Defining Patient-Specific Risk in Acute Myeloid Leukemia

In 2013, the primary clinical decision for allogeneic hematopoietic stem-cell transplantation eligible patients with acute myeloid leukemia (AML) is dichotomous, consolidation chemotherapy or stem-cell transplantation. The timing of this decision is critical because applying transplantation at relapse is associated with a poor outcome for patients whose disease is not associated with a core binding factor abnormality. This decision is informed by a growing number of diseaseand patient-specific factors. Terwijn et al reported their results from the Dutch Hemato-Oncology Group/Swiss Group for Clinical and Epidemiological Cancer Research (HOVON/SAKK) AML 42A study, a prospective study evaluating the prognostic value of assessing minimal residual disease (MRD) in adult patients with AML by multicolor flow cytometry (MFC). Disease-specific prognostication has guided AML therapy since the advent of Giemsa-banding technology and several prognostic systems became available. Though there are differences among these systems, they have some similarities in two large groups of patients with divergent outcomes. Young, good-risk AML patients have an average overall survival rate of approximately 65%, with standard induction and consolidation therapy. Poor-risk patients have a less than 20% chance of survival with chemotherapy alone and are referred for an allograft when in a first remission. Despite allogeneic stem-cell transplantation, overall survival remains poor for patients in this group, and enrollment on a clinical trial should be a priority. While there is general agreement regarding treatment approaches for the goodand poor-risk cohorts, there is significant variation in the approaches to the remaining patients. Prognostic systems undergo regular updating to include the growing list of mutations that have been reported to have prognostic relevance and function to move a patient to either a better or worse risk group. With the list of recurring mutations growing, each mutation alone and in combination with others will be queried regarding their potential to move patients in either direction. Add to the mix the growing body of mRNAand miRNA-expression profiling data and the complexity of risk assessment becomes a bit overwhelming. While disease-specific prognostic systems are informative for clinical trial design and population studies, they are limited when counseling a patient in the clinic. Therapeutic response is the summation of all “relevant cellular resistance mechanisms, pharmacokinetic resistance, dosage and compliance, and other unknown factors affecting the effectiveness of treatment.” Complete remission (CR) is the lowest hurdle in assessing response; the accepted blast count for morphologic CR is less than 5% marrow myeloblasts with resolution of extramedullary leukemia and near normalization of blood cell counts. This standard was established approximately 60 years ago, and clinicians have recognized its limited predictive value. The 2004 International Working Group revised the reporting criteria, adding cytogenetic CR and molecular CR categories. These categories, now a standard approach in the management of chronic myelogenous leukemia, are not typically applied in trials for AML in the United States. The sensitivity of fluorescent in situ hybridization is approximately 1% and most commercial mutation assays, with the exception of BCR/ABL and PML/RARA, were developed primarily for diagnostic purposes. A growing number of studies applied MFC and quantitative polymerase chain reaction (qPCR) to quantify MRD after AML treatment. Measurement of MRD provides independent prognostic information during or following treatment in adult and childhood AML using either marrow aspirate or blood specimens. Most reports involved retrospective analyses of previously collected samples and many lacked sufficient patient numbers. Also, prior validation of the assays was not included and cutoff levels for MRD-positive versus MRD-negative were not defined prestudy. In the recent past, the efforts of the European Leukemia Network and others have demonstrated that MRD studies could be performed with the rigor necessary to be employed prospectively in large collaborative trials. A growing number of these trials are being reported. Terwijn et al prospectively assessed MRD in patients treated on the HOVON/SAKK AML 42A study. They employed MFC using a predefined cutoff of 0.1% and analyzed samples following the first and second induction. The threshold of 0.1% is in the range shown to be prognostic in other studies. The choice of these time points corresponds to decision branchpoints in AML treatment algorithms and thus is informative for future trials. The analyses were performed at several labs and were highly predictive of relapse. Interestingly, an assessment of MRD after the second induction divided the intermediate-risk group into two cohorts with different outcomes. The majority of intermediate-risk patients fell into the MRD group that identified patients who might be spared the risks associated with transplantation. When specimens were analyzed by both MFC and qPCR, the results were comparable for most samples. To date, few studies have performed MRD assessment prospectively with a clinical decision made based on the results of this information in real time. These few studies also are limited by patient JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 31 NUMBER 31 NOVEMBER 1 2013