Clonal hematopoiesis and incident hematologic and myeloid malignancy in ARIC

Background: Clonal hematopoiesis (CH), defined using mutations in myeloid-lineage driver genes (N=74 [classic], Bick et al. Nature 2020), is a risk factor for hematologic malignancy (HM). Bernstein et al. ( Nature Genetics 2024) identified 17 additional genes with mutations that are positively selected in the population, and documented a positive association with HM. Continuing to refine CH mutations that convey risk is critical for clinical use and research on modifiable drivers of CH to HM. Thus, we developed a CH-calling pipeline that detects CH from population-based sequencing data using an expanded gene list as well as a variant pathogenicity score based on Google AlphaFold 3. We investigated associations between CH called by the new pipeline and HM risk in a community-based cohort with long follow up. Methods: We conducted a prospective analysis of 9,365 participants in the Atherosclerosis Risk in Communities study aged 44-80 years, without a cancer history, and with whole exome sequencing data. 410 HMs (147 myeloid) were ascertained mainly by cancer registry linkage in a median follow up of 21 years. Follow up is longer than in the original well-known CH studies. With a mean depth of 100x, mutations with a variant allele frequency (VAF)>3.8% were considered for CH. We classified definitive CH as mutations in classic CH genes and truncating mutations in genes additionally identified by Bernstein et al. For missense mutations in the additional genes, we applied AlphaFold 3 to predict pathogenicity and called probable if score≥0.9. We estimated hazard ratios (HR) and 95% confidence intervals (CI) for the association of small (3.8<vaf<10%) and="" large="" (vaf≥10%)="" ch="" (both="" vs="" no="" ch)="" with="" incident="" hm="" myeloid="" malignancy="" adjusting="" for="" age,="" sex,="" race.="" comparison,="" we="" ran="" the="" same="" analyses="" using="" classic="" (bick="" et="" al.="" defined="" vaf≥2%).</vaf<10%)>