Measurable residual disease (MRD), regardless of method of assessment, is a robust individual-level predictor for overall survival among patients with acute myeloid leukemia (AML) who have already received induction therapy, according to findings from a pooled analysis presented at the 2025 American Society of Hematology (ASH) Annual Meeting & Exposition (Abstract 343).
This study represented the largest dataset of MRD in patients with AML to date and showed strong individual-level surrogacy for use of MRD as a substitute regulatory endpoint for expedited drug development.
“The individual-level association is pretty striking—[MRD] positivity remains an independent predictor of worse outcomes,” said Jesse Tettero, MD, PhD, who led the research while at Amsterdam University Medical Center Department of Hematology in the Netherlands; he is now at Virginia Tech FBRI Cancer Research Center. “It provides stronger prognostic information than any single genetic or molecular marker we currently use. The association was very consistent, even after adjusting for other clinical factors, which really confirms the patient-level link between [MRD] and survival.”
Background and Study Methods
MRD is an increasingly commonly used prognostic biomarker for relapse and long-term survival in AML.
The U.S. Food and Drug Administration (FDA) has accepted MRD as a surrogate endpoint in clinical trials for multiple myeloma, but not yet for AML due to disease heterogeneity, variability in assessments, and a lack of trial-level validation complications.
The researchers sought out to evaluate individual and trial-level surrogacy for MRD, as assessed by multiparameter flow cytometry and quantitative polymerase chain reaction for mutant NPM1, using harmonized patient-level data from seven randomized phase II/III trials, to try to establish MRD as an acceptable surrogate endpoint.
“The quality of [MRD] testing really depends on where and how it’s done. Centralized, experienced laboratories deliver accurate and reproducible results, which are essential if [MRD] is to be used for clinical or regulatory decisions. I think the field has really developed and matured, so people are becoming more interested in using [MRD],” said Dr. Tettero.
Data were collected from 1,858 adult patients with AML who were included in the AMLSG 09-09; AML-102, AML-103, and AML-132 HOVON-SAKK studies; SAL cohort; and UK-NCRI AML17 trials, which were collected through the HARMONY Alliance group. Each trial included randomization to an experimental agent or placebo added to a standard intensive induction chemotherapy. At least 20 patients were included in each arm and in each MRD subgroup. Patients were included in the analysis if they had undergone MRD assessment after two chemotherapy cycles by either multiparameter flow cytometry or quantitative polymerase chain reaction for mutant NPM1. All data were harmonized using the OMOP common data model.
The FDA had the researchers analyze two levels of surrogacy: the association between MRD status and overall survival for individual-level surrogacy, and the relationship between treatment effects on MRD and overall survival using hazard ratios (HRs) and odds ratios (ORs) for trial-level surrogacy. A coefficient of determination (R2) of more than 0.8 with the lower bound of the 95% confidence interval (CI) being higher than 0.6 was considered strong for the trials-level surrogacy.
Subgroup analyses were conducted according to the method of MRD assessment and transplant status.
Key Findings
By multivariate analysis, MRD was associated with significantly worse overall survival outcomes across all of the collected studies (HR = 1.66; 95% CI = 1.33–2.07; P < .001). Even when patients were stratified by MRD assessment method or treatment arm, this association persisted.
The global OR for the association between MRD status and overall survival was 0.39 (95% CI = 0.32–0.47). In transplanted patients, the OR was 0.61 (95% CI = 0.42–0.81), and was 0.33 (95% CI = 0.25–0.41) for nontransplanted patients.
In terms of trial-level surrogacy, the analysis was limited to multiparameter flow cytometry–based MRD (n = 1,268) as there were a limited number of quantitative polymerase chain reaction–based studies.
The overall R2 between treatment effect on MRD and overall survival was 0.91 (95% CI = 0.56–1.00), which showed strong surrogacy, even though the lower bound for the CI fell below the predefined threshold. Among nontransplanted patients, the R2 was 0.99 (95% CI = 0.94–1.00) and the R2 for transplanted patients was 0.54 (95% CI = 0.00–1.00), which suggested that, at the trial level, transplant status may impact the association between MRD and overall survival.
“For the trial-level surrogacy, the data for nontransplanted patients show a strong correlation, but caution is warranted due to the number of trials. Trial-level surrogacy is a very high bar to reach,” said Dr. Tettero.
Disclosure: The study was conducted within the HARMONY Alliance. For full disclosures of the study authors, visit hematology.org.
