Despite relatively high rates of morphology-based complete remission (CR) after intensive chemotherapy, most patients with acute myeloid leukemia (AML) ultimately relapse, and long-term survival is only 30 to 40 percent. Thus we see the interest in improving risk stratification in AML and in the delivery of risk-adapted therapies post-remission.1 In addition to well-validated clinical parameters such as age and cytogenetic and molecular abnormalities, there is increasing evidence that the ability to identify measurable residual disease (MRD) below the morphology-based 5 percent blast threshold is an important prognostic factor.2,3 As such, the European LeukemiaNet (ELN) has updated AML response criteria to include a new category of morphological CR without MRD (CRMRD-)4 and has additionally recommended MRD monitoring as standard of care for all patients with AML.3
In their recently published article in Leukemia, Dr. Bruno Paiva and colleagues raise several practical concerns with respect to widespread or decentralized MRD monitoring. The studies investigating the prognostic value of MRD in AML have been performed at single sites or centralized in a few laboratories with extensive experience. Heterogeneity in methodology and interpretation of results is likely as MRD monitoring transitions into broader clinical practice. A correlation between heterogeneity in key aspects of MRD testing using multiparameter flow cytometry (MFC) in particular and the ability of this technique to predict outcome in AML had not previously been analyzed.
The authors set out to address these unresolved questions by analyzing a cohort of 1,076 patients with AML in first remission (CR1) reported to the PETHEMA registry, in whom MRD was evaluated immediately post-induction by MFC in local laboratories over a period of 20 years (1999-2019). Most of the patients were younger than 60 years (76%), and they were classified with favorable, intermediate, and adverse cytogenetics according to 2010 Medical Research Council (MRC) criteria, modified to include data on CEBPA, NPM1, and FLT3-ITD mutational status where available. An MRD threshold of 0.1 percent was selected for downstream analysis of cumulative incidence of relapse (CIR) and overall survival (OS). They found that patients with MRD less than 0.1 percent (n=599) had a 29 percent lower CIR and a 27 percent lower risk of death compared with patients with MRD greater than 0.1 percent (N=477). Median relapse-free survival for patients with MRD less than 0.1 percent versus greater than 0.1 percent was 28 versus 14 months, respectively, thus confirming an association between MRD detection at first remission and inferior survival, as seen in previous studies.
Further analyses yielded several interesting findings that have not been reported previously. In subgroup analysis, the prognostic value of MRD status in CR1 was limited to patients with adverse genetics and to those receiving chemotherapy alone for consolidation. MRD status had no statistically significant impact on CIR, relapse-free survival, or OS among patients undergoing hematopoietic stem cell transplantation (HSCT), suggesting that HSCT may overcome MRD positivity in risk-adapted protocols, particularly in patients with intermediate genetic risk. In multivariable analyses, together with patient age, white blood cell count, genetic risk, and postconsolidation therapy, MRD status had no independent prognostic value for CIR but did have a statistically significant impact on OS (HR, 1.30). However, stratification by time period uncovered that the independent prognostic value of MRD status in OS was only observed in the more recent PETHEMA LMA 2010 protocol, in keeping with improved MFC assessment of MRD over time. In a recursive partitioning model, the authors determined that age was the single best discriminator of OS; among cases of patients older than 60 years, MRD status did not significantly impact OS. To assess methodological heterogeneity of decentralized MRD assessment using MFC, the authors surveyed hospitals regarding techniques used in the PETHEMA LMA 1999, 2007, and 2010 protocols. The responses revealed significant heterogeneity intra- and inter-protocols, reflecting a lack of standardization at the national level.
In Brief
This study provides a critical “real-world” analysis of decentralized MRD assessment in a large patient population with a long follow-up period. While the study corroborates some observations from prior studies, it raises numerous important concerns. The authors found that, in contrast to observations from a recent meta-analysis by Dr. Sarah A. Buckley and colleagues,5 nearly all aspects of MFC-based MRD testing had an impact on its ability to predict outcomes in AML. These findings stress the need for careful design and standardization of MFC panels developed in individual laboratories, and support the ELN statement discouraging MRD assessments in centers without relevant experience.3 Adequate standardization of this technique will ultimately improve risk stratification of patients with AML, thereby facilitating clinical decision-making outside trials. These authors advocate for 0.1 percent, which may be a higher threshold than other series. Furthermore, as there is increasing interest in post-remission maintenance therapies with targeted agents, the parameters established for MFC-standardized will likely provide a framework to guide standardization for MRD with molecular results.
Competing Interests
Dr. Kagan and Dr. DeZern indicated no relevant conflicts of interest.
