Postoperative Crohn’s Disease Recurrence Risk and Optimal Biologic Timing After Temporary Diversion Following Ileocolic Resection

Abstract

Background

Postoperative recurrence of Crohn’s disease (CD) is common. While most patients undergo resection with undiverted anastomosis (UA), some individuals also have creation of an intended temporary diversion (ITD) with an ileostomy followed by ostomy takedown (OT) due to increased risk of anastomotic complications. We assessed the association of diversion with subsequent CD recurrence risk and the influence of biologic prophylaxis timing to prevent recurrence in this population.

Methods

This was a retrospective cohort study of CD patients who underwent ileocolic resection between 2009 and 2020 at a large quaternary health system. Patients were grouped by continuity status after index resection (primary anastomosis or ITD). The outcomes of the study were radiographic, endoscopic, and surgical recurrence as well as composite recurrence postoperatively (after OT in the ITD group). Propensity score–weighted matching was performed based on risk factors for diversion and recurrence. Multivariable regression and a Cox proportional hazards model adjusting for recurrence risk factors were used to assess association with outcomes. Subgroup analysis in the ITD group was performed to assess the impact of biologic timing relative to OT (no biologic, biologic before OT, after OT) on composite recurrence.

Results

A total of 793 CD patients were included (mean age 38 years, body mass index 23.7 kg/m², 52% female, 23% active smoker, 50% penetrating disease). Primary anastomosis was performed in 67.5% (n = 535) and ITD in 32.5% (n = 258; 79% loop, 21% end) of patients. Diverted patients were more likely to have been males and to have had penetrating and perianal disease, prior biologic use, lower body mass index, and lower preoperative hemoglobin and albumin (all P < .01). After a median follow-up of 44 months, postoperative recurrence was identified in 83.3% patients (radiographic 40.4%, endoscopic 39.5%, surgical 13.3%). After propensity score matching and adjusting for recurrence risk factors, no significant differences were seen between continuity groups in radiographic (adjusted hazard ratio [aHR], 1.32; 95% confidence interval [CI], 0.91-1.91) or endoscopic recurrence (aHR, 1.196; 95% CI, 0.84-1.73), but an increased risk of surgical recurrence was noted in the ITD group (aHR, 1.61; 95% CI, 1.02-2.54). Most (56.1%) ITD patients started biologic prophylaxis after OT, 11.4% before OT, and 32.4% had no postoperative biologic prophylaxis. Biologic prophylaxis in ITD was associated with younger age (P < .001), perianal disease (P = .04), and prior biologic use (P < .001) but not in recurrence (P = .12). Despite higher rates of objective disease activity identified before OT, biologic exposure before OT was not associated with a significant reduction in composite post-OT recurrence compared with starting a biologic after OT (52% vs 70.7%; P = 0.09).

Conclusions

Diversion of an ileocolic resection is not consistently associated with a risk of postoperative recurrence and should be performed when clinically appropriate. Patients requiring diversion at time of ileocolic resection are at high risk for recurrence, and biologic initiation prior to stoma reversal may be considered.

Introduction

Crohn’s disease (CD) is a chronic inflammatory condition with an ever-increasing worldwide prevalence.¹ CD may lead to major complications such as strictures, fistulas, or abscesses. Ileocolic resection (ICR) and ileocolic anastomosis is the most frequently performed surgical procedure in CD often for disease-related complications.¹ Following resection, bowel continuity can either be restored at the time of the ICR, termed a undiverted anastomosis (UA), or patients can undergo intended temporary diversion (ITD), via either a diverting loop ileostomy (DLI) or end or end-loop ileostomy (EI), with planned subsequent restoration of continuity. The clinical decision to perform a UA vs ITD is often complex, multifactorial, surgeon dependent, and attributed to the risk of ensuing anastomotic complications, such as leaks should UA be undertaken. Specifically, emergent surgery, smoking, inpatient status, malnutrition (significant weight loss and/or hypoalbuminemia), anemia, chronic steroids, and prolonged operative time are independently associated with anastomotic leaks.²

CD recurrence after resection occurs in up to 50% of patients and proceeds along a micro- to macroscopic continuum.³ Risk factors for early CD recurrence after resection include young age at diagnosis (<30 years), active smoking, >2 surgeries for penetrating disease, and <3 years since previous surgery. The American Gastroenterological Association guidelines on the management of postoperative CD favor the use of pharmacological prophylaxis soon after ICR in these high-risk patients to reduce recurrence rates; however, the data involving patients with DLI or EI are limited.⁴ Diverted patients experience delayed histologic disease activity initiation with rapid activation within 2 weeks upon fecal stream restoration.^5–8 Consequently, OT is currently thought that diverted patients are protected from downstream recurrence and medical prophylaxis initiation can be deferred until restoration of continuity. However, factors influencing the decision to divert also share commonality with risk factors for disease recurrence, and earlier initiation of prophylaxis may mitigate this risk even further. Thus, we aimed to compare disease recurrence rates among continuity groups after index resection (UA and ITD) and determine whether disease recurrence rates vary between CD patients receiving biologics before and after OT.

Methods

We performed a retrospective observational cohort study of adult (>18 years) CD patients undergoing ICR at the Cleveland Clinic Foundation (18 hospitals in Northeast Ohio and Florida) between January 1, 2009, and January 1, 2020. Our CD population was defined using standard International Classification of Diseases–Ninth Revision (555.0, 555.1, 555.2, 555.9) and International Classification of Diseases–Tenth Revision codes (K50.0, K50.1, K50.8, K50.9) for CD entered by a gastroenterologist or colorectal surgeon. Surgical intervention for ICR or redo ICR was identified using Current Procedural Terminology codes (44160, 44140, 44204, 44205). Manual chart review of clinician documentation, operative reports documented at time of surgery, preoperative laboratory values, and ileocolonoscopic and imaging reports was performed to confirm the accuracy of the CD diagnosis, surgical intervention, and relevant clinical data. Patients were excluded if (1) ICR was performed for a non-CD indication (eg, neoplasm, ischemia), (2) there was no gastroenterology follow-up (<1 outpatient clinic visits postoperatively), (3) there was permanent ostomy, or (4) there was missing endoscopy and cross-sectional imaging information.

Demographic and Clinical Data

Data were collected through manual chart review by 4 independent reviewers (A.J, S.B., R.S., J.E.B.). Patient demographics (age, age at surgery, sex, active tobacco use at surgery, diabetes mellitus, body mass index [BMI]), disease characteristics (disease duration, disease behavior, disease location, perianal disease according to Montreal classification),⁹ patient malnutrition status assessed preoperatively by a registered dietitian and classified as mild/moderate/severe protein calorie malnutrition, prior biologic use (none, anti-tumor necrosis factor [anti-TNF], vedolizumab, or ustekinumab), preoperative biologic exposure within 12 weeks of surgery, number of biologics used prior to surgery, preoperative systemic steroid or immunomodulator use within 4 weeks of surgery, and preoperative laboratory values (hemoglobin, white blood cell count, platelets, albumin, C-reactive protein) were recorded. The 12-week preoperative biologic period was chosen based on biologic pharmacokinetics and previous data on complications associated with preoperative biologic exposure.^10,11 Exposure to biologics during this period was confirmed via manual chart review. The most recent biologic exposure was recorded if a patient received multiple biologics over the 12-week period.

Surgical data included type of anastomosis (end-end, end-side, side-side), nature of surgery (elective/urgent), anastomotic technique (handsewn or stapled), surgical approach (hand-assist, laparoscopic, and/or open), diversion technique (DLI, EI), and indication of surgery and diversion (perforation/rupture, abscess, fistula/stricture) when available. Kono-S anastomosis was not offered during the time frame of the study. All surgical decision making was according to surgeon discretion.

Postoperative data included postoperative biologic prophylaxis defined as initiation or continuation of biologic (none, anti-TNF, vedolizumab, or ustekinumab) within 3 months of date of surgery. The decision and timing of biologic initiation was per the treating gastroenterologist.

Ileocolonoscopic Data

Postoperative ileocolonoscopic data were obtained via manual chart review of procedure reports. All lower endoscopic procedures performed >3 months from date of surgery were analyzed. Endoscopic severity was analyzed using the modified Rutgeerts score (RS).¹² If the RS was not prospectively available, the modified RS was retrospectively applied utilizing colonoscopy images and report text. Retrospective RS application was performed by 2 independent scorers (S.P.B., B.C.) and validated (90% accuracy, internal data) using a sample dataset prior to data collection. Scorers were blinded to imaging and clinical data. Ileoscopy through the stoma (proximal to an ileocolonic anastomosis), flexible sigmoidoscopy, and colonoscopies not reaching ileocolonic anastomosis or neoterminal ileum (unless due to stricture) were excluded. Endoscopic reports lacking prospective RS and sufficient data to apply RS were excluded. Patients who did not undergo an endoscopic assessment were excluded from the endoscopic recurrence analysis. Endoscopic postoperative recurrence was defined as a modified RS of i2b or greater,¹³ and as >5 aphthous lesions with normal mucosa between lesions scattered beyond the anastomosis (i2b) or diffuse aphthous ileitis, large ulcers, and nodules (i3, i4). Strictures at the anastomosis were not considered i4 recurrence. Corroborating radiographic data were used to make this determination when available and when endoscopic data were insufficient. Endoscopic assessments performed in the ITD group prior to OT were included for determination of presence or absence of disease activity prior to OT, but not for outcome comparison of endoscopic recurrence between continuity groups (see Outcomes).

Radiographic Data

Clinically indicated radiographic cross-sectional abdominopelvic studies including computed tomography and magnetic resonance imaging, typically enterography protocols, >3 months after surgery were reviewed for evidence of CD activity on the clinical report. Radiographic recurrence was defined as active luminal inflammation attributable to CD per the interpreting radiologist. Patients without radiographic assessments and colonic recurrence were excluded from the radiographic recurrence analysis.

Outcomes

Patients were grouped by continuity status at index resection (UA and ITD) (Figure 1). Patients who had ileocolic anastomosis constructed at time of ICR but were proximally diverted were included in the ITD group, and end ileostomy and EI patients in the EI subgroup. The primary outcome was radiographic, endoscopic, or surgical CD recurrence between the 2 groups (UA or ITD). Recurrence documented by 2 or more of the previous methods was defined as composite recurrence. Surgical recurrence was defined by subsequent ICR for disease activity ≥90 days after the index resection. Ileostomy takedown was not included in the definition of surgical recurrence. For the primary outcome, follow-up time commenced at the time of index ICR in the UA group and at the time of OT and restoration of continuity in the ITD group. To assess the association of biologic initiation timing on recurrence rates relative to OT (no biologic, biologic before OT, after OT), subgroup analysis was performed in the ITD group on composite recurrence. A singular outcome was selected for the subgroup analysis due to the relatively limited sample sizes.

Statistical Analysis

To assess the recurrence between 2 treatment groups, UA and ITD, propensity score (PS)–weighted Cox proportional hazards models were performed after controlling for potential confounders. To account for baseline differences in recurrence risk, PS weights were assessed by a logistic regression model with 17 factors (age at diagnosis, age at surgery, BMI, sex, smoking status, upper gastrointestinal involvement, prior resections, prior ICRs, prior biologic treatment, CD location, CD behavior, perianal disease, surgical approach, surgical anastomosis, preoperative hemoglobin, preoperative white blood cell count, and preoperative platelet count) as independent variables and treatment group as dependent variable. Five percent of patients with extreme weights were trimmed for the weighted Cox proportional hazards models. When calculating the PS weights, the multivariate imputation by chained equation was performed to construct a complete dataset, and Love plots were constructed to indicate the performance of the PS weighting. Weighted Kaplan-Meier survival curves with 95% confidence intervals (CIs) were constructed.

A multivariable Cox proportional hazards regression model was performed to assess the association between outcome, recurrence, and risk factors including biologic use groups. In the multivariable model development, the multivariate imputation by chained equation was also performed to impute missing values to construct a complete dataset for variable selection. The stepwise variable selection method based on Akaike information criterion was used to choose the final model. The variables that had large portion of missing value, were unbalanced between levels, or highly correlated to others were excluded from the model. Kaplan-Meier survival curves with 95% CIs were conducted. Analyses were performed using R (version 3.6.2; R Foundation for Statistical Computing). A P value <.05 was considered statistically significant.

Ethical Considerations

The Cleveland Clinic Institutional Review Board approved this study.

Results

Study Population

A total of 793 adult patients (52% female) with CD who underwent ICR were included. The median age at surgery was 35 years (IQR, 26-49 years), and the age at diagnosis was 23 years (IQR, 17-23 years) (Table 1). Active smoking was recorded in 23% patients. The median BMI was 23.7 kg/m². Disease behavior was predominately stricturing (42.6%), penetrating (19.8%), or a combination of both manifestations (30.6%). Perianal disease was seen in 29% patients at the time of surgery, and 17% patients had upper gastrointestinal involvement. Preoperative (within 4 weeks of surgery) corticosteroid and immunomodulator use rates were 29.1% and 20.2%, respectively. Of the 793 patients, 44.8% patients had not been exposed to a biologic within 12 weeks prior to index ICR, and 59.3% (n = 471) patients received a biologic postoperatively. A total of 450 (56.7%) of the 793 patients had some form of postoperative cross-sectional imaging, and 590 (74.4%) patients underwent endoscopic evaluation and were included in the PS-matching model.

Continuity Groups

UA was performed in 67.5% (n = 535) and ITD in 32.5% (n = 258; 80.2% DLI, 19.8% EI) (Table 1). Diverted patients more often were male (55% vs 45%), had penetrating (31% vs 15%) and perianal disease (37.2% vs 25%), had prior biologic use (66% vs 50%), had lower BMI (23.0 kg/m² vs 24.2 kg/m²), and had lower preoperative hemoglobin (11.4 g/dL vs 12.4 g/dL) and albumin (3.5 g/dL vs 4 g/dL) (all P < .01). A total of 18.6% (n = 148) of diverted patients were exposed to biologics postoperatively. Of the 219 diverted patients, 40.4% (n = 84), 29.3% (n = 61), 18.3% (n = 38), and 12% (n = 25) had a diagnosis of severe, moderate, mild, and no malnutrition, respectively. Diverted patients were followed up earlier between primary surgery and first radiographic or endoscopic assessment (4.5 months vs 11.6 months; P < .001). Identified reasons for diversion included structuring and/or fistulizing disease (41.1%), abdominal abscesses (19%), perforation (14.7%), and unspecified surgical decision (27.1%).

Follow-Up and Recurrence Among Continuity Groups

The median follow-up time for the study cohort was 38.1 months (95% CI, 30.7-45.9 months) for endoscopic recurrence, 43.3 months (95% CI, 36.5-52.5 months) for radiographic recurrence, and 53.5 months (95% CI, 49.4-59 months) for surgical recurrence. The median time duration between the primary surgery and the first assessment was 9.7 months among 682 patients with assessment (radiology or endoscopy). The median follow-up times for radiographic, endoscopic, and surgical recurrence were significantly shorter in ITD group compared with UA (radiographic: 27.4 months [95% CI, 24.3-32.5 months] vs 39.1 months [95% CI, 34-47.2 months]; endoscopic: 27.9 months [95% CI, 25-31.4 months] vs 45.1 months [95% CI, 36.4-49.5 months]; surgical: 42 months [95% CI, 37.9-50.3 months] vs 66.4 months [95% CI, 58.5-73.8 months]).

Postoperative recurrence was identified in 83.3% patients, with a median time to recurrence of 56.1 months. Radiographic recurrence was present in 40.4% (35.8% UA, 51.1% ITD; P = .003) of patients, with a median time to radiographic recurrence of 41 months. Endoscopic recurrence was observed in 39.5% (41.2% UA, 35.1% ITD; P = 0.20) of patients who underwent endoscopic surveillance, with a median time to endoscopic recurrence of 56.1 months. Surgical recurrence was experienced by 13.3% (12.2% UA, 15.7% ITD; P = 0.04), with a median time to surgical recurrence of 110.2 months.

After PS matching, and adjusting for recurrence risk factors, no significant differences were seen between continuity groups in radiographic (adjusted hazard ratio [aHR], 1.32; 95% CI, 0.91-1.91; P = 0.13) or endoscopic (aHR, 0.98; 95% CI, 0.68-1.41; P = 0.89) recurrence across respective follow-up periods. However, surgical recurrence was significantly more common in the ITD group compared with UA (aHR, 1.67; 95% CI, 1.062-2.62; P = .03) (Table 2; Figures 2, 3, and 4).

Biologic Groups

A total of 148 (18.6%) of the 258 ITD patients were postoperatively exposed to biologics, with most patients starting biologics after OT (83.1% [n = 123]), while 16.4% (n = 25) of patients were started on or continued biologics during the diversion period (biologic exposure before OT). Nearly one-third (32.4% [n = 71]) of ITD patients had no postoperative biologic exposure. Biologic exposure in ITD was associated with younger age (P < .001), perianal disease (P = .04), and prior biologic use (P < .001), but there was no difference in high risk for recurrence (P = .18). (Table 3; Supplementary Figures 1, 2, and 3)

Prior to OT, disease activity was identified in 14.2% (endoscopic 1.8%, radiographic 9.6%) and was significantly more common in individuals starting or continuing a biologics prior to OT (32% vs 14.2%; P = .02) (Supplementary Table 1). Following OT, patients in the no biologic group had the least composite post-OT recurrence (39.1%). Composite post-OT recurrence rate in the pre-OT biologic group was not significantly lower than the rate noted in patients in whom biologic therapy was initiated after OT (52.0% vs 70.7%; P = .11) (Table 4; Figure 5).

Discussion

In this large, single-center, retrospective study of adult patients with CD undergoing ICR, we found that utilization of temporary diversion occurred in 32.5% and was not associated with postoperative radiographic, endoscopic, or composite recurrence. However, surgical recurrence was more common in diverted individuals. To our knowledge, this is the first study assessing endoscopic and radiographic recurrence rates among continuity groups and evaluating outcomes in diverted patients based on biologic timing. These data suggest that diversion alone does not influence luminal recurrence rates, and early biologic exposure may mitigate postoperative disease recurrence after ICR in high-risk diverted CD patients.

The rate of diversion in our study population was 32.5%, consistent with prior descriptions.² Similar to previous literature, our study demonstrated that diversion was more common in patients with perianal and penetrating disease, prior biologic use, poor nutritional status indicated by low BMI, low albumin, and low preoperative hemoglobin. Though ICR with UA is the preferred approach in most patients with CD, construction of a diverting ileostomy may be performed in such patients to reduce the risk of anastomotic leaks by promoting healing before pursuing intestinal continuity.² We have previously shown rates of intraabdominal septic complications in the same study cohort. A total of 47 patients (8.8%) developed perioperative intra-abdominal septic complications within 90 days of UA (36 [6.7%] intra-abdominal abscess, 11 [2.1%] anastomotic leak).¹⁴ Despite several studies attempting to risk stratify patients undergoing ICR, the choice to divert patients is ultimately left to the surgeon’s discretion.²

Comparing continuity groups, the risk of radiographic, endoscopic, or composite recurrence was not significantly different whether diversion was employed. A recent multicenter retrospective study by Bolckmans et al¹⁵ showed reduced surgical recurrence over 9 years in a small group of high-risk patients (tobacco use and penetrating disease) after ITD. Interestingly, our study showed an overall increased risk of surgical recurrence. This finding may be due to obstructing strictures confined to the anastomosis harboring i4 disease, and surgical complications such as anastomotic strictures and not true CD recurrence, which technically are i2a and not endoscopic recurrence by our definitions.¹²

Interestingly, our study showed that despite higher rates of disease activity identified before OT, biologic exposure before OT was associated with a numerically reduced rate of composite post-OT recurrence compared with starting a biologic after OT. However, the small sample size limits definitive conclusions. Additionally, we were unable to detect if biologic use before OT was for prophylaxis/treatment, and if biologic use was newly initiated or a was a continuation of preoperative biologic therapy. Current guidelines recommend initiating high-risk patients on biologic prophylaxis with anti-TNF plus immunomodulator.⁴ Additionally, there is emerging evidence that anti-TNF therapy may be safe in the perioperative period. The PUCCINI (Prospective Cohort of Ulcerative Colitis and Crohn’s Disease Patients Undergoing Surgery to Identify Risk Factors for Post-operative Infection I) trial showed that preoperative use of anti-TNF therapy was not an independent risk factor of postoperative biologic exposure.¹⁶ Recognizing the known rapid onset of histologic activity following restoration of continuity, OT may thus be advantageous to initiate biologic prophylaxis prior to OT while not increasing perioperative risks. Thus, although our data are limited in sample size, the relative reduction in post-operative recurrence rates is hypothesis generating and deserves additional investigation.

Our study has limitations. Despite the prospectively maintained databases, the data were retrospectively analyzed and prone to known biases and limitations of retrospective studies including residual confounding despite statistical attempts to control for this problem. As this investigation was from a single-center population including likely a large referral cohort at a higher risk of recurrence, generalizability may be limited. The relatively limited sample size of subgroups may have constrained the ability to detect differences in recurrence rates. The decision to undergo ITD includes complex patient factors identified by the individual surgeon’s subjective discretion. All endoscopic and radiographic assessments were clinically indicated and may have been performed for reasons other than disease activity monitoring. Furthermore, while all endoscopic procedures were individually reviewed and scored when appropriate, radiographic recurrence was based on standard-of-care reports and did not utilize standardized protocols or reinterpretation of primary images. However, prior work from our groups has suggested moderate concordance between radiographic studies and endoscopic assessments for postoperative recurrence.¹⁷ Finally, the timing, indication, and discontinuation data on postoperative medical therapy are not well captured after the initial postoperative period and may be a source of ambiguity in the follow-up period.

Conclusions

Temporary diversion after ICR is not consistently associated with risk or protection of postoperative recurrence and should be performed when clinically appropriate. Biologic initiation prior to stoma reversal may be protective, may reduce postoperative recurrence rates, and may be considered in high-risk individuals or those with identified disease activity. Further studies with larger sample sizes and prospective data collection are needed to help determine optimal risk reduction strategies in this high-risk patient population.

Supplementary data

Supplementary data is available at Inflammatory Bowel Diseases online.

Acknowledgments

None.

Funding

None.

Conflicts of Interest

B.Co. has received financial support for serving on advisory boards and has served as a consultant for AbbVie, Celgene-Bristol Myers Squibb, Pfizer, Sublimity Therapeutics, Takeda, TARGET RWE, Cornerstones, and Vindico; and served as a speaker for AbbVie. F.R. has received research funding from the National Institutes of Health, Helmsley Charitable Trust, Crohn’s and Colitis Foundation, UCB, Pliant, BMS, AbbVie, Pfizer, Boehringer Ingelheim, Morphic; served as a consultant and on advisory boards for Adnovate, Agomab, Allergan, AbbVie, Boehringer Ingelheim, Celgene/BMS, CDISC, Cowen, Galmed, Genentech, Gilead, Gossamer, Guidepoint, Helmsley, Index Pharma, Jannsen, Koutif, Mestag, Metacrine, Morphic, Organovo, Origo, Pfizer, Pliant, Prometheus Biosciences, Receptos, RedX, Roche, Samsung, Surmodics, Surrozen, Takeda, Techlab, Theravance, Thetis, UCB, Ysios, and 89Bio. S.D.H. has received consulting fees from Shionogi and Takeda. M.R. has served on the advisory board or as a consultant for AbbVie, Janssen, UCB, Takeda, Pfizer, Miraca Labs, Amgen, Celgene, Seres, Allergan, Genentech, Gilead, Salix, Prometheus, Lilly, TARGET Pharma Solutions, ALFASIGMA, S.p.A., and Bristol Myer Squibbs. B.Cl. has served as a consultant for AbbVie, Bristol Myers Squibb, Janssen, Pfizer, Prometheus, Takeda, and TARGET-RWE. All other authors have no relevant disclosures to report.

References