Sequencing an Acute Myeloid Leukemia (AML) Genome with “Next Generation” Technologies.

The genetic events responsible for the initiation and progression of AML are unknown for most patients. Although many important mutations for pathogenesis have been discovered, unbiased genomic approaches will be required to discover all relevant mutations. The development of “Next Generation” massively parallel sequencing approaches (454 and Solexa) have substantially reduced the cost of whole genome sequencing, and have allowed us to initiate a study designed to identify all potentially relevant mutations in a case of M1 AML. The first patient selected for study was a previously healthy 57-year-old Caucasian female who presented with a white blood count of 102,500 (91% blasts); the bone marrow biopsy revealed 100% blasts. Cytogenetics were normal. Tumor and skin (normal) samples were banked with informed consent, using our Washington University IRB-approved protocol that specifically permits whole genome sequencing. Resequencing of 14 genes frequently mutated in AML revealed somatic mutations in FLT3 (ITD) and NPM. Oligomer array-based comparative genomic hybridization using the 2.1 million-oligomer NimbleGen array revealed only one small (<10Kb) region of somatic amplification on chromosome 1p. 500K SNP array analysis of both tumor and skin DNA revealed no regions of “copy number neutral” LOH. Expression analysis of tumor RNA on the Affymetrix Hu133+2 platform revealed a signature that is typical for most patients with M1 AML. The patient’s bone marrow RNA was used to prepare a polyA-primed cDNA library that was enzymatically normalized (to bias towards less abundant mRNAs). Two runs of this material on the Solexa platform yielded approximately 1.5 billion bases of sequence (8x coverage). These reads were stringently aligned to micro-repeat masked versions of the human transcriptome and the human genome (Build 36) and then scored for single-nucleotide and small (1–3bp) indel variants. We further evaluated the coverage extent, depth, and gap size across the transcriptome, using custom scripts and the Synamatix Synamer^™ algorithm. ∼4,500 sequence variants have been identified thusfar, which are currently being classified (e.g. known SNPs, non-synonymous variants, etc.) for further analysis. Non-amplified tumor and skin DNA samples were also prepared for sequencing. 25 Solexa runs have been completed on the tumor genome (∼19 billion bases; 6.3X coverage). ‘Actual’ coverage is being assessed by comparing the Solexa data to informative (i.e. heterozygous) SNPs detected in the patient’s tumor and/or skin samples on the 500K SNP array. At this time, 18% of the 135,167 informative SNPs have been detected in Solexa reads. Within a few months, we will complete ∼25x coverage of the patient’s tumor cDNA and genomic DNA. The patient’s skin DNA will then be sequenced to identify private SNPs. Potential somatic mutations will be verified with an established PCR-based resequencing pipeline, and the frequencies of validated somatic mutations will be further determined in 93 additional cases of AML. Using comprehensive array-based genomic platforms and Next Generation sequencing, we hope to define all of the inherited and acquired mutations that were relevant for AML pathogenesis in this patient.