Acute myeloid leukemia with limited differentiation (AML-LD) has been identified as an entity based on its specific immunophenotype (MPO+CD33+CD117(+) and lack of other lineage-specific markers) and was demonstrated to invariably carry an nucleophosmin (NPM1) type A mutation and to lack cytogenetic abnormalities in most cases (Leukemia 2009;23:1361-4). To further clarify the biologic background of AML-LD we performed microarray analysis (MA, array: Affymetrix HG-U133 Plus 2.0) in 27 such cases and compared gene expression profiles (GEP) to cases with similar biology or immunophenotype, respectively: AML with NPM1 type A mutation and normal karyotype (NPM1-A, n=24), AML with NPM1 mutation other than type A and normal karyotype (NPM1-other, n=12), AML without NPM1 mutation and with normal karyotype (AML-NK, n=30), and acute promyelocytic leukemia (APL, n=15). By standard diagnostic criteria, AML-LD was identical to NPM1-A with the exception of the above-mentioned specific immunophenotype. Moreover, the differences of AML-LD to NPM1-other and to AML-NK in addition only included a different type of NPM1 mutation or its lack, respectively. APL was included in the analysis because it shares some features in the immunophenotype with AML-LD, including lack of HLA-DR expression, strong MPO expression, and weak side-scatter (SSC) signal. Thus, all groups used for comparison to AML-LD had genetic or immunophenotypic similarities to AML-LD. GEP, however, yielded a signature which was highly specific for AML-LD. Thus, in pair-wise comparisons using the top-500 differentially expressed genes unsupervised cluster analysis resulted in a complete separation of AML-LD from the respective comparitive groups. Cluster analyses including AML-LD, NPM1-A, NPM1-other, and AML-NK again demonstrated a complete separation of AML-LD cases from other cases while the latter cases were not separated from each other according to the groups as defined above. A similar result was obtained when all five groups were analyzed with AML-LD cases clustering together while APL cases almost completely clustered together and remaining groups did not separate. Accordingly, applying a classification algorithm with a 10-fold leave-one-out cross-validation and 100-fold resampling approach using the top 500 differentially expressed genes AML-LD were correctly classified with 100% accuracy when analyzing all five groups as well as when considering four groups without APL, respectively. In order to clarify the pathogenetic background of AML-LD we next analyzed the top-500 genes that were differentially expressed between AML-LD and NPM1-A, NPM1-other, and AML-NK using Ingenuity Pathway Analysis (Version IPA 7.5). 432 probe sets were mapped to gene functions by the software. 306 genes were categorized into networks with associated functions such as cancer, nucleic acid metabolism, cell signaling, and development disorders. The top network was centered on CTNNB1 (beta-catenin) which was overexpressed in AML-LD. Beta-catenin, an adherens junction protein and member of the Wnt/β-catenin signaling pathway, plays a key role in the development of breast cancer, colorectal cancer, ovarian carcinomas, or hepatocellular carcinoma as well as in melanoma cell lines. It is implicated in the pathogenesis of Wilms′ tumors where mutations lead to overexpression of the gene. In the present analysis, additional overexpressed candidates of the Wnt/β-catenin pathway were CSNK1G1, CSNK1G2, and SOX4. Furthermore, several genes of the Notch signaling pathway were detected overexpressed in AML-LD: presenilin, APH1, APH1A, PSENEN, and PSEN1. NOTCH proteins are single-pass transmembrane receptors, reported to be frequently mutated in T-ALL. CD99 was also found overexpressed in AML-LD which is known to occur in T-ALL. Interestingly, MALAT1 was found among the most differentially expressed genes with lower expression exclusively in AML-LD. MALAT1, however, has not been mapped to any network. While it has been found overexpressed in various carcinomas and sarcomas it has not yet been described underexpressed in a specific malignancy. Therefore, we sequenced MALAT1 in 8 cases but no mutations were identified. In conclusion, GEP proves AML-LD as a distinct subentity of AML and reveals targets from the Wnt/β-catenin and Notch signaling pathways as candidate genes for underlying mutations implicated in the pathogenesis of AML-LD.
Kern:MLL Munich Leukemia Laboratory: Equity Ownership. Grossmann:MLL Munich Leukemia Laboratory: Employment. Kohlmann:MLL Munich Leukemia Laboratory: Employment. Schnittger:MLL Munich Leukemia Laboratory: Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Equity Ownership.
Asterisk with author names denotes non-ASH members.