Tracing the Evolution of Clonal Hematopoiesis to AML Using Longitudinal Pre-Diagnosis Blood Samples

The acquisition of somatic mutations in hematopoietic stem and progenitor cells (HSPCs) is increasingly common with age (`clonal hematopoiesis'). If sequential acquisition and clonal expansion of mutations occurs, progression to Acute Myeloid Leukemia (AML) can occur. While the mutational landscape of clonal hematopoiesis antecedent to AML development has been well-defined (Abelson et al. 2018, Desai et al. 2018), the timing of acquisition and growth dynamics of these high-risk mutations remain largely unknown. At what age are these mutations acquired? Are the fitness effects (growth rates) conferred by specific mutations predictable from person-to-person and how do fitness effects change with additive mutations? Are the clonal dynamics that precede AML development characterised by strong competition between clones (clonal interference)?

To answer these questions, we identified 220 women from the United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) who were cancer-free at enrolment but subsequently developed AML during the >12 years follow-up. 50 of these women had annual blood samples collected at multiple time-points pre-AML diagnosis (mean: 5 time-points, range: 2-11). Deep error-corrected duplex sequencing, with a variant allele frequency (VAF) detection limit of 0.1%, was performed on peripheral blood DNA from these women, as well as from age- and timepoint-matched controls who remained blood cancer free. A custom designed next-generation sequencing (NGS) panel was used to enable detection of mutations in 34 clonal hematopoiesis/AML-associated genes, genome-wide mosaic chromosomal alterations (mCAs) and AML-associated translocations.

Having samples from multiple timepoints enabled the fitness effects (growth rates) of mutations to be calculated, as well as the additive effect of further mutations. These growth rates, in combination with insights from evolutionary theory, allowed the acquisition time of many mutations to be estimated, with initiating driver mutations often arising in the first 2 decades of life in the pre-AML cases. Growth trajectory dynamics of co-occurring mutations enabled the clonal composition to be inferred in many instances and revealed linear evolution of successive mutations in some pre-AML cases, but a branching pattern with clear evidence of clonal interference in others. Specific variants, which we have previously identified as 'highly fit' in clonal hematopoiesis (Watson et al. 2020), were significantly enriched in pre-AML cases compared to controls and were often detectable at VAFs >10% more than 5 years pre-diagnosis. NPM1 mutations, which characteristically occur `late' in AML development, could be detected as early as 2 years pre-diagnosis, highlighting the benefit afforded by error-corrected low VAF variant calling, particularly in high-risk individuals.

Our findings, exploiting longitudinal blood samples collected pre-AML combined with an integrated assessment of multiple types of genetic changes, reveal key insights into the evolutionary dynamics of mutations in the years preceding AML development. Understanding which features distinguish pre-malignant from benign clonal evolution is key for risk stratification of individuals with clonal hematopoiesis to allow rational monitoring and identification of individuals that may benefit from early intervention studies.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal