Whole-Transcriptome Sequencing In Chronic Myeloid Leukemia Reveals Novel Gene Mutations That May Be Associated with Disease Pathogenesis and Progression

Abstract 885 Philadelphia-positive (Ph+) chronic myeloid leukemia (CML) has always been regarded as a genetically homogeneous disease. However, the fact that a proportion of patients (pts), especially in the high Sokal risk setting, fail tyrosine kinase inhibitor therapy and progress to blast crisis (BC) suggests that a certain degree of heterogeneity exists. It can be hypothesized that genetic factors additional to the Ph+ chromosome may be present in these pts. To address this issue, we are currently using massively parallel sequencing to perform a qualitative and quantitative survey of the whole transcriptome of Ph+ CML cells at diagnosis and at progression to BC. Results are being integrated with genome-wide search for copy number alterations by Affymetrix SNP 6.0 arrays. We used a Solexa Illumina Genome Analyzer to scan the transcriptome of a CML patient at the time of diagnosis, at the time of remission (major molecular response) and at the time of progression from chronic phase (CP) to lymphoid blast crisis (BC). Both custom scripts and published algorithms were used for read alignment against the human reference genome, for single nucleotide variant (SNV) calling, for identification of alternative splicings and fusion transcripts, and for digital gene expression profiling. Comparison of the SNVs identified in the diagnosis and relapse samples with the SNVs detected in the remission sample – representing inherited sequence variants not specific for the Ph+ clone – allowed the identification of eight missense mutations at diagnosis affecting the coding sequences of AMPD3 (encoding adenosine monophosphate deaminase 3), SUCNR1 (succinate receptor 1), FANCD2 (Fanconi anemia, complementation group D2), INCENP (inner centromere protein), BSPRY (B-box and SPRY domain containing), HEXDC (hexosaminidase containing), NUDT9 (ADP-ribose diphosphatase) and KIAA2018 (encoding a protein with predicted DNA binding and transcriptional regulation activity) genes. Six of these mutations (FANCD2, INCENP, BSPRY, HEXDC, NUDT9) were also detected in the Ph+ clone re-emerged at the time of disease progression, together with seven additional missense mutations affecting the coding sequences of IDH2 (isocitrate dehydrogenase isoform 2), DECR1 (2,4-dienoyl CoA reductase 1), C4Orf14 (mitochondrial nitric oxide synthase), MRM1 (mitochondrial rRNA methyltransferase 1), PRKD2 (protein kinase D2), TCHP (mitostatin) and ABL1 genes. Digital gene expression analysis showed downregulation of SUCNR1, that might be a consequence of the P292A mutation we detected. IDH2, MRM1, AMPD3, and KIAA2018 mutations were found in additional pts. The IDH2 R140Q mutation was detected in 3/75 (4%) myeloid BC, 1/31 (3.2%) lymphoid BC, 0/34 Ph+ ALL and 0/23 Philadelphia-negative (Ph-) ALL pts. The MRM1 C120S mutation was found in 6/70 (9%) additional BC pts (2 lymphoid and 4 myeloid). AMPD3 and KIAA2018 genes were found to harbour the same point mutations (N334S and S1818G, respectively) in 1 out of 20 additional CP patients analyzed. Massively parallel sequencing of the sample collected at diagnosis also revealed that the Bcr-Abl kinase domain was already harbouring point mutations at low levels (E308D, A344G, R386S) but not the T315I that was selected at the time of disease progression. Point mutations in untraslated regions where miRNAs are known to bind were also detected, and are currently under validation. Digital gene expression profiling comparing progression to diagnosis showed significant expression changes including upregulation of 134 genes and downregulation of 88 genes. In particular, we observed an upregulation of the B-cell developmental factor PAX5, its interactor Lef-1 and its targets IRF4, BLNK, Bik, EBF1, CD79A, CD79B, CD19, VpreB1, VpreB3, BOB1, RAG1 and RAG2; upregulation of PAX9; upregulation of WNT3A, WNT9A, GLI3 and downregulation of SFRP1, resulting in aberrant activation of the Wnt signalling pathway. In summary, our preliminary data highlighted putative key genes whose deregulation may be recurrent in a subset of CML patients and may be linked to disease pathogenesis or progression. Their actual role in CML is currently being exlored. Massively parallel sequencing of additional patients is ongoing. Supported by European LeukemiaNet, AIL, AIRC, Fondazione Del Monte di Bologna e Ravenna, FIRB 2006, PRIN 2008, Ateneo RFO grants. Disclosures: Baccarani: NOVARTIS: Honoraria; BRISTOL MYERS SQUIBB: Honoraria. Martinelli:Novartis: Consultancy, Honoraria; BMS: Consultancy, Honoraria; Pfizer: Consultancy.