Clinical, Endoscopic, and Radiological Effectiveness of Ustekinumab in Bio-naïve Versus Bio-experienced Patients With Crohn’s Disease: Real-world Experience From a Large Canadian Center

Abstract

Introduction

With the expanding therapeutic armamentarium for inflammatory bowel disease (IBD), real-world data may help inform drug positioning. We assessed clinical, endoscopic, imaging, and biochemical response/remission outcomes in patients with Crohn’s disease (CD) treated with ustekinumab in a large Canadian IBD center.

Methods

A retrospective cohort study of CD patients was treated with ustekinumab. Clinical, endoscopic, radiological, and biochemical response and remission outcomes were stratified by prior biologic exposure status. Hazard ratios for biologic exposure status were estimated using Cox proportional hazard models and subgroup-specific incidence rates for healing.

Results

A total of 231 patients (55.9% female, median 45.8 years) were identified as receiving ustekinumab during the study period, with 2 patients subsequently excluded (N = 229). Of these patients, 79.0% (181 of 229) were bio-experienced, with 38.7% (70 of 181) having failed 1 biologic and 61.3% (111 of 181) having failed ≥2 biologics. At 3 months of follow-up after induction, clinical remission (Harvey-Bradshaw Index ≤4) was achieved by 59.1% (62 of 105) of bio-experienced patients and 79.4% (27 of 34) of bio-naïve patients (relative risk [RR], 1.34; 95% CI, 1.06-1.70; P = .013). Endoscopic remission (absence of mucosal ulcers) was achieved in 37.9% (33 of 87) cases. Rate of endoscopic healing (either endoscopic response or remission) per 1000 person-months was 72.7 (95% CI, 42.4-125.1) and 50.2 (37.9-66.4); and the median time to endoscopic response was 8.4 months (95% CI, 6.4-9.8) and 15.4 months (95% CI, 10.3-17.9) in bio-naïve vs bio-experienced patients, respectively. Imaging response/remission and steroid-free remission rates were higher in bio-naïve patients.

Conclusion

In this large real-world cohort of CD patients with complex phenotypes and high rates of prior biologic exposure, we observed that ustekinumab was effective and safe with higher rates of improvement in bio-naïve subjects across a range of end points.

Lay Summary

In this large real-world study of patients with Crohn’s Disease treated with ustekinumab, we observed high rates of clinical, endoscopic, radiological, and biochemical response and remission rates. Effectiveness was greater in bio-naïve compared with bio-experienced patients.

real-world, ustekinumab, Crohn’s disease

Key Messages

What is already known?

The efficacy of ustekinumab for CD is well-established in clinical trials.

What is new here?

There is a need to further evaluate ustekinumab effectiveness in real-world settings. Our study assessed ustekinumab in a large real-world cohort of Crohn’s disease patients with complex phenotypes, comparing its efficacy in bio-naïve vs bio-experienced patients.

How can this study help patient care?

This real-world data will help inform drug positioning in the treatment paradigm of Crohn’s disease, providing supportive data for payers.

Introduction

Crohn’s disease (CD) is a relapsing and remitting inflammatory bowel disease (IBD) of uncertain etiology that can lead to impaired quality of life, loss of schooling and work productivity, and disability. Without a cure, the mainstay of treatment is medical therapy to induce and maintain remission and prevent long-term complications.^1–5 Despite best available medical therapy, intestinal resection remains necessary in a substantial proportion of patients to manage medically refractory disease.^6–8 Even though tumor necrosis factor-alpha (TNF-α) antagonists have been a cornerstone in the management of CD, a substantial proportion of patients do not respond to this therapeutic class. Approximately 30% of patients are primary nonresponders due in part or wholly to non-TNF-mediated disease, and 20% to 40% subsequently experience secondary loss of response due to the development of antibodies or immune escape mechanisms.^9,10 Given that alternative inflammatory pathways may be involved in both types of TNF-antagonist nonresponse, several other biologic classes have been developed. This includes anti-integrins, inhibitors of the interleukin (IL)-12/23 and IL-23 pathways, and oral janus kinase inhibitors.^11–14

Ustekinumab is a monoclonal antibody that targets the p40 subunit of IL-12 and IL-23. It is approved for the intravenous induction of remission in moderate to severe CD, followed by subcutaneous dosing every 8 to 12 weeks for maintenance of remission.^12,15 Following the pivotal registration trials, a subsequent head-to-head trial comparing the efficacy and safety of ustekinumab vs adalimumab in 368 biologic-naïve patients with moderate to severe CD showed that rates of clinical remission and endoscopic response at week 52 were similar for both agents.¹⁶

Although the efficacy of ustekinumab for CD is well-established in clinical trials, there is a need to further assess its effectiveness in real-world settings, particularly in patients who are bio-naïve compared with those who are bio-experienced. A recently published retrospective real-world study in 156 patients with CD, including only 35 bio-naïve patients, found that bio-experienced patients required dose escalation more frequently than bio-naïve patients (16 of 87 [18.4%] vs 1 of 31 [3.2%]; P = < .001), respectively); however, there were similar clinical response and persistence rates between the 2 groups. Real-word data may also be used to help inform drug positioning in the treatment paradigm and provide supportive data for payers.^17,18 Accordingly, we conducted a real-world study to assess the clinical, endoscopic, radiological, and biochemical response, and remission outcomes achieved in patients with CD treated with ustekinumab in a large Canadian IBD center, with a focus on differences between bio-naïve and bio-experienced patients.

Methods

Study Design and Patient Data Collection

A retrospective cohort study was conducted in patients with CD followed at London Health Sciences Center (LHSC), a large tertiary referral center in Southwestern Ontario, Canada, affiliated with Western University. The center has a catchment population of approximately 2 million people and cares for approximately 5000 patients with IBD. Adult patients (18 years and older) with CD receiving ustekinumab between July, 2013, and December, 2018, were identified. Study inclusion criteria were (1) diagnosis of CD for at least 3 months, confirmed by clinical, endoscopic; and histological criteria; (2) failure of at least 1 conventional therapy (corticosteroids, immunosuppressants TNF-antagonist; or anti-integrin inhibitor); (3) receipt of at least a single dose of ustekinumab intravenously or subcutaneously for active CD; and (4) at least 1 clinical and/or endoscopic/radiological follow-up after initiation of therapy. Patients who were started on ustekinumab for ulcerative colitis, pouchitis, or indeterminate colitis were excluded.

Data Extraction

Complete chart review was performed by 1 author (C.M.) and reviewed by a second author (R.S.); discrepancies were resolved by a third author (V.J.). Sources included (1) electronic and paper medical records (clinic letters, endoscopy and radiology records, laboratory results, prescriptions, diagnostic imaging, surgical and pathology reports); and (2) infusion clinic reports that provided timing and dosing of ustekinumab. Data collection was performed using a standardized case report form with prespecified definitions and criteria for coding. All eligible data were entered into a Microsoft Excel spreadsheet. Baseline patient data collected included age, gender, smoking status, duration of CD, disease activity at ustekinumab induction (assessed using the Harvey-Bradshaw Index [HBI]), CD phenotype assessed by Montreal Classification of Disease,¹⁹ endoscopic disease activity at induction, serologic markers of inflammation (C-reactive protein [CRP]), concurrent and previous therapies for CD (including mesalamine, azathioprine, methotrexate, or biologic agents), date of ustekinumab induction, and ustekinumab dosing regimen for induction/maintenance or need for dose escalation. Electronic patient charts were further reviewed to assess clinical and endoscopic response/remission to ustekinumab induction and maintenance therapy, ustekinumab dose-escalation, tapering or re-introduction of systemic corticosteroids (CSs) and immunomodulators, surgery, ustekinumab discontinuation, and any adverse events. Considering the retrospective real-world nature of the study, all follow-up visits at 12 weeks ± 4 weeks (3 months), 24 weeks ± 4 weeks (6 months), and 52 weeks ± 12 weeks (12 months) were included in the analysis.

Outcomes and Definitions

The primary objective of the study was to evaluate the effectiveness of ustekinumab in terms of clinical response and remission at 3, 6, and 12 months after initiation of therapy, stratified by biologic exposure status. Clinical response was assessed by physician global assessment (PGA), defined as a ≥50% reduction in CD-related symptom activity/severity consistent with the definition used in the US multicenter VICTORY consortium studies.²⁰ Clinical remission was defined as an HBI of ≤4 points.²¹

Secondary objectives of the study were to determine the proportion of patients achieving different endpoints: (1) endoscopic response, defined as improvement in the endoscopic appearance compared with baseline; endoscopic remission, defined as the absence of mucosal ulcers as reported by the local endoscopist; and endoscopic healing, defined as either endoscopic response or remission. (2) Imaging response, defined as improvement in the computerized tomography [CT] scan or magnetic resonance imaging [MRI] from previous studies before initiating ustekinumab, assessed by improvement in 1 or more features of bowel wall thickness/mural hyperenhancement/fat stranding/other inflammatory features; imaging remission, defined as the absence of any features of active mucosal inflammation on imaging. (3) Corticosteroid-free clinical response and remission (as per the definitions for clinical response and remission, but also off steroids for 4 weeks before endpoint assessment). (4) Biochemical or CRP response, defined as a decrease in CRP concentration to ≤5 mg/L. Safety outcomes of interest were serious infections or adverse drug reactions attributed to ustekinumab.

Statistical Methods

Patient and clinical characteristics at baseline were summarized by standard descriptive statistics, specifically mean (standard deviation [SD]) for continuous variables or frequency (%) for categorical variables. Time-to-event analyses were used for endoscopic and imaging response end points. Censoring occurred at the earlier time of discontinuation of ustekinumab, death, or the end of the study period. These analyses were stratified by biologic exposure status. Kaplan-Meier curves of achieving healing were generated. Differences in overall time to healing according to biologic exposure status were assessed by the log-rank test. Hazard ratios (HRs) for bio-native status were estimated using Cox proportional hazard models with a robust variance estimator, as well as the overall and subgroup-specific incidence rates (IRs) for healing. Using a univariable Cox proportional hazards model, we examined these baseline characteristics: age, duration, gender, CRP status and/or concentration, smoking status, prior hospitalization or surgery for CD, location of disease involvement, perianal involvement, presence of EIMs, 5-ASA use, immunosuppressant use, and steroid use (Tables 1 and 2, Supplementary Appendix). Using a multivariable model with bio-naïve status, we examined elevated CRP and perianal disease for endoscopic outcomes. For imaging outcomes, we examined elevated CRP, perianal disease, and prior surgery for CD. The Mantel-Haenszel χ²-test and rate ratio was estimated for bio-naïve vs bio-experienced status. Statistical analyses were performed using Stata 16.1 (StataCorp LLC., College Station, TX).

Results

Baseline Characteristics

A total of 231 patients were identified as receiving ustekinumab during the study period. Two patients were excluded for beginning ustekinumab induction on March 1, 2018 (the last day of the study period). Patients’ baseline demographics are summarized in Table 1. Of 229 patients, 128 (55.9%) were female, mean age was 45.8 years (SD, 14.7), and mean disease duration was 14.6 years (SD, 11.6). The most frequent CD phenotype was inflammatory, nonstricturing/penetrating (43.7%, 100 of 230); 35.8% (82 of 229) had prior CD-related surgery; 32.8% (75 of 229) had a history of perianal CD; and 41.5% (95 of 229) had a history of extra-intestinal manifestations, of whom, 53.7% (51 of 95) presented arthralgia, 16.8% (16 of 95) arthritis, 6.3% (6 of 95) psoriasis, 5.3% (5 of 95) skin lesions or oral ulcers, 4.2% (4 of 95) erythema nodosum, and 4.2% (4 of 95) pyoderma gangrenosum, among others (Table 6 Supplementary Appendix).

Table 1.

Baseline (induction) characteristics (N = 229).

Factor	Bio-naïve Status		Overall	P
	No	Yes
N	181 (79.0%)	48 (21.0%)	229 (100.0%)
Sex, n (%)				0.36
Male	77 (42.5%)	24 (50.0%)	101 (44.1%)
Female	104 (57.5%)	24 (50.0%)	128 (55.9%)
Age, mean (SD)	45.9 (14.7)	45.3 (14.9)	45.8 (14.7)	0.79
Smoking status, n (%)				0.90
Never	121 (66.9%)	31 (64.6%)	152 (66.4%)
Former	22 (12.2%)	7 (14.6%)	29 (12.7%)
Current	38 (21.0%)	10 (20.8%)	48 (21.0%)
Prior hospitalization status, n (%)				0.18
Never	94 (51.9%)	32 (66.7%)	126 (55.0%)
> 1 year ago	66 (36.5%)	12 (25.0%)	78 (34.1%)
<=1 year ago	21 (11.6%)	4 (8.3%)	25 (10.9%)
Prior surgery status, n (%)				0.001
Never	54 (29.8%)	28 (58.3%)	82 (35.8%)
> 1 year ago	115 (63.5%)	19 (39.6%)	134 (58.5%)
<=1 year ago	12 (6.6%)	1 (2.1%)	13 (5.7%)
Location (Montreal), n (%)				0.027
Terminal Ileum	27 (14.9%)	16 (33.3%)	43 (18.8%)
Colon	36 (19.9%)	5 (10.4%)	41 (17.9%)
Ileocolon	109 (60.2%)	26 (54.2%)	135 (59.0%)
Upper GI	9 (5.0%)	1 (2.1%)	10 (4.4%)
Behaviour (Montreal), n (%)				0.52
Non-stricturing, nonpenetrating	76 (42.0%)	24 (50.0%)	100 (43.7%)
Stricturing	59 (32.6%)	15 (31.3%)	74 (32.3%)
Penetrating	46 (25.4%)	9 (18.8%)	55 (24.0%)
Perianal disease, n (%)				0.016
No	115 (63.5%)	39 (81.3%)	154 (67.2%)
Yes	66 (36.5%)	9 (18.8%)	75 (32.8%)
Presence of extra-intestinal manifestations, n (%)				<0.001
No	95 (52.5%)	39 (81.3%)	134 (58.5%)
Yes	86 (47.5%)	9 (18.8%)	95 (41.5%)
Missing
Disease duration, years, mean (SD)	15.7 (11.0)	10.7 (13.0)	14.6 (11.6)	0.008
Harvey-Bradshaw Index, mean (SD)	6.1 (3.6)	4.3 (3.1)	5.7 (3.6)	0.003
Harvey-Bradshaw Index status, n (%)				0.028
Remission	58 (37.2%)	24 (57.1%)	82 (41.4%)
Mild	43 (27.6%)	11 (26.2%)	54 (27.3%)
Moderate-Severe	55 (35.3%)	7 (16.7%)	62 (31.3%)
Missing	25	6	31
C-Reactive Protein [mg/L], mean (SD)	16.2 (31.6)	9.4 (12.6)	14.7 (28.5)	0.17
5-ASA use, n (%)				0.32
Never	125 (69.1%)	38 (79.2%)	163 (71.2%)
Former	47 (26.0%)	9 (18.8%)	56 (24.5%)
Current	9 (5.0%)	1 (2.1%)	10 (4.4%)
Methotrexate use, n (%)				0.064
Never	73 (40.3%)	28 (58.3%)	101 (44.1%)
Former	84 (46.4%)	14 (29.2%)	98 (42.8%)
Current	24 (13.3%)	6 (12.5%)	30 (13.1%)
6-MP use, n (%)				0.30
Never	174 (97.2%)	48 (100.0%)	222 (97.8%)
Former	3 (1.7%)	0 (0.0%)	3 (1.3%)
Current	2 (1.1%)	0 (0.0%)	2 (0.9%)
Missing	2	0	2
Azathioprine use, n (%)				<0.001
Never	70 (38.7%)	36 (75.0%)	106 (46.3%)
Former	89 (49.2%)	9 (18.8%)	98 (42.8%)
Current	22 (12.2%)	3 (6.3%)	25 (10.9%)
Currently using immunosuppressive therapy, n (%)				0.26
No	133 (73.5%)	39 (81.3%)	172 (75.1%)
Yes	48 (26.5%)	9 (18.8%)	57 (24.9%)
Corticosteroid use, n (%)				0.91
Never	54 (29.8%)	14 (29.2%)	68 (29.7%)
Former	81 (44.8%)	23 (47.9%)	104 (45.4%)
Current	46 (25.4%)	11 (22.9%)	57 (24.9%)
Current and steroid-dependent	34 (73.9%)	8 (72.7%)	42 (73.7%)
Current and refractory	2 (4.3%)	0 (0.0%)	2 (3.5%)

Note: P values are derived from a χ² test for categorical variables, or 2-sample t test for continuous variables.

At the time of ustekinumab initiation, 24.9% (57 of 229) of patients were using CS. Of these patients, 73.7% (42 of 57) were considered steroid-dependent (ie, patients who respond to steroid therapy but could not taper the treatment); 3.5% (2 of 57) were steroid-refractory; and 24.9% (57 of 229) were receiving concurrent immunosuppressives. The majority of patients were biologic-experienced (79.0%, 181 of 229). Among bio-experienced patients, 68% (123 of 181) had received infliximab as the first TNF-antagonist therapy, and 32% (58 of 181) adalimumab. Furthermore, 38.7% (70 of 181) failed 1 biologic due to loss of response or drug intolerance, and 61.3% (111 of 181) had failed 2 or more biologics. Baseline CRP was ≥5 mg/L in 52.4% (98 of 187) of patients.

According to the PGA assessment at baseline, 18.6% (36 of 194), 39.2% (76 of 194), and 42.3% (82 of 194) of patients were judged as having inactive, mild, or moderate to severe disease activity, respectively. The mean HBI was 5.7 (SD, 3.6)

All patients underwent intravenous weight-based dosing induction with ustekinumab (260 mg for weight ≤55 kg; 390 mg for weight >55 kg to 85 kg; and 520 mg for weight >85 kg) and continued maintenance with 90 mg subcutaneously every 8 weeks. Initial dosing schedule was maintained in 70.7% (162 of 229) of patients, with 30.1% (69 of 229) requiring dose escalation to every 4 weeks. In the overall cohort, bio-naïve patients were twice as likely to have a dose escalation compared with bio-experienced (odds ratio [OR], 2.54; 95% CI, 1.12-5.77; P = .026; Table 7 Supplementary Appendix). Given the relatively small number of patients that required dose escalation, the rates of endoscopic and imaging healing events are broadly similar among bio-naïve and bio-experienced patients (Tables 8 and 9 Supplementary Appendix).

Clinical Response and Remission

At 3 months of follow-up, PGA was assessed in 65% (149 of 229) of patients, of whom 39.6% (59 of 149) achieved clinical response; the HBI was assessed in 60.7% (139 of 229) of patients, with 64% (89 of 139) achieving clinical remission defined as HBI ≤4. At 6 months of follow-up, PGA was assessed in 53% (121 of 229) of patients, with 49.6% (60 of 121) of patients achieving clinical response; the HBI was assessed in 87.6% (106 of 121) of patients, with 59.4% (63 of 106) achieving clinical remission. At 12 months of follow-up for PGA assessment, 51.7% (31 of 64) of patients achieved clinical response, with HBI based clinical remission achieved in 57.1% (28 of 49) of patients.

Based on change in PGA status, achievement of clinical remission was rare at any point of follow-up, so clinical remission and response are combined for the following description. At 3 months of follow-up, the rate of patients achieving clinical response or remission according to the PGA was 59.4% (19 of 32) in the bio-naïve group and 37.6% (44 of 117) in the bio-experienced group (relative risk [RR], 1.45; 95% CI, 0.99, 2.11, P = .056). At 6 months of follow-up, the rate of clinical response was 81.0% (17 of 21) in the bio-naïve group and 46.0% (46 of 100) in the bio-experienced group (RR, 1.93; 95% CI, 1.38, 2.68; P < .001). However, by 12 months follow-up, few patients were assessed in the bio-naïve group (n = 4) with no attempt to make a comparison between biologic exposure status groups.

Based on HBI status at 3 months of follow-up, the rate of clinical remission was 79.4% (27 of 34) in the bio-naïve group and 59.1% (62 of 105) in the bio-experienced group (RR, 1.34; 95% CI, 1.06-1.70; P = .013). At 6 months follow-up, the rate of clinical remission was 71.4% (15 of 21) in the bio-naïve group and 56.5% (48 of 85) in the bio-experienced group (RR, 1.26; 95% CI, 0.91-1.76; P = .16).

Endoscopic Response and Remission

Overall, 38% (87 of 229) of patients had an endoscopic assessment at some point during the follow-up period. Of these, 33.3% (29 of 87) had an endoscopic response (7 of 17 [41.2%] bio-naïve and 21 of 70 [30%] bio-experienced); 37.9% (33 of 87) achieved endoscopic remission (6 of 17 [35.3%] bio-naïve and 28 of 70 [40%] bio-experienced); and 71.3% (62 of 87) achieved endoscopic healing (13 of 17 [76.5%] bio-naïve and 49 of 70 [70%] bio-experienced). The Kaplan-Meier plots showing time to endoscopic healing by biologic exposure status are shown in Figure 1. The median time to endoscopic healing (Table 2) was 8.4 months (95% CI, 6.4-9.8 months) in the bio-naïve group and 15.4 months (95% CI, 10.3-17.9 months) in the bio-experienced group; the proportion with endoscopic healing at 6 months was similar, with approximately 20% achieving this end point. The absolute rate difference of endoscopic healing in the bio-naïve vs bio-experienced group was −0.099 (95% CI, −0.240 to 0.042) at 6 months and 0.425 (95% CI, 0.196-0.654) at 12 months of follow-up (Table 2). The overall incidence rate ratio (IRR) in the bio-experienced group compared with the bio-naïve group was 1.45 (95% CI, 0.79-2.67; P = .23) and was not statistically different. There were no differences in the time to endoscopic healing (log-rank χ²(1) = 2.15; P = .14). The HR for bio-experienced relative to bio-naïve status was 1.59 (95% CI, 0.73-3.44; P = .24).

Table 2.

Rates and median time to healing assessed by endoscopic or imaging, overall and stratified by bio-naïve status.

Group	Number of People With Any Response	Person-Months on Drug	Rate of Healing per 1000 Person-months (95% CI)	Median Time (95% CI) to Healing, Months
Endoscopic healing*
Overall	62	1155.4	53.7 (41.8, 68.8)	12.5 (9.3, 16.5)
Bio-experienced	49	976.5	50.2 (37.9, 66.4)	15.4 (10.3, 17.9)
Bio-naïve	13	178.9	72.7 (42.2, 125.1)	8.4 (6.4, 9.8)
Imaging healing
Overall	22	545.2	40.4 (26.6, 61.3)	18.6 (8.6, 26.9)
Bio-experienced	19	523.0	36.3 (23.2, 57.0)	19.2 (10.7, 26.9)
Bio-naïve	3	22.2	135.1 (43.6, 419.0)	3.5 *

Note: too few events in the bio-naïve group for imaging healing outcomes to estimate a confidence interval for median time to healing.

^*Endoscopic healing = either endoscopic response or remission.

Figure 1.

KM plot of survival for endoscopic healing (A) overall and (B) stratified by bio-naïve status.

Imaging Response and Remission

Overall, 19.7% (45 of 229) of patients had imaging assessment for imaging. Of these, 48.9% (22 of 45) had imaging response (3 of 5 [60%] bio-naïve and 19 of 40 [47.5%] bio-experienced), with 35.6%, (16 of 45) achieving imaging remission (3 of 5 [60%] bio-naïve and 13 of 40 [32.5%] bio-experienced). The Kaplan-Meier plot showing time to imaging healing according to biologic exposure is shown in Figure 2. The median time to imaging healing (Table 2) was 18.6 months (95% CI, 10.7-26.9 months). Most patients assessed by imaging were bio-experienced (88.9%, 40 of 45); therefore, median time to imaging healing in the bio-naïve group is imprecisely estimated. The observed overall incidence rate of imaging healing per 1000 person-months of therapy was 40.4 (95% CI, 26.6-61.3; Table 2). The IRR in the bio-experienced group compared with the bio-naïve group was 3.72 (95% CI, 1.10-12.57; P = .023). The log-rank test suggested faster time to imaging healing in the bio-naïve group (log-rank χ²(1) = 3.78; P = .052). The HR for bio-experienced relative to bio-naïve status was 3.44 (95% CI, 1.27-9.32; P = .015).

Figure 2.

KM plot of survival for imaging healing (A) overall and (B) stratified by bio-naïve status.

Corticosteroid-free Response/Remission and Biochemical Response

During follow-up at 3, 6, and 12 months, 21.4% (36 of 168), 23.3% (30 of 129), and 11.8% (8 of 68) of patients were using CS, respectively. At 3 months follow-up, bio-naïve patients (13.5%, 5 of 37) were less likely to be using CS compared with bio-experienced patients (23.7%, 31 of 131), though this difference was not statistically significant (RR, 0.57; 95% CI, 0.29-1.37; P = .21). By 6 months, no differences were observed between bio-naïve (21.7%, 5 of 23) and bio-experienced (23.6%, 25 of 106) patients (RR, 0.92; 95% CI, 0.39-2.16; P = .85).

At 3, 6, and 12 months of follow-up, the overall trend in CRP response in terms of normalization (CRP < 5 mg/L) were 47% (47 of 100), 50% (36 of 72) and 45% (18 of 40). At 3 months, bio-naïve patients were at similar risk of CRP normalization (52.4%, 11 of 21) compared with bio-experienced (45.6%, 36 of 79) patients (RR, 1.14; 95% CI, 0.71-1.85; P = .57). Despite bio-experienced patients beginning ustekinumab with a higher average CRP concentration compared with the bio-naïve group, average CRP concentration between groups over all collected follow-up time points were similar (P = .85).

Adverse Events

Adverse events (AEs) were reported in 15.7% (36 of 230) of patients. The most frequently described AEs were others, including dry mouth, peripheral edema, rash, and erythema nodosum in 4.8% (11 of 229) of patients, followed by infections in 3.1% (7 of 229), infusion reaction in 2.6% (6 of 229), abdominal pain in 2.2% (5 of 229), arthralgia in 1.7% (4 of 229), and headache in 0.9% (2 of 229) of patients. Serious AEs were rare. One patient presented with basal cell carcinoma, with no apparent relation to using ustekinumab. No patient died during follow-up.

Discussion

In this real-world retrospective cohort study of patients with moderate to severe CD followed in a large Canadian center, the majority of patients treated with ustekinumab in clinical practice were bio-experienced (79%, 181 of 229), with high rates of CD-related surgeries (64.2%, 147 of 229) and a long disease duration of 14 years, collectively indicating a refractory population reflective of a referral center. Despite this, we found that overall patients achieved an early and high rate of clinical response (39.6% and 53% at 3 and 6 months, respectively) and clinical remission at all timepoints. When stratified by prior biologic exposure status, a consistently higher treatment effect in terms of clinical response and remission was observed in the bio-naïve group compared with the bio-experienced group, with similar trends observed across all other end points including endoscopic response and remission, imaging response and remission, and corticosteroid-free remission.

Higher rates of clinical response were observed in this cohort compared with pivotal studies such as the phase 2b randomized controlled trial (RCT) CERTIFI²² and subsequently, the phase 3 RCTs, UNITI-1 and UNITI-2.¹² However, it is not uncommon to observe better outcomes in real-world cohorts than those observed in clinical trials. Common reasons include initiation of treatment in patients with less severe disease at baseline, lack of restriction on use of concomitant medication such as steroids, and less stringent measurement of outcomes. Clinical rates of response and remission observed in the present study are in keeping with those reported in other real-world studies. The Group d’Etude Therapeutique des Affections Inflammatoires du Tube Digestif (GETAID) reported a 65% clinical response rate at 3 months in a cohort of 79 patients with CD treated with ustekinumab.²³ A Spanish cohort of 116 patients with CD reported a clinical response of up to 84% between the first 8 to 12 weeks after induction.²⁴ A Spanish long-term real-world effectiveness and safety study (The SUSTAIN study) including 463 patients with CD reported clinical response and remission rates of 57.6% and 44% at week 8 and 70.2% and 66.1% at week 16, respectively.²⁵ Furthermore, a Canadian experience published by Ma et al in 167 patients with CD reported a clinical response of 60.3% after 6 months of treatment with ustekinumab.¹⁸ Recently published results of the head-to-head trial SEAVUE in 386 patients with moderate to severe CD that were randomized 1:1 to ustekinumab or adalimumab showed a clinical remission rate of 65% at week 52 for ustekinumab, which was a clinical trial design more closely aligned with clinical practice than the pivotal registration trials.¹⁶

Although clinical guidelines recommend routine endoscopic assessment to assess response to therapy after initiation of a biologic, it was notable that only 38% of patients in this cohort had this done in routine practice. Despite this, we observed endoscopic healing of over 60% during follow-up, which is akin to rates between 54% and 76%, reported by other cohorts.^18,26,27 Overall, the bio-naïve population achieved higher rates of clinical response and remission compared with the bio-experienced group. This difference was greater at 3 months of follow-up (RR, 1.34; P = .013) than at 6 months (RR1.26; P = .16). Furthermore, 79% (49 of 62) of the patients that achieved endoscopic healing were bio-naïve. After 12 months of follow-up, the bio-naïve population achieved higher rates of endoscopic healing than the bio-experienced group (absolute rate difference, 0.425; 95% CI, 0.196-0.654). The median time to healing was also earlier in the bio-naïve compared with bio-experienced group (8.4 vs 15.4 months), although the log-rank test suggested that the survival curves were not overall different between groups (log-rank χ²(1) = 2.15; P = .14).

Cross sectional imaging in the form of magnetic resonance enterography or computerized tomography enterography are widely available; these are noninvasive tools to assess disease activity or response to therapy. Although only one-fifth of subjects underwent imaging assessment after initiation of ustekinumab, almost half achieved any grade of imaging response, with 35.6% (16 of 45) of patients achieving imaging remission. Even though most patients with imaging assessment were bio-experienced, the log-rank test suggested faster time to imaging healing in the bio-naïve group (log-rank χ²(1) = 3.78; P = .052). Use of imaging outcomes is likely to become more prominent given the array of validated imaging tools that allow more detailed assessment of disease activity (transmural healing, measurement of bowel wall thickening, differentiation between inflammation and fibrosis) such as the MAGNIFI-CD index for the assessment of perianal fistulizing CD and the simplified MaRIA score to assess activity of luminal CD.^28,29 Recently, the STARDUST trial, a treat-to-target RCT of patients with CD receiving ustekinumab, used intestinal ultrasound as an imaging tool to assess response to treatment; the trial noted that up to 25.5% (14 of 55) of patients achieved imaging response at week 4 of treatment. At week 48 of follow-up, 46.3% (25 of 54) and 24.1% (13 of 54) of patients achieved imaging response and complete transmural healing, respectively.³⁰

Overall, we found a very low rate of AE in our cohort, with only 15.7% of patients reporting an AE, compared with other cohorts that have reported AEs in 30% to 40% of their patients. Furthermore, only 1 patient reported a serious AE, with no apparent relation to ustekinumab. Our results are in keeping with the known favorable safety profile for ustekinumab in patients with CD. The safety of ustekinumab was analyzed in a systematic review of 6 real-world evidence CD studies (N = 578). This study showed a pooled proportion of AEs of 21% (95% CI, 0.12-0.35) and SAEs of 5% (95% CI, 0.03-0.08), similar to our results.³¹ Furthermore, the low rate of antibody formation associated with ustekinumab¹² allows clinicians its use as monotherapy, further reducing the potential risks associated with biologic therapy in these patients. In our cohort, only 24.9% (57 of 229) of patients were using immunosuppressive treatment at baseline, which could have contributed to the low rate of AEs.

Given its mechanism of action, ustekinumab could offer an alternative for treating extraintestinal manifestations or different complex disease phenotypes such as perianal disease, whose treatments are large unmet needs. Both conditions were prevalent in our study, with 41.5% (95 of 229) and 32.8% (75 of 229) of patients having a previous history of extraintestinal manifestations and perianal disease at baseline, respectively. Recently, Guillo et al performed a systematic review including nine studies (N = 254) assessing extraintestinal manifestations in IBD patients. They found that ustekinumab effectively treated arthralgia, psoriatic arthritis, psoriasis, pyoderma gangrenosum, and erythema nodosum, even though it showed no role in spondyloarthritis.³² Moreover, a multicenter retrospective cohort study evaluated the effectiveness of ustekinumab in 207 patients with perianal CD. Among patients with active perianal CD, 38.5% (57 of 148) experienced clinical success after 6 months of treatment (no medical or surgical further intervention of the perianal disease).³³

Our study has many strengths. First, we included a large population of patients with CD treated with ustekinumab, constituting one of the largest real-world studies published to date. Second, we assessed a range of end points including clinical, endoscopic, imaging, and biomarker response and remission, using pragmatic definitions relevant to real-world clinical practice. Limitations to our study are the retrospective design, which limited the possibility of establishing consistent follow-up of patients and outcome assessment. Additionally, information regarding ancestry or ethnicity were not routinely available, which could limit the generalizability of our results. Given the variability in clinical practice, there is lack of standardized timing to perform endoscopic or imaging evaluation. Also, we focused on CRP as the main systemic biomarker of inflammation and were not able to record fecal calprotectin values, as this is not routinely collected and is inconsistently recorded in our health care institution. Furthermore, the scoring systems and outcome definitions in this type of cohort are inherently subjective to the investigator’s interpretation and reporting practice variability.

In conclusion, in this large retrospective real-world cohort of CD patients with high rates of prior biologic therapy exposure, we found ustekinumab was effective for achieving response and remission across clinical, endoscopic, and radiological measurements. When differentiated by prior biologic exposure, we observed that the bio-naïve population achieved higher rates of clinical remission in a shorter period of time, with similar rates of endoscopic healing among both groups but higher rates than those observed in pivotal studies. Furthermore, the safety profile was consistent with the known profile of this agent. These results may help inform positioning and guidance in real life clinical practice.

Conflicts of Interest

R.S. has no conflicts to declare.

L.G. is a statistical consultant and has no conflicts to declare.

C.M.D. has no conflicts to declare.

M.B. has received consulting/advisory board fees from AbbVie, Takeda, Janssen, Lupin, Gilead, Pfizer, Fresenius Kabi Ltd; speaker’s fees from Shire, Allergan, and Pfizer.

N.C. has received honoraria for speaking/consulting from AbbVie, Janssen, Takeda, Pfizer, Novartis, Ferring, Pharmascience, Allergan, Lupin, Shire, and Fresenius Kabi.

J.G. has received consulting/advisory board fees from AbbVie, Amgen, JAMP Pharma, Janssen, Fresenius Kabi, Pfizer, and Takeda; speaker’s fees from AbbVie, Avir, Lupin, Janssen, Pfizer, and Takeda.

M.S. has received honoraria for speaking/consulting from Pendopharm and Medtronic and research grants from Pendopharm.

A.W. has received speaker’s fees from Takeda.

V.J. has received consulting/advisory board fees from AbbVie, Alimentiv Inc (formerly Robarts Clinical Trials), Arena pharmaceuticals, Asahi Kasei Pharma, Asieris, Bristol Myers Squibb, Celltrion, Eli Lilly, Ferring, Flagship Pioneering, Fresenius Kabi, Galapagos, GlaxoSmithKline, Genentech, Gilead, Janssen, Merck, Mylan, Pandion, Pendopharm, Pfizer, Protagonist, Reistone Biopharma, Roche, Sandoz, Second Genome,Takeda, Teva, Topivert, and Vividion; speaker’s fees from, Abbvie, Ferring, Galapagos, Janssen Pfizer Shire, Takeda, and Fresenius Kabi.