Real-World Evidence of the Effectiveness and Safety of Ustekinumab for the Treatment of Crohn’s Disease: Systematic Review and Meta-Analysis of Observational Studies

(1) Background: Evidence on the outcomes of ustekinumab treatment in real-world Crohn’s disease (CD) patients is needed. Our aim was to evaluate the effectiveness and safety of ustekinumab in CD, reported by observational studies. (2) Methods: bibliographical searches were performed (PubMed, EMBASE). Selection: observational studies assessing the effectiveness and safety of ustekinumab in CD. Exclusion criteria: studies using ustekinumab as a prophylaxis for postoperative recurrence or perianal disease. Data synthesis: effectiveness by intention-to-treat (random-effects model). Data were stratified by study design, population included, administered dose, and prior biologic exposure. (3) Results: A total of 63 studies (8529 patients) were included. Response was achieved in 60% (95% CI, 54–67%) in the short term (8–14 weeks); 64% (57–71%) in the medium term (16–24 weeks); and 64% (52–74%) in the long term (48–52 weeks). Remission was achieved in 37% (28–46%) in the short term; 42% (36–49%) in the medium term; and 45% (37–53%) in the long term. The endoscopic remission rate was 33% (25–40%) in the long term. Eighteen percent of patients lost response during follow-up. Nearly one-third of the patients needed dose optimisation, and in 59% of them it was effective. Twenty-five percent of patients developed adverse events, leading to treatment withdrawal in seven percent of the cases. (4) Conclusions: Ustekinumab is an effective and safe therapy in real-world refractory CD patients. Dose optimisation is frequently required, being effective in a high percentage of cases.


Introduction
Inflammatory bowel disease (IBD) comprises a series of chronic disorders of unknown cause affecting the gastrointestinal tract, and it is associated with a complex immune response. Treatment options for IBD are rapidly expanding because the currently available treatments are still ineffective in many patients [1][2][3][4].
Ustekinumab is a fully human monoclonal IL-12/23 p40 antibody. IL-12 and IL-23 play a key role in the inflammatory cascade in Crohn's disease (CD). According to the summary of product characteristics, ustekinumab is approved for the treatment of patients with moderate-to-severe CD who have had an inadequate response with, lost response to, or were intolerant to either conventional therapy or to anti-tumour necrosis factor (anti-TNF), or have medical contraindications to these therapies [5].
The efficacy of ustekinumab has been shown in the UNITI-1, UNITI-2 (induction) and IM-UNITI (maintenance) clinical trials [6][7][8]. As with other biological treatments, efficacy was greater in patients naïve to anti-TNF drugs (UNITI-2) than in previous non-responders to such drugs (UNITI-1). The safety profile of the drug is favourable. Nevertheless, randomised clinical trials may not represent the real-world IBD population because an important proportion of IBD patients do not meet their strict inclusion criteria [9,10].
In this scenario, in the last few years, observational studies reporting the effectiveness and the safety of ustekinumab for CD have been conducted. However, the published studies are scarce, with few patients mainly included in unicentric cohorts. Therefore, the effectiveness and safety of ustekinumab are not yet clear. The aim of this study was to evaluate, through a systematic review and meta-analysis, the effectiveness and safety of ustekinumab in the treatment of CD reported by observational studies.

Literature Search and Study Selection
Bibliographic searches were performed in PubMed and EMBASE up to December 2021. The search strategy (with corresponding keywords in all fields) was: ("inflammatory bowel disease" OR "crohn's disease") AND ustekinumab. Additional hand searches were performed by cross-referencing eligible studies in order to identify further relevant publications. We also included conference proceedings of the last five years from Digestive Diseases Week (DDW), United European Gastroenterology Week (UEGW), the European Crohn's and Colitis Organisation (ECCO), and the World Gastroenterology Organisation (WGO). Abstracts were screened to discard duplicates, and when the literature search yielded two or more studies by the same author assessing the same populations, only the most recent one was chosen, irrespective of the time interval, as it was assumed that the last one published would include the most comprehensive and complete data. The corresponding authors of the studies without sufficient data were contacted for additional information. The process of study selection is depicted in a flow diagram following the PRISMA statement [11]. The present systematic review was registered in PROSPERO (CRD42021273274).
Two reviewers (CR and MC) selected the articles, first by title and abstract and then by full-text review and following the selection criteria. Any discrepancy during the selection of references was solved by consensus with a third reviewer (JPG).

Selection Criteria
Prospective and retrospective studies assessing the effectiveness or the safety of ustekinumab in CD were selected for inclusion. There were no language restrictions, and studies focused on paediatric patients could be included. Articles in which ustekinumab had been prescribed exclusively as prophylaxis for postoperative recurrence in CD or for perianal CD were excluded. Systematic or narrative reviews and clinical trials were excluded from this systematic review.

Data Extraction
A predefined data-extraction form was used to collect the data. The variables recorded were: year of publication; study design (prospective or retrospective); age of the study population (adults or children); sample size; previous biologic exposure (naïve or nonnaïve); use of concomitant immunomodulator therapy; administered dose of ustekinumab; ustekinumab as induction or maintenance therapy; length of follow-up (in months); outcome measures (clinical and endoscopic response, clinical and endoscopic remission, and corticosteroid-free clinical remission); and predictors of response (if any). Outcome measures were reported in the short term (8-14 weeks), in the medium term (16-24 weeks), and in the long term: 48-52 weeks, where available. We also collected the need for dose optimisation, median time of initiation of therapy intensification, and effectiveness of dose optimisation. Dose escalation was defined as a shortening of the administration interval from every 12 weeks to every 8 weeks; dose intensification was defined as a shortening of the ustekinumab administration interval to less than 8 weeks (every 4 or 6 weeks) or administration of a reinduction IV dose of ustekinumab. Adverse events (AEs) associated with ustekinumab treatment were also recorded (including AEs that required ustekinumab dis-continuation and serious AEs related to ustekinumab). We defined serious AEs following the criteria of the Food and Drug Administration (FDA): the need for hospitalisation, disability or permanent damage, life-threatening, required intervention to prevent permanent impairment or damage, congenital anomaly/birth defect, or death.

Quality Assessment
To assess the quality of the observational studies (only the full-text ones) we used the "Newcastle-Ottawa Scale" (NOS), which is considered the most reliable method for outcome assessment [12]. The evaluated items of NOS are detailed in Figure S1. Clinical and endoscopic response and remission in CD patients treated with ustekinumab in the short, medium, and long term, in a real-life setting.

2.
Safety of ustekinumab in CD patients.

1.
Effectiveness of intensification of ustekinumab treatment (either by decreasing the intervals of ustekinumab administration or by dose intensification (shortening of the ustekinumab administration interval to less than 8 weeks or administration of a reinduction IV dose of ustekinumab)).

2.
Predictive factors of response in a real-life setting.

Data Synthesis and Statistical Analysis
All analyses were pre-planned a priori. The outcomes were thereafter combined using the inverse variance method, providing 95% confidence intervals (CIs). Due to the expected high heterogeneity in the design and the results of the studies, a random effects model was used. Heterogeneity was analysed using I 2 statistic: according to I 2 values, the heterogeneity was considered: not important (I 2 < 40%), moderate (40-75%), and considerable (>75%). Such interpretations were also adjusted for the magnitude of the effect and/or the strength of the evidence given (i.e., p-value <0.1 of the Chi 2 test). Safety data were reported as the proportion of AEs per patient.
Begg's funnel plot was used to estimate the possibility of publication bias [13]. Post hoc sensitivity analyses were performed for each meta-analysis subgroup by excluding those studies that were identified as potentially introducing a critical risk of bias that could likely modify the outcome. Data were analysed using the Review Manager software (version 5.4.1, Copenhagen, Denmark).

Study Selection and Characteristics
A total of 63 studies (including 8529 patients) met the inclusion criteria and were finally included in the systematic review and meta-analysis ( Figure 1) . Forty-four (70%) were reported as full-text articles and nineteen (30%) were abstracts.

Effectiveness of Ustekinumab
The effectiveness of ustekinumab in the selected studies is summarised in Table 2.

Effectiveness of Ustekinumab
The effectiveness of ustekinumab in the selected studies is summarised in Table 2.
Subgroup analyses for clinical response based on the proportion of patients who were biologic-naïve, by type/design of study (full text vs. abstract; prospective vs. retrospective), by study population (children vs. adults), or by ustekinumab induction regimen did not explain between-study heterogeneity.
Subgroup analyses for clinical response based on the proportion of patien were biologic-naïve, by type/design of study (full text vs. abstract; prospective v spective), by study population (children vs. adults), or by ustekinumab induction r did not explain between-study heterogeneity.
Subgroup analyses for endoscopic remission based on the proportion of patients w were biologic-naïve, by type/design of study (full text vs. abstract/prospective vs. retr spective), by study population (children vs. adult), or by ustekinumab induction regim did not explain between-study heterogeneity.

Loss of Response to Ustekinumab
The loss of response rates and the median time to loss of response of each stu (when available) are detailed in Table 3. The loss of response rate ranged from 0% to 67
Subgroup analyses for endoscopic remission based on the proportion of patients who were biologic-naïve, by type/design of study (full text vs. abstract/prospective vs. retrospective), by study population (children vs. adult), or by ustekinumab induction regimen did not explain between-study heterogeneity.

Loss of Response to Ustekinumab
The loss of response rates and the median time to loss of response of each study (when available) are detailed in Table 3. The loss of response rate ranged from 0% to 67%, with an overall pooled rate of 18% (95% CI, 14-22, I 2 = 98%). Ustekinumab secondary loss of response rate in the medium term (six months) was 18% (95% CI, 13-24, I 2 = 88%). In the long term (12 months), the overall loss of response to ustekinumab rate was 18% (95% CI, 13-23, I 2 = 98%).

Quality of Included Studies and Risk of Publication Bias
Quality of evidence grading of the studies ranged from five (eight studies) to nine (one study). Four studies (9.3%) were graded as 'high quality' studies (Newcastle-Ottawa score ≥ 7). The quality of each included study is detailed in Table 5.
Begg's funnel plot was used, and we did not identify publication bias ( Figure S4).

Discussion
To the best of our knowledge, this systematic review and meta-analysis summarises the largest collection of real-world evidence assessing the effectiveness and safety of ustekinumab in CD, including 63 studies and 8529 patients. Our analysis reports relatively high remission and response rates in refractory CD patients (i.e., those previously exposed to other biological agents), supporting that ustekinumab represents a relevant therapeutic option in the management of CD.
The efficacy and safety of induction therapy with ustekinumab in patients with moderately to severely active CD was demonstrated in the IM-UNITI trials, in which ustekinumab was shown to be effective and safe in the long term (up to 5 years in preliminary data) [6][7][8].
Despite the benefit of ustekinumab in the aforementioned clinical trials, real-world studies are needed to confirm the effectiveness and the safety reported by randomised studies.
Our systematic review almost exclusively includes refractory CD patients: almost all were biologic-experienced, and the vast majority had experienced failure of two or more anti-TNF drugs ( Table 1). The overall clinical response and remission rates in our study in the short term were relatively high (60% and 37%, respectively). Moreover, approximately one-third of the patients included in our analysis were in corticosteroid-free clinical remission at week 14. However, the percentage of patients who had a response at week 6 in the UNITI-1 clinical trial was lower (34%) [6]. This difference may be partially explained by the less stringent definition of response in real-world studies included in our review (employing mainly the HBI and/or PGA) in contrast with clinical trials, in which CDAI is generally used. Additionally, short-term clinical response was evaluated after 6 and 8 weeks in the UNITI-1 clinical trial, while in the studies included in our meta-analysis, it was evaluated between 8 and 14 weeks. In the UNITI-1 trial, the one-year remission rate was 41% [6]; in the long term (week 52), 34% of patients in the every-8-weeks group and 29% in the every-12-weeks group were in clinical remission [8]. In the long term, our pooled clinical remission and response rates were 45% and 64% at one year. Additionally, we could not confirm the higher clinical response rates associated with subcutaneous compared with intravenous induction previously reported by Macaluso et al. [77].
Regarding endoscopic outcomes, results should be interpreted with caution, because the number of patients with available endoscopic data was quite limited (Table 2). Currently, data from clinical trials on the ability of ustekinumab to induce mucosal healing are scarce. In IM-UNITI, a sub-study showed rates of endoscopic response and remission of 17% and 11%, respectively [78]. Despite the heterogeneity of our results (due to the different criteria for the assessment of endoscopic response and the limited availability of these data in clinical practice), remarkably significant endoscopic remission rates were reported in the medium (six months, 24%) and in the long term (12 months, 33%).
Previous biologic exposure, stricturing, and penetrating CD and higher HBI at baseline were associated with a lower probability of achieving clinical remission in the observational studies included in our systematic review. While these clinical outcomes are promising (and could be taken into account in the future for the selection of a treatment schedule), controlled studies are necessary to confirm these data.
Medical treatment options for anti-TNF refractory patients are eagerly needed, and ustekinumab has recently emerged as a new therapeutic target. Despite the effectiveness of ustekinumab reported in short-and long-term studies, a significant percentage of patients treated with this drug may lose response (18% in our study). In this scenario, the need for dose optimisation is relatively high in clinical practice, as shown by our results, in which dose optimisation was needed in 30% of cases.
In our review, 60% of the intensified patients achieved clinical remission with different dose optimisation regimens: dose escalation to every 8 weeks and/or intensification every 6-4 weeks, or intravenous reinduction. Reinduction with intravenous ustekinumab after secondary loss of response in CD is a relatively new strategy to regain efficacy. In this scenario, an observational and multicentre study of 53 patients in Spain reported that 49% and 43% of the cohort were in clinical remission at week 8 and week 16 after reinduction, respectively, whereas 64% and 53% had clinical response [79].
Finally, two recently published systematic review and meta-analyses evaluated the safety of ustekinumab in clinical trials [8,80]. Rolston et al. compared rates of AEs in randomised controlled trials of ustekinumab compared to placebo among a spectrum of autoimmune diseases. Of the 30 studies included, 5 were conducted in IBD patients. In this subgroup, there were no differences in the rates of serious or mild/moderate AEs for ustekinumab vs. placebo or in the rates of AEs for low vs. high-dose ustekinumab [80]. Likewise, Sandborn et al. reported that safety events were similar in the placebo and in the ustekinumab groups for all AEs, serious AEs, infections, and serious infections [8].
To our knowledge, the safety of ustekinumab in observational studies was recently summarised in three reviews [77,81,82]. In the systematic review of Macaluso et al., including thirteen studies (1450 patients), the pooled incidence rate of total AEs was 19.1 per 100 patients-year [77]. Similarly, Engel et al. combined the results of 578 patients from six different cohorts, and a total of 134 AEs were reported with a pooled proportion of 21%. Finally, in the study of Honap et al., 498 AEs were reported in 2977 patients (17%) resulting in a pooled estimate of incidence rate of 14% [82].
Our meta-analysis of 8529 patients shows similar results, with a pooled incidence rate of 26%. However, it is necessary to emphasise that AEs were reported heterogeneously in the observational studies included. We defined serious AEs following the criteria of the FDA to minimise this bias. Thus, despite the heterogeneity of reported AEs, our study confirms that the proportion of serious AEs was low, and only a minority of them (6.8%) led to discontinuation of ustekinumab treatment.
The limitations of the present study are mainly those associated with observational studies. Firstly, the majority of included studies had a retrospective study design and, as such, were susceptible to bias and confounding. Secondly, it is necessary to underline the high degree of between-study statistical heterogeneity in data analyses. In our systematic review and meta-analysis, data were stratified by study design, population included, administered dose of ustekinumab, and prior treatment with biological drugs. All the performed sub-analyses did not adequately explain between-study heterogeneity.
Nevertheless, our study has several strengths. First, this is the largest and longest real-world meta-analysis evaluating the effectiveness and safety of ustekinumab in CD published to date. Furthermore, we also provide data regarding the need for dose optimisation and its effectiveness, which is also a relevant issue in clinical practice.
In conclusion, our meta-analysis emphasises that ustekinumab is effective and safe for the treatment of refractory CD, and that its clinical benefit seems to be higher in real-life observational studies compared with controlled clinical trials. Finally, dose optimisation of ustekinumab is frequently required in clinical practice, and achieves clinical response in a high percentage of cases.