Baker et al.37 showed that glucocorticoid receptor polymorphisms influence the functional response to glucocorticoids. Consequently, the effects in populations other than Japanese remain to be determined. These observations clarify the need and direction for further research. A large RCT study is needed to determine whether the results of Sato et al.7 can be reproduced. It is recommended to use the same weight-dependent dosing and timing of methylprednisolone administration as Sato et al.7 because this produced the most promising results to date. The outcome ‘pulmonary complications’ should be defined carefully in order to discern whether the effect differs for the different types of pulmonary complications. Ideally, patients undergoing minimally invasive oesophagectomy with postoperative epidural anaesthesia should be included to improve external validity to future surgical populations.
Conclusion In this meta-analysis, no significant effect of perioperatively administered glucocorticoids on the risk of pulmonary complications after transthoracic oesophagectomy, or adverse effects, was found. Subgroup analyses of studies that used weight-dependent dosing within 30 min preoperatively suggest that methylprednisolone 10 to 30 mg kg1 administered within 30 min preoperatively might be a promising dosing regimen for further research. Future studies should further investigate this dosing regimen and determine effects in patients undergoing minimally invasive surgery with adequate postoperative analgesia.
Acknowledgements relating to this article Assistance with the systematic review: the authors wish to acknowledge the valuable contribution of Nanami Pronk, who helped with the data extraction and critical appraisal of two articles written in Japanese. Financial support and sponsorship: none. Conflicts of interest: none. Presentation: these data were presented as an oral presentation to the European Society for Diseases of the Esophagus Congress, Rotterdam, 31 October to 2 November 2013.
References 1 2
3
4 5 6
GLOBOCAN 2008 v1.2, Cancer incidence and mortality worldwide: IARC CancerBase. http://globocan.iarc.fr [Accessed 20 March 2014]. Omloo JM, Law SY, Launois B, et al. Short and long-term advantages of transhiatal and transthoracic oesophageal cancer resection. Eur J Surg Oncol 2009; 35:793–797. Boshier PR, Anderson O, Hanna GB. Transthoracic versus transhiatal oesophagectomy for the treatment of esophagogastric cancer: a metaanalysis. Ann Surg 2011; 254:894–906. Eichenbaum KD, Neustein SM. Acute lung injury after thoracic surgery. J Cardiothorac Vasc Anesth 2010; 24:681–690. Ng JM. Update on anesthetic management for esophagectomy. Curr Opin Anaesthesiol 2011; 24:37–43. Licker M, Fauconnet P, Villiger Y, Tschopp JM. Acute lung injury and outcomes after thoracic surgery. Curr Opin Anaesthesiol 2009; 22:61– 67.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
694 Weijs et al.
7
8
9
10
11
12
13
14
15
16
17 18
19
20 21
22
Sato N, Koeda K, Ikeda K, et al. Randomized study of the benefits of preoperative corticosteroid administration on the postoperative morbidity and cytokine response in patients undergoing surgery for esophageal cancer. Ann Surg 2002; 236:184–190. Takeda S, Kim C, Ikezaki H, et al. Preoperative administration of methylprednisolone attenuates cytokine-induced respiratory failure after esophageal resection. J Nippon Med Sch 2003; 70:16–20. Yano M, Taniguchi M, Tsujinaka T, et al. Is preoperative methylprednisolone beneficial for patients undergoing oesophagectomy? Hepatogastroenterology 2005; 52:481–485. Okawa K, Onda M, Miyashita M, Sasajima K. Systemic and pulmonary responses of inflammatory cytokines following esophagectomy. Nippon Ika Daigaku Zasshi 1998; 65:42–49. Takemura S, Miya K, Takao H, et al. Influence of corticosteroid preoperative administration for surgical stress of the oesophageal cancer patients during peri-and postoperative periods. Jpn J Gastroenterol Surg 1999; 32:1133– 1141. Matsutani T, Onda M, Sasajima K, Miyashita M. Glucocorticoid attenuates a decrease of antithrombin III following major surgery. J Surg Res 1998; 79:158–163. Sayama J, Shineha R, Yokota K. Control of the excessive reaction after surgery for esophageal carcinoma with preoperative administration of the corticosteroids. Jpn J Gastroenterol Surg 1994; 27:841–848. Shimada H, Ochiai T, Okazumi S, et al. Clinical benefits of steroid therapy on surgical stress in patients with esophageal cancer. Surgery 2000; 128:791–798. Engelman E, Maeyens C. Effect of preoperative single-dose corticosteroid administration on postoperative morbidity following esophagectomy. J Gastrointest Surg 2010; 14:788–804. Raimondi AM, Guimaraes HP, do Amaral JLG, Leal PH. Perioperative glucocorticoid administration for prevention of systemic organ failure in patients undergoing esophageal resection for esophageal carcinoma. Sao Paulo Med J 2006; 124:112–115. Higgins JP, Green S, Cochrane handbook for systematic reviews of interventions, 2008. http://www.cochrane.org/handbook Wells GA, Shea B, O’Connell D et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. 2011. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp Friedrich JO, Adhikari NK, Beyene J. Inclusion of zero total event trials in meta-analyses maintains analytic consistency and incorporates all available data. BMC Med Res Methodol 2007; 23:5. Borenstein M, Hedges L, Higgins J, Rothstein H. Comprehensive metaanalysis version 2 Biostat, Englewood NJ (2005) Takeda S, Ogawa R, Nakanishi K, et al. The effect of preoperative high dose methylprednisolone in attenuating the metabolic response after oesophageal resection. Eur J Surg 1997; 163:511–517. Tsukada K, Miyazaki T, Katoh H, et al. Effect of perioperative steroid therapy on the postoperative course of patients with oesophageal cancer. Dig Liver Dis 2006; 38:240–244.
23
24
25 26
27
28
29 30 31
32
33
34
35
36
37
Nakamura M, Iwahashi M, Nakamori M, et al. An analysis of the factors contributing to a reduction in the incidence of pulmonary complications following an esophagectomy for esophageal cancer. Langenbecks Arch Surg 2008; 393:127–133. Park SY, Lee HS, Jang HJ, et al. Efficacy of intraoperative, single-bolus corticosteroid administration to prevent postoperative acute respiratory failure after oesophageal cancer surgery. Interact Cardiovasc Thorac Surg 2012; 15:639–643. Schulz KF, Grimes DA. Allocation concealment in randomised trials: defending against deciphering. Lancet 2002; 359:614–618. Wan S, DeSmet JM, Antoine M, et al. Steroid administration in heart and heart-lung transplantation: is the timing adequate? Ann Thorac Surg 1996; 61:674–678. Lodge AJ, Chai PJ, Daggett CW, et al. Methylprednisolone reduces the inflammatory response to cardiopulmonary bypass in neonatal piglets: timing of dose is important. J Thorac Cardiovasc Surg 1999; 117:515– 522. Shum-Tim D, Tchervenkov CI, Laliberte E, et al. Timing of steroid treatment is important for cerebral protection during cardiopulmonary bypass and circulatory arrest: minimal protection of pump prime methylprednisolone. Eur J Cardiothorac Surg 2003; 24:125–132. Zen M, Canova M, Campana C, et al. The kaleidoscope of glucocorticoid effects on immune system. Autoimmun Rev 2011; 10:305–310. Buttgereit F. Mechanisms and clinical relevance of nongenomic glucocorticoid actions. Z Rheumatol 2000; 59:119–123. Dieleman JM, Nierich AP, Rosseel PM, et al. Intraoperative high-dose dexamethasone for cardiac surgery: a randomized controlled trial. JAMA 2012; 308:1761–1767. Biere SS, van Berge Henegouwen MI, Maas KW, et al. Minimally invasive versus open oesophagectomy for patients with oesophageal cancer: a multicentre, open-label, randomised controlled trial. Lancet 2012; 379:1887–1892. Michelet P, D’Journo XB, Roch A, et al. Protective ventilation influences systemic inflammation after esophagectomy: a randomized controlled study. Anesthesiology 2006; 105:911–919. Jarral OA, Purkayastha S, Athanasiou T, et al. Thoracoscopic esophagectomy in the prone position. Surg Endosc 2012; 26:2095– 2103. Cense HA, Lagarde SM, de Jong K, et al. Association of no epidural analgesia with postoperative morbidity and mortality after transthoracic esophageal cancer resection. J Am Coll Surg 2006; 202:395– 400. van Hagen P, Hulshof MC, van Lanschot JJ, et al. Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med 2012; 366:2074–2084. Baker AC, Chew VW, Green TL, et al. Single nucleotide polymorphisms and type of steroid impact the functional response of the human glucocorticoid receptor. J Surg Res 2013; 180:27–34.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
ORIGINAL ARTICLE
The effect of perioperative administration of glucocorticoids on pulmonary complications after transthoracic oesophagectomy A systematic review and meta-analysis Teus J. Weijs, Jan M. Dieleman, Jelle P. Ruurda, A. Christiaan Kroese, Hans J.T.A. Knape and Richard van Hillegersberg BACKGROUND Severe pulmonary complications occur frequently following transthoracic oesophagectomy. An exaggerated immunological response is probably a main driving factor, and this might be prevented by perioperative administration of a glucocorticoid. OBJECTIVE To determine the clinical benefits and harms of perioperative glucocorticoid during transthoracic oesophagectomy, using pulmonary complications as the primary outcome. Mortality, anastomotic leakage rate and infection were secondary outcomes. METHODS A systematic review of interventional trials with a meta-analysis of randomised controlled trials (RCTs). RESULTS The search retrieved seven RCTs and four interventional nonrandomised studies. In total, 367 patients received perioperative glucocorticoid and 415 patients did not. A meta-analysis of the RCTs showed no significant effect
of glucocorticoid. For pulmonary complications, the pooled risk ratio was 0.69 [95% confidence interval (CI) 0.26 to 1.79], for anastomotic leakage 0.61 (95% CI 0.23 to 1.61) and for infections 1.09 (95% CI 0.41 to 2.93). A subgroup analysis of RCTs that used weight-dependent dosing within 30 min preoperatively showed a pooled risk ratio of 0.28 (95% CI 0.10 to 0.77) for pulmonary complications compared with placebo. CONCLUSION In this meta-analysis, perioperative administration of glucocorticoid did not affect the risk of pulmonary complications after transthoracic oesophagectomy, nor did it cause adverse effects. A subgroup analysis showed that a weight-dependent dose of methylprednisolone 10 to 30 mg kg1 within 30 min preoperatively might be the most promising dosing regimen for further research. Published online 16 May 2014
Background Oesophageal cancer is the eighth most common cancer and the sixth most common cause of cancer-related death globally.1 A transthoracic surgical approach offers the best chance for long-term survival. It allows meticulous mediastinal dissection resulting in extended lymph node retrieval and radical resection.2 However, this approach is accompanied by severe pulmonary complications, usually pneumonia and acute lung injury, which lead to increased perioperative mortality.3 The intense inflammatory response induced by surgery, in combination with local inflammation due to one-lung ventilation, is thought to play an
important role in the development of these complications.4–6 Perioperative glucocorticoid administration reduces the postoperative inflammatory response after transthoracic oesophagectomy.7–13 In addition, it has been suggested that administration of a glucocorticoid can reduce the incidenceofpulmonarycomplications.7 However,thepotential of glucocorticoid to increase the risks of infection, tumour recurrence and wound healing abnormalities has limited widespread use during transthoracic oesophagectomy.9,14
From the Department of Surgery (TJW, JPR, RVH), and Department of Anaesthesiology, Intensive Care and Emergency Medicine, University Medical Center Utrecht, Utrecht, The Netherlands (JMD, ACK, HJTAK) Correspondence to Jelle P. Ruurda, Department of Surgery, University Medical Center Utrecht, Heidelberglaan 100, 3508 GA Utrecht, The Netherlands Tel: +088 7558074; e-mail: [email protected] 0265-0215 ß 2014 Copyright European Society of Anaesthesiology
DOI:10.1097/EJA.0000000000000093
Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
686 Weijs et al.
Two previous meta-analyses have investigated the outcomes of perioperative glucocorticoid administration for oesophagectomy.15,16 However, the most recent metaanalysis included several studies with a high risk of bias. In the other meta-analysis three randomised controlled trials were not included. This makes the results difficult to interpret. Consequently, we undertook an updated systematic review and meta-analysis of RCTs on perioperative glucocorticoid administration for transthoracic oesophagectomy. The primary objective of this meta-analysis was to determine the effect of perioperative glucocorticoid administration for transthoracic oesophagectomy on pulmonary complications. Secondary objectives were to determine the risks for adverse outcomes: mortality, anastomotic leakage and infections.
Materials and methods Eligibility criteria
All articles that investigated perioperative administration of glucocorticoids for transthoracic oesophagectomy were sought. In order to present a systematic review of available evidence, we considered all randomised and nonrandomised comparative interventional studies for inclusion. Studies not reporting on the primary outcome were excluded. Information sources and search strategy
We conducted a systematic search of Medline (via Pubmed), Embase and Cochrane databases. The search strategy combined all synonyms regarding the intervention, ‘glucocorticoids’ and the domain, ‘oesophagectomy’ (e-supplement 1, http://links.lww.com/EJA/A48). No limitations were applied. The sensitivity of the search strategy was improved by screening all references of included articles for relevant publications that were not retrieved in the initial search. Also, all citing articles and related articles were screened using Web of Science version 5.7. Authors were contacted by e-mail when articles were not available in full text. When additional eligible articles were identified, the search strategy was evaluated and improved if possible. Study selection
Two independent investigators (T.W., S.D.) screened all articles obtained from the database search on title and abstract. Studies were excluded when their population consisted of nonhumans or patients not undergoing oesophagectomy, and if the intervention investigated was not glucocorticoid administration. The remaining articles were fully reviewed. We included all articles fulfilling the eligibility criteria for the systematic review, but only RCTs were included for meta-analysis. No language restrictions were applied to the selection process.
Risk of bias in individual studies
Two investigators (T.W., S.D.) independently assessed the risk of bias at the study level in the included studies. For the studies published in Japanese only, the necessary data were extracted by a Japanese medical student (N.P.) on forms designed by the lead investigator (T.W.). The Cochrane collaboration’s tool for assessing risk of bias was used for RCTs and the Ottawa quality assessment scale for nonrandomised studies.17,18 In addition, the risk of bias on outcome level was assessed by extracting the outcome definitions of individual studies. Disagreements were resolved by discussion. Data collection process
All data were extracted by the leading reviewer (T.W.) and checked for errors by the second reviewer (S.D.). Only published data were extracted; however, authors were contacted by e-mail when verification of data was necessary. Data items
We extracted data on study location and period to prevent inclusion of duplicate publications. Data on study design that were considered important were type, dosing regimen and timing of glucocorticoid administration; oesophagectomy approach and type of intestinal tract reconstruction; use of postoperative epidural analgesia; location of extubation; outcome definitions and duration of follow-up. Finally, the number of events in the intervention and placebo group was extracted for primary and secondary outcomes. The outcome ‘pulmonary complications’ was defined as respiratory failure and/or pneumonia because these are most probably linked to the inflammatory response. For ‘mortality’, primarily in-hospital mortality rates were sought, and secondarily, 30-day mortality rates. For the outcome ‘infections’, all events due to pneumonia were excluded because these were already included in the primary outcome. Data analysis
Absolute risks and risk ratios were calculated and compared using the Mantel–Haenszel method for dichotomous outcomes. A P value of less than 0.05 was considered significant. For individual studies with no events in one or more groups, a continuity correction of 0.5 was added to each cell for each effect measure.19 Publication bias was evaluated for the primary outcome using funnel plot analysis of all RCTs. A meta-analysis of all RCTs was performed. Statistical heterogeneity was investigated using the x2 test, wherein a P value of less than 0.1 was considered to indicate significant heterogeneity. Pooled weighted risk ratios with 95% confidence intervals (CIs) were calculated for all investigated outcomes except mortality because reported mortality rates were too low for reliable calculations. A sensitivity analysis was used to determine the
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
Glucocorticoids for oesophagectomy
impact of single studies on the pooled effect for the primary outcome. Some studies used a fixed-dose, and other a weightdependent dose, without presenting the mean weight of their patients. Because in general, a weight-dependent dosing regimen (10 or 30 mg kg1) would result in higher dosage than a fixed dose (125, 250 or 500 mg), a weightdependent dosing regimen was considered as a surrogate for ‘high dose’, and a fixed dose as a surrogate for ‘low dose’. Studies were grouped by this variable and pooled risk ratios were calculated. No other subgroup analyses were planned or performed. In all analyses without significant statistical and methodological heterogeneity, a fixed effects model was used; in all other analyses, a random effects model was used. All analyses were performed with Comprehensive Metaanalysis software, version 2.20
Results Search results
The results of the search strategy are shown in Fig. 1. The final search was performed on 17 September 2013. From the 30 articles screened in full text, 16 articles investigated irrelevant populations, outcomes or interventions, and two were nonsystematic reviews. Screening references from a meta-analysis retrieved one additional article.13 Finally, 11 articles were included: seven RCTs and four nonrandomised comparative studies.7–9,11–14,21–24
687
Characteristics of included studies
The baseline characteristics of the included studies are summarised in Table 1. This review includes 367 patients receiving perioperative glucocorticoid and 415 patients who did not. In all studies, the glucocorticoid used was methylprednisolone. All seven RCTs were conducted more than a decade ago in Japan.7–9,11–13,21 Methylprednisolone was administered as a fixed dose of 250 or 500 mg, or as a weightdependent dose of 10 or 30 mg kg1. The dosage most frequently used was 10 mg kg1.7–9,12 Timing of methylprednisolone administration ranged from 3 h preoperatively to just before induction of anaesthesia. In all RCTs, a thoracotomy was performed and the stomach was the preferred organ for reconstruction. Data on postoperative epidural analgesia were missing in most RCTs. All RCTs reported on pulmonary complications, but only two provided clear definitions for pulmonary complications (e-supplement 2, http://links.lww.com/EJA/ A48).7,13 Mortality rates were reported in three RCTs.7,9,21 None of the five RCTs reporting anastomotic leakage rates provided clear-cut outcome definitions, and for infection, only one of the five did so.7,9,11,18,21 The length of follow-up was only 3 days in one RCT; in all other RCTs, follow-up was for between 7 days and 5 years. The four nonrandomised studies14,22–24 were conducted more recently, three in Japan, and one in Korea (Table 1). The dosage of methylprednisolone ranged from 125 to
Fig. 1
. Pubmed N = 175
Embase N = 486
Cochrane N = 15
N = 676 Excluding duplicates: N = 183 N = 493 Exclusion based on the title and abstract: N = 462 N = 31
Screening references, citing and related publications: 1 additional articles retrieved
N = 10
Exclusion based on FT: Different outcomes: 4 Different population: 7 Different intervention: 6 Review: 2 Meta-analysis: 2
Included: 11 primary studies
Flowchart showing results of study selection process. FT, full text.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
Location
2003 to 2004 vs. 2002 1995 to 1996 vs. 1997 to 1998 107 (57/50)
36 (21/15)
184 (100/84)
234 (83/151)
17 (8/9)
30 (15/15)
33 (14/19)
18 (9/9)
66 (33/33)
17 (7/10)
40 (20/20)
N (I/C)
Physiological saline Physiological saline ? Physiological saline Physiological saline ?
10 mg kg1
10 mg kg1
250 mg
n.a.
n.a.
n.a.
2 125 mg
250 mg
n.a.
125 mg
250 mg
30 mg kg1
250 mg
10 mg kg1
Physiological saline
Placebo
500 mg
MP Dose
After construction of anastomosis At start of thoracotomy and at closure 60 min preoperatively 60 min preoperatively and 125 mg on POD 1 and 2
Before anaesthesia 30 min preoperatively 30 min preoperatively At induction of anaesthesia Before anaesthesia 120 to 180 min preoperatively
120 min preoperatively
Time of administration
Thoracotomy
Thoracotomy
Thoracotomy or VATS
Thoracotomy
Thoracotomy
Thoracotomy
Thoracotomy
Thoracotomy
Thoracotomy
Thoracotomy
Thoracotomy
Approach
2 or 3-field
3-field
2 or 3-field
2 or 3-field
2 or 3-field
3-field
3-field
2 or 3-field
3-field
3-field
2 or 3-field
Lymph node dissection
?
?
Gastric tube or colon Gastric tube or colon
Stomach or small intestine
Gastric tube
Stomach
Stomach
Stomach
Stomach, small intestine or colon Stomach
Reconstruction
?
?
Cervical or intrathoracic Cervical
Cervical or intrathoracic Cervical or intrathoracic
?
Intrathoracic
Cervical or intrathoracic Cervical
?
Level of anastomosis
?
?
Yes
?
?
Yes
?
?
?
?
?
Epidural analgesia
ICU
ICU
ICU
OR
ICU
ICU
?
ICU
ICU
ICU
?
Location of extubation
?, unknown; C, control group; I, intervention group; i.v., intravenously; MP, methylprednisolone; N, total number of included patients; n.a., not applicable; OR, operation room; POD, postoperative day; VATS, video-assisted thoracoscopy.
Japan
Japan
?
Japan Sayama et al.13 Nonrandomised studies Park et al.24 Korea
Tsukada et al.22 Shimada et al.14
?
Japan
Japan
?
Japan
Nakamura et al.23
1996 to 1997
Japan
Takemura et al.11 Matsutani et al.12 Takeda et al.21
2007 to 2008 vs. 2005 to 2006 2001 to 2005 vs. 1991 to 2000
1996 to 1999
Japan
Sato et al.7
?
Japan
Takeda et al.8
1997 to 1999
Period
Characteristics of included interventional studies
Randomised studies Japan Yano et al.9
Study
Table 1
688 Weijs et al.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
Glucocorticoids for oesophagectomy
250 mg, and the timing of administration ranged from 1 h preoperatively to directly after construction of the anastomosis. Additional doses of methylprednisolone were administered after surgery in two nonrandomised studies.14,23 In general, the oesophagectomy was conducted through a thoracotomy, but in one study,23 a videoassisted thoracoscopic approach was also used. All four nonrandomised studies provided clear definitions for pulmonary complications, but all lacked definitions for anastomotic leakage and infections (e-supplement 2, http://links.lww.com/EJA/A48).14,22–24 The duration of follow-up was 30 days in two of the nonrandomised studies, and the entire admission time in the other two studies. Risk of bias in individual studies
The results of the critical appraisal on study level are summarised in Table 2. Among the seven RCTs, five did not describe whether they concealed allocation and blinded participants, personnel and outcome assessment.8,11–13,21 Thus, the risk for bias was high in these studies.25 The study by Yano et al.9 was deemed at risk of bias because the authors did not describe allocation concealment. The only study considered to be at low risk of bias was conducted by Sato et al.7 Among the nonrandomised studies, the study of Shimada et al.14 did not investigate nor correct for confounding factors. All nonrandomised studies14,22–24 compared patients from different time periods, and therefore, the risk for bias was considered to be very high. Results of individual studies
The individual study results are shown in Figs 2–5. Table 2
689
Seven RCTs, which included 221 patients, reported on the primary outcome, pulmonary complications (Fig. 2). The risk ratios for pulmonary complications after perioperative administration of methylprednisolone compared with placebo ranged from 0.09 (95% CI 0.01 to 1.51) to 3.00 (95% CI 0.14 to 65.16).7–9,11–13,21 The two large and high-quality RCTs obtained somewhat contradictory results. In the study by Sato et al.,7 a risk ratio of 0.30 (95% CI 0.09 to 0.99) favouring methylprednisolone was found. In contrast, Yano et al.9 found a nonsignificant increase in risk in the methylprednisolone group to 2.0 (95% CI 0.58 to 6.91). Regarding the secondary outcomes, mortality rates were reported in three RCTs, which included 136 patients (Fig. 3). Takeda et al.21 reported the only two cases of postoperative mortality, both in the placebo group. In five RCTs that included 171 patients, anastomotic leakage was reported (Fig. 4). The risk ratios for anastomotic leakage ranged from 0.20 (95% CI 0.01 to 3.85) to 3.00 (95% CI 0.14 to 65.16).7,9,11,13,21 Infection rates were reported in the same five RCTs (Fig. 5). The risk ratios for infections after perioperative administration of glucocorticoid compared with placebo ranged from 0.16 (95% CI 0.01 to 2.67) to 2.33 (95% CI 0.70 to 7.76). All four nonrandomised studies,14,22–24 which included 561 patients, reported on the primary outcome, pulmonary complications (Fig. 2). The risk ratios for pulmonary complications ranged from 0.24 (95% CI 0.12 to 0.50) to 0.70 (95% CI 0.41 to 1.20). Regarding the secondary outcomes, all reported on mortality (Fig. 3). The risk ratios for postoperative mortality ranged from 0.42 (95% CI 0.08 to 2.24) to 0.88 (95% CI 0.06 to 13.66). On anastomotic leakage, three nonrandomised studies,14,22,24 which included 377 patients, found that the risk ratios for
Critical appraisal of studies
Randomised studies Item Randomisation? Allocation concealment? Blinding of participants and personnel? Blinding of outcome assessment? Outcome data complete? No selective outcome reporting? Risk of bias Nonrandomised studies Item Representative exposed cohort? Nonexposed selected from same population? Ascertainment of exposure reliable? Outcome of interest not present from start? Adequately dealt with confounding factors? Reliable outcome assessment? Adequately dealt with loss to follow-up? No selective outcome reporting? Risk of bias
Sayama et al.13
Takeda et al.21
Matsutani et al.12
Takemura et al.11
Sato et al.7
Takeda et al.8
Yano et al.9
þ ? ? ? þ ? High
þ ? ? ? þ þ High
þ ? ? ? þ High
þ ? ? ? þ ? High
þ þ þ ? þ þ Low
þ ? ? ? þ High
þ ? þ ? þ þ High
Shimada et al.14
Tsukada et al.22
Nakamura et al.23
Park et al.24
þ þ þ ? þ High
þ þ þ þ ? þ High
þ þ þ þ ? þ High
þ þ þ þ ? þ High
þ, yes; , no; ?, unclear.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
690 Weijs et al.
Fig. 2
Study
Year
Randomised studies Yano9 2005 Takeda8a 2003 2002 Sato7 1999 Takemura11a Matsutani12a 1998 1997 Takeda21a Sayama13 1994 Pooled result
Control-group
MP-group
Risk ratio
Events
No events
Events
No events
6 0.5 3 1.5 0.5 0.5 1
14 7.5 30 8.5 14.5 15.5 7
3 0.5 10 0.5 2.5 5.5 1
17 10.5 23 9.5 17.5 10.5 8
Non-randomised studies Park24 2012 2008 Nakmura23 2006 Tsukada22 2000 Shimada14
Forest plot
2.00 (0.58 – 6.91) 1.38 (0.03 – 62.25) 0.30 (0.09 – 0.99)b 3.00 (0.14 – 65.16) 0.27 (0.01 – 5.15) 0.09 (0.01 – 1.51) 1.13 (0.08 – 15.19) 0.69 (0.26 – 1.79)
2 8 3 16
81 92
16 28 4 20
18 41
135 56 11 30
0.23 (0.05 – 0.96)b,c 0.24 (0.12 – 0.50)b,c 0.54 (0.14 – 2.05) 0.70 (0.14 – 1.20) 0.01 0.1 Favours methylprednisolone
1
10 Favours control
100
Table and forest plot summarising study results concerning pulmonary complications. Numbers between brackets are lower and upper limits of 95% confidence intervals. MP, methylprednisolone. aContinuity correction applied. bSignificant result. cSignificant result in multivariate analysis.
postoperative anastomotic leakage ranged from 0.22 (95% CI 0.05 to 0.99) to 1.36 (95% CI 0.49 to 3.80) (Fig. 4). Infection rates were reported in one nonrandomised study,22 which included 36 patients (Fig. 5); the risk ratio for infections was 1.07 (95% CI 0.20 to 5.65).
Secondary outcomes
The pooled risk ratio for anastomotic leakage in five RCTs7,9,11,13,21 was 0.61 (95% CI 0.23 to 1.61; P ¼ 0.32) (Fig. 4). Pooling the results for infections from the same five RCTs resulted in a risk ratio of 1.09 (95% CI 0.41 to 2.93; P ¼ 0.86) (Fig. 5).
Meta-analysis
The results of the meta-analysis of RCTs are shown in Figs 2–5. Primary outcome
The results of the seven RCTs that reported on pulmonary complications7–9,11–13,21 were pooled using a random effects model (Fig. 2). This resulted in a pooled risk ratio of 0.69 (95% CI 0.26 to 1.79; P ¼ 0.44) for pulmonary complications when glucocorticoid was administered perioperatively compared with placebo.
Ancillary analyses
Statistically, there was no significant interstudy heterogeneity. However, the studies were heterogeneous in their study methods; therefore, a random effects model was used for pooling of the results. No indication for publication bias regarding the primary outcome was found in the funnel plot (e-supplement 3, http://links. lww.com/EJA/A48). The studies of Yano et al.9 and Sato et al.7 had the greatest impact on the pooled result, in opposite directions
Fig. 3
Study
Year
Randomised studies 2005 Yano9 2002 Sato7 21 Takeda 1997 Non-randomised studies Park24 2012 Nakmura23 2008 2006 Tsukada22 Shimada14 2000
MP-group
Control group
Risk ratio
Events
No events
Events
No events
0 0 0
20 33 15
0 0 2
20 33 15
n.a. n.a. n.a.
2 2 0 1
83 100 20 56
8 4 0 1
151 84 15 49
0.45 (0.10 – 2.09) 0.42 (0.08 – 2.24) n.a. 0.88 (0.06 – 13.66)
Forest plot
0.01
0.1
Favours methylprednisolone
1
10
100
Favours placebo
Table and forest plot summarising study results concerning mortality. Numbers between brackets are lower and upper limits of 95% confidence intervals. MP, methylprednisolone; n.a., not applicable.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
Glucocorticoids for oesophagectomy
691
Fig. 4
Study
MP-group
Year
Randomised studies 2005 Yano9 Sato7 2002 11a Takemura 1999 21a Takeda 1997 13a 1994 Sayama Pooled result Non-randomised studies Park24 2012 Tsukada22 2006 2000 Shimada14
Control-group
Risk ratio
Events
No events
Events
No events
3 1 1.5 0.5 0.5
17 32 9.5 15.5 8.5
4 2 0.5 2.5 2.5
16 31 9.5 15.5 9.5
0.75 (0.19 – 2.93) 0.50 (0.05 – 5.25) 3.00 (0.14 – 65.16) 0.20 (0.01 – 3.85) 0.22 (0.01 – 4.04) 0.61 (0.23 – 1.61)
6 2 2
77 19 55
8 2 8
143 13 42
0.36 (0.49 – 3.80) 0.71 (0.11 – 4.52) 0.22 (0.05 – 0.99)b
Forest plot
0.01 0.1 1 Favours methylprednisolone
10 Favours placebo
100
Table and forest plot summarising study results concerning anastomotic leakage. Numbers between brackets are lower and upper limits of 95% confidence intervals. MP, methylprednisolone. aContinuity correction applied. bsignificant result.
(e-supplement 4, http://links.lww.com/EJA/A48). The pooled risk ratio excluding the results of Yano et al.9 was 0.45 (95% CI 0.15 to 1.40); the pooled risk ratio excluding the results of Sato et al.7 was 0.96 (95% CI, 0.31 to 3.00). Subgroup analysis
The impact of using a weight-dependent dosage of glucocorticoid rather than a fixed dose on the risk for pulmonary complications is shown in Fig. 6. A fixed dose was used in three studies,9,11,13 which included 75 patients. Pooling their results resulted in a risk ratio for pulmonary complications after a fixed dose of glucocorticoid of 1.91 (95% CI 0.67 to 5.46; P ¼ 0.23) compared with placebo. A weight-dependent dosage was used in four studies,7,8,12,21 which included 146 patients. The pooled risk ratio for pulmonary complications was 0.28 (95% CI 0.10 to 0.77; P ¼ 0.01) after a weight-dependent dosage of glucocorticoid compared with placebo. A random effects model was used for the subgroup analysis.
Discussion This systematic review and meta-analysis summarises the evidence for the benefits and harms of perioperative glucocorticoid administration for transthoracic oesophagectomy. In total, 782 patients were included in 11 studies, which were heterogeneous in both design and results. In a pooled analysis of RCTs, we were unable to demonstrate either clinical benefits or adverse effects of perioperative administration of glucocorticoid. A subgroup analysis of RCTs using a weight-dependant dose of 10 to 30 mg kg1 of glucocorticoid within 30 min preoperatively suggests that this regimen might be effective in reducing pulmonary complications compared with a fixed dose of 250 or 500 mg given 120 to 180 min preoperatively. Recently, Engelman and Maeyens15 published a metaanalysis on this subject. The authors found significantly decreased risks of pulmonary complications and infections in their pooled analyses, and their analysis showed a trend towards decreased risks of anastomotic leakage and
Fig. 5
Study
Year
Randomised studies 2005 Yano9 2002 Sato7 11 1999 Takemura 21a 1997 Takeda 13a 1994 Sayama Pooled result Non-randomised studies Tsukada22 2006
MP-group Events
Control-group
No events Events
Risk ratio
Forest plot
No events
7 3 1 0.5 0.5
13 30 8 15.5 8.5
3 2 1 2.5 3.5
17 31 8 13.5 6.5
2.33 (0.70 – 7.76) 1.50 (0.27 – 8.40) 1.00 (0.07 – 13.84) 0.20 (0.01 – 3.85) 0.16 (0.01 – 2.67) 1.09 (0.41 – 2.93)
3
18
2
13
1.07 (0.20 – 5.65) 0.01 0.1 1 Favours methylprednisolone
10 Favours placebo
100
Table and forest plot summarising study results concerning infections. Numbers between brackets are lower and upper limits of 95% confidence intervals. MP, methylprednisolone. aContinuity correction applied.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
692 Weijs et al.
Fig. 6 Study
Year
MP-group Events
Fixed dose Yano9 2005 1999 Takemura11a Sayama13 1994 Pooled result
Control-group
No events Events
Risk ratio
Forest plot
No events
5 1.5 1
14 0.5 7
3 0.5 1
17 9.5 8
2.00 (0.58 – 6.91) 3.00 (0.14 – 65.16) 1.13 (0.06 – 15.19) 1.91 (0.67 – 5.46)
0.5 3 0.5 0.5
7.5 30 14.4 15.5
0.5 10 2.5 5.6
10.5 23 17.5 10.5
1.38 (0.03 – 62.25) 0.30 (0.09 – 0.99)b 0.27 (0.01 – 5.15) 0.09 (0.01 – 1.51)b 0.28 (0.10 – 0.77)
Weight dependent dose Takeda8a Sato7 Matsutsni10a Takeda21a Pooled result
2003 2002 1998 1997
0.01 0.1 Favours methylprednisolone
1
10 Favours control
100
Table and forest plot in which the effect of a fixed dose of methylprednisolone is compared with a weight-dependent dose. Numbers between brackets are lower and upper limits of 95% confidence intervals. MP, methylprednisolone. aContinuity correction applied. bSignificant result.
early mortality. However, their results should be interpreted with caution. In their meta-analysis, two nonrandomised studies,14,22 highly at a risk for bias, were included. In addition, they did not identify one other relevant study.11 Furthermore, studies with zero events in the study group were not included in this metaanalysis. In these cases, it has been shown that including such studies by performing continuity correction (of 0.5 events to each arm) for the meta-analysis leads to less bias.19 Finally, the results of the RCT by Yano et al.9 were presented, and probably analysed, the wrong way round, making their results difficult to interpret. Consequently, we decided to conduct the present review and metaanalysis, avoiding these pitfalls. Our new approach, excluding studies highly at risk of bias, may explain the different results obtained in this meta-analysis, compared with the results obtained by Engelman and Maeyens.15 The strengths of our review are its systematic search, thorough critical appraisal and restriction of the metaanalysis to RCTs only. The use of two independent reviewers minimised the risk of mistakes. We further decreased the risk of introducing bias in the analyses by performing a continuity correction. A weakness of our review is that the outcomes ‘pneumonia’ and ‘respiratory failure’, which we combined in the outcome ‘pulmonary complications’, were poorly defined by the authors of the RCTs. This might, for example, explain the opposite results obtained by Sato et al.,7 who carefully defined their outcomes, and Yano et al.,9 who did not provide outcome definitions. However, at the present time, these are the highest quality data available. In addition, the included RCTs were somewhat heterogeneous in terms of timing and dosage of glucocorticoid administration. However, all RCTs administered the glucocorticoid preoperatively, when the greatest effect is expected, and this did not lead to statistical significant heterogeneity. Because these variables are important determinants of the effect of
glucocorticoids on the immune response, we conducted the meta-analysis using a random effects model.26–28 We hypothesised that the perioperative administration of glucocorticoids would reduce pulmonary complications after transthoracic oesophagectomy through dampening of the inflammatory response. This hypothesis was not supported by the results of our meta-analysis of RCTs. However, we did observe variation in effects among studies. Our hypothesis was that there might be interstudy differences concealing a significant effect when all studies were pooled.29,30 The RCTs differed in dosing and timing of methylprednisolone. Weight-dependent doses investigated were a single dosage of 10 or 30 mg kg1 within 30 min of surgery in four RCTs, and fixed doses were a single dose of 250 or 500 mg within 120 to 180 min preoperatively in three RCTs. Mean weight of the patients was not provided, preventing a formal assessment of a dose–response relationship. However, we used fixed dosing as a surrogate for a low dosage, and weightdependent dosing as a surrogate for a high dosage, because, in general, weight-dependent dosing would result in higher dosages. Although this is a surrogate variable, it is interesting that none of the studies that used fixed dosing found a decreased risk of pulmonary complications. Only one RCT using weight-dependent dosing with a very short follow-up (3 days) did not find a decreased risk of pulmonary complications.8 The two subgroups did not differ in other variables. However, because we were not able to perform a formal dose– response assessment and had very small subgroups, we can only suggest that a weight-dependent dosing regimen of 10 mg kg1 within 30 min before surgery seems the most promising regimen for future studies. There is only one high-quality RCT, conducted by Sato et al.7 They found a significantly reduced risk ratio of 0.30 (95% CI 0.09 to 0.99) of pulmonary complications after a dosage of methylprednisolone 10 mg kg.1 In our institution, we base our policy on these results and treat all our
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
Glucocorticoids for oesophagectomy
patients undergoing transthoracic oesophagectomy with this dosage 30 min preoperatively. The mortality rates were provided in only three RCTs. The overall mortality was very low. Takeda et al.21 reported the only two cases, and therefore, a reliable meta-analysis could not be performed. With such low mortality rates, much larger trials are needed for a reliable estimation of the effect on mortality of perioperative glucocorticoid administration. Concerning anastomotic leakage, the meta-analysis points to a potential beneficial effect of glucocorticoids. Although not statistically significant, this effect is the opposite of the hypothesis that perioperative glucocorticoid administration might increase anastomotic leak rates.9,14 However, the CI around the pooled effect was broad (0.23 to 1.61), and no hard conclusion can be drawn. The results regarding infections were somewhat conflicting. The trials by Takeda et al.21 and Sayama et al.13 reported statistically nonsignificant decreased risks for infections after glucocorticoid administration. Sato et al.7 and Yano et al.9 reported statistically nonsignificant increased risks. The reason for these somewhat conflicting results might be lack of blinding in the trials by Takeda et al.21 and Sayama et al.,13 different definitions or chance. These results caused the interval of the pooled result to be broad (0.41 to 2.93), thus preventing drawing of hard conclusions. The results of this review and meta-analysis cannot be directly extrapolated to all patients undergoing transthoracic oesophagectomy in 2014 without some comments. All RCTs were conducted more than a decade ago. We do not know why no RCTs have been conducted more recently. However, a recent large RCT on this subject in cardiothoracic surgery31 showed promising effects of glucocorticoids on the incidence of postoperative pneumonia. This trial may reopen this research field for oesophageal surgery, in which pneumonia is encountered much more frequently. Nowadays, the perioperative inflammatory response is reduced by the use of minimally invasive surgery and protective ventilation, which could lead to reduced pulmonary complications.32,33 Transthoracic minimally invasive oesophagectomy in the prone position is a new procedure that aims to further reduce pulmonary inflammation. This innovative approach does not require one-lung ventilation.34 In addition, postoperative epidural analgesia is now administered more routinely.35 In contrast, the recent introduction of neoadjuvant therapy might negatively affect the inflammatory response.36 Finally, all the RCTs were conducted in Japan. A possible explanation is the locally high incidence of oesophageal cancer, resulting in more research on this issue by Japanese surgeons. The same trend can be observed for neutrophil elastase inhibitors, which are currently under investigation for the same patient group,
693
again mainly in Japan. Recently, Baker et al.37 showed that glucocorticoid receptor polymorphisms influence the functional response to glucocorticoids. Consequently, the effects in populations other than Japanese remain to be determined. These observations clarify the need and direction for further research. A large RCT study is needed to determine whether the results of Sato et al.7 can be reproduced. It is recommended to use the same weight-dependent dosing and timing of methylprednisolone administration as Sato et al.7 because this produced the most promising results to date. The outcome ‘pulmonary complications’ should be defined carefully in order to discern whether the effect differs for the different types of pulmonary complications. Ideally, patients undergoing minimally invasive oesophagectomy with postoperative epidural anaesthesia should be included to improve external validity to future surgical populations.
Conclusion In this meta-analysis, no significant effect of perioperatively administered glucocorticoids on the risk of pulmonary complications after transthoracic oesophagectomy, or adverse effects, was found. Subgroup analyses of studies that used weight-dependent dosing within 30 min preoperatively suggest that methylprednisolone 10 to 30 mg kg1 administered within 30 min preoperatively might be a promising dosing regimen for further research. Future studies should further investigate this dosing regimen and determine effects in patients undergoing minimally invasive surgery with adequate postoperative analgesia.
Acknowledgements relating to this article Assistance with the systematic review: the authors wish to acknowledge the valuable contribution of Nanami Pronk, who helped with the data extraction and critical appraisal of two articles written in Japanese. Financial support and sponsorship: none. Conflicts of interest: none. Presentation: these data were presented as an oral presentation to the European Society for Diseases of the Esophagus Congress, Rotterdam, 31 October to 2 November 2013.
References 1 2
3
4 5 6
GLOBOCAN 2008 v1.2, Cancer incidence and mortality worldwide: IARC CancerBase. http://globocan.iarc.fr [Accessed 20 March 2014]. Omloo JM, Law SY, Launois B, et al. Short and long-term advantages of transhiatal and transthoracic oesophageal cancer resection. Eur J Surg Oncol 2009; 35:793–797. Boshier PR, Anderson O, Hanna GB. Transthoracic versus transhiatal oesophagectomy for the treatment of esophagogastric cancer: a metaanalysis. Ann Surg 2011; 254:894–906. Eichenbaum KD, Neustein SM. Acute lung injury after thoracic surgery. J Cardiothorac Vasc Anesth 2010; 24:681–690. Ng JM. Update on anesthetic management for esophagectomy. Curr Opin Anaesthesiol 2011; 24:37–43. Licker M, Fauconnet P, Villiger Y, Tschopp JM. Acute lung injury and outcomes after thoracic surgery. Curr Opin Anaesthesiol 2009; 22:61– 67.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
694 Weijs et al.
7
8
9
10
11
12
13
14
15
16
17 18
19
20 21
22
Sato N, Koeda K, Ikeda K, et al. Randomized study of the benefits of preoperative corticosteroid administration on the postoperative morbidity and cytokine response in patients undergoing surgery for esophageal cancer. Ann Surg 2002; 236:184–190. Takeda S, Kim C, Ikezaki H, et al. Preoperative administration of methylprednisolone attenuates cytokine-induced respiratory failure after esophageal resection. J Nippon Med Sch 2003; 70:16–20. Yano M, Taniguchi M, Tsujinaka T, et al. Is preoperative methylprednisolone beneficial for patients undergoing oesophagectomy? Hepatogastroenterology 2005; 52:481–485. Okawa K, Onda M, Miyashita M, Sasajima K. Systemic and pulmonary responses of inflammatory cytokines following esophagectomy. Nippon Ika Daigaku Zasshi 1998; 65:42–49. Takemura S, Miya K, Takao H, et al. Influence of corticosteroid preoperative administration for surgical stress of the oesophageal cancer patients during peri-and postoperative periods. Jpn J Gastroenterol Surg 1999; 32:1133– 1141. Matsutani T, Onda M, Sasajima K, Miyashita M. Glucocorticoid attenuates a decrease of antithrombin III following major surgery. J Surg Res 1998; 79:158–163. Sayama J, Shineha R, Yokota K. Control of the excessive reaction after surgery for esophageal carcinoma with preoperative administration of the corticosteroids. Jpn J Gastroenterol Surg 1994; 27:841–848. Shimada H, Ochiai T, Okazumi S, et al. Clinical benefits of steroid therapy on surgical stress in patients with esophageal cancer. Surgery 2000; 128:791–798. Engelman E, Maeyens C. Effect of preoperative single-dose corticosteroid administration on postoperative morbidity following esophagectomy. J Gastrointest Surg 2010; 14:788–804. Raimondi AM, Guimaraes HP, do Amaral JLG, Leal PH. Perioperative glucocorticoid administration for prevention of systemic organ failure in patients undergoing esophageal resection for esophageal carcinoma. Sao Paulo Med J 2006; 124:112–115. Higgins JP, Green S, Cochrane handbook for systematic reviews of interventions, 2008. http://www.cochrane.org/handbook Wells GA, Shea B, O’Connell D et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. 2011. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp Friedrich JO, Adhikari NK, Beyene J. Inclusion of zero total event trials in meta-analyses maintains analytic consistency and incorporates all available data. BMC Med Res Methodol 2007; 23:5. Borenstein M, Hedges L, Higgins J, Rothstein H. Comprehensive metaanalysis version 2 Biostat, Englewood NJ (2005) Takeda S, Ogawa R, Nakanishi K, et al. The effect of preoperative high dose methylprednisolone in attenuating the metabolic response after oesophageal resection. Eur J Surg 1997; 163:511–517. Tsukada K, Miyazaki T, Katoh H, et al. Effect of perioperative steroid therapy on the postoperative course of patients with oesophageal cancer. Dig Liver Dis 2006; 38:240–244.
23
24
25 26
27
28
29 30 31
32
33
34
35
36
37
Nakamura M, Iwahashi M, Nakamori M, et al. An analysis of the factors contributing to a reduction in the incidence of pulmonary complications following an esophagectomy for esophageal cancer. Langenbecks Arch Surg 2008; 393:127–133. Park SY, Lee HS, Jang HJ, et al. Efficacy of intraoperative, single-bolus corticosteroid administration to prevent postoperative acute respiratory failure after oesophageal cancer surgery. Interact Cardiovasc Thorac Surg 2012; 15:639–643. Schulz KF, Grimes DA. Allocation concealment in randomised trials: defending against deciphering. Lancet 2002; 359:614–618. Wan S, DeSmet JM, Antoine M, et al. Steroid administration in heart and heart-lung transplantation: is the timing adequate? Ann Thorac Surg 1996; 61:674–678. Lodge AJ, Chai PJ, Daggett CW, et al. Methylprednisolone reduces the inflammatory response to cardiopulmonary bypass in neonatal piglets: timing of dose is important. J Thorac Cardiovasc Surg 1999; 117:515– 522. Shum-Tim D, Tchervenkov CI, Laliberte E, et al. Timing of steroid treatment is important for cerebral protection during cardiopulmonary bypass and circulatory arrest: minimal protection of pump prime methylprednisolone. Eur J Cardiothorac Surg 2003; 24:125–132. Zen M, Canova M, Campana C, et al. The kaleidoscope of glucocorticoid effects on immune system. Autoimmun Rev 2011; 10:305–310. Buttgereit F. Mechanisms and clinical relevance of nongenomic glucocorticoid actions. Z Rheumatol 2000; 59:119–123. Dieleman JM, Nierich AP, Rosseel PM, et al. Intraoperative high-dose dexamethasone for cardiac surgery: a randomized controlled trial. JAMA 2012; 308:1761–1767. Biere SS, van Berge Henegouwen MI, Maas KW, et al. Minimally invasive versus open oesophagectomy for patients with oesophageal cancer: a multicentre, open-label, randomised controlled trial. Lancet 2012; 379:1887–1892. Michelet P, D’Journo XB, Roch A, et al. Protective ventilation influences systemic inflammation after esophagectomy: a randomized controlled study. Anesthesiology 2006; 105:911–919. Jarral OA, Purkayastha S, Athanasiou T, et al. Thoracoscopic esophagectomy in the prone position. Surg Endosc 2012; 26:2095– 2103. Cense HA, Lagarde SM, de Jong K, et al. Association of no epidural analgesia with postoperative morbidity and mortality after transthoracic esophageal cancer resection. J Am Coll Surg 2006; 202:395– 400. van Hagen P, Hulshof MC, van Lanschot JJ, et al. Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med 2012; 366:2074–2084. Baker AC, Chew VW, Green TL, et al. Single nucleotide polymorphisms and type of steroid impact the functional response of the human glucocorticoid receptor. J Surg Res 2013; 180:27–34.
Eur J Anaesthesiol 2014; 31:685–694 Copyright © European Society of Anaesthesiology. Unauthorized reproduction of this article is prohibited.
