Uncovering AML Subtypes and Their Biology: Applying Genome-Wide Profiling.

Abstract

Classification of AML subtypes based on chromosomal abnormalities is clinically relevant, as several of them are associated with distinct clinical behavior. For instance, recurring reciprocal translocations t(15;17), t(8;21), or inv(16)(p13q22)/t(16;16) predict favorable prognosis, whereas other chromosomal aberrations, such as t(6;9), −7/7q- or inv(3;3)/t(3;3), are associated with inferior outcome. In approximately 40–50% of AML patients no chromosomal aberrations have been identified. In ~75% of this latter patient group molecular lesions have been found, particularly in CEBPA, NPM1, or FLT3 (FLT3-ITD). Importantly, CEBPA or NPM1 single mutants associate with a favorable outcome, whereas cases that carry NPM1 as well as FLT3-ITD mutations correspond with unfavorable prognosis. In approximately 25% of AML no (cyto)genetic abnormalities have been found so far. We have carried out gene expression profiling (GEP) to uncover novel AML subtypes and obtain insight on the biology of these leukemias. By applying GEP to a large cohort of human AML (n>550), we identified subgroups of patients with a gene expression signature similar to that of cases with known chromosomal or molecular aberrations. We will discuss two examples. 1) AML cases that carry chromosome 3q26 lesions show high expression of EVI1, the gene that resides in this locus. Gene expression analysis revealed a significant number of patients without 3q26 cytogenetic lesions but with high EVI1 expression. Importantly, although those cases appeared frequently cytogenetically normal, fluorescent in situ hybridization (FISH) using EVI1 genomic probes revealed complex 3q26 abnormalities in ~50% of those cases. In the other half, no aberrations have been found so far. These data appear clinically relevant, as high EVI1 expression levels associate with poor treatment outcome. Consequently, we have developed an EVI1 specific Q-PCR assay to be used as a diagnostic test in AML. 2) Using gene expression profiling, we identified a novel AML subgroup that showed a gene expression profile highly similar to that of CEBPAmut AML cases. A closer look at those leukemias revealed that the CEBPA gene was switched off by promoter hypermethylation, explaining the high similarity with CEBPAmut AML. In depth analysis of the GEP data further demonstrated that this particular AML subgroup carried myeloid as well as T-lymphoid features, in particular high levels of the T cell marker CD7. Interestingly, hematopoietic stem cells isolated from conditional CEBPA knock out mice showed high CD7 expression as well, which could be reverted by reintroduction of CEBPA. These results indicate that absence of CEBPA in the leukemia subgroup explains, at least in part, the unique biological features identified by GEP. Finally, using a novel genome-wide promoter methylation assay we found that CEBPA methylation was associated with aberrant hypermethylation of many genes. Taken together, these data suggest the existence of a previously unrecognized subtype of AML, which results from aberrant epigenetic programming.