Mitochondrial heteroplasmy improves risk prediction for myeloid neoplasms

Abstract

AbstractThe most well-known pathogenic risk factor for myeloid neoplasms (MN) is clonal hematopoiesis of indeterminate potential (CHIP)1. However, MN can develop in CHIP negative individuals, indicating that additional markers of clonal expansion might also be informative. Heteroplasmy, defined as the presence of mitochondrial DNA (mtDNA) mutations in a subset of cellular mtDNA, has been associated with hematological malignancies2and could represent a marker of clonal expansion3. However, the relationship between heteroplamsy and CHIP, as well as its association with the incidence of MN in the general population is not known. In this study, we explored the association between somatic mtDNA and nuclear DNA (nDNA) mutations (mito-nuclear interaction), its impact on MN incidence, and whether its inclusion to the latest CHIP-based MN prediction algorithm could improve risk stratification in over 440,000 participants in the UK Biobank and Atherosclerosis Risk in Communities (ARIC) studies. We found that heteroplasmy count and heteroplasmic variants predicted to be more deleterious were enriched in individuals with CHIP, particularly in those with significantly expanded clones (VAF ≥20%), with more than one CHIP mutation, and with mutations in the spliceosome machinery. Individuals with both heteroplasmy and CHIP were more likely to develop MN than participants with either entity alone. Furthermore, we found a significant and independent association of predicted pathogenic effect of heteroplasmic variants with incident MN, suggesting a causal role of mtDNA variations in MN pathogenesis, even in the absence of CHIP. Finally, incorporating heteroplasmy into an existing risk score model for MN in individuals with CHIP significantly improved the sensitivity by 13.1% and identified 34.4% more cases in the high-risk group (10-year risk ≥10%). In sum, our findings suggest that heteroplasmy, in addition to being a marker of clonal expansion, may be a causal biomarker of MN development, with clinical utility in the general population.