Early Drivers of Clonal Hematopoiesis Shape the Subsequent Evolutionary Trajectories of De Novo AML

Background: Acute myeloid leukemia (AML) is a heterogeneous disease characterized by expansion of myeloid progenitor cells and the acquisition of recurrent somatic mutations. Mutations commonly found in AML such as DNMT3A and TET2 are sometimes present in the peripheral blood of otherwise healthy adults – a phenomenon referred to as clonal hematopoiesis (CH). CH-associated mutations are thought to represent the earliest genetic events in the evolution of AML. Genomic studies on samples acquired at diagnosis, remission, and at relapse have demonstrated significant stability of CH mutations following induction chemotherapy. Meanwhile later mutations in genes such as NPM1 and FLT3, have been shown to contract at remission and in the case of FLT3 often are absent at relapse. We sought to understand how early CH mutations influence subsequent evolutionary trajectories throughout remission and relapse in response to induction chemotherapy.

Cohort description: We retrospectively compiled all patients diagnosed with de novo AML at our institution that had two or more next-generation sequencing (NGS) studies, performed at least 30 days apart, on blood or bone marrow specimens. The final cohort comprised 182 patients. All patients were profiled at diagnosis, with 119 sequenced at first remission and 76 sequenced at first relapse. The average age at diagnosis was 58.06 ± 1.03 (mean ± s.e.m.) years, and 53.3% of patients were female. In total, 84.6% of patients received induction chemotherapy with combination anthracycline and nucleoside analog therapy (i.e. “7+3”).

Results: To investigate the genomic changes in AML disease course, we calculated the change in variant allele fraction (VAFs) for all identified variants, evaluating matched samples from individual patients across time. Comparing VAFs at remission and diagnosis, we observed a wide spectrum of VAF changes varying across genes. CBL, IDH1, KRAS, NPM1, NF1, and NRAS variants predominantly had decreased VAFs at the time of remission. FLT3 and NPM1 variants that were initially detected at diagnosis were generally undetectable at remission (variants lost: FLT3: 35/45 NPM1: 26/29). Conversely, only one FLT3 and no NPM1 variants were newly identified at remission compared to diagnosis. Variants identified at diagnosis in the CH-associated genes ASXL1 (1/17 variants), DNMT3A (6/41 variants), and TET2 (10/40 variants) were much less likely to have been eliminated at remission. This is in contrast with FLT3 and NPM1 mutations, which rarely persist at the time of remission. Similar analyses were undertaken at first relapse: FLT3 and NPM1 mutations disappear at the time of remission and reemerge at relapse, consistent with their function as potent AML driver alterations, Meanwhile DNMT3A, TET2 and ASXL1 were generally maintained throughout the disease course in spite of chemotherapy and observed cytologic remission. Overall, these data indicate that CH-associated mutations likely exist synchronously in a preleukemic cell compartment that is not effectively eliminated by AML-directed therapy.

We next sought to determine if early DNMT3A and TET2 mutations subsequently impact the genomic trajectories in the course of AML remission and relapse. Comparing VAFs between diagnosis and relapse for DNMT3A^mut vs TET2^mut AML, we observed that FLT3 and NPM1 variants were similarly split between increased and decreased VAFs at relapse vs diagnosis. Of note, NRAS mutations were both gained and lost in DNMT3A^mut samples between diagnosis and relapse, but no NRAS mutations were identified at all in TET2^mut samples. In contrast, CBL variants were both gained and lost between diagnosis and relapse in TET2^mut samples, but no CBL variants were observed in DNMT3A^mut samples. Finally, we applied the CALDER algorithm to infer phylogenetic relationships from matched longitudinal AML sequencing data in individual patients. We observed that DNMT3A^mut samples were more likely to undergo subclonal swapping from diagnosis to relapse, whereas TET2^mut samples relapsed with a more stable mutation profile.

Conclusions: We demonstrate that DNMT3A and TET2 exhibit distinct constellations of co-occurring genetic alterations at the time of diagnosis and at relapse. Thus, early CH-associated mutations that precede malignant transformation can subsequently shape the evolutionary trajectories of AML through diagnosis, therapy, and relapse.