The Hematologist

July-August 2018, Volume 15, Issue 4

The Usual Suspects Aren't Always the Bad Guys! Molecular MRD Comes of Age in AML

Annette S. Kim, MD, PhD Associate Professor of Pathology
Harvard Medical School; Brigham and Women's Hospital, Boston, MA

Published on: July 13, 2018

Jongen-Lavrencic M, Grob T, Hanekamp D, et al. Molecular minimal residual disease in acute myeloid leukemia. N Engl J Med. 2018;378:1189-1199.

While the morphologic designation of complete remission (CR) remains a widely used benchmark in the treatment of patients with leukemia,1 the persistence of submorphologic levels of disease is a strong predictor of relapse.2 However, despite well-defined requirements of testing for minimal residual disease (MRD) to predict outcomes in precursor lymphocytic leukemia or lymphoma, in the case of acute myeloid leukemia (AML), the field has lagged in codifying the meaning, methods, and requisite sensitivity of MRD. A recent consensus document from the European LeukemiaNet MRD Working Party focuses on the method requirements for multiparametric flow cytometry (MFC) and for some molecular methods, in particular, real-time quantitative polymerase chain reaction from cDNA for well-documented molecular lesions (e.g., NPM1 mutation and fusions of RUNX1-RUNXT1, CBFB-MYH11, and PML-RARA).

However, not all patients have a discernable aberrant immunophenotype by MFC; 46 to 100 percent of patients have an aberrant, trackable immunophenotype that is heavily dependent on the design of the MFC MRD panels.2 Even more limiting, the well-characterized molecular markers listed above are found in only approximately 40 percent of patients.2 So how can we level the playing field and provide meaningful MRD testing for all AML patients?

Next-generation sequencing (NGS) simultaneously interrogates large numbers of genes with either focused or whole coding sequence coverage as biologically indicated, thereby enabling the examination of AML in a mutation agnostic fashion. In 2015, Dr. Jeffery Klco and colleagues introduced molecular MRD by NGS, demonstrating different patterns of mutation clearance after induction therapy in a small group of patients: 1) all variants cleared; 2) variants partially cleared, but returned at relapse; and 3) a subset of mutations cleared while the remaining persisted during CR without significant change in their variant allele frequency (VAF).3 The persistence of at least one leukemia-specific mutation during morphologic CR was associated with reduced event-free survival (EFS; p=0.003; hazard ratio [HR], 3.32), and reduced overall survival (OS; p=0.02; HR, 2.88).

Recently, Dr. Mojca Jongen-Lavrencic and colleagues demonstrated broad clinical relevance of molecular MRD by NGS using a widely used, commercially available platform at both diagnosis and CR on a large cohort of 482 patients. Using just a small, focused, 54-gene panel, these researchers identified at least one myeloid-associated mutation in 89.2 percent of cases at diagnosis. When samples from these patients were tested in morphologic CR (between day 21 and 4 months postinduction), 51.4 percent had molecular persistence of diagnostic mutations. As in the study by Dr. Klco and colleagues, various patterns of MRD were identified, with most cases with original mutations in DNMT3A, TET2, and ASXL1 retaining those mutations during morphologic CR (78.7%, 54.2%, and 51.6%, respectively), often as the only persistent mutation (67.7% of cases). Collectively, these mutations are referred to as DTA mutations by the authors. In some cases, the VAF of these variants was as high as 47 percent, which is consistent with a heterozygous mutation in essentially all cells. Other mutations, particularly those in the RAS pathway, were more likely to be cleared or present at low VAF (< 2.5% VAF).

The distinct behavior of the DTA mutations corroborates what is known about patterns of clonal evolution in myeloid neoplasms, with age-related clonal hematopoiesis (or clonal hematopoiesis of indeterminate potential [CHIP]) presumably underlying/preceding the leukemic clone in a subset of cases of AML. In those cases, the leukemogenic mutations are subclonal to the CHIP mutations in the aged stem cell. Hence, while the leukemia-specific variants decrease or are cleared in CR, the recovering marrow may repopulate with the CHIP stem cell, which has a clonal advantage. Not surprisingly, the most common CHIP variants, or the “usual suspects,” are DTA mutations, but during the study’s 40-month average follow-up time, these usual suspects did not carry an increased risk of progression. Although the persistence of any mutation, independent of VAF, was associated with an increased four-year relapse rate (RR; p=0.003, training cohort), when DTA and non-DTA variants were separated, only the non-DTA mutations retained the significant association with relapse (p=0.001; Figure, part A), while DTA mutations did not (p=0.29; Figure, part B) unless accompanied by a non-DTA mutation (p=0.002; Figure, part C). The risks associated with persistent non-DTA mutations were confirmed in a separate validation cohort and stood up in multivariate analysis (RR HR, 1.89, p<0.001; EFS HR, 1.64, p=0.001; OS HR, 1.64, p = 0.003) without correlation to type of disease or treatment.

The authors also compared NGS with MFC in 340 patients for whom data were available. Concordant results were found in 69.1 percent of patients, with a 73.3 percent RR in NGS+/MFC+ cases and a 26.7 percent RR in NGS–/MFC– cases. Importantly, MFC identified MRD in 12.1 percent of cases while NGS did not (NGS–/MFC+), and NGS alone identified MRD in 18.8 percent of cases (NGS+/MFC–) with similar intermediate RRs (49.8% and 52.3%, respectively).

As demonstrated by this study, NGS methods for MRD assessment in AML demonstrate clear prognostic value for non-DTA mutations. The study also highlights critical tenants of clonal evolution in AML ontogeny, with underlying CHIP variants (the “usual suspects” or DTA mutations) that are independent of the leukemia disease burden and progression/leukemogenic variants, or non-DTA mutations, whose presence (although not levels/VAFs) predict risk of relapse. Finally, the research highlights the independent prognostic risk associated with MFC and NGS methods. Increases in the number of genes covered by the panel could increase the overall clinical applicability of this method beyond 90 percent of patients. Additional follow-up would be required to understand the long-term significance of DTA mutations in AML patients. Thus, at this time, the following conclusions can be drawn: 1) which mutation matters, 2) how much of the mutation does not appear to matter, and 3) both MFC and NGS methods are required for optimal MRD assessment.


  1. Cheson BD, Bennett JM, Kopecky KJ, et al. Revised recommendations of the International Working Group for Diagnosis, Standardization of Response Criteria, Treatment Outcomes, and Reporting Standards for Therapeutic Trials in Acute Myeloid Leukemia. J Clin Oncol. 2003;21:4642-4649.
  2. Schuurhuis GJ, Heuser M, Freeman S, et al. Minimal/measurable residual disease in AML: a consensus document from the European LeukemiaNet MRD Working Party. Blood. 2018;131:1275-1291.
  3. Klco JM, Miller CA, Griffith M, et al. Association between mutation clearance after induction therapy and outcomes in acute myeloid leukemia. JAMA. 2015;214:811-822.
  4. Jaiswal S, Fontanillas P, Flannick J, et al. Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371:2477-2487.

Conflict of Interests

Dr. Kim indicated no relevant conflicts of interest. back to top