Pattern of Mutational Changes in Patients with Chronic Phase CML Who Are Treated with Frontline TKI and Transform to Blast Phase CML

Introduction: CML patients (pts) who transform to blast phase (CML-BP) or accelerated phase (AP) from a chronic phase (CML-CP) have a poor prognosis and exhibit a poor response to therapy with TKI. Increased usage of newer TKI's has remarkably reduced the incidence of CML-BP, however, the genetic mechanisms involved in the transformation remain unclear. In this study, using paired whole-exome sequencing, we compared the pattern of gene mutations and chromosomal allelic imbalances (CAI) between CML-CP and CML-BP in patients who received front-line TKI therapy.

Methods: We analysed 13 CML patients who received front-line TKI therapy in CML-CP and eventually transformed to CML-BP. In 7 of 13 patients, sufficient DNA was available for sequencing both at the time of CP and AP/BP while in the remaining 6 patients, DNA was only available at 1 time point, either BP (n=4) or CP (n=2). Whole exome sequencing (WES) was performed using Agilent exome capture kit on Illumina HiSeq. For exome mutation analysis, read alignment was done using BWA and GATK, and variants were called using GATK's UnifiedGenotyper. CAI profiles were generated using hapLOHseq. For CAI, chromosomal arm events were called if they spanned at least 20% of the genomic region of a chromosomal arm. Somatic status was assigned to variants after filtering for potential germline variants using variants identified in the ExAC project. For assessing biological processes driving each cancer's mutation profile, mutation signatures were deconvoluted using the R package deconstructSigs into a set of previously identified, weighted set of COSMIC mutation signatures. For reporting of specific mutations, only somatic variants reported in the COSMIC database were selected.

Results: The median age of the study group at the time of BP was 54 years (range, 24 to 73); 53% were male. Seven pts had e13a2 and 6 had e14a2 BCR-ABL transcripts. Eleven pts (85%) were treated with imatinib and 2 pts received nilotinib as their initial therapy. The median time to transformation to BP was 19 months (range, 3.5-63). The immunophenotype of blasts was myeloid in 9 (69%), lymphoid in 3 (23%) and mixed lineage in 1 pts (8%). The cytogenetic abnormalities detected by conventional karyotyping at the time of BP included: classic Ph alone in 5 pts, 2 pts each with variant Ph, double Ph and isochromosome 17q and one pt each had inversion 3 and another with cryptic Ph. The median OS was 16 months; 9 pts had died at the time of last follow up. Both CML-CP and CML-BP samples exhibited deficiencies in DNA-mismatch repair and DNA double-strand break repair through mutation signature analyses. The median number of mutations at the time of BP was 374. The most frequently mutated genes detected exclusively during BP included ABL1 and RUNX1 . ABL1 mutation was significantly associated with CML-BP (5/10; p=0.012). Similarly suggests that RUNX1 mutation may be associated with CML-BP (4/10) although not at statistical significance (p=0.14; odds ratio of CML-BP = 6 (95% CI: 0.46 - 359). CAI profiles show that BP samples have increased genomic instability. CAI was frequent at the time of BP. CAI was significantly associated with BP (8/10; p = 0.0019; odds ratio of BP = 30 (95% CI: 2.3 - 1911). Chromosomes 7 and 8 were the most frequently altered chromosomes (each altered in 3/10 BP samples) while 0/3 (0%) AP and 1/8 (12.5%) CP samples had CAI. Gene mutation results are shown in Figure 1.

Conclusions: Bypaired exome sequencing of CML patients who received frontline TKI therapy, our analyses suggest that CML-BP is associated with (1) acquisition of additional gene mutations during BP, frequently in ABL1 and RUNX1 and (2) increased chromosomal instability. Defects in DNA mismatch repair and double-strand break repair were identified in all CML phases. This may suggest that these deficient repair processes, evident in the CP phase, are early contributors to the transition to CML-BP. This hypothesis is supported by the increased chromosomal instability observed in the BP samples that may result from these cellular defects. As such, identification of these repair deficiencies may have an early predictive potential in detecting CML patients who can potentially transform to blast phase. We are performing additional studies using targeted NGS and WES in samples that did not undergo BP transformation to substantiate these hypotheses. Further studies are needed to explore the modalities to target these alterations in CML.

View largeDownload slide

View largeDownload slide

Close modal