Leukemia Fusion Gene Detection in the Clinical Molecular Laboratory Using RNA-Based Targeted Next-Generation Sequencing

Introduction: The diagnosis and risk stratification for patients with B-lymphoblastic leukemia (B-ALL) requires the accurate detection of fusion genes such as BCR-ABL1, ETV6-RUNX1,or Philadelphia-like (Ph-like) B-ALL kinase fusions, or gene rearrangements (e.g. KMT2A). No single assay can detect all relevant alterations, so costly and inefficient testing algorithms that combine karyotyping, fluorescent in situ hybridization (FISH), and reverse transcriptase PCR (RT-PCR) are often required. A comprehensive and sensitive RNA-based next-generation sequencing (NGS) assay that could consolidate diagnostic and prognostic B-ALL gene fusion and rearrangement testing into a single clinical test is an attractive alternative.

Methods: We obtained RNA from 15 clinical specimens collected from 14 patients [11 bloods (2 from the same patient) and 4 bone marrows] with hematologic malignancies and known genomic alterations by RT-PCR, FISH, or cytogenetics, and 1 Ph-positive B-ALL cell line (SUP-B15). Samples harbored either B-ALL-associated alterations [BCR-ABL1 (3), ETV6-RUNX1 (2), intrachromosomal amplification 21 (iAMP21) (2), and KMT2A (2) or PDGFRB (1) rearrangements], or alterations detected in other hematologic malignancies [PML-RARA (2, same patient), RUNX1-RUNX1T1, t(5;9)(q33;q22), t(1;4)(p13;q12), and monosomy 7]. We prepared NGS libraries from extracted total RNA (100 ng) using the Archer® FusionPlex® Heme v2 anchored multiplex PCR-based NGS library with molecular barcoding protocol that targets 553 exons in 87 genes associated with B-ALL and other hematologic malignancies for detection of expressed fusions regardless of gene partner. Illumina paired-end indexed libraries were multiplexed (8 per run) and sequenced on a MiSeq (2x150 bp, v2) in two separate runs. Data were analyzed using vendor-provided virtual-machine based analysis pipelines and custom-developed scripts. ANNOVAR was used for breakpoint annotation, and fusion calls were compared to previous molecular results.

Results: A mean total of 5.1x10⁵ unique and 6.26x10⁴ unique RNA paired-end reads were generated. One sample [t(1;4)(p13;q12)] failed QC due to inadequate RNA reads and was excluded from analysis. NGS detected the fusion or gene rearrangement in 10/11 samples (91%) with mean supporting split read counts of 175 reads (range: 6-1209). NGS detected both ETV6-RUNX1 fusions in samples with 4% and 5% blasts, respectively, defined binding partners for the KMT2A and PDGFRB gene rearrangements (KMT2A-MLLT10, KMT2A-MLLT3, and CCDC88C-PDGFRB), and detected novel P2RY8-CRLF2 fusions in both iAMP21 samples. NGS failed to detect a PML-RARA fusion in a post-treatment low-level RT-PCR-positive, FISH-, flow-, and morphology-negative blood sample, and no fusion was detected in the t(5;9)(q33;q22) sample. A low-level KMT2A-MLLT10 fusion was detected in an ETV6-RUNX1 sample suggesting a minor subclone or false-positive result.

Conlcusions: In summary, RNA-based targeted NGS detects gene fusions and rearrangements comparable to conventional methods, accurately detects fusions in samples with ~5% blasts, and highlights previously unknown fusions and fusion partners. Additional studies are required to establish clinical specimen requirements, limit of detection, sensitivity, and specificity. Experiments are underway to confirm novel fusions, sequence additional recurrent B-ALL alterations (e.g. JAK2, TCF3, and PDGFRA fusions), and assess assay performance in formalin-fixed, paraffin-embedded tissue samples. RNA-based targeted NGS may be a clinically useful method to detect gene fusions and rearrangements in B-ALL and other hematologic malignancies.