Effects of vedolizumab in Japanese patients with Crohn’s disease: a prospective, multicenter, randomized, placebo-controlled Phase 3 trial with exploratory analyses

The incidence of Crohn’s disease (CD) is lower in Asian countries than in Western ones. However, CD prevalence in Japan has rapidly increased over recent decades, from 5.9 per 100,000 people in 1991 to 30.1 per 100,000 people in 2013 [1]. As no curative treatment for CD has been established, therapeutic strategies focus on control of disease activity via induction and maintenance therapy, and measures to improve quality of life [2]. These strategies include: treatment with 5-aminosalicylic acid, immunomodulators, steroids, or biologics (e.g., anti-TNFα); surgery; and nutrition therapy [2-4]. Notably, although anti-TNFα is generally effective, there have been reports of safety concerns and primary loss or loss of response [5, 6].

Vedolizumab is a humanized immunoglobulin G1 monoclonal antibody that binds exclusively to α₄β₇ integrin, blocking its interaction with mucosal addressing cell adhesion molecule-1 and thereby specifically modulating gut lymphocyte trafficking [7]. In the GEMINI 2 (assessing outcomes of vedolizumab induction and maintenance therapies in patients with active CD for whom ≥ 1 prior CD therapies had failed) and GEMINI 3 trials (assessing outcomes of vedolizumab induction therapy in active CD, focusing on patients with previous anti-TNFα failure), vedolizumab demonstrated efficacy in moderate-to-severe CD, leading to wide approval for the treatment of CD [8, 9]. As Japanese patients were not included in GEMINI 2 or 3, efficacy and safety of vedolizumab in this population have not yet been evaluated.

Genetic backgrounds and phenotypes of inflammatory bowel disease differ considerably between Asian and Western patients [10, 11]; it is therefore important to determine treatment targets, outcomes, and responses in populations with different genetic backgrounds [12]. Multivariable analysis of GEMINI 2 data identified that ethnicity was a predictive factor of remission (American Indian or Alaskan Native, Asian, Black, and Native Hawaiian or other Pacific Islander vs non-Hispanic/Latino; odds ratio [OR] and 95% confidence interval [CI], 0.30 [0.12–0.75]) [13], and a post hoc analysis of data from GEMINI 2 and 3 also showed that race (non-White vs White) tended to predict rapid response to vedolizumab at Week 2 [14]. Therefore, it is meaningful to evaluate vedolizumab efficacy and associated predictive factors in patients with CD who are of different race to the more widely studied Western patient population.

Here, we evaluated efficacy, safety, pharmacokinetics and immunogenicity of vedolizumab versus placebo as induction and maintenance therapy in Japanese patients with moderate-to-severe CD. Additionally, as subgroup or post hoc exploratory analyses, we assessed factors affecting efficacy and time courses of disease activity, to contribute to the improvement of vedolizumab treatment in clinical practice.

Mean CDAI scores decreased from baseline for both treatment groups at all measured time points during the induction phase (Fig. 3). However, the mean change from baseline in CDAI score for the vedolizumab group was greater than the placebo group at each measured time point (Fig. 3a). The difference in mean change from baseline in CDAI score between vedolizumab and placebo groups was greater in the subgroup without prior anti-TNFα use (Fig. 3b, − 104.6 ± 82.7 vs − 36.6 ± 87.2 at Week 10) than in that with prior anti-TNFα use (Fig. 3c, − 31.7 ± 109.8 vs − 22.6 ± 82.4 at Week 10). Mean CDAI score in the vedolizumab group at Week 10 was 166.0 ± 82.9 in the subgroup without prior anti-TNFα use (Fig. 3b).

Post hoc subgroup analysis showed a numerically greater difference in mean change from baseline in CDAI score between treatment groups in the subgroup with inadequate response to prior anti-TNFα (Fig. 3d, − 49.0 ± 70.0 vs  13.0 ± 35.1 at Week 10) versus the subgroup with loss of response to prior anti-TNFα (Fig. 3e,− 28.6 ± 121.5 vs − 37.6 ± 86.4 at Week 10).

Fig. S4 shows post hoc analysis of CDAI score stratified by CRP level at baseline (a cutoff of 1.6 mg/dL was used to ensure similar patients number between subgroups). In the subgroup with baseline CRP > 1.6 mg/dL, change in CDAI score at Week 10 from baseline was − 57.5 ± 116.8 in the vedolizumab group and − 5.3 ± 90.8 with placebo (Fig. S4a); no notable difference was observed in the subgroup with baseline CRP ≤ 1.6 mg/dL. In the subgroup with baseline CRP > 1.6 mg/dL, mean CDAI score in the vedolizumab group reached the lowest level at Week 6 (256.0 ± 98.2). A similar trend was observed in CDAI subscores for abdominal pain (Fig. S4b, − 19.1 ± 35.3 vs − 0.7 ± 20.4 at Week 10) and number of liquid or very soft stools (Fig. S4c, − 12.9 ± 40.2 vs − 2.1 ± 34.2).

The change over time in CRP concentration during the induction phase in patients with baseline CRP concentration > 0.30 mg/dL is shown in Fig. S5. From baseline to Week 10, median (1st quartile, 3rd quartile) CRP concentration continuously decreased by 0.295 (− 1.600, 0.550) mg/dL in the vedolizumab group (n = 60) and increased by 0.220 (− 0.900, 1.390) mg/dL in the placebo group (n = 59).

This is the first randomized, placebo-controlled, Phase 3 study of vedolizumab in Japanese patients with moderate-to-severe CD. CDAI-100 response rate at Week 10 was numerically higher in the vedolizumab group than in the placebo group, although the difference between treatment groups was not statistically significant. The timing of the primary endpoint assessment was changed from Week 6 in GEMINI 2 and 3 to Week 10 in our study because differences in clinical outcomes between vedolizumab and placebo groups were further increased at the Week 10 timepoint in GEMINI 2 and 3 [9, 18]. In patients without prior anti-TNFα use, CDAI-100 response rates for vedolizumab and placebo at Week 10 were comparable to those in GEMINI 3 (50.0% vs 25.0% in our study; 51.0% vs 22.0% in GEMINI 3), which had a similar design with approximately three quarters of patients with prior anti-TNFα exposure, and similar baseline CDAI score to our study [9]. Furthermore, CR rate at Week 10 in our study was higher than that in GEMINI 3 (50.0% vs 12.5% in our study; 28.7% vs 13.0% in GEMINI 3). However, CDAI-100 response rate at Week 10 in patients with prior anti-TNFα use in our study was lower than that in GEMINI 3 for both treatment groups (19.7% vs 14.5% in our study; 46.8% vs 24.8% in GEMINI 3), particularly for vedolizumab. This may possibly be due to an imbalance in baseline disease localization between the treatment groups in patients with prior anti-TNFα use; the proportion of patients with colonic type was lower in the vedolizumab group than in the placebo group (9.8% vs 25.8%). Although subgroup analysis indicated higher CDAI-100 response rate at Week 10 with vedolizumab versus placebo in patients with colonic type (66.7% vs 18.8%), differences between treatment groups were lower in patients with ileal and ileocolonic type. Therefore, there was a possibility to interfere the data of efficacy by fewer patients of colonic type in the vedolizumab group. Two recent prospective trials have reported that endoscopic efficacy of vedolizumab for the ileum was inferior to that for the colon [19, 20]. However, these trials were assessed by conventional ileocolonoscopy, and there is a possibility that proximal ileal lesions can be missed with this technique [21, 22]. Furthermore, endoscopic active ileal lesions are not always successfully identified by CDAI, or by assessment of biomarkers of CD severity, for instance CRP or fecal calprotectin [23]. Further research is needed to evaluate the efficacy of vedolizumab in ileal lesions by using capsule endoscopy or balloon-assisted enteroscopy. Another possibility is the higher proportion of loss of response to prior anti-TNFα in our study versus GEMINI 3 (~ 70% vs 44%). In common with other biologics, including ustekinumab [24], efficacy of vedolizumab in CD with loss of response to anti-TNFα is lower than in bio-naïve CD. Subgroup analysis in patients with loss of response to prior anti-TNFα showed little difference in CDAI-100 response rate at Week 10 between the two treatment groups in our study. However, in patients with inadequate response to prior anti-TNFα, this response rate was higher with vedolizumab versus placebo. Loss of response to prior anti-TNFα may help to account for the lower CDAI-100 response rate at Week 10 in patients with prior anti-TNFα use in our study versus GEMINI 3. Meanwhile, in patients with inadequate response, mainly primary non-responders, to prior anti-TNFα, the inductive efficacy of vedolizumab is certainly expected, because the mode of action for anti-TNFα is different.

The data sources for our study are considered reliable because this study was conducted in compliance with good clinical practice. Therefore, we performed exploratory subgroup and post hoc analyses to evaluate factors that could contribute to the improvement of vedolizumab treatment in daily clinical practice. In the subgroups of patients aged ≥ 35 years, with inadequate response to prior anti-TNFα, prior corticosteroid failure only, concomitant use of oral corticosteroids, and CRP > 1.6 mg/dL, CDAI-100 response was superior with vedolizumab versus placebo. Similarly, in the subgroups aged ≥ 35 years, without prior anti-TNFα failure, and prior corticosteroids failure only, CR rate was superior with vedolizumab versus placebo. Moreover, increasing age may be a predictive factor for CR. But the results of the post hoc univariate analyses have a risk of false positive due to the multiplicity of 44 statistical tests, warranting careful interpretation. Vedolizumab may be more effective in patients with non-intractable CD (for instance, prior corticosteroid failure only) or older patients with CD.

CDAI-100 response rate at Week 2 was numerically higher with vedolizumab versus placebo. Evaluation of temporal changes in CDAI scores in the induction phase showed that changes from baseline were greater with vedolizumab than with placebo from Week 2 onwards. Similar results for the difference between the two treatment groups from Week 2 were also recently reported [14]. Subgroup analysis showed that CDAI-100 response rate at Week 10 was higher with vedolizumab versus placebo in all subgroups, except for those related to prior anti-TNFα use. Among subgroups with prior anti-TNFα use, the difference in mean change from baseline in CDAI score and CDAI-100 response between the vedolizumab group and the placebo group were greater in the subgroup with inadequate response to prior anti-TNFα than in that with loss of response to prior anti-TNFα. To our knowledge, no previous reports have shown changes in the CDAI score during the induction phase in CD patients with inadequate response to prior anti-TNFα. In contrast, patients with higher baseline CRP tended to show more pronounced changes in the CDAI score. In the VICTORY Consortium, patients with severe disease activity shown to be less likely to achieve CR (hazard ratio [HR] 0.54; 95% CI: 0.31–0.95) and mucosal healing (HR 0.54; 95% CI: 0.31–0.95) [25]. Therefore, vedolizumab may be more effective for remission induction in patients with moderately active CD than in those with severe active disease.

In the maintenance phase, vedolizumab treatment resulted in a numerically greater rate of CR, CDAI-100 response, durable remission and corticosteroid-free remission at Week 60 compared with placebo. A similar trend for CR was observed between our study (41.7% vs 16.7%) and GEMINI 2 (39.0% vs 21.6%). In addition, regardless of prior anti-TNFα use, the vedolizumab group exceeded the placebo group for all endpoints, except durable remission in patients without prior anti-TNFα use. The GEMINI LTS study also revealed long-term maintenance of vedolizumab efficacy regardless of prior anti-TNFα exposure [26]. In patients who achieved remission induction after administration of vedolizumab, maintenance of remission may be improved based on the lower immunogenicity of vedolizumab [27].

Vedolizumab treatment in Japanese patients with CD was well tolerated in our study, with most AEs mild to moderate in intensity. The most frequent AE in both phases was upper viral respiratory tract infection. No marked differences were reported between the vedolizumab group and the placebo group. Importantly, there were no cases of PML, a known toxicity with nonspecific α4 integrin inhibition [28]. In a recent retrospective cohort study, the number of concomitant immunosuppressive agents (corticosteroids or immunomodulators; OR, 1.72 per agent) used was associated with infections [29]. This study also reported that gastrointestinal infections were observed in the inflammatory bowel disease patients who were treated with vedolizumab (2.4 per 100 patient years of exposure). In contrast, 4 mild or moderate infectious enteritis patients were identified in our study. All 4 patients were recovered and judged by investigators not related to study drug. From the perspective for the mode of action, we should still carefully observe it as the possible specific adverse event of vedolizumab. Although it depends on the clinical course, introduction of vedolizumab as monotherapy is preferred when considering the safety profile. Furthermore, vedolizumab might be a preferred option over other immunosuppressive agents for patients who should avoid systemic immunosuppression or the elderly [30].

Pharmacokinetics and immunogenicity results in our study were consistent with those in previous studies in non-Japanese [8, 9] and Japanese patients [31, 32]. CR at Week 10 was 13.3% in the subgroup with minimum to 1st quartile of serum vedolizumab concentration, and 40.0% in the subgroup with 4th quartile to maximum of serum vedolizumab concentration. Similar trends were observed in the quartile analysis using GEMINI 3 data [33]. In general, higher concentrations of biologics provide better outcomes. Baseline albumin and body weight were reported as factors affecting vedolizumab linear clearance [17], and higher serum albumin and lower body mass at baseline demonstrated better outcomes [34]. In our analysis, vedolizumab serum concentrations were increased in the subgroup with higher albumin at Week 0, but not in the subgroup with lower body weight.

Our study is limited by the small number of patients per treatment subgroup, for instance as stratification for prior anti-TNFα use. GEMINI 2 and 3 were global trials with > 1000 or > 400 subjects, respectively [8, 9]. Our study was performed locally in Japan for a regional registration, so it was not feasible to recruit on a similar scale to the GEMINI studies. Consequently, the results of the subgroup analysis should be interpreted carefully. A further limitation is that the study was not designed to statistically evaluate efficacy during the maintenance phase; the small sample size in the maintenance phase limits the interpretation of the results. Additionally, there was no endoscopic assessment as objective monitoring.

In conclusion, vedolizumab showed numerically higher efficacy compared with placebo as induction therapy in Japanese patients with CD, but the difference was not statistically significant. Vedolizumab also showed numerically higher efficacy compared with placebo as maintenance therapy. Vedolizumab was well tolerated in this treatment setting. In patients without previous anti-TNFα exposure and those with inadequate response to anti-TNFα, vedolizumab showed improved outcomes versus placebo during the induction phase; this improvement was shown after Week 2. Age might be a possible predictive factor of remission induction for future research. Vedolizumab constitutes a new treatment option for Japanese patients with moderate-to-severe CD.