Meta-analysis: effect of preoperative infliximab use on early postoperative complications in patients with ulcerative colitis undergoing abdominal surgery

Authors


  • As part of AP&T's peer-review process, a technical check of this meta-analysis was performed by Dr P. Collins.

Correspondence to:

Prof. K. Wu or Prof. D. Fan, 127 West Changle Road, Xi'an 710032, China.

E-mails: kaicwu@fmmu.edu.cn; fandaim@fmmu.edu.cn

Summary

Background

Infliximab is widely used in severe and refractory ulcerative colitis (UC). The results of clinical studies are inconsistent on whether preoperative infliximab use increases early postoperative complications in UC patients.

Aim

To determine the clinical safety and efficacy of preoperative infliximab treatment in UC patients with regard to short-term outcomes following abdominal surgery.

Methods

PubMed, Embase databases were searched for controlled observational studies comparing postsurgical morbidity in UC patients receiving infliximab preoperatively with those not on infliximab. The primary endpoint was total complication rate. Secondary endpoints included the rate of infectious and non-infectious complications. We calculated pooled odds ratios (ORs) with 95% confidence intervals (CIs) as summary measures.

Results

A total of 13 studies involving 2933 patients were included in our meta-analysis. There was no significant association between infliximab therapy preoperatively and total (OR = 1.09, 95% CI: 0.87–1.37, P = 0.47), infectious (OR = 1.10, 95% CI: 0.51–2.38, P = 0.81) and non-infectious (OR = 1.10, 95% CI: 0.76–1.59, P = 0.61) postoperative complications respectively. Infliximab might be a protective factor against infection for the use within 12 weeks prior to surgery (OR = 0.43, 95% CI: 0.22–0.83, P = 0.01). No publication bias was found.

Conclusion

Preoperative infliximab use does not increase the risk of early postoperative complications in patients with ulcerative colitis undergoing abdominal surgery.

Introduction

The anti-TNF-α agent infliximab has become an important step forward in the management of refractory inflammatory bowel disease (IBD) during the past decade.[1] However, the expanding use of infliximab has not reduced the need for operation for patients with IBD.[2, 3] The safety of infliximab in the peri-operative setting is a major cause for concern. Although we conducted a meta-analysis of five studies with similar topic previously, little conclusive evidence was obtained.[4] It is still controversial with regard to preoperative infliximab treatment among surgeons and gastroenterologists.[5] Recently, a meta-analysis of eight studies in patients with Crohn's disease (CD) indicated that preoperative infliximab treatment was associated with an increased risk of postoperative infectious complications, and a trend towards an increased risk of non-infectious and overall complications.[6] Therefore, we decided to perform an updated meta-analysis of the existing observational studies to determine the relationship between preoperative infliximab use and early postoperative complications in patients with ulcerative colitis (UC) undergoing abdominal surgery.

Materials and Methods

Search strategy

We searched PubMed and Embase databases using the terms ‘(infliximab or anti-TNF-α or TNF-α inhibitor or TNF-α antagonist) and (postoperative or surgical complications) and (inflammatory bowel disease or ulcerative colitis or Crohn's disease)'. The computer search was supplemented with a manual search of reference lists for all available review articles and primary studies. We performed the final search on 3 August, 2012.

Study selection

Two investigators (Z. Y. and Q. W.) independently screened the titles and abstracts of studies to identify those that fulfilled the inclusion criteria: (i) population: patients with UC undergoing abdominal operation; (ii) intervention: infliximab treatment prior to surgery; (iii) comparator: controls not receiving infliximab preoperatively; (iv) outcomes: the primary outcome was the total complication rate in a short term after surgery (usually 30 days). Secondary outcomes were the rate of infectious and non-infectious complications in the above period. We summarised a list of the definition of infectious and non-infectious complications in each study in Supplemental Table 1. We did not include information solely available in abstract form, because the amount of provided data was usually not sufficient for performing the necessary analyses, and evaluating the studies' scientific quality properly. All studies included should have sufficient information to calculate the unadjusted odds ratio (OR) and 95% CI.

Data extraction and quality assessment

Two independent reviewers (Q. W. and F. W.) extracted data and assessed the quality of the selected studies in a prespecified form (Table S2); disagreements were resolved by discussion with a third reviewer (K. W.). The following data were collected: country of origin, years of study and year of publication, duration of infliximab use before surgery, duration of follow-up after surgery, number of patients, demographic characteristics (gender, age at surgery, body mass index, smoking status), disease characteristics (duration, severity, concomitant medications) and surgical characteristics (indication, type and approach of operation). The quality of the study reports was evaluated using the Newcastle-Ottawa scale (NOS), which was developed to assess the quality of nonrandomised studies with its design, content and ease-of-use directed to the task of incorporating the quality assessment in the interpretation of meta-analytic results.[7] The full score of NOS is 9. Study with 7 or above on the NOS was considered as high quality.

Statistical analysis

Pooled ORs and 95% CIs were calculated for evaluating the risk of postoperative complications in patients with UC on infliximab using a fixed-effect model. We assessed heterogeneity using the Chi-squared-based Q-test and the I2 measure of inconsistency. If significant heterogeneity had a Q-test P-value <0.10 or I2 >50%, a random-effect model was replaced. Sensitivity analyses were carried out without the less-qualified studies or the study, which has shown the most significant difference accordingly. Publication bias was examined using the Egger's test as well as the funnel plots. All analyses listed above were conducted using the software Reviewer Manager (version 5.1. Copenhagen, Denmark: The Nordic Cochrane Centre, The Cochrane Collaboration, 2011) and stata (version 12.0; Stata Corporation, College Station, TX, USA).

Results

Search results

Our search retrieved 434 potentially relevant citations, of which 408 were discarded after title and abstract screening. Twenty-six articles were retained for full-text review. Thirteen studies were excluded for not addressing data on a UC population alone. In particular, four studies provided their data for a mixed cohort of IBD patients that included both UC and CD patients.[8-11] We contacted the corresponding authors of these studies to access the data in their UC patients separately. However, no response was received all the while. Finally, 13 observational studies with 2933 patients (median: 81; range: 21–1226) were included in the meta-analysis (Figure 1).[12-24] Total complication rates could be accessed in 12 of these studies. Infectious and non-infectious complications were assessed in nine and eight studies respectively.

Figure 1.

Study flow diagram.

Study and patient characteristics

The characteristics of studies and patients are presented in Table S2. Seven studies came from North America, whereas the others were from Europe. In all, 516 out of 2933 (17.6%) patients included in our meta-analysis received infliximab and 2417 (82.4%) did not. Only one study included a minority of patients treated with other anti-TNF-α agents (32/199, 16.1%).[23] Infliximab was given within 12 weeks prior to surgery in all the patients of seven studies.[12, 16, 18, 20-23] It is generally believed that 12-week is chosen as cutoff point to assess the washout effect for infliximab before surgery based on its half-life.[25, 26] The duration of postoperative follow-up was reported at 30 days in 10 studies[13-18, 20, 22-24] and at 60 days in one study,[21] and without available information in two studies.[12, 19]

There were 1557 (53.5%) males from the available data about the gender in 12 studies. Two studies compared outcomes solely in paediatric patients (≤18 years).[20, 21] Although data on the surgical characteristics were available in most of the included studies, we could not conduct a meta-analysis for specific operative interventions due to the variability of definitions employed by the studies.

Data on concomitant medications, stratified by infliximab treatment, were available in 10 studies.[13-18, 20, 22-24] Corticosteroids were equi-frequently used between infliximab and non-infliximab groups (OR = 1.18, 95% CI: 0.95–1.45, 10 studies, 2199 patients). Patients in the infliximab group had a more frequent use of 5-aminosalicylates (OR = 1.91, 95% CI: 1.21–3.01, 4 studies, 482 patients) and azathioprine (OR = 2.80, 95% CI: 1.76–4.45, 4 studies, 446 patients), but a less frequent use of cyclosporine A (OR = 0.33, 95% CI: 0.18–0.59, 5 studies, 471 patients).

Given the nature of all included studies were retrospective, the major limitation of these studies was the incomparability between cases and controls (Table S2). Only 7 out of 13 studies were considered as high-quality ones based on the NOS.[14-17, 22-24]

Postoperative complications

Figure 2 displays the pooled outcome for development of early total postoperative complications in patients exposed to infliximab preoperatively vs. controls (494 vs. 2298). The pooled OR was 1.09 (95% CI: 0.87–1.37; Pheterogeneity = 0.17, I2 = 28%), suggesting that there was no significant overall association between preoperative infliximab therapy and total complications.

Figure 2.

Pooled outcome for development of total complications in patients exposed to infliximab preoperatively vs. controls.

Figure 3 displays the pooled outcome for development of early infectious postoperative complications in patients exposed to infliximab preoperatively vs. controls (265 vs. 721). The pooled OR was 1.10 (95% CI: 0.51–2.38; Pheterogeneity = 0.002, I2 = 67%), suggesting that there was no significant association of preoperative infliximab therapy and infectious complications, but a significant heterogeneity among studies.

Figure 3.

Pooled outcome for development of infectious complications in patients exposed to infliximab preoperatively vs. controls.

Figure 4 displays the pooled outcome for development of early non-infectious postoperative complications in patients exposed to infliximab preoperatively vs. controls (243 vs. 602). The pooled OR was 1.10 (95% CI: 0.76–1.59; Pheterogeneity = 0.18, I2 = 31%), suggesting that there was no significant association of preoperative infliximab therapy and non-infectious complications.

Figure 4.

Pooled outcome for development of non-infectious complications in patients exposed to infliximab preoperatively vs. controls.

Sensitivity analysis and publication bias

When the only one study with a few patients treated with other anti-TNF antibodies was excluded, no significant association for the risk of total complications was found either (OR = 1.25, 95% CI: 0.93–1.68, P = 0.15, 11 studies). When we just included the studies with infliximab use in 12 weeks prior to operation, infliximab was not associated with an increased risk of total (OR = 0.89, 95% CI: 0.66–1.20, P = 0.45, six studies) and non-infectious (OR = 1.16, 95% CI: 0.62–2.18, P = 0.65, three studies) postoperative complications, but a significantly decreased risk for infectious complications (OR = 0.43, 95% CI: 0.22–0.83, P = 0.01, four studies). The pooled OR did not reach statistical significance in the analyses of total, infectious and non-infectious complications, respectively, after the studies without reporting complications for 30 postoperative days were removed from the meta-analysis, and also after the studies dealing with children population solely were excluded (data not shown). When low-quality studies (NOS < 7) were excluded, there was still no significant difference between infliximab and control groups in total, infectious and non-infectious complication rates respectively.

No publication bias was discovered in comparing total, infectious and non-infectious postoperative complications, respectively, based on the symmetry of the funnel plots and the P-values of Egger's linear regression approach (Figure 5).

Figure 5.

Funnel plots of published studies for detecting publication bias.

Discussion

This meta-analysis showed there was no significant disparity in the risk of early postoperative complications following abdominal surgery between UC patients exposed to infliximab preoperatively and controls. Infliximab might be a protective factor against infection when we took its durability of biologic activity (12 weeks) into account in the subgroup analysis. However, our results need to be interpreted with caution due to several limitations.

First, because individuals were not randomly allocated to treatment, it was impossible to fully estimate the influence of confounding variables. Many potential confounders could not be controlled ideally in these observational studies (Table S2). For example, disease severity, nutritional deficiency, disease duration prior to operation, body mass index and individual surgeon etc. might be predictive of developing postsurgical morbidity.[18, 27] However, the absence of randomized controlled trials could be explained by the fact that it remains unethical to perform such a study.[28]

Secondly, there was a significant heterogeneity when combining these studies with regard to the infectious complications. The high heterogeneity in the synthesis could be partly due to the minor differences on the definition of infectious and non-infectious complications between studies (Table S1).

Thirdly, some important surgical parameters such as timing (urgent or elective), type (one-stage, two-stage, or three-stage) and approach (open or laparoscopic) of the operative procedures needed to be concerned in the context of postsurgical morbidity in UC patients.[4, 10] There could be variations between the studies on the criteria of patient selection for each surgery and surgical technique used would vary between the studies. These patients would therefore have their own individual risk profiles among the different studies.

Fourthly, another important limitation in our meta-analysis was the concomitant use of non-infliximab immunomodulators. Patients treated with infliximab were more likely to be treated with azathioprine and 5-aminosalicylates, but less likely to receive ciclosporin. It is not unreasonable to assume that causality of the increased postoperative complications might be owing to the combination or multitude of medications rather than infliximab alone.[20, 29] Although corticosteroids have been proved to increase the likelihood of surgical complications in patients with IBD,[28] the usage frequency was almost equal between the infliximab population and controls in our meta-analysis (54% vs. 50%). In addition, patients on infliximab treated with more concomitant medications might indicate that they had more severe disease. Thus, any potential adverse effects detected in the patients with infliximab undergoing surgery might be related to the fact that they were sicker.[20, 22] However, it was impossible to stratified the meta-analysis results for these vital confounders because of insufficient data.

Finally, all the studies included in this meta-analysis were from Western countries without any appropriate published data from other regions. It is well known that UC is more and more prevalent and highly incident than before in the Far East as a result of the westernised environment and lifestyle in this area.[30]

Despite the above-mentioned limitations, this updated meta-analysis is more powerful than the study we performed before. The current synthesis included 13 individual studies with a large number of target population (2933 patients). We did sensitivity analyses as possible as we could by excluding those studies, which would affect the pooled results mostly. It is noteworthy that the outcomes of the different subgroups were much consistent with the overall association in statistics.

Why there are different results for UC vs. CD in terms of postoperative infection, although the regimens of infliximab treatment for the two types of IBD are similar? In our opinion, the extent of aggressive medical therapy in an UC patient is often less than in a CD patient, as the former disease has a surgical cure, which often prompts earlier surgical consultation, and which prevents the degree of medical ramping that might occur when there is no cure whatsoever, as in CD. The complication rates for UC patients may be different as, for example, the number of surgeries for this phenotype are smaller, and very often involve an end or a loop ileostomy.[5, 31] In addition, when low-quality studies in the systematic review by Kopylov et al. were excluded, the pooled OR for infectious complications did not reach statistical significance. It is still worth discussing whether infliximab will increase the postoperative infection in CD patients.

In conclusion, preoperative infliximab use does not increase the risk of early postoperative complications in patients with UC undergoing abdominal surgery. The current practice of some doctors to delay operation and to discontinue infliximab for 8–12 weeks prior to elective surgery in UC patients may not be warranted. A globally multicentre prospective study is required to confirm the findings of this meta-analysis.

Acknowledgements

Declaration of personal interests: None. Declaration of funding interests: This study was funded in part by the National Natural Science Foundation of China (grant no. 81172062 and 81000988). The funding source had no role in the design and conduct of the study; collection, analysis, and interpretation of the data; preparation or review of the manuscript; or the decision to submit the manuscript for publication.

Ancillary