Effectiveness and safety of vedolizumab and infliximab in biologic-naive patients with Crohn’s disease: results from the EVOLVE study

Abstract

Objectives

This study compared the real-world effectiveness and safety of α₄β₇-integrin inhibitor vedolizumab and anti-tumor necrosis factor alpha (anti-TNFα) inhibitor infliximab in biologic-naive patients with Crohn’s disease (CD).

Methods

EVOLVE was a retrospective, multicenter, medical chart review of biologic-naive adults with inflammatory bowel disease receiving vedolizumab or anti-TNFα treatment as first-line biologics in Canada, Greece, and the USA. Twelve-month outcomes were analyzed in vedolizumab- or infliximab-treated patients with moderate-to-severe CD (and subgroups with complicated and noncomplicated CD) including cumulative rates of clinical response, clinical remission, and mucosal healing, and incidence rates of serious adverse events (SAEs) and serious infections (SIs). Inverse probability weighting (IPW) was used to account for baseline differences between treatment groups.

Results

Data were analyzed from 167 patients. In the IPW dataset (99 vedolizumab-treated and 63 infliximab-treated), adjusted 12-month clinical remission rates were 73.1% and 55.2%, respectively (P = 0.31). Overall, effectiveness rates were similar across treatment and complicated/noncomplicated disease subgroups. Adjusted 12-month incidence rates (first occurrence/1000 person-years) of SAEs for vedolizumab vs. infliximab: 43.6 vs. 200.9 [hazard ratio (HR) 0.36 (0.09–1.54)]; SIs: 10.8 vs. 96.0 [HR 0.08 (<0.01–2.64)]. AE incidence was significantly lower in vedolizumab- vs. infliximab-treated patients for complicated [131.6 vs. 732.2; HR 0.19 (0.05–0.65)] and noncomplicated [276.3 vs. 494.8; HR 0.59 (0.35–0.99)] disease subgroups.

Conclusion

These real-world data on first-line biologics show no differences in 12-month effectiveness outcomes for vedolizumab- vs. infliximab-treated biologic-naive patients with CD. Vedolizumab may have a more favorable safety profile vs. infliximab in patients with complicated and noncomplicated disease.

Introduction

Crohn’s disease (CD) is characterized by chronic inflammation of the gastrointestinal (GI) tract of unknown etiology, although environmental, immunologic, and genetic factors play a role [1–3]. Patients with CD often require lifelong treatment and cause substantial health care resource utilization, including frequent hospitalization, resulting in impaired quality of life [4–6]. Optimized anti-inflammatory therapy is crucial for effective long-term patient management [7]. Advanced targeted therapies, designed to inhibit particular inflammatory processes [8], have been approved in patients with moderately to severely active CD [3,9–11]. Anti–tumor necrosis factor alpha (anti-TNFα) biologics include infliximab [12,13] and adalimumab, among others used as treatment of patients with moderate-to-severe CD escalated to advanced therapy [3,9–11]; however, analyses suggest that between 20% and 40% of patients do not respond to induction with anti-TNFα treatment, whereas loss of response occurs in ~40% of patients after the first year of anti-TNFα maintenance therapy [14–16]. Furthermore, anti-TNFα treatment has been associated with an increased risk of infections [17,18].

Vedolizumab is a gut-selective anti-lymphocyte trafficking anti-integrin agent used to treat moderately to severely active CD [[19–21]]. In the GEMINI 2 [22] phase III clinical trial, vedolizumab demonstrated efficacy as a first-line biologic in biologic-naive and bio-experienced patients [22,23]. Most real-world studies have examined vedolizumab treatment outcomes in patients who had previously been exposed to anti-TNFα treatment [24–26]. The EVOLVE (Entyvio Outcomes in Real-World Bio-Naive Ulcerative Colitis and Crohn’s Disease Patients) real-world study compared the effectiveness and safety of vedolizumab with anti-TNFα agents in a large multicenter cohort of biologic-naive patients with CD over 24-months of follow-up from treatment initiation [27]

The choice of advanced treatment in patients with CD who have failed steroids and/or conventional immunomodulators has important consequences when optimizing long-term treatment outcomes for individual patients [3,7]. As head-to-head clinical trials in this area are rare, comparative effectiveness and safety studies with real-world data can be useful when positioning biologic drugs in the treatment algorithm for CD [11,28,29–31].

Findings from the EVOLVE study and other real-world evidence indicate that vedolizumab has similar effectiveness to anti-TNFα treatments in biologic-naive patients with CD [24,27,32,33], while exhibiting a more favorable safety profile [23,27].

To date, little is known about the comparative long-term effectiveness and safety of vedolizumab vs. specific anti-TNFα treatments in biologic-naive patients with CD. Some meta-analyses have suggested that infliximab appears to be the most effective among the different anti-TNFα treatments currently available [28,30,34].

The aim of the current study was to evaluate effectiveness, safety, and healthcare resource utilization outcomes in vedolizumab-treated vs. infliximab-treated biologic-naïve patients with moderate-to-severe CD during 12 months of follow-up after treatment initiation.

Materials and methods

Study design

EVOLVE was a multicenter, retrospective, medical chart review study conducted across multiple sites in three countries (USA, Canada, and Greece) [27]. A total of 37 study sites participated across the three countries (USA, 15 sites; Canada, 13 sites; Greece, nine sites). The sites comprised geographically dispersed university/teaching hospitals, community hospitals, and clinics/outpatient facilities of various sizes. The study received local ethics committee approval at each participating site. Patients alive at the time of chart abstraction (99% of patients) had signed an informed consent form prior to study participation if there was no waiver of informed consent. The study received local ethics approval in line with country-specific requirements.

The follow-up period began 1 day after the start of index treatment with vedolizumab or infliximab and ended at the date of death, loss to follow-up, start of chart abstraction, or 6 months post treatment discontinuation, whichever was earliest. Patient data from the time between disease diagnosis to treatment initiation (pre index) and at the time of treatment initiation (baseline) were also collected. A minimum 6-month follow-up period was used for all patients. Outcomes were analyzed during a 12-month period from initiation of index treatment; those patients who were lost to follow-up during the 12-month period (or between 6 and 12 months) were censored in the time-to-event analysis.

Participants

Patients eligible for study inclusion were aged ≥18 years with a chart-documented diagnosis of CD; they were biologic treatment-naive and had initiated first-line biologic treatment receiving at least one dose of either vedolizumab or infliximab (i.e. the index treatment, including originator and biosimilars) between 20 May 2014 and 31 July 2017 (the eligibility period). Patients receiving concomitant nonbiologic therapies (aminosalicylates, corticosteroids, and immunomodulators) were also included. Eligible patients had moderate-to-severe disease at index treatment initiation (baseline) and had ≥6 months of follow-up data from initiation of index treatment. Additionally, patients who were treated with vedolizumab through a compassionate use access program, prior to approval in Canada, were also eligible for study inclusion. In Greece, only patients who were treated with vedolizumab were enrolled. Patients were excluded if their index treatment was administered as part of a clinical trial or was initiated as combination therapy with another biologic agent, or if they had received prior treatment with biologic agents for conditions other than inflammatory bowel disease. Patients who met eligibility criteria were randomly selected for inclusion in the study so that similar numbers of vedolizumab- and anti-TNFα–treated patients were included.

Data collection and outcome measures

All data were collected retrospectively from patient medical charts. Data abstraction from patients’ charts took place between 21 September 2017 and 14 December 2018. Available data on patient demographics, clinical history (including CD-related hospitalizations in the past 12 months of index date), treatment history (including prior systemic nonbiologic therapy for inflammatory bowel disease such as corticosteroids, aminosalicylates or immunomodulators within the past 24 months of index), disease characteristics at diagnosis (intestinal location, disease behavior, active fistulae status), and biochemical disease markers (laboratory assessment of fecal calprotectin [FCP] ≥ 250 mg/kg, or if unavailable, C-reactive protein [CRP] ≥ 5 mg/l, or if unavailable albumin <35 g/l within 3 months of index), were collected. Baseline disease severity was determined according to standard clinical assessments for endoscopic and symptomatic CD severity conducted nearest to index treatment initiation (≤6 months prior to initiation of vedolizumab or infliximab treatment), as available in patients’ medical charts (Simple Endoscopic Index for CD [SES-CD] score: 7–15 for moderate and ≥16 for severe disease, or if unavailable, Crohn’s Disease Activity Index [CDAI] score: 220–450 for moderate disease and >450 for severe disease; or if unavailable, endoscopic findings: 2 for moderate and 3 for severe disease, or if unavailable, Harvey-Bradshaw Index [HBI] score: 8–16 for moderate and ≥16 for severe disease; or if unavailable, modified HBI score [excludes abdominal mass]: 8–16 for moderate and ≥16 for severe disease; or if unavailable, Physician’s Global Assessment [PGA] score: 2 for moderate and 3 for severe disease) see Supplementary Table 1, Supplemental digital content 2, http://links.lww.com/EJGH/A964 for hierarchical algorithm of standard clinical assessments used to determine baseline disease severity. Additional details on patient treatment patterns, effectiveness outcomes (clinical assessments of disease activity and treatment response including endoscopy findings), disease exacerbations, hospitalizations, surgical procedures, concomitant nonbiologic treatment, and safety events were collected after index treatment.

Effectiveness outcomes included clinical response, clinical remission, and mucosal healing. Clinical response was defined as a positive change in CDAI score category from baseline (baseline score CDAI categories 220–450 or>450). If CDAI scores were unavailable, a decrease of ≥3 points from baseline in overall HBI score was used, or if unknown, a decrease of ≥3 points from baseline in the modified HBI score, or if unknown, a medical chart entry of ‘complete’ or ‘partial response’ was used. Clinical remission was defined as a CDAI score of <150 points, or if unknown, a HBI score of ≤4 points, or if unknown, a modified HBI score of ≤4 points was used, or if unknown, an ‘in remission’ status as recorded in the medical chart was used [27]. Mucosal healing was defined as an endoscopic assessment score of 0 or 1 (normal or inactive or mild disease), or if unknown, a Simple Endoscopic Score for CD of <3, or if unknown, one or more endoscopic procedural findings indicating lack of ulcerations (i.e. ‘no ulcers’ or free-text indication of lack of ulceration), or, if unknown, one or more endoscopic procedural findings indicating inactive disease (i.e. no findings, no active disease, no erosions, no inflammation, no inflammatory activity, or no pathologic findings). The hierarchical algorithm used to classify effectiveness outcomes based on standard measures of disease activity [27] is also detailed in Supplementary Table 1, Supplemental digital content 2, http://links.lww.com/EJGH/A964.

Treatment patterns analyzed included treatment persistence (did not discontinue vedolizumab or infliximab treatment for any reason during the study follow-up period) and dose escalation (defined as an increase in treatment dosing frequency with vedolizumab or dose and/or dosing frequency with infliximab for two or more consecutive drug administrations). Disease exacerbations were defined as the worsening of symptoms attributed to CD. Primary nonresponse (PNR) was defined as discontinuation of treatment due to no response or only partial response, disease exacerbation, surgical procedure in the bowel, an inflammatory bowel disease–related hospitalization that was not related to treatment or diagnostic procedures and tests, or a dose escalation ≤14 weeks after initiation of vedolizumab or infliximab treatment. The definition of secondary loss of response (SLOR) or need for dose escalation was similar to PNR but with a dose escalation after >14 weeks from initiation of vedolizumab or infliximab treatment.

The subgroup of patients with complicated disease included patients with penetrating disease and those with a mixed inflammatory/stenotic phenotype. Patients with noncomplicated CD were defined as those with no active fistula at vedolizumab or infliximab treatment initiation, no CD-related surgeries since diagnosis, and no CD-related hospitalizations within 12 months prior to treatment initiation. In order to evaluate safety outcomes within the timeframe of detectable serum levels, serious adverse events (SAEs) and serious infections (SIs) from the start of treatment up to 5 half-lives after treatment discontinuation were examined (half-lives: vedolizumab, 125 days [18 weeks] and infliximab, 47.5 days [6.8 weeks] http://links.lww.com/EJGH/A964). Adverse events (AEs) or infections were classified as serious if they were life-threatening, required hospital admission, resulted in significant disability/incapacity, or were recorded as an important medical event.

Statistical analyses

Descriptive statistics using t-tests or nonparametric tests were used for unadjusted comparisons of continuous data, and chi-square tests were used for categorical data. Time-to-event analyses of treatment patterns and effectiveness over 12 months were conducted using Cox proportional hazard models. Dose escalation was used as a censoring criterion in all analyses. Kaplan–Meier curves of the vedolizumab and infliximab treatment groups were compared using log-rank tests. Unadjusted analyses were also conducted to compare first-line biologic infliximab treatment with second-line biologic infliximab treatment after discontinuation with vedolizumab.

Cumulative rates were calculated for patients with at least one outcome assessment. As this analysis focused on patients receiving the standard dose of treatments, patients who dose-escalated were censored in time-to-event models at the time of escalation. Effectiveness outcomes were assessed at time points of 6, 9, and 12 months. Incidence rates per 1000 person-years of disease exacerbation, disease-related surgery, SAEs, and SIs were calculated as counts of first occurrences for disease by exposure and risk assessment evaluated using adjusted hazard ratios (HRs) with 95% confidence interval (CIs).

To account for differences in baseline characteristics between groups, all adjusted analyses were performed using stabilized inverse probability weighting (IPW) [35]. Age, sex, disease location, disease duration (≤2 years, 2–5 years, 5–10 years, >10 years, and unknown), disease-related hospitalizations, disease severity, steroid dependency status, fistula status, and a composite biochemical marker (FCP, CRP, and albumin) were included in the model as baseline covariates (see Fig. 1 and Supplementary Table 2, Supplemental digital content 2, http://links.lww.com/EJGH/A964 for details on the IPW adjustments). Corticosteroid and other nonbiologic treatment utilization at baseline were controlled in the regression model implemented to develop IPW weights and both comparison groups were balanced on these factors with a standardized difference of less than 10%. The distributions of the propensity scores before and after weighting are shown in Supplementary Fig. 1 (Supplemental digital content 2, http://links.lww.com/EJGH/A964). IPW weights were not trimmed for this analysis. A P value <0.05 was considered statistically significant. All analyses were conducted using SAS version 9.4 software (SAS Institute, Cary, NC, USA).

Fig. 1.

Standardized differences of variables included in the logistic regression model among vedolizumab-treated and infliximab-treated patients with moderate-to-severe CD in the study cohort before (unweighted) and after application of inverse probability weighting (weighted). The composite biochemical marker was based on laboratory assessment data available for the following biochemical disease indicators: fecal calprotectin ≥250 mg/kg, or if unavailable, c-reactive protein (≥5 mg/L), or if unavailable, albumin (<35 g/L). Index event is the day of treatment initiation with vedolizumab or infliximab. CD, Crohn’s disease; GI, gastrointestinal.

Results

Patient population

A total of 454 patients (252 vedolizumab-treated and 202 infliximab-treated patients) were screened and 167 (101 vedolizumab-treated and 66 infliximab-treated patients) were included in the dataset available for analysis. Patients were excluded from the analysis dataset for failure to meet study eligibility criteria (n = 50), including criteria related to patient recruitment into the study (n = 56), not having documented moderate to severe disease (n = 179), or as screening failures identified after study enrollment (n = 2).

Baseline and clinical characteristics

Unadjusted analysis

In the unadjusted analysis, the median (range) duration of follow-up in this study was 16.2 (4.2–45.9) months for the vedolizumab treatment group and 17.8 (6.0–51.0) months for the infliximab treatment group (P = 0.06). Of note, patients included in the analysis (by definition) had ≥6 months between the date of treatment initiation and the date of chart abstraction. Males comprised 51.5% and 42.4% of vedolizumab-treated and infliximab-treated patients, respectively (Table 1); mean (SD) age at baseline was 49.1 (16.4) years and 37.6 (13.0) years for the respective treatment groups (P < 0.01). The vedolizumab treatment group had a higher proportion of patients with disease duration of ≥2 years: 71.1% (64/90) compared with 32.8% (19/58) for infliximab-treated patients (P < 0.01). Of vedolizumab-treated patients with known CD location at index, 22.7% (22/97) had colonic with/without upper GI disease, and approximately equal proportions of the remaining vedolizumab-treated patients had either ileal or ileocolonic disease with/without upper GI involvement. Infliximab-treated patients most frequently had ileocolonic disease with/without upper GI involvement [43.9% (25/57)], followed by colonic or ileal disease with/without upper GI involvement [33.3% (19/57) and 22.8% (13/57), respectively]. Most patients receiving vedolizumab treatment (83.2%) had moderate disease, whereas there was an even distribution of moderate and severe disease among infliximab-treated patients (51.5% moderate and 48.5% severe; difference between treatment cohorts, P < 0.01). The majority of patients from both treatment groups had nonstricturing, nonpenetrating disease, constituting 49 of 90 (54.4%) of the vedolizumab cohort and 29 of 49 (59.2%) of the infliximab cohort. Patients with penetrating disease at baseline comprised 7 of 90 (7.8%) in the vedolizumab treatment cohort and 6 of 49 (12.2%) in the infliximab-treated cohort. Fewer patients in the vedolizumab treatment group had active fistula at index: 3 of 91 (3.3%) compared with 13 of 61 (21.3%) in the infliximab cohort (P < 0.01). The number of patients receiving corticosteroid bridging therapy with available data (n = 41, defined as patients receiving corticosteroids ≤1 month [≤30.5 days] prior to index, at index treatment initiation and discontinuing corticosteroids within 3 months [≤91 days] of index) were 5/20 (25.0%) in the vedolizumab treatment group and 11/21 (52.4%) in the infliximab treatment group. Overall, 21% of vedolizumab-treated patients and 16% of infliximab-treated patients received concomitant immunomodulator treatment after the initiation of vedolizumab or infliximab treatment.

Table 1.

Unadjusted baseline and clinical characteristics of biologic-naive patients with CD treated with vedolizumab or infliximab

Baseline characteristics	Vedolizumab n = 101	Infliximab n = 66	P value
Demographics
Mean (SD) age, years	49.1 (16.4)	37.6 (13.0)	<0.01
Male, n (%)	52 (51.5)	28 (42.4)	0.25
BMI, n with available data	57	43	0.74
<18.5 (underweight), n (%)	3 (5.3)	1 (2.3)
18.5–24.9 (normal), n (%)	18 (31.6)	18 (41.9)
25.0–29.9 (overweight), n (%)	21 (36.8)	14 (32.6)
≥30 (obese), n (%)	15 (26.3)	10 (23.3)
Smoking status, n with available data	89	61	0.23
Current, n (%)	15 (16.9)	6 (9.8)
Former, n (%)	15 (16.9)	17 (27.9)
Never smoked, n (%)	59 (66.3)	38 (62.3)
Clinical characteristics
Disease duration, n with available data	90	58	<0.01
<2 years, n (%)	26 (28.9)	39 (67.2)
2 to <5 years, n (%)	14 (15.6)	8 (13.8)
≥5 years, n (%)	50 (55.6)	11 (19.0)
Median (min–max) observation period, monthsa	16.2 (4.2–45.9)	17.8 (6.0–51.0)	0.06
CD location at index, n with available data	97	57	0.02
Colonic, with/without upper GI disease, n (%)	22 (22.7)	19 (33.3)
Ileal, with/without upper GI disease, n (%)	38 (39.2)	13 (22.8)
Ileocolonic, with/without upper GI disease, n (%)	37 (38.1)	25 (43.9)
Disease severity at index, n with available data	101	66	<0.01
Moderate, n (%)	84 (83.2)	34 (51.5)
Severe, n (%)	17 (16.8)	32 (48.5)
Disease behavior, n with available data	90	49	0.05
Nonstricturing, nonpenetrating, n (%)	49 (54.4)	29 (59.2)
Penetrating, with or without perianal disease, n (%)	7 (7.8)	6 (12.2)
Stricturing, with or without perianal disease, n (%)	34 (37.8)	14 (28.6)
Active fistula at index, n with available data	91	61	<0.01
Active fistula, n (%)	3 (3.3)	13 (21.3)
Prior nonbiologic therapy, n (%)	83 (82.2)	57 (86.4)	0.47
Most common types of prior nonbiologic therapy, n with available data	83	57
Prednisone (CS), n (%)	36 (43.4)	38 (66.7)
Mesalazine (5-ASA), n (%)	34 (41.0)	8 (14.0)
Azathioprine, n (%)	25 (30.1)	19 (33.3)
CS bridging therapyb, n with available data	20	21	0.07
n (%)	5 (25.0)	11 (52.4)
Steroid-dependent, n with available data	89	57	0.21
n (%)	18 (20.2)	7 (12.3)
Composite biochemical marker, n with available data	58	34	0.08
Within normal range, n (%)	15 (25.9)	3 (8.8)
Outside normal range, n (%)	43 (74.1)	31 (91.2)
Prior CD-related surgeries (since diagnosis), n (%)	5 (6.2)	1 (2.4)	0.66
CD-related hospitalizations (12 months prior), n (%)	10 (12.3)	9 (21.4)	0.19

ASA, aminosalicylate; CD, Crohn’s disease; CS, corticosteroids; GI, gastrointestinal.

^aAlthough all patients were required to have 6 months of follow-up, based on the difference between time of treatment initiation to time of data abstraction, some patients were lost to follow-up; therefore, the minimum duration during the observation period was <6 months.

^bPatients were classified as having CS bridging therapy at index treatment initiation if CS started ≤1 month (≤30.5 days) prior to index AND was ongoing at index initiation AND was discontinued within 3 months (≤91 days) following index treatment initiation.

Adjusted analysis

A total of 162 patients were included (99 vedolizumab-treated and 63 infliximab-treated patients) in the logistic regression model and further to develop inverse probability treatment weights. None of the covariates that were included in the IPW model were significantly different between the vedolizumab and infliximab treatment groups, and there were no differences in baseline characteristics between the treatment groups after weighting (Fig. 1 and Supplementary Table 2, Supplemental digital content 2, http://links.lww.com/EJGH/A964).

Effectiveness

For the adjusted analyses, 12-month cumulative rates of clinical response were 75.3% for vedolizumab-treated and 69.2% for infliximab-treated patients (P = 0.77); cumulative rates of clinical remission were 73.1% for vedolizumab and 55.2% for infliximab (P = 0.31); and cumulative rates of mucosal healing were 68.1% for vedolizumab and 65.8% for infliximab (P = 0.76) (Fig. 2). Although rates of these effectiveness endpoints were numerically higher for vedolizumab- compared with infliximab-treated patients with CD, there were no statistical differences between these treatment groups.

Fig. 2.

Adjusted cumulative rates of clinical response, clinical remission, and mucosal healing in biologic-naive patients with Crohn’s disease treated with vedolizumab or infliximab. Effectiveness outcomes were assessed based on a hierarchical algorithm utilizing standard measures of disease activity (see Methods section and Supplementary Table 1, Supplemental digital content 2, http://links.lww.com/EJGH/A964). Annotated data are the rates at each timepoint as indicated by the dashed line. P values are unadjusted log-rank values. ^aThe sum of the patient body weights for each group still receiving treatment at timepoint and at clinical outcome can still be assessed.

Treatment patterns

In the unadjusted analyses, the probability of treatment persistence over 12 months was similar between the vedolizumab and infliximab treatment groups (vedolizumab 91.0% vs. infliximab 72.8%; P = 0.07). The main reasons for discontinuation of index vedolizumab treatment were PNR [10/101 (9.9%)] and SLOR [9/101 (8.9%)]. The main reasons for discontinuation of index infliximab treatment were nonserious AEs [5/66 (7.6%)], SLOR [3/66 (4.5%)], and SAEs [3/66 (4.5%)]. No vedolizumab-treated patients discontinued treatment due to SAEs. Over 12 months, 17 of 101 (16.8%) vedolizumab-treated patients were dose escalated compared with 17 of 66 (25.8%) infliximab-treated patients. The primary reasons for dose escalation in the vedolizumab group were worsening of CD signs/symptoms or disease not being controlled satisfactorily [6/101 (5.9%)], PNR [4/101 (4.0%)], SLOR [4/101 (4.0%)], and adjustment of dose for therapeutic drug monitoring-related purposes [2/101 (2.0%)]. The most frequently documented primary reasons for dose escalation in the infliximab group were worsening of CD signs/symptoms or disease not being controlled satisfactorily [6/66 (9.1%)], adjustment of dose for therapeutic drug monitoring–related purposes [5/66 (7.6%)], PNR [3/66 (4.5%)], and SLOR [3/66 (4.5%)].

Adjusted incidence rate of disease exacerbation (first occurrence per 1000 person-years) was not statistically different between vedolizumab and infliximab [230.2 vs. 421.0, respectively; HR 0.86 (0.42–1.75)] (Fig. 3). For the incidence of surgeries, there was one surgery in the vedolizumab treatment group and none in the infliximab treatment group.

Fig. 3.

Adjusted first incidence of disease exacerbations, serious adverse events (SAEs), and serious infections (SIs): comparison between biologic-naive patients with Crohn’s disease (CD) treated with vedolizumab vs. infliximab. No CD-related surgeries data are shown on the graph as there were no surgeries in the infliximab treatment group and one surgery in the vedolizumab treatment group. Data are the adjusted incidence rates [95% confidence interval (CI)] and hazard ratios (HRs) (95% CI) for disease exacerbations, SAEs, and SIs. Cox proportional hazards model HRs are from adjusted Cox models (number of patients experiencing an event of interest divided by total time in years that the patients were at risk, multiplied by 1000). Models were adjusted for the following covariates: age, sex, disease location, disease duration (≤2 years, 2–5 years, 5–10 years, >10 years, and unknown), disease-related hospitalizations, disease severity, steroid dependency status, fistula status, and a composite biochemical marker using a hierarchical algorithm for the three baseline biochemical disease indicators, based on data availability, with fecal calprotectin (FCP) at the top of the hierarchy [FCP (≥250 mg/kg) → C-reactive protein (≥5 mg/L) → albumin (<35 g/L)].

Safety

Unadjusted results showed that overall, four SAEs occurred in the vedolizumab group and seven occurred in the infliximab group. Each type of SAE had a single occurrence in the vedolizumab group except malignancy [2/101 (2.0%)]. The most frequently documented SAEs in the infliximab group were GI infections [2/66 (3.0%)] and lower respiratory tract infections [2/66 (3.0%)] (see Supplementary Table 3, Supplemental digital content 2, http://links.lww.com/EJGH/A964 for additional information on SAEs and SIs). In the adjusted analysis, incidence (first occurrence per 1000 person-years) of SAEs was 43.6 vs. 200.9 for vedolizumab- vs. infliximab-treated patients, respectively, and incidence of SIs was 10.8 vs. 96.0 in vedolizumab vs. infliximab treatment cohorts, respectively (Fig. 3).

Effectiveness outcomes in patients with complicated and noncomplicated disease

A subgroup analysis was conducted in patients with complicated CD (n = 57) or noncomplicated disease (n = 109) at baseline, respectively, in vedolizumab-treated patients (n = 28; n = 73) or infliximab-treated patients (n = 29; n = 36). There were no differences in 12-month effectiveness outcomes between vedolizumab and infliximab treatment groups for patients with complicated and noncomplicated disease (Fig. 4). In patients treated with vedolizumab, the rate of clinical response (59.3% vs. 73.6%; P = 0.41), clinical remission (43.9% vs. 69.6%; P = 0.08), and mucosal healing (40.1% vs. 67.1%; P = 0.08) were similar in subgroups of patients with complicated vs. noncomplicated disease. The rate of treatment persistence (90.9% vs. 86.2%; P = 0.45) was also similar across these disease subgroups in vedolizumab-treated patients. No differences were found for these outcomes in infliximab-treated patients with complicated vs. noncomplicated disease. The likelihood of clinical response, remission, mucosal healing, and treatment persistence in patients with noncomplicated disease was similar in both treatment groups (Table 2 and Fig. 4).

Fig. 4.

Adjusted cumulative rates of clinical response, clinical remission, mucosal healing, and treatment persistence for vedolizumab- vs. infliximab-treated patients in the noncomplicated and complicated disease subgroups. Vedolizumab cohort, left-hand graphs; infliximab cohort, right-hand graphs. The complicated disease cohort was defined as patients with penetrating Crohn’s disease (CD) and included patients with a mixed inflammatory/stenotic phenotype. Noncomplicated CD was defined as patients who had no active fistula at index treatment initiation, no CD-related surgeries since diagnosis, and no CD-related hospitalizations within 12 months prior to treatment initiation. Effectiveness outcomes were assessed based on a hierarchical algorithm utilizing standard measures of disease activity (see Methods section and Supplementary Table 1, Supplemental digital content 2, http://links.lww.com/EJGH/A964). Treatment persistence was defined as: did not discontinue index treatment for any reason during the study follow-up period. C, complicated disease. ^aNumber of patients still receiving treatment at clinical outcome can still be assessed.

Table 2.

Comparison of outcomes in patients with complicated and noncomplicated disease^a treated with vedolizumab vs. infliximab

	Infliximab vs. vedolizumab, point estimate (95% Wald confidence limits)
	Complicated	Noncomplicated
Clinical response	2.086 (1.084–4.096)	1.224 (0.728–2.057)
Clinical remission	2.188 (1.025–5.670)	1.097 (0.630–1.912)
Mucosal healing	1.862 (0.708–4.897)	1.499 (0.796–2.823)
Treatment persistence	2.629 (0.510–13.557)	2.120 (0.816–5.506)

^aComplicated disease cohort was defined as patients with penetrating Crohn’s disease (CD) and included patients with a mixed inflammatory/stenotic phenotype. Noncomplicated CD was defined as patients who had no active fistula at index treatment initiation, no CD-related surgeries since diagnosis, and no CD-related hospitalizations within 12 months prior to treatment initiation. See Fig. 3 for patient numbers in the subgroups of vedolizumab- or infliximab-treated patients with complicated and noncomplicated disease. Effectiveness outcomes were assessed based on a hierarchical algorithm utilizing standard measures of disease activity (See Methods and Supplementary Table 1, Supplemental digital content 2, http://links.lww.com/EJGH/A964).

Safety outcomes in patients with complex and noncomplicated disease

In both the complicated and noncomplicated disease groups, the likelihood of experiencing AEs was significantly lower in vedolizumab-treated patients vs. infliximab-treated patients. For AEs, incidence rate of first occurrence per 1000 person-years was 131.6 vs. 732.2 [HR 0.19 (0.05–0.65)] for the complicated disease group and 276.3 vs. 494.8 [HR 0.59 (0.35–0.99)] for the noncomplicated disease group for vedolizumab-treated vs. infliximab-treated patients, respectively (Fig. 5).

Fig. 5.

Incidence of adverse events for vedolizumab- vs. infliximab-treated patients in noncomplicated and complicated disease subgroups. ^aDenotes a significant difference between cohorts. Data are the unadjusted incidence rates and adjusted hazard ratios (HRs) (95% confidence interval). Cox proportional hazards model HRs are from adjusted Cox models (number of patients experiencing an adverse event of interest divided by total time in years that the patients were at risk, multiplied by 1000).

Discussion

Findings from this comparative analysis of real-world effectiveness and safety data in biologic-naive patients with moderate-to-severe CD suggest no differences between vedolizumab and infliximab in effectiveness outcomes of clinical response, clinical remission, mucosal healing and treatment persistence of over 12 months. Disease exacerbations and the incidence of surgeries were low in both treatment groups in this study.

Effectiveness results are in line with data from other retrospective, real-world studies [24,32], including propensity score–matched observational data from anti-TNFα–naive patients in the VICTORY consortium (no significant difference between vedolizumab and infliximab for achieving clinical remission, steroid-free clinical remission, or endoscopic remission) [24] and the main EVOLVE study (including real-world data for other anti-TNFα treatments as well as infliximab) [27].

Rates of treatment persistence analyzed in this study showed a trend for being higher in vedolizumab vs. infliximab treatment groups over time, but this did not reach statistical significance at 12 months, which may have been a reflection of restricted sample size in this unadjusted analysis. Other retrospective studies have showed higher treatment persistence at 12 months and at later timepoints in biologic-naive patients with CD treated with vedolizumab vs. infliximab [36,37–39], and vs. other anti-TNFα treatments [39–41]. In this study, there may have been other factors affecting persistence that were not considered (e.g. patient decision or reimbursement factors).

Similarly, no differences were found for effectiveness outcomes analyzed across treatment groups for patients with complicated vs. noncomplicated disease.

Regarding safety, in both the adjusted and unadjusted analyses of SAE and SI frequency, the HRs suggested a difference between the vedolizumab and infliximab treatment groups: incidence of SAEs [HR 0.36 (0.09–1.54)] and SIs [HR 0.08 (<0.01–2.64)] for adjusted analysis. However, the relatively small sample size and few events resulted in widened 95% CIs and an absence of a demonstrable statistical difference between treatments. These results were consistent with the VICTORY observational cohort analysis of 1266 patients who received vedolizumab or anti-TNFα treatment for CD (n = 659) between May 2014 and December 2017; there were no significant differences between vedolizumab and anti-TNFα treatment cohorts with regard to risk of SAEs [odds ratio (OR) 0.751; 95% CI, 0.519–1.806] or SIs (OR 1.183; 95% CI, 0.786–1.795), although rates of noninfectious SAEs were significantly lower with vedolizumab vs. anti-TNFα treatment (OR 0.072; 95% CI, 0.012–0.242) [24].

In this study, the unadjusted analysis of treatment effect in subgroups of patients with complicated or noncomplicated disease at 12 months found that vedolizumab-treated patients had a significantly lower likelihood of experiencing AEs than infliximab-treated patients in both subgroups. In the main EVOLVE study, the incidence of SAEs and SIs was lower in vedolizumab-treated vs. anti-TNFα–treated patients (HR 0.42, 95% CI, 0.28–0.62 for SAEs; HR 0.40, 95% CI, 0.19–0.85 for SIs) [27].

There are several limitations associated with the retrospective nature of this study; these include, most importantly, that patients were not randomized. Although use of weighted analyses controlled for possible confounding factors (including differences in age, disease duration, and disease severity between treatment cohorts, as identified in the unadjusted data), the possibility of unknown factors with the potential to bias the results cannot be excluded. For example, corticosteroids and other nonbiologic treatments were included in the logistic regression model to generate propensity scores and IPWs, and Table 1 indicates that there was not a statistically significant difference between treatment groups in terms of prior nonbiologic therapy. However, the analysis was adjusted for the use of corticosteroids and other nonbiologic treatments at baseline only and not during the follow-up period. There were also difficulties in distinguishing between PNR and SLOR in the study, reflecting the differences in dose escalation treatment practices and restrictions across countries. In addition, effectiveness outcomes used to compare treatments were not prespecified or collected systematically, and only vedolizumab-treated patient data was collected from Greece; not collecting infliximab treatment data from all sites is a limitation to the comparative analysis. It is also possible that bias was introduced by only including patients with available data in the analysis. Another limitation is that the study was not specifically powered for the comparison of effectiveness outcomes between treatment groups. The SPIRIT guidelines on optimal endpoints for measuring disease progression in CD were not available at the time that the EVOLVE study protocol was developed [42]. Nevertheless, comparative effectiveness and safety data derived from real-world studies such as EVOLVE [27] can provide a large amount of retrospective data that are useful for optimizing the positioning of vedolizumab in the sequence of advanced therapies available for moderate-to-severe CD. The restrictions applied to extracting the data allow for definition of clear-cut endpoints and criteria for remaining in the study. Analysis of treatment outcomes achieved with vedolizumab compared with anti-TNFα treatments in biologic-naive patients from EVOLVE has shown that the use of anti-TNFα biologics after vedolizumab does not impact the effectiveness of anti-TNFα treatment. In contrast, vedolizumab is less effective after anti-TNFα treatment [43]. This dataset has facilitated analysis of the relative benefit of vedolizumab as a first-line advanced therapy, and has contributed to the limited data collected in complicated and noncomplicated disease populations [27].

Results from the EVOLVE dataset, including this study, strengthen existing evidence that the clinical efficacy/effectiveness of vedolizumab is more pronounced in patients with moderate to severe CD without prior anti-TNFα exposure. This was true for clinical remission (CDAI score of ≤150) after induction in patients from the phase III GEMINI 3 trial [44], and for mucosal healing outcomes from the VERSIFY phase IIIb study, in which anti-TNFα–naive patients were more likely to achieve endoscopic remission and complete healing than those who had failed on anti-TNFα treatment [45]. Evidence-based guidelines have also suggested that vedolizumab could be useful as a first-line advanced therapy in patients with moderate to severe CD [9,10,46]. Clinical response, examined retrospectively from real-world data in EVOLVE, suggested that biologic-naive patients with noncomplicated CD were more likely to achieve a clinical response with vedolizumab at 12 months than patients who failed on anti-TNFα treatment [47].

When patients with moderate-to-severe CD who have failed on standard therapy are escalated to advanced therapy, the choice of treatment is important to optimize long-term treatment outcomes for individual patients [7,14–16]. The results from this analysis of real-world effectiveness and safety data in biologic-naive patients with CD over 12 months suggest that there were no differences in effectiveness outcomes between vedolizumab and infliximab treatment groups, and that the tolerability profile of vedolizumab, analyzed separately in noncomplicated and complicated disease, is likely to be more favorable. These results add to available evidence supporting the use of vedolizumab as a first-line advanced therapy for patients with CD.

Acknowledgements

This study was sponsored by Takeda. Editing support was provided by Excel Medical Affairs, and funded by Takeda.

G.J.M., B.B., S.A., M.L., P.K., S.W., and A.Y. contributed to the conception and design of the study; G.J.M., B.B., and A.Y. were involved in the data collection; G.J.M., B.B., S.A., C.C., N.R.B., S.S., P.K., S.W., and A.Y. performed the data analysis; and all authors interpreted the data, drafted/revised the manuscript critically for intellectual content, and gave final approval for it to be submitted.

The datasets, including the redacted study protocol, redacted statistical analysis plan, and individual participants’ data supporting the results reported in this article, will be made available within 3 months from initial request to researchers who provide a methodologically sound proposal. The data will be provided after its de-identification, in compliance with applicable privacy laws, data protection, and requirements for consent and anonymization.

Part of the work was presented as a poster at the European Crohn’s and Colitis Organisation Congress, 12–15 February 2020, Vienna, Austria, and as a poster presented at Digestive Disease Week, 2–5 May 2020, virtual congress.

Conflicts of interest

G.J.M. reports being an advisor/speaker for AbbVie, Aenorasis, Angelini, Celgene, Celltrion, Dr Falk Pharma, Ferring, Hospira, Janssen, Merck Sharp & Dohme, MYLAN, Pfizer, Takeda, and Vianex; and research support from AbbVie, Ferring, Galenica, Genesis, Merck Sharp & Dohme, MYLAN, Takeda, and Vianex. B.B. reports being an advisor/speaker for AbbVie, Bristol Myers Squibb, Ferring, Janssen, Merck, Novartis, Pfizer, and Takeda; an advisor for Alimentiv, Allergan, Amgen, AMT, Bristol Myers Squibb, Celgene, Fresenius Kabi, Gilead, Genentech, Iterative Scopes, Microbiome Insights, Merck, Mylan, Pendopharm, and Protagonist; research support from AbbVie, Alvine, Amgen, Boehringer Ingelheim, Bristol Myers Squibb, Celgene, GlaxoSmithKline, Genentech, Janssen, Merck, and Qu Biologic; and stock options in Qu Biologic. S.A., S.S., P.K., and S.W. are employees of and hold stock/stock options in Takeda. C.C. and N.R.B. are employees of Evidera, which received funding from Takeda. M.L. is a former employee of Takeda and holds stock/stock options in Takeda. A.Y. reports being on advisory boards for Arena, Bristol Myers Squibb, Procise, Prometheus Laboratories, and Takeda; and speaker bureau for Bristol Myers Squibb.

Supplementary Material

ejgh-36-281-s001.pdf ^{(381.5KB, pdf)}