Comparative Speed of Early Symptomatic Remission With Advanced Therapies for Moderate-to-Severe Ulcerative Colitis: A Systematic Review and Network Meta-Analysis

Abstract

INTRODUCTION: 

Rapidity of symptom resolution informs treatment choice in patients with moderate-severe ulcerative colitis (UC). We conducted a systematic review and network meta-analysis comparing early symptomatic remission with approved therapies.

METHODS: 

Through a systematic literature review to December 31, 2022, we identified randomized trials in adult outpatients with moderate-severe UC treated with approved therapies (tumor necrosis factor α antagonists, vedolizumab, ustekinumab, janus kinase inhibitors, or ozanimod), compared with each other or placebo, reporting rates of symptomatic remission (based on partial Mayo score, with resolution of rectal bleeding and near-normalization of stool frequency) at weeks 2, 4, and/or 6. We performed random-effects network meta-analysis using a frequentist approach and estimated relative risk (RR) and 95% confidence interval values.

RESULTS: 

On network meta-analysis, upadacitinib was more effective than all agents in achieving symptomatic remission at weeks 2 (range of RR, 2.85–6.27), 4 (range of RR, 1.78–2.37), and 6 (range of RR, 1.84–2.79). Tumor necrosis factor α antagonists and filgotinib, but not ustekinumab and vedolizumab, were more effective than ozanimod in achieving symptomatic remission at week 2, but not at weeks 4 and 6. With approximately 10% placebo-treated patients achieving symptomatic remission at 2 weeks, we estimated 68%, 22%, 23.7%, 23.9%, 22.2%, 18.4%, 15.7%, and 10.9% of upadacitinib-, filgotinib-, infliximab-, adalimumab-, golimumab-, ustekinumab-, vedolizumab-, and ozanimod-treated patients would achieve early symptomatic remission, ustekinumab and vedolizumab achieving rapid remission only in biologic-naïve patients.

DISCUSSION: 

In a systematic review and network meta-analysis, upadacitinib was most effective in achieving early symptomatic remission, whereas ozanimod was relatively slower acting.

INTRODUCTION

Rapid resolution of symptoms is a key treatment attribute for patients with moderate-severe ulcerative colitis (UC) to improve quality of life and avoid corticosteroids (^1,2). While this is essential for patients hospitalized with acute severe UC to decrease the risk of colectomy, it is also important for outpatients experiencing disruptive symptoms of active disease, particularly bloody diarrhea and urgency. Recent head-to-head trials and network meta-analyses have informed comparative efficacy of different therapies for induction of remission, but this outcome is typically assessed around weeks 8–12 (^3–5). Patients with debilitating symptoms often desire symptomatic relief before that time point, and earlier resolution of symptoms may help avoid concomitant corticosteroids. Therapies that block cytokines systemically may in theory have faster onset of action compared with therapies that act by inhibiting lymphocyte trafficking. Post hoc analyses of randomized controlled trials (RCT) have suggested that janus kinase inhibitors (JAKi) may start improving symptoms of rectal bleeding and stool frequency within 1–3 days of treatment initiation (^6–8). However, little is known about the comparative speed of different therapies for providing rapid symptom relief in patients with moderate-severe UC.

We conducted a systematic review and network meta-analysis of RCT of advanced therapies (biologic agents and oral small-molecule drugs) approved for the use of moderate-severe UC to examine comparative efficacy in achieving early symptomatic remission within 2 weeks of treatment initiation.

METHODS

This systematic review was performed using an a priori established protocol and is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension statement for systematic reviews incorporating network meta-analyses (⁹). We followed good research practices outlined in the International Society for Pharmacoeconomics and Outcomes Research report on interpreting indirect treatment comparisons and network meta-analysis for healthcare decision-making (¹⁰).

Study selection

We included phase II or III RCT that met the following inclusion criteria: (i) patients: adult (aged older than 17 years) outpatients with moderate to severely active UC (Mayo Clinic score 6–12, with an endoscopic subscore of 2 or 3); (ii) intervention: US Food and Drug Administration–approved and/or European Medicines Agency–approved advanced therapies including tumor necrosis factor (TNF)–α antagonists (infliximab, adalimumab, and golimumab), vedolizumab, ustekinumab, JAKi (including tofacitinib, filgotinib, and upadacitinib), or ozanimod, with a minimum duration of therapy of 14 days; (iii) comparator: another active intervention or placebo; and (iv) outcome: achieving symptomatic remission (based on partial Mayo score or 2-item patient-reported outcome [PRO-2] score with resolution of rectal bleeding and normalization or near-normalization of stool frequency) at weeks 2, 4, and/or 6. Data from these trials could be reported in the primary report, trial registry, post hoc analyses from the trial committee, or secondary analysis of patient-level data from these trials reported by independent investigators.

We excluded the following studies: (i) trials where outcome was not reported categorically, based on partial Mayo score or PRO-2 (i.e., trials in which outcome was reported on a continuous scale from Mayo score or where data were reported separately for rectal bleeding score and stool frequency score); (ii) trials of novel agents in development such as etrasimod or interleukin-23 antagonists; (iii) trials of agents not approved for use in moderate-severe UC (such as etrolizumab); or (iv) trials conducted in hospitalized patients with acute severe UC. We also excluded trials of methotrexate or thiopurines for moderate-severe UC (not recommended as monotherapy for induction of remission).

Search strategy

We updated our prior literature review on the comparative efficacy of approved therapies for the management of moderate-severe UC, with a focused search between October 1, 2019, and December 31, 2022, for phase II and III RCT of approved therapies, cross-referencing these with a recent systematic review and network meta-analysis of approved therapies for moderate-severe UC by Burr et al (^4,5). We performed a supplementary search for all post hoc analyses of these RCT including conference proceedings (Digestive Diseases Week, annual meeting of the European Crohn's and Colitis Organization, annual meeting of the American College of Gastroenterology, Crohn's and Colitis Congress, between 2020 and 2022). We also reviewed protocols registered on open science platforms such as Vivli and the Yale University Open Data Access project, to look for published aggregate results on the topic of interest. Two investigators independently reviewed these studies for inclusion. Details of the search strategy are shown in the Supplementary Appendix. In several instances, results from multiple induction trials of the same drug were presented together in post hoc analysis in a single manuscript (e.g., data from U-ACHIEVE and U-ACCOMPLISH rows in Tables 1 and 2) (⁸).

Data abstraction and risk of bias assessment

Data on study-, participant-, disease-, and treatment-related characteristics were abstracted onto a standardized form by 2 investigators independently, and discrepancies were resolved by consensus, referring back to the original article. One study investigator independently rated the quality of included trials using the Cochrane Risk of Bias Tool, updating our prior review.

Outcomes

The primary outcome was the comparative efficacy of different drugs in achieving early symptomatic remission, based on the partial Mayo score (most commonly defined as partial Mayo score ≤2 with all subscores ≤1) or PRO-2 (most commonly defined as stool frequency score ≤1 + rectal bleeding score 0) at week 2 (Table 1). The secondary outcomes were comparative efficacy of different drugs in achieving early symptomatic remission at weeks 4 and/or 6. When data for multiple doses of the same medication were available, only data for approved dose were considered. The denominator used in all trials was based on intention-to-treat analysis, and all dropouts were assumed to be treatment failures for the primary outcome.

Table 1.: Characteristics of trials included in the synthesis

Given potential differences in the speed of symptomatic relief in patients with prior exposure to biologics, we performed subgroup analyses in biologic-naïve and biologic-exposed patients.

Data synthesis and statistical analysis

Direct meta-analysis to calculate pooled relative risk (RR) and 95% confidence intervals (CI) was performed using the DerSimonian-Liard random-effects model (¹¹). We assessed statistical heterogeneity using the I² statistic, with values more than 50% suggesting substantial heterogeneity (¹²). Publication bias was assessed by evaluating small study effects by examining funnel plot asymmetry (¹³). Direct comparisons were performed using Comprehensive Meta-Analysis version 2 (Biostat, Englewood, NJ).

We performed a network meta-analysis using the frequentist model, with the statistical package “netmeta” (version 0.9-0, https://cran.r-project.org/web/packages/netmeta/index.html) in R (version 4.0.2) (¹⁴). This frequentist approach provides a point estimate from the network along with 95% CI from the frequency distribution of the estimate. We examined incoherence between results of head-to-head trials and indirect comparisons. We used the P score, which is a value between 0 and 1, for all advanced therapies vs placebo or each other, to rank them (¹⁵). P scores are based solely on point estimates and SE from the network estimates and measure the mean extent of certainty that one intervention is better than another, averaged over all competing interventions. Higher scores indicate a greater probability of the intervention being ranked as best.

To understand the potential absolute effect of different agents against a harmonized placebo, we calculated the pooled placebo rate of achieving symptomatic remission at weeks 2, 4, and 6 from the included studies (placebo rate: 10%, 16%, and 22%, respectively). Subsequently, we used RR of each agent vs placebo from the network meta-analysis for weeks 2, 4, and 6 to estimate the absolute risk rate of achieving early symptomatic remission with each agent, using GRADEpro version 3.6.1 (McMaster University, 2014).

RESULTS

Data to calculate symptomatic remission at week 2 were available from 14 RCT of 8 agents, including infliximab (ACT 1 and 2, reported in a secondary analysis of patient-level data) (¹⁶), adalimumab (ULTRA 1 and 2, NCT00853099, and VARSITY) (^3,17,18), golimumab (PURSUIT-SC, reported in a secondary analysis of patient-level data) (¹⁶), vedolizumab (GEMINI I and VARSITY) (^3,19), ustekinumab (UNIFI) (²⁰), ozanimod (TRUE NORTH) (²¹), filgotinib (SELECTION) (⁶), and upadacitinib (U-ACHIEVE and U-ACCOMPLISH) (⁸). Data on speed of early symptomatic remission with tofacitinib could not be included in the analysis because findings were not reported as partial Mayo score or PRO-2 at the desired time points; this data were available only for week 8. Instead, results in these trials were reported separately for achieving resolution of rectal bleeding and for normalization of stool frequency (⁷). The investigators observed up to one-third patients would have 1-point decline in stool frequency and/or rectal bleeding score within 3 days with tofacitinib (⁷). By 2 weeks, approximately 41% patients treated with tofacitinib had complete resolution of rectal bleeding, compared with 21% patients treated with placebo. The schematic diagram of study selection is shown in eFigure 1 (see Supplementary Digital Content 1, https://links.lww.com/AJG/C917), and available direct comparisons and network of trials are shown in eFigures 2A–C (see Supplementary Digital Content 2–4, https://links.lww.com/AJG/C918, https://links.lww.com/AJG/C919, https://links.lww.com/AJG/C920).

Trial and patient characteristics are summarized in Tables 1 and 2. Overall, the median of mean age of patients was 41 years (interquartile range [IQR] of mean age, 40–41.7), and 61.2% (IQR, 57.7–62.8) were males. The median disease duration was 7.3 years (IQR, 6.4–8.0), and 47.8% (IQR, 42.9–52.9) patients had extensive colitis. A median of 54.9% (IQR, 40.5–72.2) patients were treated with concomitant corticosteroids at baseline. A median of 19.7% (IQR, 0–42.2) of patients had prior exposure to TNF-α antagonists. Patients across all trials and treatment arms were similar for baseline prognostic variables, inclusion/exclusion criteria, and cointerventions. All outcomes were uniformly assessed based on standard definition of symptomatic remission with partial Mayo Clinic score (≤2 with all subscores ≤1) (9 trials) or PRO2-UC (5 trials). Overall, the studies were deemed to be at low risk of bias, and all included studies were industry sponsored.

Table 2.: Characteristics of patients included in the synthesis

Symptomatic remission at week 2

On direct meta-analysis, all agents except ozanimod were superior to placebo in achieving symptomatic remission at week 2, with the strongest effect size for upadacitinib (RR, 6.80; 95% CI, 3.94–11.76), with no heterogeneity in individual comparisons (see eFigure 3A, Supplementary Digital Content 5, https://links.lww.com/AJG/C921). These findings were confirmed on network meta-analysis comparing all agents against placebo. With approximately 10% of placebo-treated patients achieving symptomatic remission at week 2, we estimate 68%, 22%, 23.7%, 23.9%, 22.2%, 18.4%, 15.7%, and 10.9% of upadacitinib-, filgotinib-, infliximab-, adalimumab-, golimumab-, ustekinumab-, vedolizumab-, and ozanimod-treated patients, respectively, would achieve symptomatic remission (Figure 1).

Figure 1.: Estimated proportion of patients with moderate-severe ulcerative colitis achieving early symptomatic remission with different advanced therapies, based on relative risk estimates from network meta-analysis. These estimated rates are based on a pooled remission rate in the placebo arms at weeks 2, 4, and 6 of 10%, 16%, and 22%, respectively.

On comparison of active interventions, upadacitinib was more effective than all other medications in achieving symptomatic remission at week 2 (Table 3). In addition, filgotinib, infliximab, adalimumab and golimumab, but not ustekinumab and vedolizumab, were more effective than ozanimod in achieving symptomatic remission at week 2. Upadacitinib had the highest probability of being best (p score, 1.0), followed by adalimumab, infliximab, golimumab, and filgotinib (p score, 0.61–0.71) for achieving early symptomatic remission. On subgroup analysis examining biologic-naïve patients, all agents for which data were available (filgotinib, infliximab, golimumab, ustekinumab, and vedolizumab) were more effective than placebo in achieving rapid symptomatic remission (see eTable 1, Supplementary Digital Content 8, https://links.lww.com/AJG/C924). However, in a subset of patients with prior exposure to TNF-α antagonists, filgotinib, but not ustekinumab or vedolizumab, was effective in achieving rapid symptomatic remission at week 2; these data were not available for trials of TNF-α antagonists, upadacitinib or ozanimod.

Table 3.: Comparative efficacy of advanced therapies for achievement of rapid symptomatic remission at week 2 and week 6 in patients with moderate-severe ulcerative colitis, using network meta-analysis

Symptomatic remission at weeks 4 and 6

Week 4.

On direct meta-analysis, all agents were superior to placebo in achieving symptomatic remission at week 4 (data not available for infliximab) (see eFigure 3B, Supplementary Digital Content 6, https://links.lww.com/AJG/C922). These findings were confirmed on network meta-analysis comparing all agents against placebo. With approximately 16% placebo-treated patients achieving symptomatic remission at week 4, we estimate 59.8%, 28%, 33.6%, 25.4%, 28.6%, 26.7%, and 25.3% of upadacitinib-, filgotinib-, adalimumab-, golimumab-, ustekinumab-, vedolizumab- and ozanimod-treated patients, respectively, would achieve symptomatic remission (Figure 1). On comparison of active interventions, upadacitinib was more effective than all other medications in achieving symptomatic remission at week 4 (see eTable 2, Supplementary Digital Content 8, https://links.lww.com/AJG/C924). No other comparison of active interventions was significant. On subgroup analysis of biologic-naïve patients, all agents were more effective than placebo in achieving symptomatic remission at 4. However, in a subset of patients with prior exposure to TNF-α antagonists, filgotinib and ustekinumab, but not vedolizumab, were effective in achieving symptomatic remission at week 4; these data were not available for trials of TNF-α antagonists, upadacitinib or ozanimod.

Week 6.

On direct meta-analysis, all agents, except vedolizumab, were superior to placebo in achieving symptomatic remission at week 6 (data not available for ustekinumab) (see eFigure 3C, Supplementary Digital Content 7, https://links.lww.com/AJG/C923). Network meta-analysis confirmed all agents are more effective than placebo in achieving symptomatic remission. With approximately 22% placebo-treated patients achieving symptomatic remission at week 4, we estimate 83.8%, 44.7%, 45.5%, 30.1%, 37.2%, 36.5%, and 37% of upadacitinib-, filgotinib-, infliximab-, adalimumab-, golimumab-, vedolizumab-, and ozanimod-treated patients, respectively, would achieve symptomatic remission (Figure 1). On comparison of active interventions, upadacitinib was more effective than all other medications in achieving symptomatic remission at week 4 (Table 3). No other comparison of active interventions was significant, although there was a trend toward superiority of filgotinib over adalimumab (RR, 1.52; 95% CI, 0.99–2.31) and of infliximab over adalimumab (RR, 1.49; 95% CI, 0.99–2.24). On subgroup analysis of biologic-naïve patients, all agents were more effective than placebo in achieving symptomatic remission at week 6 (see eTable 3, Supplementary Digital Content 8, https://links.lww.com/AJG/C924).

Publication bias

There was no evidence of small study effects on the evaluation of funnel plot; however, the number of studies for each comparison was small, and we cannot reliably detect publication bias.

DISCUSSION

In this systematic review and network meta-analysis combining direct and indirect evidence from 14 trials comparing the speed of achieving rapid symptomatic resolution in patients with moderate-severe UC with different advanced therapies for, we made several key observations. First, we observed that upadacitinib is significantly most effective compared with all other advanced therapies in providing rapid symptomatic relief within 2 weeks. Second, ozanimod, a lymphocyte-trafficking inhibitor, has a relatively lower rate of achieving rapid symptomatic remission compared with JAKi and TNF-α antagonists. Third, some medications such as vedolizumab and ustekinumab may take longer to achieve symptom resolution in patients with prior exposure to TNF-α antagonists compared with biologic-naïve patients; this difference in speed of symptom resolution is not significant with JAKi. Overall, these findings provide useful data to facilitate shared decision-making in choosing therapies and inform clinical guidelines where speed of onset of action is considered an important outcome. Medications that provide rapid symptomatic relief may be used without the need for concomitant corticosteroids. Future head-to-head trials or well-designed observational studies comparing these therapies in patients with moderate-severe UC, those hospitalized with acute severe UC, are desired to inform positioning of these agents more comprehensively in achieving rapid symptom resolution and decreasing short-term risk of colectomy.

Rapid relief of symptoms is an important consideration of patients and physicians, particularly in those with severe symptoms, when choosing therapies (^1,2). Rapid achievement of remission is also a very favorable prognostic factor, with patients who achieve remission within 6 weeks achieving more durable long-term remission (^22,23). Typical period of symptomatic improvement with corticosteroids in patients with UC, frequently used in the setting of flares, is approximately 2 weeks but is associated with significant side effects, including mood changes, irritability, weight gain, and an increased risk of infections, and is not used for long-term maintenance (²⁴). In our study, we observed similar rapid improvement in symptoms with JAKi and TNF-α antagonists, particularly upadacitinib, which serve as both induction and maintenance therapies. Fast onset of action is also observed with tofacitinib, although the data could not be incorporated in this network meta-analysis. Hanauer and colleagues observed up to one-third patients would have 1-point decline in stool frequency and/or rectal bleeding score within 3 days with tofacitinib (⁷). By 2 weeks, approximately 41% patients treated with tofacitinib had complete resolution of rectal bleeding, compared with 21% patients treated with placebo. Although our study focused primarily on symptomatic remission, these agents have also been shown to induce rapid decline in biomarkers of inflammation (^8,19,20). This rapid improvement in symptoms and biomarkers within 2 weeks is associated with higher rates of achieving remission within 8–12 weeks and more durable remission with long-term therapy (^6,8).

We observed that lymphocyte-trafficking inhibitors may have lower rates of rapid symptomatic remission, compared with medications that directly block cytokine release such as JAKi or TNF-α antagonists, although these differences disappear at later time points of assessment such as reported in typical trials of induction therapy by weeks 8–14. This may reflect limited impact of trafficking inhibitors on existing activated inflammatory cells till exhaustion of local cytokines; by contrast, JAKi and TNF-α antagonists block downstream effects of cytokines and tend to be faster acting (²⁵). However, lymphocyte trafficking achieve increasing rates of symptomatic remission at later time points. For example, we observed the RR of achieving symptomatic remission with adalimumab vs vedolizumab decreased from 1.30 at week 2 to 0.98 at week 4 and 0.82 at week 6. Similarly, in comparing adalimumab vs ozanimod, RR decreased from 2.20 at week 2 to 1.11 at week 4 and 0.81 at week 6. Even in the VARSITY trial comparing vedolizumab vs adalimumab in patients with moderate to severely active UC, while the week 2 symptomatic response rates were very similar between the 2 agents, 17.1% of adalimumab-treated patients vs 13.6% of vedolizumab-treated patients achieved symptomatic remission by week 2; over a period of time, the trial demonstrated that vedolizumab was more effective than adalimumab in achieving remission at week 52³.

We observed differential speed of symptom resolution in patients with prior exposure to TNF-α antagonists, with vedolizumab and ustekinumab, but not with JAKi. The likelihood of achieving remission within 2 weeks among patients with prior exposure to TNF-α antagonists was no different with vedolizumab or ustekinumab vs placebo, although this was not observed with filgotinib. While similar data were not available for upadacitinib and tofacitinib, post hoc analyses of U-ACHIEVE, U-ACCOMPLISH, and OCTAVE trials have suggested a comparably rapid onset of action in those with vs without prior exposure to TNF-α antagonists (^7,8). Further understanding of how prior exposure to biologics modifies the local inflammatory milieu and likelihood of response to subsequent therapies is required.

The strengths of the study are inclusion of RCT and associated post hoc analyses to maximize robust available data; examining speed of symptom resolution using well-defined, fairly consistent definitions and at similar time points, and the use of a network meta-analysis approach to contextualize comparative speed of onset of different advanced therapies for moderate-severe UC. Across trials of induction therapy, key inclusion/exclusion criteria, outcome definitions, patient and clinical characteristics, and cointerventions were comparable across trials, which facilitated this network meta-analysis. However, there are important limitations to our analyses. First, with study-level synthesis, we were unable to account for potential differences in patient-level characteristics between these trials that can influence the speed of rapid symptom resolution. Second, symptoms may not always correlate with endoscopic findings, such that rapid resolution of symptoms may not imply achievement of mucosal healing. However, network meta-analyses of trial of induction therapy for moderate-severe UC have consistently demonstrated similarity between symptomatic remission and mucosal healing (⁵). Third, data for some medications were incomplete. For example, we were unable to synthesize data on tofacitinib in this review because the remission endpoint data based on stool frequency and rectal bleeding were not reported as such. We were unable to consistently examine differences in speed of onset of action by prior exposure to TNF-α antagonists, a key effect modifier.

In summary, based on a systematic review and network meta-analysis, we found that upadacitinib was superior to all other agents in achieving early symptomatic remission within 2 weeks. By contrast, ozanimod may take longer to achieve symptomatic remission compared with JAKi and TNF-α antagonists, though these differences resolve at later time points by weeks 4–6. These findings may help inform choice of pharmacotherapy for patients with moderate-severe UC regarding their presenting symptom burden—patients with severe limiting symptoms may benefit from JAKi or TNF-α antagonists, particularly upadacitinib. Given their rapidity of onset, JAKi should be explored in hospitalized patients with acute severe UC.

CONFLICTS OF INTEREST

Guarantor of the article: Siddharth Singh, MD, MS.

Specific author contributions: Study concept and design: S.S. Acquisition, analysis, and interpretation of data: D.A., M.H.M., and S.S. Draft of initial manuscript: S.S. Critical revision of the manuscript for important intellectual content: D.A., M.H.M., C.M., and V.J. Approval of the final manuscript: all the authors.

Financial support: S.S. is supported by NIDDK K23DK117058 and R03DK129631. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Potential competing interests: D.A. and M.H.M.: no relevant disclosures. C.M.: has received consulting fees from AbbVie, Alimentiv, Amgen, AVIR Pharma, BioJAMP, Bristol Myers Squibb, Celltrion, Ferring, Fresenius Kabi, Janssen, McKesson, Mylan, Takeda, Pendopharm, Pfizer, Prometheus Biosciences, Roche, Sanofi; speaker's fees from AbbVie, Amgen, AVIR Pharma, Alimentiv, Bristol Myers Squibb, Ferring, Fresenius Kabi, Janssen, Takeda, Pendopharm, and Pfizer; royalties from Springer Publishing; research support from Ferring, Pfizer. V.J.: has received has received consulting/advisory board fees from AbbVie, Alimentiv, Arena pharmaceuticals, Asahi Kasei Pharma, Asieris, Astra Zeneca, Bristol Myers Squibb, Celltrion, Eli Lilly, Ferring, Flagship Pioneering, Fresenius Kabi, Galapagos, GlaxoSmithKline, Genentech, Gilead, Janssen, Merck, Mylan, Pandion, Pendopharm, Pfizer, Protagonist, Prometheus, Reistone Biopharma, Roche, Sandoz, Second Genome, Takeda, Teva, Topivert, Ventyx, Vividion; speaker's fees from, Abbvie, Ferring, Galapagos, Janssen Pfizer Shire, Takeda, Fresenius Kabi. S.S.: has received research grants (AbbVie, Pfizer) and personal fees (Pfizer for ad hoc grant review).