Has gene expression profiling improved diagnosis, classification, and outcome prediction in AML?

Gene expression profiling, the simultaneous measurement of the global patterns of gene expression in cells using microarray technologies, has held promise to improve diagnosis, risk classification, and outcome prediction in acute leukemias as well as in other human cancers. But how much has gene expression profiling impacted or improved current diagnosis and risk classification schemes and therapeutic targeting in acute myeloid leukemia (AML)? To date, gene expression profiling of AML has largely confirmed the presence of well-established recurring cytogenetic abnormalities, such as translocations, deletions, point mutations, or normal karyotypes, and has led to the identification of sets of genes reflective of these abnormalities. The use of expression profiling and statistical modeling to develop molecular classifiers that improve outcome prediction beyond traditional prognostic variables or the targeting of patients to specific therapies or transplantation has been less successful. Similarly, expression profiling has not led to the identification of many novel therapeutic targets in AML. Using advanced statistical modeling techniques, our group has focused on expression profiling in adult AML patients with poor risk features in order to identify novel cluster groups and potential targets for therapy in this highly resistant form of disease. To further advance the application of gene expression profiling and other comprehensive genomic and phosphoproteomic techniques to the study of AML, it would be ideal for the NCI Cooperative Groups to register all patients to common or intergroup collaborative clinical trials or registration studies in order to develop large, well-characterized cohorts of uniformly treated patients for future research studies.

Key words: acute myelogenous leukemia, AML, gene expression profiling, molecular classifiers