Systematic review and meta-analysis of randomized clinical trials of self-expanding metallic stents as a bridge to surgery versus emergency surgery for malignant left-sided large bowel obstruction

Abstract

Background:

Use of self-expanding metallic stents (SEMS) as a bridge to surgery has been suggested as an alternative management for acute malignant left-sided colonic obstruction, as emergency surgery has a high risk of morbidity and mortality. This meta-analysis evaluated high-quality evidence comparing preoperative SEMS with emergency surgery.

Methods:

Relevant randomized clinical trials (RCTs) were identified from the Cochrane Central Register of Controlled Trials, MEDLINE, Embase and PubMed (1990–2011). Primary outcomes were primary anastomosis, stoma and in-hospital mortality rates. Secondary outcomes included anastomotic leak, 30-day reoperation and surgical-site infection rates.

Results:

Four RCTs with 234 patients were included. Technical and clinical success rates for stenting were 70·7 per cent (82 of 116) and 69·0 per cent (80 of 116) respectively. The clinical perforation rate was 6·9 per cent (8 of 116) and the silent perforation rate 14 per cent (11 of 77). SEMS intervention resulted in significantly higher successful primary anastomosis (risk ratio (RR) 1·58, 95 per cent confidence interval 1·22 to 2·04; P < 0·001) and lower overall stoma (RR 0·71, 0·56 to 0·89; P = 0·004) rates. There was no difference in primary anastomosis, permanent stoma, in-hospital mortality, anastomotic leak, 30-day reoperation and surgical-site infection rates. Three trials were stopped prematurely, one because the emergency surgery group had a significantly increased anastomotic leak rate, and two others because of stent-related complications and increased 30-day morbidity following SEMS management.

Conclusion:

Technical and clinical success rates for stenting were lower than expected. SEMS is associated with a high incidence of clinical and silent perforation. However, as a bridge to surgery, SEMS has higher successful primary anastomosis and lower overall stoma rates, with no significant difference in complications or mortality. Copyright © 2012 British Journal of Surgery Society Ltd. Published by John Wiley & Sons, Ltd.

Introduction

Colorectal carcinoma can present as an emergency with bowel obstruction, perforation and bleeding requiring urgent surgical intervention. Some 10–30 per cent of patients with colorectal carcinoma present with acute obstructive symptoms1. Large bowel obstruction results in massive colonic distension, bacterial translocation, electrolyte and fluid imbalance, and an increased risk of colonic necrosis and perforation; urgent surgical decompression is needed. Large bowel obstruction is more common with left-sided lesions than with lesions of the right colon2. Emergency presentations of colorectal carcinoma are also associated with worse oncological outcomes and a higher incidence of local spread and metastatic disease3.

The general consensus for treatment of acute right-sided colonic obstruction is resection and primary anastomosis4. However, there is still significant debate regarding the best surgical treatment for acute malignant left-sided large bowel obstruction. Previously, surgical treatment involved staged procedures with resection and end colostomy followed by later reversal, but one-stage procedures (primary resection and anastomosis) are becoming increasingly popular. Not surprisingly, emergency surgery for acute obstruction is associated with an increased risk of medical and surgical morbidity of up to 40 per cent3, 5, 6. Furthermore, a recent retrospective cross-sectional population-based study7 of the National Cancer Data Repository reported that operative urgency was associated with increased postoperative mortality; the 30-day postoperative mortality rate for emergency surgery was 14·9 per cent, compared with 5·8 per cent for patients operated on electively. This provides the rationale for the use of self-expanding metallic stents (SEMS) as a bridge to surgery for acute malignant left-sided large bowel obstruction. The aim of this meta-analysis was to compare SEMS as a bridge to surgery versus emergency surgery in the management of acute left-sided large bowel obstruction.

Methods

Methodology was developed from the standard guidelines outlined in the Cochrane Handbook for Systematic Reviews of Interventions (version 5.1.0)8 and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement9.

Search strategy and study selection

A detailed electronic search was carried out from the following databases: Cochrane Central Register of Controlled Trials, MEDLINE, Embase and PubMed. The search was performed using the Medical Subject Headings (MeSH): ‘stents’, ‘colectomy’ and ‘intestinal obstruction’ combined with Boolean operator ‘AND’. Other free-text search terms used were: ‘colonic stent’, ‘colorectal stent’, ‘SEMS’, ‘large bowel obstruction’, ‘comparative studies’, ‘Hartmann's procedure’ and ‘Hartmann's’. No language limitation was applied to the search. All studies published from 1990 to 2011 were considered.

Abstracts of potentially relevant publications based on the titles were read and comparative studies of SEMS versus emergency surgery retrieved. A review of all the comparative studies was performed and only randomized clinical trials (RCTs) were included. A hand-search of the references of all comparative studies retrieved was undertaken for any further potential studies; these were then reviewed.

Inclusion and exclusion criteria

Inclusion criteria were: RCTs, use of SEMS as bridge to surgery versus emergency surgery, and study participants with acute malignant left-sided large bowel obstruction. Exclusion criteria were: non-randomized trials, use of SEMS for palliative treatment, and non-malignant large bowel obstruction.

Data extraction

All data were extracted independently by two reviewers using a paper data extraction pro forma. The information extracted from each study was: year of publication, study design, inclusion criteria, exclusion criteria, type of SEMS used, number of participants in each group, types of surgery used, and data on efficacy and safety of SEMS such as technical success, failure and stent complication rates. Data to assess methodological quality of the studies were also extracted.

Primary outcomes were rates of primary anastomosis, successful primary anastomosis (number of technically successful primary anastomoses where clinical anastomotic leakage had not occurred), overall stoma (permanent and temporary stoma formation), permanent stoma (all types of stoma still present at the conclusion of the trial) and in-hospital mortality.

Secondary outcomes were anastomotic leak, reoperation and surgical-site infection (SSI; defined as the presence of pathogenic organisms in a wound site or space, giving rise to local signs and symptoms of infection10) rates. Data for these outcomes were extracted using the data extraction pro forma. Authors of all included RCTs were contacted for any unclear or missing information, and asked to complete a questionnaire regarding duration of follow-up, hospital stay, and short- and long-term complications.

Assessment of risk of bias

Two raters independently assessed the methodological quality of the included studies using the Cochrane Collaboration's tool for assessing risk of bias8. This included random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, assessment of incomplete data outcome, selective reporting and other source bias.

Statistical analysis

Meta-analysis was performed using the software RevMan 5·1 provided by the Cochrane Collaboration (http://ims.cochrane.org/revman). All the assessed outcomes were dichotomous variables. Statistical analysis was performed using risk ratio (RR) as the summary statistic with corresponding 95 per cent confidence interval (c.i.). Meta-analysis using a fixed-effect model and a random-effects model was performed for each outcome and the analysis compared. When there was no discrepancy between the two analyses, the fixed-effect model was used. When there was a discrepancy, both models were presented. P < 0·050 was considered statistically significant.

Statistical heterogeneity was calculated using the χ² test, where significance was set at P < 0·100 and quantified by measuring I². An I² value above 50 per cent indicated significant statistical heterogeneity. The random-effects model was employed when there was significant heterogeneity.

Results

Four randomized controlled trials were identified and included in the meta-analysis (Fig.1).

Figure 1. Flow chart of study selection according to PRISMA guidelines. SEMS, self-expanding metallic stents; RCT, randomized clinical trial

Alcántara and colleagues11 compared the use of SEMS as a bridge to a planned elective one-stage procedure versus emergency one-stage open surgery with intraoperative colonic lavage. Cheung and co-workers12 compared the use of SEMS as a bridge to elective laparoscopic surgery with emergency open surgery; the choice of type of surgery was left to the surgical team. Pirlet et al.13 and van Hooft and colleagues14 described multicentre trials comparing SEMS as a bridge to open surgery with emergency open surgery; again, the type of operation was left to the surgical team. The characteristics of these trials are shown in Table1. All analyses were performed on an intention-to-treat principle. Assessment of risk of bias is summarized in Table2.

Table 1. Details of randomized clinical trials included in meta-analysis

						Findings
Reference	Year	Type of study	Intervention	No. of patients (SEMS:surgery)	Type of stent	Significant difference	No significant difference	Notes
* Stent placement was successful in 33 of the 47 patients: 27 had one stent and six had two stents, giving a total of 39 stents. SEMS, self-expanding metallic stents; RCT, randomized clinical trial; IOCL, intraoperative colonic lavage.
Alcántara	2011	RCT	SEMS and planned 1-stage open surgery versus emergency 1-stage open surgery and	15:13	Wallstent^® (Boston Scientific, Natick, Massachusetts, USA)	SEMS: reduced overall morbidity and anastomotic leak	SSI, hospital stay, mortality	Trial stopped as emergency surgery group had significantly increased rate of anastomotic leak
			IOCL					Time from SEMS to
			Duration: February 2004 to December 2006					surgery: 5–7 days
Cheung et al.12	2009	RCT	Endolaparoscopic versus emergency open surgery	24:24	Wallstent^®	SEMS: reduced blood loss, pain, wound infection, anastomotic		Time from SEMS to surgery: < 2 weeks
			Duration: January 2002 to May 2005			leak rates, stoma rate; increased 1-stage procedure rate
Pirlet et al.13	2011	Multicentre RCT	SEMS and open surgery versus emergency open surgery	30:30	Nitinol (C. R. Bard, Voisins le Bretonneux, France)		Stoma, colonic resection, in-hospital mortality,	Trial stopped owing to 3 colonic perforations during stent placement and
			Duration: December 2002 to October 2006				surgical and medical morbidity	high rate of technical failure of stent placement (16 of 30)
							rates	Time from SEMS to surgery: not given
van Hooft et al.14	2011	Multicentre RCT	SEMS and open surgery versus emergency open surgery	47:51	31 Wallstent^®; 8 Wallflex^® (Boston Scientific)*	SEMS: reduced initial stoma rates; increased stoma-related	Mean global health status, mortality,	Trial stopped as SEMS group had increased absolute risk of 30-day morbidity
			Duration: March 2007 to August 2009			problems	morbidity, stoma rates	(0·19 in first 60 patients; 0·19 in first 90 patients) on interim analysis
								Time from SEMS to surgery: < 4 weeks

Table 2. Risk of bias summary for the four randomized clinical trials

	Alcántara et al.11	Cheung et al.12	Pirlet et al.13	van Hooft et al.14
+, Low risk of bias; ± , unclear risk of bias; −, high risk of bias.
Random sequence generation (selection bias)	+	+	+	+
Allocation concealment (selection bias)	+	+	+	+
Blinding of participants and personnel (performance bias)	±	±	±	+
Blinding of outcome assessment (detection bias)	−	−	−	+
Incomplete outcome data (attrition bias)	+	+	+	+
Selective reporting (reporting bias)	+	+	+	+
Other bias	+	+	+	+

Some 234 patients (118 men and 116 women) were included in the meta-analysis: 116 patients in the SEMS group and 118 in the emergency surgery group. Inclusion criteria for the studies were adult patients aged over 18 years, clinical features of left colonic obstruction confirmed radiologically, obstruction at or distal to the splenic flexure, and cause of obstruction a malignant tumour. Exclusion criteria were obstruction with a non-malignant cause, signs of peritonitis and perforation, and obstruction proximal to the splenic flexure.

Stent efficacy and safety

Stent placement in all four RCTs was performed under radiological guidance or endoscopically. Clinical success of stent decompression was defined as evidence of intestinal transit or passing flatus or stool within 24 h in one study12, within 48 h in another11, and within 3 days in the two other RCTs13, 14. Radiological investigations (serial abdominal X-rays11, 12, 14 and water-soluble contrast enema13) were used to confirm successful decompression and correct deployment of stents in all four RCTs.

In terms of efficacy of SEMS placement, technical success was achieved in 82 (70·7 per cent) of the 116 patients, with clinical success in 80 (69·0 per cent). In terms of safety of SEMS placement, clinical perforation occurred in eight patients (6·9 per cent). Two RCTs13, 14 described histological examination of resected specimens and reported silent perforation in 14 per cent (11 of 77).

Primary outcomes

Primary anastomosis

Primary anastomosis rates were determined to investigate whether SEMS as a bridge to elective surgery increases the likelihood of having a one-stage procedure (RR greater than 1 indicates that SEMS has a positive effect on the primary anastomosis rate). All four RCTs reported the primary anastomosis rate. Meta-analysis showed a significant difference in favour of the SEMS group using a fixed-effect model (RR 1·46, 95 per cent c.i. 1·17 to 1·82; P < 0·001). However, there was no significant difference when the random-effects model was used (RR 1·40, 0·84 to 2·35; P = 0·20). There was considerable heterogeneity between the studies (I² = 87 per cent) (Fig.2).

Figure 2. Meta-analysis of primary anastomosis rates using a fixed-effect and b random-effects Mantel–Haenszel models. Risk ratios are shown with 95 per cent confidence intervals. SEMS, self-expanding metallic stents

Data on rates of successful primary anastomosis were extracted from all four RCTs. Meta-analysis showed a significant difference in the overall successful primary anastomosis rate in favour of the SEMS group using the fixed-effect model (RR 1·58, 95 per cent c.i. 1·22 to 2·04; P < 0·001) (Fig.3). The random-effects model gave similar results and there was no heterogeneity between studies (I² = 0 per cent).

Figure 3. Meta-analysis of successful primary anastomosis rates using a Mantel–Haenszel fixed-effect model. Risk ratios are shown with 95 per cent confidence intervals. SEMS, self-expanding metallic stents

Stoma rates

Overall and permanent stoma rates were reported by all four RCTs. Meta-analysis of overall rates showed a significant difference in favour of the SEMS group (RR 0·71, 95 per cent c.i. 0·56 to 0·89; P = 0·004) (Fig.4a). The random-effects model gave similar results, with no heterogeneity between studies (I² = 0 per cent). Meta-analysis of permanent stoma rates showed no significant difference between the two interventions (RR 0·75, 0·55 to 1·01; P = 0·06); there was moderate heterogeneity (I² = 47 per cent) (Fig.4b).

Figure 4. Meta-analysis of a overall and b permanent stoma rates using a Mantel–Haenszel fixed-effect model. Risk ratios are shown with 95 per cent confidence intervals. SEMS, self-expanding metallic stents

In-hospital mortality

The in-hospital mortality rate, reported by all four RCTs, was 6·9 per cent (8 of 116 patients) in the SEMS group versus 5·9 per cent (7 of 118) in the emergency surgery group. This difference was not statistically significant (RR 1·17, 95 per cent c.i. 0·46 to 2·99; P = 0·74). There was no heterogeneity between studies (I² = 0 per cent).

Secondary outcomes

Anastomotic leak

Anastomotic leak rates were reported by all four RCTs. Meta-analysis showed no significant difference for both fixed-effect and random-effects models. Owing to substantial heterogeneity (I² = 51 per cent), the result of the random-effects model was used (RR 0·72, 95 per cent c.i. 0·13 to 4·00; P = 0·71).

Thirty-day reoperation rate

Meta-analysis of 30-day reoperation rates, also reported by all four RCTs, showed no significant difference in either the fixed-effect or random-effects model. As there was substantial heterogeneity (I² = 54 per cent), the random-effects model was used (RR 0·82, 95 per cent c.i. 0·15 to 4·57; P = 0·82).

Surgical-site infection

All four RCTs reported SSI rates: 12·9 per cent (15 of 116) for the SEMS group versus 22·9 per cent (27 of 118) following emergency surgery. There was no significant difference (overall RR 0·56, 95 per cent c.i. 0·31 to 0·99; P = 0·05) and no significant heterogeneity (I² = 33 per cent) between the studies.

Discussion

The overall technical (70·7 per cent) and clinical (69·0 per cent) success rates for stent placement were much lower in this review than those reported in the literature15–17. Sebastian and colleagues16 performed a systematic review using 21 case series and reported a technical success rate of 91·9 (range 33–100) per cent and a clinical success rate of 71·7 (45–84) per cent for SEMS placement as a bridge to elective surgery. The authors of the two RCTs included in this review with low technical and clinical success rates (47 and 40 per cent respectively for Pirlet et al.13, and 70 per cent for van Hooft et al.14) cited the high number of patients with complete colonic obstruction as the reason for these low rates. van Hooft and colleagues14 reported that 70 per cent of patients had complete obstruction—much higher than published data of about 50 per cent18. Small and co-workers18 have suggested that the degree of occlusion is one of the two most important risk factors for SEMS complications as the completely occluded bowel may result in friable microperforated tissue and present as a very tight stricture that makes stent deployment technically difficult.

Three of the four trials were stopped prematurely. Alcántara and colleagues11 found that overall morbidity, particularly the rate of anastomotic leak, in the emergency surgery group was significantly greater than in patients having SEMS as a bridge to surgery. In contrast, Pirlet and co-workers13 stopped their trial after finding a high number of stent-related complications. They had a 53 per cent technical failure rate for stent insertion; there were also two cases of stent perforation in the SEMS group and one perforation in a non-randomized patient leading to closure of the trial. van Hooft et al.14 also stopped their trial after interim analysis found an unexpectedly increased absolute risk of 30-day morbidity in the SEMS group.

The clinical stent perforation rate in the present meta-analysis was 6·9 per cent. This was higher than in previous systematic reviews15–17 that cited a perforation rate of 3–5 per cent, although these reviews used data from case reports, case series and non-randomized studies. Silent perforations could have oncological significance, potentially resulting in tumour cell seeding and dissemination. A recent retrospective study19 comparing oncological outcomes of SEMS as a bridge to surgery in left-sided large bowel obstruction with elective non-obstructing surgery found that SEMS had an adverse effect on 5-year overall and disease-free survival rates. However, the poorer outcomes in this group could have been due to patients with more advanced disease presenting with emergency bowel obstruction3. van Hooft and colleagues14 reported that stent placement was performed by experienced endoscopists who had placed more than 20 enteral stents, including at least ten colonic stents. The other three RCTs did not report on the level of experience of stent placement required. Operator experience and technical expertise in stent placement has been shown to reduce significantly the number of stent-related complications18.

The most common surgical procedure performed for emergency left-sided large bowel obstruction is resection and end colostomy20. Intestinal continuity can be restored later, but this requires a second major operation and is performed in only 60 per cent of cases with up to 40 per cent morbidity21, 22. Primary resection and anastomosis has the advantage of being a definitive surgical procedure, although there is a higher risk of anastomotic leak from an obstructed bowel in a critically ill patient23. Hence, preoperative placement of SEMS decompresses the colon, enabling patient optimization and allowing primary resection and anastomosis. This meta-analysis did not show any significant difference in overall primary anastomosis rates between the two interventions, although there was a trend favouring the SEMS group. However, there was a significant difference for overall successful primary anastomosis in favour of the SEMS group. This supports the hypothesis that preoperative stenting of the obstructed colon can enable conversion of an emergency procedure to an elective one, creating the right conditions for successful primary anastomosis and hence increasing the likelihood of having a one-stage procedure.

This meta-analysis found a significantly lower overall stoma rate in favour of the SEMS group after surgery for acute large bowel obstruction. However, this effect was only short-lived as in the longer term there was no significant difference in the permanent stoma rate. Evaluation of the stoma rate is essential when comparing the two different interventions, because creation of a stoma can have profound effects on the morbidity and quality of life of the patients24. Frequently, a stoma has a negative impact on the psychosocial well-being of patients and specialized counselling is needed to improve quality of life significantly25. Many stomas formed after emergency surgery are also never reversed26.

There was no difference for in-hospital mortality between the groups in this meta-analysis, in contrast to previous findings3, 5, 7 showing a significant difference in postoperative mortality after elective and emergency surgery for colorectal carcinoma. There was also no significant difference in other known surgical complications of anastomotic leak, reoperation and SSI in this meta-analysis.

Cost considerations are important when evaluating the two different intervention strategies. When Alcántara and colleagues11 evaluated cost they found no significant difference in the combined cost of surgery and hospitalization between the two intervention groups. However, the total costs, including materials such as stents and intraoperative colonic lavage equipment, were significantly higher for the SEMS group than for the emergency surgery group.

Acknowledgements

This paper is based on a dissertation submitted by C.J.T. for an MSc in Surgical Technology at Imperial College London.

Disclosure: The authors declare no conflict of interest.