Age-related mutations and chronic myelomonocytic leukemia

Abstract

Chronic myelomonocytic leukemia (CMML) is a hematologic malignancy nearly confined to the elderly. Previous studies to determine incidence and prognostic significance of somatic mutations in CMML have relied on candidate gene sequencing, although an unbiased mutational search has not been conducted. As many of the genes commonly mutated in CMML were recently associated with age-related clonal hematopoiesis (ARCH) and aged hematopoiesis is characterized by a myelomonocytic differentiation bias, we hypothesized that CMML and aged hematopoiesis may be closely related. We initially established the somatic mutation landscape of CMML by whole exome sequencing followed by gene-targeted validation. Genes mutated in ⩾10% of patients were SRSF2, TET2, ASXL1, RUNX1, SETBP1, KRAS, EZH2, CBL and NRAS, as well as the novel CMML genes FAT4, ARIH1, DNAH2 and CSMD1. Most CMML patients (71%) had mutations in ⩾2 ARCH genes and 52% had ⩾7 mutations overall. Higher mutation burden was associated with shorter survival. Age-adjusted population incidence and reported ARCH mutation rates are consistent with a model in which clinical CMML ensues when a sufficient number of stochastically acquired age-related mutations has accumulated, suggesting that CMML represents the leukemic conversion of the myelomonocytic-lineage-biased aged hematopoietic system.

Figure 1

Most prevalent genes mutated in CMML. (a) Genes ordered by prevalence of mutation in CMML (N = 69). (b) Median relative expression (% of GUSB) in CD14⁺ cells from 12 unrelated CMML patients. (c) Median relative expression (% of GUSB) in CD34⁺ cells from 12 unrelated CMML patients. Red bars indicate genes not previously associated with CMML.

Figure 2

CMML mutation characteristics and survival. (a) Distribution of number of mutations observed in each CMML patient (N = 69). (b) OS by total mutational burden in CMML patients (N = 48). (c) Concomitance of frequently mutated genes in CMML (N = 69). Light blue indicates tumor variant frequency (TVF) <30%, blue indicates 30% <TVF< 60%, dark blue indicates TVF ⩾60% and red indicates multiple mutations. (d) Concomitance of mutations in 11 recently reported ARCH genes assessed in CMML patients (N = 69).

Figure 3

Multi-hit mutation modeling. (a) CMML incidence rate in the United States and Canada (2005–2011, CINA). (b) Multi-hit mutation modeling under a constant accrual model compared with observed CMML incidence rate. (c) Total CMML mutations versus age at diagnosis (N = 35 having biopsy and diagnosis within 1 year, similar results not shown if within 1 month). Single outlier (only patient recorded with SM-CMML diagnosed at age 36 years) with no mutations not shown. (d) Variant allele frequencies by gene and sample (N = 69). Black, nonsense mutation; red, frameshift indel; blue, nonsynonymous mutation; purple, splice-site mutation; orange, nonframeshift indel.

Figure 4

Gene mutation concomittance and estimated order of appearance. (a) Concomittance of gene mutations in CMML patients (N = 69). Color shading represents estimated logit odds ratio (black and blue indicate mutations are mutually inclusive, red if mutually exclusive). Unadjusted associations with P<0.05 by Fisher’s exact test are boxed in yellow, Bonferroni adjusted association with P<0.05 boxed in green. (b) Normalized rank order of gene mutations (using first gene mutation when multiple, higher rank order values indicate later occurring mutations) in relation to total genes mutated per patient estimated from tumor variant frequencies (N = 69).