Aliment Pharmacol Ther 31, 486–492
Background Infliximab was approved for use in ulcerative colitis in recent years. It has been debated if infliximab increases the risk of post-operative complications in patients with ulcerative colitis.
Aim To perform a meta-analysis that examines the relationship between preoperative infliximab treatment and short-term post-operative complications in patients with ulcerative colitis.
Methods We searched the PubMed and MEDLINE databases to identify observational studies on the impact of pre-operative infliximab use on short-term post-operative complications in ulcerative colitis. Infectious complications mainly included wound infection, sepsis and abscess, whereas non-infectious complications included intestinal obstruction, thromboembolism and gastrointestinal haemorrhage. Pooled odds ratios (ORs) were calculated for each relationship.
Results A total of 5 studies and 706 patients were included in our meta-analysis. Overall, we did not find a strong association between pre-operative treatment of infliximab and short-term infectious [OR 2.24, 95% confidence interval (CI) 0.63–7.95] or non-infectious (OR 0.85, 95% CI 0.50–1.45) post-operative complications in ulcerative colitis patients. On the contrary, we discovered that pre-operative infliximab use increased short-term total post-operative complications (OR 1.80, 95% CI 1.12–2.87).
Conclusions Pre-operative infliximab use increased the risk of short-term post-operative complications. Subgroup analysis is underpowered to assess the nature of these complications but shows a trend towards increased post-operative infection.
Ulcerative colitis (UC) is a chronic inflammatory bowel disease (IBD) that diffusely affects the superficial mucosa of the colon and is characterized by recurrent episodes of inflammation.1 About 10–40% of patients with UC will require surgery at some point during the course of their disease.2–4 However, abdominal surgery for UC is not without risk. Short-term post-operative morbidity following colectomy is in the order of 27% and includes abdominal wound infection, anastomotic leakage, pelvic sepsis and small bowel obstruction.5
The availability of the prototypical anti-TNF agent infliximab has offered an important advance in therapy for patients with IBD.6 The drug was first approved in 1998 for use in the treatment of Crohn’s disease (CD) by US FDA and thereafter in the EU in 1999.7 Infliximab has not become available for the treatment of UC until recent years.8 Prior research implied that infliximab use might not increase post-operative complications after abdominal surgery for CD.9–11 It has been controversial whether treatment with infliximab increases the risk of post-operative complications in UC patients undergoing subsequent colectomy.
The aim of our study was to determine if pre-operative infliximab use affects short-term surgical outcomes. A meta-analysis is a method of pooling data from multiple studies to draw more definitive conclusions from a body of research.12 Using the guidelines for the reporting of meta-analysis of observational studies, which were published by the Meta-analysis of Observational Studies in Epidemiology Group in 2000,13 we performed a meta-analysis of the existing observational studies that explored the association between pre-operative infliximab treatment and short-term post-operative complications in patients with UC.
We performed a search of the PubMed database for articles published from 1950 to September 2009. Keywords used were infliximab and (post-operative or surgical complications) and (UC or IBD). Other search strategy was using the keywords IBD and surgical risk. We subsequently repeated this search using the same criteria in the MEDLINE database (1950 to September 2009). If a study could not be included based on the title and/or abstract, the full-text article was reviewed. In addition, the reference lists of any studies meeting the inclusion criteria were reviewed manually to identify additional relevant publications.
Studies were included if they met the following criteria: (i) observational studies that evaluate the association between pre-operative infliximab treatment and short-term (within 30 days) post-operative complications in patients with UC; (ii) studies that include a control group; (iii) studies that are published as a full article in the English language; and (iv) reported complication rates either as infectious complications, which mainly included wound infection, sepsis, abdominal abscess and peritonitis, or as total complications, which also included intestinal obstruction, thromboembolism and gastrointestinal haemorrhage.14 We summarized a list of the description and division of infectious and non-infectious complications in each study (Table 1). If the odds ratio (OR) for risk of complications associated with infliximab was not reported in the study, this was calculated using spss 12.0 (SPSS Inc., Chicago, IL, USA) from the data given.
|Reference||Infectious complications||Non-infectious complications||Baseline characteristics (Ifx group vs. non-Ifx group)|
|Jarnerot et al.23||Re-operation because of septic complications, high fever||Rectal bleeding, ileus||n.a.||n.a.|
|Selvasekar et al.18||Pouch-specific complications (anastomotic leak and pelvic abscess), wound infections (superficial and deep)||n.a.||Gender, body mass index, Extent of disease, operation procedure||Median age (younger in Ifx group), Severity (more severe colitis in Ifx group), fulminant colitis (none in Ifx group, 9.5% in non-Ifx group), indication for operation (more patients for intractable disease in Ifx group), pre-operative medications (higher per cent of treatment with CS, 5-ASA, AZA or combination of these drugs in Ifx group)|
|Schluender et al.19||Superficial wound infection, large peristomal abscess, infected pyoderma gangrenosum, pneumonia||Ileus, bleeding, rectal stump leak, thrombosis, acute renal failure, cerebrovascular accident, dehydration, urinary retention||Gender, median age, median duration of disease, extent of disease, pre-operative medical treatment (CS, CsA)||Pre-operative medical treatment (higher rate of treatment with MP or 6-TG in Ifx group)|
|Mor et al.20||Sepsis||Leak, haemorrhage, thrombotic event, ileus||Age, gender, indication for pouch, extent of colitis, severity, pre-operative medications (CS, CsA, MP)||n.a.|
|Ferrante et al.22||Pouch-specific complications (anastomotic leak and pelvic abscess), surgical site infectious complications (pouch-specific complications and wound infections), nonsurgical site infectious complications (respiratory or urinary tract infections)||n.a.||Gender, diagnosis, extent of disease, extraintestinal manifestations, smoking behaviour, other medical therapy within 12 weeks prior to surgery, body mass index, indication for colectomy||Median age at primary surgery (younger in Ifx group), median disease duration prior to primary surgery (shorter in Ifx group), median C-reactive protein level at primary surgery (higher in Ifx group), type of primary surgery, stages of surgery|
The following information was collected in a predefined data extraction form (Table 2): author, country where the study was conducted, duration of the study, year of publication, duration of follow-up after operation, number of patients with UC studied, concomitant medication with infliximab prior to surgery and short-term complication rates in patients receiving infliximab therapy and those not treated with this drug. Data were abstracted by two independent investigators (Z.Y. and Q.W.). The discrepancies between the two abstractors were resolved by discussion, and if necessary, by re-evaluation by the senior author (K.W.).
|Author||Country||Study years (publication year)||Condition||Duration of follow-up (days)||n||No. receiving Ifx||No. not receiving Ifx||Concomitant medication||Infectious complications||Non-infectious complications||Total complications|
|Ifx||No Ifx||OR||Ifx||No Ifx||OR||Ifx||No Ifx||OR|
|Jarnerot et al.23||Sweden and Denmark||2001–2004 (2005)||Severe to moderately severe UC||Short-term||21||7||14||CS||n.a.||n.a.||n.a.||n.a.||n.a.||n.a.||2(29)||4(29)||1.00 (0.13–7.45)|
|Selvasekar et al.18||US||2002–2005 (2007)||Chronic UC||30||301||47||254||CS, 5-ASA, AZA||13(28)||25(10)||3.50 (1.64–7.50)||16(34)||99(39)||0.81 (0.42–1.55)||29(62)||124(49)||1.69 (0.89–3.20)|
|Schluender et al.19||US||2000–2005 (2007)||Refractory UC||30||151||17||134||CS, 6-MP, 6-TG, CsA||3(18)||11(8)||2.40 (0.60–9.63)||3(18)||26(19)||0.89 (0.24–3.33)||6(35)||37(28)||1.43 (0.49–4.15)|
|Mor et al.20||US||2000–2006 (2008)||UC+IC||30||92||46||46||CS, CsA, MP||10(22)||1(2)||13.8* (1.82–105)||6(13)||6(13)||1.00 (0.30–3.37)||16(35)||7(15)||3.54* (1.51–8.31)|
|Kunitake et al.21||US||1993–2007 (2008)||188 CD + 156 UC + 69 IC||30||413||101||312||CS, AZA, MP||6(6)||32(10)||0.55 (0.22–1.36)||11(11)||17(5)||2.12 (0.96–4.69)||17(17)||49(16)||1.09 (0.59–1.99)|
|Ferrante et al.22||Belgium||1998–2008 (2009)||134 UC + 7 IBDU||30||141||22||119||CS, CsA, AZA, MP, MTX||2(9)||29(24)||0.31 (0.07–1.41)||n.a.||n.a.||n.a.||n.a.||n.a.||n.a.|
Quality assessment of studies was carried out independently by two reviewers (K.W. and D.F.). Discrepancies in the interpretation were resolved by consensus. The studies that met the following predetermined criteria had high quality: (i) The recruited patients in the infliximab and the control group came from the same period. (ii) There were little significant differences in demographic or clinical variables between the infliximab and the control group (Table 1).
Meta-analysis was performed by combining the ORs of the individual studies in a global OR, using a random effects model (DerSimonian and Laird). This model does not assume homogeneity among studies in terms of methodological or clinical characteristics, and is overall a more conservative approach than the fixed effects model. Significance and 95% confidence interval (CI) were provided for the combined OR.
We tested for heterogeneity among studies using the chi-squared test as well as the I2 test. The I2 test describes the percentage of the variability in the estimates of effect that is caused by heterogeneity rather than chance, with a value greater than 50% being considered having substantial heterogeneity. We also performed the Begg and Egger tests to evaluate for the presence of publication bias.12 All analyses were conducted using StataIC 10 (Stata Corp., College Station, TX, USA).
Our search strategy in the PubMed database yielded 150 articles, of which 142 were excluded on the basis of the title and abstract. Alternative keyword search in PubMed yielded no additional articles. A further three articles were excluded after careful review of the full text. One of the three works was a comment on a relevant article.15 There were no control groups in the other two articles.16, 17 A comprehensive review of the reference lists of all the articles from the previous searches resulted in the identification of one further study. A total of six studies were finally included.18–23 Our duplicate search of the MEDLINE database did not yield any additional studies for inclusion. Kunitake et al.21 provided their data on post-operative complications in a mixed cohort of IBD patients treated with infliximab before surgery. We contacted the first and corresponding authors in order to get the complication rate of UC cohort in their study. No response was received from both of them. So the remaining five studies were included in our meta-analysis (four studies on infectious complications, three studies on non-infectious complications and four studies on total complications, see Table 2).
Short-term infectious post-operative complications
Four observational studies on pre-operative infliximab use and infectious post-operative complications included a total of 132 UC patients who received infliximab prior to operation and 553 controls. When results from all of the studies were combined, the pooled OR was found to be 2.24 (95% CI 0.63–7.95; test of heterogeneity P = 0.015, I2 = 71.4%; Figure 1a), suggesting that there is no significant overall association between pre-operative infliximab and infectious post-operative complications, but a significant heterogeneity among studies. There was no evidence of publication bias (Begg P-value = 1.000, Egger P-value = 0.822).
Short-term non-infectious post-operative complications
Three observational studies on pre-operative infliximab use and non-infectious post-operative complications included a total of 110 UC patients who received infliximab prior to operation and 434 controls. When results from all of the studies were combined, the pooled OR was found to be 0.85 (95% CI 0.50–1.45; test of heterogeneity P = 0.953, I2 = 0.0%; Figure 1b), suggesting that there is no significant overall association between pre-operative infliximab and non-infectious post-operative complications, and no significant heterogeneity among studies. There was no evidence of publication bias (Begg P-value = 0.602, Egger P-value = 0.337).
Short-term total post-operative complications
Four observational studies on pre-operative infliximab use and total post-operative complications included a total of 117 UC patients who received infliximab prior to operation and 448 controls. When results from all of the studies were combined, infliximab treatment was found to have increased the risk of total post-operative complications in patients with UC after abdominal surgery (pooled OR 1.80, 95% CI 1.12–2.87; test of heterogeneity P = 0.683, I2 = 0.0%; Figure 1c). There was no evidence of publication bias (Begg P-value = 0.174, Egger P-value = 0.733).
Several studies9–11 and a systematic review by Subramanian et al.14 demonstrated no significant positive association between perioperative infliximab use and post-operative complications in CD. In Kunitake’s research, CD patients made up almost half of the whole patients (45.5% CD, 37.8% UC, 16.7% IC); therefore, they also drew a negative conclusion.21 The overall pooled OR in our meta-analysis suggested that pre-operative treatment with infliximab increased the short-term total post-operative complications in UC. We did not find a strong association between pre-operative infliximab use and short-term infectious complications because of the small number of studies and high degree of heterogeneity among studies. We also did not find any association between pre-operative infliximab use and short-term non-infectious complications. However, the OR for short-term infectious complications was 2.24, whereas that for non-infectious complications was only 0.85. It might imply that most of the increase in post-operative complications was because of infectious complications.
Not all the confounding factors can be controlled in these observational studies. But some important parameters need to be considered in the context of post-operative complications in UC patients. Surgery-specific factors, such as indication, type and timing of operation, might influence outcome. Ferrante et al.22 reviewed all 141 patients with UC who underwent surgery in their tertiary referral centre in the past 10 years. They found a higher occurrence of short-term post-operative infectious complications after performing a restorative proctocolectomy without defunctioning ileostomy as primary surgery. In Mor’s study,20 a total of 523 restorative proctocolectomies were performed. In the infliximab group, there were 46 two-stage and 39 three-stage procedures. In patients who underwent two-stage restorative proctocolectomy, covariate-adjusted OR of early complication for the inflixmab group was 3.54 times that of controls (P = 0.004; 95% CI 1.51–8.31), and the OR of sepsis was 13.8 times greater for infliximab (P = 0.011; 95% CI 1.82–105). The OR of requirement for three-stage procedures was 2.07 times greater in the infliximab group (P = 0.011; 95% CI 1.18–3.63), which might affect the surgical outcomes.
In all the includable studies, patients with UC took a combined medication (corticosteroids, ciclosporin A or other immunomodulators) with infliximab prior to operation. Although Selvasekar et al.18 concluded that pre-operative infliximab use increased the risk of post-operative complications, the majority of infliximab-treated patients who came to surgery were also on a combination of high-dose corticosteroids and immunomodulators, in contrast to only a small proportion of non-infliximab-treated patients. Therefore, the study could not address whether infliximab itself contributed to the observed increase in post-operative complications or if it was a surrogate marker for patients with more aggressive disease.24 Ferrante et al.22 carried out a multivariate analysis to identify the factors associated with the occurrence of short-term infectious post-operative complications. The use of a moderate-to-high dose of corticosteroids was independently associated with short-term infectious post-operative complications overall (P = 0.003; OR 5.19; 95% CI 1.72–15.66). This finding was in harmony with previous research25, 26 and a meta-analysis by Subramanian et al.27 The results of the study by Schluender et al.19 demonstrated that pre-operative infliximab use alone in the treatment of medically refractory UC does not influence short-term morbidity after colectomy. However, combined infliximab and ciclosporin A use seemed to have both higher overall (P = 0.04) and infectious (P = 0.03) complication rates than ciclosporin A treatment alone.
In addition, not all the studies clarified the duration between last infusion of infliximab and primary surgery. Some patients in the infliximab group had received a last infusion of infliximab more than 12 weeks before surgery. A sustained biological effect of infliximab beyond 12 weeks seemed unlikely;28 therefore, we might doubt whether factors other than infliximab itself were associated with postsurgical morbidity.
In conclusion, pre-operative infliximab use increased the risk of short-term total post-operative complications. Subgroup analysis is underpowered to assess the nature of these complications, but shows a trend towards increased post-operative infection. However, these results need to be interpreted with caution. Prospective studies with larger sample size and longer follow-up are needed before definitive conclusions can be drawn.
Declaration of personal and funding interests: None.