Clonal Hematopoiesis: From Mechanisms to Clinical Intervention

Clonal hematopoiesis describes the presence of somatic mutations in the bone marrow or peripheral blood resulting from clonal hematopoietic stem and progenitor cell (HSPC) expansion (1) and has garnered tremendous scientific and clinical interest since the publication of seminal papers in 2014 (2–4). Although the development of cancer from premalignant lesions or the existence of age-related clonal populations in the blood of otherwise healthy individuals was not a new observation (5), large-scale sequencing studies identified the somatic lesions associated with these clones, determined the frequency of these premalignant states in the hematopoietic system, and revealed a strong correlation of these leukemia-associated somatic variants to nonmalignant conditions such as cardiovascular disease (CVD; ref. 6).

Sequencing-based approaches have also been applied to other tissues, demonstrating the similar existence of clonally diverse cell populations carrying somatic variants (7). This mosaicism can be found in virtually all tissues and is largely a product of cell division—thus making the strong correlation of this phenomenon with age not surprising. In the hematopoietic system, cells on average acquire 1.14 mutations per genome per cell division (Fig. 1A; refs. 7, 8). Some of these mutations will confer a fitness advantage and lead to expansion of this clone (9). However, we have yet to fully grasp which factors contribute to clonal expansion and development of hematologic malignancy, as well as nonmalignant consequences of clonal hematopoiesis.

To begin to answer these questions, the field has expanded on the initial discovery of clonal hematopoiesis in multiple intriguing ways: (i) Several studies have elucidated the molecular underpinning of individual clonal hematopoiesis mutations and how the mechanisms of clonal expansion differ between these mutations. (ii) Significant work has increased the understanding of specific clinical scenarios where either heritable factors, co-occurring conditions, or treatments for unrelated conditions contribute to the expansion and progression of clonal hematopoiesis. (iii) There is persistent interest in risk stratification of individuals with clonal hematopoiesis, with a goal to develop clinically applicable tools that reliably identify high-risk individuals.

Given the potential relevance of these observations to patients, some academic centers have established clonal hematopoiesis clinics, both to centralize the collection of biospecimens and to facilitate a central point of contact for other clinical specialties (10, 11). We anticipate that the need for these specialized clinics will increase as sequencing of blood specimens for other indications continues to expand. Counseling both the referring physicians and the patients will remain a central responsibility of hematologists, although these clinics may comprise multidisciplinary teams including cardiologists and others.

Here, we aim to summarize the current state of knowledge related to the clinical consequences of clonal hematopoiesis and discuss which directions the field is pursuing—and importantly, how these preclinical and observational insights can be most efficiently translated into insightful prospective clinical trials.

Initial observations of clonality in the hematopoietic system date back more than 60 years, based on Mary Lyon's concept of X-inactivation (12, 13), when Philip Fialkow and colleagues demonstrated the clonal origin of chronic myeloid leukemia (14). Using a similar approach, pivotal studies by Martin Fey (15) as well as Kim Champion (5) and colleagues identified age-related clonal skewing in otherwise healthy women, paving the way for the recognition of clonal hematopoiesis. However, these studies—while illuminating to our understanding of the hematopoietic system and aging—lacked the resolution to pinpoint the driving genetic events responsible for the observed clonality. Decades later, the availability of high-throughput sequencing technologies reshaped our understanding of clonal hematopoiesis by facilitating detailed analysis of large datasets from sequencing studies of peripheral blood specimens (2–4). These and other studies identified the genetic variants involved in clonal hematopoiesis and demonstrated once more that the presence of clonal hematopoiesis is strongly correlated with age (Fig. 1A). They also determined that these clones persist for many decades and occur in several hematopoietic lineages, suggesting that they are likely to initiate and reside in long-lived HSPCs. A significant insight from these recent studies was that individuals with clonal hematopoiesis, particularly those with large clone sizes, experienced an increase in all-cause mortality in addition to an increased risk of developing myeloid malignancies. Although the existence of a premalignant state in blood cancers had long been postulated based on the detection of such precancerous lesions in other tissues, only the advent of such sequencing studies allowed us to pinpoint how healthy HSPCs acquire these initiating mutations and later progress to frank leukemia (16–18). Importantly, both conditions—clonal hematopoiesis and myeloid malignancies—are strongly associated with age, further supporting a mechanistic link. Finally, these studies uncovered a surprising association of clonal hematopoiesis with cardiovascular events (6), which has rightfully garnered much attention.

The spectrum of mutated genes identified from these exome sequencing–based studies largely aligned with those thought to be founding events in myeloid malignancies, with variants in DNMT3A and TET2 comprising the vast majority, followed by ASXL1, splicing factors, JAK2, and TP53. Subsequent studies using whole-genome sequencing elucidated that cases of clonal hematopoiesis without any known driver mutation can also be identified and that these cases still retain the association with increased mortality (19). Finally, clonal hematopoiesis with chromosomal alterations has been studied more comprehensively and shown to be associated with the development of hematologic malignancies (20–23). However, these cases of clonal hematopoiesis are more likely to develop lymphoid malignancies, where there has been a strong association of +12, 13q−, and 14q− clonal hematopoiesis and the development of chronic lymphocytic leukemia (CLL). Even more surprisingly, the strong association seen in clonal hematopoiesis with point mutations or small insertions/deletions and CVD is not found in individuals with clonal hematopoiesis driven by chromosomal alterations (22, 23). Finally, a recent study by Seishi Ogawa and colleagues demonstrated that clonal hematopoiesis with chromosomal alterations and clonal hematopoiesis with point mutations or small insertions/deletions frequently co-occur in the same individuals and that this co-occurrence increased the risk to develop hematologic malignancies as well as experiencing adverse cardiovascular outcomes (24). This recent study thus highlights that our current understanding of the interaction of these two types of clonal hematopoiesis warrants further study.

The initial acquisition of mutations found in clonal hematopoiesis is thought to occur through several mechanisms. For point mutations, spontaneous deamination of 5-methylcytosine to thymine is found to be one of the most common events, which is a typical age-related mutational pattern found in many cancers (25). Less commonly, mutations are acquired by errors arising during repair of double-stranded DNA breaks creating small insertions/deletions or through replication errors by DNA polymerase. Because age correlates with the number of cell divisions, these events naturally also accumulate with age (26). The general assumption is that these mutational events occur somewhat randomly and then clones carrying biologically relevant variants subsequently undergo selection by possessing increased fitness over their wild-type counterparts. However, there is limited recent evidence that enrichment for the types of mutations seen in clonal hematopoiesis might also be due to preferential formation during microhomology-mediated end joining–based DNA repair (27).

Likely to originate in a single HSPC, once the fitness advantage of the clonal hematopoiesis–associated mutation has led to an expansion of that clone—through whichever mechanism—it becomes detectable in the peripheral blood by conventional sequencing approaches. Initial studies using whole-exome sequencing used a cutoff of around 2% variant allele frequency (VAF) largely for technical reasons. Using this cutoff, up to 20% of individuals above the age of 70 have detectable clonal hematopoiesis (2–4). More recently, technological advances such as ultra-deep error corrected sequencing have enabled detection of clonal hematopoiesis–associated somatic variants in blood specimens at minute levels (28). These low levels of somatic mutations were detectable in nearly every adult and, importantly, largely remained stable over at least a decade.

These observations have led to the formulation of more unified definitions of clonal hematopoiesis categories. The detection of clonal hematopoiesis in the peripheral blood in individuals without a known hematologic or clonal disorder with somatic mutations in genes recurrently mutated in hematologic malignancies with a mutant allele fraction of ≥2% was termed clonal hematopoiesis of indeterminate potential (CHIP; ref. 29). A related but not entirely overlapping term, age-related clonal hematopoiesis (ARCH), has been defined as a gradual, clonal expansion of HSPCs carrying specific, disruptive, and recurrent genetic variants in individuals without clear diagnosis of hematologic malignancy (ref. 30; https://ncit.nci.nih.gov/). However, some authors define ARCH much less strictly as any type of detectable, acquired clonal event in the hematopoietic system without detectable malignancy (31). Although the nomenclature, cutoffs, and definitions have varied, and are still likely to be refined, by emphasizing the indeterminate potential even in cases with higher VAF and somatic variants associated with hematologic malignancies, the field wisely sought to stress that for most individuals the presence of clonal hematopoiesis may not be significant to their health and longevity.

Comprehensive reviews of the current—and evolving—understanding of the molecular underpinnings of clonal hematopoiesis can be found elsewhere (32, 33). In this review, we aim to highlight how preclinical studies have illuminated the remarkable diversity of mechanisms underlying clonal expansion in clonal hematopoiesis, but also how some convergent pathways have emerged, which might be targetable.

The two most commonly affected genes in clonal hematopoiesis, DNMT3A and TET2—while biochemically involved in opposing reactions regulating DNA 5-methylcytosine—both lead to the expansion of mutant clones in the hematopoietic system. Only recently have we begun to grasp how this might occur. Recent work has suggested that DNMT3A-mutant hematopoietic stem cells (HSC) mainly gain their competitive fitness advantage by possessing a slight self-renewal bias (34), which is further magnified under exogenous stress such as inflammation, where normal HSCs differentiate and lose their self-renewal capacity while DNMT3A-mutant HSCs show improved resilience (35). Complicating the picture further, the DNMT3A mutational spectrum found in clonal hematopoiesis diverges from that seen in myeloid malignancies, which is notably enriched for variants of the R882 residue (36, 37). The mechanism for this observation is less well understood, but indicates that within DNMT3A, different mutations confer different risk for progression to acute myeloid leukemia (AML).

For TET2, loss-of-function mutations also result in increased self-renewal of affected HSCs (38, 39); however, unlike the loss of DNMT3A, which mostly exerts a phenotype in HSCs, more differentiated progenitors with TET2 loss are also prone to clonal expansion, leading to a robust myeloproliferation in mouse models (40). The molecular underpinnings of how DNA methylation changes caused by the loss of TET2 function result in these phenotypes are yet to be fully explored, but what has been of great recent interest is that TET2-mutant HSPCs show markedly heightened responsiveness to proinflammatory cytokines and accelerated myeloproliferation by microbial induced inflammation (41). In addition, progeny of TET2-mutant HSPCs—such as mature macrophages—are affected by TET2 loss and show heightened secretion of inflammatory mediators, including IL1B (42, 43) and IL6 (44), which might in turn affect HSCs (43), suggesting that some degree of reciprocity, or feed-forward loops between different compartments, is likely to shape the clonal expansion in clonal hematopoiesis. Finally, sophisticated mathematical modeling studies supported by preclinical model systems have also suggested that the presence of atherosclerotic lesions leads to increased HSC division rates, thus further accelerating the expansion of TET2-mutant clonal hematopoiesis clones (45).

Taken together, while the exact molecular mechanisms might differ, the recurring theme of inflammation acting as an accelerant in both DNMT3A and TET2 clonal hematopoiesis suggests that inflammation is a common pathway that could prove to be a therapeutic target in both DNMT3A and TET2-mutant clonal hematopoiesis.

Several genes involved in DNA damage repair (DDR), such as the tumor suppressor TP53 and PPM1D, which is part of a regulatory feedback loop with TP53, have been recurrently found in clonal hematopoiesis (46). Importantly, the mechanism of selection for these clones differs from DNMT3A and TET2, where TP53 and PPM1D mutant clonal hematopoiesis is associated with a particularly high risk of expansion and progression in patients receiving radiotherapy or chemotherapy and are consequently enriched in therapy-related myeloid malignancies (46–50). This is furthermore supported by preclinical evidence showing that cells with these mutations are more resistant to apoptosis under the pressure of radiotherapy or chemotherapy (51–53). In particular, exposure to platinum-based chemotherapy induced a pronounced advantage for DDR-mutant clonal hematopoiesis (52). However, what is particularly puzzling is that while the presence of pretherapeutic clonal hematopoiesis is a strong risk factor for therapy-related myeloid neoplasms, the clonal hematopoiesis mutation does not universally act as the founding mutation of the myeloid neoplasm itself (48, 49), which highlights that our grasp of the pathogenic mechanisms of these associations is still limited.

Taken together, our understanding of the molecular underpinnings of some of the recurrent mutations found in clonal hematopoiesis has improved dramatically. However, this brief summary of four mutations commonly found in clonal hematopoiesis also highlights that on a mechanistic level, clonal hematopoiesis is diverse and even within individual clonal hematopoiesis genes, specific mutations might confer a differential degree of fitness (9) and risk for certain outcomes (36, 37). Accordingly, any potential therapeutic approach will have to be tailored to the specific mutation.

There has been strong interest in defining both populations and clinical contexts where clonal hematopoiesis is of particular relevance. A major focus with huge clinical ramifications is the association of clonal hematopoiesis with CVD (54). This was initially discovered by noting that individuals with CHIP who never developed any hematologic malignancy were still found to have increased mortality (2, 6). This was due to an increased frequency of cardiovascular events, which was functionally confirmed in hyperlipidemic mouse models, where Tet2 loss in HSPCs markedly increased atherosclerotic lesions (6, 55). Mechanistically, it is thought that increased inflammation—possibly driven by mature, mutant myeloid cells—is the causative link in DNMT3A and TET2 clonal hematopoiesis to these cardiovascular events (Fig. 1B; refs. 33, 54). This connection also raised the possibility that targeting this increased inflammation might reduce cardiovascular events. Indeed, the most compelling early clinical evidence to date comes from an unplanned subgroup analysis in the randomized, placebo-controlled CANTOS trial (56, 57), where administration of the anti-IL1B antibody canakinumab led to a reduction of major adverse cardiovascular events in patients with TET2 CHIP (HR, 0.36; P = 0.034; ref. 58). Additional supporting evidence comes from the UK Biobank, where a genetic proxy of IL6 receptor inhibition (presence of the IL6R D358A variant) reduced the risk of cardiovascular events in individuals with DNMT3A or TET2 CHIP with a large clone size (defined as >10% VAF) back to the baseline of non-CHIP carriers (59).

Clonal hematopoiesis with mutations in JAK2 has also been associated with an increased risk of cardiovascular events; indeed, the relative risk for this association appears to be much higher than for carriers of TET2 or DNMT3A mutations (hazard ratio 12 for JAK2 vs. 1.7 and 1.9 for DNMT3A and TET2, respectively; ref. 6). Of note, activating JAK2 mutations are thought to confer this increased risk through a different mechanism than DNMT3A and TET2 mutations. Specifically, thrombotic and cardiovascular events are likely to be due to dying granulocytes releasing strands of DNA—termed neutrophil extracellular traps—which promote thrombosis (Fig. 1B; refs. 60, 61). Indeed, thrombotic events are detected at drastically increased rates in individuals with JAK2^V617F CHIP compared with non-JAK2^V617F CHIP (61). Thromboembolic complications are also a commonly observed complication in myeloproliferative neoplasms (MPN) and some have raised whether JAK2^V617F should be excluded from clonal hematopoiesis and rather serve as a defining lesion of de facto MPN (62). However, more recent work has demonstrated that acquisition of JAK2^V617F can occur many years before diagnosis of overt malignancy (63)—in some cases even in utero (64), only to emerge as the founding lesion many decades later. This finding, in conjunction with the observation that JAK2^V617F can be detected in more healthy individuals than patients with MPN, supports the notion that JAK2^V617F should be categorized similarly to other clonal hematopoiesis mutations (33, 65). Currently, there is interest in determining whether inhibition of the enzyme peptidyl arginine deaminase 4, which is involved in facilitating chromatin decondensation during neutrophil extracellular trap formation, might prevent these thrombotic events (66); however, clinical trial data are lacking.

The risk of clonal progression and the development of a hematologic malignancy in individuals with clonal hematopoiesis is elevated compared with the general population (Fig. 1B; refs. 2–4); however, the average absolute risk remains low. Small clones carrying variants associated with clonal hematopoiesis can be found in almost every adult when highly sensitive sequencing approaches are used (28), making clonal hematopoiesis per se a poor predictor of malignant progression. Thus, significant effort has been undertaken to determine which population of individuals with clonal hematopoiesis and which exogenous triggers are most strongly correlated with the development of hematologic malignancies (67). Population-based as well as case–control studies have pinpointed several factors that are associated with markedly increased risk of progression, such as the presence of specific mutations (mutations in TP53, IDH2, and the spliceosome genes SRSF2 and U2AF1), the number of mutations, and finally the fractional abundance (68, 69). In addition, there is some evidence that other clinical laboratory parameters, such as higher red blood cell distribution width, are associated with increased mortality (2) and risk of AML development (68). However, even in combination, the sensitivity of these clinical parameters remains limited to accurately identify individuals with high risk of AML development on a population level, and currently there is no rationale to screen healthy adults without hematologic anomalies for the presence of clonal hematopoiesis.

This perspective changes in the context of external stressors. In particular, the observation that clonal hematopoiesis can frequently serve as a precursor to therapy-related myeloid neoplasms (tMN) has been studied extensively (46, 48, 49, 70–72). Particularly, patients with clonal hematopoiesis receiving autologous stem cell transplantation (ASCT) for lymphoma (48) and with solid tumors receiving radiotherapy or chemotherapy are at increased risk of tMNs (73). However, clonal progression triggered by cytotoxic therapy is not seen uniformly across clonal hematopoiesis variants, nor is it uniform across different types of therapy (Fig. 1C). The highest risk of therapy-induced expansion is seen in clonal hematopoiesis with DDR variants (specifically TP53, PPM1D, and CHEK2), and in patients receiving radiation, platinum, or topoisomerase II inhibitor therapy for unrelated malignancies (73). The most intriguing observation from these studies was that lesions in these genes that later progressed to tMN were detectable prior to the initiation of cytotoxic therapy, suggesting that the mutations were not induced by the cytotoxic therapy, but rather that these clones were selected for by the therapeutic pressure. This observation is important, because it suggests the possibility that clinicians might screen for clonal hematopoiesis in individuals prior to receiving certain cytotoxic therapies, identify high-risk patients, and modify the clinical course of action. Indeed, clinicians are already contemplating whether the detection of TP53 CHIP should inform whether or not a patient should receive adjuvant chemotherapy for an unrelated malignancy, especially if the expected survival benefit is thought to be marginal (10, 74). Very limited evidence from patients with multiple myeloma receiving ASCT provides insight that the adverse prognosis for patients with multiple myeloma with concurrent clonal hematopoiesis might be modulated (75). In this study, patients with multiple myeloma and concurrent CHIP had shortened survival, including an increased rate of multiple myeloma progression. However, this association was abrogated in a subset of patients receiving lenalidomide maintenance therapy (75). Finally, in a study of clonal hematopoiesis in patients with cancer, the presence of clonal hematopoiesis was associated with increased rates of progression of the primary malignancy (70). The nature of this association is not well understood. Given the direct effect of cytotoxic therapy on clonal hematopoiesis, the level of pretreatment might act as a possible confounder, with heavily pretreated patients exhibiting higher levels of clonal hematopoiesis expansion without a mechanistic link to the progression of the primary malignancy. However, the association could also be more mechanistic, including increased inflammation from mutant, clonal hematopoiesis–derived effector cells or impairment of immune function (76).

In addition to these two major contexts, the presence and relevance of clonal hematopoiesis have been studied in additional settings. In allogeneic transplantation, some studies have postulated a link between the presence of clonal hematopoiesis in stem cell donors and increased rates of chronic graft-versus-host disease (77), as well as unexplained cytopenias (78). Moreover, there have been case reports of donor clonal hematopoiesis giving rise to donor-derived malignancy (79). These and other observations have prompted a debate over whether donors should be screened for the presence of clonal hematopoiesis (80, 81).

Individuals with HIV appear to have higher rates of clonal hematopoiesis than HIV-negative individuals (82, 83), whereas—despite their increased risk of myeloid malignancies (84, 85)—individuals with sickle cell disease do not appear to have any increased frequency or divergent mutational spectrum of clonal hematopoiesis (86). Somatic mutations in myeloid driver genes can be detected in almost half of individuals with aplastic anemia (87); however, the mutational spectrum is only partially overlapping with clonal hematopoiesis in the general population, and whether these patterns are due to the underlying immune activation as part of the pathophysiology, or whether they are primarily related to the treatment protocols—or some combination of both—remains somewhat unclear (88). Several less well understood sequelae of clonal hematopoiesis have also emerged, including a possible correlation with insulin resistance (89), chronic obstructive pulmonary disease (90), and most recently with the severity of COVID-19 (91, 92). In addition, there is also growing interest in whether the presence of clonal hematopoiesis influences frequency or severity of neurodegenerative disorders (33). In summary, the expanding identification of and interest in the associations of clonal hematopoiesis with a vast array of conditions is likely to lead to increased interest in whether there are any mechanistic links between clonal hematopoiesis and these conditions, and whether this might open a possibility of therapeutic intervention.

Not surprisingly, the prospect of early intervention—to detect and target clonal hematopoiesis early and prevent the emergence of the associated clinical sequelae—has drawn the interest of hematologists and other clinicians. As outlined above, the molecular mechanisms governing expansion and progression of mutant clones have nominated potential therapeutic strategies in some situations, and this research will continue to advance. However, the ubiquity of clonal hematopoiesis and the latency between detection and clonal progression pose significant challenges.

One potential treatment approach is to use mutant-specific targeted therapies (reviewed in ref. 93). Relevant FDA-approved agents include inhibitors for mutant IDH1 (ivosidenib) and IDH2 (enasidenib), as well as JAK2 inhibitors (ruxolitinib and fedratinib). Hypomethylating agents (azacitidine and decitabine) have demonstrated increased efficacy against TET2-mutant myeloid malignancies (94, 95). Similarly, high-dose vitamin C supplementation has been demonstrated to limit expansion of HSPCs with deleted TET2 (96). Investigational agents include several molecules targeting splicing factor mutations (97, 98) and mutant TP53 (99), which are currently in clinical trials for various cancers including hematologic malignancies.

However, there are numerous considerations and limitations to designing clonal hematopoiesis clinical trials with such agents. First, because the overall risk conferred by clonal hematopoiesis is relatively low in most situations, any intervention needs to carefully weigh the potential for treatment-related toxicities against the risk of progression to myeloid malignancy or clonal hematopoiesis–associated disease. Even the FDA-approved agents described above carry relevant risks of toxicity that may be acceptable in patients with cancer, but might only be considered in individuals with clonal hematopoiesis at markedly increased risk of progression or disease. This highlights the need for continued research to identify and stratify high-risk individuals. Secondly, the potentially long latency from detection of clonal hematopoiesis and the emergence of clinically relevant events such as death from CVD or progression to myeloid malignancy will make trials investigating these primary endpoints difficult, if not impossible. Surrogate and/or biomarker endpoints have increasingly been used to evaluate clinical interventions, including many cancer treatments (100). In the case of clonal hematopoiesis, surrogate endpoints might include decrease in mutant VAF, decrease in the number of mutant clones, or improvement in measures of disease risk such as validated CVD assessments. However, whether these endpoints will correlate with primary endpoints of overall survival and/or progression to myeloid malignancy is unknown (100). It will be important to engage key FDA stakeholders in the design and selection of endpoints for clonal hematopoiesis clinical trials and eventual drug approvals. clonal hematopoiesis clinical trials will be faced with additional challenges compared with cancer studies including how to identify patients, given that most individuals with clonal hematopoiesis are asymptomatic, and how long to treat, given that clonal hematopoiesis does not align with conventional measures of treatment responses in hematologic malignancies. Finally, even if such trials are conducted and clonal hematopoiesis treatments are approved, this would establish clonal hematopoiesis as a public health issue requiring adoption of peripheral blood screening by next-generation sequencing in the primary care setting. Such an approach will require development of a robust infrastructure and coordination including clinical sequencing labs, primary care physicians, and public health agencies.

In this last section, we argue that the field should begin pragmatically by focusing on those subgroups and endpoints that are highest risk and clinically most relevant to begin conducting prospective clinical trials. Until we have answers from such trials, counseling individuals with clonal hematopoiesis—even with what is considered “high-risk clonal hematopoiesis”—will remain a daunting task, and any recommendation will be based on limited evidence. This is of relevance because the number of patients who will approach their physicians with questions on how to interpret and react to a finding of clonal hematopoiesis is very likely to increase dramatically over the next decade due to the ubiquity of genetic testing (Fig. 2A). For example, an increasing number of studies are performing next-generation sequencing assays to refine risk stratification or monitor treatment response using cell-free DNA (cfDNA) from blood plasma (Fig. 2B), and these assays are capable of detecting clonal hematopoiesis variants (101, 102). In parallel, clinicians are already contemplating modifying clinical care based on the finding of “high-risk clonal hematopoiesis” under the umbrella of “shared decision-making” (10). Thus, there is urgency to start designing and conducting prospective trials that will give us answers—at least for certain subgroups and sequelae of clonal hematopoiesis.

We envision two patient populations where there is a clinical need and a path to identify enough individuals at risk in a reasonable time frame. The first category is individuals with DNMT3A or TET2 clonal hematopoiesis who are at elevated risk of cardiovascular events. The association between clonal hematopoiesis and CVD was initially surprising (2), but has since been confirmed in multiple cohorts, as well as in mechanistic studies (6, 55). As outlined above, cytokine-driven inflammatory signaling seems to be a crucial mechanism through which DNMT3A and TET2-mutant clonal hematopoiesis are linked to CVD. Ongoing studies will help determine VAF thresholds to identify high-risk individuals. Although there is considerable evidence that the risk of progression to myeloid malignancies is drastically increased in clonal hematopoiesis with VAFs ≥10%, there is some evidence that a lower VAF cutoff correlates with mortality in CVD. Specifically, a recent study investigated VAF cutoffs for clonal hematopoiesis with DNMT3A and TET2 in patients with chronic ischemic heart failure (103), suggesting that mortality was increased at much lower levels of clonal hematopoiesis (VAF ≥1.15% for DNMT3A and ≥0.73% for TET2). Intervention trials such as the CANTOS trial using the anti-IL1B agent canakinumab described above have found some success in targeting that pathway, potentially even eliminating the increased CVD risk seen in TET2-mutant CHIP patients, but this result will need to be validated in a specifically designed study. Although pharmacologic interventions aimed at the likely mechanistic links between clonal hematopoiesis and CVD are of interest, designing and executing prospective trials modifying “classical” CVD risk factors, such as optimal lipid and blood pressure control, are also important. Currently, well-informed recommendations for management of CVD in patients with clonal hematopoiesis have been proposed, but these expert authors emphatically declare that implementation of these recommendations lacks appropriate prospective clinical evidence (1). Given how common both CVD and clonal hematopoiesis are, any type of risk-reducing intervention—even if it only allows for a moderate reduction in risk—would contribute to a large improvement in public health.

The second high-risk category where prospective clinical trials should be pursued are individuals with cancer undergoing chemotherapy and/or radiotherapy who harbor a significant TP53, PPM1D, or CHEK2 CHIP clone. A conceivable intervention to be tested in a prospective trial would be whether chemotherapy and/or radiotherapy should be avoided in these individuals. This could be refined to only specific agents/procedures such as platinum-based chemotherapy or radiotherapy, which are associated with the greatest risk of clonal selection (73). Indeed, using a synthetic modeling approach, Bolton and colleagues estimated that for patients with breast cancer undergoing adjuvant therapy, who were simultaneously at the highest risk of developing a myeloid neoplasm based on clonal hematopoiesis and blood count parameters, adjuvant chemotherapy increased the absolute risk of myeloid malignancy by 9%, exceeding the expected overall survival benefit of the adjuvant therapy (73). Similarly, because patients with clonal hematopoiesis receiving ASCT for lymphoma experience a markedly increased risk of tMNs (48), modifications to clinical management—including alternatives to ASCT for patients with DDR-mutant clonal hematopoiesis and lymphoma—could be conceived. Indeed, there is an emerging view, likely to be adopted by some clinicians, to “avoid prescription of medications that could further predispose patients to acquiring additional somatic mutations that could further escalate the patient's evolution to acute myeloid leukaemia” (74). However, although there is sufficient reason to “believe” that the potential benefits of receiving adjuvant chemotherapy or ASCT in some settings does not outweigh the risks of this therapy in individuals with TP53, PPM1D, or CHEK2 CHIP, we would strongly encourage the field to pursue rigorous, prospective clinical testing of such a hypothesis to obtain “proof.” Given the large number of individuals diagnosed with cancers for which they would undergo adjuvant chemotherapy, radiotherapy, and/or ASCT, and the increasing clinical sequencing of peripheral blood or plasma in patients with cancer, recruitment for such studies in a reasonable time appears feasible.

The discovery of clonal hematopoiesis and elucidation of mechanisms by which these clonal mutations contribute to development of myeloid malignancies and other nonmalignant diseases has highlighted the need to develop therapeutic approaches for this condition. Indeed, as more patients with clonal hematopoiesis are identified through sequencing of peripheral blood, the need and demand for such treatments will only increase. Hematologists will increasingly be called on for personalized risk assessment and preventative medicine related to clonal hematopoiesis, as is already happening at many centers. Although we are continually adding to our risk stratification of clonal hematopoiesis and identification of high-risk individuals, evidence-based clinical interventions are lacking. Clonal hematopoiesis presents numerous challenges for clinical trials and therapeutic development including low absolute risk of progression, long latency to development of disease, need for surrogate endpoints with regulatory uncertainty, and tolerance of only minimal toxicities in otherwise healthy individuals. However, it is only with rigorous, prospective trials that we can bring evidence-based treatments to individuals with clonal hematopoiesis.

T. Köhnke reports grants from American Society of Hematology during the conduct of the study; grants from Deutsche Forschungsgemeinschaft outside the submitted work. R. Majeti reports grants from NIH, Leukemia and Lymphoma Society, Mark Foundation, Paul G. Allen Frontiers Group, the RUNX1 Research Program, and the Edward P. Evans Foundation during the conduct of the study; personal fees and other support from BeyondSpring Inc., and Coherus Biosciences, personal fees from Acuta Capital Partners, other support from Kodikaz Therapeutics and Gilead Sciences, grants and other support from CircBio Inc, other support from Pheast Therapeutics outside the submitted work. R. Majeti is on the Board of Directors of BeyondSpring and CircBio and the Advisory Board of Coherus Biosciences, Kodikaz Therapeutic Solutions, and Syros Pharmaceuticals; has research funding from CircBio and Gilead; is a consultant for Acuta Partners; has patent royalties from Gilead; and is founder/has equity in CircBio, Pheast Therapeutics, and MyeloGene.

This work was supported by NIH grants 1R01HL142637 and 1R01CA251331, the Stanford Ludwig Center for Cancer Stem Cell Research and Medicine, and the Blood Cancer Discoveries Grant Program through The Leukemia and Lymphoma Society, The Mark Foundation for Cancer Research, The Paul G. Allen Frontiers Group, the RUNX1 Research Program, and a Discovery Research Grant through the Edward P. Evans Foundation (all to R. Majeti). T. Köhnke is a recipient of an American Society of Hematology Research Restart Award and R. Majeti is a recipient of a Leukemia and Lymphoma Society Scholar Award.