Therapy-selected clonal hematopoiesis and its role in myeloid neoplasms

Abstract

Therapy-related myeloid neoplasms (t-MN) account for approximately 10-15% of all myeloid neoplasms and are associated with poor prognosis. Genomic characterization of t-MN to date has been limited in comparison to the considerable sequencing efforts performed for de novo myeloid neoplasms. Until recently, targeted deep sequencing (TDS) or whole exome sequencing (WES) have been the primary technologies utilized and thus limited the ability to explore the landscape of structural variants and mutational signatures. In the past decade, population-level studies have identified clonal hematopoiesis as a risk factor for the development of myeloid neoplasms. However, emerging research on clonal hematopoiesis as a risk factor for developing t-MN is evolving, and much is unknown about the progression of CH to t-MN. In this work, we will review the current knowledge of the genomic landscape of t-MN, discuss background knowledge of clonal hematopoiesis gained from studies of de novo myeloid neoplasms, and examine the recent literature studying the role of therapeutic selection of CH and its evolution under the effects of antineoplastic therapy. Finally, we will discuss the potential implications on current clinical practice and the areas of focus needed for future research into therapy-selected clonal hematopoiesis in myeloid neoplasms.

Graphical Abstract

Introduction

Advances in anti-neoplastic therapy have led to more prolonged cancer survival overall. Still, an increased risk of therapy-related myeloid neoplasms (t-MN) remains, leading to an increased focus on understanding the genetic and environmental contributors to t-MN. t-MN are a well-recognized and heterogeneous group of clonal disorders occurring after treatment with anti-neoplastic therapy and have been demonstrated to clinically manifest along the spectrum of acute myeloid leukemia (AML), myelodysplastic syndromes (MDS), or myeloproliferative neoplasms (MDS/MPN) with characteristic clinical features [1-3]. t-MN have been shown to develop 3-8 years following conventional chemotherapeutics and often carry a worse prognosis than the primary malignancy that preceded them [4]. Recently, in the treatment of cancers like hematological malignancies, the overall 10- and 20-year survival rates have increased substantially [5], arising from the development of new therapeutic options [6, 7]. The improvements in treating these primary malignancies more broadly have led to the estimation that there will be 18 million cancer survivors in 2022 in the United States alone [8, 9]. As patients live longer, the risk for t-MN is projected to increase [7, 10-13]. t-MN currently comprise 10-20% of all malignant myeloid diseases [8, 14]. The prognosis of t-MN is significantly worse compared to de novo disease, with a median survival of 8 months after diagnosis and 5-year survival rates at less than 10%, further highlighting the need for enhanced focus on determining the underlyjng mechanisms and individual risk for t-MN development [4, 15-17]

Clonal hematopoiesis (CH) has been identified as a precursor to myeloid malignancies [18-22], and it has been demonstrated that t-MN is more likely to manifest in patients with antecedent clonal hematopoiesis [17, 18, 23-25]. CH is the expansion of a single population of blood cells with an advantageous somatic mutation(s) without readily identifiable clinical manifestation [26, 27]. Across all cancer patients, CH has been observed in ~25% of cancer patients, with mutations in leukemia-specific driver genes being present in 5% of these [28, 29]. Additionally, it is essential to note that the prevalence of CH varies widely after different modalities of treatment, and characterizations of the relationships between therapeutics and the development of t-MN will be discussed further during the discussion of t-MN related genomics. CH is a heterogeneous condition with different mutations conferring different growth dynamics in the face of selective pressures and may evolve from an asymptomatic state into an aggressive malignancy [3, 21, 24-26]. Unfortunately, some of these clonal populations are associated with several health issues, not limited to t-MN [4, 16, 18, 30, 31]. Many investigators have thus directed their attention toward understanding the relationship of CH with cancer therapy and whether therapies induce clonal hematopoiesis or solely provide selective pressure for fit clones. This question remains a topic of active investigation.

t-MN can arise from a broad spectrum of background treatments, including cytotoxic chemotherapy, ionizing radiation, immunosuppressive therapy, or after prolonged exposure to an environmental carcinogen. Historically, individual therapies have been associated with unique genomic abnormalities. Alkylating agents have been associated with deletions on 5q and 7q, which have been observed in 76% of all t-MN cases in patients with an abnormal karyotype [16, 32]. Foundational genetic karyotyping work described that 5-20% of patients with therapy-related AML exhibited genetic rearrangements in the form of translocations around bands 11q23 and 21q22 [33-35], which are associated with topoisomerase inhibitor therapy. However, given the genomic complexity and idiosyncratic mutational burden that distinct chemotherapies may introduce, comprehensive analyses of the breadth and resolution of copy number abnormalities and structural variation have been limited by studies focusing primarily on whole exome sequencing or targeted sequencing data [36-38]. In this review, we will further discuss the genetics of t-MN, the role of clonal hematopoiesis on their development, and describe the findings resulting from whole genome sequencing, which aids in characterizing the interaction between intrinsic and extrinsic factors in the development of t-MN.

A study conducted by Takahashi et al. discovered an increased risk for those with clonal hematopoiesis to develop into t-MN, evidenced, in part, by the presence of CH within peripheral blood samples in 74% of t-MN patients prior to primary cancer treatment, and the observation that the incidence of t-MN was four times higher in patients with pre-existing clonal hematopoiesis compared to those without it [3]. An increased risk of t-MN development due to CH has been corroborated in several other studies [23, 39-41], demonstrating that CH is a critical element in the development of t-MN. It can also be seen, through investigations by Ok et al., that the genetic alterations underlying t-MN are distinct when compared to de novo myeloid malignancies [42]. The main question has been whether chemotherapy may introduce driver events that lead to t-MN or if it merely acts as a selection bottleneck for CH with advantageous mutations. Furthermore, chemotherapy exposure can select and eventually increase the genomic complexity of CH clones, potentially leading to the perpetuation of chemotherapy-resistant cell populations and a higher subsequent risk of developing t-MN following cancer-directed therapy [39, 43, 44]. Before this is addressed, it is crucial to explore the fundamental characteristics of CH in order to provide additional context to its relevance in discussions on the development of t-MN.

Genomics of therapy-related myeloid neoplasms

To date, characterization of the genomic contributors to the development of t-MN has primarily included TDS and WES. Several characteristic genetic abnormalities have been identified in the nearly 90% of patients with t-MN [45]. In this section, we will discuss the prior literature describing the genomic landscape of t-MN.

Dating back to the use of karyotyping to characterize gross chromosomal aneuploidies, deletions within chromosomes 5 and 7 have been observed as recurrent events in t-MN. In a study conducted by Shih et al., it was observed that within a cohort of 12 patients with TP53 mutations, 83% of patients had either monosomy 5 or deletions within chromosome 5, with half also showing a concurrent loss of regions of chromosome 7 [45]. Historically, alkylating agents are associated with an increase in chromosome 5 or 7 aberrations in patients presenting with t-MN [46-49]. It has been hypothesized that the G-C-rich regions of these chromosomes are targeted explicitly for methylation of O⁶ guanines, which results in DNA breaks and subsequent deletions or truncations [50]. Consequently, a loss of part, or all, of Chromosomes 5 or 7 has been postulated as a potential driver of t-MN pathogenesis [46]. However, these observations have been mainly based on cytogenetic technologies that miss the breadth and depth of chemotherapy-induced mutational events.

Other gene-treatment interactions have been identified in t-MN with numerous studies linking topoisomerase II (TOP2) inhibitors to fusions and translocations of 11q23, the MLL (KMT2A) locus [51-53]. By blocking the ligation step of DNA replication, therapeutics targeting TOP2 induce single and double-strand breakages and cell death [54, 55]; however, cells that repair these breakages might survive and, during repair, accumulate translocations. If these translocations occur within the MLL gene, a critical regulator of HOX gene expression [56-58], they confer an increased risk of developing a secondary malignancy [53, 59-61]. In support of this, a study by Cho et al., which aimed at characterizing the genetic origins of t-MN, found that 11q23 cytogenic abnormalities were the most observed genetic aberration among 14 patients with cytogenic abnormalities [62].

Several hypotheses have been posited that provide a mechanism by which mutations resulting from TOP2 inhibitors could directly lead to the development of t-MN. During mutational events within the MLL gene, it has been postulated that fusion events may occur, which are one of the most prominent drivers of leukemia [46, 63]. From this, only fusion events that conferred a selective advantage as an oncogene would be carried on, which has been demonstrated to lead to leukemias [54, 64]. Explanation or reasoning for the enrichment of MLL rearrangements with TOP2 inhibition is unknown thus far, as the mechanism by which MLL fusion contributes to leukemias as a driver event has yet to be characterized. While studies into TOP2 inhibitor-induced translocations and genetic aberrations provided some insight into a potential mechanism for the development of t-MN, it also paved the way for further investigations into other critical genetic regions postulated to have a role in t-MN development as well.

Apart from aneuploidies and structural variations, some single nucleotide variants in driver genes are enriched in t-MN. TP53 mutations have been observed to be present with significantly higher frequency in t-MN than in de novo myeloid neoplasms and are hypothesized to be critically important in disease pathogenesis [42, 45, 65]. In the previously mentioned study by Shih et al., TP53 mutations were the most common genetic anomaly, occurring in 21% of 38 patients presenting with t-MN [45]. In another, conducted by Ok et al., TP53 mutations were observed at a rate of 35.7% within t-MDS and 33.3% within t-AML. More notably, these were significantly increased within these patients when compared with de novo disease, with a 17.7% increase observed within MDS and a 12.8% increase within AML [42]. Additionally, increases in the frequency of TP53 mutations have been observed consistently within t-MN at higher frequencies when compared to de novo malignancies [25, 38, 65, 66]. Furthermore, recent evidence has suggested that TP53 is an integral component in the development of t-MN [67].

While findings from previous sequencing efforts are significant and have provided a foundation for the genomic landscape of t-MN, there have been some limitations. For instance, while de novo AML can more easily be characterized by targeted deep sequencing and conventional cytogenetics due to their relative genomic simplicity, these techniques are limited in their ability to identify the extent of genomic complexity in t-MN [19, 68]. Additionally, many endogenous and exogenous mutational processes leave evidence of their activity across the genomes of exposed cells, known as mutational signatures [69, 70]. Certain chemotherapies cause distinct DNA damage in preferred base sequence contexts that can be linked to a discrete period of clinical exposure [71]. For example, platinum-based drugs are associated with the single base substitution signatures SBS31 and SBS35 [72-74], and melphalan is associated with SBS-MM1 [37, 75, 76]. Figure 1 provides a timeline that chronicles the discovery of the melphalan SBS-MM1 signature. Others, such as 5-fluorouracil, ganciclovir (SBSA), temozolomide (SBS11), and ionizing radiation (SBS2 and SBS13), have all similarly been linked to several characteristic mutational signatures [77-80]. Therefore, the measurement of mutational signatures serves as unique markers of specific chemotherapy and can assist in characterizing the evolution of t-MN following cancer therapy. WGS has thus emerged as an essential tool to fully understand the relationship between t-MN and their preceding cancer therapies [81].

Figure 1. Timeline of investigations into the development of the SBS-MM1, a predominant signature associated with t-MN development: 2014-present.

A temporal depiction of the timeline from the first description of APOBEC and aging signature identified in 2014 through the first new unknown signature, later termed SBS-MM1, observed in 2016 and the explosion of literature in recent years demonstrating the correlation between SBS-MM1 and melphalan exposure. Abbreviations: APOBEC = apolipoprotein B editing enzyme, catalytic polypeptide; SBS = single base substitutions. [37, 82-94]

Prior studies of the genomic landscape of t-MN have been compiled in Table 1, which divides prominent studies into relevant categories for future characterizations of t-MN. Table 1 also includes the background conditions of patients that develop t-MN, as secondary leukemias resulting from the treatment of antecedent hematologic malignancies, including myeloid neoplasms have been well documented [95-97]. Of these, a landmark study published by Wong and colleagues included a WGS analysis of 22 cases of t-MN [65]. Here, it was demonstrated that t-MN had a similar mutational burden to de novo AML. However, not all cytotoxic chemotherapies cause measurable mutagenic activity, and notably, these samples came from patients with a low incidence of treatment with agents that have known and measurable mutagenic profile (i.e., platinum, melphalan, temozolomide). In fact, Pich et al. further analyzed these genomes in conjunction with additional samples. They showed, adjusting for treatment, that the t-MN arising from individuals exposed to platinum chemotherapies did indeed have a higher mutational burden than t-MN without such exposure [38]. Importantly, it was demonstrated that positively selected genes (i.e., driver genes) were not likely to have been directly introduced by platinum therapy and seemed to predate the exposure. Similar findings related to temozolomide were seen in a pediatric cohort sequenced by Bertrums et al. [98].

Table 1.

Compiled table of various sequencing efforts to characterize t-MN

STUDY	YEAR	TOTAL # OF PATIENTS	BACKGROUND CONDITION (# WHEN SPECIFIED)	METHOD PERFORMED	CITATIONS
CLEVEN ET AL.	2015	95	MDS (60); CML (3); AML (32)	TDS	[99]
VOSO ET AL.	2015	37	HL (7); nHL (12); Breast (13); Combination of Breast and other (5)	TDS	[100]
WONG ET AL.	2015	22	AML; MDS	TDS; WGS	[65]
TAKAHASHI ET AL.	2017	112	Primary Lymphoma	TDS; MBS	[3]
GILLIS ET AL.	2017	13	Solid Tumor; Lymphoma; MM	TDS; WES	[101]
NISHIYAMA ET AL.	2018	13	MDS (9); AML (2); CMML (2)	TDS	[102]
BERGER ET AL.	2018	18	MDS	TDS; WES	[103]
SCHWARTZ ET AL.	2019	62	Hematologic; Bone/Soft Tissue; Brain	WES; RNA-Seq; WGS	[104]
SINGHAL ET AL.	2019	129	Hematologic; Breast; Prostate	TDS	[105]
KUZMANOVIC ET AL.	2020	266	Breast; Prostate; Lymphoma	TDS	[106]
SCHWARTZ ET AL.	2021	84	N/S	WES; RNA-Seq; WGS	[107]
COORENS ET AL.	2021	20	Solid Tumor	TDS; WGS	[108]
VOSO ET AL.	2021	15	CLL	TDS	[110]
LIU ET AL.	2021	196	MDS (108); AML (88)	TDS	[109]
PATEL ET AL.	2021	109	Various (N/S)	TDS	[111]
BERTRUMS ET AL.	2022	24	ALL (5)	TDS; WGS	[98]
CLAERHOUT ET AL.	2022	64	Breast, AML, MDS	TDS	[112]
DIAMOND ET AL.	2022	40	Hematologic Malignancies; Solid tumors	TDS; WGS	[81]
SHAH ET AL.	2022	342	N/S	TDS	[113]

Abbreviations: WES = Whole Exome Sequencing, TDS = Targeted Deep Sequencing, MBS = Molecular Barcode Sequencing, RNA-Seq = RNA Sequencing, WGS = Whole Genome Sequencing, ALL = Acute Lymphocytic Leukemia, AML = Acute Myeloid Leukemia, MDS = Myelodysplastic Syndrome, HL = Hodgkin’s Lymphoma, nHL = Non-Hodgkin’s Lymphoma, CMML = Chronic Myelomonocytic Leukemia,

Therapeutic Selection Pressures on Clonal Hematopoiesis

Recently, the mechanisms, characteristics, and frequency of somatic mutations and the tendency of these to accumulate gradually within the genome have been well-characterized [114-116]. Additionally, strong correlations have been established between mutations and aging [114], with data suggesting that nearly half of the somatic mutations are acquired during the aging process [117, 118]. While somatic mutations, especially within the context of a lifespan, are prevalent and mostly harmless [119], their effects can vary widely based on their location within the genome and the specific gene they impact [115]. To expand upon this idea, somatic mutations occurring in critical regulatory regions of the genome have been observed to possess the capacity to perpetuate into larger populations due to selective pressures inducing their proliferation. This, in essence, describes the mechanism from which clonal hematopoiesis is derived. In greater detail, CH has been shown to occur with mutations in genes, such as DNMT3A, TET2, ASXL1, or the previously discussed TP53 genes [120], and/or with mosaic chromosomal alterations. These alterations lead to the expansion of the clone into a significantly larger percentage of the population due to an acquired competitive advantage [1, 121, 122]. Moreover, the combination of mutations and mosaic chromosomal alterations (clonal populations with gene mutations and copy number variants) was associated with the highest risk of progression to leukemia compared to either alone [123].

Antecedent clonal hematopoiesis has been repeatedly demonstrated as a risk factor for the development of t-MN [17, 18, 23-25]. The association has fueled multiple investigations into specific gene-treatment interactions. A quintessential study examining the evolution of CH under cancer therapy was the aforementioned study performed by Wong and colleagues. The investigators identified that many TP53 mutations that were clonally present in t-MN genomes could be detected at low frequency (variant allele frequency [VAF] 0.003-0.7%) in prior samples predating the exposure to therapy [65], suggesting that they had expanded under the selective pressures of chemotherapy. A Tp53 heterozygous null mouse model further demonstrated positive selection for Tp53 aberrant hematopoietic stem and progenitor cells (HSPCs) after chemotherapeutic treatment [65]. These results, among others, led to the acceptance of the model of t-MN being a result of therapeutic selection pressure on clones with mutations conferring a survival advantage [46, 102].

Since Wong and colleagues demonstrated that TP53 mutations predated chemotherapy exposure, further large-scale efforts have been made to characterize the selective forces and gene-therapy relationships between CH and anti-cancer therapies [81]. In a targeted sequencing study of 8,810 cancer patients undergoing various therapies, Bolton and colleagues elaborated on variants preferentially selected by specific treatments using a VAF cutoff of 2%. Following therapy, there was a general increase in frequencies of CH with DNA-damage response mutations (TP53, CHEK2, and PPM1D), providing evidence of their advantage under selective pressure. Specifically, platinum and TOP2 therapies have been demonstrated to be most strongly correlated with the development of CH with putative driver genes, with increased mutations occurring in critical TP53, CHEK2, and PPM1D genes. Out of 35 patients in this cohort who went on to develop t-MN and had paired CH and t-MN samples, more than half had TP53 mutations, and >70% of those mutations were present at the time of CH testing. Interestingly, Bolton and colleagues challenged the notion that selection was the only effect of chemotherapy on CH by demonstrating an increase in additional somatic events, including aneuploidies or mutations in familiar leukemic drivers, including NRAS, KRAS, and FLT3 in sequential samples and resultant t-MN. This concept has tremendous importance in understanding the multiple routes CH may take toward expansion into malignancy [25].

Mutations in DNA damage response genes have since been interrogated for evolution under selective pressure. In sequencing data of peripheral blood samples with t-MN, PPM1D variants have appeared at an increased frequency [29, 40]. In a considerable DNA sequencing effort conducted by Hsu et al., PPM1D mutations were observed in 20% of 156 t-MDS patients, with only TP53 mutations being more prevalent (VAF cutoff of 0.4) [124]. PPM1D has been identified as having a vital role in regulating p53, demonstrating that mutations within this region would have similar effects to TP53 mutations and can predispose one to the development of t-MN [125]. Recently, Hsu et al. investigated the increased rates of PPM1D mutations within t-MN patients [124]. A strong relationship was confirmed between platinum chemotherapeutics [126] and the proliferation and expansion of PPM1D mutations within the genome.

Immunomodulatory agents have also emerged as essential arbiters of selective pressure on myeloid cells. In a recent study by Sperling et al., the role of lenalidomide treatment in selecting somatic TP53 mutations was analyzed within a cohort of 416 patients with multiple myeloma (MM), demonstrated through TP53 DNA sequencing efforts [127]. Lenalidomide specifically degrades CK1α [128], which has been demonstrated to activate p53 and lead to subsequent apoptosis [129]. Due to this, Sperling et al. investigated whether mutated TP53 cells would inherit the ability to bypass these apoptotic mechanisms and gain a selective advantage under lenalidomide treatment. Through mouse models mimicking patient cohorts with TP53 mutations, it was seen that the increased potency of lenalidomide induced an increase in apoptosis of wild-type TP53 cells compared to the mutated TP53 cells [127]. However, the study must be interpreted within the context of the concomitant effects of high-dose melphalan and autologous stem cell transplant (HDM/ASCT), a ubiquitous concurrent therapy in multiple myeloma. For example, in the recent DETERMINATION study of bortezomib, lenalidomide, and dexamethasone with or without HDM/ASCT, 2.7% of patients in the transplant arm developed t-MN compared to 0% with the lenalidomide combination only [130]. This suggests that further mutation from mutagenic chemotherapy and/or chemotherapy-induced alteration of the immune microenvironment is a critical factor in t-MN development.

It has become clear that even as cancer therapies shift away from cytotoxics and towards immunotherapeutics, t-MN has remained an ever-present specter. There are reports of an increased incidence of t-MN following treatment with CAR-T cellular therapy [131-133]. Recent studies have explored the relationship between CAR-T therapy and relapse, which has revealed striking correlations between induction of treatment and subsequent t-MN development [134, 135]. These examples of immune modulation/depletion leading to downstream t-MN reveal the delicate interplay between CH and immune checks. Exogenous forces – cytotoxic or otherwise – may favor the expansion of t-MN from precursors populations with proliferative or survival advantages. It has been shown that, following profound damage to the hematopoietic compartment, as in ASCT, unique clonal dynamics shape reconstitution with evidence that remarkably few ancestors repopulate the marrow [136-139]. In analyses of the unintended effects of chemotherapy, the immunosuppressive effects of various chemotherapies are well characterized and corroborated [140-142]. Additionally, inflammation responses are also observed within studies on the consequences of chemotherapies, which can affect the efficacy of the treatment, decrease immune system capabilities, and ultimately lead to worsened prognoses, particularly in hematological malignancies in which immune response plays a critical role [143-145]. Furthermore, recent investigations have indicated that novel therapies such as peptide receptor radionucleotide therapy and poly-ADP ribose polymerase (PARP) inhibitor therapy have recently been associated with t-MN development [146, 147].

Disparate Routes to t-MN Progression: Selection vs. Acquisition

Despite the plethora of evidence for the selective effects of chemotherapy on CH, some chemotherapies do indeed cause an increase in the mutational burden on resultant t-MN. Though it has been summarized here that single nucleotide mutations in driver genes exist in CH prior to the administration of chemotherapy, Diamond et al. sought to answer whether complex driver events might be introduced by mutagenic therapy. The authors included patients that had undergone ASCT as melphalan is delivered in an abbreviated time window and because the therapy is myeloablative, supporting that only the most advantageous CH mutations should confer the ability to survive exposure.

In line with prior evidence [38], all patients with previous platinum exposure had a resultant platinum-associated mutational signature within the t-MN genome. However, only a minority of patients exposed to high-dose melphalan had a corresponding SBS-MM1 melphalan-associated mutational signature [76, 148]. Through multiple lines of evidence, including a demonstration of CH in pre-chemotherapy apheresis products, the authors proposed that this differential presence of melphalan-induced mutagenesis was due to the ability of CH to escape direct exposure to myeloablative melphalan via the leukapheresis procedure. The reinfusion model corroborates the importance of the selective forces of chemotherapy: direct exposure to an exogenous agent is not needed to provoke t-MN progression, but cytotoxic effects on the immune system can create a background for CH to dominate, which is demonstrated in Figure 2. Conversely, t-MN that had evidence of chemotherapy-induced mutagenesis (i.e., SBS-MM1) did indeed have higher mutational burdens, increased incidence of complex cytogenetics, chromosomal aneuploidies, and complex structural variants with novel and recurrent drivers (i.e., SMARCA4). Using novel chemotherapy barcoding approaches, significant chromosomal gains and structural variants were seen to have been acquired after chemotherapy exposure and late in tumor evolution. This finding indicates that while cytotoxic therapy does indeed provide a selective force on prior CH, it can also lead to the acquisition of driver events.

Figure 2. Trajectories for Expansion and Transformation of Clonal Hematopoiesis Following High-Dose Melphalan and Autologous Stem Cell Transplantation.

CH may escape exposure to melphalan via leukapheresis and be autologously reinfused into an immunosuppressed recipient. The resultant t-MN bears a genomic profile similar to de novo AML and t-MN exposed to less/non-mutagenic therapies. Alternatively, fit CH clones (i.e., TP53-mutated) may survive exposure and accrue complex genomic drivers. Inset left: Single nucleotide variants will be accrued in a specific and predictable context (i.e., the mutational signature SBS-MM1). Inset center: example of a complex t-MN genome with multiple aneuploidies across each chromosome separated by vertical hatched lines. The horizontal solid black line represents the total copy number, and the horizontal hatched line represents copy number. Inset right: Example of a chromothripsis event on chromosome 5 in t-MN. The vertical lines represent structural variants breakpoints: blue = inversion, green = tandem duplication; red = deletion; black = translocation.

In addition to findings by Diamond and colleagues, research conducted by Sridharan et al., in which MM patients’ progenitor and stem cells were collected prior to autologous transplant, has uncovered that TP53 or RUNX1 mutations were present in HSCs prior to the onset t-MN. It was observed that TP53 mutations were harbored at a high rate in HSCs and ultimately became the dominant cell population within t-MN developed in these patients years later. From this, Sridharan and colleagues concluded that HSCs could act as reservoirs for leukemia-initiating cells, and provides further evidence favoring an alternate route for t-MN expansion that is largely unrelated to chemotherapeutic selection [149].

Summary and future implications

From the evidence presented within this review, it is clear that much is left to uncover regarding the origins of clonal hematopoiesis and its role in developing therapy-related myeloid neoplasms. Technological advances, like WGS, provide the opportunity to gain crucial insight into these questions and uncover the mechanisms by which these phenomena occur in order to shape future treatment plans and improve the otherwise dire prognosis associated with t-MN. WGS technology has increased the capacity to identify complex and novel t-MN driver events and helped illuminate the selection of CH from specific chemotherapies. It is now understood that certain chemotherapies have particular interactions and relationships with specific genes, but also that CH may evolve into t-MN without direct exposure of the cells to chemotherapy through reinfusion into a patient after high-dose chemotherapy and autologous stem cell transplant. Furthermore, it has been presented that certain chemotherapies can introduce genomic complexity and driver events, leading to the formation of the novel concept of t-MN acting as a heterogenous entity with multiple routes of CH leading to progression. These findings provide relevance to recent findings by Diamond et al. and raise questions about the role of chemotherapies on the immune microenvironment and t-MN expansion. There remains much to understand regarding the role that immunosuppression and inflammatory responses play in the development of therapy-selected CH expansion and t-MN. Additionally, the direct effect of these factors on clonal populations and the mode by which they integrate into the previously demonstrated therapy-selection hypothesis presents itself as a critical target of future research on the expansion of CH and the development of t-MN. Moving forward, it is critical to identify the particular subset of patients that demonstrate a pre-disposition or affinity towards the development of t-MN, as this information is not yet known. Additionally, it is essential to thoroughly characterize the triad of the immune system, leukemic precursors, and exogenous selective forces to understand how to tailor current and emerging therapies to patients and their specific risk for t-MN development.

Highlights.

• Therapy related myeloid neoplasms (t-MN) make up 10-20% of myeloid malignancies

• Clonal hematopoiesis (CH) occurs at high frequency in t-MN

• Whole genome sequencing characterizes complexity of t-MN genetic alterations

• Triad of CH, chemotherapy selection and immune response contributes to t-MN