Assessment of Individual versus Composite Endpoints of Acute Graft-versus-Host Disease in Determining Long-Term Survival after Allogeneic Transplantation

Abstract

The overall composite of graft-versus-host disease (GVHD)-free, relapse-free survival (GRFS), defined as survival free of grade III-IV acute GVHD (aGVHD), chronic GVHD (cGVHD) requiring systemic immunosuppressive therapy (IST), or relapse, has emerged as a useful composite in clinical trials and to capture clinically meaningful events that impact quantity and quality of survival after allogeneic hematopoietic cell transplantation (HCT). We reviewed 565 consecutive patients aged 18 years undergoing HCT for hematologic malignancy to analyze how baseline incidence, specifics of clinical definitions, and proposed reductions in any one individual event may dynamically alter the overall performance of the composite To determine the relative impact of each GRFS event (excluding death), we accounted for competing risks using Fine and Gray methods, and correlated each event with overall survival (OS) using Kaplan-Meier methods. The consequences of modulating individual or composite endpoints on OS, such as hypothesized reductions of events of an HCT interventional trial, were examined using Monte Carlo simulations. The median age of the cohort was 54 years (range, 18 to 73 years). The majority of patients received HLA-matched unrelated donor HCT (53%), consisting of peripheral blood stem cell grafts (90%) after myeloablative conditioning (68%). Relapse conferred the greatest risk for death (hazard ratio [HR], 7.89; 95% confidence interval [CI], 5.83 to 10.69), followed by grade III-IV aGVHD (HR, 6.16; 95% CI, 4.42 to 8.56) and cGVHD requiring IST (HR, 1.69; 95% CI, 1.16 to 2.46). The overall GRFS composite correlated with an HR of 4.81 (95% CI, 3.61 to 6.41), which was lower compared with either relapse or grade III-IV aGVHD. Statistical simulations found that modulating the combined risk of both relapse and grade III-IV aGVHD predicted the greatest change in 5-year OS. These simulations suggest that GRFS as currently defined may be less optimal for correlating with OS, and further refinement of composite endpoints is needed. Nonetheless, composite endpoints may be particularly helpful in mitigating potential difficulties in interpretation when competing risks are present, most commonly seen in HCT studies.

Establishment of a predictive model for GVHD-free, relapse-free survival after allogeneic HSCT using ensemble learning

“…However, composite end points using right-censored data only measure the time to the first event and GRFS can't replace OS, RFS or GVHD. Magenau et al reported that chronic GVHD had less modulating effect on OS than grade III-IV acute GVHD or relapse 39 . This is partly explained by the association of graft-versus-leukemia effect with chronic GVHD 40 .…”

“…Magenau et al reported that chronic GVHD had less modulating effect on OS than grade III-IV acute GVHD or relapse. 39 This is partly explained by the association of graft-versus-leukemia effect with chronic GVHD. 40 On the other hand, chronic GVHD has tremendous negative impact on quality of daily life after transplantation even under mild to moderate symptoms.…”

Inferences for current chronic graft-versus-host-disease free and relapse free survival

“…CGRFS has been recognized as a useful addition to the existing survival functions [11][12][13] and considered as suitable implication of transplant success [14]. Our clinical paper did not include interval estimation, which is critical for assessing precision of CGRFS estimates.…”