Interplay between chromosomal alterations and gene mutations shapes the evolutionary trajectory of clonal hematopoiesis

Abstract

Stably acquired mutations in hematopoietic cells represent substrates of selection that may lead to clonal hematopoiesis (CH), a common state in cancer patients that is associated with a heightened risk of leukemia development. Owing to technical and sample size limitations, most CH studies have characterized gene mutations or mosaic chromosomal alterations (mCAs) individually. Here we leverage peripheral blood sequencing data from 32,442 cancer patients to jointly characterize gene mutations (n = 14,789) and mCAs (n = 383) in CH. Recurrent composite genotypes resembling known genetic interactions in leukemia genomes underlie 23% of all detected autosomal alterations, indicating that these selection mechanisms are operative early in clonal evolution. CH with composite genotypes defines a patient group at high risk of leukemia progression (3-year cumulative incidence 14.6%, CI: 7-22%). Multivariable analysis identifies mCA as an independent risk factor for leukemia development (HR = 14, 95% CI: 6-33, P < 0.001). Our results suggest that mCA should be considered in conjunction with gene mutations in the surveillance of patients at risk of hematologic neoplasms.

Fig. 1. Landscape of mCA in the MSK-IMPACT cohort.

a Proportion of subjects with detected mCA increases with age. b Classification of events by copy number states. logR: log₂ ratio of the coverage depth between analyzed sample and normal comparison. logOR: log allelic ratio between major and minor alleles in heterozygous SNP loci. Events are colored by inferred alteration type. c Number of subjects with 1, 2, or 3+ mCAs. d Cell fraction of detected aberrant clones stratified by alteration type. Numbers below the violins indicate the minimum. e Genome-wide distributions of detected mCAs. f Proportion of subjects with mCA among major cancer types. Squares: observed incidence. Blue circles: expected incidence. 95% CI are shown in black. Number of patients are displayed in the middle. Source data are provided as a Source Data file.

Fig. 2. Global characteristics of mCA in relation to gene mutations.

a Overlap between CH cases carrying gene mutations (GMs) and chromosomal alterations (mCAs). Number of subjects are shown in black. b, c The prevalence of mCA increases with mutation burden and VAF. Observed proportion of mCAs within each subgroup is shown as solid dots; 95% CIs generated from a binomial distribution around the observed proportion are shown in blue shades. Gene mutations in regions affected by mCAs are excluded from this analysis. d Age distribution of subjects with mCA only, gene mutations only, and both types of mutations. Unadjusted P values derived from a two-ended Student’s t-test are shown. The lower and upper bounds of boxes denote 25^th (Q1) and 75^th (Q3) percentiles of observed ages, respectively. The lower and upper whiskers indicate the minima (Q1 − 1.5*IQR) and maxima (Q3 + 1.5*IQR). e Volcano plot of q values and odds ratios of gene representation in mCA-positive versus mCA-negative cases (Fisher’s exact test). Genes significantly enriched (FDR < 0.05) in mCA subjects are colored in red. f Frequency of gene mutations significantly enriched in subjects with mCA. Source data are provided as a Source Data file.

Fig. 3. Patterns of co-occurrence between mCAs and gene mutations reveal diverse mechanisms of selection.

a mCA events and their co-occurring gene mutations. Events are colored by alteration types. Only genes that appear more than one time are shown. b Gene loci that exhibited recurrent double hits in cis. Zoom window shows the gene loci affected and the location of the gene mutations; “2” denotes two adjacent mutations in close proximity with each other. Centromere positions are marked in gray. Gene structures are displayed below. c Pairwise associations of co-mutations between specific mCAs and gene mutations. Associations with specific chromosomal alteration types are indicated by the position of the triangle (left: AMP, top: CNLOH, right: DEL, bottom: any). FDR-corrected significance values from pairwise Fisher’s exact tests are indicated by shapes of asterisks. Odds ratios are indicated by the transparency of red shading. The chromosomal location of genes are indicated by black squares. Gene labels are colored by functional groups. Source data are provided as a Source Data file.

Fig. 4. mCA and evolution to hematologic malignancies.

a CH mutations detected at initial blood draw in patients who had a subsequent hematological cancer diagnosis. Patients in each disease category (top bar) were arranged by increasing time to diagnosis. Events are colored by mutation types (=, CNLOH; −, deletion; +, amplification). b Exemplar cases of genomic evolution from CH with detectable mCA to myeloid neoplasms. Gene mutations at the time points of blood draw and diagnosis are shown in the middle, while regions of chromosomal alterations are shown on the left and right, respectively. Solid lines indicate signal median. Aberrant regions are colored in red. BAF, B-allele frequency. Serial blood count indices are shown below, where the shaded areas in blue indicate the period between initial blood collection and MN diagnosis. c Cumulative incidence of leukemia among patients who had detectable mCA only, gene mutation putative driver (PD) only, both, or neither; 95% CIs are shown in shaded ribbons. d Association of CH features with subsequent leukemia diagnosis. Hazard ratios (HR; solid dots), 95% CIs (horizontal bars), and unadjusted P values (above horizontal bars) are derived from a multivariable cause-specific Cox regression model with a 9 month landmark. The HR for 10% increment in gene mutation PD VAF was shown. mCA status was included as a binary indicator. Source data are provided as a Source Data file.