Genetic Biomarkers in Acute Myeloid Leukemia

Will the Promise of Improving Treatment Outcomes Be Realized?

Jay Yang; Charles A Schiffer


Expert Rev Hematol. 2012;5(4):395-407. 

In This Article

Expert Commentary

Recent advances in prognostic factors in AML have led to changes in the diagnostic and treatment algorithms for newly diagnosed patients. The routine incorporation of NPM1, FLT3 and perhaps CEBPA-DM mutational testing can now be strongly recommended in addition to a standard karyotype evaluation. Patients with normal cytogenetics and mutated NPM1 or CEBPA-DM should initially be treated with intensive chemotherapy alone, sparing the potential toxicities of SCT. Even older patients with NPM1 mutations should be offered similar therapy unless prohibited by other medical reasons. Younger patients with favorable risk and IR-AML by cytogenetics should be offered induction with cytarabine and higher-dose DNR. By contrast, patients with known FLT3-ITD mutations are not likely to benefit from the dose intensification of anthracyclines during induction. They should also be considered for SCT in first complete remission, although there are no prospective data addressing this point.

Despite this progress, there is a sense that we are rapidly becoming experts in defining the prognosis but not necessarily in improving the treatment of AML. In the clinic there is a dearth of therapies that are able to take advantage of the leukemia's genetic abnormalities. This lag between novel discoveries and the fruition of effective therapies is to be somewhat expected. What is sobering, however, is that CBF leukemias were discovered more than 25 years ago as recurrent chromosomal changes, but this has not yet translated into new targeted agents for this disease. Although many of the advances in cancer therapy have been made from targeting oncogenes, less progress has been made in targeting mutations that affect transcription or inhibit differentiation. The paradigm of differentiating therapies that has been so successful in APL has been difficult to replicate in other subtypes of AML.

Even the application of novel mutations as prognostic factors comes with limitations. Most recurrent genetic aberrations are found in only a minority of patients with AML. Subgroups of patients will be further subdivided on the basis of the presence or absence of other mutations, not to mention the over- or under-expression of other genes. The categorization of AML will become increasingly complex. The hope is that one day these subgroups may be homogeneous enough so that we may refine their prognostic precision in order to provide uniquely tractable targets for therapy. The fear is that as subgroups of patients become smaller, it will become more difficult to draw firm conclusions about prognosis based on the limited numbers represented in clinical trials. The inconsistent and sometimes contradictory results regarding the prognostic impact of some of these biomarkers are a reminder of this problem.

Going forward, there are also considerable operational challenges in conducting appropriate clinical trials necessary for validating prognostic markers and new targeted therapies. Much of the available prognostic data have been the result of retrospective data mining. Prospective studies are clearly warranted, but even randomized Phase III trials with large patient numbers may not be adequately powered to detect significant differences in outcomes in small subgroups of patients. Furthermore, such trials will be near impossible to perform without unprecedented cooperation on an international scale. Drawing definitive conclusions from Phase II studies is fraught with danger due to a lack of a comparative control arm. Smaller but more nimble multiarm randomized Phase II trials have been suggested as a more effective compromise.[114]