Systematic review with meta-analysis: real-world effectiveness and safety of vedolizumab in patients with inflammatory bowel disease

Background Selective patient recruitment can produce discrepancies between clinical trial results and real-world effectiveness. Methods A systematic literature review and meta-analysis were conducted to assess vedolizumab real-world effectiveness and safety in patients with ulcerative colitis (UC) or Crohn’s disease (CD). MEDLINE, MEDLINE In-Process, EMBASE, and Cochrane databases were searched for real-world studies of vedolizumab in adult patients with UC/CD reporting clinical response, remission, corticosteroid-free remission, UC/CD-related surgery or hospitalization, mucosal healing, or safety published from May 1, 2014–June 22, 2017. Response and remission rates were combined in random-effects meta-analyses. Results At treatment week 14, 32% of UC patients [95% confidence interval (CI) 27–39%] and 30% of CD patients (95% CI 25–34%) were in remission; and at month 12, 46% for UC (95% CI 37–56%) and 30% for CD (95% CI 20–42%). For UC, the rates of corticosteroid-free remission were 26% at week 14 (95% CI 20–34%) and 42% at month 12 (95% CI 31–53%); for CD they were 25% at week 14 (95%, CI 20–31%) and 31% at month 12 (95%, CI 20–45%). At month 12, 33–77% of UC and 6–63% of CD patients had mucosal healing. Nine percent of patients reported serious adverse events. Conclusions Vedolizumab demonstrated real-world effectiveness in patients with moderate-to-severely active UC or CD, with approximately one-half and one-third of patients, respectively, in remission at treatment month 12. These findings are consistent with clinical trial data and support the long-term benefit–risk profile of vedolizumab. Electronic supplementary material The online version of this article (10.1007/s00535-018-1480-0) contains supplementary material, which is available to authorized users.

Strict inclusion criteria used in randomized controlled trials (RCTs) can limit the patient population and generalizability of trial results to clinical practice, with the latter further compromised by IBD patient heterogeneity [11,12]. Indeed, up to two-thirds of patients with IBD might be ineligible to participate in RCTs of biologics [11,13]. An additional hindrance is the increasing unwillingness of patients to accept placebo control. Randomized controlled trials are, therefore, unlikely to fully represent the real-world IBD population. However, physicians require real-world effectiveness data to complement clinical trial results and inform treatment decisions. Assessing the treatment quality and effect size in clinical practice and evaluating the strength of this evidence through systematic literature reviews and meta-analyses can provide such data. Summation can overcome potential bias associated with individual studies and address challenges associated with the transferability of RCT findings; systematic literature reviews are, therefore, at the top of the evidence hierarchy as defined by the Oxford Centre for Evidence-Based Medicine [14,15].
Data from the GEMINI 3 trial suggest that the full effect of vedolizumab-induced clinical remission in patients with CD may not be apparent before treatment week 10 [7]. The European Summary of Product Characteristics and US Prescribing Information both recommend that vedolizumab treatment of UC and CD should be discontinued if a therapeutic benefit is not observed by week 14 (by week 10 in UC in Europe) [8,9]. Real-world data allow evaluation of the optimal time points for assessing clinical effectiveness and when concomitant therapies should be adjusted based on therapeutic response outside of RCT protocoldefined assessments.
Given that vedolizumab is relatively new and the number of treated patients is increasing, ongoing safety monitoring is essential. Real-world data from large cohorts can further characterize a drug's safety profile not fully elucidated in clinical trials [16,17]. We sought to systematically review and summarize published literature on real-world effectiveness and safety of vedolizumab studies and conduct a meta-analysis of effectiveness data.

Study selection
A systematic review of MEDLINE, MEDLINE In-Process, EMBASE and Cochrane (May 1, 2014-June 22, 2017), and searches of clinicaltrials.gov and the World Health Organization International Clinical Trials Registry Platform were completed. Conference proceedings from 2015 to June 2017 were searched. Two researchers reviewed relevant publications independently, with disagreements resolved by discussion or a third reviewer. Studies were eligible if they included real-world evidence (e.g., medical record review, database, registry) and an adult patient population (C 18 years when initiating vedolizumab) receiving vedolizumab (Takeda Pharmaceuticals International, Inc., Deerfield, IL) for IBD (UC, CD, or unspecified/ indeterminate colitis) and if outcomes reported were of interest. English and non-English language studies were eligible for inclusion. Studies were excluded if the total patient population was \ 10, if vedolizumab was used offlabel, or if safety data were reported at event level only (no denominator). Investigators were contacted for unpublished data (unpublished clinical data provided courtesy of Dr. Mark A. Samaan and Dr. Peter Irving from their UK study, 2016) and conference abstracts, and manual backward citation tracking of references (including studies) were performed to identify additional relevant studies [18][19][20][21][22].

Data extraction and outcome measures
One researcher used predefined parameters to extract all data using a piloted form and, after this, a second researcher performed data checks for accuracy. Information obtained for each eligible study included author, year of publication, geographic location, and clinical outcomes reported. Patient characteristics included disease duration, age, sex, prior medication history, and IBD-related surgeries. The primary outcome measure was clinical remission; secondary outcome measures were clinical response, CS-free clinical remission, mucosal healing, endoscopic improvement, surgery and hospitalization rates, doseescalation rates, and safety. Clinical response, clinical remission, and CS-free clinical remission rates (classified according to summarized measures in Table S1) (Samaan and Irving, 2016) [13, were collected at weeks 6, 14, 26-30 (month 6), and 46-54 (month 12), where available. Subanalyses were performed to determine clinical remission rates by geographic region and in patients who were anti-TNFa-naive.

Grading of evidence
Studies were assessed using the Oxford Centre for Evidence-Based Medicine 2011 Levels of Evidence, which evaluates the strength of evidence (including quality and bias) based on study design [15,49]. One reviewer appraised each study and assigned a level from 1 (high quality or low risk of bias) to 5 (low quality or high risk of bias) (Table S2) [15,49], with uncertainty resolved by discussion with a second reviewer.

Statistical analyses
Meta-analyses were performed to combine clinical response, clinical remission, and CS-free clinical remission rates using R statistical software (version 3.2.2; R Foundation; Vienna, Austria) with the ''meta'' package (version 4.3-2). When multiple publications were available for a study, data from the most recent cohort were used for the combined analyses. Weighted mean clinical response, clinical remission, and CS-free clinical remission rates and corresponding 95% confidence intervals (CIs) were calculated using the DerSimonian-Laird random-effects model to account for between-study heterogeneity [50]. Where mean/median values and 95% CIs were reported, data were used as stated (calculated using the binomial distribution if 95% CIs were not reported). For studies reporting safety, IBD-related surgery or hospitalization, and dose escalation, the proportion of events was calculated. For mucosal healing or endoscopic improvement, analyses were based on either a cumulative incidence approach or as a proportion of those receiving endoscopy.
Study heterogeneity was determined using the I 2 statistic (which describes the variability in the effect estimate that results from heterogeneity rather than sampling error [51]) and the Q-statistic (P \ 0.05 was considered significant and suggested statistical heterogeneity). When C 10 studies reporting the same outcome were available, publication bias was assessed using Egger's weighted regression statistic, with P \ 0.05 suggesting a higher likelihood of bias [52].

Discussion
To the best of our knowledge, to date this is the most comprehensive meta-analysis of real-world clinical response and remission rates for vedolizumab over 12 months of treatment, incorporating data from both peerreviewed full-text manuscripts and abstracts. Real-world effectiveness data provide valuable evidence to support the efficacy observed in RCTs, because trial patients may not be representative of the real-world IBD population [11].
In UC, clinical remission was achieved in approximately one-third of patients at 14 weeks and in approximately one-half of patients at 12 months. In CD, clinical remission was achieved by approximately one-third of patients at both 14 weeks and 12 months. An important treatment goal in the management of patients with IBD is the achievement and maintenance of sustained CS-free clinical remission [1,2,111]. Approximately, one-quarter of patients with UC or CD achieved CS-free clinical remission at 14 weeks and 42% of patients with UC and 31% of patients with CD at 12 months. As patients comprising the 12-month cohort likely represent the earliest vedolizumab users, they could represent a more severe, treatment-refractory cohort (most are likely to have failed anti-TNFa treatment). According to RCT experiences, greater effectiveness should be achieved in biologic-naive patients. In a real-world setting, this trend could induce higher efficacy rates with continued and earlier use of the drug. Also, in our study, up to approximately one-third of patients with CD achieved clinical remission after week 14, suggesting potential benefits of therapeutic monitoring beyond this time point. Despite including patients with more complex disease versus RCTs, real-world clinical and CS-free clinical remission rates in UC and CD reported here are consistent with, and in some cases exceed, vedolizumab efficacy reported in the GEMINI trials [5][6][7]. Moreover, the findings suggest a similar treatment effect in UC and CD, despite including patients with more complex disease versus RCTs.
Subgroup analyses in UC and CD biologic-naive patients receiving vedolizumab demonstrated substantially b Fig. 2
In the current study in both UC and CD, Swedish cohorts had higher clinical remission rates, whereas cohorts in Germany and Spain had lower remission rates. The differences in remission rates based on geography need to be interpreted with caution, however, because of the small number of studies in this analysis. Several characteristics of IBD patients, including epidemiology, phenotype, and genotype, are known to vary with geography [117]. Geographic differences in study population baseline characteristics [e.g., disease severity at vedolizumab initiation, disease duration, prior anti-TNFa use (and number of prior therapies)], national treatment guidelines, and IBD management patterns may also account for variations in remission rates across geographic locations in our study. This is an area worthy of further investigation, but is beyond the current analysis.
Five publications included in the current review reported on hospitalization rates both pre-vedolizumab (6-12 months before initiation) and post-vedolizumab (6 months after initiation); 4 studies [63,100,101,110] reported a reduction in post-treatment hospitalization rates, whereas 1 study [99] reported no change in hospitalization rates. Furthermore, a recent study in biologic-naive patients (published after the prespecified date range for this review) reported lower rates of IBD-related surgery and hospitalizations at 6 and 12 months after the first infusion of vedolizumab compared with infliximab [62]. Additional studies on the long-term effects of vedolizumab treatment on hospitalization rates are warranted.
Mucosal healing is an important IBD therapy goal associated with sustained clinical remission, CS-free clinical remission, and reduced hospitalization and surgery rates [118,119]. Recent ''treat-to-target'' draft clinical guidelines state that only patients with mucosal healing (absence of macroscopic signs of active inflammation) and no/very mild signs or symptoms should be considered as remitted [120]. Among larger studies (sample size C 100) in our systematic review, more than half of patients with UC or CD achieved mucosal healing at 12 months; results for UC were better than reported in GEMINI 1 [5]. Although data were limited to 12 studies, the observed rates of mucosal healing over 12 months were greater than the combined rates of clinical remission, supporting previous reports of a lack of clear correlation between clinical symptom measures and bowel damage assessed by endoscopy/colonoscopy or diagnostic imaging modalities [121,122]. Interim results from the LOVE-CD trial demonstrated that, of 74 patients who underwent endoscopy, endoscopic remission (defined as Simple Endoscopic Score for CD B 3) was observed in 30% of patients at week 26 [123]. Patients with endoscopic response were shown to have higher median vedolizumab concentrations compared with endoscopic nonresponders [123]. Results from the phase 3b, open-label, VERSIFY study (NCT02425111) will provide additional insights into rates of mucosal healing in CD patients receiving vedolizumab (manuscript in progress) [124].
Dose escalation is used to address secondary loss of response to biologics in the clinical management of IBD [125].
The studies included here (n = 8) [23,33,37,41,46,57,75,108] reported that 4-60% of patients required dose escalation up to week 54, with lower rates reported in biologic-naive patients (n = 2; range 0-20%). However, the highest rates of dose escalation (47-60%) were observed in more complex, treatment-refractory UC and CD patients who were included as part of a compassionate-use program [23] and thus are unlikely to be representative of the general IBD population receiving biologics. In 2 studies, dose-escalation rates were lower with vedolizumab than with anti-TNFa agents [57,75]. Of the few studies reporting dose-escalation outcomes (N = 4), at least one-third of patients were able to recapture response [33,37,46,108].
A meta-analysis of 9 studies comprising 1565 adult patients with UC or CD was recently published by Engel and colleagues [126]. Investigation of rates of clinical remission, clinical response, CS-free clinical remission, and safety demonstrated that vedolizumab is efficacious in UC and CD and has a favorable safety profile. Our study results corroborate their findings. The overall AE rate b Fig. 3 Meta-analysis of CS-free clinical remission rates among patients with ulcerative colitis receiving vedolizumab at the time points: a week 6; b week 14; c 6 months; and d 12 months.

Patients in remission
Patients assessed d reported by Engel and colleagues was 30.6% (6 studies, n = 306) compared with a rate of 0-67% (23 studies, n = 2358) in the current study. Nasopharyngitis and arthralgia were among the most common AEs reported in both meta-analyses.
A notable point of differentiation between the current study and the report by Engel and colleagues is the comprehensiveness of the current study with inclusion of not only full-text articles but also congress abstracts, thus allowing for additional effectiveness data to be assessed at relevant treatment time points. The thorough and inclusive approach adopted for the current review facilitated b Fig. 4   Crohn's disease. The size of each data bubble is proportional to the corresponding country clinical remission rates. Unless specified otherwise, one study was reported at each geographical location. Random-effects meta-analysis of single proportions was used to calculate an overall proportion in cases of [ 1 study [5][6][7][127][128][129][130]. Real-world safety data reported here are consistent with those from the GEMINI trials, with no new or unexpected safety signals [128]. This tolerability profile may help to improve treatment persistence [115,116], thereby potentially positively affecting long-term outcomes. Postoperative complication rates in the current analysis ranged from 13 to 65% [89,102,[131][132][133]. A recent meta-analysis assessing the impact of preoperative vedolizumab treatment on the rate of postoperative complications in real-world patients with IBD demonstrated no increased risk of postoperative infectious or total overall postoperative complications compared with either preoperative anti-TNFa therapy or no biologic therapy [134].
In addition to the limitations of real-world studies, the limitations of this meta-analysis include potential publication bias. Egger's weighted regression statistic was calculated for only 1 analysis (CD at week 14) and in this case the P value suggested that bias was unlikely. The remaining analyses did not include enough studies (C 10) to allow an assessment of publication bias [52]. However, the inclusion of studies published as abstracts as in the current analysis may help minimize the risk of publication bias. A moderate to high degree of between-study statistical heterogeneity was detected in some analyses. Major contributory factors to this heterogeneity may include the different disease activity measures and variable thresholds used to assess clinical response and remission, which may impact the extrapolation of these findings to clinical practice. Nevertheless, the current meta-analysis attempted to address bias by combining study data using a weighted average based on sample size. Moreover, the consistency of evidence levels (i.e., most studies were level 4) did not allow for sensitivity analysis to be conducted by study quality. Finally, real-world data may be less stringent than RCT data, which are obtained by rigorous data collection and quality control of data integrity. However, real-world data provide greater insight into the effectiveness of vedolizumab in heterogenous and more complex patient populations that are more representative of clinical practice.

Conclusions
The results from this meta-analysis of real-world data confirm the effectiveness of vedolizumab in inducing longterm clinical response, clinical remission, CS-free clinical remission, and mucosal healing in patients with moderateto-severely active UC or CD. The safety data presented here support the positive long-term benefit-risk profile of vedolizumab in the treatment of IBD.
Author contributions DD and JMK: study concept and design; DD: conduct of literature review (with PHMR Limited-acknowledged); GH: analysis of data; SS, AD, LPB, MM, RC, and EVL: interpretation of data and adding clinical context; DD and JMK: drafting of the manuscript. All authors critically reviewed and revised the manuscript for important intellectual content and approved the final version of the manuscript before submission.
Funding Takeda sponsored medical writing assistance, which was provided by Khalid Siddiqui of Chameleon Communications International Ltd, UK (a Healthcare Consultancy Group Company), Rezan Sahinkaya, PhD, and Vinay Pasupuleti, PhD, of ProEd Communications, Inc. Authors from Takeda were involved in the statistical analysis and interpretation of data, which led to co-authorships according to the level of contribution.

Compliance with ethical standards
Conflict of interest Stefan Schreiber has received honoraria from AbbVie, Celltrion, MSD, Pfizer, and Takeda. Axel Dignass has no conflict of interest. Laurent Peyrin-Biroulet has received lecture fees from MSD, Abbvie, Janssen, Takeda, Celltrion, Pfizer, BMS, Pharmacosmos, Shire, Genentech, Mitsubishi, Ferring, Norgine, Tillots, Vifor, UCB Pharma, Hospira, Boehringer-Ingelheim, and Lilly. Greg Hather owns stock in Takeda Pharmaceuticals and is an employee of Takeda Global Research and Development. Dirk Demuth is an employee of Takeda International, UK Branch. Mahmoud Mosli has no conflict of interest. Rebecca Curtis is a former employee of Takeda International, UK Branch. Javaria Mona Khalid owns stock in Takeda Pharmaceuticals International and is an employee of Takeda International, UK Branch. Edward V. Loftus, Jr, has received research support from and has served as a consultant for Takeda.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://crea tivecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.