Unrelated Allogeneic Stem Cell Transplant in Relapsed/Refractory AML: Widening the Yellow Brick Road

In this issue of Clinical Cancer Research, Nagler and colleagues utilize a large observational dataset from the Acute Leukemia Working Party (ALWP) of the European Society for Blood and Marrow Transplantation (EBMT) to conduct an epidemiologic analysis of survival gains over the last 20 years among adults with relapsed or refractory acute myeloid leukemia (R/R AML) receiving an unrelated donor (UD) allogeneic hematopoietic stem cell transplantation (HSCT; ref. 1).

Acquiring complete remission (CR) was historically a requisite for HSCT in AML. However, up to 30% patients fail to achieve CR with upfront induction or low-intensity regimens (2). Treatment options for these patients are limited. HSCT can achieve long-term remission in a proportion of R/R AML (3, 4), but contemporary series demonstrate ample role for improvement with 2-year overall survival (OS) and cumulative incidence of relapse of 30% and 54% (5). In this study, the authors describe significant gains in OS and leukemia-free survival (LFS) among patients transplanted for R/R AML between 2010 and 2019 compared with the 2000 to 2009 period. The results add to the growing evidence on the value of HSCT, which represents the only potentially curative option for these patients. By focusing on UD HSCT, the authors build on scarce data for unrelated transplantation in R/R AML and contextualize the survival trends into the modern era of near universal access to suitable donors, in which most transplants are not from matched siblings (6, 7).

The multicenter, registry-based, retrospective cohort study included 3,430 patients across 600 EBMT transplant centers. A multivariable proportional hazard regression model was built to identify the effect of time periods on survival outcomes, adjusting for donor sex, UD type, donor and patient cytomegalovirus (CMV) status, stem cell source, conditioning type, total body irradiation, disease status at HSCT, graft T-cell depletion, and AML cytogenetic risk. Estimated 2-year OS and LFS from time of transplantation significantly increased by 6% in the 2010 to 2019 period. In multivariable analysis, transplantation during 2010 to 2019 was associated with 13% lower risk of leukemia-related death and 14% lower risk of overall mortality.

Though survival gains may be expected in population-based studies of historical transplant outcomes, a crucial finding in this analysis is that the predominant factor explaining improved survival over time was a decrease in posttransplant relapse. Disease relapse and nonrelapse mortality (NRM) are the two most important barriers to long-term survival in HSCT (8). More effective supportive care such as timely antimicrobials, particularly antifungal and CMV prophylaxis and treatment, and enhanced graft-versus-host disease (GVHD) prevention strategies such as posttransplant cyclophosphamide, as well the rapid identification of alternative donors and increased use of reduced intensity conditioning (RIC), have resulted in reduced NRM. Indeed, longitudinal analyses, including previous ALWP/EBMT reports, demonstrate that survival gains mostly result from improved NRM (9, 10). Contrarily, targeting risk of posttransplant relapse has been challenging and minimal to no gains have been documented in previous population-based studies (11).

In this analysis, the 2-year cumulative incidence of posttransplant relapse, calculated with death as competing risk, decreased by 5% in 2010 to 2019. Furthermore, this time period was associated with 15% lower risk of posttransplant relapse compared with 2000 to 2009. By contrast, no difference was noted in NRM in univariate or multivariate analysis. This reduction in posttransplant relapse over time is an exciting and important finding. Though relapse incidence remains unacceptably high, demonstrating such gains in longitudinal analyses should be a cause for optimism for transplantation as an evolving therapeutic option in a group of patients with dismal outcomes.

Analyses of large registries, as the EBMT and the Center for International Blood and Marrow Transplant Research, are important to scrutinize the role of transplantation in specific settings, like UD HSCT for R/R AML in this report. However, they carry an important risk of unmeasured confounding and selection bias. Comparison of time-dependent cohorts is particularly susceptible to changes attained during the study period, which affect therapy choices for individual patients. Unmeasured confounders affecting individual choices cannot be controlled for within observational registries and represent an inherit limitation in large database outcomes research. Furthermore, several factors driving HSCT accessibility and safety have considerably changed within the timeframe of this analysis. Expansion of UDs, growing RIC utilization, and shifting preferences in GVHD prophylaxis translated into substantial baseline differences between study cohorts, leading to comparison of highly heterogeneous groups. Multivariable analysis allowed statistical adjustments for some important factors, but conclusions about the impact of any individual variable on the study findings are limited.

Despite the limitations, valuable questions arise from these longitudinal trends. First, therapeutic advances positively impacting posttransplant relapse and NRM risk must be identified and expanded. These may be pre- or posttransplant measures. For example, incorporating genomic minimal residual disease (MRD) to decide between myeloablative or RIC HSCT has been associated with improved survival (12). Similarly, implementing MRD monitoring to guide early posttransplant interventions may benefit patients with R/R AML patients (13). Posttransplant prophylactic interventions including preventive donor lymphocyte infusions or treatment with hypomethylating agents, the BCL-2 inhibitor venetoclax, or tyrosine kinase inhibitors have shown efficacy in preventing posttransplant relapse (14, 15). Survival gains outside NRM in this report underscore the importance of prospectively studying these strategies.

Second, primary refractory (PREF) AML is biologically distinct from relapsed AML (2, 16). Though all patients with active disease fare poorly in epidemiologic HSCT studies, there are observed differences in clinical outcomes following HSCT in PREF and relapsed refractory AML (3, 17). These biological and clinical differences suggest that future analyses should be limited to one entity to better interpret time-dependent survival gains and hypothesize on the underlying reasons. Primary refractory disease is one of the most important causes of treatment failure in newly diagnosed AML (4). Longitudinal studies restricted to PREF AML controlling for disease burden and chemosensitivity will be particularly helpful to define the role of HSCT.

Third, cytogenetic and clonal heterogeneity are hallmarks of AML (2). Transplant value must be defined within the context of molecular prognostic markers. Recurrent cytogenetic abnormalities are independent predictors of survival and influence HSCT efficacy (18, 19). In this report, cytogenetic risk was accounted for but missing for half the patients. The heterogeneous nature of the dataset also threatens the ability to draw conclusions from regression analysis of cytogenetic risk groups. The advent of widespread next-generation sequencing has uncovered a complex molecular landscape in AML with implications in classification and prognosis (20). Co-mutation patterns classify AML subtypes into nonoverlapping clinically distinct phenotypes (2). Thus, future population-based studies may benefit from linkage between large transplant registries and sequencing databases (Fig. 1). Molecular studies usually lack granular transplantation data while transplant registries endure inconsistent mutation reporting across centers. Linking datasets will help define survival gains and define the role of HSCT across distinct molecular subgroups.

In conclusion, this report from the ALWP/EBMT group should be considered an important contribution to the body of data demonstrating proof of concept for HSCT as a beneficial and evolving treatment in R/R AML. In this challenging population with limited therapeutic options and poor prognosis, the study found significant time-dependent improvements over the last two decades in survival measures and relapse incidence associated with unrelated transplantation. Further studies are needed to identify morphologic and molecular subgroups that derive most benefit from transplant. More importantly, clinical trials are needed to prospectively examine and expand on the pre- and postallograft strategies that may explain these survival gains.

A. Shastri reports grants from Kymera Therapeutics and other support from Janssen, Rigel Pharmaceuticals, and NACE outside the submitted work. No disclosures were reported by the other author.