Real-world effectiveness and predictors of need for extended induction of mirikizumab therapy in ulcerative colitis

Abstract

Background

Mirikizumab (MIR), a selective interleukin-23p19 inhibitor, has demonstrated efficacy in randomized trials for moderate-to-severe ulcerative colitis (UC). Real-world data remain limited. Here, we evaluated the effectiveness, durability, and safety of MIR in routine clinical practice.

Methods

We conducted a retrospective, multicentre cohort study of adults with UC initiating MIR at two tertiary centres in Canada. Clinical, biomarker, endoscopic, and safety outcomes were assessed at weeks 12, 28, and 52. The primary outcome was corticosteroid-free (CSF) clinical remission at week 12. Secondary outcomes included CSF clinical remission at weeks 28 and 52, CSF clinical response, mucosal healing, treatment persistence, extended induction, and serious adverse events. Predictors of extended induction were evaluated using logistic regression.

Results

Eighty-two patients were included; 94% were previously exposed to at least one advanced therapy. CSF clinical remission was achieved in 24% of patients at week 12, 35% at weeks 28, and 33% at week 52, with corresponding CSF clinical response rates of 45%, 53%, and 40%. Moreover, 40% of the cohort received extended induction. In the extended induction subgroup, CSF clinical remission was observed in 19% at week 28 and 22% at week 52, with CSF clinical response rates observed in 37% and 30%, respectively. Early CSF clinical response at week 12 was associated with treatment persistence. The incidence of serious adverse events was low, with no new safety signals.

Conclusions

In this multicentre real-world Canadian cohort of predominantly treatment-experienced patients with UC, MIR demonstrated clinically relevant effectiveness, acceptable durability, and a favourable safety profile over 1 year.

Introduction

Ulcerative colitis (UC) is a chronic inflammatory disorder of the colon characterized by diarrhoea, rectal bleeding, and urgency.¹ Although the therapeutic landscape has expanded substantially with the introduction of biologics and small molecules, treatment failure remains common. Both primary nonresponse and secondary loss of response are common, and 10%-20% of patients ultimately require colectomy.^2,3 There remains an unmet need for novel therapeutic classes as patients frequently cycle through available treatments. Inhibition of the interleukin-23 (IL-23) ­pathway has garnered increased attention over the past two years following the licencing of three selective IL-23p19 inhibitors for the treatment of moderate-to-severely active UC.^4–6

Mirikizumab (MIR) was the first IL-23p19 inhibitor to be licenced for UC, following robust efficacy and safety data from the pivotal randomized controlled trials (RCTs) LUCENT 1 and 2.⁵ However, RCTs enrol highly selected populations, with fewer than 20% of patients with inflammatory bowel disease (IBD) typically meeting trial eligibility criteria.⁷ Therefore, it remains important to confirm the validity of these trials by assessing real-world outcomes of novel therapies in patient populations that are typically more complex, frail, and comorbid.⁷ Accordingly, we sought to evaluate the real-world effectiveness, durability and safety of MIR in patients with UC.

Methods

Study design and population

This was a retrospective, multicentre cohort study conducted at Mount Sinai Hospital, Toronto and The IBD Centre of BC, Vancouver, Canada. Adult patients with a confirmed diagnosis of UC who initiated treatment with MIR between January 2023 and August 2025 were identified through institutional databases. A diagnosis of UC was established based on standard clinical, endoscopic, radiological, and histological criteria as documented in the medical records.⁸ Patients were required to have at least 12 weeks follow-up data available following the initiation of MIR. Exclusion criteria included inadequate follow-up period, Crohn’s disease, IBD-unclassified, pouchitis, pregnancy at initiation, or treatment received as part of a clinical trial or as part of a combination advanced therapy regimen.

Data collection

Electronic patient records were reviewed to gather baseline demographic data and disease characteristics. Available C-reactive protein (CRP), faecal calprotectin (FCP), disease activity scores assessed by the Mayo score, endoscopic score (endoscopic mayo subscore), steroid use, hospitalizations, and adverse events were recorded.⁹ Baseline data were collected within 6 months of treatment initiation. Follow-up data were collected at weeks 12, 28, and 52 ± 8 weeks for each time point.

Endpoints

The primary endpoint was the rate of corticosteroid-free (CSF) clinical remission at week 12. Secondary endpoints included rates of CSF clinical remission at weeks 28 and 52; rates of CSF clinical response and mucosal healing at weeks 12, 28, and 52; and treatment persistence. Additional secondary outcomes included changes in CRP, FCP, and Mayo scores (overall, partial, and endoscopic), identification of factors associated with the need for extended induction, and the incidence of serious adverse events throughout therapy. All endpoints were evaluated using the closest clinical or endoscopic assessment within ±8 weeks of each prespecified time point.

Definitions

Baseline disease severity was determined using the closest assessment within 6 months of MIR initiation and defined hierarchically depending on data availability using overall Mayo score, partial Mayo score, or physician global assessment.^10,11 Advanced therapies were defined as biological or small molecule therapies. Clinical remission at weeks 12, 28, and 52 was defined hierarchically depending on data availability as: an overall Mayo score ≤2 with no subscore >1; or a partial Mayo score ≤2 with no subscore >1; or documented clinician impression of “in remission” or “normal” in the absence of oral corticosteroid use at time of clinical assessment. Clinical response was assessed at weeks 12, 28, and 52 and defined, in the absence of corticosteroids, as an overall Mayo score decrease of ≥3 points and ≥30% from baseline with either a ≥ 1-point improvement in rectal bleeding or a rectal bleeding score of 0-1; if unavailable, a partial Mayo score decrease of ≥2 points and ≥30% from baseline with the same rectal bleeding criteria; if unavailable, a documented complete or partial clinical response in the medical notes. Patients discontinuing MIR for lack of effectiveness or adverse events were classified as nonresponders at subsequent assessments. Mucosal healing was defined as a Mayo endoscopic score of 0 or 1 or, if unavailable, documentation of inactive disease (no inflammation, ulcers, erosions, or active pathological changes). Standard induction dosing was defined as receiving MIR 300 mg intravenously at weeks 0, 4, and 8, then transitioning to receive MIR 200 mg every 4 weeks subcutaneously. Extended induction dosing was defined as receiving a further 300 mg MIR by intravenous infusion at weeks 12, 16, and 20, then transitioning to receive MIR 200 mg every 4 weeks subcutaneously. Safety outcomes included all serious adverse events (SAEs) or serious infections (SIs) from treatment start date until treatment discontinuation or last follow-up, defined as events that were life-threatening, resulted in hospitalization, caused disability or incapacity, or were considered important medical events by the treating clinician.

Statistical analysis

Baseline characteristics were summarized descriptively. Continuous variables are reported as median with IQR and categorical variables as counts and percentages. Between-group comparisons used Wilcoxon rank-sum tests for continuous variables and Fisher’s exact tests for categorical variables. Changes in clinical, biochemical, and endoscopic disease activity measures over time were assessed using paired Wilcoxon signed-rank tests among patients with both baseline and follow-up measurements available. Predictors of extended induction were evaluated using logistic regression; results are reported as odds ratios (ORs) with 95% CIs, and multicollinearity was assessed using variance inflation factors. Treatment persistence was analyzed using Kaplan–Meier methods and compared using the log-rank test; Cox models estimated hazard ratios (HRs) for discontinuation. Adverse events were summarized descriptively and as exposure-adjusted incidence rates per 100 patient-years with 95% CIs. Two-sided P < .05 was considered significant. Analyses were performed in R (version 4.4.3).

Ethics

This study was approved by the Mount Sinai Hospital Research Ethics Board and Providence Healthcare Research Ethics Board. It was conducted in accordance with institutional guidelines.

Results

Baseline characteristics

A total of 134 MIR-exposed patients were identified in our institutional databases. After excluding patients who did not meet the inclusion criteria, 82 patients were included in the study. Baseline characteristics are summarized in Table 1. The cohort included 42 males (51.2%) with a median age of 37 years (IQR, 28-53). The median disease duration was 8 years (IQR, 3-15.25). Most patients had pancolitis (n = 57; 69.5%), with all but one of the remaining patients having left-sided disease (n = 24; 29%). Individuals had been exposed to a median of 2 (IQR, 1-3) prior advanced therapy classes. Only 5 patients (6.1%) were advanced therapy naïve. Most patients had moderate disease at baseline (n = 38; 46.3%), with one third of patients having severe disease (n = 25; 30.5%). Only 3 patients met the criteria for CSF remission at baseline. The median endoscopic Mayo subscore was 2 (IQR, 2-3). Also, 43.9% (n = 36) of patients were on oral steroids at initiation of therapy.

Table 1.

Baseline characteristics.

Characteristic	Overall, N = 82a
Sex
Male	42 (51.2%)
Female	40 (48.8%)
Age at initiation (years)	37 [28-53]
Disease duration (years)	8 [3-15.25]
Weight at initiation (kg)	69 [62-82]
Smoking
Never	61 (74.4%)
Ex-smoker	13 (15.9%)
Current	0 (0%)
Unknown	8 (9.8%)
UC disease extent
Proctitis	1 (1.2%)
Left-sided	24 (29.2%)
Pancolitis	57 (69.5%)
Previous UC-related hospitalization
Yes—within the year prior to MIR initiation	22 (26.8%)
Yes—over 1 year prior to start of MIR	25 (30.5%)
Disease severity at baseline
Quiescent	5 (6.1%)
Mild	14 (17.1%)
Moderate	38 (46.3%)
Severe	25 (30.5%)
Overall Mayo at baseline	8.00 [6-10]
Partial Mayo at baseline	6.00 [4-7]
Endoscopic Mayo at baselineb	2 [2-3]
Endoscopic Mayo 1	5 (6.1)
Endoscopic Mayo 2	40 (48.8)
Endoscopic Mayo 3	28 (34.1)
Segments inflamedb
Proctitis	6 (7.3%)
Left-sided	36 (43.9%)
Pancolitis	33 (40.2%)
Baseline biochemistry
CRP (mg/L)	4 [1-10]
Faecal calprotectin (µg/g)	799 [560-1700]
Albumin (g/L)	43 [38.5-44]
Concurrent therapies
Oral steroids at initiation	36 (43.9%)
5-ASA	43 (52.4%)
Thiopurine	3 (3.6%)
Number of previous advanced therapy classes
0	5 (6.1%)
1	25 (30.5%)
2	28 (34.1%)
≥3	24 (29.2%)
Previous therapies
S1P modulator	1 (1.2%)
JAK inhibitor	22 (26.8%)
Ustekinumab	15 (18.3%)
Vedolizumab	66 (80.5%)
TNF inhibitors	49 (59.8%)
Thiopurine	24 (29.3%)
5-ASA	38 (45.8%)
Extra-intestinal manifestations of IBD	14 (17.1%)

Advanced therapy classes included biologic and small-molecule therapies previously received prior to treatment initiation. Abbreviations: CRP = C-reactive protein; IBD = inflammatory bowel disease; JAK = Janus kinase; S1P = sphingosine-1-phosphate; TNF = tumour necrosis factor; UC = ulcerative colitis; 5-ASA = 5-aminosalicylic acid.

^a n (%); Median [Q1, Q3].

^bBaseline endoscopy data available for 73 patients.

Clinical treatment outcomes

At week 12, 24% (18/74) achieved CSF clinical remission, and 45% (33/74) achieved CSF clinical response. Eight individuals had no week 12 follow-up data due to an absence of clinical review at this time point. These individuals were included in the study as follow-up data were available for subsequent timepoints. At week 28, 35% (23/66) achieved CSF clinical remission and 53% (35/66) achieved CSF clinical response. At week 52, 33% (17/52) achieved CSF clinical remission and 40% (21/52) met criteria for CSF clinical response (Figure 1, Table S1). A sensitivity analysis restricted to assessments performed within ±2 weeks of the prespecified timepoint yielded results that were highly consistent with the main analysis for the week 12 endpoint (primary outcome) (Tables S2 and S3).

Figure 1.

Corticosteroid-free clinical response, corticosteroid-free clinical remission, and mucosal healing rates of mirikizumab at weeks 12, 28, and 52. Bars represent the proportion of patients achieving each outcome at the specified timepoint. Patients who discontinued therapy due to ineffectiveness or an adverse event were considered non-responders for all subsequent clinical outcomes. Mucosal healing was defined as an endoscopic Mayo subscore of 0-1.

Among the 15 patients with prior ustekinumab exposure, remission, or response rates did not differ compared with those without prior exposure at weeks 12, 28, or 52 (Figure S1). Similarly, no differences in treatment outcomes at the same timepoints were observed in the 49 individuals with prior anti-TNF-exposure compared to those without (Figure S2).

Extended induction

Overall, 40% (33/82) of patients received extended induction. Baseline characteristics of this subgroup are described in Table S4. In most cases (n = 23; 70%) there was objective evidence of ongoing disease activity at the week 12 assessment, as represented by an endoscopic Mayo subscore ≥2, CRP ≥5 mg/L or FCP ≥150 µg/g at this time point. In a multivariable logistic regression model (adjusting for baseline serum albumin, disease duration, endoscopic Mayo score, and number of prior advanced therapy classes) oral corticosteroid use (adjusted OR, 12.91; 95% CI, 1.7-185.28; P = .02) and higher baseline body weight (adjusted OR, 1.07; 95% CI, 1.02-1.15; P = .038) were independently associated with having received extended induction (Table S5, Figure 2). At week 12, 54.5% (18/33) of patients who proceeded to extended induction were receiving oral corticosteroids.

Figure 2.

Multivariable predictors of receiving extended induction with mirikizumab. Forest plot displaying adjusted odds ratios with 95% CIs from a multivariable logistic regression model evaluating baseline factors associated with receiving extended induction.

Patients who received extended induction had lower rates of CSF clinical remission at week 28 compared with those who received standard induction (19% vs 46%; P = .034); however, no significant difference was observed at week 52 (22% vs 41%; P = .152) (Figure 3).

Figure 3.

Clinical and endoscopic outcomes by induction strategy. Bars represent the proportion of patients achieving each outcome at the specified timepoint. Patients who discontinued therapy due to ineffectiveness or an adverse event were considered non-responders for all subsequent clinical outcomes. Mucosal healing was defined as an endoscopic Mayo subscore of 0-1.

Treatment persistence

During follow-up, 35.4% (29/82) of patients discontinued MIR. Most discontinuations were due to ongoing UC disease activity (23/29; 79.3%), with the remainder attributable to adverse events. Median follow-up was 45 weeks (95% CI, 38.4-51.4). At week 52, the estimated probability of remaining on treatment was 59% (95% CI, 47.7%-73%) (Table S6 and Figure S3).

Compared with standard induction, receiving extended induction was associated with a higher risk of treatment discontinuation by week 52 (HR, 2.33; 95% CI, 1.12-4.86; P = .02) (Figure 4A). Estimated week 52 treatment persistence was 68.2% in the standard induction group compared with 44.3% in the extended induction group.

Figure 4.

Treatment persistence on mirikizumab stratified by (A) need for extended induction and (B) early corticosteroid-free clinical response. Kaplan–Meier curves depicting the probability of remaining on mirikizumab over 52 weeks, stratified by (A) need for extended induction and (B) corticosteroid-free clinical response at week 12. Tick marks indicate censoring events. P-values were calculated using the log-rank test. Treatment discontinuation was defined as permanent cessation of mirikizumab for any reason. Patients without a documented discontinuation event were censored at their last known follow-up.

Early CSF clinical response was associated with treatment persistence: patients achieving clinical response at week 12 had an 84% lower risk of treatment discontinuation compared with non-responders (HR, 0.16; 95% CI, 0.06-0.47; P < .001) (Figure 4B). Estimated week 52 treatment persistence was 84.1% among week 12 responders compared with 40.0% among non-responders.

Endoscopy and inflammatory biomarker outcomes

Endoscopic and biomarker data were available for a subset of patients during follow-up. Mucosal healing was observed in 20% (4/20) of patients with available endoscopic assessment at week 12, 22% (4/18) at week 28, and 50% (4/8) at week 52. In paired analyses, endoscopic Mayo scores decreased significantly at week 28, but not at week 12 or week 52 (Figure 5).

Figure 5.

Mayo scores at baseline, weeks 12, 28, and 52 stratified by induction strategy. Median and IQR of total Mayo, partial Mayo, and endoscopic Mayo scores measured at baseline, weeks 12, 28, and 52. Analyses were performed using paired Wilcoxon signed-rank tests assessing change in score from baseline to weeks 12, 28, and 52 for each subgroup. Statistically significant differences (P < .05) are shown.

Median CRP decreased numerically from 3.9 mg/L (IQR, 1.0-9.6) at baseline to 1.1 mg/L (IQR, 0.5-1.95) at week 52. Median FCP decreased numerically from 799 µg/g (IQR, 424-1673) at baseline to 106 µg/g (IQR, 60.25-511.5) at week 52. Paired changes in CRP and FCP from baseline were not statistically significant. Available denominators, median [IQR] values, and paired analyses are reported in Table S7, with paired analyses stratified by induction strategy shown in Table S8.

Adverse events

A total of 17 adverse events were reported over the 56.8 patient-years of follow-up, corresponding to an incidence rate of 29.9 per 100 patient-years (95% CI, 17.4-47.9). Adverse events primarily included rash (n = 3) and injection-site reactions (n = 5), which led to treatment discontinuation in 6 of these cases. There were three infectious adverse events, all deemed nonserious and did not result in treatment discontinuation. One serious adverse event occurred during follow-up: a UC-related hospitalization. No colectomies, malignancies, or treatment-related deaths were reported during follow-up.

Discussion

This multicentre study represents one of the largest real-world cohorts of patients with UC treated with MIR reported to date. In this predominantly advanced therapy-exposed population (94%), MIR demonstrated clinically relevant effectiveness over 1 year. Extended induction was required in 40% and was independently associated with baseline corticosteroid use and higher body weight. Early clinical response at week 12 was associated with treatment persistence, suggesting that early treatment response may predict sustained clinical benefit with MIR in routine practice.

The pivotal LUCENT trials demonstrated week 12 clinical remission in 24.4% of patients overall, decreasing to 15.2% among those with prior advanced therapy exposure. These findings are broadly consistent with our observed week 12 remission rate of 24%, despite our predominantly treatment-experienced population.⁵ Maintenance of CSF clinical remission was observed in 44.9% in LUCENT-2, dropping to 40.6% in those previously exposed to an advanced therapy.⁵ In our cohort CSF clinical remission rates were lower (33%) at week 52, possibly reflecting the induction-responder re-randomization design used in the maintenance phase of the LUCENT trials. We observed no difference in clinical outcomes between those exposed to and naïve to ustekinumab, a cohort excluded from these RCTs.⁵ However, the number of subjects who were previously exposed to ustekinumab was relatively small, thus making conclusions about the point estimates of efficacy outcomes less precise.

Real-world studies have reported variable remission rates with MIR. A Japanese multicentre cohort (n = 52) reported 44.2% clinical remission at week 12, although interpretation is limited by high concomitant tacrolimus use (30.8%).¹² A small US cohort (n = 20) demonstrated higher CSF remission rates (78%) despite extensive prior therapy exposure, though baseline inflammatory burden was lower (median FCP 152 vs 799 µg/g in our cohort).¹³ Conversely, an international cohort from Israel and the Czech Republic (n = 74) reported week 12 CSF remission of 16%, in a population with higher rates of severe endoscopic disease (Mayo 3 in 55.6%).¹⁴ A recently published single-centre Japanese cohort (n = 32) reports clinical remission at week 52 in 4.8%. Notably, this cohort only included difficult-to-treat UC (failure of at least 2 advanced therapies with different mechanisms of action) treated with MIR.¹⁵ Overall, differences in baseline severity, concomitant therapy, and outcome definitions likely account for variability across studies rather than true differences in MIR effectiveness.

Strikingly, 40% of our cohort received extended induction, most with objective evidence of ongoing disease activity. Similar proportions have been reported in both RCTs and real-world cohorts. In the previously mentioned cohort from Japan, 38.3% required extended induction.¹² In the LUCENT trials 38% (317/835) patients treated with MIR in LUCENT-1 did not achieve clinical response at week 12, and subsequently 33% (272/835) of individuals treated with MIR entered LUCENT-2 in the extended induction arm.^5,16 Patients in this subgroup had numerically higher rates of severe endoscopic disease, prior advanced therapy failure, baseline corticosteroid use, individuals with a BMI ≥ 25 kg/m², pancolitis, and longer disease duration, compared with induction responders. In our cohort, corticosteroid use at baseline and higher body weight were associated with an increased likelihood of requiring extended induction. Both factors are known to affect the pharmacokinetics of monoclonal antibodies supporting the notion that inadequate drug exposure in some patients may be a contributing factor to lack of effectiveness at this early time point.¹⁷ Analysis of pharmacokinetic data from the LUCENT studies demonstrated that clearance increased with higher body weight and a lower serum albumin level, and bioavailability decreased with a higher BMI.¹⁸ Although the magnitude of these pharmacokinetic effects was modest and not considered relevant, these factors may be clinically relevant in real-world cohorts enriched with patients with severe disease, corticosteroid exposure, and higher body weight. Whether extended induction impacts clinical outcomes remains unclear, however these findings do not support universal extended induction but suggest that patients with higher body weight or baseline corticosteroid use may benefit from closer objective monitoring during induction to guide timely treatment optimization. Prospective studies are needed to determine which patients are most likely to benefit from extended induction.

Six patients (7.3%) discontinued MIR due to an adverse event. These discontinuations were due to injection-site reactions and rash rather than serious adverse events. In LUCENT-2, injection-site reactions occurred in 8.7% of MIR-treated patients compared with 4.2% receiving placebo, and rash occurred in 3.6% compared with 0%, respectively.⁵ More recently, a citrate-free formulation of MIR has been developed and has been associated with less injection-site pain and reactions. However, this formulation was unlikely to have been available to most patients during the period of treatment in our cohort.¹⁹ As such, discontinuation due to adverse-events in this study likely reflect tolerability issues rather than new safety issues.

The favourable safety profile observed in our cohort is consistent with extensive trial and real-world data supporting inhibition of the IL-23 pathway. Across RCTs and open-label extension studies in both UC and Crohn’s disease, agents targeting the IL23 pathway have demonstrated rates of serious adverse events, serious infections, and malignancies comparable to placebo.²⁰ Treatment sequencing in UC increasingly requires consideration of trade-offs between efficacy and long-term safety. In this context, the favourable safety profile associated with inhibition of the IL-23 pathway may influence future positioning of these agents within treatment algorithms, particularly when compared with anti-TNF therapies and Janus kinase inhibitors.²¹

This study has several important limitations. First, the modest sample size limits statistical power, particularly for subgroup and multivariable analyses. Second, the retrospective design restricted analyses to data available in clinical records, resulting in missing longitudinal biomarker and endoscopic follow-up data. This may introduce attrition bias, particularly at later timepoints, as patients without available follow-up assessments may differ from those with complete data. In addition, later timepoint analyses may be affected by survivor bias, as patients remaining on treatment and available for assessment are more likely to represent those deriving clinical benefit. To minimize this, the primary clinical outcome analysis used treatment-failure/adverse-event non-responder imputation, whereby patients who discontinued therapy because of persistent disease activity or adverse events were classified as non-responders at subsequent timepoints. Third, biochemical and endoscopic reassessment was not protocolized, and endoscopy in routine practice may have been preferentially performed in patients with persistent symptoms. This may introduce measurement bias. Therefore, biomarker and endoscopic outcomes should not be interpreted as definitive estimates of objective treatment response but rather as supportive data. Fourth, treatment decisions, including the necessity for extended induction, were made at the treating physician’s discretion, reflecting real-world practice but introducing heterogeneity and potential bias. Despite these limitations, the multicentre real-world nature of the cohort and inclusion of a highly treatment-experienced population enhance the clinical relevance and generalizability of our findings.

Future larger, prospective multicentre cohorts with standardized collection of clinical, biomarker, endoscopic, and patient-reported outcomes are required to better define predictors of response and remission to selective IL-23 inhibition in UC. Although exposure–response relationships were evaluated in clinical trials, prospective real-world studies incorporating therapeutic drug monitoring may help assess individualized induction strategies and clarify whether pharmacokinetic variability contributes to delayed response in selected patients, particularly considering the observed association between body weight, corticosteroid exposure, and receiving extended induction in our study.

In this Canadian multicentre cohort of predominantly advanced therapy-experienced patients with UC, MIR demonstrated clinically acceptable effectiveness, durability, and a favourable safety profile over 1 year. Extended induction was common and associated with baseline corticosteroid use and higher body weight. These findings support the role of MIR in routine clinical practice, even in a highly therapy refractory patient populations, and highlight the importance of real-world data to guide individualized treatment strategies.

Author contributions

C. McShane, I. Kalisky, and J.S. Moore contributed equally to this work. Conceptualization: C.M., I.K., J.S.M., G.R., and M.S.S. Methodology: C.M., I.K., J.S.M., G.R., and M.S.S. Data curation and analysis: C.M., I.K., J.S.M., G.R., M.S.S., R.M., J.S., P.A.O., G.N., K.C., V.W.H., L.E.T., A.V.W., A.H.S., and S.-H.L. Supervision: G.R., and M.S.S. Writing—original draft: C.M., I.K., and J.S.M. Writing—review & editing: All authors.

Supplementary material

Supplementary material is available at Journal of the Canadian Association of Gastroenterology online.

Funding

None declared.

Conflicts of interest

Conflict of interest disclosure forms (ICMJE) have been collected for all co-authors and can be accessed as supplementary material.

Data availability

The data underlying this article cannot be shared publicly due for the privacy of individuals that participated in the study.