Long-Term Clonal Dynamics upon Allogeneic Haematopoietic Stem Cell Transplantation Revealed Using Somatic Mutations As Clonal Tracking Marks in HSC Donor and Recipient Pairs

Allogeneic haematopoietic stem cell transplantation (HSCT) is a potentially curative treatment for over 40,000 patients/year in Europe and the US alone. However, substantial treatment-related mortality and morbidity, as well as risks of disease relapse give a survival rate of about 50% and leave considerable room for improvement. Despite being an established treatment for over 50 years, fundamental questions remain regarding its biology. For example, how many of the frequently >100 million transplanted CD34+ cells are true haematopoietic stem cells (HSCs), determined by long-term engraftment and contribution to multi-lineage hematopoiesis (long-term engrafting HSCs [LTE-HSCs])? What are the mutational consequences for transplanted HSCs given their proliferation and potential mutagenic insults in the post-transplant period? Most recently, the discovery of clonal haematopoiesis (CH) has raised interest in the interaction between this and HSCT. Do such clones further expand during HSCT? This may potentially lead to the devastating complication of donor-cell leukemia or other CH-related risks, e.g. cardiovascular disease. Recently, some studies have addressed this using targeted sequencing panels for myeloid cancer genes. However, many clonal expansions in normal blood are not driven by mutations in such genes, with evidence suggesting that the set of potential 'driver' genes is much larger than currently recognized (Poon et al, bioRXiv 2020).

Advances in HSC tracking methodologies - using naturally-occurring somatic mutations as clonal markers (Lee-Six et al, Nature 2018) - provide a powerful tool to simultaneously address these questions. Whole-genome sequencing of hundreds of single-cell derived haematopoietic stem and progenitor cell (HSPC) colonies from a single individual is used to compile a complete set of somatic mutations in each colony founder cell, and the pattern of shared mutations amongst cells used to infer their phylogeny or 'family tree'. The constant rate of mutation acquisition during post-development life allows estimation of the timing of mutation acquisition. Using phylodynamic approaches borrowed from pathogen biology, patterns of branching points can be used to infer important parameters such as the size of population 'bottlenecks' (in this context the number of LTE-HSCs), and the growth dynamics of expanded clones.

We selected 7 donor/ recipient (D/ R) pairs who had undergone HSCT 9-31 years previously. For each individual (D and R), peripheral blood CD34+ HSPC-derived colonies were grown on methylcellulose medium. Whole-genome sequencing (WGS) was performed on 100-300 colonies per individual - a total of 2,278 genomes. Mutations were called using established pipelines, then filtered to remove artefacts, germline variants, and in vitro mutations leaving only somatically-acquired mutations. Phylogenies for each D/R pair were inferred, using a maximum parsimony algorithm. Mutational signatures were extracted using a hierarchical dirichlet process. D/R phylogenies were compared using metrics of phylogenetic diversity. Clonal fractions of expanded clones in D/R were compared. Approximate Bayesian computation was used to estimate numbers of LTE-HSCs.

Our results reveal that HSCT engraftment is remarkably polyclonal, with thousands of transplanted HSCs (in most cases >5,000) actively contributing to haematopoiesis decades after transplant. HSCs suffer little consequence in terms of their somatic mutation burden. Recipient haematopoiesis showed decreased clonal diversity compared to their donors with a mean 20% decrease of the Shannon's Diversity Index. This may partly result from increased selective pressures during HSCT. Intriguingly, several DNMT3A-driven expansions seen in donors had lower clonal fractions in recipients. Conversely, clones with >1 driver mutation (e.g. DNMT3A/CHEK2) showed larger expansions in recipients compared to donors, despite originating in the donor. DNMT3A mutations frequently originated in early development - in one case occurring in utero.

We demonstrate the power of applying a novel clonal tracking approach to HSCT, for the first time giving a detailed picture of the clonal dynamics of engraftment. Overall, our findings are reassuring from a safety perspective, but the different clonal composition in recipients merits further investigation to better understand the factors involved.