Persistence of Somatic Mutation Spectrum and Clonal Architecture in Chronic Myelomonocytic Leukemia Treated with Hypomethylating Agents

Chronic myelomonocytic leukemia (CMML) is a clonal hematologic neoplasm that combines features of myelodysplasia and myeloproliferative disorder, and carries a poor prognosis due to progression to acute myeloid leukemia or complications of cytopenias. Hematopoietic stem cell transplantation is the only curative option, but many patients are ineligible due to comorbidities and old age. Previous genomic analyses reveal recurrent somatic driver mutations in a variety of genes, including many epigenetic regulators. Hypomethylating agents (HMAs) such as 5-azacitidine induce hematologic responses in approximately 40% of CMML patients, but most patients ultimately experience disease progression.

To understand genetic factors driving response or resistance to HMAs, we deciphered clonal architecture and dynamics in CMML patients treated with 5-azacitidine on a prospective clinical study (NCT01350947; n=11) or in analogy to the study protocol (n=2) over 2 years. Median age at presentation was 70 years. Eleven patients had CMML-1 and two had CMML-2 based on WHO classification. Three patients (23%) achieved complete response (CR), 4 (31%) had partial response (PR), 2 had stable disease (SD), and 1 had no response according to the International Working Group 2006 response criteria. Three patients (23%) progressed after a period of SD.

Whole exome sequencing was performed on mononuclear cells or sorted monocytes from serial blood/bone marrow samples collected at screening, during treatment with 5-azacitidine and at the end of study or progression (n=53, median of 4 samples/patient). Cultured mesenchymal stromal cells (n=5), skin fibroblasts (n=6) or sorted CD3⁺ lymphocytes (n=2) were used as sources of constitutional DNA. Paired-end sequencing was done on Illumina HiSeq 2500 (San Diego, CA) following library preparation and exome enrichment. Fastq files were aligned to the reference human genome (Genome Reference Consortium GRCh37) with Burrows-Wheeler Aligner (bwa_mem) algorithm. Variants were called with Freebayes and annotated by SnpEff. Non-exonic mutations were filtered and somatic mutations were analyzed to study clonal dynamics using SubcloneSeeker [Qiao et al. Genome Biology. 2014;15(8):443]. Copy number variation (CNV) and loss of heterozygosity (LOH) was analyzed by whole genome single nucleotide polymorphism arrays (Infinium Omni2.5-8 v1.3, Illumina).

TET2 (62% of patients), SRSF2 (62%) and ASXL1 (46%) were the most frequently mutated genes. Additional mutations in other genes associated with CMML were detected at lower frequencies: R UNX1 (38%), CBL (31%), KRAS (23%), DNMT3A (15%) and NRAS (15%). No obvious CNV was seen in most patients, but trisomy 8 and LOH in chromosome 11q with uniparental disomy in CBL^c.1151G>A variant were observed in a patient with disease progression.

There was no correlation between the type of mutations and clinical response. Most mutations retained similar levels of variant allele frequency (VAF) in pre- and post-treatment samples, irrespective of clinical response. Models by SubcloneSeeker also demonstrated that overall clonal architecture based on somatic mutations remained unaffected by HMA treatment, and no difference was observed between responders and non-responders to HMAs. While there were modest shifts in relative sizes between individual sub-clones over time, mutations in TET2, SRSF2 or ASXL1 were consistently identified in all the sub-clones, with median VAF of 53%, 51% and 41% respectively, indicating that mutations in these epigenetic regulator genes were acquired early as part of CMML founding clones and their allele burden was not reduced by HMA. Immunophenotyping of post-treatment samples showed re-expansion of normal lymphocytic populations to a median of 24% in responders (n=4) and 13% in non-responders (n=3), suggesting that HMA suppressed myelomonocytic expansion, with or without restoring balanced hematopoiesis.

Our data show that HMA-induced hematological responses in CMML occur despite largely preserved somatic mutation burden and clonal architecture, consistent with a predominantly epigenetic mechanism of action. Whether other therapeutic interventions will restore polyclonal hematopoiesis remains to be determined and may depend primarily on the availability of residual normal hematopoietic stem cells for reconstitution.