Introduction

The anti-tumoral response of the innate immune system when triggered by endotoxins had already been recognized 100 years ago by William B. Coley. In 1962, O'Malley et al. proposed that the tumor necrotizing effect is mediated by a cytokine1 which was termed in 1975 by Dr. Lloyd J. Old as tumor necrosis factor (TNF).2 TNFα was finally cloned3 in 1984 and subsequent studies revealed that TNFα is the prototypic member of a cytokine superfamily, the TNF ligand family.4

TNFα is a key pro-inflammatory cytokine5 that exerts pleiotropically its robust immunostimulatory signaling and orchestrates the inflammatory process in many autoimmune diseases such as Crohn's disease (CD).6 , 7 TNFα has been targeted therapeutically, and anti-TNFα antibodies are currently available and result in its neutralization.8 Many studies have proven that anti-TNFα treatment is effective in the management of CD patients.9 – 12

Apart from the therapeutic benefits deriving from TNFα neutralization, amendments in the cellular functions of the immune system may induce immunosuppressive effects resulting in a potent inhibitory outcome on the wound healing process. This hypothesis raises great concerns about the safety of perioperative administration of anti-TNF. In view of this, several studies in the literature have attempted to address the issue of postoperative complications in patients with CD under perioperative anti-TNF treatment.

The aim of this review is to critically evaluate the literature concerning the risk of postoperative complications, resulting from an attenuated anastomotic healing, in anti-TNF-treated CD patients.

Methods

Search Strategy and Data Selection

The keywords for the literature review were arranged in three groups: the first group contained the key words “anti-TNF”, “Infliximab”, “Adalimumab” and “Cetrolizumab”; the second group consisted of the key words “anastomotic complications” and “postoperative complications”; the third group consisted of the words “Inflammatory Bowel Disease” and “Crohn’s Disease”. The search terms were structured by combining one word from each group, so as all possible combinations were accounted. This process rendered 16 search terms, all of which were looked for in titles and/or abstracts of English language papers appearing in PubMed database. The last search was performed in October 2013.

Study Selection

Studies were shortlisted for inclusion if they met a basic criterion: to compare the postoperative complications rate between anti-TNF treated CD patients and CD patients not receiving biologics peri-operatively.

We included clinical studies and meta-analyses that involved CD patients. Studies with patients receiving combined treatment with immunosuppressives such as azathioprine and/or steroids were included. All studies independently of the pre-operative anti-TNF administration protocol were included.

Studies in non-English language were excluded. In addition, studies not involving adult patients of CD and studies that incorporated both CD and ulcerative colitis (UC) patients without clearly distinguishing patient groups were excluded, as well.

Quality Assessment of Selected Studies

Methodological Index for Non-Randomized Studies (MINORS) is a validated and established index for evaluating the methodological quality of non-randomized studies. This index involves 12 criteria, eight of which have been designed for non-comparative studies, whereas the other four criteria apply to comparative studies. These criteria are scored in a scale as developed by Slim et al.13: 0 for not reported, 1 for reported but inadequate and 2 for reported and adequate. The maximum score for comparative studies is 24 while the minimum score is 0. Two reviewers (CZ and KN) independently evaluated each study according to MINORS index and scoring differences were discussed till consensus was reached. Studies achieving more than 16 points were defined as well-designed.14

Evaluation of “Anastomosis-Related Complications”

Data addressed to the impact of anti-TNF treatment on “anastomosis-related complications” such as anastomotic leak, re-operation rate, intraabdominal abscess formation and fistula formation rates, were retrieved from the selected studies and were critically evaluated according to the MINORS scores of the relevant studies.

Results

Characteristics of Selected Studies

The search process retrieved 262 citations. The title and abstract of each retrieved publication was reviewed and 225 studies were excluded. The full text was retrieved and further reviewed for the remaining 37 studies. Of these, 25 were further excluded from our review: three were incorporating immunosuppressive treatment other than anti-TNFs, 14 were non-clinical studies, one meta-analysis and seven studies including both UC and CD patients. Finally, ten15 – 24 original articles and two meta-analyses25 , 26 were included in our review. Figure 1 illustrates the flow diagram of our research and selection process.

Fig. 1
figure 1

Flow diagram of the research and selection process

Full size image

The characteristics of the ten studies meeting inclusion criteria are summarized in Table 1. Unfortunately, no prospective studies were found and all included studies are retrospective. Infliximab was the most common biological agent used in all studies and there were only two studies19 , 23 evaluating the effect of Adalimumab and Cetrolizumab in post-operative complication rates. The two meta-analyses had reviewed the literature until September 2011 (Kopylov et al.)25 and until October 2012 (Rosenfeld et al.),26 respectively, and as a result they do not include the two later studies that have been included in our review.23 , 24

Table 1 Characteristics and primary end-points of the selected studies
Full size table

Methodological Quality

The 12 MINORS criteria applied to all studies. The results of MINORS scoring system are illustrated in Table 2. According to our evaluation, five studies had a MINORS score over 16, while five studies had a score under this cut-off value.

Table 2 Εvaluation of the selected studies according to MINORS Criteria
Full size table

Apart from this, MINORS scores ranged from 9 to 21, demonstrating a significant amount of methodological heterogeneity among the studies contributing to the existing controversy in the literature on this matter.

Effect of Perioperative Anti-TNF Treatment on “Anastomosis-Related Complications”

The impact of peri-operative anti-TNF treatment in abscess formation, anastomotic leak, fistula formation and re-operation rate according to data of the selected studies are illustrated in Table 3. According to the study of Appau et al.18 (MINORS score 18), there was a significant difference in abscess formation, anastomotic leak and re-operation rate. All others showed non-significance, or did not separately evaluate anastomosis-related complications.

Table 3 The impact of anti-TNF treatment in post-operative “anastomosis-related” complications according to data from the selected studies
Full size table

The outcomes of well-designed studies, addressed to the anastomosis-related complications are summarized in Table 4. Four studies with MINORS score over 16 did not find a significant difference in the anastomotic related complications between groups, while only one well designed study found a significant difference.

Table 4 The outcomes of the selected studies, addressed to the anastomosis related complications classified accordingly to MINORS Index of each study
Full size table

Discussion

Peri-operative administration of biologics is a common scenario for patients with complicated or refractory to medication CD, who require surgical treatment. Surgical procedures may involve bowel resections followed by anastomosis of the remaining intestinal tract. A robust immune response to the surgical trauma induces a sound healing of intestinal anastomosis and subsequently an uncomplicated postoperative course. Instead, an attenuated immune response may compromise the intestinal healing ability leading thus to weaker anastomosis and increased risk of anastomotic leak and septic complications (Fig. 2).

Fig. 2
figure 2

Potent Impact of immunosuppression in Wound Healing Cascade

Full size image

However, the impact of anti-TNFs in the healing cascade is not confined to its potent immunosuppressive effect; instead, TNFα affects injury healing variously27 – 30 and in a conflicting manner31 not thoroughly clarified yet. To our knowledge, it is difficult to foresee the theoretical net effect of TNFα or anti-TNF on the healing cascade.

In clinical trials (Table 1), there is a great controversy regarding the overall effect of peri-operative anti-TNF treatment on postoperative complications rate: six studies find no postoperative clinical impact, three studies concluded that anti-TNF administration is associated with an increased risk of postoperative complications and Tay et al. study revealed that patients treated with immunomodulators developed fewer intra-abdominal septic complications. Partially controversial are the outcomes of the meta-analyses as well. Kopylov et al.25 demonstrated in their meta-analysis a significant association between peri-operative anti-TNF therapy and non-local postoperative infectious complications (odd ration [OR] 2.07, 95 % confidence interval [CI] 1.30–3.30, five studies, I 2 = 0 %), whereas Rosenfeld et al.26 found no significant differences in the major and minor complications rate between Infliximab and control groups.

In our point of view, it is crucial to clarify the association between anti-TNF therapy and postoperative complications resulting from an attenuated anastomotic healing such as: anastomotic leak, intraabdominal abscess and enterocutaneous fistula and re-operation rate. In the majority of the studies, “anastomosis-related complications” are not examined independently. Instead, they are classified heterogeneously with complications of different origin in groups such as: Infectious, septic and major or minor complications.

Kopylov et al. despite the mentioned difficulties in collecting data, performed a meta-analysis of anastomotic-related septic complications solely and revealed no significant association to perioperative Infliximab therapy (OR 1.18, 95 % CI 0.61–2.30, six studies, I 2 = 43,4 %). Further data addressing to “anastomosis-related complications” come from the recent study of Syed et al.23 According to this study, there is no significant difference between anti-TNFs treated patients and the control group in the rate of postoperative anastomotic leakage (9/150 vs. 9/175, p = 0.41), intra-abdominal abscess (21/150 vs. 18/175, p = 0.4), peritonitis (6/150 vs. 8/175, p = 0.86) and local fistula (2/150 vs. 4/175, p = 0.39) formation. Additionally, in the same study a multivariate analysis revealed preoperative anti-TNF therapy as an independent predictor of infectious and surgical site complications but was not associated with increased intraabdominal septic complications. No correlation between peri-operatively anti-TNF therapy and postoperative anastomotic-related complications have also revealed in the Nasir et al. study (5.0 % in anti-TNF group vs. 7.2 % in no anti-TNF group, p = 0.44) and in Kasparek et al.22 study.

Rosenfeld et al.’s meta-analysis addressed “major” and “minor” complications without evaluating separately anastomosis-related complications.

In our study, we focused on the endpoints that were associated with anastomosis-related complications and evaluated the findings according to the quality of the relevant studies (Tables 3 and 4). The rate of intra-abdominal abscess formation and anastomotic leak was of significant difference between anti-TNF receiving and not receiving CD patients in one “well-designed” study, as defined from MINORS Criteria (MINORS Index >16), whereas four studies of MINORS score >16 revealed no statistical significant difference. In view of fistula formation there was no study supporting statistical important difference, whereas three “well-designed” studies revealed no significant difference. Accordingly to re-operation rate, one study of MINORS score over 16 found significant increased rate in anti-TNF treated patients whereas three “well-designed” studies found no difference. To sum up, the majority of studies of MINORS Index >16 revealed no significant association of anti-TNF treatment and anastomosis related complications. In view of the available evidence, in patients who have received anti-TNF therapy and in whom surgical intervention is deemed necessary, one need not delay the surgery for fear of an increased incidence of surgical complications.

However, major limitations of the included studies prohibit the extraction of safe conclusions. First of all, an important limitation of included studies is concomitant immunosuppressive treatment. Indeed, a great proportion of the IFX-treated patients had received combined therapy with immunomodulators (AZA/6-MP) and/or steroids. As a result, it is difficult to approach with accuracy the net effect of anti-TNF on postoperative complication rate as an independent risk factor.

More limitations originating from the retrospective design of the studies are pointed in the meta-analysis of Kopylov et al. Heterogeneity in patient's inclusion criteria, in the definitions of infectious/non-infectious complications and more importantly in timing of the assessment of complications may affect the outcome of each study. Indeed, inclusion of late (>30 days) post-operative complications may induce a Type I error.

Finally, interpretation errors may occur because of the lack of standardization in the timing of pre-operative cessation of IFX. This period varies in the evaluated studies (Fig. 3). IFX, as with all monoclonal antibodies, is eliminated after its degradation by proteases. Elimination half-life of IFX ranges generally from 7 to 12 days.32 Roughly, by 11–13 weeks after IFX's last dose, most patients should have under-therapeutic (<1 μg/ml) or even undetectable trough concentrations. Thus, studies with early preoperative cessation of IFX may include patients with low or undetectable IFX trough levels at the time of surgery and overestimate the incidence of post-operative complications attributed to IFX treatment. It is also worth mentioning that this error may be more profound in patients who have developed antibodies towards IFX (ATIs). ATIs influence the pharmacokinetics33 of IFX by increasing its clearance 2,7 times and as a result by shortening IFX’s half-life. Obviously, as the half-life diminishes, the greater is the possibility of lower IFX trough levels at the time of surgery. In surgical population this effect could be more obvious, as a significant proportion of patients refractory to optimization with IFX treatment is finally treated surgically. This hypothesis is confirmed by a recent study,34 which evaluated the postoperative outcomes in IBD patients receiving IFX pre-operatively. In this study, a significant proportion (9/19) of patients had undetectable IFX serum levels with a median interval time between last dose and surgery of 34 days.

Fig. 3
figure 3

Preoperative Infliximab window, and post-operative IFX administration in every study

Full size image

Large-scale trials measuring IFX trough levels at the time of surgery would approach objectively the effect of peri-operative IFX administration to the rate of postoperative complications. If future clinical and/or experimental studies demonstrate the potent adversely affect of IFX to the healing of intestinal anastomoses, then further investigations have to answer important questions:

  • Is there a dose-related effect of IFX treatment on intestinal anastomoses?

  • Which is the critical threshold of IFX trough levels that delimits the adversely effect of IFX to the healing process of intestinal anastomoses?

  • Does combination therapy with steroids and/or purine analogues lessen this threshold?

  • Could delaying the surgical intervention till IFX trough levels deescalate be an acceptable treatment strategy?

Conclusion

We used MINORS Criteria as a tool for evaluating the quality of non-randomized studies and we concluded that the majority of “well-designed” studies revealed no significant association between anti-TNF administration and anastomosis-related complications. Based on these data, in patients who have received anti-TNF therapy and in whom surgical intervention is deemed necessary, one need not delay the surgery for fear of an increased incidence of surgical complications. However, data originating from studies with major limitations and significant amount of methodological heterogeneity cannot support any definitive conclusions. In our opinion large-scale prospective studies measuring the anti-TNFs trough levels at the surgery time are needed to clarify the effect of anti-TNFs on post-operative septic anastomosis-related complications rates.