Identification of Progression-Associated Mutations in Chronic Myeloid Leukemia (CML) By Comparative Genomic Analysis

Introduction: CML progression from chronic phase (CP) to blast crisis (BC) has been related to the acquisition of additional mutations, but detailed information on the genes involved is lacking. Here we aimed to identify progression-associated driver mutations and potentially targetable pathways.

Patients and methods: Bone marrow samples from 58 patients (14 BC-CML, 2 AP and 42 newly diagnosed CP) were collected from Helsinki University Hospital and National Cancer Institute, Cairo University. For CML-CP patients, CD34+ sorted cells were used to facilitate the detection of low variant allele frequency mutations. Genetic alterations were identified by whole exome sequencing (WES) (8 BC, 1 AP and 15 CP samples) and deep targeted sequencing (Nimblegen gene panel, SeqCap EZ comprehensive cancer Design) (10 BC, 1 AP, and 32 CP diagnostic samples and 15 follow up samples) comprising target regions for 578 driver genes with a documented association with cancer. RNA sequencing was used to identify progression-associated gene expression profiles (8 BC and 5 CP diagnostic CD34+ samples).

Results: Analysis of data from all 16 advanced patients and 24 CP-diagnostic samples revealed 80 mutations in 62 driver genes in BC samples compared to 69 mutations in 54 driver genes in CP samples. All variants had high oncogenic potential (score >28) as measured by OncoScore (Piazza, R. et al . Sci. Rep. 2017). The median number of mutations per patient (WES data only) was two-fold higher in BC-CML (10, range: 7-14, n=9) than in CP samples (5, range: 1-10, n=11) (p=0.0027). In BC-CML patients, core-binding factor aberrations (RUNX1 mutations and inv(16)) were the most recurrent (n=6) followed by ABL1 (n=5) and BCOR mutations (n=4). Mutations were also discovered in ASXL1 (n=3), NF1, ALK, SETD2, SF3B2, MSH2 (n=2 each), FLT3-TKD, IKZF1 and IDH2 (n=1 each) . In CP-CML, recurrent mutations were detected in ASXL1, TET2 (n=4 each), NOTCH3 (n=3), TET1, KMT2C and PTPRD (n=2 each). CNV data analysis of BC samples identified progression-associated events including +8, inv(16) (n=2 each), i(17q) with P53 loss, LOH in 21q and IKZF1 deletion (n=1 each). Driver mutations known to be associated with hematological malignancies (RUNX1, FLT3 and BCOR) were almost exclusively limited to BC (RUNX1 mutation in one CP-patient who progressed to BC within a year). Epigenetic modifiers (ASXL1, TET1, TET2, and IDH2) represented a major subset of mutated genes in both phases. Mutations in protein tyrosine phosphatase (PTP) genes (putative negative regulators of JAK-STAT signaling) were common in CP patients (n=5), and they were associated with poor response or progression as none of the patients carrying these mutations achieved major molecular remission within 12 months. In CP patients, higher mutation burden (median mutation number from WES data) was associated with poor treatment responses (failure to achieve MMR at 12 months). The median mutation number was almost 2-fold higher in poor responders (8, range: 3-10, n= 5) than in good responders (4.5, range: 2-5, n=4). Variant allele frequencies of putative driver mutations (ex; RUNX1, MECOM, PTP genes) in poorly responding CP patients were tracked in follow up samples, and were found to correlate with BCR-ABL1 levels confirming their presence in the leukemic clone.

Conclusion:Comparative analysis of CP-CML and BC-CML patients revealed a group of genes associated with a particular disease phase (e.g. RUNX1, FLT3, and BCOR) of which some could represent therapeutic targets for future studies. In CP-CML patients, mutations in PTP genes were found to correlate with poor response warranting their screening at the time of diagnosis.