Somatic Mutations in Commonly Mutated Genes in Myeloid Malignancies May Preexist or Arise in the Course of Chronic Myeloid Leukemia - Different Scenarios of Progression Revealed By Targeted Next-Generation Sequencing

Chronic myeloid leukemia (CML) is currently a true chronic disease for majority of patients who achieve durable remission with tyrosine kinase inhibitors (TKI) and remain in chronic phase (CML-CP) for several years. However, a number of patients develop TKI resistance and may progress to blastic phase (CML-BP) which represent major therapeutic challenge. Although CML-BP resembles acute myeloid leukemia (AML) in many aspects, including selected genetic aberrations, pathogenesis of CML progression to acute phase is not fully understood. Therefore there is a strong rationale for studying underlying mechanisms of progression to CML-BP.

To gain comprehensive insight into genetic background of CML progression, we performed targeted high-throughput sequencing of sequential DNA samples from patients experiencing a treatment failure and progression to CML-BP. We inquired whether gene mutations previously described in human malignancies (with special focus on genes involved in leukemogenesis) were gained during progression or if they existed already at diagnosis. Sequential samples were collected from 5 patients, who progressed to CML-BP, despite TKI therapy and for whom samples were available from both diagnosis and progression. Roche NimbleGen SeqCap EZ custom-capture was used to acquire exonic sequences of approximately 1000 cancer-related genes, comprising genes from commercially available cancer panels (such as Illumina TrueSight Cancer, NimbleGen Comprehensive Cancer Panel) and also genes altered in hematological malignancies as reported in current literature. Common variants (>1%) gathered in ESP6500 and 1000 genomes projects and our internal exome database were filtered out.

Patient CML-1 was diagnosed in CML-BP and relapsed within 13 months despite treatment with 2 TKIs. Both at the diagnosis and progression, a DNMT3A p. P799T mutation was detected with similar allele frequency. Patient CML-2 was diagnosed in CML-CP, underwent alloHSCT within 6 months from diagnosis and achieved short-term remission. However, within few months she relapsed with lymphoid BP and despite short-lived responses to combined chemotherapy and TKI, she developed 2^nd BP and 3^rd BP with BCR-ABL1 p. Y253H and p. T315I mutations, respectively. Strikingly, p. R320* RUNX1 mutation was present at all four time points (diagnosis in CML-CP and all three BPs), though at diagnosis frequency of this mutation was approx. 1%. Patient CML-3 wasdiagnosed in CML-BP and was treated with TKIs plus chemotherapy. After 25 months patient developed resistance and progressed. We detected IDH1 p. R132S mutation, typical for AML, exclusively in the BP sample from progression, while it was absent in diagnostic sample. Patient CML-4 was diagnosed in CML-CP, developed resistance to 3 TKIs and finally progressed to CML-BP after 5 years. P. Y183C IDH1 mutation was detected in the BP sample. Patient CML-5 was diagnosed in CP and despite treatment with TKI, progressed to CML-BP within 3,5 years. P. R139L and p. R166L RUNX1 mutations were detected only in the sample from CML-BP. All detected mutations were confirmed independently by Sanger sequencing or deep amplicon sequencing for low frequency variants. Detected mutations are summarized in table 1.

Our analysis of sequential samples from CML patients proves that mutations in genes commonly mutated in myeloid malignancies (DNMT3A, IDH1, RUNX1) may be preexisting or may arise during progression, independently of BCR-ABL1 mutation. With regard to preexisting mutations, this may lead to clonal evolution of the disease. Importantly, such preexisting mutations could have been missed in the previous studies, presented or published, employing next-generation sequencing strategy, since most of those studies used algorithms to detect newly gained aberrations in CML-BP as compared to CML-CP.