Ultra-sensitive Sequencing Identifies High Prevalence of Clonal Hematopoiesis-Associated Mutations throughout Adult Life

Abstract

Clonal hematopoiesis results from somatic mutations in hematopoietic stem cells, which give an advantage to mutant cells, driving their clonal expansion and potentially leading to leukemia. The acquisition of clonal hematopoiesis-driver mutations (CHDMs) occurs with normal aging and these mutations have been detected in more than 10% of individuals ≥65 years. We aimed to examine the prevalence and characteristics of CHDMs throughout adult life. We developed a targeted re-sequencing assay combining high-throughput with ultra-high sensitivity based on single-molecule molecular inversion probes (smMIPs). Using smMIPs, we screened more than 100 loci for CHDMs in more than 2,000 blood DNA samples from population controls between 20 and 69 years of age. Loci screened included 40 regions known to drive clonal hematopoiesis when mutated and 64 novel candidate loci. We identified 224 somatic mutations throughout our cohort, of which 216 were coding mutations in known driver genes (DNMT3A, JAK2, GNAS, TET2, and ASXL1), including 196 point mutations and 20 indels. Our assay's improved sensitivity allowed us to detect mutations with variant allele frequencies as low as 0.001. CHDMs were identified in more than 20% of individuals 60 to 69 years of age and in 3% of individuals 20 to 29 years of age, approximately double the previously reported prevalence despite screening a limited set of loci. Our findings support the occurrence of clonal hematopoiesis-associated mutations as a widespread mechanism linked with aging, suggesting that mosaicism as a result of clonal evolution of cells harboring somatic mutations is a universal mechanism occurring at all ages in healthy humans.

Keywords: DNMT3A; age-related clonal hematopoiesis; mosaicism; smMIPs; somatic mutations; ultra-sensitive sequencing.

Figure 1

Validation of Mutations by Restriction Digestion and Re-sequencing (A) Mutations identified in DNMT3A Arg882 selected for additional validation by non-sequencing-based method. (B) Scheme showing recognition site for restriction digestion enzyme TauI in the genomic sequence corresponding to DNMT3A Arg882. Mutations chr2:25457242C>T and chr2:25457243G>A (hg19) leading to p.Arg882His and p.Arg882Cys, respectively, are marked below with red arrows. (C) Size analysis of restriction digestion of DNMT3A PCR products. Lanes 1 to 5 represent samples with mutations with different VAFs. C1 is a control with false positive signal for a G>A mutation at chr2:25457243, as determined statistically. C2 is a control with no DNMT3A mutation. (D) Gel trace of size analysis of digestion, with sample 4 on the left and C2 on the right. The peak corresponding to the full-size product is marked with a red triangle for both samples. Note that C2 present a small peak at 577 bp, corresponding to undigested PCR products due to digestion enzyme saturation. (E) Sequencing results of digested PCR product for sample 5. We obtain a higher ratio of mutation to wild-type reads than in the original sample due to digestion of the wild-type product.

Figure 2

Prevalence of Clonal Hematopoiesis-Driver Mutations (A) Prevalence and distribution of clonal hematopoiesis-driver mutations identified in healthy individuals aged between 20 and 69 years of age. (B) Prevalence of mutations in DNMT3A per age group. Hotspot in DNTM3A are defined as residues in which five or more mutations were identified in our cohort and include Arg326, Arg729, Tyr735, Arg736, Trp860, and Arg882. All other missense, loss-of-function, and indels are included in the non-hotspot mutations.

Figure 3

Clonal Hematopoiesis-Driver Mutations per Age Group (A) Prevalence of clonal hematopoiesis per age group, defined as the frequency of individuals with one or more CHDMs per decade of age. (B) Proportion of the population per age group with CHDMs. (C) Mutation variant allele fraction per age of the individual in which the mutation was identified. The y axis is in logarithmic scale. No significant correlation is observed between the age of the individual and the VAF of the mutation identified.

Figure 4

Type of Mutation Change per Age Group