MRD in AML: it's time to face the FACS

Improvements in the treatment outcome for childhood AML have lagged behind those achieved for childhood ALL, in part because of a dearth of well-defined prognostic factors. Whereas the prognostic impact of minimal residual disease (MRD) measurements in ALL is well established, relatively little is known about the impact of residual disease detection in childhood AML. In this issue, Sievers and colleagues (page 3398) have narrowed this gap in knowledge. Extending their previous work, the investigators used multidimensional flow cytometry to evaluate bone marrow specimens of 252 pediatric patients with AML for the presence of leukemic cells. Overall, 41 (16%) patients had immunophenotypic evidence of leukemia at one or more time points before relapse. Multivariate analysis demonstrated that residual disease was the most significant prognostic factor. Compared with patients who had no detectable leukemia by flow cytometry, those who were positive for residual disease had significantly increased risks of relapse and death. It should be noted, however, that the authors analyzed “responsive” patients, defined as those with fewer than 30% morphologically detectable blasts after one course of therapy. Because some of the patients may have had microscopically detectable blasts at the time of analysis (eg, 5% to 30% blasts), it might be more relevant to limit the analysis to patients with fewer than 5% marrow blasts, thereby identifying and analyzing only patients with “minimal” or occult residual disease.

The investigators used a standard panel of antibodies to identify phenotypic abnormalities in bone marrow specimens. Although this approach is practical and allowed the investigators to study 100% of samples at a stated sensitivity of 0.5%, its sensitivity may actually be quite variable. In fact, because the immunophenotypes of leukemic cells obtained from each patient at the time of diagnosis were not determined, it is not possible to know the sensitivity of their assay in each case. This might explain the observations that only 16% of patients were positive for residual leukemia and only one-third of subsequent relapses were predicted. Nevertheless, the authors clearly demonstrated the prognostic impact of residual disease in childhood AML: patients with detectable disease had a 3-year overall survival estimate of 41%, compared with 69% for those without measurable leukemia.

I share the authors' hope that early intensification of therapy for patients with detectable residual disease before overt relapse occurs will improve their outcome. But for this strategy to have a clinical impact, it will be important to study all patients early in therapy. Although the investigators successfully analyzed all bone marrow samples received by their laboratory, specimens were submitted from only 48% of patients before intensification therapy (ie, at a time when increased intensification could potentially improve outcome). In addition, optimal use of residual disease studies may require more sensitive assays. The use of patient-specific antibody panels, which considerably increase the sensitivity of residual disease studies, may permit the identification of more patients who are at high risk of relapse. The application of such methods to all patients early in therapy should provide a more accurate estimate of the leukemic burden at the time of morphologic remission and a rational means of selecting postremission therapy. Moreover, as suggested by the results of the present study, sequential monitoring of residual disease, which has proven to be useful in ALL, may also add prognostic information in AML by identifying patients who have slow regression of disease or increasing levels of disease and are thus at a particularly high risk of treatment failure.