Work

Type of prior genotoxic insult determines the genomic characteristics of therapy‐related myeloid neoplasms

Year: 2021

Type: article

Abstract: Therapy-related myeloid neoplasm (tMN) is a well-recognized late complication of genotoxic cancer therapy, and may also rarely arise as a result of therapies used to manage patients with autoimmune diseases or solid organ transplants.1 Clinical outcomes of patients with tMN are worse than patients with de novo myeloid neoplasm (dnMN).2 This is in part due to the higher frequency of older or frail patients in the tMN group, who may not be eligible for intense therapies such as hematopoietic cell transplant.3 However, tMN is also characterized by adverse genomic features, such as complex karyotype, MLL rearrangements, or mutations involving the TP53 gene.4 These features, in particular, TP53 mutations, can render patients less responsive to currently available therapies. Given the inferior outcomes associated with tMN, it is important to identify risk factors associated with tMN in solid tumor patients who require cytotoxic therapy to determine potential interventions that may mitigate their risk of developing tMN. Note, tMN may arise from treatment-induced selection of pre-existing clones harboring mutations (i.e., clonal hematopoiesis [CH]), or as a result of de novo mutations in healthy hematopoietic stem cells.1 A previous study analyzed pre-chemotherapy and post-chemotherapy blood and bone marrow samples from patients with tMN, and reported that the exact TP53 mutation found at the diagnosis of tMN was also present years before the diagnosis, prior to their first chemotherapy.5 A large cross-sectional analysis of 7379 solid tumor patients with CH investigated the factors associated with CH, including smoking and prior therapies.6 Serial sampling data provided evidence that cytotoxic therapy selected clones with mutations in DNA repair genes (i.e., TP53, PPM1D, and CHEK2), which otherwise had limited competitive fitness in the absence of these selective pressures. However, only a minority of patients with CH progress to MN during their lifetime. Therefore, identification of risk factors associated with tMN generation from pre-existing or treatment-induced CH clones is important for future preventative strategies. We hypothesized that age, smoking and different cytotoxic treatment modalities have unique effects on the biology of tMN, which are mediated by divergent molecular mechanisms. To gain insights into the different leukemogenic pathways of tMN, we compared this patient cohort with three other AML cohorts: (1) dnMN patients, (2) patients with prior history of malignancy not treated with cytotoxic therapy (MN with history of prior malignancy, pmMN), and (3) patients with secondary acute myeloid leukemia (sAML), defined as AML arising from an antecedent myeloid malignancy such as myelodysplastic syndrome (MDS) or myeloproliferative neoplasm (MPN). Detailed information of patient selection, study design and statistical analyses are summarized in supplementary methods. Clinical characteristics of 480 patients (75 tMN, 39 pmMN, 256 dnMN, 110 sAML) are summarized in Table S1. Patients with dnMN were younger. Patients with tMN or sAML had higher incidence of adverse risk genomic features when compared to patients with pmMN or dnMN. There were no significant differences in the frequency of patients with smoking history. Median OS was significantly worse in tMN (median, 12 months) and sAML (median, 10 months) groups when compared to patients with pmMN (median, 15 months) and dnMN (median, 20 months) (p < .001). Types of prior malignancies for tMN and pmMN groups, as well as cytotoxic therapies received by tMN patients are summarized in Table S2. A comparative analysis of myeloid mutations between the four groups is shown in Figure 1A. The incidence of TP53 mutation showed differences between these groups (p = .042), with higher frequency observed in patients with tMN (29%) and sAML (30%) when compared to patients with pmMN (18%) and dnMN (18%). Conversely, tMN patients had lower frequency of RUNX1 (4%) and FLT3-ITD (3%) mutations (p = .001). We also compared VAFs for commonly occurring mutations in the bone marrow samples of the four groups (Figure S1). We examined the association between smoking and genomic alterations in the entire cohort of 480 patients (Figure 1B). Mutations involving ASXL1 (OR = 3.39, CI = 1.77–6.48, p < .001) and SRSF2 (OR = 2.35, CI = 1.2–4.59, p = .008) were more common in smokers as compared to non-smokers. However, mutations involving JAK2 (OR = 0.21, CI = 0.07–0.66, p = .002) and SF3B1 (OR = 0.35, CI = 0.14–0.88, p = .02) were less common in smokers compared to non-smokers. We further investigated the relationship between the amount of smoking and relative risk for mutations, which showed an increasing risk for ASXL1 mutation from 1 to 30 pack-year smoking history (Figure 1C). However, we could not demonstrate a similar dose–response relationship between the amount of smoking and relative risk for SRSF2 mutation. Next, we examined the impact of prior genotoxic therapies on the clonal composition of myeloid neoplasms (Figure 1D). In univariable analysis, trisomy 8 was more common in patients treated with etoposide (57 vs. 12%, p < .001) and vinca alkaloids (40 vs. 13.5%, p = .017), while MLL rearrangement was associated with history of vinca alkaloid (30% vs. 3%, p < .001), alkylating agent (20 vs. 2.7%, p < .001), and anthracycline (20 vs. 2.8%, p < .001) therapy. Patients with history of exposure to taxanes had significantly higher incidence of del(5q)/−5 (36.8 vs. 11.6%, p = .001), complex karyotype (57.9 vs. 24.6%, p = .001), and TP53 mutation (47.4 vs. 21.7%, p = 0.009). In addition, patients with history of receiving platinum chemotherapy had higher frequency of del(17p)/−17 (28.6 vs. 8.9%, p = .012), complex karyotype (64.3 vs. 24.8%, p < .001) and TP53 (64.3 vs. 21.5%, p < .001) mutations. There was no significant relationship between a particular genetic subset and prior exposure to antimetabolites or radiation therapy. We performed multivariable logistic regression analysis in the entire cohort of 480 patients with the following covariates: age, smoking, and genotoxic therapies that were found to be significantly associated with a particular mutation in univariable analysis (Figure S2). After adjustment for age and smoking, platinum agent therapy was independently associated with TP53 mutation (OR = 6.81, CI = 1.71–27.1, p = .006) and del(17p) abnormality (OR = 3.79, CI = 1.12–12.7, p = .03). An independent association was also found between exposure to vinca alkaloids and the TET2 mutation (OR = 4.16, CI = 1.11–15.66, p = .03), alkylator therapy and MLL rearrangement (OR = 12.1, CI = 1.26–115.8, p = .03), taxane therapy and del(5q)/−5 (OR = 4.3, CI = 1.25–14.7, p = .02). Accumulating evidence suggests that most patients with dnMN may harbor hematopoietic stem cells carrying MN-associated somatic mutations years before the diagnosis of a full-blown MN. We investigated the impact of smoking and different genotoxic treatment modalities on the molecular characteristics of MNs by studying a large cohort of patients treated at a single center. We show that different genotoxic insults may enrich specific mutations in patients who develop MN, i.e., ASXL1 and SRSF2 mutations with smoking, TP53 mutations and del(17p) with platinum therapy, and MLL rearrangements with vinca alkaloids, alkylating agents and anthracyclines. The association between TET2 and vinca alkaloids has not been reported before, and should be further investigated to determine whether it represents a de novo mutation or expansion of pre-existing TET2-mutated CH clones. Identification of CH as a risk factor for MN and cardiovascular diseases led to the opening of CH clinics dedicated to monitor these individuals, and implement practices to mitigate these risks. While these clinical efforts of preventive medicine are in their infancy, our knowledge on biologic heterogeneity of CH and its variable effects on MN risk have expanded. There are several fundamental questions to be answered: Who should be screened? Which CH mutations pose the highest risk for future MN development in a given individual? The observations from this study may help guide therapeutic interventions in solid tumor patients with pre-existing high-risk CH clones prior to the initiation of cytotoxic therapies. This may include active surveillance of hematologic parameters and CH clones, substitutions of cancer chemotherapy (e.g., avoidance of platinum agents in patients with TP53-mutated CH) or the use of targeted therapies for actionable mutations in CH (e.g., JAK2 or IDH inhibitors). A new era of preventive precision medicine with active surveillance and risk mitigation strategies will likely become part of clinical hematology practice. The authors declare no conflicts of interest or sources of funding. Figure S1. The variant allelic frequencies (VAF) of individual mutations are compared between four groups. *p < 0.05, **p < 0.01, ***p < 0.001. Figure S2. Forest plot showing multivariable logistic regression analysis of the association between myeloid mutations and age, smoking and various genotoxic therapies. Age is factored in as a continuous variable. OR, odds ratio, *p < 0.05, **p < 0.01, ***p < 0.001 Appendix S1: Supplementary Methods Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Source: American Journal of Hematology

DOI: 10.1002/ajh.26170