Steroids vs. steroids plus additional agent in frontline treatment of acute graftversus-host disease: a systematic review and meta-analysis of randomized trials

Abstract

Despite extensive research in the last few decades, progress in treatment of acute graft-versus-host disease (aGvHD), a common complication of allogeneic hematopoietic cell transplantation (HCT), has been limited and steroids continue to be the standard frontline treatment. Randomized clinical trials (RCTs) have failed to find a beneficial effect by escalating immunosuppression using additional agents. Considering the small number of RCTs, limited sample sizes, and frequent early termination due to anticipated futility, we conducted a systematic review and an aggregate data meta-analysis to explore whether a true efficacy signal has been missed due to the limitations of individual RCTs. Seven reports met our inclusion criteria. The control arm in all studies was 2 mg/kg/day prednisone (or equivalent). The additional agent(s) used in the experimental arm(s) were higher dose steroids, anti-thymocyte globulin, infliximab, anti- interleukin-2 receptor antibody (daclizumab and BT563), CD5-specific immunotoxin, and mycophenolate mofetil. Random effects meta-analysis revealed no efficacy signal in pooled response rates at various times points. Overall survival at 100 days was significantly worse in the experimental arm (relative risk 0.83, 95% confidence interval 0.74–0.94, P = 0.004, data from 3 studies) and showed a similar trend (albeit not statistically significantly) at 1 year as well (RR 0.86, 95%CI 0.68–1.09, P = 0.21, data from 5 studies). In conclusion, these results argue against the value of augmented generic immunosuppression beyond steroids for frontline treatment of aGvHD and emphasize the importance of developing alternative strategies. Novel forms of immunomodulation and targeted therapies against non-immune-related pathways may enhance the efficacy of steroids in this setting and early predictive and prognostic biomarkers can help identify the subgroup of patients that would likely need treatments other than (or in addition to) generic immunosuppression.

Introduction

Acute graft-versus-host disease (aGvHD) continues to be a major complication of allogeneic hematopoietic cell transplantation (HCT) and is associated with significant morbidity and mortality¹. Despite more than five decades of research, there is no FDA-approved agent for frontline treatment of aGvHD and prednisone 2 mg/kg/day (or equivalent) is nearly universally used for grade II–IV aGvHD². Considering the limited response of aGvHD to standard frontline treatment (durable complete response [CR] <40%)³ and the excessive mortality (>70%) of steroid-refractory aGvHD⁴, more effective strategies are desirable. Unfortunately, however, randomized frontline trials are scarce and a number of trials were halted early due to unacceptable toxicity or futility on interim analysis. This raises the question of whether a true efficacy signal exists but has not been demonstrable by individual studies. We addressed this question using an aggregate data meta-analysis of randomized clinical trials (RCTs) comparing outcomes of frontline aGvHD treatment with steroids vs. steroids + additional agent(s).

Methods

This study was conducted in accordance with the PRISMA guidelines⁵. Studies were included if they (i) were RCTs published in peer-reviewed journals, (ii) examined frontline treatment of aGvHD following allogeneic HCT, (iii) randomized patients to receive 1–2 mg/kg prednisone (or equivalent; with or without placebo) daily vs. 1–2 mg/kg/day prednisone (or equivalent) plus the experimental agent, and (iv) reported at least one outcome (see below) for both arms. Using higher dose steroids was not an exclusion criterion and was considered as an experimental arm. Discontinuation of previous immunosuppressive medications at the time of enrollment was not a requirement for inclusion. However, we required the groups to receive no new anti-GvHD treatment other than steroids (control group) or steroids with the experimental agent (experimental group) before enrollment. Eligible studies had to have reported on overall response rate (ORR) or CR at any specific time point within the first 6 weeks or OS within the first 100 days after initiation of frontline therapy.

We performed electronic searches of MEDLINE and EMBASE to include articles indexed as of November 24, 2015. The following terms were used for MEDLINE search: (“acute graft-versus-host disease”[Title] OR “acute graft versus host disease”[Title] OR “acute graft vs host disease”[Title] OR “acute graft-vs-host disease”[Title] OR “acute GVHD”[Title]) AND (“randomized” OR “randomized” OR “phase 3” OR “phase III”). The following were used for EMBASE search: (‘acute graft-versus-host disease’:ti OR ‘acute graft versus host disease’:ti OR ‘acute graft vs host disease’:ti OR ‘acute graft-vs-host disease’:ti OR ‘acute GVHD’:ti) AND (‘randomized’ OR ‘randomized’ OR ‘phase 3’ OR ‘phase III’). Duplicates were first removed from the search results. The remaining reports were then screened by scanning titles and abstracts for the inclusion criteria. Two reviewers (A.R. and G.A.C.) independently reviewed the studies, collected the data, and resolved discrepancies by consensus.

A random-effects model was used for meta-analysis using study-level aggregate data⁶. Effect size was measured by relative risk (or risk ratio; RR) and reported along with the corresponding two-sided 95% confidence interval (95%CI). Response and survival by the specified time point(s) were considered events. Study heterogeneity was assessed using the Cochran’s Q statistic and quantified using I² = 100 × (Q − df)/Q, where df = k − 1, df is the degrees of freedom, and k is the number of studies⁷. Sensitivity analysis for each outcome was performed by removing individual studies (one at a time) and recalculating the pooled estimate. Under one of the following three circumstances, the removed study was considered influential: (i) If the new pooled estimate was different by >20% from the one derived from all studies combined; (ii) If the new, but not the original, pooled estimate was statistically significant; (iii) If the original, but not the new, pooled estimate was statistically significant. Review Manager (version 5.3; Cochrane Collaboration, Oxford, United Kingdom) was used for meta-analysis. P < 0.05 was considered statistically significant.

Results

Seven reports met our inclusion criteria (Supplementary Figure S1)^8–14, and a summary of these studies is provided in Table 1. An inclusion criterion was aGvHD grade II–IV in 5 studies^9–13, aGvHD grade II–III in one study¹⁴, and any grade aGvHD requiring systemic therapy in one study⁸. In the last study, aGvHD was grade II–IV in 89% of patients. The total number of patients in the included studies was 895. Two studies only included bone marrow (as graft source) transplants^10,12 and one included only matched sibling transplants¹². There were no restrictions with regards to the underlying hematologic diseases or conditioning intensity in any of the included studies. An inclusion criterion in one study was pre-randomization GvHD prophylaxis using either cyclosporine or cyclosporine and methotrexate¹³, whereas any type of prophylaxis was allowed in other studies. In general, the studies did not allow for new agents for the treatment of aGvHD prior to enrollment (exceptions detailed in Table 1). Four studies were double-blinded and placebo-controlled^8,11,13,14 while 3 studies were not blinded and did not use placebo^9,10,12. The additional agent in the experimental arm varied among studies: higher-dose steroids (10 mg/kg/day methylprednisolone)¹², anti-thymocyte globulin¹⁰, interleukin-2 receptor antibody (daclizumab and BT563)^11,14, tumor necrosis factor alpha antibody infliximab⁹, CD5-specific immunotoxin¹³, and mycophenolate mofetil⁸. The frequency of severe GvHD (grade III–IV) at the time of enrollment ranged between 0 and 33% among the included studies which reported this information (n = 5). There were 8 outcomes available each from at least 3 studies: CR at 2 weeks (4 studies)^8–10,13, CR at 4 weeks (4 studies)^8–10,13, CR at 6 weeks (4 studies)^8,10,11,13, ORR at 2 weeks (4 studies)^8–10,13, ORR at 4 weeks (5 studies)^8–10,12,13, ORR at 6 weeks (4 studies)^8,10,11,13, OS at 100 days (3 studies)^9–11, and OS at 1 year (5 studies). Additional data, not available in the original publications, were provided by the authors in one study¹⁰ and the BMT Clinical Trials Network leadership in another study⁸. Three studies were halted at interim analysis, 2 due to futility and 1 due to inferior results in the experimental arm^8,9,11.

Table 1.

Characteristics of the included studies

Study	N	aGvHD requirement	Experimental agent	Severe aGvHD at enrollment	Prior GvHD prophylaxis	Prior steroid exclusion criteria	Completed
Cahn, 1995	69	Grade III–IV	BT5631	No grade IV	CsA or CsA/MTX	None allowed	Yes
Martin, 1996	243	Grade II–IV	CD5 immunotoxin	NA	CsA/MTX: 84% Cont. through Rx	None allowed	Yes
Van Lint,1998	93	Grade II–IV	MP 10 mg/kg/day2	NA	CsA: 53% CsA/MTX: 47%	NA	Yes
Cragg, 2000	96	Grade II–IV	ATG	5% (no grade IV)	MTX/CsA: 53% MTX/ATG/Pred.	NA	Yes
Lee, 2004	102	Grade II–IV	Daclizumab	20%	NA	>1mg/kg/d within 7d	No
Couriel, 2009	57	Grade II–IV	Infliximab	33%	Tac/MTX: 88%	>2d and >2 mg/kg/d	No
Bolanos-Meade, 2014	235	Requiring systemic Rx	MMF	33%	No prior MMF allowed within 7d	>3d and >0.5 mg/kg/d	No

The experimental agent in all studies was added to standard treatment (2 mg/kg/day prednisone [or equivalent; with or without placebo]).

¹BT563 is a monoclonal antibody against interleukin-2 receptor;

²The experimental agent in this study was additional steroids. aGvHD: acute graft-versus-host disease; ATG: anti-thymocyte globulin; CsA: cyclosporine; Cont.: continued; d: days; MMF: mycophenolate mofetil; MP: methylprednisolone; MTX: methotrexate; NA: not available; Pred: prednisone; Rx: treatment

The pooled RR (95%CI) estimate for CR at 2, 4, and 6 weeks was 0.94 (0.77–1.15), 1.05 (0.74–1.49), and 1.01 (0.78–1.32), respectively, all non-significant (P = 0.52, 0.77, and 0.91, respectively; Figure 1). Similarly, the pooled RR (95%CI) estimate for ORR at 2, 4 and 6 weeks was 1.06 (0.96–1.17), 1.06 (0.96–1.18), and 0.98 (0.81–1.19), respectively, all non-significant (P = 0.28, 0.22, and 0.87, respectively; Supplementary Figure S2). Interestingly, pooled analysis of OS at 100 days revealed significantly worse outcomes in the experimental arm (RR 0.83, 95%CI 0.74–0.94, P = 0.004, Figure 2). Also, there was a trend for worse OS at 1 year in the experimental arm, although the difference did not reach statistical significance (RR 0.86, 95%CI 0.68–1.09, P = 0.21, Figure 2). There was significant study heterogeneity in the analysis of CR at 4 weeks (I² = 0.68) and OS at 1 year (I² = 0.63). Slight study heterogeneity was present also for CR and ORR at 6 weeks (I² = 0.51 and 0.52, respectively). In sensitivity analysis, none of the included studies was influential in any of the studied outcomes. The number of studies was not large enough for meta-regression over factors such as the frequency of severe aGvHD at the time of enrollment in each study.

Figure 1. Forest plots for complete response.

Complete response (CR) at 2 weeks (A), 4 weeks (B), and 6 weeks (C). Response is considered event. Size of squares in each panel is proportional to the weight assigned to the corresponding study (studies with larger weights have narrower confidence intervals). MMF: Mycophenolate mofetil, MP: Methyl-prednisolone, ATG: Anti-thymocyte globulin

Figure 2. Forest plots for overall survival.

Overall survival (OS) at 100 days (A) and 1 year (B). Survival is considered event. Size of squares in each panel is proportional to the weight assigned to the corresponding study (studies with larger weights have narrower confidence intervals). MMF: Mycophenolate mofetil, ATG: Anti-thymocyte globulin

Discussion

The fact that 3 RCTs were halted before completion raises the possibility that pooled analysis may reveal a hidden efficacy signal. However, our results demonstrate no evidence for benefit beyond steroids (at least with agents used in previous RCTs) and eliminate the possibility of low power being the underlying cause of negative results in individual RCTs. Importantly, pooled survival at 100 days (and also at 1 year, albeit not statistically significantly) was shorter with escalated immunosuppression. Ideally a meta-analysis of RCTs should pool the results of studies using the same treatment approach. Considering that only two previous RCTs with our inclusion criteria used the same treatment approach (IL-2 receptor blockade), we decided to design this meta-analysis as a steroid vs. steroid plus additional agent study. Within the limitations of this inclusive approach (due to study-specific differences in participants, diseases, transplants, steroid tapering strategies, and designs), our meta-analysis argues against the value of augmented generic immunosuppression beyond steroids for frontline treatment of aGvHD and emphasizes the importance of developing alternative strategies. Indeed, at least a subgroup of patients seems to be adequately treatable with lower dose steroids. A randomized trial exploring this possibility demonstrated that in patients with grade IIa aGvHD (i.e. upper gastrointestinal symptoms, diarrhea volume < 1 L/day, rash < 50% of body surface, and no liver dysfunction), lower dose steroids did not increase the risk of requiring secondary therapy and did not impact survival¹⁵. However, the primary endpoint of the study (33% or more reduction of the mean cumulative prednisone dose by day 42) was not met and the incidence of adverse events attributed to prednisone was not significantly different between the two groups.

The proportion of patients in each group who continued their previous prophylactic immunosuppressive medications after enrollment could not be ascertained from the included reports. Nonetheless the randomized design is expected to have minimized baseline differences between the groups including immunosuppressive medications (other than steroids and/or experimental agents) that were continued after enrollment. An interesting feature of the included studies in the present meta-analysis is the low frequency of severe aGvHD (0–33% in studies which reported this information) at the time of enrollment. There is at least a theoretical possibility that the attempted treatment strategy in the experimental arm in some of the included studies might have improved outcomes if a larger proportion of patients enrolled had severe aGvHD at the time of enrollment.

Importantly, these studies of initial aGvHD therapy did not identify or target higher risks cohorts with particularly poor outcomes using steroids alone. While additional immunosuppression has not been demonstrated to help the higher risk group, the higher mortality of high-risk patients may justify further study particularly directed to them¹⁶. An important question and a potentially valuable goal for ongoing and future research would be to distinguish between low- and high-risk patients for developing steroid refractoriness and poor outcomes. Several biomarkers have been proposed and many others are being tested^17–19. Sensitive and specific early predictive or prognostic biomarkers of severe aGvHD or of steroid refractoriness could help identify high-risk patients that may benefit from early intervention, perhaps using alternative strategies beyond generic immunosuppression. Future progress in treatment of aGvHD may depend on targeting specific inflammatory mediators^20–23, alternative immunomodulating approaches^24,25, and more attention to non-immune-related pathways^26–28 rather than escalating generic immunosuppression.

Supplementary Material

1. Figure S1: Article flow diagram.

A total of 7 studies were included in the final analysis

2. Figure S2: Forest plots for overall response.

Overall response rate (ORR) at 2 weeks (A), 4 weeks (B), and 6 weeks (C). Response is considered event. Size of squares in each panel is proportional to the weight assigned to the corresponding study (studies with larger weights have narrower confidence intervals). MMF: Mycophenolate mofetil, MP: Methyl-prednisolone, ATG: Anti-thymocyte globulin

Highlights.

• Progress in frontline treatment of acute graft-versus-host disease has been limited.

• 7 randomized trials have compared steroids vs. steroids plus additional agent.

• A meta-analysis shows no efficacy signal with increasing generic immunosuppression.

• Novel therapies are needed for frontline treatment of acute graft-versus-host disease.