Vedolizumab for the prevention of intestinal acute GVHD after allogeneic hematopoietic stem cell transplantation: a randomized phase 3 trial

Abstract

Acute graft-versus-host disease (aGVHD) of the lower gastrointestinal (GI) tract is a major cause of morbidity and mortality in patients receiving allogeneic hematopoietic stem cell transplantation (allo-HSCT). Vedolizumab is a gut-selective anti-α₄β₇ integrin monoclonal antibody that reduces gut inflammation by inhibiting migration of GI-homing T lymphocytes. The efficacy and safety of vedolizumab added to standard GVHD prophylaxis (calcineurin inhibitor plus methotrexate/mycophenolate mofetil) was evaluated for prevention of lower-GI aGVHD after unrelated donor allo-HSCT in a randomized, double-blind, placebo-controlled phase 3 trial. Enrollment closed early during the COVID-19 pandemic with 343 patients randomized (n = 174 vedolizumab, n = 169 placebo), and 333 received ≥1 intravenous dose of 300 mg vedolizumab (n = 168) or placebo (n = 165) and underwent allo-HSCT. The primary end point was met; Kaplan-Meier (95% confidence interval) estimated rates of lower-GI aGVHD-free survival by day +180 after allo-HSCT were 85.5% (79.2-90.1) with vedolizumab versus 70.9% (63.2-77.2) with placebo (hazard ratio, 0.45; 95% confidence interval, 0.27-0.73; P < 0.001). For the 5 key secondary efficacy end points analyzed by day +180 after allo-HSCT, rates of lower-GI aGVHD-free and relapse-free survival and grade C-D aGVHD-free survival were significantly higher with vedolizumab versus placebo. No significant treatment differences were found for the other key secondary end points of non-relapse mortality, overall survival and grade B-D aGVHD-free survival, respectively. Incidence of treatment-related serious adverse events analyzed in patients receiving ≥1 dose of study treatment (n = 334) was 6.5% (n = 11 of 169) vedolizumab versus 8.5% (n = 14 of 165) placebo. When added to standard calcineurin inhibitor-based GVHD prevention, lower-GI aGVHD-free survival was significantly higher with vedolizumab versus placebo. ClinicalTrials.gov identifier: NCT03657160 .

Fig. 1. Eligibility, randomization and follow-up.

Discontinuation of the study refers to all patients who discontinued before the end of the long-term follow-up safety survey period of the study, 6 months after the last dose of study treatment. ^†Withdrawn by physician is noted as reason ‘other’. ^‡Patients included in the analysis for efficacy end points per protocol were those who received ≥1 dose of study treatment and also received allo-HSCT. One patient was randomized to receive vedolizumab but did not receive allo-HSCT; per protocol, this patient was not included in the analysis of efficacy end points but was included in the analysis of safety end points.

Fig. 2. Lower-GI aGVHD-free survival by day +180 after allo-HSCT.

Analysis included all randomized patients who received ≥1 dose of study treatment and received allo-HSCT. All statistical tests were two-sided. a, Graph shows number and proportion of patients with a lower-GI aGVHD event or death; censored for patients who had not had the lower-GI aGVHD event or died or had the event after a prespecified time, for example, last contact or day +180 after allo-HSCT, whichever occurred first. If a patient had a lower-GI aGVHD event and died due to any cause, including lower-GI aGVHD, the time to event was derived as the time to the first qualifying event (lower-GI aGVHD event). b, Frequency of lower-GI aGVHD by maximum clinical stages 0–4 by day +180 after allo-HSCT for patients in vedolizumab and placebo treatment groups and also the corresponding frequency of skin aGVHD and liver aGVHD in these treatment groups by maximum clinical stages 0–4 by day +180 after allo-HSCT. CI was based on the Clopper–Pearson method. c, KM estimate for the primary study end point lower-GI aGVHD-free survival from first study treatment (day −1) to lower-GI aGVHD event or death due to any cause. Red line shows the vedolizumab group; blue line shows the placebo group; open circles indicate censored patients. HR obtained via a Cox proportional hazards model with treatment group, stratified by randomization stratification factors: HLA match (7 of 8 or 8 of 8), conditioning regimen (MAC or RIC), ATG (with or without) and P value from a log-rank test (P = 0.0009). d, Forest plot of prespecified subgroup analyses for the primary study end point of lower-GI aGVHD-free survival by day +180 after allo-HSCT: conditioning regimen MAC or RIC, with or without ATG, CNI TAC or CYS, HLA match, and stem cell source peripheral blood or bone marrow. HRs plotted with 95% CIs were obtained via a Cox proportional hazards model with treatment group stratified by randomization strata. Results for the remaining prespecified subgroup analyses are shown in Extended Data Fig. 1.

Fig. 3. Key secondary efficacy end points by day +180 after allo-HSCT.

a–e, KM estimates for the secondary efficacy end points. Analyses included all randomized patients who received ≥1 dose of study treatment and allo-HSCT. In the fixed-sequence hierarchical testing procedure, once 1 efficacy end point was not significant (P ≥ 0.05), testing of subsequent end points was not performed. P values were obtained using a log-rank test unless otherwise stated. All statistical tests were two-sided. *P value is significant for vedolizumab versus placebo. ^†HR and 95% CI values were obtained from a Cox proportional hazards model with treatment group stratified by randomization strata: HLA match (7 of 8 or 8 of 8), conditioning regimen (MAC or RIC) and ATG (with or without). ^‡Time to first documented lower-GI aGVHD, relapse of underlying malignancy or death from any cause. ^§Sensitivity analysis, excluding lower-GI aGVHD events classified as clinical grade 0 or unknown. ^¶NRM was a competing risk in this competing risk sensitivity analysis; P value for comparison of vedolizumab with placebo was obtained by a Gray’s test. ^‖Time to first documented IBMTR grade C–D aGVHD (any organ) or death from any cause. ^**Death and relapse were competing risks in this sensitivity analysis; an event was defined as IBMTR grade C–D aGVHD (any organ) or death. P value was obtained by a Gray’s test. ^††Death from first dose of study treatment without occurrence of a relapse. ^‡‡Relapse was a competing risk in this sensitivity analysis; NRM was the time from first study treatment to death without occurrence of a relapse; P value was obtained by a Gray’s test. ^§§Overall survival by day +180 was the analysis of the time from the first dose of study treatment to death from any cause. All deaths were defined as events in this analysis. ^¶¶Time to first documented IBMTR grade B–D aGVHD (any organ) or death from any cause. ^‖‖Death and relapse were competing risks in this sensitivity analysis; an event was defined as IBMTR grade B–D aGVHD (any organ) or death. P value was obtained by a Gray’s test.

Extended Data Fig. 1. Subgroup analyses for the primary end point.

Forest plot shows results of prespecified subgroup analyses for the primary efficacy end point lower-GI aGVHD-free survival by day+180 after allo-HSCT: sex; race; geographic region; primary disease; and GVHD prophylaxis, in addition to subgroup analyses reported in Fig. 2d. Analyses included all randomized patients who received ≥1 dose of study treatment and allo-HSCT; statistical tests were two-sided. Hazard ratios plotted with 95% confidence intervals were obtained via Cox proportional hazard model with treatment group stratified by randomization strata. ALL, acute lymphoid leukemia; AML, acute myeloid leukemia; ATG, anti-thymocyte globulin; CML, chronic myeloid leukemia; CYS, cyclosporin; MDS, myelodysplastic syndrome; MMF, mycophenolate mofetil; TAC, tacrolimus.

Extended Data Fig. 2. Subgroup analyses for the secondary end points.

Forest plots show results of prespecified subgroup analyses for the secondary efficacy end points: conditioning regimen MAC or RIC; with or without ATG; CNI tacrolimus or cyclosporine; HLA match; and stem cell source peripheral blood or bone marrow. Analyses included all randomized patients who received ≥1 dose of study treatment and allo-HSCT; statistical tests were two-sided. Hazard ratios plotted with 95% confidence intervals were obtained via Cox proportional hazard model with treatment group stratified by randomization strata. Allo-HSCT, allogeneic hematopoietic stem cell transplantation; ATG, anti-thymocyte globulin; aGVHD, acute graft-versus-host disease; CNI, calcineurin inhibitor; HLA, human leukocyte antigen; MAC, myeloablative conditioning; RIC, reduced intensity conditioning.