Transcriptional Profiling Identifies Genes Differentially Regulated by the BCR/ABL Fusion Oncogene.

The translocation t(9;22) is associated with chronic myeloid leukemia (CML) and also occurs in 30% of adult acute lymphoblastic leukemia (ALL). In this study, we analyzed differential gene expression using microarrays to determine if upregulation or downregulation of specific genes may explain the distinct phenotypes. Enriched monoculear cells from 218 adult patients were hybridized to Affymetrix U133 set (A+B) microarrays (discovery set). Resulting lists of differentially expressed genes were further analyzed in an independent set of 110 patients hybridized to U133 Plus 2.0 microarrays (validation set). In a first analysis ALL with t(9;22) (n=34 discovery, n=6 validation) and CML (n=75 discovery, n=49 validation) were included. Various unsupervised data analysis algorithms, e.g. hierarchical clustering and principal component analysis, clearly separated both types of t(9;22) leukemias from each other. A supervised approach, i.e. t-test statistics followed by false discovery rate estimation, identified genes that were significantly differentially expressed. Using the top differentially expressed genes in a classification algorithm (SVM) >97% of the samples were correctly assigned to their classes, both in the discovery and the validation cohort. This set of genes was further examined by pathway analysis (Ingenuity software). Numerous networks point at clear biological differences between both t(9;22) types. Higher expressed genes in CML were connected to networks related to leukotriene metabolism, immune response, integrin signaling, non-selective vesicle transport, or humoral defense mechanisms. This reflects the underlying transcriptional profile of granulation of promyelocytes in CML in contrast to the non-granulated immature ALL blasts. The aggressiveness of acute leukemic blasts is visualized by several pathways where genes with higher expression in t(9;22) positive ALL were aggregated to networks with cellular functions of DNA metabolism and replication, cell cycle progression, and protein biosynthesis. Next analyses were performed to mine for common t(9;22) target genes. CML samples were compared against an equal number of AML with normal karyotype, and t(9;22) ALL against an equal number of c-ALL/Pre-B-ALL without t(9;22). Then both lists of differentially expressed genes were compared for overlapping probe sets. Here, no statistically significant differentially expressed genes were identified as consistently associated with the presence of t(9;22) across the two lineages. In contrast, using a similar strategy where ALL and AML with t(11q23)/MLL were grouped together and were analyzed against various non-MLL positive leukemia subtypes it is possible to identify common t(11q23)/MLL target genes, e.g. a overexpressed HOXA cluster gene signature. This leads to the hypothesis that both types of t(9;22) leukemias, despite an identical underlying chromosomal aberration, trigger different genes involved in BCR/ABL-dependent leukemogenesis. Thus, depending on the cellular background, i.e. myeloid or lymphoid, translocation t(9;22) results in two types of leukemias with fundamental differences in gene expression, clinical course, and the time and quality of response to therapy which is demonstrated also if a BCR/ABL-specific tyrosine kinase inhibitor (e.g. imatinib mesylate) is administered.