The Use of Vedolizumab in Preventing Postoperative Recurrence of Crohn’s Disease

Abstract

Background

Clinical and endoscopic recurrence are common after surgery in Crohn’s disease (CD). Vedolizumab has been increasingly used to treat CD, however, its effectiveness in preventing postoperative recurrence remains unknown. We aimed to investigate the use of vedolizumab in the postoperative setting and compare the risk of recurrence between patients receiving vedolizumab and anti-tumor necrosis factor (TNF)-α agents.

Methods

Medical records of CD patients who underwent surgery between April 2014 and June 2016 were reviewed. We first analyzed how frequently vedolizumab is used to prevent postoperative recurrence and compared the patient characteristics with those being treated with other therapies. Furthermore, the rates of endoscopic remission, defined as a simple endoscopic score for CD of 0, at 6–12 months after surgery were compared between patients receiving vedolizumab and anti-TNF-α agents. Clinical, biological, and histologic outcomes such as Harvey-Bradshaw index, C-reactive protein, and histologic inflammation also were compared between the 2 groups. Risks of recurrence were assessed by univariate, multivariate, and propensity score-matched analyses.

Results

Among 203 patients that underwent a CD related surgery, 22 patients received vedolizumab as postoperative treatment. There were 58, 38, and 16 patients who received anti-TNF-α agents, immunomodulators, and metronidazole, respectively, whereas 69 patients were monitored without any medication. Patients receiving vedolizumab were young and frequently had perianal disease. Patients postoperatively treated with vedolizumab or anti-TNF-α agents were mostly treated with the same agent pre- and postoperatively. Rate of endoscopic remission at 6–12 months in the vedolizumab group was 25%, which was significantly lower as compared to anti-TNF-α agent group (66%, P = 0.01). Vedolizumab use was the only factor that was associated with an increased risk of endoscopic recurrence on both univariate (odds ratio (OR) 5.58, 95% confidence interval (CI) 1.51–24.3, P = 0.005) and multivariate analysis (OR 5.77, 95%CI 1.71–19.4, P = 0.005). The results were supported by a propensity score-matched analysis demonstrating lower rates of endoscopic remission (25 vs 69%, P = 0.03) in patients treated with vedolizumab as compared to anti-TNF-α agents.

Conclusion

In the present retrospective cohort study of real-world experience, vedolizumab was shown to be commonly used as postoperative treatment for CD especially in high risk patients. Multivariate and propensity score-matched analyses showed that postoperative endoscopic recurrence in CD was higher with vedolizumab than with anti-TNF-α agents, but further investigation including controlled trials is required before determining the utility of vedolizumab in preventing postoperative recurrence of CD.

Crohn’s disease, postoperative recurrence, vedolizumab, anti-TNF-α agents

INTRODUCTION

Crohn’s disease (CD) is a chronic inflammatory bowel disease (IBD) caused by immune cells accumulating to the inflamed intestine.^1,2 Patients with CD suffer from various morbidities and complications due to persistent intestinal inflammation. Several reports have described that achievement of mucosal and histologic healing are important factors to prevent disease progression.^3,4 Clinical and endoscopic recurrence are common following surgery, and a substantial proportion of patients require multiple surgeries.⁵ Thiopurines and nitroimidazole antibiotics are known to reduce postoperative recurrence of CD.^6–8 More recently, anti-tumor necrosis factor (TNF)-α agents have been shown to reduce recurrence following surgery.^9,10 Colonoscopy at 6–12 months after surgery is now widely accepted as a strategy to monitor endoscopic recurrence as it predicts future progressive clinical (symptomatic) recurrence.^5,11

Vedolizumab is a novel gut-specific anti-integrin inhibitor that increasingly has been used due to its efficacy and safety in refractory CD. Clinical trials have shown vedolizumab-induced clinical and endoscopic remission in patients who were naive to and previously failing anti-TNF-α agents.^12–14 However, it remains unclear whether vedolizumab reduces the risk of postoperative recurrence in CD.

In the present study of real-world data, we aimed to evaluate the use of vedolizumab in the postoperative setting and assess the clinical, biological, endoscopic, and histologic recurrence at 6–12 months after surgery in patients receiving vedolizumab in comparison to those receiving anti-TNF-α agents.

METHODS

Medical records of patients registered in a prospectively collected IBD database at the University of Chicago Medicinewere retrospectively reviewed for those who underwent surgery between April 2014 and June 2016. The study was approved by the institutional review board (IRB 16–0061). The study was undertaken based on a priori defined protocol. Study patients included adults with CD who underwent surgery during the study period and postoperatively received vedolizumab or other agents as their therapy. Due to the retrospective nature of the study, no predefined criteria were adopted in deciding whether to use vedolizumab or other agents as postoperative therapy and the decision was that of the attending physician. CD surgery included small bowel resection, colonic resection, ileocecectomy, and ileocolectomy. All surgeries were curative as determined to have no remaining major luminal disease by a combination of preoperative work-up, intraoperative findings, and pathology findings. Examination under anesthesia (EUA), ostomy creation, ileostomy takedown, stricturoplasty, etc. were excluded.

Patients with >6 months of postoperative follow-up were included when they had 1 of clinical, laboratory, endoscopic, or histologic outcomes evaluated at 6–12 months. Patients were excluded if they did not have 6 months of follow-up after their operation, they were followed at an outside hospital or had no clinic visits during the 6–12 months period, or data was missing. Data abstracted included patient demographics, smoking history, date and type of surgery, concomitant medications, previous medication use, clinical symptoms, laboratory findings, and endoscopy and histology findings. If multiple clinic visits were found during that period, the date of endoscopy was chosen for assessment.

Outcomes

The primary endpoint was endoscopic remission at 6–12 months after a CD surgery. Endoscopic remission was defined as a simple endoscopic score for CD (SES-CD) of 0. We adopted SES-CD instead of Rutgeerts score for this study so that the degree of inflammation in the ileum and colon could be quantified with a single- scoring system.^5,15 Two-thirds of SES-CD data were prospectively assessed by the endoscopist and the remaing e data were retrospectively calculated based on the reports and images by the study investigators (Akihiro Yamada and Atsushi Sakuraba) when possible. When the sites of anastomosis were not evaluated, patients were excluded from the analysis.

Secondary endpoints included clinical, biological, and histologic remission. Clinical remission was defined as Harvey-Bradshaw index (HBI) ≤4. HBI was prospectively assessed at each visit and recorded in the charts/database. Biological remission was defined as a negative C-reactive protein (CRP) (<3 mg/L, the cutoff for negative at our institution). We did not have data regarding fecal calprotectin as we have not routinely used it in the postoperative setting at our institution. Histologic remission was defined as absence of any active inflammation on biopsies¹⁶ , which was assessed by expert pathologists at our institution. Cases with insufficient information were excluded from each analysis.

Statistical Analysis

Continuous variables were expressed as median and interquartile range (IQR). Differences in the means between subgroups were compared using Mann-Whitney U test. Comparisons between categorical variables were analyzed using the Fisher’s exact test. A P value of < 0.05 was considered statistically significant.

The primary outcome of 6–12 months postoperative recurrence was analyzed using univariate and multivariate comparisons among patients treated postoperatively with vedolizumab vs anti-TNF-α agents. Univariate predictor variables with a P value of < 0.10 were included in the multivariate analysis.

In order to reduce the effect of treatment-selection bias and potential confounding in this observational study, we performed adjustments for significant differences in the baseline characteristics of patients with propensity score matching. Data were matched with concomitant immunomodulator use and age. Serving as control, 1 vedolizumab patient was matched to 1 anti-TNF-α agent patient using nearest-neighbor matching without replacement.^17,18

For all statistical analyses, data were analyzed by EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan),¹⁹ which is a graphical user interface for R (The R Foundation for Statistical Computing, version 2.13.0, Vienna, Austria).

RESULTS

Patient Characteristics

During the study period, among 203 patients who underwent a CD-related surgery, 22 patients (10.8%) received vedolizumab as postoperative treatment. There were 58, 38, and 16 patients who received anti-TNF-α agents, immunomodulators, and metronidazole, respectively, whereas 69 patients were monitored without any medication. No predefined criteria were adopted in deciding whether to use a specific postoperative therapy or not, and it was solely based on the attending physician’s decision. The background characteristics of the patients receiving each therapy are shown in Table 1. The groups were significantly different regarding age and proportion of patients with perianal disease suggesting that physicians take into account the known risk factors for postoperative recurrence when choosing postoperative therapies. Patients treated postoperatively with vedolizumab or anti-TNF-α agents mostly had been preoperatively treated with the same agent.

Table 1:

Baseline Patient Characteristics

	VDZ (N = 22)	Anti-TNF (N = 58)	IM (N = 38)	Metronidazole (N = 16)	No Medication (N = 69)	P
Age, year (IQR)	25.5 (23.0–30.7)	36.0 (28.5–48.5)	40.5 (25.0–49.5)	44.0 (34.7–53.0)	41.0(30.0–54.0)	0.001
Gender, female (%)	14 (63.6)	28 (48.3)	20 (52.6)	9 (56.2)	35 (50.7)	0.80
Disease duration, year (IQR)	9.0 (2.5–12.0)	12.0 (4.0–18.0)	9.0 (1.0–15.0)	8.0 (5.5–18.2)	8.0 (2.0–19.0)	0.43
Disease location (%)
Ileum	4 (18.2)	16 (27.6)	14 (36.8)	6 (37.5)	18 (27.7)
Colon	5 (22.7)	8 (13.8)	5 (13.2)	4 (25.0)	18 (27.7)	0.50
Ileum-colon	13 (59.1)	34 (58.6)	19 (50.0)	6 (37.5)	29 (44.6)
Upper GI	2 (9.1)	4 (6.9)	4 (10.5)	0 (0)	1 (1.4)	0.15
Disease behavior (%)
Inflammatory	6 (27.3)	9 (15.5)	4 (10.5)	3 (18.8)	17 (24.6)
Stricturing	10 (45.5)	24 (41.4)	11 (28.9)	6 (37.5)	23 (33.3)	0.28
Penetrating	6 (27.3)	25 (43.1)	23 (60.5)	7 (43.8)	29 (42.0)
Perianal disease (%)	12 (54.5)	16 (27.6)	6 (15.8)	8 (50.0)	11 (15.9)	<0.001
Current smoker (%)	3 (13.6)	7 (12.1)	4 (10.5)	1 (6.2)	15 (21.7)	0.42
Previous use of vedolizumab (%)	17 (77.3)	5 (8.6)	1 (2.6)	3 (18.8)	8 (11.6)	<0.001
Previous use of anti-TNF-α agents (%)	20 (90.9)	50 (86.2)	17 (44.7)	10 (62.5)	36 (52.2)	<0.001
Number of anti-TNF-α agents use
1	3 (13.6)	29 (50.0)	7 (18.4)	2 (12.5)	16 (23.3)	<0.001
2 or more	17 (77.3)	21 (36.2)	10 (26.3)	8 (50.0)	20 (29.0)	<0.001
Naive to biologics (%)	2 (9.1)	8 (13.8)	21 (55.3)	6 (37.5)	33 (47.8)	0.72
History of surgery (%)	13 (59.1)	37 (63.8)	17 (44.7)	8 (50.0)	46 (66.7)	0.20
Number of previous surgery, median (IQR)	1 (0–2)	1 (0–2)	1 (0–2)	0.5 (0–1.25)	1 (0–2)	0.54
Type of surgery (%)
Small bowel resection	6 (27.3)	17 (29.3)	11 (28.9)	5 (31.2)	10 (14.5)
Colon resection	9 (40.9)	21 (36.2)	5 (13.2)	4 (25.0)	36 (52.2)	0.15
Ileocecectomy or ileocolectomy	7 (31.8)	20 (34.5)	22 (57.9)	7 (43.8)	23 (33.3)
Postoperative concomitant medication
Antibiotics (%)a	5 (22.7)	7 (12.1)	5 (13.2)	16 (100)	0 (0)	<0.001
Steroid (%)b	3 (13.6)	6 (10.3)	7 (18.4)	1 (6.2)	11 (15.9)	0.73
Immunomodulator (%)	10 (45.5)	45 (77.6)	38 (100)	0 (0)	0 (0)	<0.001

VDZ; vedolizumab, Anti-TNF; anti-TNF-α agents, IM; immunomodulators, IQR; interquartile range.

^aCiprofloxacin, metronidazole, ^bContinuous usage due to steroid dependency.

As shown in Table 2, rates of clinical and biologic recurrence were not different among each treatment group (P = 0.20 and P = 0.88, respectively). Rates of endoscopic and histologic recurrence were significantly different between groups (P = 0.049 and P = 0.02, respectively) with vedolizumab demonstrating the highest values (Table 2).

Table 2:

Postoperative Outcomes at 6–12 Months

	VDZ (N = 22)	Anti-TNF (N = 58)	IM (N = 38)	Metronidazole (N = 16)	No Medication (N = 69)	P
Clinical recurrence (%)	10/21 (47.6)	18/49 (36.7)	14/36 (38.9)	10/14 (71.4)	29/61 (47.5)	0.20
Biologic recurrence (%)	6/12 (50.0)	11/29 (37.9)	8/19 (42.1)	3/5 (60.0)	9/19 (47.4)	0.88
Endoscopic recurrence (%)	15/20 (75.0)	13/38 (34.2)	14/25 (56.2)	5/10 (50.0)	26/48 (54.2)	0.049
Histologic recurrence (%)	15/17 (88.2)	15/28 (53.6)	19/28 (67.9)	2/5 (40.0)	16/32 (50.0)	0.046

Comparison of Risk of Postoperative Recurrence Between Vedolizumab and anti-TNF agents

We next sought to compare the risk of postoperative recurrence between those treated postoperatively with vedolizumab vs anti-TNF-α agents as anti-TNF-α agents are now commonly recommended as postoperative treatment in high risk patients. The majority of patients in the anti-TNF-α agent group were treated with infliximab or adalimumab (25 infliximab, 30 adalimumab, and 3 certolizumab pegol). All patients started therapy with vedolizumab or anti-TNF-α agents within 4 weeks of their surgery and continued them at the same dosage and interval as the maintenance therapy of each drug. No patient had dose escalation during the study period. The characteristics of 2 groups did not differ regarding sex, disease duration, smoking, numbers of previous surgery, and history of anti-TNF use (Table 3). Patients in the vedolizumab group were younger and more frequently had perianal disease but were less likely to be taking an immunomodulator as combination therapy. They were more likely to have been treated with vedolizumab preoperatively for a median of 5.5 months (IQR 2.0–9.5) and to continue it as postoperative treatment (90.9%). Similarly, patients in the anti-TNF-α agent group were more likely to have been treated with anti-TNF-α agents preoperatively for a median of 9.5 months (IQR 7.0–16.5) and continue the same agent as postoperative treatment (75.9%). Significantly less patients in the anti-TNF-α agent group had a history of vedolizumab use (P < 0.001). There were 2 (9.1%) and 8 (13.8%) patients who were naive to biologics in the vedolizumab and anti-TNF-α agent group, respectively. Three of 22 patients on vedolizumab (13.6%) and 13 of 58 patients on anti-TNF-α agents (22.4%) experienced adverse effects, respectively (Supplementary Table 1). There were no severe adverse events such as serious infections or severe infusion reactions leading to discontinuation of therapy in both groups during the study period.

Table 3:

Baseline Patient Characteristics of Vedolizumab and anti-TNF Users

	VDZ (N = 22)	Anti-TNF (N = 58)	P
Age, year (IQR)	25.5 (23.0–30.7)	36.0 (28.5–48.5)	0.001
Gender, female (%)	14 (63.6)	28 (48.3)	0.32
Disease duration, year (IQR)	9.0 (2.5–12.0)	12.0 (4.0–18.0)	0.08
Disease location (%)
Ileum	4 (18.2)	16 (27.6)
Colon	5 (22.7)	8 (13.8)	0.50
Ileum-colon	13 (59.1)	34 (58.6)
Upper GI	2 (9.1)	4 (6.8)	0.66
Disease behavior (%)
Inflammatory	6 (27.3)	9 (15.5)
Stricturing	10 (45.5)	24 (41.4)	0.31
Penetrating	6 (27.3)	25 (43.1)
Perianal disease (%)	12 (54.5)	16 (27.6)	0.04
Current smoker (%)	3 (13.6)	7 (12.1)	1.00
Previous use of vedolizumab (%)	17 (77.3)	5 (8.6)	<0.001
Previous use of anti-TNF-α agents (%)	20 (90.9)	50 (86.2)	0.72
Naive to biologics (%)	2 (9.1)	8 (13.8)	0.72
History of surgery (%)	13 (59.1)	37 (63.8)	0.80
Number of previous surgery, median (IQR)	1 (0–2)	1 (0–2)	0.93
Type of surgery (%)
Small bowel resection	6 (27.3)	17 (29.3)
Colon resection	9 (40.9)	21 (36.2)	0.95
Ileocecectomy or ileocolectomy	7 (31.8)	20 (34.5)
Postoperative concomitant medication
Antibiotics (%)a	5 (22.7)	7 (12.1)	0.30
Steroid (%)b	3 (13.6)	6 (10.3)	0.70
Immunomodulator (%)	10 (45.5)	45 (77.6)	0.01

VDZ; vedolizumab, Anti-TNF; anti-TNF-α agents, IQR; interquartile range.

^aCiprofloxacin, metronidazole, ^bContinuous usage due to steroid dependency.

As shown in Figure 1A, 58 of the 80 patients (72.5%) had an ileocolonoscopy at a median of 6 months (IQR 6.0–7.0) following surgery. Seventy of 80 patients were available for evaluation of clinical remission. Biological and histologic evaluation were available for 41 and 42 patients, respectively.

FIGURE 1.

(A) At 6–12 months of follow-up, rates of clinical and biologic remission were similar among patients receiving vedolizumab or anti-TNF-α agents (52% vs 63 %, P = 0.50, 50% vs 62%, P = 0.43, respectively), but rates of endoscopic and histologic remission were higher among patients receiving anti-TNF-α agents (25% vs 66%, P = 0.01 and 12% vs 46%, P = 0.005, respectively).

Assessment of Factors Influencing Postoperative Recurrence with Univariate and Multivariate analysis

As shown in Figure 1A, the crude analysis showed that endoscopic remission was seen in a smaller proportion of patients treated with vedolizumab as compared to anti-TNF-α agents (vedolizumab 25 vs anti-TNF-α agents 66%, P = 0.01). Clinical remission was observed in similar proportions of patients among the 2 groups (52% vs.. 63%, P = 0.50). Rates of biological remission were similar between the 2 groups (50 vs 62%, P = 0.43), but those of histologic remission were lower with vedolizumab (12 vs 46%, P = 0.005).

As shown in Table 4, postoperative vedolizumab use was associated with an increased risk of endoscopic recurrence on univariate analysis (odds ratio (OR) 5.58, 95% confidence interval (CI) 1.51–24.3, P = 0.005). Multivariate analysis was performed to assess the effect of multiple variables on the outcome. We incorporated univariate predictor variables with a P value of < 0.10 (perianal disease, immunomodulator, vedolizumab, and anti-TNF-α agent use) and demonstrated that postoperative vedolizumab use remained as a risk factor of endoscopic recurrence (OR 5.77, 95%CI 1.71–19.4, P = 0.005). Age, perianal disease, and previous vedolizumab use, which were covariates that differed significantly between the 2 groups, did not affect the outcome of endoscopic remission. No factors were associated with the risks of clinical and biological recurrences on univariate and multivariate analysis.

Table 4:

Factors Influencing Postoperative Recurrence Assessed with Univariate and Multivariate Analysis

Factors	Univariate			Multivariate
	OR	95%CI	P	OR	95%CI	P
Clinical
Age (>30)	1.78	0.61–5.44	0.32
Gender (female)	1.62	0.55–4.95	0.46
Smoking	1.00	0.19–4.74	1.00
Perianal disease	1.34	0.45–4.00	0.62
Immunomodulator	0.48	0.15–1.48	0.19
Vedolizumab	1.55	0.49–4.97	0.43
Anti-TNF-α agents	0.64	0.20–2.05	0.43
Biological
Age (>30)	3.54	0.84–16.7	0.06
Gender (female)	1.66	0.41–7.16	0.53
Smoking	0.93	0.07–9.25	1.00
Perianal disease	1.95	0.47–8.53	0.35
Immunomodulator	0.72	0.16–3.14	0.74
Vedolizumab	1.61	0.34–7.84	0.50
Anti-TNF-α agents	0.62	0.12–2.96	0.50
Endoscopic
Age (>30)	0.58	0.18–1.87	0.42
Gender (female)	0.89	0.27–2.87	1.00
Smoking	0.78	0.10–5.14	1.00
Perianal disease	3.70	1.08–13.8	0.03a	2.91	0.83–10.2	0.09
Immunomodulator	0.26	0.06–0.91	0.03a	0.34	0.10–1.21	0.10
Vedolizumab	5.58	1.51–24.3	0.005a	5.77	1.71–19.4	0.005a
Anti-TNF-α agents	0.18	0.04–0.66	0.005a	0.17	0.05–0.59	0.005a
Histologic
Age (>30)	0.34	0.07–1.41	0.12
Gender (female)	1.39	0.33–5.81	0.75
Smoking	0.94	0.09–12.4	1.00
Perianal disease	2.54	0.58–13.6	0.34
Immunomodulator	0.93	0.21–3.86	1.00
Vedolizumab	8.90	1.58–95.4	0.005a
Anti-TNF-α agents	0.11	0.01–0.63	0.005a

OR; odds ratio. ^a indicates P < 0.05.

A similar result was seen when the analysis was limited to those who underwent ileocecectomy or ileocolectomy (Supplementary Figure 1A). Rate of endoscopic remission, as defined by a Rutgeerts score of ≤i1, with vedolizumab trended to be lower than anti-TNF-α agents (2/4 (50%) vs. 11/12 (92%), P = 0.07), although it should be noted that the number of patients included for this outcome was small.

To exclude the influence of immunomodulators, we performed an analysis limited to those on monotherapy (12 patients on vedolizumab and 13 patients anti-TNF-α agents). Rates of clinical and biologic remission were similar between patients receiving vedolizumab or anti-TNF-α agents (41 vs. 54 %, P = 0.68, 37 vs 71%, P = 0.31, respectively), but rates of endoscopic and histologic remission were higher among patients receiving anti-TNF-α agents (0 vs. 67%, P = 0.002 and 0 vs vedolizumab 75%, P = 0.007, respectively).

Propensity Score-matched Analysis

Due to the retrospective nature of the study, the difference in the patients’ background characteristics may have influenced the outcome of each therapy. Therefore, to reduce the bias of confounding variables, we performed propensity score -matched analysis. Following propensity score matching, the characteristics of the 2 groups did not differ regarding to any clinical parameter, except that significantly fewer patients in the anti-TNF-α agent group had a history of vedolizumab use (P < 0.001) (Supplementary Table 2).

As shown in Supplementary Figure 1B, the rate of endoscopic remission (vedolizumab 25 vs anti-TNF-α agents 69%, P = 0.03) was lower in patients treated with vedolizumab compared to anti-TNF-α agents supportive of the aforementioned findings on multivariate analysis.

The rates of clinical (52% vs 72%, P = 0.22) and biological (50 vs 79%, P = 0.22) remission were similar in both groups, but those of histologic (12 vs 60%, P = 0.02) remission were statistically lower in patients treated with vedolizumab (Supplementary Figure 1B).

DISCUSSION

In the present study, we found that vedolizumab now is increasingly used as postoperative prophylactic treatment for CD, especially in those at high risk for postoperative recurrence. Vedolizumab has been increasingly used to induce and maintain remission of CD, however, our study is the first to assess its utility in the postoperative setting. We compared the risk of postoperative recurrence between CD patients treated with vedolizumab and anti-TNF-α agents and showed that both agents maintained clinical and biological remission at 6–12 months at similar rates, however, vedolizumab use was associated with reduced rates of endoscopic remission on multivariate and propensity score-matched analyses. Patients treated with vedolizumab may have been at higher risk of recurrence, but controlled studies are needed to further investigate the utility of vedolizumab in the postoperative setting.

Due to chronic progressive inflammation, approximately 60% of patients with CD will require surgery during their disease course. Furthermore, a substantial proportion of patients undergo multiple surgeries. The postoperative setting is unique in that patients are in a surgically induced remission. Most medications are evaluated for active CD, but thiopurines, nitroimidazole antibiotics, and infliximab have been shown to reduce postoperative recurrence.^6,8,9,20 In the present study, we evaluated whether vedolizumab would reduce the recurrence of CD in the postoperative setting. The results of our study indicated that whereas the rates of clinical and biological remission at 6–12 months after surgery were similar between vedolizumab and anti-TNF-α agents, those of endoscopic and possibly of histologic remission were lower among patients treated with vedolizumab. Patients in the vedolizumab group were of younger age and more frequently had perianal disease, which may have placed them at higher risk for postoperative recurrence, however, multivariate analysis and propensity score-matched analysis demonstrated that vedolizumab use remained as a risk factor for postoperative recurrence.

There are several limitations to our study. Firstly, the retrospective single center design of our study will not allow us to draw any definite conclusion. We performed multivariate analysis and propensity matched analysis to reduce the effect of confounders, however, a head-to-head randomized controlled trial between vedolizumab and anti-TNF-α agent(s) is required to compare their effectiveness as confounders may still have persisted. We are not aware of such an ongoing study or even of a placebo controlled study of vedolizumab in the postoperative setting, so the results of our study provide useful information for clinicians. Data were collected by multiple physicians, who were all experienced gastroenterologists or pathologists, however, there may still be interobserver variance in endoscopic and clinical scoring²¹ and management styles and approaches to postoperative CD. We adopted SES-CD to simultaneously assess the degree of inflammation in the ileum and colon,^5,15 but despite being used in studies that included patients with a history of surgery,²² it has not been formally validated in the postoperative setting. Of note, analysis limited to those with ileocolectomy/ileocecectomy and adopting the Rutgeerts score showed similar results. The majority of our patients were treated postoperatively without an induction dosing, which may have reduced the effect of vedolizumab, although, this also applies to anti-TNF-α agents. Vedolizumab is reported to have a rather delayed onset of action compared to other biologics, but it is unlikely that it has not fully exerted its effectiveness at 6–12 months. The effects of leukocyte migration may have a different role in initiation vs perpetuation of disease, although more research is required to elucidate this hypothesis. Most patients treated postoperatively with vedolizumab alrady had been treated preoperatively with both vedolizumab and anti-TNF-α agents, which may indicate that they had a more medical refractory disease, although it remains to be determined whether preoperative failure of biologics is a risk factor of postoperative recurrence.²³ Preoperatively being treated with vedolizumab may also indicate that they were nonresponders to this class of drugs or may have developed antidrug antibodies, although the same scenario applies to patients treated pre- and postoperatively with anti-TNF-α agents, and it has been reported that formation of antidrug antibody is rare with vedolizumab.¹² Furthermore, most patients continued the same agent pre- and postoperatively, which is not surprising considering the lack of available options in patients previously failing multiple agents. Postoperative treatment with infliximab has been suggested to be effective in patients preoperatively exposed to it,²⁴ but the same may not apply to vedolizumab. Another major limitation is the short follow-up period and studies with longer follow-up may better address clinical recurrence, although it has been suggested that early endoscopic recurrence is an indicator for subsequent clinical recurrence.⁵

Collectively, our retrospective cohort study of real-world data showed that vedolizumab is commonly used as postoperative treatment in high risk CD patients. Our results suggested that risk of endoscopic recurrence appeared to be higher with vedolizumab as compared to anti-TNF-α agents. As there were differences in the background characteristics of the patients treated with vedolizumab and anti-TNF-α agents, further investigation including controlled trials is required before recommending the use of vedolizumab in preventing postoperative recurrence of CD.

SUPPLEMENTARY DATA

Supplementary data are available at Inflammatory Bowel Diseases online.

Specific Author contribution

Akihiro Yamada, Yuga Komaki, and Fukiko Komaki, analysis of data and drafting of manuscript; Nayan Patel, analysis of data and approval of manuscript; Joel Pekow, Sushila Dalal, Russell D. Cohen, Lisa Cannon, Konstantin Umansky, Radhika Smith, Roger Hurst, Neil Hyman, and David T. Rubin, critical review and approval of manuscript; Atsushi Sakuraba, study concept and design, analysis of data, and writing of manuscript.

Conflict of Interest

Akihiro Yamada, Yuga Komaki, Fukiko Komaki, Nayan Patel, Lisa Cannon, Konstantin Umanskiy, Radhika Smith, Roger Hurst, and Neil Hyman have no conflicts to report. Joel Pekow research grants from Takeda and Abbvie. Sushila Dalal; investigator-initiated research grant from Pfizer. Russell D. Cohen; consultant and/or scientific advisory board for Abbvie, Celgene, Entera Health, Hospira, Janssen, Pfizer, Sandoz Biopharmaceuticals, Takeda, and UCB Pharma. Speaker’s bureau for Abbvie, and Takeda. David T. Rubin; consultant and grant support from Takeda, Janssen, and AbbVie, consultant for Pfizer and Amgen. Atsushi Sakuraba; collaborative research with Abbvie. Consultant fee from Abbvie and Celltrion.

Supported by

Funding for Yuga Komaki was provided by the Pediatric Oncology Research Fellowship of the Children’s Cancer Association of Japan. No other grants are related to this study.