The dynamic fitness landscape of ageing haematopoiesis through clonal competition

Clonal haematopoiesis (CH) - the existence of large mutant clones in blood - is a prime example of somatic evolution. Yet how evolution shapes CH with age remains to be understood. Here, we show that clonal competition can explain the complex dynamics observed in vivo. In this paradigm, numerous fit clones continually appear and compete, driving an evolving fitness landscape of stem cells. This naturally explains shrinking expanded clones, varying driver efficacy across individuals, and transitions of site frequency spectra with age. Inferences of evolutionary parameters from variant trajectories and site frequency spectra converge to nearly identical estimates of a non-exponential fitness distribution with mean 0.08, and an arrival rate of 2-20 advantageous clones per year. Inferring innate fitnesses from single trajectories, we find that 80% of the variance on identical mutations is explained by clonal competition, with fitness estimates of most common drivers between 0.14 and 0.18. Strikingly, we find clones with much higher fitness to occur only later in life, with an arrival time distribution well-described by a multi-step model of clonal evolution. Overall, a quantitative clonal competition model predicts many aspects of ageing haematopoiesis and allows a personalized identification of high-risk clones potentially important for patient stratification.