Reimbursement Review

Risankizumab (Skyrizi)

Sponsor: AbbVie Corporation

Therapeutic area: Ulcerative colitis

This multi-part report includes:

Clinical Review

Pharmacoeconomic Review

Clinical Review

Abbreviations

Executive Summary

An overview of the submission details for the drug under review is provided in Table 1.

Table 1: Background Information of Application Submitted for Review

NOC = Notice of Compliance; UC = ulcerative colitis.

Introduction

Ulcerative colitis (UC) is a chronic disease characterized by inflammation, predominantly of the mucosal layer of the large intestine (colon) most often involving the rectum and frequently extends continuously into the proximal colon.¹ The cause of UC remains uncertain, but a combination of genetic and environmental factors contributes to immune dysregulation and up-regulation in response to micro-organisms in the gastrointestinal (GI) tract.² UC is characterized by blood in the stool with mucus, frequent diarrhea, urgency, loss of appetite, and tenesmus (severe rectal cramp or spasm).³ Although UC principally affects the GI tract, extraintestinal manifestations, such as arthritis, may also occur.⁴ While most patients have a mild to moderate disease course, about 10% to 15% experience an aggressive course.⁵ Relapse is common, with the cumulative risk of relapse being 70% to 80% at 10 years.⁵ The chronic nature of UC has a considerable impact on patients’ health-related quality of life (HRQoL), including psychological, physical, sexual, and social domains of HRQoL, due to chronic symptoms such as urgency, frequency, and incontinence.^6,7

The prevalence of UC in Canada was estimated to be 414 per 100,000 in 2023.⁸ It is estimated that 32% to 46% of Canadians with UC have moderate disease, and 13% to 14% have severe disease.⁹

The general goal of pharmacotherapy is to achieve complete remission, defined as both symptomatic and endoscopic remission without corticosteroid therapy; to preserve HRQoL; and to prevent disability.^10,11 Guidelines generally recommend a standard step-up approach to the medical management of moderate to severe UC. It is further recommended that for maintenance of remission, the same drug should be used in patients whose disease has responded to induction therapy with that drug.

The objective of this report is to review and critically appraise the evidence submitted by the sponsor on the beneficial and harmful effects of risankizumab in the treatment of adult patients with moderately to severely active UC who have experienced an inadequate response, loss of response, or were intolerant to conventional therapy, a biologic treatment, or a JAK inhibitor.

Perspectives of Patients, Clinicians, and Drug Programs

The information in this section is a summary of input provided by the patient and clinician groups who responded to our call for input and from clinical experts consulted for the purpose of this review.

Patient Input

Canada’s Drug Agency (CDA-AMC) received input from 2 patient groups, the Gastrointestinal (GI) Society and Crohn’s and Colitis Canada. Both advocate for patients with UC and their caregivers. The GI Society gathered information from online surveys and round tables, and through a one-to-one conversation with a patient living with UC and receiving risankizumab. Crohn’s and Colitis Canada gathered information from a survey, from its 2023 Impact of Inflammatory Bowel Disease in Canada report, and through interviews with patients who have had experience with risankizumab.

Respondents reported that sustained remission or treatment response is more important than relieving any 1 symptom and the constant concern that there will be future flares, possibly worse than the last, at unpredictable times, can disastrously disrupt patients’ lives. Patients also noted that they need treatments that improve quality of life, not cause more symptoms, pain, frustration, or hardship. They also emphasized that improved outcomes included improved quality of life; access to different treatment options; access to a treatment that works, results in less suffering, and involves less unnecessary usage of health care resources (hospital stay, surgeries, and diagnostic procedures); access to a treatment with no side effects; reduction of symptoms; and ease of administration (fewer injections, self-injection, or oral administration).

The GI Society interviewed a patient with UC who started taking risankizumab through a clinical trial in December 2022. After 2 to 3 weeks, the patient noticed a significant improvement. The patient’s injections were every 8 weeks, and the patient noticed that, by about week 6, they could feel that they needed the medication again.

Clinician Input

Input From Clinical Experts Consulted for This Review

According to the clinical experts consulted by CDA-AMC, more effective and safer treatments are welcomed, as the currently available options are limited. The clinical experts noted that treating UC can be challenging and the goal is to control both symptoms (e.g., frequency, urgency) and signs of UC (e.g., clinical and endoscopic remission) with AT as quickly as possible. There are many unmet needs in patients with UC because UC responds to the available treatments only in some patients, and UC may become refractory to current treatment options in some patients. In addition, patients need new treatments that are better tolerated, with rapid onset and enhanced safety profiles. Furthermore, there have been few SC options for treatment of UC to date.

The clinical experts noted that risankizumab should be available as a first-line option for treatment of moderate to severe UC and that previous treatment failure should not be required to begin this drug. The only previous therapy that they would consider as a reasonable option would be corticosteroids, but these are not an option for long-term maintenance therapy.

The clinical experts noted that any patient with moderate to severe UC would be a suitable candidate for risankizumab. Patients unable to complete a self-injection may require support administering the medication. With regards to the diagnosis, and staging of UC, both the diagnosis and treatment need to be made by a gastroenterologist. Misdiagnosis is unlikely, as endoscopy is needed to establish the diagnosis.

The clinical experts noted that the important outcomes can be categorized as follows:

• improvement in symptoms, including decreasing stool frequency, urgency, and bleeding; these have a direct impact on patients’ HRQoL and important end points, such as work or school participation, fatigue (by hindering patients from sleeping through the night), and overall function or activities of daily living

• sustained clinical remission, avoiding the need to add steroids (e.g., steroid-sparing and steroid-free remission with ongoing clinical remission)

• endoscopic bowel healing, which is the best marker to ensure that adverse outcomes such as surgery and hospitalization are avoided.

In practice, they noted that clinical scoring systems (e.g., Mayo and partial Mayo scores, rectal bleeding score), endoscopic outcomes, and histopathological evaluation are used to evaluate response to treatment. In addition, biomarkers (e.g., fecal calprotectin) and corticosteroid-free remission are used to monitor ongoing treatment response and during maintenance therapy. From the patient’s perspective, improvement in HRQoL is prudent to monitor.

The main issue with scoring systems is that there can be a level of subjectivity to them. Hard end points would include Mayo endoscopic subscore and histopathological evaluation. Following the Selecting Therapeutic Targets in Inflammatory Bowel Disease-II (Stride II) guidelines is imperative to ensuring appropriate ongoing clinical management.

The clinical experts noted that treatment discontinuation can be organized into 3 main categories:

• primary nonresponse during induction treatment, with ongoing evidence of objective disease, despite a robust attempt at induction treatment (and possible reinduction) and no meaningful improvement with symptom persistence.

• early improvement in symptoms and objective disease, followed by relapse with persistent symptoms and disease activity requiring corticosteroid treatment, either persistently or recurrently

• serious adverse events (SAEs) (e.g., anaphylactic reactions).

The clinical experts noted that, following diagnosis and staging of UC by a gastroenterologist, treatment can be prescribed by a gastroenterologist or general internist with experience and expertise in treatment of inflammatory bowel disease (IBD). Treatment can occur in any clinical setting, but most patients will be initiated in the outpatient setting.

Clinician Group Input

CDA-AMC received input from 1 clinician group, the Canadian IBD Physicians Group, which consists of 25 clinicians who met through videoconference in order to contribute to this submission.

Regarding the place in therapy, the clinician group believed that risankizumab should be used as the first choice in patients with moderately to severely active UC in whom conventional therapies have failed, who are not candidates for other conventional therapy, or who have failed 1 or more ATs. They also noted that UC patients with current or previous mild to moderate active disease, whose remission can be adequately maintained with conventional treatments, are least suitable for treatment with risankizumab.

Drug Program Input

The drug plans identified jurisdictional implementation issues related to relevant comparators, considerations for initiation of therapy, considerations for prescribing of therapy, generalizability, care provision issues, and system and economic issues. The clinical experts consulted by CDA-AMC weighed evidence from the INSPIRE and COMMAND trials and other clinical considerations to provide responses to the drug plans’ questions.

Clinical Evidence

Systematic Review

Description of Studies

The systematic review included 2 pivotal trials (INSPIRE and COMMAND). The INSPIRE trial was a phase III, randomized, placebo-controlled, double-blind, parallel-group induction study in adult patients with moderately to severely active UC who experienced an inadequate response or intolerance to conventional therapies or other ATs. Patients in the INSPIRE trial were randomized to risankizumab 1,200 mg IV (650 patients) or placebo (325 patients). Patients who did not achieve clinical response at week 12 were eligible to enter another blinded risankizumab treatment period (i.e., Induction Period 2, which lasted for an additional 12 weeks).

The COMMAND trial was a phase III, randomized, placebo-controlled, double-blind, parallel-group maintenance study in adult patients with UC previously enrolled in the INSPIRE induction trial whose disease had an adequate response to risankizumab. Patients were randomized to risankizumab 180 mg subcutaneously (SC) (179 patients), risankizumab 360 mg SC (186 patients), or placebo (183 patients). The placebo group were patients who discontinued risankizumab after the induction phase (risankizumab withdrawal). Combined, the 2 studies lasted up to 81 weeks and included a screening period that lasted up to 35 days, a 12-week double-blind induction period, and a 52-week maintenance period. Randomization was stratified by history of inadequate response to advanced therapy (AT) (yes, no), last risankizumab induction dose (IV 600 mg, 1,200 mg, or 1,800 mg), and clinical remission status (per local evaluation) at the last visit of the induction trial (yes, no). Patients who experienced inadequate response were eligible for risankizumab rescue therapy starting at week 16 of maintenance (1 dose of 1,200 mg or 1,800 mg followed by 360 mg every 8 weeks). Loss of response or inadequate response was based on clinical and endoscopic symptoms (rectal bleeding score at least 1 point greater than the week 0 value or endoscopic subscore of 2 or 3). Final patient follow-up occurred in April 2023.

The following efficacy outcomes were assessed to be most relevant to inform expert committee deliberations and finalized in consultation with members of the expert committee: clinical remission per adapted Mayo score; clinical response per adapted Mayo score; endoscopic improvement; histologic, endoscopic, and mucosal improvement (HEMI); histologic, endoscopic, and mucosal remission (HEMR); discontinuation of corticosteroid use in patients taking steroids at baseline (maintenance only); UC-related hospitalization; and the patient-reported outcome Inflammatory Bowel Disease Questionnaire (IBDQ). Select notable harms outcomes considered important were treatment-emergent adverse events (TEAEs), SAEs, and adverse events of special interest (AEOSIs).

The baseline characteristics of patients in the risankizumab and placebo treatment arms in the INSPIRE trial were generally well balanced. UC was highly refractory in the entire study population. This is shown by the proportion of patients who had extensive UC or pancolitis at baseline, in addition to a high proportion in whom ATs had previously failed (close to 51%). Furthermore, this study population had a high proportion of patients with a baseline adapted Mayo score greater than 7, and most patients had a baseline endoscopic subscore of 3.

Similarly, in the population in the maintenance study (COMMAND), UC was also highly refractory, as evinced by severe disease characteristics: extensive UC or pancolitis, mean disease duration, and previous failure of an AT (close to 75%), including failure of more than 2 ATs (around 25%) and failure of JAK inhibitors (around 15%).

Efficacy Results

In the induction trial (INSPIRE), all efficacy outcomes were analyzed using all randomized patients who received at least 1 dose of study drug, according to the treatment that they were randomized to. Similarly, in the maintenance trial (COMMAND), all efficacy outcomes were analyzed using all randomized patients who received at least 1 dose of study drug after receiving IV risankizumab for only 1 period of 12 weeks in the induction trial, according to the treatment that they were randomized to.

The primary end point of the INSPIRE trial was clinical remission at 12 weeks using the adapted Mayo score. The risk difference was 14.0% (95% CI, 10.0% to 18.0%) for risankizumab compared to placebo. Similarly, in the COMMAND trial, the risk difference at 52 weeks was 16.3% (95% CI, 7.4% to 25.3%) for risankizumab 180 mg and 14.2% (95% CI, 5.3% to 23.2%) for risankizumab 360 mg compared to placebo (risankizumab withdrawal).

The secondary outcomes were in line with the primary end point. In the INSPIRE trial, the risk difference for clinical response was 28.6% (95% CI, 22.3% to 34.8%) for risankizumab compared to placebo. In the COMMAND trial, the risk difference was 17.1% (95% CI, 7.5% to 26.6%) and 11.5% (95% CI, 1.7% to 21.2%) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal).

For endoscopic improvement, in the INSPIRE trial, the risk difference was 24.3% (95% CI, 19.3% to 29.4%) for risankizumab compared to placebo. In the COMMAND trial, the risk difference was 20.1% (95% CI, 10.6% to 29.6%) and 17.4% (95% CI, 7.9% to 26.9%) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal).

For HEMI, in the INSPIRE trial, the risk difference was 16.6% (95% CI, 12.3% to 21.0%) for risankizumab compared to placebo. In the COMMAND trial, the risk difference was 20.2% (95% CI, 11.2% to 29.2%) and 19.8% (95% CI, 10.8% to 28.8%) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal).

For HEMR, in the INSPIRE trial, the risk difference was 5.6% (95% CI, 3.5% to 7.7%) for risankizumab compared to placebo. In the COMMAND trial, the risk difference was 4.0% (95% CI, −2.2% to 10.3%) and 6.1% (95% CI, −0.3% to 12.5%) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal).

For discontinuation of corticosteroid use in patients taking steroids at baseline, which was measured only in the COMMAND trial, the risk difference was 28.4% (95% CI, 14.0% to 42.8%) and 20.7% (95% CI, 4.9% to 36.6%) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal).

Total IBDQ score ranges between 32 and 224, with higher scores representing better HRQoL. Using the IBDQ scale, the mean difference was 18.3 (95% CI, 13.38 to 23.25) points for risankizumab compared to placebo in the INSPIRE trial. In the COMMAND trial, the mean difference was 17.5 (95% CI, 8.01 to 27.06) and 15.2 (95% CI, 5.18 to 25.31) points for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal).

In the INSPIRE trial, the risk difference for UC-related hospitalization was −4.8% (95% CI, −7.3% to −2.2%) for risankizumab compared to placebo. In the COMMAND trial, the incidence rate difference was −2.5 per 100 patient-years (95% CI, −5.4 to 0.4) and −1.8 per 100 patient-years (95% CI, −5.0 to 1.3) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal).

Harms

The safety analyses in both trials included all randomized patients who received at least 1 dose of study drug, analyzed according to the treatment that they received. In the INSPIRE trial, the number of patients with at least 1 TEAE was similar between the risankizumab (42.4%) and placebo (49.4%) groups. The most common TEAE in the risankizumab group was COVID-19 (4.8%), and the most common TEAE in the placebo group was UC (refers to the worsening of the underlying disease) (10.2%). The proportion of patients with at least 1 SAE was higher in the placebo group (10.2%) than the risankizumab group (2.3%). The most common serious TEAEs in the risankizumab group were anemia, UC, and pulmonary embolism (2 patients [0.3%] each). In the placebo group, the most common serious TEAEs were UC (4.9%), anemia (0.6%), and anal fistula (0.6%). A higher proportion of patients in the placebo group (3.7%) discontinued treatment due to a TEAE than in the risankizumab group (0.6%). One patient in the risankizumab group died due to COVID-19. The most frequently occurring AEOSIs in both arms were hypersensitivity (risankizumab: 3.8%; placebo: 2.2%) and hepatic events (risankizumab: 1.5%; placebo: 4.0%).

In the COMMAND trial, the number of patients with at least 1 TEAE were similar between the risankizumab 180 mg (72.5%), risankizumab 360 mg (70.8%), and placebo (76.5%) groups. The proportion of patients with at least 1 SAE was similar across groups (5.2% with risankizumab 180 mg, 5.1% with risankizumab 360 mg, and 8.2% with placebo). In the risankizumab 180 mg group, the only serious TEAE occurring in more than 1 patient was appendicitis (2 patients, 1.0%). In the risankizumab 360 mg group, no serious TEAE occurred in more than 1 patient. In the placebo (risankizumab withdrawal) group, the most common serious TEAE was UC (2.6%); no other serious TEAE occurred in more than 1 patient. The proportions of patients who discontinued treatment due to a TEAE were highest in those taking risankizumab 360 mg (2.6%) compared to risankizumab 180 mg (1.6%) and placebo (1.5%) groups. Regarding AEOSIs, the risankizumab 180 mg group had the highest proportion of patients with hypersensitivity (10.4%). The risankizumab 360 mg group had the highest proportion of patients with hepatic events (6.7%).

Critical Appraisal

Internal Validity

INSPIRE trial: Randomization in the INSPIRE trial took place using a 2:1 computer-generated block randomization via interactive response technology. This is an adequate form of randomization and allocation concealment. Participants were stratified for major confounding factors (e.g., corticosteroid use, baseline Mayo score, and prior AT exposure). The baseline characteristics of the randomized groups were balanced, reflecting a proper randomization procedure. The trial was double-blind and placebo-controlled, with a detailed protocol including which concomitant medications were allowed and which changes in dosage were allowed after randomization. This decreases the probability of any deviations from the intended interventions during the trial. Concomitant medications were balanced across the treatment groups and were not expected to impact the efficacy results. The primary and secondary efficacy end points in INSPIRE were tested using a graphical multiple-testing procedure to ensure control of family-wise type I error at a significance level of 0.05 (2-sided), which was deemed adequate.

The sponsor considered the occurrence of appropriate intercurrent events (ICEs) that affect the interpretation of the effect of interest within the trial estimands. Data for patients who discontinued prematurely continued to be collected (i.e., treatment policy) while initiation or escalation of UC-related corticosteroids and the occurrence or UC-related surgery were considered as a nonresponse for binary end points and return to baseline for continuous end points. Regulatory authorities consider this approach adequate when the therapeutic intent is to induce remission in the short term.^12,13

With regard to missing outcome data, the sponsor was unable to provide information requested by the review team about ICEs before study discontinuation or sporadic missing data for reasons other than study discontinuation (e.g., missing visit). Therefore, it was not possible to perform a complete appraisal of the potential for bias due to missing data arising from reasons that are incompatible with the nonresponder imputation approach used by the sponsor for binary end points (and missing at random for continuous end points). There was a higher discontinuation rate in the placebo group (8.3%) than in the risankizumab group (2%), which can introduce attrition bias when nonresponder imputation is used and the reasons for study discontinuation (and possibly missing data) are incompatible with lack of efficacy. More than half of patients in the placebo group who withdrew did so because of lack of efficacy or AEs, suggesting that the assumption of nonresponse is acceptable to account for withdrawals. Sensitivity analyses were conducted for the primary end point, including a tipping-point analysis which supported that the statistical significance of the results would be maintained across a range of assumptions about the missing data. The outcome of UC-related hospitalization used an as-observed analysis; this approach considers that data are missing completely at random, which is unlikely to be reasonable and may introduce bias if there is an increased proportion of missing data (which could not be assessed).

To minimize detection bias, the trial used central endoscopic review with blinded assessors for objective outcomes, but it is important to note that this is only 1 component of the adapted Mayo score. For subjective outcomes, such as patient-reported outcomes (e.g., other components of the adapted Mayo score, IBDQ) and safety outcomes, these results are unlikely to be biased because the trial was double-blind. A prespecified protocol, published on ClinicalTrials.gov (NCT03398148), minimized selective outcome and analysis reporting bias. Overall, the INSPIRE trial is at low risk of bias for most end points, although uncertainty remains about potential attrition bias, which could not be fully assessed due to insufficient reporting.

The collection of information on harms included worsening UC as well as UC-related symptoms. Given that these are likely to be increased in the placebo group, the interpretation of the difference between groups is challenging. Further, it is implausible that there were significantly more SAEs reported in the placebo group than in the active intervention group, which would imply that the placebo effect is associated with SAEs.

With regard to the subgroup analyses, randomization in the INSPIRE trial was stratified by the presence of baseline corticosteroid use (yes, no), baseline adapted Mayo score (≤ 7, > 7), and a history of intolerance or inadequate response to ATs (0, 1, > 1 treatment). All other subgroup analyses may have broken randomization and therefore could have been affected by biases, including selection bias and confounding due to unadjusted differences in baseline characteristics. The purpose of these analyses was to demonstrate consistency, and they cannot be used to draw credible conclusions about effect modification.

COMMAND trial: The trial protocol was publicly available (NCT03398135) and the randomization and allocation concealment methods were similar to those in the INSPIRE trial, but with a 1:1:1 allocation ratio stratified by prior advanced therapy response and induction dose. The baseline characteristics suggested imbalances in some disease characteristics, though these did not appear to systematically favour any treatment group. Therefore, they are not expected to have an important impact on the results.

While the trial was double-blind and open-label, rescue with risankizumab 1,200 mg IV was allowed after week 16. Unblinded therapy may have introduced performance bias in the collection of data related to harms. For efficacy end points, a patient needing rescue therapy was assumed to be a nonresponder, which is appropriate and would not introduce bias. Concomitant medications were balanced across the treatment groups and are not expected to impact the efficacy results. In the COMMAND trial, the testing began with testing the primary end point at the prespecified significance level of 0.025 (2-sided) for each risankizumab dosage group compared to placebo. This was deemed to be an adequate approach.

The sponsor considered the occurrence of appropriate ICEs that affect the interpretation of the effect of interest within the trial estimands. Data for patients who discontinued prematurely continued to be collected (i.e., treatment policy) while initiation or escalation of UC-related corticosteroids, the occurrence or UC-related surgery, and the need for rescue treatment were considered as a nonresponse for binary end points and return to baseline for continuous end points. Given that the intent of the COMMAND trial is to demonstrate maintenance of treatment efficacy in the long term, considering treatment discontinuations as nonresponse may be a preferred approach.¹³ Nonetheless, a supplemental analysis using the nonresponder imputation approach for treatment discontinuations showed similar results to the primary analysis on the primary and all secondary end points.

It was not possible to perform a comprehensive appraisal of the risk of bias due to missing outcome data as a result of incomplete reporting in the COMMAND study report. The concern for potential unreported missing data may be higher for the COMMAND trial than for the INSPIRE trial, resulting from the longer period of follow-up. There was a higher discontinuation rate in all groups compared to the INSPIRE trial, with 9.8% of patients in the placebo group discontinuing, and 7.8% to 12.4% of patients in the risankizumab groups discontinuing. This high discontinuation rate can introduce attrition bias when nonresponder imputation is used and the reasons for study discontinuation are incompatible with lack of efficacy (if these patients’ data are considered missing). The potential for bias cannot be appraised because it is unclear how many patients were already imputed as nonresponders before study discontinuation (due to ICE, which appeared to be common; for example 19% to 43% of patients received rescue medication). On request from the review team, the sponsor was unable to provide this information. Sensitivity analyses were conducted for the primary end point, including a tipping-point analysis indicating that the statistical significance of the results would be maintained across a range of assumptions about the missing data. The outcome of UC-related hospitalization used an as-observed analysis, which considers data to be missing completely at random, which is unlikely to be reasonable and may introduce bias if there is an increased proportion of missing data (which could not be assessed).

In addition, the outcome “discontinuation of corticosteroid use at week 52 in patients taking steroids at baseline” was not assessed in the full cohort of randomized participants, and it is unclear whether the baseline characteristics in the patients are similar between the groups. Overall, it is considered that bias is possible in the harms data due to open-label rescue treatment and in the corticosteroid discontinuation end point due to possible loss of randomization. The risk of bias due to missing outcome data could be increased because of attrition and lack of information on the proportion of data missing. This concern was mitigated for the primary outcome via sensitivity analysis.

Information on harms included worsening UC as well as UC-related symptoms. Given that these are likely to be increased in the placebo group, the interpretation of the difference between groups is challenging. Additionally, the comparison may not reflect a true placebo group due to the potential for risankizumab rescue therapy, which was received by 43% of patients in the placebo group.

In the COMMAND trial, randomization was stratified by history of inadequate response to advanced therapy (yes, no), last risankizumab induction dose (IV 600 mg, 1,200 mg, or 1,800 mg), and clinical remission status (per local evaluation) at the last visit of the induction trial (yes, no). All other subgroup analyses may have broken randomization and therefore could have been affected by biases, including selection bias and confounding due to unadjusted differences in baseline characteristics.

It is also important to note that the placebo group in the COMMAND trial received risankizumab during the induction phase (risankizumab withdrawal). As such, there could have been a carry-over effect due to an inadequate washout.

External Validity

The populations of both trials had a history of disease that was highly refractory to previous UC treatments, and the clinical experts consulted by the review team confirmed that patients in the newer UC trials are expected to have UC that is more refractory to treatment. In addition, the trials excluded patients with a history of IL-23 inhibitor use. With regard to co-interventions, in the INSPIRE trial, initiating or increasing the dose of comedications was prohibited, while in the COMMAND trial, patients undergoing corticosteroid therapy were required to taper by week 8. With regard to the comparator, no direct comparison to an active intervention was available (only compared to placebo). Furthermore, the main primary and secondary outcomes used the Mayo scoring system, while this score is not commonly used in clinical practice to inform decision-making. In the subgroup analyses, several subgroups, including patients in North America, showed less clinically significant results. As a result, the results may not be generalizable and credible conclusions on effect modification cannot be drawn. These specific trial characteristics and results may impact generalizability and implementation in a real-world setting.

GRADE Summary of Findings and Certainty of the Evidence

The selection of outcomes for Grading of Recommendations Assessment, Development and Evaluation (GRADE) assessment was based on the sponsor’s Summary of Clinical Evidence, consultation with clinical experts, and input received from patient and clinician groups and public drug plans. The following list of outcomes was finalized in consultation with expert committee members.

INSPIRE trial:

• Clinical remission

• Clinical response

• Endoscopic improvement

• HEMI

• HEMR

• IBDQ

• SAEs

COMMAND trial:

• Clinical remission

• Clinical response

• Endoscopic improvement

• HEMI

• HEMR

• Discontinuation of corticosteroid use in patients taking steroids at baseline

• IBDQ

• SAEs

Table 2: Summary of Findings for Risankizumab vs. Placebo for Patients With UC (INSPIRE)

CI = confidence interval; HEMI = histologic, endoscopic, and mucosal improvement; HEMR = histologic, endoscopic, and mucosal remission; IBDQ = Inflammatory Bowel Disease Questionnaire; MD = mean difference; MID = minimal important difference; NR = not reported; RCT = randomized controlled trial; vs. = versus.

Note: Study limitations (which refer to internal validity or risk of bias), indirectness, imprecision of effects, and publication bias were considered when assessing the certainty of the evidence. All serious concerns in these domains that led to the rating down of the level of certainty are documented in the table footnotes. The proportion of missing data was not reported and therefore cannot be fully appraised.

^aAn empirically derived MID of 11% was identified in the literature¹⁴ for the between-group difference for this outcome. Rated down 1 level for imprecision (as the confidence intervals [CIs] show both potential benefit and little to no clinically important difference between risankizumab and placebo).

^bAn empirically derived MID of 14% was identified in the literature¹⁴ for the between-group difference for this outcome.

^cAn empirically derived MID of 12.5% was identified in the literature¹⁴ for the between-group difference for this outcome.

^dAn empirically derived MID was not identified for the between-group difference for this outcome; effects were appraised using the null.

^eAn empirically derived improvement of 15 points or more greater than placebo was considered be the MID for this outcome, as identified in the literature.^15,16 Rated down 1 level for imprecision, as the CIs show both potential benefit and little to no clinically important difference between risankizumab and placebo.

^fAn empirically derived MID of 6% was identified in the literature¹⁴ for the between-group difference for this outcome. Rated down 1 level for imprecision (as the CIs show both potential benefit and little to no clinically important difference between risankizumab and placebo). Rated down 1 level for serious indirectness, as the SAEs include symptoms of worsening UC, which complicates interpretation. The effect was judged as not different from placebo, given that a reduction in SAEs compared to placebo is implausible.

Source: INSPIRE Clinical Study Report.¹⁷ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Table 3: Summary of Findings for Risankizumab 180 mg SC vs. Placebo (Risankizumab Withdrawal) for Patients With UC (COMMAND)

CI = confidence interval; HEMI = histologic, endoscopic, and mucosal improvement; HEMR = histologic, endoscopic, and mucosal remission; IBDQ = Inflammatory Bowel Disease Questionnaire; MD = mean difference; MID = minimal important difference; NR = not reported; RCT = randomized controlled trial; vs. = versus.

Note: Study limitations (which refers to internal validity or risk of bias), indirectness, imprecision of effects, and publication bias were considered when assessing the certainty of the evidence. All serious concerns in these domains that led to the rating down of the level of certainty are documented in the table footnotes. The proportion of missing data was not reported and therefore cannot be fully appraised.

^aAn empirically derived MID of 11% was identified in the literature¹⁴ for the between-group difference for this outcome. Rated down 1 level for serious imprecision; the point estimate suggests clinically important benefit while the lower bound of the CI suggests little to no difference.

^bAn empirically derived MID of 14% was identified in the literature¹⁴ for the between-group difference for this outcome. Rated down 1 level of serious imprecision; the CI includes the potential for benefit and little to no difference.

^cAn empirically derived MID of 12.5% was identified in the literature¹⁴ for the between-group difference for this outcome. Rated down 1 level for imprecision, as the CIs shows both potential benefit and little to no difference between risankizumab and placebo.

^dAn empirically derived MID was not identified for the between-group difference for this outcome; effects were appraised using the null.

^eAn empirically derived MID was not identified for the between-group difference for this outcome; effects were appraised using the null. Rated down 1 level for imprecision, as the CIs show both potential benefit and little to no difference between risankizumab and placebo.

^fNo threshold of clinical importance could be established from the literature. An empirically derived MID was not identified for the between-group difference for this outcome. A difference of 15% between the groups was identified by the clinical experts consulted by CDA-AMC as a threshold of clinical importance for this outcome. Rated down 1 level for serious study limitations; the randomization was not stratified by baseline corticosteroid use; therefore, randomization in this population might not be upheld. Rated down 1 level for imprecision, as the CIs show both potential benefit and equivalence between risankizumab and placebo.

^gAn empirically derived improvement of 15 points or more greater than placebo was considered be the MID for this outcome. Rated down 1 level for imprecision, as the CIs show both potential benefit and little to no clinically important difference between risankizumab and placebo.

^hAn empirically derived MID of 6% was identified in the literature¹⁴ for the between-group difference for this outcome. Rated down 1 level for imprecision, as the CIs show both potential benefit and little to no clinically important difference between risankizumab and placebo. Rated down 1 level for serious indirectness, as SAEs include symptoms of worsening UC, which complicates interpretation.

Source: COMMAND Clinical Study Report.¹⁸ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Table 4: Summary of Findings for Risankizumab 360 mg SC vs. Placebo (Risankizumab Withdrawal) for Patients With UC (COMMAND)

CI = confidence interval; HEMI = histologic, endoscopic, and mucosal improvement; HEMR = histologic, endoscopic, and mucosal remission; IBDQ = Inflammatory Bowel Disease Questionnaire; MD = mean difference; MID = minimal important difference; NR = not reported; RCT = randomized controlled trial; vs. = versus.

Note: Study limitations (which refers to internal validity or risk of bias), inconsistency across studies, indirectness, imprecision of effects, and publication bias were considered when assessing the certainty of the evidence. All serious concerns in these domains that led to the rating down of the level of certainty are documented in the table footnotes. The proportion of missing data was not reported and therefore cannot be fully appraised.

^aAn empirically derived MID of 11% was identified in the literature¹⁴ for the between-group difference for this outcome. Rated down 1 level for serious imprecision; the point estimate suggests clinically important benefit while the lower bound of the CI suggests little to no difference.

^bAn empirically derived MID of 14% was identified in the literature¹⁴ for the between-group difference for this outcome. Rated down 1 level of serious imprecision; the CI includes the potential for benefit and little to no difference.

^cAn empirically derived MID of 12.5% was identified in the literature¹⁴ for the between-group difference for this outcome. Rated down 1 level for imprecision, as the CIs show both potential benefit and equivalence between risankizumab and placebo.

^dAn empirically derived MID was not identified for the between-group difference for this outcome; effects were appraised using the null.

^eAn empirically derived MID was not identified for the between-group difference for this outcome; effects were appraised using the null. Rated down 1 level for imprecision, as the CIs show both potential benefit and little to no difference between risankizumab and placebo.

^fNo threshold of clinical importance could be established from the literature. An empirically derived MID was not identified for the between-group difference for this outcome. A difference of 15% between the groups was identified by the clinical expert consulted by CDA-AMC as a threshold of clinical importance for this outcome. Rated down 1 level for serious study limitations; the randomization was not stratified by baseline corticosteroid use, therefore randomization in this population might not be upheld. Rated down 1 level for imprecision, as the CIs show both potential benefit and equivalence between risankizumab and placebo.

^gAn empirically derived improvement of 15 points or more greater than placebo was considered be the MID for this outcome. Rated down 1 level for imprecision, as the CIs show both potential benefit and little to no clinically important difference between risankizumab and placebo.

^hAn empirically derived MID of 6% was identified in the literature¹⁴ for the between-group difference for this outcome. Rated down 1 level for imprecision, as the CIs show both potential benefit and little to no clinically important difference between risankizumab and placebo). Rated down 1 level for serious indirectness, as SAEs include symptoms of worsening UC, which complicates interpretation.

Source: COMMAND Clinical Study Report.¹⁸ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Long-Term Extension Studies

The sponsor submitted an abstract¹⁹ for the COMMAND open-label extension study. Because of the lack of access to complete methods and results, the summary of this study is not reported in this section.

Indirect Evidence

Description of Studies

No direct comparison exists between risankizumab and other publicly reimbursed ATs or those recommended for reimbursement in Canada for moderately to severely active UC. In the absence of head-to-head randomized controlled trials (RCTs), the sponsor submitted an indirect treatment comparison (ITC). A network meta-analysis (NMA) was used to compare the efficacy and safety of risankizumab to other publicly reimbursed ATs or those recommended for reimbursement in Canada for moderately to severely active UC: adalimumab, infliximab, golimumab, vedolizumab, tofacitinib, ustekinumab, ozanimod, upadacitinib, mirikizumab, and etrasimod. The NMA included 28 phase III or higher randomized, double-blind trials. The NMA reported on clinical remission, clinical response, endoscopic improvement, adverse events (AEs), SAEs, and serious infections.

Induction

Efficacy: Results from the indirect comparison showed that, for the induction phase, risankizumab 1,200 mg was favoured over adalimumab (odds ratio [OR] █████ ███ ███ ████ ██ ████) and mirikizumab (OR █████ ███ ███ ████ ██ ████) for clinical response and to adalimumab (OR █████ ███ ███ ████ ██ ████), golimumab (OR █████ ███ ███ ████ ██ ████), mirikizumab (OR █████ ███ ███ ████ ██ ████), tofacitinib (OR █████ ███ ███ ████ ██ ████), and ustekinumab (OR █████ ███ ███ ████ ██ ████), with more patients in the risankizumab group achieving endoscopic improvement in the patient population who had not previously received AT (non-AT). Upadacitinib was favoured over risankizumab 1,200 mg for clinical remission (OR █████ ███ ███ ████ ██ ████), clinical response (OR █████ ███ ███ ████ ██ ████) and endoscopic improvement (OR █████ ███ ███ ████ ██ ████) in the population with previous inadequate response or intolerance to AT (AT-IR). There was no evidence of difference between risankizumab 1,200 mg and the other interventions in terms of efficacy during the induction period.

Safety: Risankizumab 1,200 mg was favoured over adalimumab (OR █████ ███ ███ ████ ██ ████), etrasimod (OR █████ ███ ███ ████ ██ ████), ozanimod (OR █████ ███ ███ ████ ██ ████), and tofacitinib (OR █████ ███ ███ ████ ██ ████), with fewer patients in the risankizumab 1,200 mg group having SAEs during the induction period. Etrasimod was favoured over risankizumab 1,200 mg (██ █████████ ███ ███ █████ ██ ████████) with regard to serious infections. There was no evidence of difference between risankizumab 1,200 mg and the other interventions in terms of safety during the induction phase.

Maintenance

Efficacy: For the maintenance phase, in the population with no previous inadequate response or intolerance to AT (non–AT-IR), upadacitinib 30 mg was favoured over risankizumab 180 mg (██ █████ ███ ███ ████ ██ ████) and risankizumab 360 mg (██ █████ ███ ███ ████ ██ ████), with fewer patients in the risankizumab groups having a clinical response in the non–AT-IR population during the maintenance period. There was no evidence of a difference between risankizumab and the other interventions in terms of efficacy during the maintenance phase.

For the maintenance phase, in AT-IR populations, risankizumab 180 mg was ████ ████████ ████████ ██ ████████████ █████████████ ███ ██ ██████ ███ ███ ████ ██ ██████ ████████████ ███ ██ █████ ███ ███ ████ ██ █████ ███ ███████████ ████████████ ███████████ ██ █████ ███ ███ ████ ██ ██████ ███████████ ███████ ██ █████ ███ ███ ████ ██ █████ ███ ████████ ██████████ ███████████ ███ █████ ███ ███ ████ ██ █████ ███ ████████████ █████████████ ███ ██ █████ ███ ███ ████ ██ █████ ████████████ ███ ██ █████ ███ ███ ████ ██ █████ ███ ████████ █████████ ███ ████████████ █████████████ ███ ██ █████ ███ ███ ████ ██ █████ ████████████ ███ ██ █████ ███ ███ ████ ██ █████ ███ ███████████ ███ █████ █████ ████ ██ █████ ███ ██████████ █████████. There was no evidence of a difference between risankizumab 180 mg and the other interventions in terms of efficacy during the maintenance phase.

Risankizumab 360 mg was ████ ████████ ████████ ██ ███████████ ███ █████ ███ ███ ████ ██ ██████ ████████████ █████████████ ███ ██ █████ ███ ███ ████ ██ █████ ████████████ ███ ██ █████ ███ ███ ████ ██ █████ ███ ███████████ ████████████ ███████████ ██ █████ ███ ███ ████ ██ ██████ ███████████ █████ ███ █████ ███ ███ ████ ██ ██████ ███████████ ██████ ███ █████ ███ ███ ████ ██ █████ ███ ████████ ██████████ ███████████ ███ █████ ███ ███ ████ ██ █████ ███ ████████████ █████████████ ███ ██ █ ███ ███ ███ ████ ██ █████ ████████████ ███ ██ █████ ███ ███ ████ ██ █████ ███ ████████ █████████ ███ █████████ ███ █████ ███ ███ ████ ██ ██████ ████████████ █████████████ ███ ██ █████ ███ ███ ████ ██ █████ ████████████ ███ ██ █████ ███ ███ ████ ██ █████ ███ ███████████ ████████████ ███████ ██ █████ ███ ███ ████ ██ ████ ███ ███████████ ███████ ██ █████ ███ ███ ████ ██ █████ ███ ██████████ █████████. There was no evidence of a difference between risankizumab 360 mg and the other interventions in terms of efficacy during the maintenance phase.

Safety: Risankizumab 360 mg was favoured over golimumab (██ █████ ███ ███ ████ ██ ████), with fewer patients with AEs in the risankizumab group during the maintenance phase. There was no evidence of a difference between risankizumab 180 mg or risankizumab 360 mg and the other interventions in terms of safety during the maintenance phase.

Critical Appraisal of the NMA

In the sponsor-submitted ITC, relevant RCTs were identified using a systematic review produced following accepted methodological guidance and based on an a priori protocol that outlined the inclusion and exclusion criteria. The search is 2 years old, and it is not known whether more recent studies are available, nor what impact that might have on the results. Selection, data extraction, and risk-of-bias assessments were conducted in duplicate by independent researchers, which is considered adequate. Risk of bias of the included studies was conducted using the Cochrane Risk-of-Bias tool. The overall bias of all included studies was deemed to be low. The assessments occurred at the study level, and not the outcome level, as suggested by Cochrane, and bias concerns may vary by effect estimate. The appraisal may not be applicable across all outcomes. There was no assessment of the risk of publication bias; therefore, its presence or absence cannot be confirmed.

The methods used for the NMA were considered to be appropriate, including the use of Bayesian framework, as its conduct adhered to relevant guidance. However, there were no sensitivity analyses to understand the potential impact of the chosen priors for between-study standard deviation. Overall, the patient populations, interventions, comparators, and outcomes (PICO) of the included studies were consistent with the overall review’s objective. The clinical experts consulted by the review team did not expect any major issues regarding the representativeness of the study populations enrolled in the RCTs that were included in the ITC in relation to the populations in Canada that may be eligible for treatment with risankizumab.

As is common in NMAs of treatments for UC, 1 overall concern was the heterogeneity in the patient populations, which included their baseline characteristics, average disease severity (and how this was defined across trials), disease duration, and the extent of disease. In addition, the lengths of the induction and maintenance periods were not uniform across the trials. There were some differences in the AT-IR definitions used across studies, and due to changes in the standard of care and available treatments over time, the number and type of prior treatments to which patients were exposed likely varied. There was also heterogeneity in the use of concurrent immunomodulators and corticosteroids. Additionally, the timing of the outcome assessments varied substantially, and the definitions of the efficacy end points of interest were not always consistent. Some relevant evidence was excluded due to misalignment in the assessment time points; while this was necessary to reduce heterogeneity, the impact on results is not known.

There were additional sources of heterogeneity in the maintenance phase. In treat-through (TT) studies, patients were randomized to treatment or placebo at baseline and outcomes were measured at the end of an induction phase and again at the end of a maintenance phase. In rerandomization studies (e.g., risankizumab studies), patients were randomized to induction treatment or placebo at baseline, with outcomes measured at the end of the induction phase; induction responders were then rerandomized to maintenance treatment or placebo, with outcomes measured at the end of the maintenance phase strictly among induction responders. Adjustments were applied to include both TT and rerandomized trial designs in the same NMA. This was necessary and has been applied to other NMAs for UC but relies on assumptions that cannot be fully validated. Placebo was a common comparator in the NMAs, although the placebo groups across studies are likely not equivalent, due to carry-over effects from different induction treatments. Differences in the corticosteroid tapering strategy (and whether this was included) may also differ across the studies. For both induction and maintenance phases, the information needed to comprehensively assess the level of heterogeneity was not always available for all studies. The notable clinical and methodological heterogeneity raises concern for intransitivity, which undermines the validity of the indirect comparisons, as it indicates that the assumption of exchangeability may not hold.

Relevant efforts were made to reduce or account for heterogeneity across the NMAs, including stratification of the results by AT-IR status, baseline risk meta-regression, and sensitivity analyses. The stratification of analyses by AT-IR and non–AT-IR populations for efficacy analyses is appropriate but limits generalizability of the results of either analysis to the overall indicated population. The interpretation of these subgroups faces some limitations. The baseline characteristics within subgroups were not always fully known (i.e., the full population was used for comparison), making it difficult to comprehensively assess comparability. It is not clear whether randomization was stratified by AT-IR status across all trials. Confounding is possible, as the effect estimates might not be arising from fully randomized groups. None of the baseline risk-adjusted models converged; therefore, it became impossible to apply this adjustment. Sensitivity analyses of some sources of heterogeneity were not possible.

The networks were sparsely populated, with several nodes centred around a single connection (placebo) in a star geometry, with only 1 to 2 trials against placebo per treatment and only a few head-to-head trials (i.e., within-study comparisons of different doses of the same treatment). This reduces the robustness of the NMA and makes most comparisons underpowered by the lack of direct evidence. Further, all evidence for risankizumab was against placebo, increasing uncertainty in the estimates for each outcome, and the consistency assumption could not be assessed. The heterogeneity and network sparsity are reflected in imprecise credible intervals (CrIs). Fixed-effects models were chosen for a small number of outcomes based on model fit; when it was a better fit than the random-effects model. However, given the expected heterogeneity, the fixed-effects model has the potential to underestimate the uncertainty in the results (i.e., width of the CrIs). There was poor model fit for the analysis of serious infections in the induction phase. Last, based on the ITC feasibility assessment, 1 study was not included in the NMA because the drug is not approved in Canada.

Eight outcomes of interest to this review were reported: clinical remission, clinical response, endoscopic improvement, AEs, SAEs, and serious infections. Interpretation of harms outcomes was complicated by inclusion of UC worsening and related symptoms as harms in the placebo groups, and potential differences in how harms were collected across studies. Some outcomes of importance to patients, such as corticosteroid-free clinical remission and HRQoL, were not included.

Studies Addressing Gaps in the Evidence From the Systematic Review

Description of Studies

This study by Panaccione et al. (2025).²⁰ was post hoc analysis of the pivotal INSPIRE and COMMAND trials specifically for AT-IR status. The objective of study was to assess the efficacy and safety of risankizumab induction and maintenance therapy in patients with moderately to severely active UC based on prior inadequate response or intolerance to AT (i.e., AT-IR status). For the efficacy analysis, the study included 472 patients in the non–AT-IR population and 503 patients in the AT-IR population for the induction phase, and 137 patients in the non–AT-IR population and 411 patients in the AT-IR population for the maintenance phase.

Induction Phase Efficacy Outcomes

During the induction phase, clinical outcomes at week 12 showed higher proportion of patients responding in the risankizumab 1,200 mg group than in the placebo group across both AT-IR and non–AT-IR subgroups. In the AT-IR subgroup, clinical remission rates were 11.4% for risankizumab and 4.3% for placebo (difference = 7.2%), while in the non–AT-IR subgroup, clinical remission rates were higher, at 29.7% and 8.4% (difference = 21.3%), respectively. Clinical response rates followed a similar trend, with 55.2% in the AT-IR subgroup and 73.8% in the non–AT-IR subgroup for risankizumab, compared to 31.2% and 40.6% for placebo (difference = 24.0% for AT-IR and 33.2% for non–AT‑IR subgroups).

Endoscopic outcomes also were higher in the risankizumab group. Endoscopic improvement at week 12 was observed in 25.9% of the AT‑IR subgroup and 47.6% of the non–AT‑IR subgroup receiving risankizumab, compared to 10.1% and 14.2% in the placebo groups (difference = 15.8% for AT‑IR and 33.2% for non–AT‑IR subgroups). Endoscopic remission rates were 4.8% (AT‑IR subgroup) and 16.7% (non–AT‑IR subgroup) for risankizumab, versus 3.0% and 3.9% for placebo (difference = 1.8% for AT‑IR and 12.8% for non–AT‑IR subgroups). Additionally, HEMI rates were 16.0% in the AT‑IR subgroup and 33.4% in the non–AT‑IR subgroup for risankizumab, compared to 7.1% and 8.4% for placebo, respectively (difference = 8.9% for AT‑IR and 25.1% for non–AT‑IR subgroups).

Maintenance Phase Efficacy Outcomes

During the maintenance period at week 52, the differences in clinical remission rates in the AT-IR subgroup versus placebo were 6.3% for risankizumab 360 mg and 13.4% for risankizumab 180 mg. In the non–AT-IR subgroup, remission rates versus placebo were 30.6% for risankizumab 360 mg and 19.8% for risankizumab 180 mg. Clinical response rates in the AT-IR subgroup versus placebo were 11.2% for risankizumab 360 mg and 17.8% for risankizumab 180 mg, while in the non–AT-IR subgroup, they were 7.5% and 11.1%, respectively.

Endoscopic outcomes were also higher in the risankizumab groups. Endoscopic improvement in the AT-IR subgroup versus placebo was observed in 8.4% for risankizumab 360 mg and 17.3% for risankizumab 180 mg, while in the non–AT-IR subgroup, rates were 40.6% and 24.2%, respectively. Endoscopic remission rates in the AT-IR subgroup versus placebo were 2.1% for risankizumab 360 mg and 5.6% for risankizumab 180 mg, and in the non–AT-IR subgroup were 31.6% for risankizumab 360 mg and 16.6% for risankizumab 180 mg. HEMI rates in the AT-IR subgroup versus placebo were 11.4% for risankizumab 360 mg and 17.1% for risankizumab 180 mg, and in the non–AT-IR subgroup were 40.4% for risankizumab 360 mg and 25.9% for risankizumab 180 mg.

Safety Outcomes

The proportions of patients with any TEAE in the AT-IR subgroup in the induction study was 44.1% and 53.5% in the risankizumab 1,200 mg and placebo groups, respectively.

The proportions of patients with any TEAE in the non–AT-IR subgroup in the induction study was 39.9% and 45.5% in the risankizumab 1,200 mg and placebo groups, respectively.

The proportions of patients with any TEAE in the AT-IR subgroup in the maintenance study was 69.4%, 74.7%, and 76.5% in risankizumab 360 mg, risankizumab 180 mg, and placebo groups, respectively.

The proportions of patients with any TEAE in the non–AT-IR subgroup in the maintenance study, was 75.0%, 66.0%, and 76.6% in risankizumab 360 mg, risankizumab 180 mg, and placebo groups, respectively.

Critical Appraisal of the Studies Addressing Gaps in the Evidence From the Systematic Review

This is a post hoc analysis. There is no type I error control, which increases the risk of type I error. Conversely, subgroups might not be powered to find significant differences. There is a risk of randomization being broken for some of the subgroups (those not included as stratification factors). Previous ATs were stratification factors, but classified differently in the post hoc analysis. The analysis could not distinguish whether observed differences across groups were due to AT-IR status or other differences between the groups. There are no tests for subgroup differences. Based on the previously noted reasons, credible conclusions about effect modification cannot be drawn, but the analysis can be used for hypothesis generation. In several cases the analysis suggests consistency in the direction of effect with the overall population in the trials, although magnitudes of effects differed.

Conclusions

Two pivotal phase III, randomized, placebo-controlled, double-blind, parallel-group trials for induction (INSPIRE) and maintenance (COMMMAND) phases in adult patients with moderately to severely active UC who experienced an inadequate response or intolerance to conventional therapies or other ATs were included in the CDA-AMC review. During the induction phase, moderate- to high-certainty evidence favoured risankizumab with regard to clinical response, clinical remission, endoscopic improvement, HEMI, and HEMR. Further, there was moderate-certainty evidence favouring risankizumab with regard to IBDQ and low-certainty evidence for little to no difference in SAEs. During the maintenance phase, there was high-certainty evidence favouring risankizumab for endoscopic improvement and HEMI. Further, there was moderate-certainty evidence favouring risankizumab for clinical response (for the 180 mg dose only, whereas the 360 mg dose showed little to no difference), clinical remission, HEMR, and IBDQ. Lastly, there was low-certainty evidence favouring risankizumab for discontinuation of corticosteroid use in patients taking steroids at baseline, and for little to no difference in SAEs.

There was some concern resulting from the inability to comprehensively appraise for risk of bias due to missing data, as this was not reported in the submitted materials. While the results of most subgroup analyses were similar in direction to the main analyses, they should be viewed as exploratory for hypothesis generation because the trial was not stratified by these subgroups. The occurrence of harms was relatively limited and aligned with the known safety profile of risankizumab. A study addressing gaps submitted by the sponsor suggested similar directions of effect for patients in the AT-IR and non–AT-IR groups, but the magnitude varied. Magnitudes of benefit appeared smaller for patients in the AT-IR group compared to those in the non–AT-IR groups in both the induction and maintenance phases. Conclusions on effect modification were not possible to draw from this study.

The sponsor-submitted NMA comparing risankizumab to other relevant comparators was affected by several limitations, including likely violation of the underlying assumptions, network sparsity, imprecision, and inconsistency of the results across outcomes. As a result, the relative treatment effects of risankizumab versus relevant comparators are subject to uncertainty and there is no conclusive evidence of which treatment may provide the preferred balance of benefits and harms.

Introduction

The objective of this report is to review and critically appraise the evidence submitted by the sponsor on the beneficial and harmful effects of risankizumab:

• 600 mg in 10 mL (60 mg/mL) by IV infusion

• 180 mg in 1.2 mL (150 mg/mL) by SC injection

• 360 mg in 2.4 mL (150 mg/mL) by SC injection.

The indication is for the treatment of adult patients with moderately to severely active UC who experienced an inadequate response, loss of response, or were intolerant to conventional therapy, a biologic treatment, or a JAK inhibitor.

Disease Background

Contents within this section have been informed by materials submitted by the sponsor and clinical expert input. The following have been summarized and validated by the review team.

Overview of the Condition

IBD is a group of diseases characterized by chronic recurrent, progressive inflammation of the GI tract.²¹ There are 2 main types of IBD: Crohn disease and UC. UC is a chronic disease characterized by inflammation, predominantly of the mucosal layer of the large intestine (colon) most often involving the rectum and frequently extends continuously into the proximal colon.¹ The cause of UC remains uncertain, but a combination of genetic and environmental factors contributes to immune dysregulation and up-regulation in response to micro-organisms in the GI tract.² UC is characterized by blood in the stool with mucus, frequent diarrhea, urgency, loss of appetite, and tenesmus (severe rectal cramp or spasm).³ Although UC principally affects the GI tract, extraintestinal manifestations may also occur, such as arthritis.⁴ There is no notable difference in frequency of UC among males and females.²² Although the risk of mortality from UC itself is low, the disease is associated with increased risk of other complications (e.g., respiratory diseases, colorectal cancer, lymphoma, and skin cancer) that result in higher premature mortality compared to the general population.²³ About 30% to 60% of UC patients first present with isolated proctitis (involvement is limited to the rectum).^24,25 Patients with proctitis are prone to proximal extension, associated with more colon involved in active disease, higher colectomy rates, increased need for AT, and higher hospitalization rates than patients who start with extensive colitis.^24,26,27 Among patients with isolated proctitis who are untreated for 1 year, the relapse rate is between 47% and 86%.²⁸ Resource utilization and direct and indirect costs are higher among patients with UC compared with matched controls, and in patients with moderate to severe UC the reported excess burden has been greatest.²⁹

There is no single test used to diagnose UC, and diagnosis is based on clinical examination, laboratory results, endoscopic findings, histopathology results, and exclusion of alternative diagnoses.³⁰

While most patients have a mild to moderate disease course, about 10% to 15% experience an aggressive course.⁵ Relapse is common, with the cumulative risk of relapse being 70% to 80% at 10 years.⁵ Achieving endoscopic healing earlier may be associated with reduced risk of future colectomy.⁵ The chronic nature of UC has a considerable impact on patients’ HRQoL, including psychological, physical, sexual, and social domains, due to chronic symptoms such as urgency, frequency, and incontinence.^6,7 The medical and surgical treatments for UC (e.g., colectomies) and their potential accompanying complications can also negatively impact HRQoL and productivity.^31-35 Individuals with UC are at greater risk of comorbid anxiety, depression, and impaired social interactions.^6,7,36,37 Patients with UC frequently report fatigue and sleep disturbance, as well as an inability to perform regular daily routines such as jobs or domestic chores.^32,38-40 Furthermore, the disease can impact the patients’ caregivers, family, workplace, and community.²³

Estimated Disease Prevalence

The prevalence of UC in Canada was estimated to be 414 per 100,000 in 2023.⁸ It is estimated that 32% to 46% of Canadians with UC have moderate disease, and 13% to 14% have severe disease.⁹

Standards of Therapy

Contents within this section have been informed by materials submitted by the sponsor and clinical expert input. The following have been summarized and validated by the review team.

Pharmacotherapy regimens for UC are evolving rapidly, but the general goal is to accomplish complete remission, defined as both symptomatic and endoscopic remission, without corticosteroid therapy; to preserve HRQoL; and to prevent disability.^10,11 The most recent clinical practice guidelines used in Canada for the medical management of UC are the 2015 Toronto Consensus for patients with UC who are not hospitalized,¹⁰ the 2021 European Crohn’s and Colitis Organisation (ECCO)¹¹ guideline, and the 2024 American Gastroenterological Association (AGA) guideline for the pharmacological management of moderate to severe UC.⁴¹ One of the main differences of the AGA guidelines is that they made recommendations specific to individual drugs rather than to classes, as they noted there are emerging data suggesting differences in efficacy even within a therapeutic class.

The guidelines generally recommend a standard step-up approach to the medical management of moderate to severe UC. The Toronto and ECCO guidelines recommend that patients should undergo a combination of oral corticosteroid therapy, 5-aminosalicylates, and/ or anti–tumour necrosis factor (TNF) agents (i.e., infliximab, adalimumab, and golimumab). Thiopurine, methotrexate, vedolizumab, tofacitinib, or ustekinumab may also be considered for induction of remission. It is further recommended that for maintenance of remission, the same drug should be used in patients who have responded to induction therapy with that drug. The AGA guidelines classified:

• infliximab, vedolizumab, ozanimod, etrasimod, upadacitinib, risankizumab, and guselkumab as high-efficacy medications

• golimumab, ustekinumab, tofacitinib, filgotinib, and mirikizumab as moderate-efficacy medications

• adalimumab as having lower efficacy.

In Canada, the following drugs are available for patients with moderate to severe UC:

• 5-aminosalicylates (e.g., mesalamine, sulfasalazine, olsalazine)

• corticosteroids

• immunosuppressive therapy (e.g., azathioprine, 6-mercaptopurine)

• anti-TNF therapies

• anti-integrin therapies (e.g., vedolizumab)

• anti–IL-12 and anti–IL-23 therapies (e.g., ustekinumab)

• other targeted therapies, such as JAK inhibitors (e.g., tofacitinib and upadacitinib) and sphingosine‑1‑phosphate (S1P) receptor agonists (e.g., ozanimod).^42,43

Mirikizumab, an anti–IL-23 therapy, received a Notice of Compliance in July 2023 and a recommendation for reimbursement with clinical criteria and/or conditions by CDA-AMC in November 2023.^44,45 The pan-Canadian Pharmaceutical Alliance (pCPA) negotiations for mirikizumab concluded in 2024, and it is currently being reimbursed in some jurisdictions.⁴⁶ Additionally, another S1P receptor modulator, etrasimod, received a Notice of Compliance in January 2024 and a final CDA-AMC recommendation in August 2024 to reimburse with clinical criteria and/or conditions, and it is currently undergoing pCPA negotiations.^47,48

Drug Under Review

Key characteristics of risankizumab are summarized in Table 5 with other treatments available for the treatment of adult patients with moderately to severely active UC who experienced an inadequate response, loss of response, or were intolerant to conventional therapy, a biologic treatment, or a JAK inhibitor.

The recommended dose of risankizumab is 1,200 mg administered by IV infusion at week 0, week 4, and week 8, followed by 180 mg or 360 mg administered by SC injection at week 12, and every 8 weeks thereafter.

Other indications include the treatment of adult patients with:

• moderate to severe plaque psoriasis

• active psoriatic arthritis

• moderately to severely active Crohn disease.

Risankizumab has been previously reviewed by CDA-AMC and received a recommendation to reimburse with clinical criteria and/or conditions for the following indications:

• the treatment of moderate to severe plaque psoriasis in adults

• the treatment of adults with moderately to severely active Crohn disease who experienced an inadequate response, intolerance, or a demonstrated dependence to corticosteroids; or an inadequate response, intolerance, or loss of response to immunomodulators or biologic therapies.

Risankizumab is a humanized immunoglobulin G1 (IgG1) monoclonal antibody that selectively binds with high affinity to the p19 subunit of human interleukin 23 (IL-23) cytokine and inhibits IL-23 signalling in cell-based assays, including the release of the proinflammatory cytokine IL-17. IL-23 is a naturally occurring cytokine that is involved in inflammatory and immune responses.

The sponsor’s reimbursement request is per the Health Canada–approved indication, for the treatment of adult patients with moderately to severely active UC who experienced an inadequate response, loss of response, or were intolerant to conventional therapy, a biologic treatment, or a JAK inhibitor. Risankizumab underwent review by Health Canada through a standard review pathway, and a Notice of Compliance was granted on October 10, 2024.

Table 5: Key Characteristics of Risankizumab and Main Comparators

6-MP = 6-mercaptopurine; AE = adverse event; b.i.d.= twice a day; CNS = central nervous system; CV = cardiovascular; DMARD = disease-modifying antirheumatic drug; HC = Health Canada; IgG1 = immunoglobulin G1; IgG1k = immunoglobulin G1-kappa; IL = interleukin; PRES = posterior reversible encephalopathy syndrome; S1P = sphingosine 1-phosphate; S1P1 = sphingosine 1-phosphate receptor subtype 1; SC = subcutaneous; TNF = tumour necrosis factor; TYK2 = tyrosine kinase 2; UC = ulcerative colitis; vs. = versus.

^aHealth Canada–approved indication.

Note: All the comparators in this table were included in ITC as well as pharmacoeconomic analyses. Adalimumab was included in the ITC, but it was not broken down by adalimumab biosimilars vs. reference biologic drug. Infliximab was included in the ITC, but it was not broken down by infliximab biosimilars vs. reference biologic drug. Tofacitinib was included in the ITC, but it was not broken down by tofacitinib generic vs. reference biologic drug.

Source: Product monographs.^45,50-72

Perspectives of Patients, Clinicians, and Drug Programs

The full patient and clinician group submissions received are available in the consolidated patient and clinician group input document for this review on the project website.

Patient Group Input

This section was prepared by the review team based on the input provided by patient groups.

CDA-AMC received 2 patient group inputs from GI Society and Crohn’s and Colitis Canada. The GI Society is a national charity committed to improving the lives of people with GI and liver conditions, supporting research, advocating for appropriate patient access to health care, and promoting GI and liver health. Crohn’s and Colitis Canada is a national health charity focused on finding cures for Crohn disease and UC through research, advocacy, and increasing awareness.

The GI Society gathered information through 1-to-1 conversation with a patient who is living with UC and is receiving the drug under review; a round table with gastroenterologists, patients, and patient groups across Canada in 2025; an online survey about the unmet needs of individuals living with IBD, with 514 respondents from Canada in 2024; a follow-up survey focusing on opinions regarding biologics and biosimilars with 55 respondents; interviews with 7 individuals living with IBD in 2023; a survey about the IBD patient journey with 54 respondents from Canada in 2022; a focus group with several patients with IBD in 2022; and 2 surveys regarding the unmet needs of patients with IBD in 2018 and 2020, with 432 and 579 respondents, respectively.

Crohn’s and Colitis Canada gathered information from Crohn and Colitis Canada’s report through its 2023 Impact of Inflammatory Bowel Disease in Canada report,- as well as through a survey conducted in 2022, in which 354 out of 1,706 survey respondents were patients with moderate to severe UC, and through interviews with patients who have had experience with the drug under review.

In terms of disease experience, the GI Society explained that diarrhea, pain, rectal bleeding, and loss of bowel control are common recurring symptoms of UC and anemia can result from blood loss due to ulcerations in the intestine and from general malnutrition due to decreased nutrient absorption. Some patients have extra-intestinal manifestations, including fever, inflammation of the eyes (uveitis) or joints (arthritis), ulcers of the mouth or skin, tender and inflamed nodules on the shins, and numerous other conditions. Anxiety, stress, suicidal thoughts, and poor mental health are major factors. UC often has a profound effect on an individual’s life — physically, emotionally, and socially — both at home and at school or in the workplace. Symptoms can be relentless, embarrassing, and scary. The severity of the disease can fluctuate, making it necessary to go through routine testing, reassessments, and medication changes. It is particularly difficult for children and young adults, because it often affects a person’s sense of self. According to the input, patients reported that sustained remission or treatment response is more important than relieving any single symptom. The constant concern about future flares, possibly worse than the last, at unpredictable times, can disastrously disrupt patients’ lives.

The GI Society noted that patients need treatments that improve quality of life, rather than cause more symptoms, pain, frustration, or hardship.

Regarding the currently available treatments, the patient group noted that patients with IBD still have a lot of difficulty achieving remission or adequate symptom relief. In the 2020 survey, 33% of respondents did not believe that their IBD was well-controlled by their current medications. In the 2024 survey, this number was 29%, compared with 38% who found it well-controlled and 33% who were unsure. When asked how concerned they were about running out of treatment options, 82% were at least somewhat concerned. Patients are still suffering, and each person living with IBD has a different experience. This is why it is so important that patients living with IBD have access to various effective treatments.

According to the patient group input, improved outcomes included improved quality of life; access to different treatment options; access to a treatment that works, which results in less patient suffering and involves less unnecessary usage of health care resources (hospital stay, surgeries, and diagnostic procedures); access to a treatment with no side effects; reduction of symptoms; and ease of administration (fewer injections, self-injection, or oral administration).

The GI Society had an interview with 1 patient with UC who started taking risankizumab through a clinical trial in December 2022. After 2 to 3 weeks, the patient noticed a significant improvement. The patient’s injections are every 8 weeks, and the patient noticed that, by about week 6, they could feel that they needed the medication again.

Clinician Input

Input From Clinical Experts Consulted for This Review

All CDA-AMC review teams include at least 1 clinical specialist with expertise regarding the diagnosis and management of the condition for which the drug is indicated. Clinical experts are a critical part of the review team and are involved in all phases of the review process (e.g., providing guidance on the development of the review protocol, assisting in the critical appraisal of clinical evidence, interpreting the clinical relevance of the results, and providing guidance on the potential place in therapy). The following input was provided by 2 clinical specialists with expertise in the diagnosis and management of UC.

Unmet Needs

The clinical experts noted that treating UC can be challenging. The goal is to control symptoms (e.g., frequency, urgency) and signs of UC (e.g., clinical and endoscopic remission) with AT as quickly as possible. There are many unmet needs in patients with UC because UC responds to the available treatments in only some patients and UC may become refractory to current treatment options in some patients. In addition, patients need new treatments that are better tolerated, with rapid onset and enhanced safety profiles. Furthermore, there have been few SC options for treatment of UC to date.

Place in Therapy

The clinical experts noted that risankizumab should be available as a first-line option for treatment of moderate to severe UC and that previous treatment failure should not be required to begin this drug. The only previous therapy that they would consider as a reasonable option would be corticosteroids, but these are not an option for long-term maintenance therapy.

Patient Population

The clinical experts noted that any patient with moderate to severe UC would be a suitable candidate for risankizumab. Patients unable to self-inject the drug may require support to administer the medication.

Both diagnosis and treatment need to be made by a gastroenterologist. Misdiagnosis is unlikely, as endoscopy is needed to establish the diagnosis.

Assessing the Response Treatment

The clinical experts noted the following important outcomes:

• improvement in symptoms, including decreasing stool frequency, urgency, and bleeding; these have a direct impact on patients’ HRQoL and important end points, such as work or school participation, fatigue (by hindering patients from sleeping through the night), and overall function or activities of daily living

• sustained clinical remission, avoiding the need to add steroids (e.g., steroid-sparing and steroid-free remission with ongoing clinical remission)

• endoscopic bowel healing, which is the best marker to ensure that adverse outcomes such as surgery and hospitalization are avoided.

In practice, they noted that clinical scoring systems (e.g., Mayo and partial Mayo scores, rectal bleeding score), endoscopic outcomes, and histopathological evaluation are used to evaluate response to treatment. In addition, biomarkers (e.g., fecal calprotectin) and corticosteroid-free remission are used to monitor ongoing treatment response, including during maintenance therapy. From the patient’s perspective, it is prudent to monitor improvement in HRQoL.

The main issue with scoring systems is that they may be somewhat subjective. Hard end points include the Mayo endoscopic subscore and histopathological evaluation. Clinicians must follow the Stride II guidelines to ensure appropriate ongoing clinical management.

Discontinuing Treatment

The clinical experts noted that treatment may be discontinued for 3 main reasons:

• primary nonresponse during induction treatment, with ongoing evidence of objective disease and no meaningful improvement of symptoms, despite a robust attempt at induction treatment (and possible repeated attempts)

• early improvement in symptoms and objective disease, followed by relapse and with persistent symptoms and disease activity requiring corticosteroid treatment, either persistently or recurrently

• SAEs (e.g., anaphylactic reactions).

Prescribing Considerations

The clinical experts noted that, following a diagnosis and staging of UC by a gastroenterologist, treatment can be prescribed by a gastroenterologist or general internist with experience and expertise in treatment of IBD. Treatment can be provided in any clinical setting, but, in most patients, it will be initiated in the outpatient setting.

Clinician Group Input

This section was prepared by the review team based on the input provided by clinician groups.

CDA-AMC received 1 clinician group input from Canadian Inflammatory Bowel Disease (IBD) Physicians Group. This group consists of more than 30 gastroenterologists in Canada with specific interest and expertise in IBD and with broad representation from across Canada. Twenty-five clinicians who met through videoconference contributed to this submission.

The clinician group strongly endorsed induction of remission with advanced targeted agents in patients with moderately to severely active UC. These ATs include anti-TNF drugs, alpha4beta7 integrin inhibitor vedolizumab, the JAK inhibitors tofacitinib and upadacitinib, the IL-12 and IL-23 inhibitor ustekinumab, sphingosine-1-phosphate receptor modulators ozanimod or etrasimod, or the IL-23 p19 inhibitor mirikizumab. The clinician group added that ATs were indicated in patients who were intolerant or had an inadequate response to conventional therapies (corticosteroids and immunomodulators such as thiopurines, azathioprine, or methotrexate).

According to the clinician group input, histologic healing is a potential future therapeutic target. The clinician group discussed that histologic remission, in conjunction with endoscopic remission (histo-endoscopic mucosal improvement and remission), has become increasingly important for new drugs to achieve. These measures are associated with a decreased likelihood of clinical relapse. The clinician group believes that rapid symptom relief, durability of treatment, and safety of treatment are important treatment goals, which is similar to patients’ wishes.

In terms of treatment gaps, the clinician group noted that, for moderately to severely active UC, conventional treatments for UC lack sustained efficacy and have long-term safety concerns. While ATs with varying mechanisms of action are available and have improved efficacy, a considerable proportion of patients (up to 20% to 30%) have disease that does not respond to induction therapy or loses response over time. As a result, there remains a significant medical need for additional efficacious, durable, and safe treatment options that provide symptom control and mucosal healing, improve quality of life, and reduce the risk of hospitalization and surgery in the long term for patients with moderately to severely active UC who have experienced inadequate response or intolerance to conventional therapies or ATs.

Regarding the place in therapy for patients with moderately to severely active UC, the clinician group believed that risankizumab could be used as the first choice in such patients in whom conventional therapy has failed, who are not candidates for other conventional therapy, or in whom 1 or more ATs have failed. The clinician group added that risankizumab is best suited to treat any such adult patient who have experienced an inadequate response, loss of response, or were intolerant to conventional therapy, a biologic treatment, or a JAK inhibitor. These patients are in most need of intervention, as they have active symptoms, lack long-term treatment options, and are at high risk of disease progression. On the other hand, UC patients with current or previous mild to moderate active disease, whose remission can be adequately maintained with conventional treatments, are least suitable level for treatment with risankizumab. The outcomes used to determine whether a patient is responding to treatment in clinical trials included clinical remission, clinical response, disease activity, patient-reported symptoms, stringent endoscopic and histologic outcomes that included mucosal improvement and healing, and HRQoL outcomes.

The Canadian IBD Physicians Group recommends discontinuing treatment with risankizumab if symptoms worsen or if there is an inadequate response, but this is not broadly anticipated. Available clinical data demonstrate that a large proportion of patients will respond to therapy during the first 6 months.

The clinician group noted that, during the induction phase, risankizumab needs to be administered in a clinic by a trained health care professional who is experienced in the management of patients with UC or is supervised by a health care professional with this experience.

Drug Program Input

The drug programs provide input on each drug being reviewed through the Reimbursement Review processes by identifying issues that may impact their ability to implement a recommendation. The implementation questions and corresponding responses from the clinical experts consulted for this review are summarized in Table 6.

Table 6: Summary of Drug Plan Input and Clinical Expert Response

AT = advanced therapy; IL = interleukin; UC = ulcerative colitis; vs. = versus.

Clinical Evidence

The objective of this clinical review report is to review and critically appraise the clinical evidence submitted by the sponsor on the beneficial and harmful effects of risankizumab:

• 600 mg in 10 mL (60 mg/mL) by IV infusion

• 180 mg in 1.2 mL (150 mg/mL) by SC injection

• 360 mg in 2.4 mL (150 mg/mL) by SC injection.

The indication is for the treatment of adult patients with moderately to severely active UC who have experienced an inadequate response, loss of response, or were intolerant to conventional therapy, a biologic treatment, or a JAK inhibitor. The focus is on comparing risankizumab to relevant comparators and identifying gaps in the current evidence.

A summary of the clinical evidence included by the sponsor in the review of risankizumab is presented in 4 sections with the CDA-AMC critical appraisal of the evidence included at the end of each section. The first section, the systematic review, includes pivotal studies and RCTs selected according to the sponsor’s systematic review protocol. The CDA-AMC assessment of the certainty of the evidence in this first section using the GRADE approach follows the critical appraisal of the evidence. The second section includes sponsor-submitted long-term extension studies; however, none were available. The third section includes indirect evidence from the sponsor. The fourth section includes additional studies that were considered by the sponsor to address important gaps in the systematic review evidence.

Included Studies

Clinical evidence from the following is included in the review and appraised in this document:

• 2 pivotal studies or RCTs identified in systematic review

• 1 NMA

• 1 additional study addressing gaps in evidence.

Systematic Review

Contents within this section have been informed by materials submitted by the sponsor. The following have been summarized and validated by the review team.

Description of Studies

Two pivotal trials, INSPIRE and COMMAND, were included in the systematic review. Characteristics of the included studies are summarized in Table 7.

The objective of these trials was to evaluate the efficacy and safety of risankizumab when administered as an induction (INSPIRE) or maintenance (COMMAND) therapy for patients with UC. The INSPIRE trial was a phase III, randomized, placebo-controlled, double-blind, parallel-group induction study in adult patients with moderately to severely active UC who experienced an inadequate response or intolerance to conventional therapies or other ATs. Patients in the INSPIRE trial who did not achieve clinical response at week 12 were eligible to enter another blinded risankizumab treatment period (i.e., Induction Period 2, which lasted for an additional 12 weeks). The COMMAND trial was a phase III, randomized, placebo-controlled, double-blind, parallel-group maintenance study in adult patients with UC previously enrolled in the INSPIRE induction trial who experienced an adequate response to risankizumab. Combined, the 2 studies lasted up to 81 weeks and included a screening period that lasted up to 35 days, a 12-week double-blind induction period, and a 52-week maintenance period. In both trials, randomization was performed using a web-based interactive response technology with block randomization methods.

During the INSPIRE trial, 977 patients were randomized in a 2:1 ratio to receive risankizumab or placebo across 261 sites in 41 countries, with 11 sites in Canada. Randomization was stratified by the presence of baseline corticosteroid use (yes, no), baseline adapted Mayo score (≤ 7, > 7), and a history of intolerance or inadequate response to ATs (0, 1, > 1 treatment). Study follow-up visits were scheduled at 4, 8, 12, and 24 weeks, and the final patient follow-up occurred in May 2023.

During the COMMAND trial, 584 patients were randomized in a 1:1:1 ratio to receive risankizumab 180 mg, risankizumab 360 mg, or placebo (risankizumab withdrawal) across 238 sites in 37 countries, with 11 sites in Canada. Randomization was stratified by history of inadequate response to advanced therapy (yes, no), last risankizumab induction dose (IV 600 mg, 1,200 mg, or 1,800 mg), and clinical remission status (per local evaluation) at the last visit of the induction trial (yes, no). Study follow-up visits were scheduled at 8, 16, 24, 32, 40, 48, and 52 weeks, and the final patient follow-up occurred in April 2023.

Table 7: Details of Studies Included in the Systematic Review

AE = adverse event; AT = advanced therapy; CRP = C-reactive protein; FACIT-F = Functional Assessment of Chronic Illness Therapy–Fatigue; HEMI = histologic, endoscopic, and mucosal improvement; IBDQ = Inflammatory Bowel Disease Questionnaire; SC = subcutaneously; SF-36 = Short Form (36) Health Survey; TEAE = treatment-emergent adverse event; UC = ulcerative colitis; WPAI-UC = Work Productivity and Activity Impairment Questionnaire-Ulcerative Colitis.

Sources: INSPIRE Clinical Study Report,¹⁷ COMMAND Clinical Study Report.¹⁸

Populations

Inclusion and Exclusion Criteria

The INSPIRE trial enrolled adult patients with moderately to severely active UC and a history of intolerance or inadequate response to 1 or more conventional or ATs for UC. Additionally, eligible patients had to have a diagnosis of UC for at least 3 months, an adapted Mayo score of 5 to 9 points, and an endoscopic subscore of 2 to 3 points. Exclusion criteria included prior exposure to p40 inhibitors (e.g., ustekinumab) or p19 inhibitors (e.g., risankizumab or mirikizumab).

The COMMAND trial included participants enrolled in a risankizumab induction trial who experienced adequate response to risankizumab at the 12- or 24-week follow-up. An adequate response was defined as a decrease of 30% or more from baseline and of 2 points or more on the adapted Mayo score, and a decrease of 1 or more in rectal bleeding score or an absolute rectal bleeding score of 1 or less.

Interventions

For the induction trial (INSPIRE), patients were randomized to receive 1,200 mg of IV risankizumab or matching placebo at weeks 0, 4, and 8. IV treatment was administered over 3 hours in an inpatient setting. Patients who failed to respond at the end of the 12-week period entered the Induction Period 2. In this phase, risankizumab dosing was based on treatment received during the initial induction period. Patients who received IV risankizumab during the first induction period were randomized to receive risankizumab 1,200 mg IV, risankizumab 180 mg SC, or risankizumab 360 mg SC. IV doses were administered at weeks 12, 16, and 20, and SC doses were administered at weeks 12 and 20. Patients who received placebo during the first induction period received risankizumab 1,200 mg IV at weeks 12, 16, and 20.

For the maintenance trial (COMMAND), patients were assigned to receive 180 or 360 mg of risankizumab or matching placebo administered SC every 8 weeks for 52 weeks.

In the induction trial, patients taking stable doses of corticosteroids, aminosalicylates, or immunomodulators at baseline continued these treatments throughout the trial. Initiating or increasing the dose of these medications was prohibited, except in the event of moderate to severe treatment-related toxicities. UC-related antibiotics could be discontinued in the blinded Induction Period 2; oral corticosteroids could also be tapered during this period but not increased above baseline. Rectal aminosalicylates were prohibited from 14 days before the screening period and for the entire duration of the study.

In the maintenance trial, patients taking stable doses of aminosalicylates and/or immunomodulators at baseline continued these treatments throughout the trial. Initiating or increasing the dose of these medications was prohibited, except in the event of moderate to severe treatment-related toxicities. Patients undergoing corticosteroid therapy were required to taper by week 8. Patients who lost satisfactory clinical response after starting to taper steroids could increase their corticosteroid dose again, up to the dose used at baseline of the induction study. Starting at week 16, patients in any treatment group could receive open-label risankizumab therapy (i.e., 1 single dose of risankizumab administered IV followed by 360 mg administered SC every 8 weeks) if they lost adequate clinical response (defined as a rectal bleeding score of ≥ 1 point greater than the week 0 value on 2 consecutive assessments 7 to 14 days apart or a Mayo endoscopic subscore of 2 or 3).

Outcomes

A list of efficacy end points assessed in this clinical review report is provided in Table 8, followed by descriptions of the outcome measures. Summarized end points are based on outcomes included in the sponsor’s Summary of Clinical Evidence as well as any outcomes identified as important to this review according to the clinical expert(s) consulted and input from patient and clinician groups and public drug plans. Using the same considerations, CDA-AMC selected end points considered to be most relevant to inform expert committee deliberations and finalized this list of end points in consultation with members of the expert committee. All summarized efficacy end points were assessed using GRADE. Select notable harms considered important for informing expert committee deliberations were also assessed using GRADE.

Table 8: Outcomes Summarized From the Studies Included in the Systematic Review

HEMI = histologic, endoscopic, and mucosal improvement; HEMR = histologic, endoscopic, and mucosal remission; IBDQ = Inflammatory Bowel Disease Questionnaire; TEAE = treatment-emergent adverse event; UC = ulcerative colitis.

^aP value adjusted for multiple testing.

Sources: INSPIRE Clinical Study Report,¹⁷ COMMAND Clinical Study Report.¹⁸ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Adapted Mayo Score

The adapted Mayo score consists of stool frequency, rectal bleeding, and endoscopic subscores, and is the most used method for measuring disease activity in UC (Table 9). Each subscore is scored from 0 to 3, and the total score ranges from 0 to 9, with higher scores indicating greater UC severity.^76,77 Stool frequency and rectal subscores were based on patient-derived assessments captured in their daily diaries, and the endoscopic subscore was centrally read based on endoscopic imaging assessments.

Clinical Remission per Adapted Mayo Score

Clinical remission was determined using the adapted Mayo score (stool frequency score ≤ 1 and not greater than baseline, rectal bleeding score of 0, and endoscopic subscore ≤ 1 without friability), which is consistent with regulatory guidance.^12,78,79 Notably, this definition for clinical remission is more stringent than definitions used in earlier UC trials.^80-83 No published estimate of the MID for the between-group difference has been identified for this outcome. Clinical experts consulted by the review team suggested that a 10% difference between groups would be considered clinically important.

In a recently published article that assessed the effect size thresholds to inform research, guidelines, and clinical decisions in patients with IBD, it was reported that the consensus results of thresholds for clinical remission are 11% ± 6% for trivial to small, 20% ± 8% for small to moderate, and 31% ± 13% for moderate to large.¹⁴

Clinical Response per Adapted Mayo Score

Clinical response on the adapted Mayo score was defined as a decrease of 30% or more and 2 points or more from baseline and a decrease in rectal bleeding score of 1 or more or an absolute rectal bleeding score of 1 or less (Table 9).¹² No published estimate of the MID for the between-group difference has been identified for this outcome. Clinical experts consulted by the review team suggested that a 10% difference between groups would be considered clinically important.

In a recently published article that assessed the effect size thresholds to inform research, guidelines, and clinical decisions in patients with IBD, it was reported that the consensus results of thresholds for clinical response are 14% ± 7% for trivial to small, 24% ± 10% for small to moderate, and 36% ± 15% for moderate to large thresholds.¹⁴

Endoscopic Improvement

Endoscopic improvement was defined as an endoscopic subscore of 1 or less without friability on the adapted Mayo score.¹² No published estimate of the MID for the between-group difference has been identified for this outcome.

In a recently published article that assessed the effect size thresholds to inform research, guidelines, and clinical decisions in patients with IBD, it was reported that the consensus results of thresholds for endoscopic response are 12.5% ± 7% for trivial to small, 22% ± 9% for small to moderate, and 33 ± 14% for moderate to large thresholds.¹⁴

Histologic, Endoscopic, and Mucosal Improvement

The Geboes score is the most commonly used histologic scoring system in UC. It classifies structural (architectural) changes as grade 0, chronic inflammatory infiltrate as grade 1, lamina propria neutrophils and eosinophils as grade 2, neutrophils in the epithelium as grade 3, crypt destruction as grade 4, and erosion as ulceration as grade 5. International consensus has defined histological improvement as a Geboes score ≤ 3.1.^84-88 HEMI was defined as an endoscopic subscore of 1 or less without friability on the adapted Mayo score and a Geboes score of 3.1 or less. No published estimate of the MID for the between-group difference has been identified for this outcome.

Histologic, Endoscopic, and Mucosal Remission

There is currently no universal consensus on how to define mucosal healing.¹² Mucosal healing was measured as HEMR, which was defined as achieving an endoscopic subscore of 0 on the adapted Mayo score and a Geboes score of less than 2.0. No published estimate of the MID for the between-group difference has been identified for this outcome.

Discontinuation of Corticosteroid Use at Week 52 in Patients Taking Steroids at Baseline

This end point involved discontinuation of corticosteroids at week 52 and for the entire 52-week maintenance period in patients taking corticosteroids at baseline in the induction study. No published estimate of the MID for the between-group difference has been identified for this outcome.

Inflammatory Bowel Disease Questionnaire

The IBDQ is a disease-specific, multidimensional, self-reported questionnaire comprising 32 Likert-scaled items (Table 9). The total score ranges from 32 to 224 points using 7-point response options, with higher scores indicating better HRQoL. The IBDQ contains 4 component subscales: bowel symptoms, systemic symptoms, emotional function, and social function. Each subscale can be computed, with total scores ranging from 10 to 70, 5 to 35, 12 to 84, and 5 to 35 points, respectively.^89-91 The IBDQ has been validated in a variety of settings, countries, and languages, and an absolute score of 170 points or greater indicates clinical remission.^89,92 An absolute score change of 16 to 30 points is considered the MID for the change from baseline and a score of 15 points or greater above the placebo score is considered the MID for the between-group difference.^15,16

UC-Related Hospitalization

This outcome included any UC-related event that resulted in an admission to the hospital for any length of time. Events per 100 person-years were also calculated for the maintenance trial. No MID for the between-group difference has been identified for this outcome.

Safety Evaluation Plan

TEAEs were recorded using the Medical Dictionary for Drug Regulatory Activities (MedDRA) System Organ Classes (SOCs) and preferred terms. The safety end points included any TEAE, TEAEs related to COVID-19, serious TEAEs, TEAEs leading to study drug discontinuation, deaths, and AEOSIs. SAEs included any of the following:

• death

• life-threatening events

• hospitalization or prolongation of hospitalization

• congenital anomaly

• persistent or significant disability or incapacity

• important medical event requiring medical or surgical intervention to prevent serious outcome.

Table 9: Summary of Outcome Measures and Their Measurement Properties

ES = endoscopic subscore; HRQoL = health-related quality of life; IBD = inflammatory bowel disease; IBDQ = Inflammatory Bowel Disease Questionnaire; ICC = intraclass correlation; MID = minimal important difference; OR = odds ratio; PGA = Physician’s Global Assessment; RBS = rectal bleeding subscore; SE = standard error; SFS = stool frequency subscore; SIBDQ = Short Inflammatory Bowel Disease Questionnaire; UC = ulcerative colitis; WPAI = Work Productivity and Activity Impairment Questionnaire.

Statistical Analysis

Sample Size and Power Calculation

In the INSPIRE trial, the sponsor calculated the sample size based on the expected proportion of patients who achieved clinical remission per the adapted Mayo score at week 12. A sample size of 966 patients (644 patients in the risankizumab group and 322 in placebo group) provided 90% power or greater to detect a treatment difference of 10% for the primary outcome, based on an assumption of 6% for clinical remission in the placebo group and 16% in the risankizumab group, using the 2-sided Miettinen and Nurminen test at a significance level of 0.05.⁸⁰ The sample size was determined to provide adequate power for the primary end point and selected ranked secondary end points (clinical remission per full Mayo score at week 12, endoscopic response at week 12, endoscopic remission at week 12, clinical response per adapted Mayo score at week 12, and clinical response per partial adapted Mayo score at week 4) and adequate patients with response to meet the requirement for this study. The assumed remission rates were based on substudy 1 and on ustekinumab data.

In the COMMAND trial, the sponsor calculated the sample size based on the expected proportion of patients who would achieve clinical remission per the adapted Mayo score at week 52. Assuming a clinical remission rate of 22% in the placebo group and 42% in the risankizumab groups, a sample size of 573 patients (191 in each treatment group) provided 90% power or greater to detect a treatment difference of 20% for the primary outcome using the 2-sided Miettinen and Nurminen test at a significance level of 0.025, adjusting the 0.05 significance level for the 2 dose comparisons.⁸⁰

Statistical and Analytic Plans

Estimands

A summary of statistical analysis methods for all efficacy end points in the INSPIRE and COMMAND trials is presented in Table 10. The INSPIRE study estimands for the primary and secondary end points considered the following ICEs: premature discontinuation of the study drug, initiation or dose escalation of UC-related corticosteroids, and UC-related surgery. For ICE1, data were collected regardless of premature discontinuation. For binary end points, patients with ICE2 and/or ICE3 were considered to have no response. For continuous end points, measurements on or after the date of the ICE were not used (considered returned to baseline).

The COMMAND study estimands for the primary and secondary end points considered the following ICEs: premature discontinuation of study drug, UC-related corticosteroid therapy, risankizumab rescue therapy, and UC-related surgery. For ICE1, data collected were used regardless of premature discontinuation in the primary analysis. In supplementary analyses (except for UC-related hospitalization), patients were considered to have no response at all visits after premature discontinuation. For ICEs 2, 3, and/or 4, patients were considered to have no response for binary end points; for continuous end points, measurements on or after the date of the ICE were not used (considered returned to baseline).

Analysis Methods

In the INSPIRE trial, categorical variables were analyzed using the Cochran-Mantel-Haenszel test for common risk difference, stratified by AT-IR status, baseline steroid use, and baseline adapted Mayo score. In the COMMAND trial, the same approach was used, but the stratification factors included AT-IR status, clinical remission status at week 0, and the last risankizumab induction dose. If there was a stratum for a treatment group that had no patient in it, a value of 0.1 was added to all cells in the corresponding table to prevent dividing by 0. Greenland and Robins’ variance estimator was used for the common risk difference.¹⁰¹ Values for all stratification factors used in Cochran-Mantel-Haenszel analyses were based on actual values. The results of binary outcomes were presented as risk differences with 95% CIs.

Continuous variables collected longitudinally (more than 1 post baseline visit) were analyzed using the mixed-effects model for repeated measurements. The results of continuous outcomes were presented as least squares mean differences with 95% CIs.

Missing Data Handling

Evaluations of patients with ICE were handled as mentioned previously in the Estimands section. Missing data could occur due to various reasons, including missing visits or assessments, early withdrawal from the clinical trials, COVID-19 infection or logistic restrictions, or geopolitical conflict in Ukraine and its surrounding areas.

In both the INSPIRE and COMMAND trials, for binary end points, a nonresponder imputation was used (except for UC-related hospitalization), with multiple imputation methods used for missing data due to COVID-19 or geopolitical conflict in Ukraine (assuming to be missing at random). For UC-related hospitalization, an “as-observed” analysis was undertaken with no imputation of missing values. Additionally, all values were included regardless of ICEs.

For continuous end points, “return to baseline” imputation was used for patients with ICEs 2, 3, or 4 (in the COMMAND trial only) in the primary analysis. Missing data were considered to be missing at random in the mixed-effects model for repeated measurements.

Sensitivity Analyses

Sensitivity analyses were conducted for the primary end point and continuous secondary end points, as described in Table 10.

Multiple-Testing Procedure

The primary and secondary efficacy end points in INSPIRE were tested using a graphical multiple-testing procedure to ensure control of family-wise type I error at a significance level alpha = 0.05 (2-sided).¹⁰² The secondary efficacy end points were divided into 2 groups. The first group included the first 10 secondary end points. The second group included all of the remaining 5 secondary end points, which were tested using the Holm procedure.¹⁰³ If the primary end point achieved significance, continued testing of the secondary outcomes followed a prespecified weight of allocation. No type I error control was applied to other study end points. The outcomes presented in the report for this trial are the results of the final analysis.

In the COMMAND trial, the testing began with testing the primary end point at the prespecified significance level of 0.025 (2-sided) for each risankizumab dose group compared to placebo. The secondary efficacy end points were divided into 2 groups. The first group included the first 12 secondary end points, and the second group included all of the remaining 5 secondary end points, which were tested using the Holm procedure.¹⁰³ If the primary end point achieved significance, continued testing of the secondary outcomes followed a prespecified weight of allocation as indicated in the graphical multiple-testing procedure. No type I error control was applied to other study end points. The outcomes presented in the report for this trial are the results of the final analysis.

Subgroup Analyses

In the INSPIRE trial, subgroup analyses were assessed using a chi-square test or Fisher's exact test if 20% or greater of the cells had an expected cell count less than 5. The nonresponder imputation method was used for primary analysis. In this method, clinical remission for patients with missing data at scheduled assessment visits was considered “not achieved.” No control for the type I error rate was applied. These analyses aimed to understand whether treatment effects were consistent across subgroups. Forest plots were used to graphically depict treatment effect estimates in the subgroups. No inferential statistics (P values) were produced. The prespecified subgroups included:

• sex

• age

• race

• baseline corticosteroid use

• baseline immunosuppressant use

• baseline adapted Mayo score

• baseline partial adapted Mayo score

• prior exposure to AT

• prior response to AT

• baseline weight

• baseline presence of pancolitis

• baseline disease duration

• baseline high-sensitivity C-reactive protein (hs-CRP)

• baseline albumin

• baseline calprotectin

• geographic region.

Subgroup analyses in the COMMAND trial were conducted using the same methods as in the INSPIRE trial. In addition to the subgroups presented in the INSPIRE trial, the COMMAND trial also included the following prespecified subgroups:

• week 0 clinical remission status

• last risankizumab induction dose.

Table 10: Statistical Analysis of Efficacy End Points

ANCOVA = analysis of covariance; AT-IR = inadequate response or intolerance to advanced therapy; CMH = Cochran-Mantel-Haenszel; HEMI = histologic, endoscopic, and mucosal improvement; HEMR = histologic, endoscopic, and mucosal remission; IBDQ = Inflammatory Bowel Disease Questionnaire; ICE = intercurrent event; UC = ulcerative colitis; vs. = versus.

Sources: INSPIRE Statistical Analysis Plan,¹⁰⁴ COMMAND Statistical Analysis Plan.¹⁰⁵ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Analysis Populations

The analysis populations in the INSPIRE and COMMAND trials included an intention-to-treat (ITT) population and safety analysis set (Table 11). In both trials, the ITT population was used for all efficacy analyses. The ITT population used in INSPIRE was the ITT2 population, which included all randomized patients who received at least 1 dose of study drug. The ITT population used in COMMAND was the ITT1RN_A population, which included all randomized patients who received at least 1 dose of study drug after receiving IV risankizumab for only 1 period of 12 weeks in the induction trial. Additionally, an efficacy analysis was conducted in patients who entered Induction Period 2 of the INSPIRE trial, which included all patients who received at least 1 dose of risankizumab during this period (ITT2_P2). The COMMAND trial also included supplementary efficacy analyses in 2 subsets of patients. The first subset was those in the ITT1RN_A population who received risankizumab 1,200 mg IV in the INSPIRE trial (ITT1RN_B). The second subset was those who received at least 1 dose of study drug in the COMMAND trial after receiving risankizumab 180 mg or 360 mg SC during Induction Period 2 in the INSPIRE trial or placebo during Induction Period 1 in the INSPIRE trial (ITT1NRN).

The safety analysis sets used in the safety analyses were the SA2 population and SA1RN population in the INSPIRE and COMMAND trials, respectively. For both trials, this population included all randomized patients who received at least 1 dose of study drug.

Table 11: Analysis Populations of the INSPIRE and COMMAND Trials

ITT = intention to treat; SA = safety analysis.

Sources: INSPIRE Clinical Study Report,¹⁷ COMMAND Clinical Study Report.¹⁸ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Results

Patient Disposition

A total of 1,430 patients were screened and 977 were randomized in the INSPIRE trial (Table 12). The reasons for screening failure in 453 participants (31.7%) were not reported. Two patients randomized to risankizumab did not receive the study drug; 1 patient was unable to receive IV treatment and the other withdrew from the study before treatment. The ITT population included 650 patients in the risankizumab group and 325 in the placebo group. By the end of the study, discontinuation of the study drug was higher in the placebo group (8.3%) than the risankizumab group (2.0%). The most common primary reasons for withdrawal in the placebo group were AEs, withdrawal of consent, and lack of efficacy; in the risankizumab group, these were AEs, withdrawal of consent, and “other.” A similar number of patients discontinued for similar reasons in the 2 groups.

In the COMMAND trial, 754 patients were screened and 584 were randomized. Of the 584 randomized patients, 14 patients in the risankizumab 180 mg group, 9 in the risankizumab 360 mg group, and 13 in the placebo (risankizumab withdrawal) group had a clinical response before 24 weeks during the induction trial and were excluded from the ITT population; as the ITT population was defined as all randomized patients who received at least 1 dose of study drug after receiving IV risankizumab for only 1 period of 12 weeks in the induction trial. As such, 548 patients were included in the ITT population: risankizumab 180 mg = 179 patients, risankizumab 360 mg = 186 patients, placebo (risankizumab withdrawal) = 183 patients. A similar proportion of patients discontinued the study drug across the treatment groups, including 6.7% in the risankizumab 180 mg group, 11.3% in the risankizumab 360 mg group, and 9.8% in the placebo group. The most common reasons for discontinuation were lack of efficacy and withdrawal of consent in the risankizumab groups, and withdrawal of consent and “other” in the placebo group. A similar number of patients discontinued for similar reasons among the groups. A higher proportion of patients in the placebo group (42.6%) compared to the risankizumab 180 mg (19.0%) and risankizumab 360 mg (36.3%) groups required rescue treatment during the study.

Baseline Characteristics

The baseline characteristics outlined in Table 13 are limited to those that are most relevant to this review or were felt to affect the outcomes or interpretation of the study results. Across both risankizumab and placebo treatment arms in the INSPIRE trial, baseline characteristics were generally well balanced. The total study population had highly refractory UC. This is shown by the proportion of patients who had extensive UC or pancolitis at baseline, in addition to a high proportion in whom ATs had previously failed (close to 51%). Specifically, these patients had prior failure of anti-TNFs, vedolizumab, or JAK inhibitors. Furthermore, this study population had a high proportion of patients with a baseline adapted Mayo score greater than 7, and most patients had a baseline endoscopic subscore of 3. Similarly, the maintenance (COMMAND) study population also had highly refractory UC with severe disease characteristics, indicated by the proportion of patients with extensive UC or pancolitis, mean disease duration, and a large proportion of patients in whom ATs had previously failed (close to 75%), including the failure of more than 2 ATs (around 25%) and failure of JAK inhibitors (around 15%).

Table 12: Summary of Patient Disposition From the INSPIRE and COMMAND Trials

ITT = intention-to-treat; NA = not applicable; NR = not reported; SC = subcutaneous.

Sources: INSPIRE Clinical Study Report,¹⁷ COMMAND Clinical Study Report,¹⁸ Louis et al.⁷³ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Table 13: Summary of Baseline Characteristics From the INSPIRE and COMMAND Trials

AT = advanced therapy; AT-IR = inadequate response or intolerance to advanced therapy; BMI = body mass index; IQR = interquartile range; SC = subcutaneous; SD = standard deviation; TNF = tumour necrosis factor.

Notes: Where indicated, study participants with missing data for a given characteristic were excluded from the calculation of summary statistics.

^aN = 647 in the risankizumab 1,200 mg IV arm; N = 177 in the risankizumab 180 mg SC arm.

^bN = 649 in the risankizumab 1,200 mg IV arm.

^cN = 602 in the risankizumab 1,200 mg IV arm; N = 302 in the placebo arm (INSPIRE); N = 151 in the risankizumab 180 mg SC arm; N = 166 in the risankizumab 360 mg SC arm; N = 162 in the placebo arm (COMMAND).

Sources: INSPIRE Clinical Study Report,¹⁷ COMMAND Clinical Study Report,¹⁸ Louis et al.⁷³ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Exposure to Study Treatments

Exposure and adherence to study treatments in the INSPIRE (SA2 population) and COMMAND trials (SA1RN population) are summarized in Table 14. Treatment adherence was high and exposure to study treatments was similar between treatment arms in both trials.

Table 14: Summary of Patient Exposure From the INSPIRE and COMMAND Trials

NA = not applicable; SC = subcutaneous; SD = standard deviation.

^aTreatment compliance defined as 100% × (1 – (absolute value of (n of kits received – n of kits planned)) / n of kits planned during induction period).

Sources: INSPIRE Clinical Study Report,¹⁷ COMMAND Clinical Study Report.¹⁸ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Concomitant Medications and Co-interventions

In the INSPIRE trial, most patients (99.7% in both treatment groups) used at least 1 concomitant medication (Table 15). Concomitant medications were generally well balanced between treatment groups.

In the COMMAND trial, patients undergoing corticosteroid therapy were required to taper by week 8. Starting at week 16, patients in any treatment group could receive open-label risankizumab rescue therapy (i.e., 1 single dose of IV risankizumab followed by 360 mg administered subcutaneously every 8 weeks) in the event of loss of adequate clinical response. Most patients (> 98% in each treatment group) used at least 1 concomitant medication (Table 15).

Table 15: Summary of Concomitant Treatment From the INSPIRE and COMMAND Trials

SC = subcutaneous.

Sources: INSPIRE Clinical Study Report,¹⁷ COMMAND Clinical Study Report.¹⁸ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Efficacy

A summary of the key efficacy results are presented in Table 16.

Table 16: Summary of Key Efficacy Results From Studies Included in the Systematic Review

CI = confidence interval; HEMI = histologic, endoscopic, and mucosal improvement; HEMR = histologic, endoscopic, and mucosal remission; IBDQ = Inflammatory Bowel Disease Questionnaire; LS = least squares; NA = not applicable; RD = risk difference; SC = subcutaneous; UC = ulcerative colitis; vs. = versus.

^aP value adjusted for multiple testing.

Sources: INSPIRE Clinical Study Report,¹⁷ COMMAND Clinical Study Report.¹⁸ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Clinical Remission per Adapted Mayo Score

The primary end point of the INSPIRE trial was clinical remission at 12 weeks. The risk difference was 14.0% (95% CI, 10.0% to 18.0%) compared to placebo, favouring risankizumab. Patients who did not experience clinical response to risankizumab 1,200 mg IV at week 12 (ITT2_P2 population) continued treatment with risankizumab up to week 24. At week 24, the proportion of patients who achieved clinical remission were risankizumab 180 mg SC: 9 of 71 (12.7%), risankizumab 360 mg SC: 11 of 70 (15.7%), risankizumab 1,200 mg IV: 6 of 68 (8.8%). For patients who received placebo during induction and were then rerandomized to receive risankizumab 1,200 mg IV during extended induction, 22% (38 of 173) achieved clinical remission at week 24.¹⁷

In the COMMAND trial, the risk differences versus placebo at 52 weeks were 16.3% (95% CI, 7.4% to 25.3%), favouring risankizumab 180 mg and 14.2% (5.3% to 23.2%), favouring risankizumab 360 mg.

Subgroup analyses in the INSPIRE and COMMAND studies showed similar results to the main analyses, except for a few notable exceptions. In the INSPIRE trial, patients aged 65 years or older, patients in North America, patients with baseline immunosuppressant use, patients with a history of more than 1 AT, and patients with AT-IR (with or without baseline steroid use) had a lower magnitude of effect. In the COMMAND trial, the effect estimates were consistent in direction.

In both the INSPIRE and COMMAND studies, sensitivity analyses using as-observed data, with FDA rules for calculating stool frequency and rectal bleeding subscores, and excluding patients impacted by geopolitical conflict had similar results as the primary analyses. The tipping-point analyses showed that the results remained statistically significant across a range of assumed values for patients with missing data in the placebo and risankizumab groups (including the most extreme case). In COMMAND, a supplemental analysis using the composite strategy (i.e., patients who discontinued treatment were classified as nonresponders) also had similar results as the primary analysis.

Clinical Response per Adapted Mayo Score

In the INSPIRE trial, the risk difference for clinical response was 28.6% (95% CI, 22.3% to 34.8%) compared to placebo, favouring risankizumab. In the COMMAND trial, the risk difference was 17.1% (95% CI, 7.5% to 26.6%) and 11.5% (95% CI, 1.7% to 21.2%) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal), favouring risankizumab. A supplemental analysis using the composite strategy (i.e., patients who discontinued treatment were classified as nonresponders) had similar results as the primary analysis. Similar results were reported for the subgroup of patients who received risankizumab 1,200 mg induction therapy in the INSPIRE trial (ITT1RN_B population).¹⁸

Among patients who did not experience clinical response to risankizumab 1,200 mg IV at week 12 (ITT2_P2 population), treatment with risankizumab up to week 24 demonstrated potential benefit for clinical response after Induction Period 2 (Table 17). At week 24, the proportion of patients who achieved clinical response per adapted Mayo score were comparable across all risankizumab treatment arms. Furthermore, for subjects who received placebo during induction and were then rerandomized to receive risankizumab 1,200 mg IV during extended induction, the proportion of subjects who achieved clinical response per adapted Mayo score at week 24 was comparable to that in the risankizumab 1,200 mg IV treatment arm at week 12 of the induction phase.¹⁷

Table 17: Clinical Response Per Adapted Mayo Score in Patients Who Entered Induction Period 2 in the INSPIRE Trial (ITT2_P2 Population)

SC = subcutaneous.

Sources: INSPIRE Clinical Study Report.¹⁷

Endoscopic Improvement

In the INSPIRE trial, the risk difference for endoscopic improvement was 24.3% (95% CI, 19.3% to 29.4%) compared to placebo, favouring risankizumab. In the COMMAND trial, the risk difference was 20.1% (95% CI, 10.6% to 29.6%) and 17.4% (95% CI, 7.9% to 26.9%) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal), favouring risankizumab. A supplemental analysis using the composite strategy (i.e., patients who discontinued treatment were classified as nonresponders) had similar results as the primary analysis. Similar results were reported for the subgroup of patients who received risankizumab 1,200 mg induction therapy in the INSPIRE trial (ITT1RN_B population).¹⁸

Histologic, Endoscopic, and Mucosal Improvement

In the INSPIRE trial, the risk difference for HEMI was 16.6% (95% CI, 12.3% to 21.0%) compared to placebo, favouring risankizumab. In the COMMAND trial, the risk difference was 20.2% (95% CI, 11.2% to 29.2%) and 19.8% (95% CI, 10.8% to 28.8%) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal), favouring risankizumab. A supplemental analysis using the composite strategy (i.e., patients who discontinued treatment were classified as nonresponders) had similar results as the primary analysis. Similar results were reported for the subgroup of patients who received risankizumab 1,200 mg induction therapy in the INSPIRE trial (ITT1RN_B population).¹⁸

Histologic, Endoscopic, and Mucosal Remission

In the INSPIRE trial, the risk difference for HEMR was 5.6% (95% CI, 3.5% to 7.7%) compared to placebo, favouring risankizumab. In the COMMAND trial, the evidence was not sufficient to show a benefit for either arm, with a risk difference of 4.0% (95% CI, −2.2% to 10.3%) and 6.1% (95% CI, −0.3% to 12.5%) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal). A supplemental analysis using the composite strategy (i.e., patients who discontinued treatment were classified as nonresponders) had similar results as the primary analysis. Similar results were reported for the subgroup of patients who received risankizumab 1,200 mg induction therapy in the INSPIRE trial (ITT1RN_B population).¹⁸

Discontinuation of Corticosteroid Use in Patients Taking Steroids at Baseline

In the COMMAND trial, the risk difference was 28.4% (95% CI, 14.0% to 42.8%) and 20.7% (95% CI, 4.9% to 36.6%) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal), favouring risankizumab.

Inflammatory Bowel Disease Questionnaire

In the INSPIRE trial, the mean difference in IBDQ scores was 18.3 points (95% CI, 13.38 to 23.25 points) compared to placebo, favouring risankizumab. In the COMMAND trial, this end point was not formally tested due to earlier failure of the statistical hierarchy (the comparison for HEMR was not significant). The mean difference was 17.5 points (95% CI, 8.01 to 27.06 points) and 15.2 points (95% CI, 5.18 to 25.31 points) for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal), favouring risankizumab. Similar results were reported for the subgroup of patients who received risankizumab 1,200 mg induction therapy in the INSPIRE trial (ITT1RN_B population).¹⁸

UC-Related Hospitalization

In the INSPIRE trial, the risk difference was −4.8% (95% CI, −7.3% to −2.2%) compared to placebo, favouring risankizumab. In the COMMAND trial, the evidence was insufficient to show a difference in the hospitalization rate. The incidence rate difference was −2.5 (95% CI, −5.4 to 0.4) per 100 patient-years and −1.8 (95% CI, −5.0 to 1.3) per 100 patient-years for risankizumab 180 mg SC and risankizumab 360 mg SC, respectively, compared to placebo (risankizumab withdrawal).

Harms

In both trials, the safety analysis sets included all randomized patients who received at least 1 dose of study drug, analyzed according to the treatment that they received. A summary of the overall TEAEs experienced by patients is provided in Table 18.

Table 18: Summary of Harms Results From the INSPIRE and COMMAND Trials

SAE = serious adverse event; SC = subcutaneous; UC = ulcerative colitis.

Note: Outcomes that occurred in at least 2 participants are reported.

Sources: INSPIRE Clinical Study Report,¹⁷ COMMAND Clinical Study Report.¹⁸ Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Treatment-Emergent Adverse Events

In the INSPIRE trial, the number of patients with at least 1 TEAE were similar between the risankizumab (42.4%) and placebo (49.4%) groups. The most common TEAE in the risankizumab group was COVID-19 (4.8%), and the most common TEAE in the placebo group was UC (10.2%). In the COMMAND trial, the number of patients with at least 1 TEAE were similar between the risankizumab 180 mg (72.5%), risankizumab 360 mg (70.8%), and placebo (76.5%) groups.

Serious Adverse Events

In the INSPIRE trial, the proportion of patients with at least 1 SAE was higher in the placebo group (10.2%) than the risankizumab group (2.3%). The most common serious TEAEs in the risankizumab group were anemia, UC, and pulmonary embolism (2 patients [0.3%] each). In the placebo group, the most common serious TEAEs were UC (4.9%), anemia (0.6%), and anal fistula (0.6%). In the COMMAND trial, the proportion of patients with at least 1 SAE was similar across groups (5.2% with risankizumab 180 mg, 5.1% with risankizumab 360 mg, and 8.2% with placebo). In the risankizumab 180 mg group, the only serious TEAE occurring in more than 1 patient was appendicitis (2 patients, 1.0%). In the risankizumab 360 mg group, no serious TEAE occurred in more than 1 patient. In the placebo (risankizumab withdrawal) group, the most common serious TEAE was UC (2.6%); no other serious TEAE occurred in more than 1 patient.

Withdrawals Due to AEs

In the INSPIRE trial, a higher proportion of patients in the placebo group (3.7%) discontinued treatment due to a TEAE than in the risankizumab group (0.6%) (Table 18). In the COMMAND trial, the proportions of patients who discontinued treatment due to a TEAE were highest in patients taking risankizumab 360 mg (2.6%) compared to risankizumab 180 mg (1.6%) and placebo (1.5%) groups (Table 18).

Deaths

One patient in the risankizumab group died from COVID-19 in the INSPIRE trial (Table 18). In the COMMAND trial, 1 patient (in the risankizumab 360 mg group) died during the trial from colon adenocarcinoma (Table 18).

Adverse Events of Special Interest

In the INSPIRE trial, the most frequently occurring AEOSIs in both arms were hypersensitivity (risankizumab: 3.8%; placebo: 2.2%) and hepatic events (risankizumab: 1.5%; placebo: 4.0%). (Table 18). In the COMMAND trial, the risankizumab 180 mg group had the highest proportion of patients with hypersensitivity (10.4%) (Table 18). The risankizumab 360 mg group had the highest proportion of patients with hepatic events (6.7%).

Critical Appraisal

Internal Validity

INSPIRE Trial

Randomization in the INSPIRE trial involved 2:1 computer-generated, block randomization via interactive response technology. This is an adequate form of randomization and allocation concealment. Participants were stratified for major confounding factors (e.g., corticosteroid use, baseline Mayo score, and prior AT exposure). The baseline characteristics of the randomized groups were balanced, reflecting a proper randomization procedure. The trial was double-blind and placebo-controlled, with a detailed protocol including what concomitant medications were allowed and what changes in dosing were allowed after randomization. This decreases the probability of any deviations from the intended interventions during the trial. Concomitant medications were balanced across the treatment groups and not expected to impact the efficacy results. The primary and secondary efficacy end points in INSPIRE were tested using a graphical multiple-testing procedure to ensure control of family-wise type I error at a significance level alpha = 0.05 (2-sided), which was deemed adequate.

The sponsor considered the occurrence of appropriate ICEs that impact the interpretation of the effect of interest within the trial estimands. Data for patients who prematurely discontinued continued to be collected (i.e., treatment policy). Starting or escalating corticosteroids due to UC and surgery related to UC were considered a nonresponse for binary end points and return to baseline for continuous end points. This approach is considered adequate by regulatory authorities when the therapeutic intent is to induce remission in the short term.^12,13

With regards to missing outcome data, on request by the review team the sponsor was unable to provide information about the occurrence of ICEs before study discontinuation nor the potential for sporadic missing data for reasons other than study discontinuation (e.g., missing visit). Therefore, it was not possible to perform a complete appraisal of the potential for bias due to missing data arising from reasons that are incompatible with the nonresponder imputation approach used by the sponsor for binary end points (and missing at random for continuous end points). There was a higher discontinuation rate in the placebo group (8.3%) than in the risankizumab group (2%), which can introduce attrition bias when nonresponder imputation is used. The reasons for study discontinuation (and possibly missing data) are incompatible with lack of efficacy. More than half of patients in the placebo group who withdrew did so because of lack of efficacy or AEs, suggesting that the assumption that withdrawals were due to nonresponse is acceptable. Sensitivity analyses were conducted for the primary end point, including a tipping-point analysis indicating that the statistical significance of the results would be maintained across a range of assumptions about the missing data. The outcome of UC-related hospitalization used an as-observed analysis; this approach considers that data are missing completely at random, which is unlikely and may introduce bias if there is an increased proportion of missing data (which could not be assessed).

To minimize detection bias, the trial used central endoscopic review with blinded assessors for objective outcomes, but it is important to note that this is only 1 component of the adapted Mayo score. For subjective outcomes, like patient-reported outcomes (e.g., other components of the adapted Mayo score, IBDQ) and safety outcomes, as the trial was double-blinded then it is unlikely that these results would be biased. Selective outcome and analysis reporting bias was marginalized by the publication of a prespecified protocol in ClinicalTrials.gov (NCT03398148). Overall, the INSPIRE trial is at low risk of bias for most end points, although uncertainty remains about potential attrition bias, which could not be fully assessed due to insufficient reporting.

The collection of information on harms included worsening UC as well as UC-related symptoms. Given that these are likely to be increased in the placebo group, the interpretation of the difference between groups is challenging. Further, it is implausible that there were significantly more SAEs reported in the placebo group than in the active intervention group, which would imply that the placebo effect is associated with SAEs.

With regards to the subgroup analyses, randomization in the INSPIRE trial was stratified by the presence of baseline corticosteroid use (yes, no), baseline adapted Mayo score (≤ 7, > 7), and a history of intolerance or inadequate AT response (0, 1, > 1 treatment). All other subgroup analyses may have broken randomization and therefore could have been affected by biases, including selection bias and confounding, due to unadjusted differences in baseline characteristics. The purpose of these analyses was to demonstrate consistency, and they cannot be used to draw credible conclusions about effect modification.

COMMAND Trial

The trial protocol was publicly available (NCT03398135) and the randomization and allocation concealment methods were similar to those in the INSPIRE trial, except that the trial used a 1:1:1 allocation ratio stratified by prior AT response and induction dose. The baseline characteristics suggested imbalances in some disease characteristics, although these did not appear to systematically favour any treatment group. Therefore, they are not expected to have an important impact on the results.

While the trial was double-blind, open-label rescue with risankizumab 1,200 mg IV was allowed after week 16. Unblinded therapy may have introduced performance bias in the collection of data related to harms. For efficacy end points, the need for rescue therapy resulted in an assumption that the patient was a nonresponder, which is appropriate and would not introduce bias. Concomitant medications were balanced across the treatment groups and not expected to impact the efficacy results. In the COMMAND trial, the testing began with testing the primary end point at the prespecified significance level of 0.025 (2-sided) for each risankizumab dose group compared to placebo. This was deemed an adequate approach.

The sponsor considered the occurrence of appropriate ICEs that impact the interpretation of the effect of interest within the trial estimands. Data for patients who prematurely discontinued continued to be collected (i.e., treatment policy). Starting or escalating corticosteroids related to UC, surgery related to UC, and rescue treatment were considered as nonresponse for binary end points and return to baseline for continuous end points. Given that the intent of COMMAND is to demonstrate maintenance of treatment efficacy in the long term, considering treatment discontinuations as nonresponse may be a preferred approach.¹³ Nonetheless, a supplemental analysis using the nonresponder imputation approach for treatment discontinuations showed similar results to the primary analysis on the primary and all secondary end points.

It was not possible to perform a comprehensive appraisal of the risk of bias due to missing outcome data as a result of incomplete reporting in the COMMAND study report. The concern for potential unreported missing data may be higher for the COMMAND trial compared to the INSPIRE trial as a result of the longer period of follow-up. There was a higher discontinuation rate in all groups compared to the INSPIRE trial, with 9.8% of patients in the placebo group discontinuing, and discontinuations in the risankizumab groups ranging from 7.8% to 12.4%. This can introduce attrition bias when nonresponder imputation is used and the reasons for study discontinuation are incompatible with lack of efficacy (if these patients’ data are considered missing). It is unclear how many patients were already imputed as nonresponders before study discontinuation (due to ICE, which appeared to be common; for example, 19% to 43% received rescue medication). As a result, the appraisal of potential for bias cannot be completed with certainty. On request from the review team, the sponsor was unable to provide this information. Sensitivity analyses were conducted for the primary end point, including a tipping-point analysis, which supported that the statistical significance of the results would be maintained across a range of assumptions about the missing data. The outcome of UC-related hospitalization used an as-observed analysis that considers data to be missing completely at random, which is unlikely to be reasonable and may introduce bias if there is an increased proportion of missing data (which could not be assessed).

In addition, the outcome “discontinuation of corticosteroid use at week 52 in patients taking steroids at baseline” was not assessed in the full cohort of randomized participants, and it is unclear whether the baseline characteristics in these individuals are similar between the groups. Having said that, the outcome “clinical remission with no corticosteroid use for 90 days” was assessed in the full cohort of randomized participants, and the results favoured risankizumab. Overall, bias is possible in the harms data due to open-label rescue treatment and in the corticosteroid discontinuation end point due to possible loss of randomization. The risk of bias due to missing outcome data could be increased because of attrition and lack of information to understand the proportion of data that may be missing. This concern was mitigated for the primary outcome via sensitivity analysis.

The collection of information on harms included worsening UC as well as UC-related symptoms. Given that these are likely to be increased in the placebo group, the interpretation of the difference between groups is challenging. Additionally, the comparison may not reflect a true placebo group due to the potential for risankizumab rescue therapy, which was received by 43% of patients in the placebo group.

In the COMMAND trial, randomization was stratified by history of inadequate AT response (yes, no), last risankizumab induction dose (IV 600 mg, 1,200 mg, or 1,800 mg), and clinical remission status (per local evaluation) at the last visit of the induction trial (yes, no). All other subgroup analyses may have broken randomization and therefore could have been affected by biases, including selection bias and confounding due to unadjusted differences in baseline characteristics. It is also important to note that the placebo group in the COMMAND trial received risankizumab during the induction trial (risankizumab withdrawal). As such, there could have been a carry-over effect due to an inadequate washout.

External Validity

The populations of both trials had a history of UC that was highly refractory to previous UC treatments, and the clinical experts consulted by the review team confirmed that patients in the newer UC trials are expected to have UC that is more refractory to treatment. In addition, the trials excluded patients with a history of IL-23 inhibitor use. With regard to co-interventions, in the INSPIRE trial, initiating or increasing the dose of comedications was prohibited, while in the COMMAND trial, patients undergoing corticosteroid therapy were required to taper by week 8. With regard to the comparator, no direct comparison to an active intervention was available (risankizumab was only compared to placebo). Furthermore, the main primary and secondary outcomes used the Mayo scoring system, which is not commonly used in clinical practice to inform decision-making. In the subgroup analyses, several subgroups showed less clinically significant results, including patients in North America. This may impact generalizability, although credible conclusions on effect modification cannot be drawn. These specific trial characteristics and results may impact generalizability and implementation in a real-world setting.

GRADE Summary of Findings and Certainty of the Evidence

The selection of outcomes for GRADE assessment was based on the sponsor’s Summary of Clinical Evidence, consultation with clinical experts, and input received from patient and clinician groups and public drug plans. The following list of outcomes was finalized in consultation with expert committee members:

INSPIRE trial:

• Clinical remission

• Clinical response

• Endoscopic improvement

• HEMI

• HEMR

• IBDQ

• SAEs

COMMAND trial:

• Clinical remission

• Clinical response

• Endoscopic improvement

• HEMI

• HEMR

• Discontinuation of corticosteroid use in patients taking steroids at baseline

• IBDQ

• SAEs

Methods for Assessing the Certainty of the Evidence

For pivotal studies and RCTs identified in the sponsor’s systematic review, GRADE was used to assess the certainty of the evidence for outcomes considered most relevant to inform expert committee deliberations, and a final certainty rating was determined as outlined by the GRADE Working Group:^106,107

• High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.

• Moderate certainty: We are moderately confident in the effect estimate — The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. We use the word “likely” for evidence of moderate certainty (e.g., “X intervention likely results in Y outcome”).

• Low certainty: Our confidence in the effect estimate is limited — The true effect may be substantially different from the estimate of the effect. We use the word “may” for evidence of low certainty (e.g., “X intervention may result in Y outcome”).

• Very low certainty: We have very little confidence in the effect estimate — The true effect is likely to be substantially different from the estimate of effect. We describe evidence of very low certainty as “very uncertain.”

• For RCTs: Following the GRADE approach, evidence from RCTs started as high-certainty evidence and could be rated down for concerns related to study limitations (which refers to internal validity or risk of bias), inconsistency across studies, indirectness, imprecision of effects, and publication bias.

When possible, certainty was rated in the context of the presence of an important (nontrivial) treatment effect; if this was not possible, certainty was rated in the context of the presence of any treatment effect (i.e., the clinical importance is unclear). In all cases, the target of the certainty of evidence assessment was based on the point estimate and where it was located relative to the threshold for a clinically important effect (when a threshold was available) or to the null. Literature-based effect size thresholds were available for clinical response and remission, endoscopic improvement, and SAEs.¹⁴ A literature-based MID was available for IBDQ, with a change of more than 30 points considered to be associated with clinical benefits and an improvement of 15 points or greater above placebo required among patients with IBD, including those with UC (Table 7). For discontinuation of corticosteroid use in patients taking steroids at baseline, between-group differences were identified by the clinical expert consulted by the review team as a threshold of clinical importance for these outcomes. The certainty for all other outcomes was based on the presence or absence in any effect, as a threshold of clinical importance could not be established.

Results of GRADE Assessments

Risankizumab Versus Placebo

Table 2 presents the GRADE summary of findings for risankizumab versus placebo for patients with UC (INSPIRE). Table 3 presents the GRADE summary of findings for risankizumab 180 mg SC versus placebo for patients with UC (COMMAND). Table 4 presents the GRADE summary of findings for risankizumab 360 mg SC versus placebo for patients with UC (COMMAND).

Long-Term Extension Studies

Contents within this section have been informed by materials submitted by the sponsor. The following have been summarized and validated by the review team.

The sponsor submitted an abstract¹⁹ for the COMMAND open-label extension study. Because of the lack of access to complete information of methods and results, the summary of this study is not reported in this section.

Indirect Evidence

Contents within this section have been informed by materials submitted by the sponsor. The following have been summarized and validated by the review team.

Objectives for the Summary of Indirect Evidence

No direct comparison exists between risankizumab and other publicly reimbursed ATs or those recommended for reimbursement in Canada for moderately to severely active UC. In the absence of head-to-head RCTs, an NMA was used to compare the efficacy and safety of risankizumab to other publicly reimbursed ATs or those recommended for reimbursement in Canada for moderately to severely active UC: adalimumab, infliximab, golimumab, vedolizumab, tofacitinib, ustekinumab, ozanimod, upadacitinib, mirikizumab, and etrasimod.

Specifically, the objectives were the following:

• to determine the comparative efficacy of risankizumab versus relevant AT comparators separately, as induction and maintenance therapies for adults with moderately to severely active UC, and also separately for patients with non–AT-IR and AT-IR, per the risankizumab approved indication

• to determine the comparative safety of risankizumab versus relevant AT comparators separately, as induction and maintenance therapies for adults with moderately to severely active UC, in the combined sample of overall safety populations.

Description of Indirect Comparisons

Study Selection Methods

A systematic review was conducted as per standardized guidance from Cochrane, the Centre for Reviews and Dissemination, and the National Institute for Health and Care Excellence (NICE).^108,109 A search for English-language RCTs was conducted in MEDLINE, Embase, and Cochrane Library (including Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Database of Abstracts of Reviews of Effects, and health technology assessment database) reporting the clinical efficacy or safety of ATs in adult patients with moderately to severely active UC. Searches were from database inception to June 27, 2023. Search strategies included search terms (free text) and controlled vocabulary for UC and the interventions. Search terms from previously published systematic reviews and NICE guidance were used to comprehensively capture RCT search terms.¹¹⁰

In addition to the bibliographic database searches, grey literature searches included relevant conference proceedings (Digestive Disease Week, Crohn’s & Colitis Congress, ISPOR–The Professional Society for Health Economics and Outcomes Research, United European Gastroenterology Week, American College of Gastroenterology, ECCO) for the last 5 years and searching trial registries keyword searches of the annual proceedings of scientific meetings and clinical trial registers (ClinicalTrials.gov, Health Canada’s Clinical Trial Database, EU Clinical Trials Register, International Clinical Trials Registry Platform). Last, the following websites of the regulatory and health technology assessment authorities in UK (England [NICE] and Scotland [Scottish Medicines Consortium]), Canada (CDA-AMC), France (Haute Autorité de Santé), Australia (Pharmaceutical Benefits Advisory Committee), and Germany (Federal Joint Committee [Gemeinsamer Bundesausschuss]) were searched for additional studies.

Finally, the bibliographies of systematic reviews and meta-analyses identified through database searches were used to identify key studies. Bibliographies from selected key studies were also reviewed to ensure literature saturation.

Title and abstract screening were conducted by 2 blinded, independent researchers in parallel using the predefined population, intervention, comparators, outcomes, and study design (PICOS) criteria presented in Table 19. The study selection was performed by 2 independent reviewers and based on a 2-step approach according to a predefined set of eligibility criteria: screening of abstracts and titles, followed by in-depth review of full-text articles. Each reviewer independently assessed the studies for inclusion according to the criteria, structured using the PICOS framework. Discrepancies between researchers were resolved by a third independent researcher. Data from included studies were extracted into a predefined Excel template by a single analyst and all results were 100% quality checked.

The risk of bias of individual full-text studies included in the systematic review was assessed using the Cochrane Collaboration risk-of-bias tool for RCTs.¹¹¹ The risk-of-bias assessment was performed independently by 2 reviewers, and any discrepancies were resolved through discussions to reach a consensus.

Table 19: Study Selection Criteria and Methods for ITCs Submitted by the Sponsor

AE = adverse event; AT = advanced therapy; EU = European Union; RCT = randomized controlled trial; S1P = sphingosine 1-phosphate.

^aWhile filgotinib was included in the global systematic review, it was excluded from the feasibility assessment, as it is not approved in Canada.

Source: Details included in the table are from the sponsor’s Summary of Clinical Evidence.

ITC Design

ITC Analysis Methods

Feasibility Assessment

Network plots were created to demonstrate the network connectivity of all included induction and maintenance RCTs. Then, relevant study and patient characteristics (e.g., age, sex, weight, duration and extent of disease, baseline Mayo scores, and concurrent medications for UC) were reviewed across the included induction and maintenance RCTs to get a sense of their comparability and identify potential sources of cross-RCT heterogeneity. The potential effect modifiers were identified based on clinical opinion. The mean baseline (placebo) effects across the included induction and maintenance RCTs were also assessed by performing a meta-analysis of all placebo arms included in each network.

Model Specifications

For each feasible network, NMAs were conducted in a generalized linear model framework using Bayesian Markov chain Monte Carlo simulations and 3 chains with 100,000 runs each. The burn-in period was set to half of the convergence sequence, until convergence (minimum set size of 25,000).^112-114 Convergence was assessed with the Brooks-Gelman-Rubin method using the potential scale reduction factor.

Per NICE Decision Support Unit Technical Support Document 2, all binary response outcomes were modelled with a binomial likelihood and logit link function.^112,113 All posterior distributions for ORs were summarized by their medians and 95% CrIs.

By default, RCT-specific baselines were modelled as independent, such that an unrelated model parameter was specified for each one. However, in networks with 1 or more placebo arm(s) having a value of 0 (i.e., no events), an exchangeable baseline assumption with a half-normal (0, 0.32²) prior for heterogeneity was used to aid parameter estimation and numerical stability or convergence.¹¹³

For all networks, both fixed-effects and random-effects models were tested.

Priors

Vague or flat prior distributions were given to the parameters to be estimated by default.^112,113 For parameters assumed to be specified on a continuous scale, namely, the relative treatment effects (d), RCT-specific baselines (mu), and baseline adjustment regression term (beta) (for models with baseline risk adjustment), a normal (0, 100²) prior distribution was used. For the between-study standard deviation (SD) (for random-effects models), a uniform (0, 2) prior distribution was used. For the between-study heterogeneity SD (or sigma) associated with the random-effects model, when most (≥ 50%) interventions in a network are informed by a single study, a half-normal (0, 0.32²) prior distribution, which gives a low prior weight to unfeasibly large SDs on the logit scale, was used.^113,115

Baseline Risk Adjustment

For each fixed-effects and random-effects model tested, a baseline risk-adjusted version was also tested that adjusted for differences in mean placebo effects across studies using the code provided in NICE Decision Support Unit Technical Support Document 3.¹¹⁶ A common regression term (beta) was assumed for all adjustments (i.e., the relationship between the placebo response and the active treatment response was assumed not to depend on treatment). The model with the baseline risk covariate was selected if, because of its inclusion, the median posterior SD (for random-effects models) decreased and the 95% CrI of the regression term beta excluded 0.¹¹³

Model Fit

Models’ global fits were assessed and compared using their overall posterior mean residual deviance and deviance information criteria.^112,113 All else being equal between fixed-effects and random-effects models, the random-effects model was selected to account for the expected heterogeneity in outcomes, study design, and study populations across included RCTs.

Heterogeneity

The relevant study and patient characteristics were compared (age, gender, weight, duration of disease, extent of disease, baseline Full Mayo score and Adapted Mayo score, and concurrent medications for UC). Any disparities were evaluated in sensitivity analyses.

The statistical assessment of homogeneity was based on the I² statistic, derived from a direct head-to-head meta-analysis of the treatment comparisons in each network that were evaluated in more than 1 study.¹¹⁷

All comparisons with an I² higher than 50% were investigated. Possible sources of heterogeneity were reviewed and, if applicable, studies introducing heterogeneity to the model were excluded via a sensitivity analysis where possible.

Consistency

To verify the consistency of the direct and indirect comparisons, when available, a node-splitting approach was considered. Whenever applicable, a sensitivity analysis was performed based on the inconsistency assessment to determine the impact of treatment comparisons with significant inconsistency between the direct and indirect evidence.

A list of all sensitivity analyses is provided in Table 20.

Included Treatments and Nodes

The interventions of interest for the NMA included ATs publicly reimbursed or recommended for reimbursement in Canada for moderately to severely active UC (adalimumab, golimumab, infliximab, vedolizumab, ustekinumab, tofacitinib, upadacitinib, ozanimod, mirikizumab, etrasimod), as well as risankizumab (Table 20). Of note, although filgotinib was included in the systematic review, it was excluded from the NMA results, as it is not approved in Canada. As previously mentioned, although not publicly reimbursed yet, etrasimod has been recommended by CDA-AMC and is currently subject to negotiations with the pCPA.⁴⁸ Only RCTs identified by the clinical systematic review that reported useful relative efficacy and safety estimates for the licensed dose(s) were included. Where the drug licence allows for dose escalation during the maintenance phase, both the low and high doses were included if they had been assessed in the RCTs. In RCTs that assessed both licensed and unlicensed doses, arms using unlicensed doses were excluded from the networks. Of note, different doses of treatment were treated as separate treatment nodes in the NMA networks. Last, placebo was included as a common comparator in the NMA.

Following a feasibility assessment, the NMA was deemed feasible for the following efficacy and safety outcomes:

• Efficacy (both induction and maintenance phases, both non–AT-IR and AT-IR populations):

  • Clinical remission

  • Clinical response

  • Endoscopic improvement

• Safety (both induction and maintenance phases, overall population):

  • % of AEs

  • % of SAEs

  • % of serious infections

TT Versus Rerandomized Responder Design

Two maintenance study designs were employed across the UC RCTs: TT and rerandomized responder (RR). In TT studies, patients were randomized to treatment or placebo at baseline, and outcomes were measured at the end of an induction phase and again at the end of a maintenance phase. In RR studies (e.g., risankizumab studies), patients were randomized to induction treatment or placebo at baseline, with outcomes measured at the end of the induction phase; induction responders were then rerandomized to maintenance treatment or placebo, with outcomes measured at the end of the maintenance phase strictly among induction responders.

Because these 2 study designs are incompatible in a standard NMA, the sponsor re-calculated data from the TT RCTs to mimic a RR design to conduct the NMA.

Table 20: Indirect Comparison Analysis Methods

AE = adverse event; AT = advanced therapy; AT-IR = inadequate response or intolerance to advanced therapy; DIC = deviance information criteria; NMA = network meta-analysis; PSRF = potential scale reduction factor; RCT = randomized controlled trial; SAE = serious adverse event; SD = standard deviation; vs. = versus.

Note: Not all the planned sensitivity analyses were possible due to lack of data.

Source: Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Results of the ITC

Summary of Included Studies

Thirty-nine studies (with 118 reports) with potential for NMA inclusion were selected for data extraction. Per the NMA PICOS criteria, all 21 induction RCTs randomized patients to relevant induction treatment(s) versus placebo for 6 to 12 weeks in a double-blind manner. Likewise, all 16 maintenance RCTs randomized patients to relevant maintenance treatment(s) versus placebo (except for SERENE-C¹¹⁸) for 40 to 54 weeks in a double-blind manner. A total of 28 original RCTs reported by 77 records were included in the NMA, 21 in the induction phase and 16 in the maintenance phase. Table 21 presents an overview of the included studies.

Induction

A network plot of the included induction RCTs is shown in Figure 1. The network is star-shaped, in that all treatments are anchored on placebo as the common comparator. The network plots for individual outcomes differed slightly, as not all outcomes were reported by all the included trials.

Figure 1: Induction Network Plot

ADA = adalimumab; ETR = etrasimod; GOL = golimumab; INF = infliximab; MIR = mirikizumab; OZA = ozanimod; PBO = placebo; RIS = risankizumab; TOF = tofacitinib; UPA = upadacitinib; UST = ustekinumab; VED = vedolizumab.

Notes: The nodes are weighted based on the number of participants. The lines are weighted based on the number of studies (numbers) with a direct comparison between treatments.

Source: Details included in the table are from the sponsor’s Summary of Clinical Evidence.

The key findings from the feasibility assessment, along with the list of resulting sensitivity analyses, were as follows:

• NMA connectivity was feasible for both non–AT-IR and AT-IR populations in induction studies.

• A sensitivity analysis excluding predominantly Asian RCTs was planned to assess the impact on the clinical response of risankizumab relative to other treatments.

• Due to the limited number of studies, no relevant sensitivity analyses based on outcome definitions could be performed. However, the sponsor considered the outcome definitions across studies to be sufficiently similar for inclusion in the NMA.

• Table 22 outlines the key findings of the baseline characteristics’ heterogeneity and the approach taken to address them. The sponsor considered that baseline patient characteristics were comparable across the induction RCTs. However, some heterogeneity was observed.

• No other sensitivity analyses were deemed necessary by the sponsor based on the feasibility assessment process.

Additionally, there are sources of heterogeneity that extend beyond the patient baseline characteristics (e.g., differences in the studies’ inclusion or exclusion criteria, outcome definitions, and time points). For example, the included studies used varying definitions of clinical remission and response (e.g., central endoscopy versus local investigator readings), durations of follow-up for outcome assessment (e.g., 6 to 12 weeks), differing definition of AT-IR, number and type of patients with AT were previously exposed differed (as the commercially available options changed over time), differences in how disease severity inclusion criteria was defined across the trials, differences in disease duration (e.g., ranging from 3.7 to 9.1 years), variation in extent of disease (e.g., from 32 to 80% with extensive colitis or pancolitis), heterogeneity in concurrent corticosteroid use (particularly for the non–AT-IR population), and wide variation in placebo risk.

Table 21: Overview of the RCTs Included In the Clinical Systematic Review

ADA = adalimumab; AMS = adapted Mayo score; EMS = endoscopic Mayo subscore; ETR = etrasimod; FMS = full Mayo score; GOL = golimumab; HIR = higher induction dosing regimen; INF = infliximab; MIR = mirikizumab; NR = not reported; OZA = ozanimod; RBS = rectal bleeding subscore; RCT = randomized clinical trial; RIS = risankizumab; RR = rerandomized responder; TDM = therapeutic drug monitoring; TOF = tofacitinib; TT = treat through; UC = ulcerative colitis; UPA = upadacitinib; UST = ustekinumab; VED = vedolizumab; X = applicable.

^aAM response = decrease in AMS ≥ 2 points and ≥ 30% from baseline, and a decrease in RBS ≥ 1 or an absolute RBS ≤ 1.

^bFM response = decrease in FMS ≥ 3 points and ≥ 30% from baseline, and a decrease in RBS ≥ 1 or an absolute RBS ≤ 1.

Source: Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Table 22: Key Findings from the Baseline Patient Characteristics Assessment (Induction Population)

AMS = adapted Mayo score; AT-IR = inadequate response or intolerance to advanced therapy ETR = etrasimod; INF = infliximab; FMS = full Mayo score; MIR = mirikizumab; NA = not applicable; OZA = ozanimod; RCT = randomized clinical trial; SA = sensitivity analysis; UC = ulcerative colitis; UPA = upadacitinib.

Source: Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Maintenance

A network plot of the included maintenance RCTs is shown in Figure 2. Like the induction network, the maintenance network is also star-shaped. All maintenance treatments are anchored on placebo as the common comparator.

Figure 2: Maintenance Network Plot

ADA = adalimumab; ETR = etrasimod; GOL = golimumab; INF = infliximab; MIR = mirikizumab; OZA = ozanimod; PBO = placebo; Q#W = every # weeks; RIS = risankizumab; SC = subcutaneous; TOF = tofacitinib; UPA = upadacitinib; UST = ustekinumab; VED = vedolizumab.

Notes: The nodes are weighted based on the number of participants. The lines are weighted based on the number of studies (numbers) with a direct comparison between treatments.

Source: Details included in the table are from the sponsor’s Summary of Clinical Evidence.

The key findings from the feasibility assessment, along with the list of resulting sensitivity analyses, were as follows:

• NMA connectivity was feasible for both non–AT-IR and AT-IR populations in maintenance phase.

• A sensitivity analysis excluding RCTs in predominantly Asian populations was planned to assess the impact on the clinical response of risankizumab relative to other treatments.

• Two sensitivity analyses were planned for clinical response and endoscopic improvement due to a stricter time point requirement (i.e., excluding PURSUIT-J for clinical response and PURSUIT-M for endoscopic improvement). Due to the limited number of studies, no other relevant sensitivity analyses based on outcome definitions could be performed. However, the outcome definitions were considered sufficiently similar by the sponsor to being included in the NMA.

• A sensitivity analysis was planned based on the study design, using TT data instead of RR for risankizumab for efficacy outcomes (true placebo analysis).

• Table 23 outlines the key findings about the baseline characteristics’ heterogeneity and the approach taken to address them. In summary, the sponsor considered that baseline patient characteristics were comparable across the maintenance RCTs, which supports the feasibility of the NMA. However, some heterogeneity was observed.

• No other sensitivity analyses were deemed necessary by the sponsor based on the feasibility assessment process.

Additionally, there are sources of heterogeneity that extend beyond the patient baseline characteristics (e.g., differences in the studies’ inclusion and exclusion criteria, outcome definitions, and time points). For example, the included studies used varying definitions of clinical remission and response (e.g., central endoscopy versus local investigator readings), durations of follow-up for outcome assessment (e.g., 40 to 54 weeks), differing definition of AT-IR, number and type of patients with AT were previously exposed differed (as the commercially available options changed over time), differences in how disease severity inclusion criteria was defined across the trials, differences in disease duration (e.g., ranging from 3.7 to 9.1 years), variation in extent of disease (e.g., from 32% to 80% with extensive colitis or pancolitis), heterogeneity in concurrent corticosteroid use (particularly for the non–AT-IR population), and wide variation in placebo risk. It is unclear how the effect of these variations was mitigated or could have biased the analyses.

Table 23: Key Findings from the Baseline Patient Characteristics Assessment (Maintenance Population)

ADA = adalimumab; AMS = adapted Mayo score; AT-IR = inadequate response or intolerance to advanced therapy; ETR = etrasimod; FMS = full Mayo score; MIR = mirikizumab; NA = not applicable; OZA = ozanimod; RCT = randomized clinical trial; SA = sensitivity analysis; UC = ulcerative colitis; UPA = upadacitinib; VED = vedolizumab.

Source: Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Induction

Efficacy

Risankizumab 1,200 mg was favoured over adalimumab ███ █████ ███ ███ ████ ██ █████ and mirikizumab ███ █████ ███ ███ ████ ██ ████) for clinical response and to adalimumab (██ █████ ███ ███ ████ ██ ████), golimumab (██ █████ ███ ███ ████ ██ ██████ mirikizumab ███ █████ ███ ███ ████ ██ ██████ tofacitinib (██ █████ ███ ███ ████ ██ ████), and ustekinumab ███ █████ ███ ███ ████ ██ █████ for achieving endoscopic improvement in the non-AT population during induction (Table 24). Risankizumab 1,200 mg was less favoured compared to upadacitinib for clinical remission (██ █████ ███ ███ ████ ██ ████), clinical response (██ █████ ███ ███ ████ ██ █████ and endoscopic improvement (██ █████ ███ ███ ████ ██ ████) in the AT-IR population during induction (Table 24). There was no evidence of a significant difference between risankizumab 1,200 mg and the other interventions in the efficacy NMAs during the induction phase (Table 24).

Table 24: League Table of ORs for Clinical Remission, Clinical Response, Endoscopic Improvement in Non–AT-IR and AT-IR Induction NMAs

ADA = adalimumab; AT-IR = inadequate response or intolerance to advanced therapyETR = etrasimod; GOL = golimumab; INF = infliximab; MIR = mirikizumab; OR = odds ratio; OZA = ozanimod; RIS = risankizumab; TOF = tofacitinib; UPA = upadacitinib; UST = ustekinumab; VED = vedolizumab.

Note: All NMAs utilized a random-effects model except clinical response in the non–AT-IR induction NMA, which utilized a fixed-effects model. Asterisks (**) indicate significance (95% CrI does not cross 1). Results interpretation: OR > 1 favours risankizumab in the column.

Source: Details included in the table are from the sponsor’s Summary of Clinical Evidence.

Safety

Risankizumab 1,200 mg was favoured over adalimumab (██ █████ ███ ███ ████ ██ ██████ █████████ ███ █████ ███ ███ █████ ██████ ████████ ███ █████ ███ ███ ████ ██ ██████ ███ ███████████ ███ █████ ███ ███ ████ ██ ████) with regard to SAEs during the induction phase (Table 25). Risankizumab 1,200 mg was less favoured than etrasimod ███ █████████ ███ ███ █████ ██ █████████ with regard to serious infections during the induction phase (Table 25). There was no evidence of a difference between risankizumab 1,200 mg and the other interventions in the safety NMAs during the induction phase (Table 25).

Table 25: League Table of ORs for AEs, SAEs, and Serious Infections in Induction NMAs