Impact of Individual Vs. Composite Endpoints for Prediction of Long-Term Survival in Gvhd Clinical Trials Research

INTRODUCTION The composite endpoint of GVHD-free, relapse-free survival (GRFS), defined as survival free from grade 3–4 acute GVHD (aGVHD), chronic GVHD (cGVHD) requiring systemic immunosuppressive therapy (IST), or relapse has emerged as an outcome to capture clinically meaningful events. In the present study, we validated GRFS as a useful surrogate endpoint by examining the relative importance of individual events and the GRFS composite itself on overall survival (OS) at 5-years after allogeneic HCT. METHODS We conducted an IRB-approved retrospective cohort study on 565 consecutive patients (age ≥18 years) undergoing first-time allogeneic HCT for hematologic malignancy (January 2007–March 2013). Data abstraction of GRFS events was performed through chart review of the EHR. To determine the impact of each event of the GRFS composite endpoint, we accounted for competing risks using Fine and Gray methods, and correlated each event with OS using Kaplan-Meier methods. Using statistical simulations, the relative consequences of successfully modulating individual or composite endpoints on OS were examined to further elucidate the role of GRFS in clinical trials research. RESULTS The median age of the cohort was 54 years (18–73 years), comprised predominantly of males (58%) and Caucasians (94%). The most common indication for transplant was AML (41%), followed by lymphoma (21%) and MDS/MPD (14%). Patients received HLA-matched related (47%), and matched unrelated (53%) donors, and mostly peripheral blood stem cells (90%) and myeloablative conditioning (68%). Relapse conferred the greatest risk for death (HR 7.89; 95% CI:5.83–10.69). Grade 3–4 aGVHD was associated with HR 6.16 (95%CI:4.42–8.56), while cGVHD requiring IST was associated with lower HR 1.69 (95%CI:1.16–2.46). Indeed, the GRFS composite correlated with less HR 4.81 (95% CI:3.61–6.41) compared with either relapse or grade 3–4 aGVHD events. To determine the impact of hypothetical improvements on individual events or combined events (composite) on OS, we next performed Monte Carlo simulations. Compared with individual events, modulating composite endpoints predicted the greatest change in 5-year OS ( Figure 1 ). The greatest gain in OS was seen by proportionally reducing both grade 3–4 aGVHD and relapse (modified GRFS). Reducing chronic GVHD requiring IST proportionally with grade 3–4 aGVHD and relapse events (conventional GRFS) did not change 5-year OS. CONCLUSION These findings highlight the importance of influencing OS by targeting the major early adverse post-HCT events (i.e., grade 3–4 acute GVHD and relapse). Moreover, the simulations herein suggest that GRFS as currently defined may be less optimal for predicting OS than using early grade 3–4 aGVHD and relapse as composite endpoints. Importantly, further refinement of these endpoints will depend on the a priori defined study population as well as intervention.