A View of Myeloid Transformation through the Hallmarks of Cancer

Myeloid malignancies are a group of blood disorders that include acute myeloid leukemia (AML), myelodysplastic syndromes (MDS), and myeloproliferative neoplasms (MPN). These diseases originate from genetic and epigenetic changes in hematopoietic stem and progenitor cells (HSPC) that are acquired and selected over time (1). Characterizing clonal hematopoiesis (CH) as a premalignant state and increasing the use of single-cell technologies have improved our understanding of the development and evolution of myeloid malignancies (2, 3). CH refers to the presence of an expanded HSPC clone in blood cell output, which propagates to the point of detection without a clinically overt myeloid malignancy. The occurrence of CH significantly increases with age, and it is caused by somatic mutations in individual genes or gains and/or losses of chromosomal segments. The mutations commonly detected in CH are disease-initiating mutations for myeloid malignancies (4).

However, the malignancy risk for patients with CH is highly variable and depends on the context in which it occurs, the mutated gene(s), and clone size, making its clinical significance challenging to interpret when incidentally found (4, 5). In addition, CH is frequently detected in patients with solid tumors, and most studies have shown a higher risk of developing hematologic malignancies and increased mortality in patients with CH compared with those without mutant hematopoietic clones (6, 7). These observations suggest that cell-extrinsic factors may play a critical role in disease progression. Although much research has been devoted to understanding the role of genetic drivers in myeloid transformation, the role of cell-extrinsic factors in the evolution of myeloid malignancies needs to be better comprehended. Presently, we do not fully understand how to weigh HSPC-intrinsic and -extrinsic factors and their relative and cooperative role in promoting disease progression and the potential for therapeutic intervention in preventing overt transformation. Cell-extrinsic factors can manifest in individuals with or without a preexisting cancer diagnosis, including current or former smoking history, exposure to environmental mutagens, comorbidities like inflammatory or metabolic syndromes, and, inevitably, aging (2, 8).

In 2000, Hanahan and Weinberg introduced their seminal conceptualization of the hallmarks of cancer, in which they identified six essential cell characteristics or abilities to transition to a neoplastic state. These included (i) evading apoptosis, (ii) self-sufficiency in growth signaling, (iii) insensitivity to antigrowth signaling, (iv) sustained angiogenesis, (v) limitless replicative potential, and (vi) metastasis, which were mainly attributed to cell-intrinsic factors (i.e., mutations; ref. 9). In 2011, the same authors updated their framework to incorporate emerging hallmarks, including a set of cell-extrinsic factors such as tumor-promoting inflammation and the interplay between cancer cells and the immune system (10). Almost a decade later, in 2022, Hanahan updated the existing framework by proposing four more hallmarks. This time, the focus was on microenvironmental conditions, including the senescence-associated secretory phenotype (SASP), microbiota, and microenvironmental mechanisms of epigenetic reprogramming (11). These updates reflect the continued increase in our understanding of different cell-intrinsic and -extrinsic factors in cancer risk and development.

Hematologic malignancies and solid tumors have noteworthy differences that require attention within the framework of the hallmarks of cancer. Genetics, immune response, and the microenvironment play a vital role in the pathogenesis of both hematologic and nonhematologic malignancies (12, 13). However, hematologic malignancies have a distinct cell composition and three-dimensional microenvironment (14). Although many studies have focused on understanding the microenvironment in which malignant HSPC clones develop, they have prioritized overall cell composition, commonly without investigating the spatial location of cells within the resident niche (15). Although hematopoietic malignancies and solid tumors share some common features within the tumor microenvironment (TME), the term “TME” is not entirely accurate in the context of hematopoietic malignancies, partly due to the circulating and homing properties of hematopoietic cells. The human bone marrow (BM) is a complex and heterogeneous environment, and a malignancy increases this complexity further. Deeper knowledge is critical because understanding the role of the microenvironment in myeloid transformation can help us better prevent and manage myeloid malignancies, especially AML, in which resistance and relapse remain a concern despite significant progress in the development of new therapies.

In this review, we aim to provide a comprehensive summary of the impact of cell-intrinsic and cell-extrinsic factors, with a particular focus on the latter, on the evolution of CH to myeloid malignancies and subsequent disease progression and how such factors can be therapeutically targeted to halt clonal expansion and malignant transformation. Here, we view myeloid malignancies through the lens of the newest classification of hallmarks of cancer. We explore how they fit into this model, given their connection with solid tumors, as well as the differences between these entities with respect to malignant transformation and therapeutic opportunities.

Cell-intrinsic hallmarks—often regarded as cell-autonomous—refer to the inherent characteristics of a cell that contribute to its oncogenic phenotype, whereas extrinsic features are elements of the surrounding microenvironment that affect cell behavior and, thus, the course of neoplastic disease. The former includes genome instability, mutations, resistance to cell death, replicative immortality, altered metabolism, and epigenetic mechanisms. Cancer cells use these features to progress through the cell cycle despite DNA damage, replicate infinitely, and evade death (16). Both solid tumors and myeloid malignancies have alterations in common proliferation/survival signaling pathways, including the JAK/STAT, Wnt/β-catenin, Hedgehog, Notch signaling, and PI3k/Akt/mTOR pathways, with deregulations providing a growth advantage and survival to cancer cells, as well as resistance to apoptosis (17). The evasion of apoptosis plays a crucial role in the development and progression of AML and drug resistance (18). The discovery of promising new classes of targeted therapies and treatment strategies relies on researching the structural and biochemical aspects of apoptosis proteins and their regulators. For example, venetoclax, a BCL2 inhibitor, has revolutionized the management of patients with AML who are ineligible for intensive chemotherapy (19).

Hematologic malignancies generally have fewer somatic mutations than solid tumors (20), which may account for their relatively poor response to immune checkpoint inhibition, except for lymphoid malignancies with PDL1 overexpression (21). A small subset of patients with high tumor mutational burden and/or mismatch repair signatures may be biological outliers and considered for immune checkpoint inhibition trials (20); however, this has not been studied to date. Several reports have shown that AML has a unique pattern of development and disease progression (3, 22). Yet, there is still very little data on immune activation or tolerance mechanisms in myeloid malignancies, which contrasts with solid tumors, in which immune evasion strategies continue to be extensively studied (23). It is possible that an improved understanding of cell-intrinsic and -extrinsic factors, which regulate the immune visibility of leukemic cells, will open up new therapeutic opportunities for patients with myeloid malignancies.

Epigenetic alterations are a key contributing factor to the pathogenesis of myeloid malignancies. Hematologic malignancies frequently involve mutations in genes that regulate DNA methylation (TET2 and DNMT3A), chromatin modifications (MLL, EZH2, and ASXL1), higher-order chromatin structure (cohesin family members), and key hematopoietic transcription factors (GATA2 and CEBPA; refs. 24, 25). Moreover, recent studies have shown that commonly mutated genes with pleiotropic functions, such as NPM1, can directly affect chromatin state and epigenetic regulation (26, 27). Although mutations in epigenetic regulators are seen in a spectrum of solid tumors, their mutational distribution is distinct from hematologic malignancies. Epigenetic therapies have also shown greater therapeutic benefits in hematologic malignancies than those in solid tumors, especially when given as monotherapy in myeloid malignancies (28). It remains to be seen if epigenetic therapies can be used to augment other therapeutic modalities in hematologic malignancies or solid tumors, which is a major area of translational investigation.

Another key feature of hematologic cancers, including myeloid malignancies, is metabolic plasticity. In much the same way as solid tumors, different leukemic subtypes rely on different metabolic pathways, which can impact the function of genes and proteins that contribute to cancer growth (11, 29). Mutations in the isocitrate dehydrogenase (IDH) 1 and IDH2 genes are common in AML and other myeloid malignancies and, by contrast, are mostly concentrated in a narrow spectrum of solid tumors, including gliomas, chondrosarcomas, and cholangiocarcinomas (30). The accumulation of 2-hydroxyglutarate —a cell-autonomous oncometabolite resulting from the gain-of-function ability of IDH-mutant cells to produce it—induces cytokine independence, further blocks hematopoietic differentiation, and suppresses the ten-eleven translocation (TET) family of DNA demethylases and Jumonji family of histone demethylases (31). This causes epigenetic changes that lead to altered gene expression, which may be a driving force in the development of AML. In addition, insightful studies in leukemic stem cells have shown that altered metabolism is a key feature of primitive leukemia-propagating cells, which led to the use of venetoclax to alter mitochondrial function and induce apoptosis in AML (32, 33). Seminal studies in patients with solid tumors have shown that altered metabolic function is a key feature of epithelial malignancies (34, 35); however, this has yet to be leveraged for therapeutic benefit.

Numerous studies have demonstrated the crucial role played by extrinsic regulators in both normal and malignant myelopoiesis. Changes in key environmental molecules can directly influence HSPC fate and behavior, indicating that cell-extrinsic mechanisms contribute to the maintenance of hematopoietic homeostasis (36). Short-range factors such as stromal cells and vasculature, as well as systemic factors such as aging, exposure to therapies, inflammation, and metabolism, can all impact hematopoietic cells at different levels, from the cellular level to the BM niche and beyond. Extrinsic factors that impact normal and malignant hematopoiesis may arise from activated signals in the malignant cells, normal hematopoietic cells, or stromal BM niche cells. These signals may impinge on all these cell types, thereby having a significant impact on the cellular environment and the malignant myeloid cells. This section highlights the primary research findings that support the role of external factors in the clonal evolution of myeloid malignancies from CH to full-blown leukemia. We focus on factors directly linked to the newest classification of the hallmarks of cancer: (i) inflammation, (ii) microbiota, (iii) cellular senescence, and (iv) leukemic BM niche and invasiveness in extramedullary niches (referring to the ability of malignant myeloid cells to infiltrate distant organs, mirroring the metastatic cascade of solid tumor cells).

In the hematopoietic compartment, inflammation can have both positive and negative effects on HSPCs, promoting normal hematopoiesis and function in some cases while hindering it in others (37). Specific inflammatory signals, such as cytokines and Toll-like receptor (TLR) ligands, can influence HSPC fate and blood output (38). However, excessive production of these signals has been identified in myeloid disorders, leading to aberrant activation of the TLR axis, secretion of proinflammatory cytokines by malignant myeloid cells, and promotion of disease progression (39). For example, in AML, IL1 receptor signaling has been shown to be upregulated by TLR activation, resulting in overexpression of the IL1 receptor accessory protein on malignant HSPCs. This enhances AML cell viability and clonal output in patients (40). In the BM of patients with MDS, inflammatory pathways become increasingly activated, resulting in elevated expression of proinflammatory cytokines such as TNFα, IL6, and IL1. Furthermore, heightened levels of damage-associated molecular patterns, including S100A8/9, and activation of the NLRP3 inflammasome have also been observed. These factors have been associated with enhanced clonal expansion and inflammatory pyroptotic cell death, which has been suggested to be conducive to MDS pathogenesis (41–43).

MPNs are also characterized by elevated proinflammatory cytokines, which can lead to adverse outcomes like decreased overall survival (44). In patients with myelofibrosis, production of the proinflammatory cytokine IL8 is elevated, and its secretion by mutant clones is associated with increased severity of symptoms and fibrosis grade. Overexpression of IL8/CXCR1/2 is also linked to poorer prognosis in MDS and AML (45). Given these findings, IL8/CXCR1/2 signaling inhibition represents a promising therapeutic strategy.

Recent scientific research has shown that mutations in genes linked to CH and myeloid malignancies, such as Tet2 and Dnmt3a, increase the susceptibility of HSPCs to inflammation. In preclinical models, inflammation-induced expansion of HSPCs results in sustained myelopoiesis, providing a competitive advantage of mutant HSPCs over their wild-type counterparts (46, 47). Several studies have revealed the importance of differentiated hematopoietic cells in activating inflammatory signaling in the context of CH. For example, the loss of TET2 or DNMT3A leads to an upregulation of a type I IFN signature in mature macrophages. This is caused by the loss of integrity in mitochondrial DNA, which activates the cyclic GMP–AMP synthase–stimulator of IFN gene pathway (48). In vitro and murine studies have shown that Tet2 or Dnmt3a-mutant macrophages can promote the increased release of inflammatory cytokines, such as IL1, IL6, and CXCL1/2, upon inflammatory stimuli (49–51). This excessive production by CH-mutant effector cells, as well as an increased CH-mutant HSPC self-renewal, may explain the link between CH and inflammatory disorders and the role of inflammation as a driving force for clonal expansion. Several forms of inflammation, such as obesity-induced inflammation (52) and aging-related inflammation (inflammaging; ref. 53), have been shown to enhance clonal fitness in CH.

Furthermore, CH has been associated with inflammation in the context of solid tumors (6). CH has been commonly observed in individuals with solid tumors and is associated with poorer overall survival and a heightened risk of tumor progression (54). This increases the possibility that CH may alter the tumor-immune microenvironment. In patients with solid tumors, CH is also linked with an increased risk of therapy-related myeloid malignancies (55). The emergence of CH in these patients may result from elevated inflammation originating from the development of the solid tumor itself (11). This inflammatory state may increase the chances of CH progressing to myeloid malignancies. This hypothesis is bolstered by mounting evidence of a greater occurrence of CH and myeloid malignancies in individuals with a history of autoimmune disorders, including aplastic anemia and rheumatoid arthritis, repeated or chronic infections, or metabolic disorders such as type 2 diabetes (2). Moreover, inflammation-inducing behaviors, such as smoking, have also been correlated with an increased likelihood of developing CH (56) and MPN (57). Still, it remains unclear whether the higher prevalence of CH in these patient populations is due to an underlying chronic inflammatory state and/or impaired immune surveillance with reduced clearance of mutant clones.

The observation that increased inflammatory signaling occurs in the context of progression from CH to AML suggests that it is a significant factor in developing myeloid malignancies (Fig. 1). Consistent with this hypothesis, recent studies point out that chronic inflammation reinforces the fitness advantage of mutant cells, as reviewed elsewhere (8). For example, patients who develop TP53-mutant secondary AML have been shown to have TP53-mutant HSPCs that undergo dysregulated inflammation–associated gene expression changes (58). Several studies also suggest that mutant myeloid cells are resistant to inflammation and are more likely to survive than their wild-type counterparts (59, 60). This process could be accompanied by the accumulation of HSPCs with somatic mutations, resulting in clonal evolution and disease progression. However, it is unclear whether inflammation is necessary to establish CH and early clonal emergence or if it mainly promotes the selection of preexisting clones with increased malignant potential. Interestingly, new research using congenic murine AML models indicates that healthy HSPC subpopulations can also undergo proinflammatory conversion in the leukemic niche and contribute to disease (61). A recent preprint study also provides evidence that both CH-mutant and healthy HSPCs express a similar inflammatory memory gene program (bioRxiv 2023.09.11.557271), suggesting that cell-extrinsic factors may be upstream of its activation. Although more research is needed to understand the interactions between wild-type and mutant cells and the relationship between cell-extrinsic and -intrinsic mechanisms in inflammatory signaling, inflammation is evidently a critical factor in myeloid diseases. Advancements in understanding these processes in CH and myeloid transformation could lead to more effective strategies for targeting inflammation and halting disease progression.

The human microbiome has emerged as a new cancer hallmark due to its contribution to cancer development (11). Microbes interact with the immune system and colonize body surfaces, impacting cancer development in various ways. Microbes can contribute to cancer by damaging DNA, disrupting DNA repair processes, or causing inflammation. The formation of harmful substances by certain microbes is a known way by which cancer can develop, and the hindrance of eliminating carcinogens can also increase exposure to them, leading to cancer. Microbiota can drive genome instability that promotes cancer, inflammation that supports tumor growth, and immune system suppression, impacting all aspects of cancer development (62). Recent “-omics” advances have led to a better understanding of how the microbiome affects cancer (63, 64). Whereas most studies have focused on the microbiome in solid tumors, the microbial content in hematologic malignancies remains understudied.

A recent report has shown evidence of significant dysbiosis in the circulating microbiome of patients with myeloid malignancies. Although there were no significant differences in microbial content between patient samples from peripheral blood and BM, a considerable shift in dominant bacterial phyla and reduced diversity was detected in patients compared with healthy individuals (65). This dysbiosis may be due to intestinal permeability, which is often altered in patients with cancer due to malabsorption and other disease-related causes even before treatment (66). However, recent work has linked this effect to the development of leukemia and shrunk diversity of bacteria due to the loss of butyrate, a microbial metabolite, in the intestinal barrier (Fig. 2). In mouse models of myeloid malignancy with decreased intestinal barrier function and altered expression of tight junction proteins (e.g., claudins), the administration of butyrate resulted in the reversal of intestinal barrier damage. Gut microbiota dysbiosis, caused by antibiotic treatment, was also shown to promote murine AML progression, with fecal microbiota transplantation reversing this process (67). Indeed, there are significant differences in the bacterial makeup among AML subtypes, with overall patients with AML showing a higher bacterial burden but lower microbial diversity when compared with healthy individuals (65). The gut microbiota can influence the efficacy and toxicity of chemotherapy drugs, which can further alter microbiome composition. After undergoing chemotherapy, the microbiota of patients with AML becomes even less diverse, with a reduction in anaerobic bacteria (68, 69). The lessened diversity in patients with AML is akin to microbiome changes observed in solid tumors (70, 71), indicating a similarity between myeloid malignancies and solid cancers. Additional studies are needed to delineate the role of the microbiome in AML pathogenesis and therapeutic response in different settings.

Bacterial infections such as Clostridium difficile and Streptococcus/Enterococcus are more prevalent in individuals with CH (72). Preclinical studies have shown that expansion of CH clones occurs in response to bacterial stimuli in Tet2 and Dnmt3a-deficient HSPCs both in vitro and in vivo (73, 74). Additionally, antibiotic treatment can reduce the expansion of Tet2-deficient cells in mice, leading to significant changes in the expression of TNFα-related genes, achieving similar results to small-molecule inhibitors that abrogate inflammatory signaling (75). These findings point toward infection-associated inflammation as a critical driver for clonal expansion. However, it has not yet been confirmed whether changes in the microbiome, like those seen in AML, also occur in individuals with CH and increase their susceptibility to colonization by pathogenic bacteria due to displacement of the commensal gut microbiome. If true, targeting the gut microbiome through dietary intervention or alternative approaches could attenuate the progression from CH to myeloid malignancies.

Aging leads to the gradual functional decline of organs and tissues and is the highest risk factor for cancer development (76). As people age, they become more susceptible to different hematologic complications, such as anemia and defects in innate and adaptive immunity, and there is an increased prevalence of myeloid malignancies (77). Furthermore, the prevalence of CH is high in the elderly population, with a 10% to 20% prevalence or higher after 70 years of age (78, 79). The risk of developing CH was initially thought to be associated with random acquisition of somatic mutations in HSPCs during aging. However, CH etiology is more complex, and it may be promoted by key pathogenic features of the aging process. In this sub-section, we will focus on cellular senescence, considered a hallmark of cancer (11) and a major actor of aging (76). However, aging is a complicated process that leads to somatic mosaicism and involves many hallmarks on its own (76, 80). Many of these hallmarks are shared with cancer, such as cellular senescence, chronic inflammation, and dysbiosis. In brief, aging affects multiple aspects and reveals complex connections among mechanisms that cannot be separated as individual units. However, the interdependence of aging hallmarks implies that manipulating one specific hallmark can also affect others (76).

Senescence is a permanent cell-cycle arrest state critical to aging processes (11). Senescent cells are highly proinflammatory and secrete the SASP, which is composed of cytokines and extracellular matrix remodeling enzymes (81). In physiologic conditions, the SASP helps to recruit senescence-targeting immune cells to restore tissue homeostasis, like in wound healing (82). By contrast, during aging, the combination of accumulated tissue damage and the decline in the strength of the immune system hinders this clearance, resulting in a chronic proinflammatory milieu (83). With many of the SASP factors being proinflammatory cytokines (such as IL1α, IL1β, IL6, and IL8) and chemokines (such as CXCL-1/3 and CXCL-10; ref. 84), the SASP contributes to the chronic low-grade inflammation during aging and is also a central mechanism by which senescent cells can promote tumor phenotypes. Senescent cells play a role in proliferative signaling, avoiding apoptosis, inducing angiogenesis, stimulating invasion and metastasis, and inhibiting tumor immunity. Senescence is, therefore, a complex process that influences the TME (85). Although it remains unclear how cellular senescence interacts with immune infiltration in cancer or how it can be used to assess malignancies, a senescent microenvironment seems to promote the progression of malignant hematopoietic cells, as seen with several solid tumors.

Mesenchymal stromal cells (MSC) within the aging microenvironment also undergo senescence (86). These altered characteristics and functions of aging stromal cells can build a permissive milieu for hematologic transformation. Although hematologic malignancies originate from different cell types and display diverse genetic abnormalities and clinical characteristics, the same alterations in MSCs observed across various hematologic malignancies suggest shared molecular mechanisms that reshape the BM microenvironment (87). Aged MSCs have heightened NF-κB activity (88, 89) and SASP production (90). This may enhance and spread senescence to the healthy surrounding cells through induction of reactive oxygen species, DNA damage, and further NF-κB activation. Expression of p16INK4a on senescent stromal cells can also promote the survival and proliferation of AML blasts through the SASP. Eliminating such senescent cells has improved survival in a murine p16-3MR AML model (91). These preclinical results align with the work of two independent groups that have developed a senescence gene signature score and validated its value as a prognostic marker of poor overall survival in patients with AML (92, 93).

The aging BM microenvironment, with increased levels of senescence, may facilitate the clonal emergence of preleukemic HSPCs at the expense of their wild-type counterparts (94). Dr. Jennifer Trowbridge’s laboratory recently identified a link between Dnmt3a-driven CH and senescence, in which mutant HSPCs induce senescence in BM stromal cells through IL6 signaling. Deleting these senescent cells using the senolytic drug navitoclax reduced myeloproliferation and transformation of Dnmt3a- and Npm1-mutant HSPCs to AML in transplantation studies (bioRxiv 2024.03.28.587254). However, further research is needed to determine the potential for the senescent milieu to cause CH and promote progression to myeloid malignancies. Conversely, CH could also accelerate hematopoietic aging to establish a senescent milieu (Fig. 3; ref. 94). Delineating these mechanisms could also be especially relevant to the context of AML relapse, which is observed in 40% to 50% of patients (95). Dr. Ari Melnick’s group found that AML cells that survive initial chemotherapy show a senescence-like phenotype with the potential to repopulate leukemia (96). This study is consistent with other solid tumor findings, suggesting that cancer cells may exploit senescence to persist throughout chemotherapy (97). However, we have yet to establish the most relevant cellular source of senescence and affected cell population(s) (i.e., cancer cells or stromal cells) to cancer progression and resistance to therapy, as well as the extent to which senescence can promote malignant transformation during aging.

The BM niche is essential for maintaining HSPC homeostasis throughout our lifetime. The niche comprises various cell types, including osteoblasts, perivascular or MSCs, endothelial cells, and noncellular components, including extracellular matrix proteins and soluble factors, such as CXCL12, angiopoietin 1, and TGFβ, which are essential for HSPCs (98). Hematologic malignancies can alter the components of the BM niche. These malignant niches play a significant role in the initiation, progression, and resistance to therapy of the disease (99). However, due to the complexity of this environment, the molecular composition and mechanisms of malignant niches still need to be better understood. The HSPC niche is not limited to a single space within the BM but comprises distinct niches or regions that differentially modulate HSPCs (100). These regions are dynamic and can change with aging (101) and/or different pathologic states (102–104). Recently, Dr. Simón Méndez-Ferrer’s group discovered that JAK2^V617F-mutated HSPCs interact differently with the BM niche in essential thrombocythemia versus polycythemia vera, which may explain the different clinical phenotype and outcome despite sharing the same primary oncogenic driver (105). Therefore, understanding how BM niches change in the context of myeloid malignancies may be essential for therapeutic BM microenvironment modulation upon the initial diagnosis and during therapy.

Leukemic and normal HSPCs can compete for the same microenvironmental support, which has implications for treatment. Niche occupation may alter blood vessel formation, making it easier for leukemia or metastatic tumors to grow. Dr. Cristina Lo Celso’s group has shown that blood vessels can undergo remodeling to favor leukemic cell expansion in an MLL-AF9–driven murine AML model (103). Like solid tumors, leukemias also use VEGF signaling to induce angiogenesis and promote proliferation (106). These shared characteristics, along with the fact that the BM is also a hotspot for solid tumor metastasis (107), indicate remarkable similarities between solid tumors and leukemias in their spreading and invasion.

Leukemic cells can also create a proleukemic niche by impairing the differentiation status of MSCs. In addition, leukemic cells can disrupt the regulation of the HSPC niche by the sympathetic nervous system. Thus, AML-induced neuropathy impairs the differentiation status of MSCs, facilitating a proleukemic niche (102). This illustrates how MSCs may be reprogrammed by leukemic cells. This reprogramming may occur via direct cell-to-cell contact or secreted factors (108, 109). However, it may also be blocked as, for example, inhibition of NADPH oxidase 2 has been shown to reduce MSC changes caused by AML and improve survival in a xenograft model (110). Additionally, AML cells may leverage MSCs to acquire resistance to chemotherapy by activating signaling pathways involved in cell proliferation (e.g., Notch and Wnt) and suppressing apoptosis (111).

Single-cell RNA sequencing is a widely used method to analyze the molecular and cellular composition within the leukemic BM niche (112). However, it has limitations as it cannot provide spatial context, which is essential in understanding the leukemias and their interaction with the niche. Combining single-cell RNA sequencing and spatial transcriptomics will be pivotal to better understanding the changes occurring in the leukemic niche, as it can provide the required spatial context (Fig. 4). The malignant BM niche remains poorly understood. This knowledge is, however, of utmost importance as different spatial distribution across leukemia subtypes or uneven spatial distribution of immunotherapy-related target molecules could potentially explain unsatisfactory therapeutic results. In a recent preprint article, spatial transcriptomics was employed to identify the presence of large aggregates of T and B cells in the BM of patients with pediatric AML, resembling the secondary lymphoid organ structures, also known as tertiary lymphoid structures, that are observed in solid cancers (medRxiv 2023.03.03.23286485). Although more work is needed to understand the origin and frequency of these structures or similar phenomena, this study indicates that these regions act as concentrated locations of activated cytotoxic T cells, memory B cells, and plasma cells, mostly devoid of AML cells, even though leukemia is present in other BM areas. This finding indicates a possible space-dependent antileukemia effect by the activity of healthy immune cells.

Extramedullary (non-BM) AML, which spreads outside the conventional hematopoietic niche, has a poor prognosis and shorter survival (113). Despite the rare ability of leukemic cells to form solid tumors, such as myeloid sarcomas derived from myeloid leukemias (114), and the recognized clinical similarity of the development of leukemic tumors to that of some solid tumors like breast cancer (115), how leukemic cells “metastasize” and invade the extramedullary tissue is not well understood. Recent studies have shown that leukemic tumors in individual extramedullary sites behave like solid cancers originating in those same sites (115). The “metastatic” invasion of extramedullary sites is essential for the progression, severity, and disease outcomes in patients with high-risk leukemia (116). Given the adverse prognosis of extramedullary disease, one can hypothesize that interactions between leukemic cells and the surrounding microenvironment of different organs may play a role in the reduced response to antileukemic therapy at extramedullary sites and intensification of immune escape. A greater understanding of these mechanisms and distinction from their BM counterparts might lead to new therapeutic approaches to disrupting leukemia/niche interactions at extramedullary sites.

Myeloid malignancies are influenced by a combination of cell-intrinsic and -extrinsic factors. Although much research has been conducted on the genetic causes of these diseases, a deeper understanding of how cell-extrinsic factors contribute to their clonal evolution is necessary. In his 2022 review (11), Professor Hanahan proposed that new additions to the hallmarks of cancer, such as senescence (i.e., the SASP) and the microbiome, may be interconnected with other factors like the TME and tumor-promoting inflammation. Our review has explored how some of these cell-extrinsic factors are interconnected, like a puzzle. Inflammation, microbiota, cellular senescence, and the malignant or extramedullary niche all contribute to selecting CH/preleukemic clones in the spectrum of myeloid diseases (Fig. 5). Ongoing research should investigate how these “puzzle pieces” connect, as well as with the presence of CH mutations and other cell-intrinsic factors or missing pieces. Recognizing the importance of cell-extrinsic factors in myeloid malignancies is crucial, as therapies that target them specifically may enhance existing antileukemic therapies and/or open up totally new therapeutic opportunities. We posit that the rapidly expanding study of cell-extrinsic factors that promote clonal evolution and myeloid transformation will be important in diagnosing and treating premalignant/malignant hematopoiesis in the spectrum of human malignancies.

S.F. Cai reports consultancy for Daiichi Sankyo and Ursamin. He was previously a consultant for Dava Oncology and held equity interest in Imago Biosciences, none of which are directly related to the content of this article. R.L. Levine reports that he is on the Supervisory Board of Qiagen (compensation/equity), a co-founder/board member at Ajax (equity), and a scientific advisor to Mission Bio, Kurome, Anovia, Bakx, Syndax, Scorpion, Zentalis, Auron, Prelude, and C4 Therapeutics; for each of these entities he receives equity/compensation. He has received research support from the Cure Breast Cancer Foundation, Calico, Zentalis, and Ajax and has consulted for Jubilant, Goldman Sachs, Incyte, Astra Zeneca and Janssen. No disclosures were reported by I. Fernández-Maestre.

This work was supported by a Momentum Fellowship from the Mark Foundation for Cancer Research, a Scholarship of Excellence Rafael del Pino, and an NCI F99 award (CA284253-01) to I. Fernández-Maestre, a Career Development Award from the NCI (K08 CA241371-01A1) to S.F. Cai, and a MSKCC Support Grant/Core Grant P30 CA088748 and a R35 grant from the NCI (CA197594) to R.L. Levine.