Comparative Effectiveness of Biologic Therapies in Chronic Pouchitis and Crohn’s-Like Disease of the Pouch: A Bi-center Retrospective Cohort Study

Abstract

Background

Chronic pouchitis and Crohn’s-like disease of the pouch are frequent complications following restorative total proctocolectomy with ileal pouch–anal anastomosis (IPAA) for ulcerative colitis (UC). Biologics are increasingly used in refractory cases, but head-to-head comparative effectiveness data between biologic therapies are limited.

Aims

We aimed to compare week 52 clinical response and long-term treatment persistence among patients treated with vedolizumab, ustekinumab, or adalimumab.

Methods

This retrospective bi-center cohort study included adult patients with chronic pouchitis or Crohn’s-like disease of the pouch with mPDAI ≥ 5 at biologic initiation. The primary endpoint was week 52 clinical response based on cPDAI, defined as a ≥ 1-point reduction from baseline. Key secondary endpoints were treatment persistence and week 52 mPDAI response (≥ 2-point reduction). Additional secondary endpoints included cPDAI remission, mPDAI remission, ePDAI response and remission, and biochemical outcomes. Between-group comparisons were performed using chi-square tests for categorical variables and Kruskal–Wallis tests for continuous variables, with post hoc pairwise comparisons adjusted for multiple testing. Treatment persistence was analyzed using Kaplan–Meier survival curves with log-rank test.

Results

Seventy-seven patients were included (vedolizumab n = 36; ustekinumab n = 22; adalimumab n = 19). Baseline characteristics including distribution of chronic pouchitis and Crohn’s-like disease of the pouch were broadly comparable across groups, although ustekinumab was more frequently used as third-line therapy (p = 0.001). At week 52, cPDAI response occurred in 88.9% of patients treated with ustekinumab, 44.8% treated with vedolizumab, and 56.3% treated with adalimumab (p = 0.01). cPDAI remission was achieved in 33.3% of ustekinumab-treated patients, 13.8% of vedolizumab-treated patients, and 6.3% of adalimumab-treated patients (p = 0.09). No statistically significant differences were observed in biochemical remission across groups. Three-year treatment persistence was highest in the ustekinumab group (81.5%) compared with adalimumab (52%) and vedolizumab (35%) (p = 0.01).

Conclusions

In this real-world cohort, ustekinumab was associated with numerically higher long-term clinical response and treatment persistence compared with vedolizumab or adalimumab. These findings support the need for prospective trials to guide biologic selection in chronic pouchitis and Crohn’s-like disease of the pouch.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10620-026-09804-x.

Introduction

Pouchitis is the most common long-term complication following restorative total proctocolectomy with ileal pouch–anal anastomosis (IPAA) for ulcerative colitis (UC), affecting up to 30% of patients [1–3]. It is thought to involve microbial dysbiosis and immune dysregulation in genetically susceptible individuals, with postoperative fecal stasis and colonic metaplasia contributing to a pro-inflammatory environment [4–6]. Although acute pouchitis typically responds to antibiotics, a subset of patients develop chronic pouchitis (CP), which may be antibiotic-dependent or antibiotic-refractory, whereas others develop Crohn’s-like disease of the pouch (CLDP), a distinct phenotype characterized by stricturing, fistulizing, or inflammatory involvement of the prepouch ileum [2]. Both entities may require biologic therapy or surgical intervention7.⁷

Biologic therapies—including anti–tumor necrosis factor (anti-TNF) agents, vedolizumab, and ustekinumab—are increasingly used [8–11]. Randomized data are emerging, including the EARNEST trial demonstrating the efficacy of vedolizumab in CP [12]. However, comparative effectiveness data across biologic agents remain limited, and direct head-to-head studies are scarce, although a prospective multicenter study suggested a potential benefit of ustekinumab over anti-TNF therapy in CLDP ^[[13]].

We therefore conducted a bi-center retrospective cohort study to compare long-term clinical response and treatment persistence among patients with CP or CLDP treated with vedolizumab, ustekinumab, or adalimumab.

Methods

Study Design and Population

This was a retrospective, bi-center cohort study approved by the Tel Aviv Medical Center institutional review board (IRB approval #376-17). Informed consent was waived due to the anonymized study design. We reviewed the medical records of all adult patients (≥ 18 years) with a history of UC and IPAA who were followed at the comprehensive pouch clinics of Tel Aviv Medical Center and Shaare Zedek Medical Center between 2015 and 2025. Inclusion criteria were: (1) diagnosis of CP or CLDP; (2) a modified Pouchitis Disease Activity Index (mPDAI) score ≥ 5 at treatment initiation; (3) treatment with vedolizumab, ustekinumab, or adalimumab; and (4) availability of ≥ 8 weeks of follow-up data.

Exclusion criteria included: (1) histological evidence of Crohn’s disease in the colectomy specimen; (2) presence of a permanent end-ileostomy; (3) Koch pouch; (4) familial adenomatous polyposis (FAP); (5) documented non-adherence to biologic therapy during routine clinical follow-up, defined as self-discontinuation or inconsistent treatment use.

Data Collection

Data were collected at baseline and at approximately weeks 12, 26, and 52 following biologic initiation. Follow-up evaluations—including pouchoscopies, outpatient clinic visits, and laboratory assessments—were performed as part of routine clinical care according to physician discretion and clinical indication, rather than within a predefined prospective protocol. Study time points were defined using prespecified windows to reflect real-world practice: ± 4 weeks for the induction assessment (week 12) and ± 8 weeks for the week 26 and week 52 assessments. Treatment persistence was assessed for up to 3 years from initiation. Variables included demographics, comorbidities, pouch phenotype, disease activity, prior and concurrent treatments, laboratory results, endoscopic findings, and reported adverse events. Prior biologic exposure refers to agents administered after colectomy for pouch inflammation. Data regarding biologic exposure prior to colectomy were not consistently available and were therefore not included in the analysis.

Definitions

CP was defined as either ≥ 4 weeks of continuous antibiotic use for a single episode or ≥ 5 antibiotic courses per year. CLDP was defined based on documentation of strictures and/or fistulas not attributable to surgical complications, or by endoscopic evidence of pre-pouch ileitis—specifically, ≥ 5 cm of diffuse inflammation, more than five aphthous ulcers, stellate or linear ulcers, or circumferential ulceration at the pouch inlet, as recorded in the medical record [14,15]. Pre-pouch ileitis was documented based on endoscopic reports. However, severity and extent were not systematically quantified in a standardized manner due to the retrospective design.

Antibiotic response was categorized as responsive, dependent, or refractory. A responsive course was defined as clinical improvement following a 2-week antibiotic regimen; antibiotic dependence was defined as ≥ 4 relapses per year or the need for ≥ 4-week antibiotic courses or combination antibiotic therapy; and antibiotic-refractory disease was defined by a lack of clinical response to antibiotic treatment [16].

Treatment persistence was defined as the duration of uninterrupted biologic therapy without discontinuation or interruption.

Assessment of Pouchitis Activity

Pouchitis was assessed using the PDAI [17], which includes clinical, endoscopic, and histologic components. The clinical sub-score (cPDAI) evaluates stool frequency, rectal bleeding, urgency, and fever. The endoscopic sub-score (ePDAI) assesses edema, granularity, friability, loss of vascular pattern, mucus exudates, and ulceration. The histologic sub-score (hPDAI) considers neutrophilic infiltration and ulcerations per low-power field. In this study, the mPDAI—which combines the clinical and endoscopic components—was used for disease activity assessment [18].

Outcomes

The primary endpoint was clinical response at week 52, defined according to the cPDAI as a ≥ 1-point reduction from baseline. [19]

Key secondary endpoints included treatment persistence for up to 3 years from treatment initiation and week 52 mPDAI response, defined as a ≥ 2-point reduction from baseline.

Additional secondary endpoints comprised endoscopic ePDAI response (≥ 1-point reduction from baseline); clinical and endoscopic remission, defined as a ≥ 1-point reduction from baseline resulting in a score of 0 for cPDAI or ePDAI [19]; mPDAI remission, defined as a ≥ 2-point reduction from baseline with a total score ≤ 4; and biochemical remission, defined as fecal calprotectin < 250 µg/g [20].

Statistical Analysis

All statistical analyses were conducted using IBM SPSS Statistics version 29 (IBM Corp., Armonk, NY) and GraphPad Prism version 8 (GraphPad Software, San Diego, CA). Shapiro–Wilk was used to test for normality. Continuous variables were not normally distributed; therefore, non-parametric tests were used: the Kruskal–Wallis test for comparisons across multiple groups, and the Mann–Whitney U test for pairwise comparisons. Categorical variables were compared using the chi-square test or Fisher’s exact test, as appropriate. When overall between-group differences were statistically significant, post hoc pairwise comparisons between treatment groups were performed with Bonferroni correction applied to adjust for multiple comparisons (α = 0.0167). The primary endpoint (week 52 cPDAI response) was analyzed using complete-case data. Sensitivity analyses were performed using non-responder imputation (NRI), whereby patients who discontinued therapy due to lack of response prior to week 52 were considered non-responders at week 52. Treatment persistence was evaluated using Kaplan–Meier survival analysis with log-rank testing. Biochemical outcomes were analyzed using available data without imputation.

This study was conducted and reported in accordance with the STROBE guidelines for observational studies (Supplementary Table S1).

Results

Study Cohort

A total of 77 patients met the inclusion criteria and were included in the study. The cohort included patients treated with vedolizumab (n = 36), ustekinumab (n = 22), and adalimumab (n = 19). Baseline demographic characteristics, including age, sex, BMI, and smoking status, as well as disease-related factors such as phenotype, indication for pouch surgery, prior treatments, pouchitis activity scores, and inflammation indices were comparable across all treatment groups (Table 1). However, ustekinumab was more frequently used as a third-line treatment compared with vedolizumab and adalimumab (45.5% vs. 2.8% and 5.3%, respectively; p = 0.001).

Table 1.

Baseline demographic and clinical characteristics of the study cohort

	Vedolizumab N = 36	Ustekinumab N = 22	Adalimumab N = 19	P-value
Gender (females)	15 (41.7%)	11 (50%)	7 (36.8%)	0.684
Age (years)	48 [31.5–64]	39 [30.75–51.75]	49 [33.75–59.5]	0.392
Pouch age (years)	12.5 [6.25–20]	15.5 [5.75–21]	15 [7–21]	0.982
BMI	22.4 [19.8–23.8] (of 31)	19 [18.3–22.1] (of 19)	22.6 [19.4–25] (of 15)	0.024
Smokers	28.1% (of 32)	5 (26.3%) (of 19)	1 (5.9%) (of 17)	0.176
Indication for surgery
Refractory disease	34 (94.4%)	20 (90.9%)	17 (89.5%)
Dysplasia / malignancy	2 (5.6%)	2 (9.1%)	2 (10.5%)	0.779
Phenotype:
Chronic pouchitis	17 (47.2%)	9 (40.9%)	9 (47.4%)
CLDP	19 (52.8%)	13 (59.1%)	10 (52.6%)	0.880
Antibiotic response:
Refractory	20 (57.1%) (of 35)	9 (42.9%) (of 21)	7 (38.9%) (of 18)
Dependent	10 (28.6%) (of 35)	10 (47.6%) (of 21)	10 (55.6%) (of 18)
Responsive	5 (14.3%) (of 35)	2 (9.5%) (of 21)	1 (5.6%) (of 18)	0.355
Previous biologic treatment post-colectomy
0	18 (50%)	2 (9.1%)	14 (73.3%)
1	16 (44.4%)	8 (36.4%)	4 (21.1%)
2	1 (2.8%)	10 (45.5%)	1 (5.3%)
> 2	1 (2.8%)	2 (9.1%)	0 (0%)	< 0.001
Antibiotics	10 (27.8%)	7 (31.8%)	7 (36.8%)	0.786
VSL #3	9 (25%)	5 (22.7%)	5 (26.3%)	0.887
Budesonide:	3 (8.3%)	1 (4.5%)	0 (0%)	0.070
Disease activity
cPDAI	3 [2–4]	4 [2–4]	3 [3, 4]	0.899
ePDAI	4 [3–5] (of 35)	3.5 [2.25–6] (of 20)	3 [3–5]	0.825
mPDAI	7 [6–8] (of 35)	7 [6–8.75] (of 20)	7 [5–8]	0.766
CRP	7.8 [2.6–14.7] (of 34)	9.5 [3.97–20.85] (of 16)	8.6 [3.85–22.4] (of 17)	0.731
Calprotectin	616.5 [260.5–1215.25] (of 26)	782 [444.75–1282.25] (of 14)	338 [192–447.5] (of 9)	0.107

Values are presented as n (%) or median (interquartile range), unless otherwise indicated. Denominators are shown where data were missing. CLDP Crohn’s-like disease of the pouch, BMI body mass index, PDAI pouchitis disease activity index, cPDAI clinical PDAI, ePDAI endoscopic PDAI, mPDAI modified PDAI, CRP C-reactive protein. P values were calculated using the chi-square test for categorical variables and the Kruskal–Wallis test for continuous variables

Primary Endpoint: Week 52 cPDAI Response

Between-group differences in cPDAI response and remission rates at weeks 12, 26, and 52 are shown in Fig. 1.

Fig. 1.

Clinical response and remission rates over time. (A) Clinical response rates and (B) clinical remission rates based on the cPDAI are shown for each treatment group at weeks 12, 26, and 52. At weeks 12, 26, and 52, cPDAI response and remission data were available for 31, 27, and 29 vedolizumab-treated patients; 15, 19, and 18 ustekinumab-treated patients; and 12, 11, and 16 adalimumab-treated patients, respectively. Group comparisons were performed using the chi-square test. *P < 0.05

Complete cPDAI data at week 52 were available for 29 of 36 vedolizumab-treated patients, 18 of 22 ustekinumab-treated patients, and 16 of 19 adalimumab-treated patients. Response rates were 88.9% for ustekinumab, 44.8% for vedolizumab, and 56.3% for adalimumab (overall Pearson chi-square p = 0.01).

Post hoc pairwise comparisons were performed with Bonferroni correction for multiple testing. A statistically significant difference was observed between ustekinumab and vedolizumab (raw p = 0.003). Comparisons between ustekinumab and adalimumab (raw p = 0.03) and between vedolizumab and adalimumab (raw p = 0.463) did not meet the adjusted significance threshold.

Sensitivity analysis using non-responder imputation (NRI), in which patients who discontinued therapy due to lack of response were classified as non-responders at week 52, yielded response rates of 81.8% for ustekinumab (22 patients), 38.9% for vedolizumab (36 patients), and 61.1% for adalimumab (18 of 19 patients), with an overall p-value of 0.005, consistent with the primary analysis.

cPDAI Remission

In complete-case analysis, week 52 cPDAI remission rates were 33.3% for ustekinumab, 13.8% for vedolizumab, and 6.3% for adalimumab (p = 0.09). In the NRI analysis, remission rates were 31.8, 11.1, and 5.6%, respectively (p = 0.04). Post hoc pairwise comparisons yielded raw p-values of 0.05 for ustekinumab versus vedolizumab, 0.04 for ustekinumab versus adalimumab, and 0.51 for vedolizumab versus adalimumab. After correction, none of these comparisons met the adjusted significance threshold.

mPDAI Outcomes

Week 52 mPDAI response was assessed in 19 vedolizumab-treated patients, 11 ustekinumab-treated patients, and 11 adalimumab-treated patients. Response rates were 81.8% for ustekinumab, 73.7% for vedolizumab, and 54.5% for adalimumab (p = 0.35). Week 52 mPDAI remission rates were 63.6% for ustekinumab, 52.6% for vedolizumab, and 45.5% for adalimumab (p = 0.69).

ePDAI Outcomes

Week 52 ePDAI response was assessed in 19 vedolizumab-treated patients, 12 ustekinumab-treated patients, and 11 adalimumab-treated patients. Response rates were 84.2% for vedolizumab, 83.3% for ustekinumab, and 90.9% for adalimumab (p = 0.85). Week 52 ePDAI remission rates were 10.5, 8.3, and 9.1%, respectively (p = 0.98).

Although no statistically significant between-group differences were observed, within-group analysis demonstrated a statistically significant reduction in ePDAI scores over time in the vedolizumab group, with median ePDAI decreasing from 4 [IQR 2.5–5] at baseline to 1.5 [IQR 1–2] at week 52 (p = 0.02). No statistically significant within-group changes were observed in the ustekinumab or adalimumab groups (Supplementary Figure S1). Given the absence of significant between-group differences and incomplete endoscopic follow-up, these findings should be interpreted as exploratory.

Biochemical Outcomes

Biochemical remission at week 52 was assessed in 19 vedolizumab-treated patients, 9 ustekinumab-treated patients, and 4 adalimumab-treated patients (Fig. 2). Remission rates were 44.4% for ustekinumab, 42.1% for vedolizumab, and 25% for adalimumab. Between-group differences were not statistically significant. Analyses were performed using available data without imputation due to variable laboratory follow-up inherent to the observational design.

Fig. 2.

Fecal calprotectin remission rates over time. Remission rates, defined as fecal calprotectin < 250 µg/g, are shown for each treatment group at weeks 12, 26, and 52. At weeks 12, 26, and 52, fecal calprotectin remission data were available for 15, 17, and 19 vedolizumab-treated patients; 9, 10, and 9 ustekinumab-treated patients; and 4 adalimumab-treated patients at each timepoint, respectively. Group comparisons were performed using the chi-square test

Treatment Persistence

Treatment persistence was evaluated using Kaplan–Meier survival analysis (Fig.3). Differences between groups were observed (log-rank p = 0.01). Three-year persistence rates were 81.5% for ustekinumab, 52% for adalimumab, and 35% for vedolizumab.

Fig. 3.

Treatment persistence over time. Kaplan–Meier curves depict treatment persistence over a 3-year follow-up period for each treatment group. Data were available on 36 vedolizumab-treated patients, 22 ustekinumab-treated patients, and 19 adalimumab-treated patients. Comparisons were performed using the log-rank (Mantel–Cox) test. *P < 0.01

Week 12 and week 26 outcomes showed similar directional trends across treatment groups and are presented in Supplementary Figures S1.and Supplementary Figures S2

Discussion

In this retrospective bi-center cohort of patients with CP and CLDP, early clinical response and remission rates were broadly comparable across vedolizumab, ustekinumab, and adalimumab. By week 52, differences in clinical response and treatment persistence were observed. cPDAI response at week 52 was observed in 88.9% of ustekinumab-treated patients, compared with 44.8% and 56.3% in the vedolizumab and adalimumab groups, respectively. Three-year treatment persistence was 81.5, 35, and 52%. These findings suggest potential differences in durability of clinical benefit; however, causal inference is limited by the observational design.

Biologic therapy has been extensively studied in inflammatory bowel diseases [21–29]. However, evidence guiding biologic selection in pouch disorders remains limited and is derived predominantly from observational cohorts with heterogeneous patient populations and outcome definitions [8–11]. The remission rates observed with adalimumab and vedolizumab in our cohort are broadly consistent with prior real-world data. A meta-analysis of anti-TNF therapy reported long-term remission rates of approximately 50% [8], comparable to the 45.5% week 52 mPDAI remission observed with adalimumab in our study. Similarly, multi-center vedolizumab cohorts have reported 12-month response rates in the 40–50% range [9], aligning with our findings.

Comparisons with randomized trials should be interpreted cautiously. In the EARNEST trial, week 34 mPDAI response and remission rates were 51% and 35%, respectively [12]. In our study, week 26 rates were 58.8% and 29.4%. Differences in follow-up timing, inclusion criteria, and prior biologic exposure limit direct comparison. Notably, our cohort consisted largely of biologic-experienced and treatment-refractory patients, which may influence observed response magnitude.

Prospective ustekinumab cohorts have reported 6–12 month response rates of approximately 50–60% [10,13]. The week 52 response observed in our study may reflect differences in treatment sequencing, patient selection, and evolving clinical practice. Ustekinumab was introduced later in the study period and was frequently used after failure of multiple prior biologics. Secular changes in patient management may therefore have influenced both response and persistence estimates.

Endoscopic outcomes did not differ significantly between treatment groups at week 52. Although a statistically significant within-group reduction in ePDAI was observed in the vedolizumab cohort (median decrease from 4 to 1.5), this finding should be interpreted cautiously in the absence of between-group differences. Endoscopic follow-up was non-protocolized and incomplete, with fewer evaluable patients than for clinical endpoints. The greater availability of endoscopic data in certain groups may have affected statistical power and may introduce follow-up bias. Clinical response and endoscopic healing represent related but distinct domains of disease activity; therefore, conclusions regarding comparative mucosal outcomes should be drawn with caution.

Biochemical remission rates were similar across groups and were analyzed using available data without imputation. Sensitivity analyses using non-responder imputation were applied to clinical endpoints but were not feasible for endoscopic or biochemical outcomes due to data structure and sample size.

The comparatively lower response observed with adalimumab may relate to extensive prior anti-TNF exposure in this cohort. A substantial proportion of patients had prior anti-TNF exposure, and reduced effectiveness of subsequent anti-TNF therapy in biologic-experienced populations is well described. [30,31] Mechanistic hypotheses involving IL-23–associated pathways are biologically plausible [32–35] but remain hypothesis-generating within the context of this observational analysis.

This study has several limitations. It was retrospective, conducted at two tertiary centers within a single healthcare system, and treatment allocation was not randomized. Group sizes were unequal, and ustekinumab was introduced later in the study period, raising the possibility that secular changes in management may have influenced response and persistence. Pre-pouch ileitis was recorded qualitatively but not quantitatively scored, limiting assessment of its association with treatment outcomes. CLDP behavior (inflammatory, stricturing, or fistulizing) was not systematically subclassified, thus residual confounding by disease phenotype cannot be excluded. Pre-colectomy biologic exposure data were incomplete, and endoscopic assessments were not standardized. These factors introduce potential confounding and limit generalizability.

Despite these limitations, this study provides comparative real-world data across three commonly used biologic agents in CP and CLDP using predefined outcome measures and correction for multiple comparisons. In a clinical area lacking head-to-head trials, such observational data may help inform therapeutic decision-making while underscoring the need for prospective comparative studies with standardized phenotyping and follow-up.

In conclusion, differences in long-term clinical response and treatment persistence were observed among biologic therapies in this retrospective cohort of patients with CP and CLDP. Given the observational design and potential confounding, these findings should be considered hypothesis-generating and warrant confirmation in prospective comparative studies.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

NM conceived and designed the study and supervised all aspects of the work. YU contributed to data collection, data analysis, and primary manuscript writing. NAC contributed to data collection and manuscript writing and revisions. LW contributed to manuscript writing and revisions. AH, HT, MZ, THS, and BK contributed patient data. All authors reviewed and approved the final manuscript.

Funding

Open access funding provided by Tel Aviv University. The study was supported through educational grants by Jansen Pharmaceuticals and Takeda Pharmaceuticals.

Data Availability

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Conflict of interest

NM has received speaking and/or consulting fees from Pfizer, AbbVie, Eli Lilly, Takeda, Janssen, Ferring, BMS, Nestle, and grant support from Takeda, Janssen, Abbott, AbbVie, Pfizer, BMS, and Nestle. NAC has received speaking and/or consulting fees from Abbvie, Takeda, BMS, Rafa Pharmaceuticals and Iterative Health, a grant support from Takeda and travel support from Pfizer and Takeda.