Effectiveness of Third-Class Biologic Treatment in Crohn’s Disease: A Multi-Center Retrospective Cohort Study

Background: Multiple studies have described the effectiveness of ustekinumab (UST) and vedolizumab (VDZ) in patients with Crohn’s disease (CD) failing anti- Tumor necrosis factors (TNFs); however, the effectiveness of VDZ or UST as a third-class biologic has not yet been described. Aims and Methods: In this retrospective multicenter cohort study, we aimed to investigate the effectiveness of VDZ and UST as a third-class biologic in patients with CD. Results: Two-hundred and four patients were included; 156/204 (76%) patients received VDZ as a second- and UST as a third-class therapy (group A); the remaining 48/204 (24%) patients received UST as a second- and VDZ as a third-class therapy (group B). At week 16–22, 87/156 (55.5%) patients and 27/48 (56.2%) in groups A and B, respectively, responded to treatment (p = 0.9); 41/156 (26.2%) and 15/48 (31.2%) were in clinical remission (p = 0.5). At week 52; 89/103 (86%) patients and 25/29 (86.2%) of the patients with available data had responded to third-class treatment in groups A and B, respectively (p = 0.9); 31/103 (30%) and 47/29 (24.1%) were in clinical remission (p = 0.5). Conclusion: Third-class biological therapy was effective in more than half of the patients with CD. No differences in effectiveness were detected between the use of VDZ and UST as a third-class agent.


Introduction
In recent years, our arsenal of therapeutic options in Crohn's disease (CD) has been expanding. Despite the diversity of therapeutic options, loss of response remains a significant challenge across all therapeutic agents [1,2]. Tumor necrosis factor (TNF)-alpha inhibitors are the most frequently used first-class biologics; however, approximately 30% of patients experience primary and at least 13% experience secondary loss of response per year [3,4]. Both ustekinumab (UST) and vedolizumab (VDZ) are frequently used as second-class options, with response rates of 40-70% and loss of response rates approximating 20-30% within the first year . As we are currently lacking knowledge regarding predictors of the response to either of these biologic classes of treatment [7], the decision is largely empirical.
Recently, two real-world studies compared the effectiveness of VDZ and UST in CD patients following loss of response to anti-TNFs [27,28]. In both studies, UST was superior to VDZ. However, loss of response to a subsequent second-class biologic is frequent. Moreover, the number of previously utilized biologics is associated with a diminished likelihood of response to biological therapy [29]. To date, there is scarce evidence regarding the effectiveness of a third-class biologic after the failure of two previous classes in CD [30].
The current study aimed to evaluate the likelihood and potential predictors of response to a third-class biologic in CD patients with a previous failure of two biologic classes.

Patient Population
This was a multicenter retrospective cohort study. We included adult patients with an established CD diagnosis who received three different biologic classes for disease treatment. We included all patients who received at least one dose of the third-class biologic. Only patients with active disease (defined as Harvey-Bradshaw index (HBI) ≥ 5) were included. Patients with ostomy or pouch surgery were excluded. Patients that had not yet reached the primary endpoint (week 16) while on active third-class therapy were also excluded.
The study was approved by the Sheba Medical Center ethics committee and written, informed consent was obtained from each patient included in the study. Approval was granted for Helsinki protocol SMC-5598-08 on 28 January 2009. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki, as reflected in a priori approval by the institution's human research committee.

Clinical Scores and Outcomes
The primary outcome of the study was clinical response at week 16-22 (defined by a reduction of HBI ≥ 3). Secondary outcomes included clinical remission (HBI ≤ 4) at week [16][17][18][19][20][21][22], clinical remission and response in week 52, steroid-free clinical remission at week 16-22, C-Reactive Protein (CRP) normalization (CRP serum concentration levels less than normal range as per the cut-off used in the corresponding institutions).

Statistical Methods
Continuous variables were articulated as the median and interquartile range (IQR). Categorical variables were analyzed using Chi-squared/Fisher's exact tests and continuous variables by the t-test/Mann-Whitney test as appropriate. A p-value < 0.05 was considered statistically significant. We constructed a multivariate logistic regression model to identify the independent predictors of week 16-22 response and remission, as well as treatment continuation after induction. Variables with a significance level <0.1 based on univariate analysis were included in the multivariate model. To investigate the effect of the variables on treatment discontinuation, we planned a survival analysis using a Kaplan Meyer survival curve or Cox proportional hazard analysis as appropriate. The analysis was performed using IBM SPSS (version 22.0; Armonk, NY, USA).

Baseline Characteristics
Two-hundred and four patients from 27 centers in 15 countries (22 Europe, 4 Israel, and 1 Canada) were included in the study. Baseline characteristics are detailed in Table 1. All patients had received anti-TNFs as a first-class biological therapy (51 patients had previously received one, 130 had received two, and 23 had received three anti-TNFs). One hundred and fifty-six out of 204 (76%) patients received VDZ as a second-class therapy (median treatment duration on VDZ was 12 months (IQR 5-20)) and UST as a third-class therapy (group A); the remaining 48 of 204 (24%) patients received UST as a second-class therapy (median treatment duration on UST was 10 months (IQR 6.5-18)) and VDZ as a third-class therapy (group B). Median disease duration before the start of third-class treatment was 16 years (IQR 10-22). All patients had active CD (median HBI: 8 (IQR 7-12)) at the time of third-class onset.
One-hundred and thirty-two (64.7%) had elevated CRP at treatment onset. The patient characteristics were similar between the groups, with the exception of the prevalence of perianal disease (group A: 37.8% vs group B: 54.1%, p = 0.04). Figure 1 describes the patient flow during the study.  There was a significant negative association between elevated CRP and clinical response (p = 0.02). Clinical remission was negatively associated with elevated CRP and smoking (p = 0.03 and p = 0.01, respectively) ( Table 2).

Treatment Discontinuation
Seven patients stopped the treatment before week 16, with a median of 10 weeks (IQR 8-12), mainly due to a primary non-response. The overall treatment discontinuation was observed in 19.11% of the patients (36/204), with a median follow-up duration of 48 (IQR 21-52) weeks. Twenty-seven patients out of 156 (17.3%) in group A with a median time of 21 weeks (IQR 16-20) and 9/48 (18.75%) in group B with a median time of 36 weeks (20-45) discontinued treatment (hazard ratio, 1.01; 95% confidence interval (CI), 0.49-2, p = 0.9). The Kaplan-Meier curve for discontinuation-free survival is presented in Figure 3. The main reason for the discontinuation of both treatment groups was a lack of response (group A: 85%, group B: 77.7%). The rest discontinued the treatment because of: patient decision (n = 2), adverse event (3), active extraintestinal manifestation (1).

Safety
Adverse events during the follow-up occurred in 19/204 (9.31%) patients. Thirty patients out of the 204 (14.7%) were hospitalized; the reasons for hospitalization are detailed in Table 3. Twenty-seven patients out of the 204 (13.2%) required surgical intervention, as detailed in Table 3.

Discussion
Our study demonstrates that third-class biologic therapy is a feasible, effective, and safe option in Crohn's disease. It is well established that the number of prior treatments is negatively associated with therapeutic success [29,31,32], and that biologic-naïve patients have a substantially higher likelihood of a response to treatment in inflammatory bowel diseases (IBDs) [17,[33][34][35][36][37]. Until a few years ago, our therapeutic armamentarium in IBD was limited to TNF-alpha inhibitors only, but since then it has expanded to include anti-integrins, IL12/23 blockers, and JAK inhibitors. The natural history of utilization of any new drug naturally favors its use in patients who failed previous and better established treatments: thus, early real-world experience (RWE) series with VDZ included few biologic-naïve patients [18,21,26,38], whereas later series addressing biologic-naïve patients demonstrated higher effectiveness in comparison to those early publications [17,39,40]; for UST, data for biologic-naïve patients is still sparse. Similarly, the literature regarding the use of third-class biologics in IBD is limited to small numbers of patients in RWE series of biologic-experienced patients [27,[41][42][43]. To date, the personalization of therapeutic decisions in IBD is underdeveloped, with limited clues to suggest which succession of therapeutic regimens is superior and subsequently likely to work best in patients failing several consecutive biologic treatments.
Our study focused on the effectiveness and safety of a third-class biologic in patients who had failed two previous classes s of biologics. The overall effectiveness of the thirdclass agent was quite comparable to previous RWE reports on UST and VDZ [15,44]. In several previous studies, UST appeared to be more effective than VDZ in CD patients refractory to anti-TNFs. However, these studies included UST or VDZ as second-class agents and not as third-class agents [27,28].
Adverse events were rare and in line with the previous reports as well. In our study, concomitant immunomodulatory therapy was not associated with the likelihood of response or remission, in similarity with the recently published data [45,46]; nonetheless, the risk of treatment discontinuation was higher in patients on monotherapy. The effect of concomitant immunomodulators in the present study could be explained by a decrease in the risk of anti-drug antibody formation in line with the recently reported data on sequential anti-TNF use [47] or due to an additional anti-inflammatory effect [48,49]. Another somewhat surprising finding was the lack of association of the number of previously used anti-TNFs with therapeutic outcomes. A possible explanation for this observation could stem from the fact that all our patients have already experienced failure of multiple biologics, potentially obviating the subtle differences that could have been observed in a more treatment-naïve cohort.
Despite the fact that 19.11% of the patients discontinued the therapy, the majority continued the therapy with a high response (87.1%) and remission rate (28.7%). Importantly, as the patients included in this cohort have already experienced multiple classes of biological therapy and have quite exhausted their treatment options, it is plausible that in a proportion of cases the treatment was continued for a longer duration of time despite the lack of a clear early therapeutic response.
Unfortunately, the groups (VDZ or UST as a third-class agent) were dissimilar in size; the likely explanation for this is the earlier introduction of VDZ into clinical practice. Although we could not demonstrate any significant difference between the groups in either clinical characteristics or outcomes, the comparison is compromised by the disproportional cohort size. Some of the additional limitations of our study are inherent to an RWE multicenter retrospective study design (including a small number of patients with available endoscopic data; missing laboratory data, including drug levels and anti-drug antibodies; and heterogeneity in the scheduled visit dates, follow up data about perianal disease, and extraintestinal manifestations). Therefore, no conclusion could be drawn on the effect of these third-class biologicals on endoscopic endpoints and perianal disease. The study cohort was narrow in terms of age (41.5 years old (IQR 32-53)) and may not well reflect the responses of all patients. There is no representation for the elderly nor the young under the age of 30.
Despite the aforementioned limitations, our study supports the use of third-class biologics in CD. Further research is required to identify the most effective succession of treatment regimens in IBD, and to further personalize decision-making on those challenging patients. Informed Consent Statement: Informed consent was obtained from each patient included in the study. Approval was granted for Helsinki protocol SMC-5598-08 on 28 January 2009. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institution's human research committee.