Cellular Stressors Contribute to the Expansion of Multiple Hematopoietic Clones of Varying Leukemic Potential

Hematopoietic stem cells (HSCs) are genetically heterogeneous with each HSC possessing its own unique mutations. Some mutations confer a fitness advantage, allowing the HSCs harboring them to clonally expand. The role of different cellular stressors on this expansion remains unknown, as does the nature of the expanding hematopoietic populations.To address these issues, we designed an error-corrected sequencing assay able to detect somatic variants in 46 genes associated with clonal hematopoiesis or AML/MDS at a variant allele frequency (VAF) of 0.1%. We then characterized clonal hematopoiesis after two distinct types of stress: cytotoxic therapy and hematopoietic transplantation.

We first assessed mobilized pheresis samples from three groups: lymphoma or myeloma patients exposed to cytotoxic therapy (n=81), myeloma patients with no such exposure (n=38), and normal donors (n=19). Prior cytotoxic therapy was associated with a significantly increased incidence of clonal hematopoiesis, which was observed in 66/81 (81.5%) of these patients. In fact, most patients with clonal hematopoiesis after cytotoxic therapy had two or more variants detected (median: 3; range: 1-11), suggesting an expansion of multiple clones. Of the 46 genes assessed, we identified six commonly associated with DNA damage response (TP53, PPM1D, ATM, BRCC3, SRCAP, and RAD21) . Variants in these genes were seen in 36/81 (44%) of patients following cytotoxic therapy compared to only 9/57 (16%; P<0.001) of individuals lacking such exposure. In contrast, variants in the other 40 genes were not significantly increased after cytotoxic therapy. These data suggest that cytotoxic therapy provides a fitness advantage to HSCs harboring mutations in certain genes involved in the DNA damage response.

To investigate the nature of expanded mutant clones, we sorted pheresis samples with DNMT3A (n=2), TP53 (n=3), or PPM1D (n=3) variants into myeloid and lymphoid populations. In general, variants were detected in both myeloid and lymphoid lineages, suggesting that they arose in HSCs. However, in most cases with TP53 or PPM1D mutant clones, the percentage of cells with the identified variant was higher in myeloid versus T cells (median: 14.2-fold; range: 2.3 to 88.3-fold). This may reflect the quicker turnover of myeloid compared to lymphoid populations and short time period (< 1 year in most cases) between cytotoxic therapy initiation and pheresis collection.

We next asked how transplantation influences HSC expansion. We sequenced peripheral blood leukocytes collected 6-12 months after transplantation from 40 of our lymphoma patients (with 104 detected pheresis variants). The VAFs of most of these variants did not change significantly with transplant while variants not initially identified in the pheresis samples often became detectable post-transplant. Several trends were observed. Of 46 DNMT3A variants detected before or after transplant, 15 (32.6%) significantly increased ≥ 2-fold in VAF after transplantation, while 2 (4.3%) decreased. Interestingly, all three clones with R882 codon variants expanded. In contrast, of 21 PPM1D variants, only 2 (9.5%) significantly increased in VAF with transplantation, while 7 (33.3%) decreased (P=0.002). These data suggest that clones expanding after cytotoxic therapy are often long-lived and persist after transplant. As with cytotoxic therapy, the behavior of hematopoietic clones following transplant depends, in part, on what mutations they harbor; however, transplantation-induced selection pressure doesn't necessarily favor the same clones as cytotoxic therapy.

Finally, to assess the leukemic potential of expanded clones following cytotoxic therapy, we sequenced bone marrow samples from 134 t-AML/t-MDS patients. They were enriched in TP53 mutations with 34.3% carrying a mutation in this gene. In contrast, variants in other interrogated DNA damage response genes were infrequent in t-AML/t-MDS despite being common after cytotoxic therapy. For example, although variants in PPM1D were detected in a similar percentage of patients following cytotoxic therapy as those in TP53 (17.3% vs. 21.0%), they were only present in 3.0% of t-AML/t-MDS cases (P = 0.008). In total, our data suggests a model in which distinct clones carrying specific mutations expand in response to different cellular stressors. Each clone differs in its leukemic potential due, in part, to the mutations it harbors.