The Effect of Anti-Tumor Necrosis Factor-Alpha Therapy within 12 Weeks Prior to Surgery on Postoperative Complications in Inflammatory Bowel Disease: A Systematic Review and Meta-Analysis

The rate of abdominal surgical interventions and associated postoperative complications in inflammatory bowel disease (IBD) patients is still substantially high. There is an ongoing debate as to whether or not patients who undergo treatment with anti-tumor necrosis factor-alpha (TNF-α) agents may have an increased risk for general and surgical postoperative complications. Therefore, a systematic review and meta-analysis was conducted in order to assess the effect of anti-TNF-α treatment within 12 weeks (washout period) prior to abdominal surgery on 30-day postoperative complications in patients with IBD. The results of previously published meta-analyses examining the effect of preoperative anti-TNF-α treatment on postoperative complications reported conflicting findings which is why we specifically focus on the effect of anti-TNF-α treatment within 12 weeks prior to surgery. PubMed, Cochrane, Scopus, Web of Science, World Health Organization Trial Registry, ClinicalTrials.gov and reference lists were searched (June 1995–February 2022) to identify studies, investigating effects of anti-TNF-α treatment prior to abdominal surgery on postoperative complications in IBD patients. Pooled odds ratios (OR) with 95% confidence intervals (CI) were calculated and subgroup analyses were performed. In this case, 55 cohort studies (22,714 patients) were included. Overall, postoperative complications (OR, 1.23; 95% CI, 1.04–1.45; p = 0.02), readmission (OR, 1.39; 95% CI, 1.11–1.73; p = 0.004), and intra-abdominal septic complications (OR, 1.89; 95% CI, 1.44–2.49; p < 0.00001) were significantly higher for anti-TNF-α-treated patients. Significantly higher intra-abdominal abscesses and readmission were found for anti-TNF-α-treated CD patients (p = 0.05; p = 0.002). Concomitant treatment with immunosuppressives in <50% of anti-TNF-α-treated patients was associated with significantly lower mortality rates (OR, 0.32; 95% CI, 0.12–0.83; p = 0.02). Anti-TNF-α treatment within 12 weeks prior to surgery is associated with higher short-term postoperative complication rates (general and surgical) for patients with IBD, especially CD.


Introduction
With worldwide increasing incidence and prevalence, inflammatory bowel diseases (IBD), encompassing Crohn's Disease (CD) and Ulcerative Colitis (UC), belong to a group of diseases with substantial social and economic burden on health care systems and governments. [1] Compared to the 1990s, the life-time risk of undergoing a major abdominal surgery for IBD has decreased in the last decade. [2] Consequently, the treatment with 2 of 23 biological drugs has continuously increased over these years and has become an established practice in the treatment of IBD nowadays. [2] Tumor necrosis factor-alpha (TNF-α) inhibitors were among the first FDA approved biologic drugs in the treatment for IBD. Today, up to 40% of CD patients and up to 16% of UC patients are under treatment with anti-TNF-α biologics such as Adalimumab, Infliximab, Golimumab, Certolizumab (pegol) or one of its biosimilars. However, in comparison with the general population, the 5-year surgery rates with its accompanying potential postoperative complications remain substantial for IBD patients. [2] Previously conducted meta-analyses, analyzing the effect of preoperative anti-TNF-α treatment on postoperative complications has reported conflicting findings [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17].
Therefore, the aim of this study was to evaluate the effect of anti-TNF-α treatment exclusively within the washout period of 12 weeks (3 months) [18] prior to abdominal surgery, on a 30-day general and surgical postoperative complication rate in IBD patients. A systematic review and meta-analysis were conducted, comparing summary effect sizes, calculating the pooled odds ratio (OR) with 95% confidence intervals (CI) and performing subgroup analyses with subgroups stratified by IBD subtype, surgical approach (open versus (vs.) laparoscopic), elective vs. emergency surgery, protective ileostomy use and concomitant use of corticosteroids and/or immunomodulatory drugs.

Methods
This systematic review and meta-analysis were conducted and reported according to the recommendations in the Cochrane Handbook for Reviews of Interventions [19] and the Preferred Reporting Items for Systematic review and Meta-Analyses (PRISMA) statement 2020 [20].

Eligibility Criteria
All observational studies (prospective or retrospective comparative cohort or casecontrol studies), nested case-control studies, non-randomized controlled trials, randomized controlled trials and cross-sectional studies were included based on the following criteria: examined humans (of which the majority were ≥18 years old), being published in English or German language, being available as full-text article in the electronic medical databases between 1 June 1995 and 17 February 2022, patients undergoing any intestinal surgical procedures for IBD (CD/UC/Indeterminate Colitis (IC)), an intervention group including patients who either received anti-TNF-α biologics within the washout period of 12 weeks (3 months) [18] prior to surgery or had a detectable serum concentrations (≥0.98 µg/mL) [21] of these drugs at surgery (regardless of anti-TNF-α biologic preparation and dose), a control group including patients who either did not received any biologic therapy before surgery or had no detectable serum concentrations of an anti-TNF-α biologic (<0.98 µg/mL) [21] at surgery, and a reported exactly 30-day postoperative complication rate. If studies reported two or more discrete data sets, these sets were separately included for the analysis.
The exclusion criteria comprised studies without a control group, abstracts and conference proceedings, reviews and meta-analyses, case-reports and case-series, in-vivo, ex-vivo and in-vitro studies, animal studies, predominantly pediatric patients being studied, concomitant use of any biologic drug other than anti-TNF-α biologics, the intervention groups' last exposure to anti-TNF-α biologics more than 12 weeks (3 months) [18] prior to intestinal surgery, and a follow-up period below or above 30 days after surgery for both the intervention and control groups.

Search Strategy
A systematic literature search was performed for studies published in the electronic medical databases PubMed (MEDLINE), Web of Science, Cochrane Library and Cochrane central register of controlled trials, Scopus, ClinicalTrials.gov and the World Health Organization Trials Registry, using predefined search items for each database (Supplementary Tables S1.1.-S1.6.). The reference lists of the included studies were examined manually, and an additional web search was conducted to ensure that potentially relevant studies were not missed (Supplementary Table S1.7.) . In the case of insufficient or inadequate data presentation, authors were contacted to provide the required information. The final search was conducted on 17 February 2022.

Selection Process
All studies were assessed manually and independently by two investigators (surgical resident: K.C., radiology resident: S.R.) and were exported to the reference management tool EndNote X9 (EndNote X9; The EndNote Team, Clarivate 2013; Philadelphia, PA, USA). According to this studies' predefined eligibility criteria, titles and abstracts were assessed, excluding both duplicates and studies not coinciding with the eligibility criteria. Finally, the remaining full-text articles were retrieved and evaluated for eligibility. The disagreements concerning eligibility were resolved in consensus with a third investigator (surgical specialist: P-A.N.), who independently assessed the accuracy of search results.

Data Collection Process
The data were collected and analyzed independently by two investigators (K.C., S.R.) onto a Microsoft Excel spreadsheet (Home and Student 2019 edition; Microsoft, Redmond, WA) and was assessed independently for accuracy by a third investigator (P-A.N.). The emerging discrepancies in relation to data extracted were discussed and resolved by consensus of all three investigators.

Data Items
The following data were collected for each study if available: author, year, and country of publication, study design and inclusion period, medical treatment and time frame in which drugs were used preoperatively, number of patients in the intervention and control group, patients' baseline and surgical characteristics, and postoperative follow-up period and postoperative complication rates.
In order to capture all complications while ensuring comparability of data, postoperative complication rates were analyzed according to the definitions used in the included studies.
For this review, 30-day postoperative complication rates were quantified and analyzed, including: (1) general overall complication rates: a.
overall postoperative complications b.
overall infectious postoperative complications c.
overall Clavien-Dindo minor and major complications d.
readmission rates, reoperation rates, mortality rates In order to serve as an orientation for surgeons and provide an implication into surgical practice, this review and the resulting meta-analysis primarily focused on 30-day postoperative general and surgical complication rates.
Potential risk factors for postoperative complications within each study group were defined a priori as: tobacco use in >50% of patients, open surgical approach and/or conversion to open surgery in >50% of patients (OA), elective and emergency surgery (ELEMS) vs. exclusively elective surgery (EL), performance of a temporary protective ileostomy in < 50% of patients or no use at all (PRI), Body-Mass-Index (BMI) of >25 kg/m 2 or <18.5 kg/m 2 , dysplasia or malignancy in >50% of patients; perforating or penetrating disease as indication for surgery in >50% of patients, concomitant corticosteroid and/or immunomodulatory agent use in >50% of patient (CSIM), laboratory values (median or mean C-reactive protein concentration of >10 mg/L, white blood cell count > 11 × 10 9 cells/L or <4 × 10 9 cells/L, hemoglobin value of <10 g/dL, and albumin value of <3 g/dL and platelet count of <150 × 10 3 /µL).

Synthesis Methods
The heterogeneity between studies was analyzed using the statistical I 2 test [44], considering a I 2 of ≥50% as substantial heterogeneity. [44] In case of significant or substantial heterogeneity among the included studies, sensitivity analysis was conducted by evaluating the effect of excluding one study at a time on the pooled OR. According to the recommendations of the Cochrane Handbook for Reviews of Interventions [19], potential publication biases were assessed for outcomes reported by ≥10 studies by applying the Egger's test [45] for funnel plot asymmetry.
The prespecified subgroup analyses were performed for outcomes reported by ≥ 5 studies to evaluate the influence of IBD subtypes and potential risk factors on the 30-day postoperative complications. The differences in the outcomes between these subgroups were assessed and reported using the test for subgroup differences (TSD). For subgroups stratified by IBD subtypes, sensitivity analysis was conducted by excluding IBD mixed populations, for which data of the separate IBD subtypes could not be retrieved. Further sensitivity analysis was conducted for subgroups stratified by prespecified risk factors by excluding studies, for which data of studied risk factor were not available.
The results with a p-value of <0.05 were considered significant. All statistical analyses were carried out using the Review Manager software version 5.3 (Nordic Cochrane Centre, Copenhagen, Denmark) and JASP Team (2021; JASP (Version 0.16) [Computer software]). In the case of significant or substantial heterogeneity (I 2 ≥ 50%) of included studies, randomeffects model (RE) was used to conduct the meta-analysis. With a non-significant I 2 < 50% of included studies, fixed-effects model (FE) was used to conduct the meta-analysis. Using either RE or FE meta-analysis, pooled OR with 95% CI were summarized and depicted in a forest plot.

Study Selection
As depicted in Figure 1, 914 studies were identified through electronic medical database search and another 21 studies through manual search (other methods). After removing 181 duplicates, titles and abstracts of 733 articles were screened manually of which 590 were excluded. For the full-text review 137 out of 143 studies (electronic medical database search) and 9 out of 21 studies (other methods) could be retrieved.

Study Selection
As depicted in Figure 1, 914 studies were identified through electronic medical database search and another 21 studies through manual search (other methods). After removing 181 duplicates, titles and abstracts of 733 articles were screened manually of which 590 were excluded. For the full-text review 137 out of 143 studies (electronic medical database search) and 9 out of 21 studies (other methods) could be retrieved.
All included studies analyzed the short-term postoperative complication and 30-day complication rates. (Table 2 and Supplementary Tables S2.1
The subgroup analyses found no subgroup difference for subgroups stratified by IBD subtypes, OA, ELEMS, PRI and CSIM. (Supplementary Table S4).  The subgroup analyses found no subgroup difference for subgroups stratified by IBD subtypes, OA, ELEMS, PRI and CSIM (Supplementary Table S4).
The subgroup analyses found no subgroup difference for subgroups stratified by IBD subtypes, OA, ELEMS, PRI and CSIM (Supplementary Table S7).
The subgroup analyses found a significant subgroup difference for subgroups stratified by CSIM (TSD: p = 0.02; sensitivity analysis: p = 0.006). Mortality was significantly lower for patients in the intervention group if concomitant corticosteroids or immunomodulatory drugs were used in just a minority of patients or not at all (OR, 0.32; 95% CI, 0.12-0.83; p = 0.02). No sensitivity analysis stable subgroup differences were found for subgroups stratified by IBD subtypes, OA, ELEMS and PRI (Supplementary Table S8).
The studies showed a significant substantial heterogeneity (I 2 = 76%; p < 0.0001), still results remained stable throughout sensitivity analysis. No significant publication bias was observed (Egger's test: p = 0.439).
No subgroup differences were found for subgroups stratified by OA, ELEMS, and CSIM (Table 3).
The subgroup analysis showed no subgroup differences for subgroups stratified by the IBD subtype, ELEMS and CSIM (Supplementary Table S11).
No subgroup differences were found for subgroups stratified by OA, ELEMS, and CSIM (Table 3).

Analysis of Other Postoperative Complications
No significant differences were found between the intervention and control group for: (1) Overall non-infectious postoperative complications [48,54,71,89] Table S19).

Discussion
This systematic review and meta-analysis give an extensive overview on the 30-day postoperative complication rates in IBD patients following preoperative anti-TNF-α treatment within the washout period of 12 weeks [18] prior to abdominal surgery.
In this specific study, the authors reported an increased risk for overall morbidity and intra-abdominal septic complications for anti-TNF-α treated CD patients after ileocolic resection, regardless of disease severity. Since the studies' authors did not specify overall postoperative morbidity further, this study might have introduced the observed heterogeneity [49].
In contrast, no significant differences were found between the studied groups for overall infectious complications, Clavien-Dindo major complications, reoperation rates and AL, even after performing thorough subgroup analysis stratified by disease subtypes and potential risk factors (Supplementary Figures S1-S3 and S6; Tables S5-S7 and S11).
The significantly higher postoperative intra-abdominal abscesses and readmission rates were found in the subgroup analyses for patients in the intervention group, when the underlying disease was CD or when only a minority of patients received a protective ileostomy after surgery. In addition, an open surgical approach in >50% of patients were found to be associated with significantly higher readmission rates for patients in the intervention group ( Figure 3; Table 3; Supplementary Figure S7). The major surgical procedure for the anti-TNF-α treated group with low protective ileostomy formation (<50% of patients) and significantly higher intra-abdominal abscess rates were ileocolic resections [46,50,60,61,[78][79][80] and colectomies [54] with primary anastomosis, mostly conducted in CD patients [46,50,60,61,79,80] (Table 3).
Interestingly, Myrelid et al., 2012 [94] investigated 132 CD patients who underwent ileocecal or ileocolonic resection for CD and found the 30-day postoperative anastomotic complications and reoperation rate to be significantly lower for patients undergoing highrisk resections with split-stoma and delayed anastomosis compared to those with primary anastomosis. Furthermore, the authors reported the risk of anastomotic complications to increase with the number of identified preoperative risk factors [94].
Even though this meta-analysis found no association between preoperative anti-TNFα treatment and AL (Supplementary Figure S6; Supplementary Table S11), significantly increased intra-abdominal abscesses were found in the anti-TNF-α treated CD group with low protective ileostomy formation (<50% of patients). The benefit of a temporary ileostomy for patients with CD has not been conclusively clarified and is overall rarely used in everyday surgical practice.
In fact, previously conducted meta-analyses [6,7] had reported significantly higher intra-abdominal abscess rates, but significantly increased readmission rates have not been related to preoperative anti-TNF-α therapy previously [15].
Likewise, none of the previously conducted meta-analyses [11,15] reported significant differences in postoperative mortality between the examined groups, which corresponds to the primary finding in this study. However, subgroup analysis found significantly lower mortality rates for the anti-TNF-α-treated patient group when <50% of them were treated concomitantly with corticosteroids and/or immunomodulatory drugs (Supplementary Figure S4; Supplementary Table S8).
This result suggests that the concomitant perioperative immunosuppression in patients preoperatively treated with anti-TNF-α biologics could be detrimental and should therefore be kept to a minimum, in order to increase the chance of overall postoperative survival. On the other hand, a certain protective effect of perioperative immunosuppression can be presumed for IBD patients not treated with anti-TNF-α biologics preoperatively. Figures S8 and S9), surgical (Supplementary Figures S10-S15; Supplementary Tables S12-S15) or non-surgical complications ( Supplementary Figures S16-S21; Supplementary Tables S16-S19) were found to be associated with preoperative anti-TNF-α therapy.
Furthermore, the influence of preoperative timing of anti-TNF-α biologic drug administration on postoperative complications could not be statistically evaluated as the included studies only depict a preoperative time frame, rather than a specific time point at which these biologics were administered [21][22][23].
Finally, potential confounding factors (such as disease severity, comorbidities, concomitant immunosuppression, to name a few) may have influenced or even biased the outcomes of the analyses. This important limitation was addressed by conducting extensive subgroup analysis according to the most commonly reported risk factors. Furthermore, in case of significant or substantial heterogeneity sensitivity analysis was performed to evaluate the effect of excluding one study at a time, on pooled OR.
In addition, we eliminated many sources of bias, still found in other meta-analysis, e.g., The outcomes of this systematic review and meta-analysis present some clinical implications, not at least due to the above-mentioned strengths and transparent working processes. The outcomes of this study suggest that the treatment with anti-TNF-α biologic drugs within the washout period of 12 weeks [18] (or detectable serum concentrations [21]) prior to abdominal surgery are associated with important short-term postoperative general and surgical complications for patients with IBD, in particular CD. Still, future prospective studies need to be conducted to assess the magnitude of the influence of preoperative timing of anti-TNF-α treatment and potential confounding factors on postoperative outcomes.

Conclusions
In conclusion, current evidence suggests that the treatment with anti-TNF-α agents within the washout period of 12 weeks (3 months [18] or detectable serum concentrations [21]) prior to abdominal surgery is associated with important short-term general and surgical postoperative complications for patients with IBD, in particular with CD. With regard to postoperative mortality, on the one hand, the concomitant treatment with corticosteroids and/or immunomodulatory drugs could be detrimental for patients preoperatively treated with anti-TNF-α biologics, which is why those patients might benefit from a reduction of perioperative immunosuppression, if possible. On the other hand, perioperative immunosuppressive therapy could have a protective effect for IBD patients not treated with anti-TNF-α preoperatively.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/jcm11236884/s1, Figure S1: Random-effects model meta-analysis for 30-day overall infectious postoperative complications in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S2: Fixed-effects model meta-analysis for 30-day overall postoperative Clavien-Dindo major complications in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S3: Fixed-effects model meta-analysis for 30-day postoperative reoperation rate in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S4: Fixed-effects model meta-analysis for 30-day postoperative mortality in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S5: Random-effects model meta-analysis for 30-day overall infectious surgical-site postoperative complications in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S6: Fixed-effects model meta-analysis for 30-day postoperative anastomotic leakage in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S7: Fixedeffects model meta-analysis for 30-day postoperative intra-abdominal abscesses in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S8: Fixed-effects model meta-analysis for 30-day overall non-infectious postoperative complications in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S9: Fixed-effects model meta-analysis for 30-day overall postoperative Clavien-Dindo minor complications in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S10: Random-effects model meta-analysis for 30-day postoperative superficial surgical-site infections in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S11: Fixed-effects model meta-analysis for 30-day postoperative deep or organ space surgical-site infections in the anti-TNFα (intervention) and control group. Forest plot of all studies included Figure S12: Fixed-effects model meta-analysis for 30-day postoperative fistula formations in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S13: Fixed-effects model meta-analysis for 30-day postoperative ileus in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S14: Fixed-effects model meta-analysis for 30-day postoperative small bowel obstructions in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S15: Fixed-effects model meta-analysis for 30-day postoperative hemorrhage in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S16: Fixed-effects model meta-analysis for 30-day overall infectious non-surgical-site postoperative complications in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S17: Fixed-effects model meta-analysis for 30-day postoperative thrombosis in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S18: Fixed-effects model meta-analysis for 30-day postoperative cardiovascular complications in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S19: Fixed-effects model meta-analysis for 30-day postoperative pneumonia in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S20: Fixed-effects model meta-analysis for 30-day postoperative urinary tract infections in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Figure S21: Fixed-effects model meta-analysis for 30-day postoperative sepsis in the anti-TNF-α (intervention) and control group. Forest plot of all studies included Tables S1.1.-S1.7.: Search strategy; Tables S2.1-S2.51: Newcastle Ottawa Quality (NOS) [43] Assessment to assess the quality of included cohort studies; Tables S3.1.-S3.2.: Other 30-day postoperative complications; Table S4: Random-effects model meta-analysis for 30-day overall postoperative complications in the anti-TNF-α (intervention) and control group. Analysis for publication bias; sensitivity analysis and subgroup analysis; Table S5: Random-effects model meta-analysis for 30-day overall infectious postoperative complications in the anti-TNF-α (intervention) and control group. Analysis for publication bias; sensitivity analysis and subgroup analysis; Table S6: Fixed-effects model meta-analysis for 30-day overall postoperative Clavien-Dindo major complications in the anti-TNF-α (intervention) and control group. Subgroup analysis; Table S7: Fixed-effects model meta-analysis for 30-day postoperative reoperation rate in the anti-TNF-α (intervention) and control group. Analysis for publication bias and subgroup analysis; Table S8: Fixed-effects model meta-analysis for 30-day postoperative mortality in the anti-TNF-α (intervention) and control group. Analysis for publication bias and subgroup analysis; Table S9: Random-effects model meta-analysis for 30-day overall postoperative infectious surgical-site compli-cations in the anti-TNF-α (intervention) and control group. Analysis for publication bias; sensitivity analysis and subgroup analysis; Table S10: Fixed-effects model meta-analysis for 30-day postoperative intra-abdominal septic complications in the anti-TNF-α (intervention) and control group. Subgroup analysis; Table S11: Fixed-effects model meta-analysis for 30-day postoperative anastomotic leakage in the anti-TNF-α (intervention) and control group. Analysis for publication bias and subgroup analysis; Table S12: Fixed-effects model meta-analysis for 30-day postoperative superficial surgicalsite infections in the anti-TNF-α (intervention) and control group. Analysis for publication bias; sensitivity analysis and subgroup analysis; Table S13: Fixed-effects model meta-analysis for 30-day postoperative ileus in the anti-TNF-α (intervention) and control group. Analysis for publication bias and subgroup analysis; Table S14: Fixed-effects model meta-analysis for 30-day postoperative small bowel obstruction in the anti-TNF-α (intervention) and control group. Analysis for publication bias and subgroup analysis; Table S15: Fixed-effects model meta-analysis for 30-day postoperative hemorrhage in the anti-TNF-α (intervention) and control group. Subgroup analysis; Table S16: Fixedeffects model meta-analysis for 30-day postoperative thrombosis in the anti-TNF-α (intervention) and control group. Subgroup analysis; Table S17: Fixed-effects model meta-analysis for 30-day postoperative pneumonia in the anti-TNF-α (intervention) and control group. Subgroup analysis; Table S18: Fixed-effects model meta-analysis for 30-day postoperative urinary tract infections in the anti-TNF-α (intervention) and control group. Subgroup analysis; Table S19: Fixed-effects model meta-analysis for 30-day postoperative sepsis in the anti-TNF-α (intervention) and control group. Subgroup analysis.

Data Availability Statement:
The data used to support the findings of this systematic review and meta-analysis are included within the article.

Conflicts of Interest:
The authors declare no conflict of interest.