Abstract
Abstract 1278
Introduction:
Acute myeloid leukemia (AML) represents a heterogeneous group of malignancies with great variability in response to therapy. In recent years, an increasing list of molecular markers with prognostic significance in AML has been identified; nonetheless, new prognostic markers and therapeutic targets are still needed. The aim of this study was to identify and verify fusion transcripts using RNA-Sequencing (RNA-Seq) that would be otherwise undetectable by conventional karyotyping.
Methods:
Transcriptome Sequence data is generated by high-throughput short-read RNA-Seq performed for each AML sample on the Illumina HiSeq. Poly(A) RNA is captured with poly(T) magnetic beads, fragmented, copied to cDNA libraries with reverse transcriptase and random primers. Each library is subjected to 50-cycle paired-end sequencing on the Illumina HiSeq at Hudson Alpha. Filtered Fastq files are processed with TopHat-Fusion [Kim2011,Trapnell2009] alignment software to discover cryptic fusions in RNA-Seq data without relying on known, annotated models. This process yielded an average of 20 million alignable reads per sample. Cord blood blast cell transcripts are also processed and serve as normal controls. A series of filtering steps eliminate junctions commonly found to be in error. Filtered junctions found in at least 3 AML samples and no normal controls are retained as AML-associated candidate junctions. Visual curation of candidates is performed using Integrative Genomics Viewer. Candidate fusions were verified by RT-PCR amplification of the AML-associated fusions in the index cases. Fusion transcript product, as well as the break point junction was verified by Sanger sequencing
Results:
Diagnostic specimens from 70 patients with de novo AML that included patients with normal karyotype (NK, N=31), core-binding factor (CBF) AML (N=33) and other (N=6) were sequenced. Age at diagnosis varied from 10 months to 69 years (Median 12 years). White blood cell count (WBC) and blast percentage were 49×109/L (range, 2.4 to 496×109/L) and 78% (40% to 100%), respectively. Bioinformatic evaluation of the RNA-Seq data revealed 67 high-value novel fusions that were not detected by conventional karyotyping: 54 (80.6%) were intra- and 13 (19.4%) inter-chromosomal junctions. The number of novel translocations varied in different cytogenetic groups, with 22 novel fusions detected in those with NK (16 intra and 6 inter-chromosomal junctions), 37 CBF (32 intra and 5 inter-chromosomal junctions) and 8 in “other” (6 intra and 2 inter-chromosomal junctions). Thirteen novel fusions (19.4%) were found in at least 2 or more screened-patients: two (15.4%) inter- and 11 (84.6%) intra-chromosomal junctions. Median number of fusions identified per patient was 2 (range, 1 to 6).
Novel fusions involving PDGFR-β gene were identified in two patients, each with a different translocation partner (G3BP1 and ETV6, which was an intra and inter-chromosomal fusions, respectively). Sequencing of the fusion transcript junctions verified the fusion junctions and demonstrated in frame fusions of G3BP1 and ETV6 to the kinase domain coding region of PDGFR-β, identical junction to that seen in cases of imatinib sensitive idiopathic hypereosinophilic syndrome (IHES). Frequency validation in 100 adult and 100 pediatric cases identified one additional patient with G3BP1-PDGFR-β. Cryptic NUP98/NSD1 was identified and verified in two patients with normal karyotype as well as NUP98/HOXD13 translocation in one patient. Frequency determination of NUP98/NSD1 demonstrated prevalence of 7.8% in patients with NK, and that of 13% in patients with FLT3/ITD. Patients who harbored both NUP98/NSD1 fusion and FLT3/ITD had a dismal remission induction rate (CR rate in FLT3/ITD with and without NUP98/NSD1 was 28% vs. 73%; p=0.002).
Conclusion:
Our data show the applicability of RNA-Seq as a tool to discover cryptic fusion transcripts in AML. These novel fusions may define new independent prognostic markers and potential therapeutic targets for patients with this highly treatment-resistant disease.
Disclosures:
No relevant conflicts of interest to declare.
Comprehensive evaluation of fusion transcript detection algorithms and a meta-caller to combine top performing methods in paired-end RNA-seq data