Biologics and small molecules for perianal Crohn’s disease: a systematic review and meta-analyses of randomized placebo-controlled trials

Abstract

Background

Although several advanced therapies (ATs) for luminal Crohn’s disease (CD) have been approved, their efficacy for perianal fistulizing Crohn’s (pCD) has not been comprehensively reviewed.

Methods

A literature search was conducted in MEDLINE, Embase, and Cochrane CENTRAL up to October 2, 2025. Placebo-controlled randomized controlled trials (RCTs) evaluating treatment efficacy in pCD were included. Outcomes of interest were fistula response and fistula remission during the induction and maintenance trials. Pooled relative risks (RRs) were calculated using a random-effects model.

Results

Fourteen studies were eligible; only 4 RCTs were specifically designed for pCD, while the remainder were post hoc analyses of luminal CD studies with moderate to serious risk of bias. In pooled analyses of induction trials (4-28 weeks), ATs were associated with significantly higher induction of fistula response (RR 1.65, 95% CI 1.22-2.22) and remission (RR 1.91, 95% CI 1.37-2.67) compared to placebo. Infliximab and upadacitinib demonstrated significantly greater fistula response and remission, while ustekinumab significantly increased rates of fistula remission. For the maintenance phase (22-52 weeks), ATs were associated with significantly higher maintenance of fistula response (RR 1.54, 95% CI 1.21-1.97) and remission (RR 2.08, 95% CI 1.49-2.90) compared to placebo. Infliximab showed significantly greater maintenance of fistula response and remission compared with placebo, while adalimumab was superior for fistula remission.

Conclusion

Findings from our meta-analysis indicate that infliximab is effective for pCD, with limited evidence of benefit with adalimumab, ustekinumab, and upadacitinib. Well-designed RCTs investigating newer advanced therapies are needed to better guide targeted treatment strategies for this patient population.

Key Messages.

What is already known? 

Perianal fistulizing Crohn’s disease (pCD) is a debilitating complication associated with substantial morbidity and impaired quality of life.

What is new here? 

This systematic review and meta-analysis summarizes the current evidence on the efficacy of advanced therapies for pCD, with most data derived from post hoc analyses and few dedicated studies. Infliximab appears effective, while evidence for adalimumab, ustekinumab, and upadacitinib remains limited, highlighting the need for well-designed RCTs.

How can this study help patient care? 

The findings of this study can support evidence-based treatment decision-making in the management of pCD.

Introduction

Fistulas are a common manifestation of Crohn’s disease (CD), occurring in approximately 25%-35% of patients, with perianal fistulas being the most frequent type.^1,2 It is estimated that the number of prevalent cases with fistulizing CD in the United States was ∼76 600 in 2017 (∼52 900 anal, ∼7400 rectovaginal, ∼2300 enterocutaneous, and ∼14 100 internal).³ Perianal fistulizing CD (pCD) is associated with impaired quality of life and is a predictor of poor long-term outcomes.^4,5

Over the past decades, substantial progress has been made in surgical approaches for managing complex fistulizing CD, whereas medical therapy has remained primarily centered on antitumor necrosis factor (anti-TNF) agents.⁶ Infliximab was approved by the U.S. Food and Drug Administration in 1998 for treating active and fistulizing CD. Patients who achieved fistula healing were found to have higher serum levels of infliximab and adalimumab compared with those who continued to have active drainage.⁷ However, the efficacy of other medical therapies remains insufficiently studied.

The introduction of biologic therapies has transformed the management of CD. In recent years, numerous RCTs have evaluated the efficacy of advanced therapies for CD, expanding the therapeutic landscape for this condition. The recent American Gastroenterological Association (AGA) evidence synthesis supports the use of advanced therapies—such as TNF antagonists, anti-integrins, interleukin (IL)-12/23 or IL-23p19 antagonists, and Janus kinase (JAK) inhibitors—in both biologic-naive and biologic-exposed patients with moderate-to-severe CD for the induction and maintenance of clinical remission.⁸ Nevertheless, the effects of these newer advanced therapies on pCD have not been comprehensively studied.

The efficacy of nonsurgical therapies has been examined in 4 systematic reviews (SRs). The earliest SR evaluated ustekinumab and included only one RCT and 24 observational studies.⁹ Two were network meta-analyses (NMAs): one assessed biologic therapies alone,¹⁰ while the other included all nonsurgical treatments such as stem cells and fibrin glue.¹¹ The most recent SR included all pharmacological therapies, with literature searched up to May 2022.¹² Collectively, these SRs concluded that anti-TNFs and small molecules are effective for pCD, albeit with low- to moderate-certainty evidence.

It is noteworthy that these SRs combined various types of fistulizing CD (eg, perianal, enterocutaneous, enteroenteric, enterovesical, enterovaginal, and enterouterine fistulas) and that data from induction and maintenance trials were sometimes pooled in the same analysis. Heterogeneous designs of maintenance trials—for example, inclusion of all fistulizing patients regardless of drainage status, treat-through design of draining fistulas, or re-randomization of clinical (not fistula) responders from induction anti-TNF trials—were not fully examined in the published SRs.

In this updated SR and meta-analysis, we summarize the latest available evidence on the efficacy of biologic therapies and small molecules, focusing on placebo-controlled RCTs, with the aim of identifying optimal treatments for pCD. We also review the study designs and methodologies of included trials.

Materials and methods

This systematic review and meta-analysis were conducted in accordance with the Cochrane Handbook for Systematic Reviews of Interventions¹³ and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement¹⁴ (Supplementary Appendix 1).

Search strategy and selection criteria

We searched MEDLINE (Ovid), EMBASE (Ovid), and the Cochrane Central Register of Controlled Trials (CENTRAL, Ovid) from 2022 (the latest search dates of previously published SRs) to October 2, 2025. The search strategy was initially developed in MEDLINE and subsequently adapted for EMBASE and CENTRAL. Both controlled vocabulary (eg, MeSH and Emtree terms) and free-text keywords related to CD and fistulas were used. Eligible conference proceedings were included. Non-English language records were excluded (Supplementary Appendix 2). Studies published before 2022 were identified through a review of reference lists from prior systematic reviews, followed by recursive searches of the bibliographies of all included studies.

Inclusion criteria: Adult patients with active pCD who received biologic or small-molecule therapy in placebo-controlled induction or maintenance trials, with efficacy outcomes assessed according to the definitions in the primary studies. Eligible interventions included any advanced therapy (ie, TNF antagonists, anti-integrins, IL (IL-12/23, IL-23p19) antagonists, JAK inhibitors, and S1P receptor modulators) compared with placebo. We included studies that enrolled patients with both perianal and nonperianal fistulas CD, provided that the majority of participants had perianal fistulas.

Studies were excluded if they (1) only studied nonbiologic and non-small-molecule therapies (eg, antibiotics, immunosuppressants, mesenchymal stem cell transplantation, and spherical carbon adsorbent) or combination therapies, (2) were not placebo-controlled trials, (3) did not provide separate data for patients with fistulizing disease to calculate the effect estimate, (4) included only nonperianal fistulas, (5) included only pediatric populations, (6) included fistulas involving the pouch, (7) non-English publications, or (8) duplicates or preliminary publications.

References identified by the search strategy were imported into the screening tool Covidence (https://www.covidence.org/). Two reviewers (YY, SV) independently reviewed the search results based on titles and abstracts, followed by a thorough examination of full-text articles. Disagreements were resolved through discussion. If consensus was not achieved, a third reviewer acted as the arbiter.

Outcome assessment

Our efficacy outcomes of interest included the perianal fistula response rate and perianal fistula remission rate in both induction and maintenance trials. The primary outcome was fistula response rate in induction trials. The other 3 were secondary outcomes. Outcome definitions were extracted as reported by the original study authors and summarized in Table 1. The measurement time points were defined according to the primary studies up to 28 weeks for induction trials and up to 56 weeks for maintenance trials.

Table 1.

Study characteristics of induction trials.

References	Analysis type	Fistula type (%)	Previous medications in fistulizing CD (%)	Prevalence of fistulizing CD in primary RCT	Intervention (n)	Placebo (n)	Age—mean	Male (%)	Intervention	Definition of fistula response	Definition of fistula remission	Measurement time point at week
Present et al.21	Primary	Perianal (90) Abdominal (10)	CS (35) 5-ASAs (55) Antibiotics (30) IM (40)	–	63	31	37	47	Infliximab 5 mg or 10 mg/kg iv, at week 0, 2, 6.	Reduction of ≥50% from baseline in the number of draining fistulas, at ≥2 consecutive visits	Complete response (absence of any draining fistulas at two consecutive visits)	18
Sandborn et al.56,57 (GAIN)	Post hoc	Abdominal (enterocutaneous; NS) or perianal (NS)	NA	14% (45/325)	20	25	NA	NA	Adalimumab 160 mg sc at week 0, 80 mg at week 2	Decrease ≥50% in the number of draining fistulas at weeks 2 and 4 vs baseline	Closure of all fistulas at weeks 2 and 4 that were draining at screening and at baseline	4
Hanauer et al.55 (CLASSIC-1)	Post hoc	Enterocutaneous or perianal fistula (NS)	NA	11% (32/299)	26	6	NA	NA	Adalimumab 40/20 mg, 80/40 mg, 160/80 mg sc at week 0 and 2	A decrease in the number of draining fistulas ≥50% for at least 2 consecutive visits, at 4 weeks	100% closure of all draining fistulas for ≥2 consecutive visits, at 4 weeks	4
Sandborn et al.16	Post hoc	Perianal (NS) or abdominal enterocutaneous (NS)	NA	22% (86/395)	60	26	NA	NA	CDP571 10 mg/kg IV q8ws to week 24	50% closure of fistulas where no drainage on gentle compression, (also reported data for “at any visit and on two consecutive visits over six weeks”)	Closure of 100% of fistulae where no drainage on gentle compression (also reported data for “at any visit and on two consecutive visits over six weeks”)	28
Sands et al.29 (UNITI-1, UNITI-2, CERTIFI)	Post hoc	Perianal (100)	NA	13% (238/1894)	161	77	NA	NA	Ustekinumab 1 mg, 3 mg, 6 mg/kg or 130 mg iv, week 0	≥50% reduction in draining fistulas	Complete fistula resolution: 100% reduction	8
Wils et al.22 (USTAP)	Primary	Perianal (100)	NA	–	16	16	NA	NA	Ustekinumab 6 mg/kg iv followed by 90 mg sc q8w	Closure of at least 50% of all treated external openings that were draining at baseline (no data reported)	100% of the fistula tracts without any drainage by the external openings (ie, absence of any drainage by all fistula openings occurring spontaneously or after gentle finger compression)	12
Gao et al.27 (ADVANCE, MOTIVATE)	Post hoc	Perianal (NA, most)	NA	7% (96/1419)	72	24	NA	NA	Risankizumab 600 or 1200 mg sc at week 0, 4, 8	Absence of draining fistulas among patients with draining fistulas at baseline (number of fistulas draining noted on gentle compression)	NA	12
Reinisch et al.23 (DIVERGENCE 2)	Primary	Perianal (100; 20% of placebo users vs 2% of filgotinib users had zero number of draining perianal fistula at baseline)	IMM (65); anti-TNFs (65), VED (19), ustekinumab (28); vedolizumab concomitant (7); CS -IMM concomitant (33)	–	42	15	40	60	Filgotinib 100 mg or 200 mg od × 24 weeks	Combined fistula response: a reduction of at least one from baseline in the number of draining external openings, as determined by physical assessment, and no fluid collections of >1 cm in at least two dimensions on pelvic MRI at week 24	Closure of all draining EOs present at baseline, as determined by physical assessment, and no fluid collections of >1 cm in at least two dimensions on pelvic MRI, at Week 24 [among participants who had at least one draining EO at baseline]	24
Colombel et al.15 (U-EXCEL U-EXCEED)	Post hoc	Perianal (90), others (10—also reported outcome, not included in this SR)	Anti-TNF (83), CS (31).	14% (143/1201) fistulas; 13% perianal (128/1021)	86 (38 draining)	42 (18 draining)	34	56	Upadacitinib 45 mg od × 12 weeks	Resolution of drainage in perianal draining fistulas (a lack of drainage)	Closure of external fistula openings	12

Abbreviations: CS, corticosteroids; EO, external opening; IMM, immunomodulators; iv, intravenous; NA, nonavailable; od, once daily; po, per os-orally; q8w, every 8 weeks; sc, subcutanous; SR, systematic review.

Data extraction

The same two reviewers independently extracted data from the included studies using a standardized data extraction form. Extracted information included the author, publication year, country, analysis type (primary or post hoc), trial phase (induction or maintenance), fistula type, patient characteristics, interventions, outcome definitions, measurement time points, and outcome data (number of patients treated and those achieving the specified outcome). When a study reported results for both perianal fistulas and any-type fistulas, data for perianal fistulas were extracted.¹⁵

We used the sample size for patients with fistulas as reported in the primary studies, rather than adopting an intention-to-treat approach. In several previous meta-analyses, the number of patients with fistulas at baseline was used as the denominator, and those without outcome data were considered treatment failures. Since most included studies were post hoc analyses of primary RCTs and did not clearly report the number of patients who were not assessed after treatment, we included only those who received treatment and had measurable outcomes, avoiding assumptions about missing data.

A study reported outcome data at any visit and at any two consecutive visits over 6 weeks; we extracted data from any visit.¹⁶ Several trials randomized patients immediately after a 4- to 14-week open-label anti-TNF induction phase^17,18 regardless of fistula status, while some studies randomized only clinical responders [defined as those with a reduction in Crohn’s Disease Activity Index (CDAI) of at least 100 points from baseline], which included some patients who still had draining fistulas after induction therapy.^19,20 These studies, which typically reported outcomes at weeks 22-26, were inconsistently categorized as either induction trials¹⁰ or maintenance trials¹² in previous meta-analyses. In the current meta-analysis, we classified these studies as maintenance trials, as randomization occurred immediately following the induction phase as defined by the primary investigators. When multiple time points were reported, data from the latest time point were extracted, taking into account fistula recurrence, remission, and the development of new fistulas.

Data synthesis and statistical analysis

Relative risks (RRs) with 95% confidence intervals (CIs) were pooled using the DerSimonian–Laird random-effects model. Heterogeneity was assessed using both the Chi-squared test with a P value < .10 used to define a significant degree of heterogeneity and the I² statistic, ranges between 0% and 100% with interpretation as: 0%-40% (may not be important), 30%-60% (moderate heterogeneity), 50%-90% (substantial heterogeneity), and 75%-100% (considerable heterogeneity).¹³ Analyses were conducted separately for induction and maintenance trials. Subgroup analyses were performed by drug and by drug class (TNF antagonists, anti-integrins, IL-12/23 antagonists, IL-23p19 antagonists, and JAK inhibitors). When more than two studies reported data for the same drug class, additional subgroup analyses were planned by analysis type (primary vs post hoc) for induction trials and by trial design (randomization after open-label induction, re-randomization after induction RCT, or treat-through design) for maintenance trials.

We conducted a complete-case analysis as the primary analysis, restricted to participants with available data for both the intervention and outcome. This approach was chosen because most included studies (10 of 14) were post hoc analyses of RCTs. These analyses only reported data for patients with pCD who both received the intervention and had outcome data available. By design, these post hoc analyses function as observational studies. It is not possible to perform intention-to-treat (ITT) analyses for these studies. Our review included only 4 RCTs that prospectively randomized patients with pCD. Two of these trials had no missing outcome data, although their sample sizes were small.^21,22 The remaining two RCTs had minimal missing data: the induction trial by Reinisch et al.²³ (41/42 vs 12/15 with outcome data) and the maintenance trial by Sands et al.¹⁸ (134/139 vs 142/143 with outcome data). To maintain analytic consistency across all studies, we used only participants with available outcome data in the 4 RCTs, aligning them with the 10 post hoc analyses that also reported outcomes in available-case populations. We also performed a sensitivity analysis using the ITT population for the 2 RCTs with missing outcome data, assuming that all patients without outcome data experienced treatment failure.

Review Manager version 5.4.1 (RevMan for Windows 2014, the Nordic Cochrane Centre, Copenhagen, Denmark) was used to generate forest plots of pooled RRs with 95% CIs for each outcome. Potential publication bias and small-study effects were planned to be assessed using funnel plots and Egger’s test if there were ≥10 eligible studies included in the meta-analysis.²⁴

Risk of bias assessment

For randomized controlled trials (RCTs), risk of bias was assessed independently by two reviewers using the Cochrane Risk of Bias 2 (RoB 2) tool.²⁵ For post hoc analysis of RCTs, there is no validated risk of bias tool, risk of bias tools designed for nonrandomized studies have been used for subgroups of RCTs. Since randomization was not stratified for the subgroup of patients with fistulas, we used the ROBINS-I (Risk of Bias in Nonrandomised Studies of Interventions) tool for observational studies,²⁶ version 2 (https://www.riskofbias.info/welcome/robins-i-v2). We resolved disagreements through arbitration by a third investigator.

Results

The updated search identified 149 potential records; 109 references were screened. After screening titles and abstracts, 86 studies were excluded. 21 full-text articles were reviewed, and 17 were excluded with documented reasons. Four new studies met the inclusion criteria: 1 RCT (USTAP)²²; 1 post hoc analysis of data from the global phase 3 ADVANCE, MOTIVATE, and FORTIFY trials of risankizumab,²⁷ and 2 were full publications of previously included conference abstracts^15,23 One conference abstract reporting additional maintenance data for a previously included study (which was also presented as a conference abstract) was identified^28,29 (Figure S1).

Thirty-five studies were included in previous meta-analyses; 25 were excluded from the current meta-analysis. These included trials that compared antibiotic therapies^30,31; immunosuppressants^32–38; mesenchymal stem cell transplantation^39–42 or AST-120 (spherical carbon adsorbent)^43,44 vs placebo; studies that compared two immunosuppressants⁴⁵; immunosuppressants vs steroids⁴⁶; anti-TNFs plus antibiotics vs anti-TNF monotherapy^47,48; combined two biologics^49,50 or two biologic regimens^51,52; and conference abstracts that have since been published in full.^53,54

Finally, 10 studies included in a previous meta-analysis were retained after assessment (Figure S1). In total, 14 studies (15 reports) met our inclusion criteria, including 3 conference abstracts.^22,28,29 Studied medications comprised anti-TNFs infliximab,^18,21 adalimumab,^17,55,56 certolizumab Pegol,^19,57 and CDP571,¹⁶ integrin antagonist vedolizumab,²⁰ IL-12/23 inhibitor ustekinumab,^22,29 IL-23 inhibitor risankizumab,²⁷ and JAK inhibitors filgotinib²³ and upadacitinib¹⁵ Detailed definitions of outcomes can be found in Tables 1 and 2.

Induction trials

Eight induction trials met our inclusion criteria. All were multinational studies. Only two studies randomized patients with active fistulas to receive active treatment or placebo^21,23; the remaining 6 studies reported data for patients with fistulas as a post hoc subgroup analysis of pivotal RCTs, including GAIN⁵⁶; CLASSIC-1⁵⁵; UNITI-1, UNITI-2, and CERTIFI²⁹; U-EXCEL and U-EXCEED¹⁵ and ADVANCE, MOTIVATE, and FORTIFY,²⁷ and CDP571.¹⁶

The prevalence of fistulizing CD among participants in the primary RCTs ranged from 7% to 22% (Table 1). Three studies included only patients with perianal fistulas,^22,23,29 one reported separate data for perianal fistula,¹⁵ while the remaining studies included both perianal and a small proportion of nonperianal fistulas. Most studies included only draining fistulas. In the DIVERGENCE 2 trial, 20% (3/15) of placebo users vs 2% (1/42) of filgotinib users had nondraining perianal fistula at baseline,²³ although the difference was not significant. These patients were not included in the fistula response analysis. One study enrolled both nondraining and draining fistulas (44%), and reported outcomes of “resolution of drainage” and “closure of external opening.”¹⁵

Except for two published RCTs that specifically randomized fistulizing patients,^21,23 only one post hoc analysis reported baseline characteristics for the fistulizing CD subgroup.¹⁵ Most studies presented baseline data for the overall trial population, rather than for the subset with fistulizing disease. Among the 3 studies that reported baseline characteristics for patients with fistulas, the mean age ranged from 34 to 40 years, and 47% to 60% were male. Apart from the 1999 study that studied infliximab,²¹ two studies reported that 65% to 83% of patients had prior anti-TNF exposure, while others had received corticosteroids and/or immunomodulators.

All studies reported outcome data for fistula response, most commonly defined as a ≥50% reduction in draining fistulas or ≥50% closure of fistulas. Seven studies also reported fistula remission, defined as complete (100%) closure of all draining fistulas. Three post hoc analyses pooled data from multiple phase 3 trials.^15,27,29 The measurement time points varied across studies, ranging from 4 to 28 weeks (Table 1).

Induction of fistula response

Eight studies reported data on the induction of fistula response across 7 interventions. The studied medications included anti-TNFs (infliximab, adalimumab, CDP571),^16,21,55,56 IL-12/23 antagonist (ustekinumab),²⁹ IL-23 antagonist (risankizumab),²⁷ and JAK inhibitors (filgotinib and upadacitinib).^15,23 Three studies were RCTs^21–23; others were post hoc analyses. Most studies reported the outcome of induction within 12 weeks, except one reported at 18 weeks, one at 24 weeks, and one at 28 weeks.

Except for two studies that reported data on adalimumab, only one study each provided fistula response data for other drugs. A significantly higher fistula response rate compared with placebo was observed only for infliximab (RR 2.40, 95% CI 1.28-4.49)²¹ and upadacitinib (RR 8.05, 95% CI 1.16-55.88).¹⁵ Overall, the pooled fistula response rate across all studies was significantly higher for interventions compared with placebo (36.6% [172/470] vs 20.7% [44/213]; RR 1.65, 95% CI 1.22-2.22, P = .001; test for heterogeneity I² = 4%, P = .40; Figure 1). However, no significant class effect was observed for anti-TNFs (infliximab, adalimumab, CDP571; n = 4; RR 1.51, 95% CI 0.91-2.51; I² = 32%, P = .22) or oral JAK inhibitors (filgotinib, upadacitinib; n = 2; RR 2.90, 95% CI 0.49-17.25; I² = 63%, P = .10) in subgroup analyses (Figure S2). No significant difference was observed between ustekinumab or risankizumab and placebo. Test for subgroup differences indicated no significant heterogeneity across the 7 drugs (I² = 15.4%, P = .31). Owing to the limited number of studies for each drug, additional subgroup analyses by analysis type (primary vs post hoc) were not conducted.

Figure 1.

Advanced therapy vs placebo for perianal fistula CD response in induction trials.

Induction of fistula remission

Eight studies reported data for the induction of fistula remission across 6 interventions. Two studies each reported data for adalimumab and ustekinumab; only one study provided outcome data for each of the following: infliximab, CDP571, ustekinumab, filgotinib, and upadacitinib. A significantly higher fistula remission rate compared with placebo was observed only for infliximab (RR 3.57, 95% CI 1.38 to 9.25; one RCT),²¹ ustekinumab (RR 1.91, 95% CI 1.16, 3.15; one RCT, one post hoc)^22,29 and upadacitinib (RR 4.64, 95% CI 1.13-18.99; one post hoc)¹⁵

Overall, the pooled fistula remission rate was significantly higher for all interventions compared with placebo (30.3% [140/462] vs 14.8% [34/229]; RR 1.91, 95% CI 1.37-2.67, P = .0001; test for heterogeneity, I² = 0%, P = .52; Figure 2). No class effect was observed for anti-TNFs (infliximab, adalimumab, CDP571; n = 4; RR 1.67, 95% CI 0.83-3.35; I² = 29%, P = .24), whereas a significant difference was observed with oral JAK inhibitors (filgotinib and upadacitinib combined; n = 2; RR 3.02, 95% CI 1.14-7.95, P = .03; test for heterogeniety, I² = 0%, P = .40; Figure S3). Test for subgroup differences indicated no significant heterogeneity among the 6 drugs (I² = 14.3%, P = .32). Owing to the limited number of studies for each drug, additional subgroup analyses by analysis type (primary vs post hoc) were not conducted.

Figure 2.

Advanced therapy vs placebo for perianal fistula CD remission in induction trials.

Maintenance trials

Study design

Nine studies contributed data to the maintenance trials. The studied medications included anti-TNFs (infliximab, adalimumab, and certolizumab pegol),^17–19,57 vedolizumab,²⁰ ustekinumab,^28,29 risankizumab,²⁷ and the JAK inhibitor upadacitinib.¹⁵

The study designs of the maintenance trials and their measurement time points (22-56 weeks) varied considerably. Only one study specifically randomized patients with pCD after induction therapy,¹⁸ while the others reported pCD outcomes as post hoc subgroups from RCTs. Five trials randomized patients following an open-label induction phase, 2 trials re-randomized patients after an induction RCT, and 1 followed a treat-through design that reported both induction and maintenance data. However, the induction regimens, treatment durations, fistula status, and clinical response status of the randomized patients differed across studies (Table 2). It should be noted that patients with pCD enrolled in maintenance trials do not necessarily have achieved fistula response or remission in the preceding induction phase.

Table 2.

Study characteristics of maintenance trials.

References	Study design	Analysis type	Fistula type (%)	Eligibility of fistulas, at baseline and at randomization	Induction therapy and duration	Maintenance therapy and duration	Intervention arm (n)	Placebo (n)	Measurement time point at week	Definition of fistula response	Definition of fistula remission/complete response
Sands et al.18 (ACCENT-II)	Randomization after 14-week infliximab induction, combined responders + nonresponders	Primary	Perianal fistulas (87); abdominal (14); rectovaginal (9; could be multiple locations)	One or more draining abdominal or perianal fistulas at baseline, then included nonresponders + responders after induction	Infliximab 5 mg/kg IV, weeks 0, 2, and 6.	Infliximab 5/mg q8w, week 14 to week 54	96 induction responders + 43 induction nonresponders	99 induction responders + 44 nonresponders	54	Reduction of ≥50% from baseline in the number of draining fistulas at consecutive visits ≥4 weeks apart. Responders: if a response was observed at both weeks 10 and 14	Complete response: the absence of draining fistulas
Colombel et al.17 (CHARM) and Colombel et al.59	Randomization after 4-week adalimumab induction, all patients, irrespective of fistula status	Post hoc	Perianal (97), abdominal (3)	Draining fistulas at baseline, about 18% had healed fistulas after 4 weeks of induction (Figure 3)	Adalimumab 80 mg sc at baseline, followed by 40 mg at week 2 (open label)	Adalimumab 40 mg eow or weekly from week 4 to week 56	70	47	56	NA	Complete fistula healing/closure (assessed at every visit) at 26 weeks, who continued to have fistula closure at 56 weeks
Sandborn et al.56,57 (PRECISE-1)	Treat-through design	Post hoc	NA	NA	Certolizumab pegol 400 mg or placebo at weeks 0, 2, and 4 and then every 4 weeks, through week 26.	Certolizumab pegol 400 mg sc at weeks through week 26.	46	61	26	Same as Present 1999-reduction of ≥50% from baseline in the number of draining fistulas, at two or more consecutive study visits. Only reported (fistula closure, not sure if complete/remission)	Same as Present 1999-complete response (absence of any draining fistulas at two consecutive visits), but did not report outcome data.
Schreiber et al.19 (PRECise 2)	Randomization after 6-week certolizumab pegol, only clinical (CDAI) responders with draining fistulas were included	Post hoc	Perianal (95); NS (5)	Draining fistulas at baseline and at randomization	Certolizumab pegol 400 mg sc at weeks 0, 2 and 4	Certolizumab pegol 400 mg q4w sc across weeks 8-24.	28	30	26	Fistula closure: closure of ≥50% of fistulas at any two consecutive post-baseline visits ≥3 weeks apart	100% closure of fistulas open at baseline
Feagan et al.20 (GEMINI 2)	Randomization after 6-week vedolizumab induction, only clinical responders and with draining fistula were included	Post hoc	perianal (79), others (21)	Draining fistulas at baseline and at randomization	Vedolizumab 300 mg iv at weeks 0 and 2 (6 weeks)	Vedolizumab 300 mg iv q8w or q4w × 46 weeks	39	18	52	NA	Fistula closure, clinically no draining fistulae at the given visit
Sands et al.29 (IM-UNITI, CERTIFI);  Sands et al.28 (IM-UNITI)	CERTIFI: responders and nonresponders after 8 weeks ustekinumab induction were randomized IM-UNITI: response to UST induction in UNITI-1 and UNITI-2 were randomized	Post hoc	Perianal (100)	Draining fistula at baseline. NA at randomization	CERTIFI: ustekinumab 1, 3, or 6 mg/kg iv IM-UNITI: ustekinumab 130 mg or 6 mg/kg iv	CERTIFI: ustekinumab 90 mg sc at weeks 8 and 16. IM-UNITI: ustekinumab 90 mg q8w or q12w sc, weeks 8 to 40	CERTIFI 24; IM-UNIT: 15	CERTIFI 30; IM-UNIT: 11	CERTIFI 22; IM-UNIT 44	≥50% reduction in draining fistulas	Complete fistula resolution: 100% reduction
Gao et al.27 (FORTIFY)	Rerandomized clinical responders after a 12 weeks induction (risankizuma vs placebo) RCT	Post hoc	Perianal (NA, but most)	Draining fistula	Risankizumab 600-1200 mg sc vs placebo × 12 weeks	Risankizumab 180 mg vs 360 mg sc vs placebo, sc q8w for 52 week	46	21	52	Absence of draining fistulas among patients with draining fistulas at baseline (number of fistulas draining noted on gentle compression)	“Fistula closure” was only reported for risankizumab
Colombel et al.15 (U-ENDURE)	Rerandomized clinical responders after a 12-week induction (upadacitinib vs placebo) RCT	Post hoc	Perianal (83 for outcome “resolution of drainage,” 89 for outcome “closure of external openings”), we only included perianal in this SR	Fistulas at baseline, n = 38 for “resolution of drainage” n = 87 for “closure of external openings.” Upon randomization to maintenance, 24.5% had achieved external closure of fistulas and 50% achieved resolution of draining fistulas	Upadacitinib 45 mg od vs placebo × 12 weeks	Upadacitinib 15 mg vs UPA 30 mg vs placebo od for 52 weeks	66 randomized (57 for “closure of external opening,” 27 for “resolution of drainage”)	32 randomized (30 for “closure of external opening,” 11 for “resolution of drainage”)	52	Resolution of drainage in perianal draining fistulas (a lack of drainage) in patients with draining fistulas	Closure of external fistula openings

Abbreviations: CS, corticosteroids; EO, external opening; IMM, immunomodulators; iv, intravenous; NA, nonavailable; od, once daily; po, per os-orally; q8w, every 8 weeks; sc, subcutanous; SR, systematic review.

The ACCENT II study included patients with draining pCD at baseline. Over 55% of patients presented with 2 or more draining fistulas. After 14 weeks of infliximab induction therapy, fistulizing patients either responders (≥50% reduction from baseline in the number of draining fistulas) or nonresponders were randomized to infliximab or placebo for maintenance therapy. Fistula responses were reported separately for responders and nonresponders, while complete response was reported only for patients who were responders at randomization.¹⁸ Previous meta-analyses treated data from induction nonresponders as induction outcomes, while induction responders were included in maintenance outcomes. Considering that all patients had received active treatment during the 14-week induction period, that 8% of patients were not randomized at the maintenance phase, and that the measurement time point was 54 weeks after randomization, we considered data from both groups under maintenance therapy, consistent with the approach of Sharma et al. 2023.⁵⁸ Furthermore, this study was the only trial that permitted crossover from week 22 for patients who had loss of response either from placebo to infliximab 5 mg (50%), or from infliximab 5 mg to 10 mg (29%); data for patients who crossed over from placebo to infliximab were censored before crossover occurred.¹⁸ Instead of using an intention-to-treat analysis as in previous meta-analyses, we used the sample size reported by the primary authors.

The CHARM trial randomized patients after 4 weeks of adalimumab induction therapy, irrespective of fistula status.¹⁷ The study reported the percentage of patients achieving fistula closure at multiple time points from week 4 to week 56. The remission rate peaked at week 16 (36% vs 15%) and decreased to its lowest at week 26 (30% vs 13%) between weeks 16 and 36. Among patients with complete fistula closure at week 26, 100% maintained closure at week.^56,59 However, it should be noted that only 18% of patients had healed fistulas after the 4-week induction period when entering the maintenance phase.¹⁷

Conversely, two studies included only clinical responders after induction therapy who still had draining fistulas. In PRECISE-2, clinical responders with draining fistulas after 6 weeks of certolizumab pegol induction therapy were studied,¹⁹ and in GEMINI2, clinical responders with draining fistulas after 6 weeks of vedolizumab induction therapy were assessed.²⁰ GEMINI2 reported data at week 14 and week 52: the fistula closure rate for placebo remained the same (11%), while closure rates for vedolizumab increased slightly from 28.2% to 30.8%. The authors also reported the development of new draining fistulas (3%-4%; 63%-80% had a history of fistulizing disease) and closure rates in nonresponders after induction who received vedolizumab. Closure rates observed in week-6 responders are not generalizable to vedolizumab nonresponders, for whom the rates were 12% at week 6 and 18% at week 52.

Two post hoc analyses reported both induction and maintenance data for ustekinumab from the UNITI-1, UNITI-2, IM-UNITI, and CERTIFI trials, with data specifically from IM-UNITI and CERTIFI included.^28,29 The two original maintenance trials had different study designs and measurement time points. Both trials included anti-TNF–refractory patients. In the CERTIFI trial, IV ustekinumab responders and nonresponders with open fistulas at Week 0 were re-randomized to receive SC ustekinumab or placebo (n = 30) at Week 8⁶⁰; only 72% (39/54) of these fistulizing patients had fistula outcome data available at Week 22. In contrast, in the IM-UNITI trial, patients who responded to ustekinumab at Week 8 were randomly assigned to receive subcutaneous maintenance injections of ustekinumab or placebo, with fistula outcomes assessed at Week 44.⁶¹ Fistula outcome was assessed on week 44.²⁹ Fistula resolution after 44 weeks in the UNITI program was reported separately.²⁸

Two trials rerandomized patients after an induction therapy RCT. In the U-ENDURE trial, clinical responders after a 12-week induction RCT (upadacitinib vs placebo) were re-randomized. The post hoc analysis included all re-randomized patients with fistulas (draining or nondraining, perianal, or any fistulas) at induction baseline who received ≥1 dose of upadacitinib during the maintenance period, with outcomes reported at 52 weeks.¹⁵ Outcomes “resolution of drainage in perianal draining fistulas” and “closure of external perianal fistula openings” were reported for patients with draining fistula or with any fistulas at rerandomization. At the start of maintenance therapy, 25% (24/98) of patients had achieved external fistula closure, and 50% (23/46) had resolution of draining fistulas. This study also reported data on fistula recurrence and new-onset fistulas. In the FORTIFY trial, patients achieving clinical response to 12 weeks of risankizumab in the ADVANCE and MOTIVATE placebo-controlled trials were re-randomized to risankizumab (180 or 360 mg) or placebo (withdrawal) every 8 weeks for up to 52 weeks.²⁷

One trial, PRECISE-1,⁵⁷ used a treat-through design, randomizing patients with CD to certolizumab pegol or placebo for 26 weeks. Patients with fistulas were reported as a subgroup. Although the authors referenced fistula evaluation methods from previous studies,^18,21 they reported fistula outcomes as “fistula closure” without reporting the percentage of fistula closure, which we considered equivalent to fistula response. The primary study defined week 6 as the induction period and week 26 as the maintenance period; only the week-26 data were reported. This study was classified as an induction trial in two previous meta-analyses^10,58 and as a maintenance trial in another.¹² Considering the treat-through design and the measurement time point, which is consistent with PRECISE-2,¹⁹ we included the week-26 data in the maintenance meta-analysis.

Maintenance of fistula response

Six studies reported fistula response data in maintenance trials for 5 interventions. Studied medications included anti-TNFs (infliximab, certolizumab),^18,19,57 ustekinumab,²⁹ risankizumab,²⁷ and upadacitinib.¹⁵ Data from the IM-UNITI and CERTIFI trials were included as two separate studies in the meta-analysis due to different measurement time points.²⁹

Biologics and small molecules were associated with a significantly increased fistula response rate in maintenance trials: 40.0% (114/286) vs 25.2% (72/286) for placebo, with RR 1.54 (95% CI 1.21-1.97; test for heterogeneity, I² = 0%, P = .61; Figure 3). Only infliximab, which was evaluated in a single, older trial, showed a significant increase in response rate (RR 1.80, 95% CI 1.3-2.65). No significant differences were observed for certolizumab pegol (RR 1.19, 95% CI 0.71-2.00) or ustekinumab (RR 1.68, 95% CI 0.99-2.86), each of which included 2 trials. One study each reported data for risankizumab (RR 1.29, 95% CI 0.61-2.77) and upadacitinib (RR 6.43, 95% CI 0.40-103.80), with no significant difference vs placebo. When the two anti-TNFs (3 studies) were pooled, there was no significant difference vs placebo (RR 1.47, 95% CI 0.99-2.18); no significant differences between drugs were detected (I² = 0%, P = .58; Figure S4). Excluding the only study with a treat-through design, the overall RR was not substantially changed (RR 1.70, 95% CI 1.30, 2.23). Due to clinical heterogeneity across studies and the limited number of trials, no further subgroup analyses were performed for study design or patients’ baseline characteristics.

Figure 3.

Advanced therapy vs placebo for perianal fistula CD response in maintenance trials.

Maintenance of fistula remission

In the maintenance trials, 6 studies reported fistula remission data for 6 medications, including anti-TNFs (infliximab, adalimumab, certolizumab pegol),^17–19 vedolizumab,²⁰ ustekinumab,²⁸ and upadacitinib.¹⁵

In the maintenance trials, biologics and small molecules were associated with a significantly increased fistula remission rate: 33.1% (99/299) vs 15.5% (36/232) for placebo, with RR 2.08 (95% CI 1.49-2.90; test for heterogeneity, I² = 0%, P = .80; Figure 4). Both infliximab (RR 1.87, 95% CI 1.15-3.04) and adalimumab (RR 2.57, 95% CI 1.13-5.84) showed significant increases in remission vs placebo. No significant differences were observed for certolizumab pegol (RR 2.14, 95% CI 0.84-5.50), ustekinumab (RR 1.77, 95% CI 0.81-3.86), or upadacitinib (RR 11.22, 95% CI 0.68-185.19)or vedolizumab (RR 2.77, 95% CI 0.69-11.11). When the 3 anti-TNFs were pooled, a significant difference vs placebo was observed (RR 2.05, 95% CI 1.40-3.01, P = .0002); no significant differences between individual drugs were detected (I² = 0%, P = .80; Figure S5). Due to heterogeneity in study designs and the limited number of studies, no further subgroup analyses were performed for study design or patients’ baseline characteristics.

Figure 4.

Advanced therapy vs placebo for perianal fistula CD remission in maintenance trials.

Sensitivity analysis

In the sensitivity analysis using the ITT population for the two RCTs^18,23 with missing outcome data, and assuming that all patients without outcome data experienced treatment failure, the results were nearly identical to the primary analyses across all outcomes (Table S1).

Risk of bias assessment

Three induction trials^21–23 and one maintenance trial¹⁸ randomized fistula patients to interventions and placebo. Using the Cochrane ROB2.0 tool, two RCTs were low risk of bias. One was rated as high risk of bias given the nonresponder group was randomized and followed, but complete response, which was a key clinical outcome, was not reported.¹⁸ One has some concerns, as fistula response was an outcome in the protocol; no data was reported in the conference abstract²² (Figure S6).

For the 10 studies (11 reports) that reported post hoc analyses of one or multiple RCTs, since patients with fistulas were not randomized to receive advanced therapies vs placebo, we assessed the risk of bias using ROBINS-I v2 for nonrandomized studies due to the observational nature of these studies. The primary RCTs were considered at low risk of bias for the randomization process, baseline confounding present before the start of the intervention and predictive of the intervention received is unlikely to be critical. However, because randomization was not stratified for the subgroup, and most studies did not report baseline characteristics of fistula patients, there were concerns regarding bias. Therefore, all studies were judged to be at moderate risk of bias for the domain bias due to confounding.

As all studies were double-blinded, with both patients and physicians blinded to treatment allocation, we assumed that fistula outcome assessors were also unaware of treatment received. For studies with missing outcome data, data were only reported for patients with outcome data, while information on how missing fistula outcomes were handled was generally not reported; therefore, these studies were considered at moderate risk of bias for missing data after taking into account the proportion of missing outcomes. Two studies were judged to be at serious risk of bias for selective reporting of results, as fistula closure was only reported for the active treatment group,²⁷ or fistula remission was measured but not reported.⁵⁷ For post hoc analyses, all studies were considered to have a moderate risk of bias for outcome assessment, as fistula outcomes were assessed only indirectly through the CDAI, which includes a single fistula-related component (draining vs nondraining). Because the CDAI provides a very limited reflection of fistula severity, any post hoc analysis using CDAI as a surrogate for fistula outcomes is inherently limited. This is consistent with systematic reviews highlighting the need for dedicated fistula-specific scoring indices.⁶² For studies that have 3 moderate risk of bias domains, the study was considered to have a serious risk of bias. In summary, 5 studies were considered at serious risk of bias, while 6 were judged to be at moderate risk of bias overall (Figure S7).

Discussion

Perianal fistulizing CD is a distinct and heterogeneous phenotype of CD, characterized by variability in anatomical features, clinical manifestations, and disease course. It carries a high risk of recurrence, markedly reduces quality of life, and requires a targeted and multidisciplinary approach.^18,63 In this updated meta-analysis, we included 14 studies assessing the efficacy of advanced therapies, including new data of efficacy of upadacitinib and risankizumab compared with placebo.

We found that infliximab was the only drug to demonstrate significant fistula induction and remission rates in both induction and maintenance trials; however, these data came from only one RCT for each phase²¹ (Figures 1 and 2). A Cochrane review including the same RCT concluded that “no conclusions could be drawn on the effects of infliximab compared to placebo for fistulizing participants regarding fistula remission, response, or withdrawals due to adverse events; the evidence was of very low certainty.”⁶⁴ In a subsequent Cochrane review on maintenance of remission,⁶⁵ the same RCT was included, and the authors concluded that infliximab may be superior to placebo for preventing loss of response in patients with fistulizing disease (not formally graded). Since then, no new RCTs have evaluated the efficacy of infliximab vs placebo for pCD. In summary, current evidence indicates that infliximab is effective for pCD, but this conclusion is based on only one dedicated RCTs exclusively enrolling patients with pCD for each phase, both published over 20 years ago using endpoints and trial design which may not be relevant today.

In the last decade, newer classes of advanced therapies have shown promise in patients with luminal CD in phase 3 clinical trials. Similar evidence is lacking for pCD, a population that remains underrepresented in clinical trials. In this review, both ustekinumab and upadacitinib were associated with significant improvements compared with placebo. Although risankizumab did not demonstrate a significant benefit, the available data were limited by the very small number of patients. In clinical trials assessing newer advanced therapies pCD, prior exposure to anti-TNF agents was often not reported, although most patients had likely failed anti-TNF therapy. In a SR of 10 observational studies including 333 patients with active perianal disease at baseline, the vast majority (>95% across all studies) had previous anti-TNF exposure, and 35.4% of patients achieved a response or fistula healing with ustekinumab.⁶⁶

In adult outpatients with CD and active perianal fistula, the 2021 AGA guideline recommended infliximab over no treatment and suggested adalimumab, ustekinumab, or vedolizumab over no treatment for the induction and maintenance of fistula remission.⁶⁷ The literature search, conducted in 2020, included only 6 placebo-controlled RCTs of biologic therapies and no data on small molecules.⁶⁸ Furthermore, the certainty of evidence was rated down to moderate to very low, mainly due to imprecision (as the optimal information size was not met or the 95% CIs were wide) and/or risk of bias (since randomization was not stratified by the presence or absence of fistula). Similarly, the ECCO guidelines recommend infliximab and adalimumab for both induction and maintenance of remission in complex pCD, while insufficient evidence supporting the use of ustekinumab, vedolizumab, and upadacitinib.⁶⁹ The same concerns remain today, although a few additional studies have been published: the number of events remains small, the evidence base for each advanced therapy is sparse, and most studies did not stratify randomization by the perianal fistula status. Several key aspects of surgical management remain also poorly defined, including the indication for seton placement, the optimal duration of drainage, and the timing and methods of treatment response assessment. The risk of bias was moderate to serious in post hoc analyses.

We acknowledge that although the point estimates varied across drugs (eg, RR ranging from 0.72 to 8.05 for induction of fistula response), the 95% CIs were wide and largely overlapping, which contributes to the low I² value (15.4% between drugs, Figure 1). This is likely due to the small sample sizes and the limited number of studies available for each drug. When few studies are included or sample sizes are small, I² may underestimate true heterogeneity, even when the effect estimates appear to differ substantially. Furthermore, although the low I² suggests limited statistical heterogeneity, it may not adequately reflect the clinical and methodological diversity across studies. Therefore, despite the variation in point estimates, the current evidence does not support a statistically significant difference between drugs. These findings highlight the need for larger, well-designed randomized controlled trials.

While the meta-analysis suggests beneficial effects of certain therapies, particularly infliximab, the strength of the conclusions should be tempered given the limitations of the underlying evidence base. There were noticeable differences in study designs. Eight and 9 studies were included in the induction and maintenance trials, respectively; however, only 3 and 1 RCTs directly randomized patients with pCD to active advanced therapy vs placebo in the induction and maintenance phases, respectively. Most studies relied on post hoc analyses of RCTs, in which patients with pCD were analyzed as a subgroup of the primary trial. In these cases, randomization was not stratified by the presence or absence of pCD, and the parent study was not specifically designed to evaluate the efficacy of advanced therapy in this population. We acknowledge that in clinical trials, and particularly in post hoc subgroup analyses of studies not specifically designed to evaluate perianal disease, missing outcome data may not be random. Patients without available outcome data are more likely to represent those with poorer clinical evolution, treatment discontinuation, adverse events, or inadequate assessment of fistula outcomes. Although our review found that two small RCTs had no missing outcome data and that the proportion of patients with missing outcome data in another two RCTs was not greater than what is commonly observed in IBD trials, this remains a potential concern. Notably, our sensitivity analysis using an intention-to-treat population for the two RCTs with missing outcome data, assuming that all patients without outcome data experienced treatment failure, yielded results nearly identical to the primary analyses across all outcomes. Nevertheless, given the small number of RCTs and the limited sample sizes, there remains a possibility that the treatment effect was overestimated in this review.

The heterogeneity in study design was more pronounced among maintenance trials, which varied in approach in the post hoc analyses. Some rerandomized patients after an open-label induction phase, some rerandomized patients following an induction RCT, and one used a treat-through design. These differences may result in clinically nonequivalent populations. Induction regimens, treatment durations, fistula status, and clinical response status of the randomized patients differed across studies. While certain maintenance trials included both clinical responders and nonresponders, others restricted inclusion to clinical responders after induction. Results obtained from induction responders may not be generalizable to nonresponders. In the study that reported outcomes for both responders and nonresponders, 12% and 18% of vedolizumab nonresponders achieved fistula closure at weeks 6 and 52, respectively, whereas 28% and 31% of vedolizumab responders achieved fistula closure at weeks 14 and 52, respectively.²⁰ After induction therapy, although all patients were clinical responders based on the CDAI criteria, baseline imbalances between the intervention and placebo groups may have existed. For example, in the GEMINI 2 trial, the proportion of patients with fistula closure at week 6 was 15% and 22% among those randomized to vedolizumab and placebo, respectively, following 6 weeks of vedolizumab induction therapy. Considering the substantial clinical heterogeneity of the included studies and their patient populations, we elected not to perform an updated NMA due to concerns about violating the fundamental assumption of transitivity.

As noted, the definitions of outcomes varied across the included trials, including differences in thresholds for reduction in drainage, criteria for closure of external openings, and definitions of clinical remission. These endpoints may not be entirely clinically equivalent, and pooling them could reduce the interpretability of the results. For example, in Sandborn 2004, outcomes were reported using two different response and two different remission definitions. In our analysis, we selected the definitions most commonly used across studies 50% closure of fistulas with no drainage on gentle compression for response, and 100% closure of fistulas for remission.¹⁶ Additionally, the outcomes pooled in this review primarily reflect clinical control of drainage rather than complete healing of the fistula tract, and these are related but distinct therapeutic goals in pCD. The effectiveness of interventions on fistulas may be more accurately assessed using quantitative imaging approaches, such as magnetic resonance imaging (MRI) using scoring systems like the Van Assche Index (VAI) or the novel Perianal Fistula Activity Index (PFAI). However, further studies are needed to validate the use of these scoring systems in clinical trials.⁷⁰ Future trials should follow a standardized framework for trial design and conduct, including eligibility criteria, use of composite endpoints, patient-reported outcomes, and timing of outcome assessments.⁶²

We acknowledge that efficacy data derived from RCTs or post hoc analyses of RCTs may not fully reflect outcomes observed in real-world clinical practice. For example, a systematic review and meta-analysis focused on ustekinumab combined one RCT with 24 observational studies, including case series.⁹ At 12 months, 54% and 42% of patients with pCD achieved a response or remission with ustekinumab. In contrast, in our meta-analysis, the rates of fistula response and fistula remission with ustekinumab from two small maintenance trials were as high as 62% (21/23) and 79% (11/14), respectively (Figures 3 and 4). In a real-world cohort of highly refractory CD patients, upadacitinib induction resulted in a clinical response of pCD in two-thirds of patients (67%),⁷¹ while only one study was included in our meta-analysis, with a 44% fistula response rate after upadacitinib in the induction trial. Conversely, the wide confidence intervals seen with upadacitinib reflect small sample sizes for newer agents. More robust data on the efficacy of newer advanced therapies are essential to optimize management for patients with pCD.

This review has several strengths. In addition to providing an updated review and including data for drugs not covered in previously published systematic reviews, we reassessed the eligibility of all previous included studies, re-extracted data from all eligible studies, removed concomitant therapies from placebo arms, evaluated the risk of bias using updated tools (ROB2.0 and ROBINS-I v2), and carefully reviewed the study designs and methodologies of the included trials. Induction and maintenance data were separated according to study design rather than solely by the timing of outcome assessment, given the potential overlap between these phases. Together, these efforts provide a comprehensive and accurate overview of the current evidence on this topic. Furthermore, in the primary RCTs, CDAI-defined clinical response after induction therapy encompassed patients with and without draining fistulas. To prevent confusion between these distinct concepts, the terms “clinical response” and “clinical remission” were not used to describe fistula response or fistula remission.

We acknowledge that this review has several limitations. Owing to the lack of available data, some drugs evaluated in CD guidelines, such as guselkumab and mirikizumab⁸ were not included in this review of pCD, whereas older agents such as CDP571 have not been investigated in more recent RCTs. Several ongoing trials are addressing this gap, notably the phase 3 FUZION CD placebo-controlled RCT assessing guselkumab in patients with fistulizing pCD,⁷² and the complete results from the USTAP trial assessing ustekinumab are awaited later this year.⁷³ Results from these studies, which focus specifically on pCD, will help inform the role of newer advanced therapies in this population. Novel therapeutic agents, including sphingosine-1-phosphate receptor modulators, miR-124 upregulators, anti-fractalkine (CX3CL1), anti-TL1A, peroxisome proliferator-activated receptor gamma agonists, TGFBRI/ALK5 inhibitors, anti-CCR9 agents, and other innovative small molecules, as well as combination therapies, are currently under investigation for pCD.⁷⁴ We noted that fistula complexity (simple vs complex) was not consistently reported across trials and therefore could not be incorporated analytically, even though it influences response in clinical practice. Finally, we only included placebo-controlled studies. RCTs comparing active treatments were not considered, including the ENTERPRISE study, which evaluated two vedolizumab regimens.⁵² This study reported rapid and sustained fistula closure with both regimens, with 54% of patients achieving ≥50% reduction in draining fistulas and 43% of patients achieving complete fistula closure at week 30.

Conclusions

Despite the growing availability of advanced therapies for CD, few dedicated trials have specifically evaluated their efficacy in pCD. Findings from our meta-analysis indicate that infliximab is effective for pCD, while limited evidence suggests potential benefits with adalimumab, ustekinumab, and upadacitinib. However, the overall efficacy of advanced therapies remains uncertain due to heterogeneity in study designs and variability in baseline patient characteristics. Well-designed RCTs or robust subgroup analyses focusing specifically on pCD and investigating newer advanced therapies are needed to better inform targeted treatment strategies for this challenging patient population, with appropriate stratification according to fistula phenotype and the adoption of standardized endpoints and patient-reported outcomes.

Supplementary material

Supplementary data are available at Inflammatory Bowel Diseases online.

Funding

None declared.

Conflicts of interest

Y.Y. None to declare. V.J. has received has received consulting/advisory board fees from AbbVie, Alimentiv, Anaptyis Bio, Arena pharmaceuticals, Asahi Kasei Pharma, Asieris, Astra Zeneca, Bristol Myers Squibb, Celltrion, Eli Lilly, Endpoint Health, Enthera, Ensho, Ferring, Flagship Pioneering, Fresenius Kabi, Galapagos, GlaxoSmithKline, Genentech, Gilead, Innomar, JAMP, Janssen, Merck, Metacrine, Mylan, MRM Health, Pandion, Pendopharm, Pfizer, Protagonist, Prometheus Biosciences, Reistone Biopharma, Roche, Roivant, Sandoz, Second Genome, Sorriso, Spyre, Synedgen, Takeda, Teva, Ventyx, Vividion; speaker’s fees from, Abbvie, Ferring, Bristol Myers Squibb, Eli Lilly, Fresenius Kabi, Janssen, Pfizer, Shire, Takeda, Tillotts. J.H. has received speaker’s fees from Abbvie, Eli Lilly, Janssen, and Takeda; consulting fees from Alimentiv Inc. M.H. has received advisory board fees and speaker’s fees from: Takeda, Celltrion Healthcare, Amgen, Abbvie, Pfizer, Johnson & Johnson. R.S. has received consulting fees from: AbbVie, Alimentiv, Pendopharm and Takeda. M.S. has received Speaker/advisory board fees from: Abbvie, Hikma, Ferring, Janssen, Acino, Pfizer, Sandoz, & Takeda. O.M.N. has received advisory board fees from: Eli Lilly, Nestlè, Janssen; Speaker fees from AbbVie, Janssen, Eli Lilly, Ferring, Alfa Sigma, Recordati, Noòs, and Pfizer. F.M. has received fees from: AbbVie, Amgen, Arena, Celgene/Bristol Myers Squibb, Falk, Ferring, Fresenius Kabi, Galapagos, Gilead, high5MD, Janssen, Johnson & Johnson, Lilly, Materia Prima, Merck Sharp & Dohme, Pharmacosmos, Pfizer, Sandoz/Hexal, Takeda, Tillotts, and Vifor. A.H. has received fees from: AbbVie, Amgen, Arena, Bristol Myers Squibb, Falk, Ferring, Galapagos, Janssen, Johnson & Johnson, Eli Lilly, Materia Prima, Pharmacosmos, Pfizer, Takeda, and Tillotts. S.K.V. has received consulting fee from Alimentiv Inc.

Data availability

The data supporting the findings of this review are available in the manuscript and its supplementary materials. All data used in this systematic review and meta-analysis were extracted from publicly available published studies.