INTRODUCTION: Clonal hematopoiesis (CH) is a clonal expansion in blood cells driven by somatic mutations and has been associated with hematologic malignancy, coronary artery disease and death. Drivers of CH include rare somatic point mutations in leukemogenic genes, somatic structural variation, and somatic chromosomal aneuploidy. Whether recurrent somatic variation exists outside specific leukemogenic loci and its relationship to CH is presently unknown.

OBJECTIVES: Somatic variant calls on diverse samples from the TOPMed consortium enables an unbiased survey of somatic variation in 46,080 deep whole genomes. Here we identify a novel class of recurrently observed somatic variants. We use the burden of recurrent somatic variants to define a new clonal phenomenon and describe its genetic causes and clinical consequences.
METHODS: We identified somatic point mutations using Mutect2. We performed stringent filtering to exclude germline variants including any germline variant called in TOPMed Freeze 5 and any variant with allele fraction > 35%. We identified variants associated with age using elastic-net regression. We performed a GWAS of the recurrent variant burden using SAIGE to identify the germline basis of this phenomenon, including age, study, and the first 10 genetic ancestry principal components as covariates. We derived a biological aging clock that estimates chronological age from the variant allele fraction of specific recurrent somatic variants.

RESULTS: We identified 87,608 recurrent somatic variants (present in >= 50 individuals and in >= 5 studies) in 46,080 samples, of which 1,192 are associated with age. An increase in age by a decade is associated with 28.4 additional mutations (95% CI: 27.3-29.5). We found that CH with rare driver mutations is not associated with recurrent variant burden after adjusting for age, suggesting the clonal phenomena are distinct. We performed mutation signature analysis and found an enrichment of COSMIC signatures SBS40 and SBS3, implicating DNA damage repair. A GWAS of recurrent variant burden identifies 13 genome-wide significant loci, suggesting germline variation influences acquisition of recurrent somatic variants. We defined age acceleration as the residual between the aging clock and chronological age. A standard deviation increase in age acceleration was associated with 35% increased risk of incident CAD (pvalue = 2.9e-15) and increased osteoprotegerin (pvalue = 6e-3).

CONCLUSION: Our compendium of recurrent somatic variation identifies a novel clonal phenomenon distinct from CH with leukemogenic mutations. We identified predisposing molecular pathways and clinical consequences.