Clonal Hematopoiesis Drives Therapy-Related Myeloid Neoplasms Following Autologous Stem Cell Transplantation and Propagates during Disease Evolution

Introduction: Clonal hematopoiesis (CH) denotes somatic mutations in genes related to myeloid neoplasms present at any variant allele frequency (VAF). Clonal hematopoiesis increases the risk of cardiovascular disease, de novo myeloid neoplasms and therapy-related myeloid neoplasms (tMN). It is well established that CH can be detected years before disease onset. Furthermore, the impact of specific mutations with regards to progression from CH to tMN is currently being unraveled. When exposed to cytoreductive therapy, a proliferative advantage of stem cells with CH over normal hematopoietic stem cells (HSCs) has been demonstrated. However, it remains unclear whether CH is to be considered a mere tMN risk factor, or if the mutations directly impact or even drive the development of tMN. We hypothesized that CH contributes to the development of tMN, and pursued this by investigating the evolution of CH, present in patients with lymphoma and multiple myeloma, prior to autologous stem cell transplantation (ASCT) and at time of tMN diagnosis.

Methods: Patients included were treated with ASCT at the Department of Hematology, Aarhus University Hospital, Denmark, between 1989 and 2016. Inclusion criteria were (i) treatment with ASCT on the indication of a non-myeloid primary disease; (ii) subsequent development of tMN, and (iii) available mononuclear cells (MNCs) at pre-ASCT and time of tMN. All tMN diagnoses were reviewed by an experienced pathologist. Data from time of ASCT of this cohort has previously been reported (Soerensen et al., 2020, PMID: 32150606). Twelve patients with available MNCs at both time points were identified out of 36 tMN patients. Samples (either leukapheresis products or bone marrow MNCs) were subjected to targeted next-generation sequencing, utilizing a 30-gene panel (Myeloid Tumor Solution, SOPHiA Genetics, Saint Sulpice, Switzerland). Variant exclusion criteria were (1) read depth < 3000; (2) VAF < 0.003; (3) variant location outside ±25 nucleotides of coding region; (4) indel present in homopolymeric stretch, and (5) potential germline variants at pre-ASCT with VAF > 0.95 or between 0.45 and 0.55, representing homo- and heterozygosity, and reported in the Exome Aggregation Consortium (ExAC) database.

Results: The cohort included 12 patients with a median age at ASCT of 63 years (range 37­–69) and male predominance (75%). Median time to tMN following ASCT was 3.9 years (range 0.7–15.3), with 7 patients developing therapy-related myelodysplastic syndrome and 5 therapy-related acute myeloid leukemia. A total of 36 and 38 mutations were detected at ASCT and tMN, respectively. Prior to ASCT, DNMT3A (39%) and TET2 (19%) were the most frequently mutated genes, whereas the mutational landscape at tMN proved to be more heterogenous, with TP53 (21%), DNMT3A (18%), RUNX1 (13%) and ASXL1 (13%) comprising the majority of mutated genes. Nine patients (75%) had one or more mutations that could be detected at pre-ASCT as well as at tMN. Seven patients (58%) had CH at pre-ASCT that were present at higher VAF (>0.15 VAF) in bone marrow samples at tMN. Of these, 6 patients had CH at VAF < 0.02 at baseline. We found a total of 14 mutations that were detected at both prior to ASCT and tMN diagnosis, distributed among TP53, SRSF2, DNMT3A, ASXL1, TET2, NRAS and EZH2. Importantly, all clones harboring mutations in non-DNMT3A genes expanded until diagnosis of tMN to VAF > 0.30, with the exception of TET2, which displayed only a modest increase in VAF from 0.01 to 0.15.

Conclusion: In this cohort of patients treated with ASCT and who subsequently developed tMN, we found the majority of patients to harbor CH in HSCs pre-ASCT that, at time of tMN, completely dominated the malignant clone. Our data suggests both a persistency of CH identified in HSCs in peripheral blood prior to ASCT to the leukemic stem cells in bone marrow at tMN diagnosis, as well as an expansion of the clones over time. These findings provide evidence to support the emerging theories that tMNs are instigated by subsets of hematopoietic cells that gain malignant somatic mutations and drive the pathogenesis years before onset disease.