Clonal hematopoiesis, somatic mosaicism, and age-associated disease

Abstract

Somatic mosaicism, the occurrence of multiple genetically distinct cell clones within the same tissue, is an evitable consequence of human aging. The hematopoietic system is no exception to this, where studies have revealed the presence of expanded blood cell clones carrying mutations in preleukemic driver genes and/or genetic alterations in chromosomes. This phenomenon is referred to as clonal hematopoiesis and is remarkably prevalent in elderly individuals. While clonal hematopoiesis represents an early step toward a hematological malignancy, most individuals will never develop blood cancer. Somewhat unexpectedly, epidemiological studies have found that clonal hematopoiesis is associated with an increase in the risk of all-cause mortality and age-related disease, particularly in the cardiovascular system. Studies using murine models of clonal hematopoiesis have begun to shed light on this relationship, suggesting that driver mutations in mature blood cells can causally contribute to aging and disease by augmenting inflammatory processes. Here we provide an up-to-date review of clonal hematopoiesis within the context of somatic mosaicism and aging and describe recent epidemiological studies that have reported associations with age-related disease. We will also discuss the experimental studies that have provided important mechanistic insight into how driver mutations promote age-related disease and how this knowledge could be leveraged to treat individuals with clonal hematopoiesis.

CLINICAL HIGHLIGHTS.

• Throughout life, we accumulate somatic mutations in all cells throughout the body. While most of the time these mutations have little phenotypic consequence to the cell, sometimes they provide the cell with a competitive advantage allowing for its clonal expansion. As a result, tissues are composed of cells with slightly different genotypes, a phenomenon known as somatic mosaicism.

• It is now appreciated that somatic mosaicism is a major hallmark of aging and occurs across a wide range of tissues without overt malignancy. Despite this, we are really only at the tip of the iceberg in understanding the impact of somatic mosaicism on human health and disease.

• Over the past decade, considerable progress has been made in understanding the implications of somatic mosaicism in the blood, which is often referred to as clonal hematopoiesis. Various drivers of clonal hematopoiesis have been identified, including blood cancer driver gene mutations, mosaic chromosome alterations, and loss of sex chromosomes, particularly loss of the Y chromosome. However, it is also understood that there are numerous unidentified drivers that are associated with clonal hematopoiesis.

• Large-scale epidemiological studies have revealed that clonal hematopoiesis is incredibly prevalent in the elderly. In fact, studies estimate that by middle age almost all individuals will have small clones with known driver mutations.

• Clonal hematopoiesis has been associated with an increase in all-cause mortality. This increase in mortality likely reflects associations with a myriad of age-related diseases, particularly cardiovascular disease, the leading cause of death worldwide.

• Animal models of clonal hematopoiesis have proven to be highly useful in understanding the relationship between clonal hematopoiesis and disease and, in some instances, have been predictive of subsequent clinical findings. It has been revealed that mutations in certain drivers can causally contribute to disease progression, particularly cardiovascular conditions. Moreover, the models and findings have shed light on the mechanisms by which different drivers contribute to disease pathology.

• From these studies, it is evident that clonal hematopoiesis is a novel risk factor for the development of many age-associated diseases.

• It is the hope that understanding the effects of somatic mosaicism on human health will not only help to provide a better understanding of the processes that contribute to disease but will help pave the way for personalized therapies for individuals with specific mutations.

1. INTRODUCTION

Aging is a complex biological process, characterized by the progressive decline in tissue and organ function (1). Over the past century, there has been a substantial increase in life expectancy, in part due to improvements in living conditions and medical care, although this has not translated into a proportional increase in healthy life expectancy (2, 3). Indeed, aging is associated with numerous diseases, many of which are chronic, including cardiovascular disease, stroke, type 2 diabetes, dementia, and cancer. Moreover, it is not uncommon for individuals to have more than one condition, as age-related diseases are often interconnected and share common risk factors (4). Overall, age-related diseases pose an enormous financial burden on healthcare systems worldwide (5). With the number of older individuals expected to double worldwide by 2050, this problem will undoubtedly worsen (3). While experimental and clinical evidence suggests that slowing biological aging leads to the prevention and/or delay in many age-related diseases (6), the underlying molecular mechanisms connecting aging with age-related disease remain incompletely understood.

Aging is a heterogeneous process where the rate of decline varies enormously between individuals. For instance, some people succumb to age-related disease in their 60s, while others are still active at 100 years of age or older. Understanding the heterogeneity of human aging has been of great interest, and research over the past two decades has focused on identifying specific genes and mutations that control aging, influence age-related disease, and extend human life span (1). Much of this research has focused on common germ-line mutations, which are those inherited from one’s parents and are present in all tissues throughout the body. However, mutations also arise spontaneously in somatic cells over the course of an individual’s lifetime. For instance, it has been estimated that by 50 years of age an individual has an average of 1,000 or more single nucleotide mutations within each cell (7) with some estimates in excess of 10,000 mutations per cell (7, 8). These mutations are known as somatic or postzygotic mutations and, in most cases, result in no or little phenotypic consequences to the individual (9). This is because the mutation can be deleterious to the function of the cell, resulting in immediate death or senescence (9). The mutation may also occur in a noncoding region, thus having very little impact on the function of the cell (i.e., neutral mutation). In a few instances, however, a somatic mutation confers an advantage to the cell, such as by promoting its proliferation, self-renewal, and/or survival, leading to the progressive expansion of the mutant clone (i.e., clonal expansion) (10). As a result, the individual becomes a mosaic of cells with different genotypes, a phenomenon known as somatic mosaicism (9–11). Somatic mutations, which confer a competitive advantage to cells, are often referred to as driver mutations. Like germ-line mutations, somatic mutations can give rise to disease and a well-known example of this is cancer (10, 12). With advances in sampling techniques, sequencing methodology, and the establishment of large tissue repositories, we are becoming increasingly aware that somatic mutations occur in otherwise healthy tissues and may contribute to nonmalignant diseases (i.e., diseases other than cancer) (7, 13–15). The following section of this review will focus on somatic mosaicism and its observation in human aging, as this provides a framework for better understanding the large number of studies on somatic mosaicism in the hematopoietic system.

2. SOMATIC MOSAICISM: AN AGE-RELATED PHENOMENON

With advancing age, we accumulate mutations in our somatic cells and thus somatic mosaicism can be viewed as a hallmark of aging in many tissues (16). For decades, it has been thought that most mutations occur due to errors made during the DNA replication process and thus are particularly prevalent in highly proliferative tissues (17). However, newer research using advanced sequencing techniques has reported that the rates of mutagenesis are similar between nondividing cells and those that are mitotically active, suggesting that mutations occur independently of cell division (7). Despite this, it is likely that tissues with higher proliferation rates exhibit greater amounts of clonal expansion of cells with mutations, as it is easier for the mutations to spread to progeny cells within the tissue. It is also noteworthy that mutations are also common in tissues exposed frequently to mutagens, such as the skin, which is regularly exposed to ultraviolet (UV) radiation from the sun (15, 18). Comparative studies of mammals show that the somatic mutation rate varies dramatically between species and appears to inversely correlate with life span, perhaps suggesting that mutational rate can be a key determinant of life span (19).

Somatic mutations can be classified into different types depending on the effect they may have on genetic structure. For instance, small-scale mutations generally affect a single gene or a couple of nucleotides and include point mutations (i.e., mutations involving a single nucleotide). Some examples of small-scale mutations include insertions and deletions (indels) within a gene and substitution mutations. Large-scale mutations involve mutations to chromosome structure such as amplifications, deletions, rearrangements, changes in copy number, and loss of heterozygosity. Another type of mutation that is frequently observed in human tissue is aneuploidy, which refers to an abnormal number of chromosomes, due to loss or gain of a chromosome. In somatic cells, aneuploidy is thought to arise from errors made in chromosome segregation during cell division (20). Aneuploidy tends to be more deleterious to cells and appears to affect certain chromosomes more frequently than others, presumably as some chromosomes contain genes more critical for cell survival (21). As would be expected, the phenotypic consequences of the different types of somatic mutations differ and are largely dependent on the size of the event and the genomic region affected (20).

2.1. Somatic Mosaicism in Blood

One of the earliest indications of somatic mosaicism in otherwise healthy human tissue arose from a cytogenetic study examining chromosomal abnormalities in human blood across the life span. In this study, the authors counted the number of chromosomes within the blood cells of 247 individuals, reporting that chromosomal abnormalities in human blood cells increase with age (22). It was found that the proportion of aneuploid cells increased with age in both males and females, but interestingly the proportion of hypodiploid cells (i.e., cells with <45 chromosomes) was more common in males (22). While the authors were unable to determine exactly which chromosomes were aneuploid, a partial analysis could be made using the Denver classification, whereby chromosomes are grouped together based on their size and position of the centromere. It was suggested that the increased hypodiploid cells observed in males may be due to loss of the Y chromosome. Later studies confirmed that healthy men are indeed susceptible to loss of the Y chromosome in their blood cells and this increases in frequency with age (23, 24). This phenomenon is known as mosaic loss of the Y chromosome (mLOY), and we now understand it is the most frequently observed somatic change in human leukocytes (25, 26). In particular, recent studies indicate that by 70 years of age, between 10 and 40% of men have lost their Y chromosome in at least 10% of their blood cells (25–30). While mosaic loss of the X chromosome (mLOX) in blood cells also occurs in women, it appears to be less frequent than mLOY and tends to preferentially affect the inactive X chromosome (31, 32). It is currently unclear as to whether mLOY represents a clonal event (i.e., cells with this type of mutation expanding over time) or whether LOY in cells is a common event that is tolerated by leukocytes. In addition to mLOY, several studies have reported the occurrence of mosaic alterations in autosomal chromosomes (mCA) in human blood cells, including losses and gains of sections or entire chromosomes as well as loss of heterozygosity (32). Studies suggest these types of aberrations are less common than mLOY but similarly increase in frequency with age (32–34).

Another type of somatic mosaicism frequently observed in the blood of otherwise healthy individuals is the type driven by mutations in genes recurrently mutated in hematological malignancies (35–37). This has been termed clonal hematopoiesis of indeterminate potential (CHIP) (38). Initial studies using whole exome sequencing found that the frequency of these mutations increases with age, with ∼10% of individuals over the age of 70 possessing at least one clonal mutation in their white blood cells (35). However, more recent studies using deeper sequencing technologies suggest that clonal hematopoiesis is ubiquitous by middle age or earlier (39, 40), while mutations in numerous candidate driver genes have been associated with these clonal events within the hematopoietic system, most are known to occur in just three genes, specifically DNA methyltransferase 3A (DNMT3A), ten eleven translocation 2 (TET2), and additional sex combs like 1 (ASXL1) (35, 37). Although it has been identified that individuals that harbor these mutations have an increased risk of subsequently developing blood cancer, epidemiological studies show that most will never develop a malignancy as these conditions are relatively rare and typically require the acquisition of additional oncogenic mutations (35, 41). Further, to meet the proposed criteria for CHIP, individuals must have an expanded blood cell clone, i.e., be carrying a mutation in a known driver gene of hematological malignancy, at a variant allele frequency (VAF) of 2% or greater without meeting the standard diagnostic criteria for malignancy (i.e., cytopenia and/or abnormal blood cell counts) (38). However, this distinction is largely based on the limits of mutation detection by the sequencing methodology employed rather than on epidemiological data or biological principles. As the depth of sequencing technology has improved since these criteria were developed, we are starting to gain insight into small clones and their impact on human health and disease (42, 43). Thus this definition may need to be revised to reflect our improved understanding of the impact of CHIP on human disease (38, 44). It is also important to note that clonal hematopoiesis can also be detected in the absence of a known driver mutation. This can result from a mutation in an unidentified driver gene, by genetic drift in the hematopoietic stem and progenitor cell (HSPC) pool during normal aging, or by a heritable epigenetic trait (30, 32, 45). In fact, studies estimate that unknown driver mechanisms may account for a substantial proportion of clonal events within the hematopoietic system (30, 37, 46, 47). It also should be noted that it has been reported that mutations may coexist within the same clone, particularly mutations in known driver genes and Y chromosome loss (48, 49). Therefore, age-related clonal hematopoiesis is an alternative term used to describe the presence of clonal events observed in the hematopoietic system that is associated with advancing age (45). Somatic mosaicism within the hematopoietic system, including those that arise from mCAs and mLOY, will be discussed in greater depth in later sections of the review.

In addition, studies have detected the presence of somatic mutations in T cells, which appear to have a different mutational spectrum than what is typically observed in CHIP and many lymphoid cancers (50–56). The exact origin of these mutations remains unclear; it is unknown whether they originate from HSCs or further downstream. However, several studies have found these mutations occur mainly within CD8+ T cells, and it has been suggested that they arise during clonal expansion following antigen exposure (50, 52–54). Nevertheless, mutations have also been detected in CD4⁺ T cells (56). It is hypothesized that these mutations provide the T cells with an advantage in disease settings, as it has been reported that many of the mutations occur in immune and proliferative genes, such as signal transducer and activator of transcription 3 (STAT3) (50–56). In support of this, studies have reported that T cells with somatic mutations are present in individuals with immune conditions, such as aplastic anemia (53), immunodeficiency (51), rheumatoid arthritis (54), multiple sclerosis (52, 55), celiac disease (57), and chronic graft versus host disease (56) perhaps suggesting they play a role in disease pathogenesis. However, there is a dearth of studies addressing causality. While it has been documented that these mutations can also occur in healthy controls (50), the extent to which they occur in the general population has received limited attention and thus is an area for future investigation. There is also evidence to suggest that T-cell somatic mosaicism is associated with advancing age (54), like other forms of somatic mosaicism in the blood described above.

2.2. Somatic Mutations in Other Organ Systems

We are also beginning to appreciate that somatic mosaicism is an age-related phenomenon that occurs across a wide range of tissues that are otherwise histologically normal, as shown in FIGURE 1 (17, 58). Some of tissues include the skin (18, 58–63), esophagus (58, 64–66), liver (58, 67–70), colon (7, 58, 70, 71), endometrium (72), bladder (58, 73, 74), bronchus (75), prostate (58, 76), and pancreas (58, 77). Furthermore, clonally expanded de novo mutations have been reported in neurons (7, 78, 79), cardiomyocytes (8), smooth muscle (7), and even in embryonic or induced human pluripotent stem cells (80). The level of detectable somatic mosaicism and clonality varies across tissues and is influenced by exposure to mutagens and by the level of cell proliferation (17, 18, 58, 67, 71, 81). Moreover, recent work would suggest that, for some tissues, clones grow and expand in parallel and thus many tissues are composed of multiple clones with mutations in different driver genes (18, 61, 64, 67, 72). It is also important to highlight that while the presence of a driver gene mutation represents an early step toward malignancy, in many cases, mutant cells do not become malignant, as this requires the acquisition of multiple driver mutations (41). Thus somatic mosaicism can be viewed as a relatively benign phenomenon (15). It is also noteworthy that somatic mutations may also occur during embryonic development, leading to mosaicism (78, 82). These mutations usually occur in specific lineages or tissues and often have a higher variant allele fraction (VAF) than what is observed for somatic mutations occurring later in life (15, 83). Moreover, they may give rise to rare congenital and developmental diseases, such as vascular malformations, Turner’s syndrome, and many others. The following paragraphs will discuss some of the recent work documenting the occurrence of somatic mutations between normal tissue and its association with aging.

FIGURE 1.

Summary of some of the organs where somatic mosaicism has been documented.

It is perhaps not surprising that human skin, which is frequently exposed to UV light, harbors a substantial number of clonally derived mutations. Several studies have reported the presence of clonal patches of skin cells carrying mutations in the tumor suppressor protein 53 gene, TP53, and these appear to increase upon UV exposure (59, 60, 84). Moreover, a study led by Martincorena et al. (18) documented that 18–32% of normal eyelid skin cells had positively selected “driver” mutations, including those which are frequently observed in squamous cell carcinomas, including notch receptor (NOTCH) 1, NOTCH2, FAT atypical cadherin 1 (FAT1), and TP53 (18). The presence of mutations in genes not commonly mutated in skin cancers such as RNA-binding motif 10 (RBM10) and fibroblast growth factor receptor 3 (FGFR3) were also noted, suggesting that there may be a class of genes in which somatic mutations provide a clonal selective advantage in normal tissue but do not necessarily cause disease (18). A more recent study performed deep sequencing on skin epidermis excised from multiple locations on the body. As would be predicted, the total number of mutations was highest in areas more often exposed to sunlight, such as the head when compared to areas less frequently exposed, such as the leg or abdomen (61). It appeared that the number of mutations increased with age and interestingly different sites appeared to harbor mutations in different sets of genes. Specifically, TP53 mutations were more commonly selected in the head whereas mutations in FAT1 and NOTCH1 were more frequent in the leg (61). This may indicate that exposure to UV light not only functions as a mutagen but can also modulate the fitness of different clones thus contributing to the clonal architecture of skin.

The skin and inner layer of the esophageal lumen have a very similar structure with both being composed of stratified epithelial cells. These epithelial cells are continuously shed from the tissue surface and are replaced by dividing stem cells in the deep basal layer (85). However, the external environment of these tissues is considerably different, which would conceivably alter the mutational landscape. A recent study examined the mutational landscape of normal human esophagus from nine healthy donors across the life span (64). It was revealed that the normal esophagus is composed of large patches of clones with mutations in cancer driver genes and these appeared to increase with age. Some of these driver genes included NOTCH1, NOTCH2, NOTCH3, FAT1, and TP53, which are similar to what was observed in normal skin (64). Surprisingly, it was noted that there was a higher frequency and size of mutant clones in normal esophagus than in sun-exposed skin, although the number of mutations per cell in normal esophagus was less than 10% of what was found in skin (64). This may suggest that there is a stronger environment for positive selection of certain clones within the esophagus although cells with mutations are less likely to become malignant due to the smaller mutational burden in each cell. It was also noted that the number of mutations in the driver gene NOTCH1 was several times higher in normal tissues when compared with esophageal squamous cell carcinoma (ESCC), raising the possibility that ESCCs may be less likely to develop from cells with NOTCH1 mutations (64). Another study reported similar findings for the esophagus using a larger cohort of samples and documented an age-associated increase in clone size and number, particularly involving NOTCH1 mutations (65). It was also found that lifestyle factors such as smoking and heavy alcohol consumption greatly accelerated clone size and number (65). Interestingly, it has been recently reported that highly fit mutant clones in normal epithelium may function to eliminate emerging tumors (86). As ESCCs occur almost exclusively in smokers or heavy drinkers, it could be postulated that exposure to these risk factors may hinder the growth of “normal” mutant clones and allow “cancer-driving clones” to expand, leading to malignancy (65, 86).

Studies have also examined somatic mosaicism in the human intestine and colon. The turnover of cells within intestinal tissue differs from that of the esophagus and skin, whereby cells are replaced by a small number of epithelial stem cells that reside in the bottom of structures called crypts. Using a laser-capture microdissection approach, one study isolated 2,035 individual colonic crypts from the normal epithelium of 42 individuals aged 11 to 78 and sequenced their genomes (71). The mutational load for many mutational signatures appeared to increase with age, as would be expected. It was found that the proportion of normal colorectal epithelial cells with driver mutations was ∼1%, which is substantially lower than that observed in normal skin (30%) and esophagus (>50%) (71). The authors postulated that this may be because each crypt contains a very small number of stem cells, which reduces the probability of a cell with a driver mutation outcompeting its wild-type neighbors (71). However, if a mutant stem cell does manage to colonize the crypt, the mutant cells will not expand beyond the crypt unless crypt fission occurs, which is rare (71). It was also found that driver mutations commonly observed in colorectal cancer, such as adenomatous polyposis coli (APC), TP53, and kirsten rat sarcoma viral oncogene homolog (KRAS), were infrequent in normal colonic crypts (71). Further, normal colonic crypt cells tended to have smaller mutational loads in comparison with colorectal cancer cells. Collectively, this may suggest that colorectal cancer is a result of a mutation in a specific driver gene, and it appears that further studies are warranted to understand why the incidence of colorectal cancer is considerably higher within the general population, given the small proportion of normal crypt cells with driver mutations when compared to other tissues. In comparison with healthy colonic epithelium, colonic tissue from inflammatory bowel disease (IBD) patients exhibits substantially accelerated tissue remodeling and somatic mosaicism (87). IBD, such as ulcerative colitis and Crohn’s disease, is characterized by recurring cycles of inflammation and tissue destruction (88), providing a unique environment for clones to expand. It has been found that IBD has more than double the mutational rate compared to normal colonic epithelium and accelerates age-related mutational processes (89). Patients with IBD frequently harbor mutations in PIGR and genes in Toll-like receptor and interleukin-17 (IL-17) pathways, which appear to provide a competitive advantage for expansion in the presence of inflammation (90, 91). These mutations are not those typically observed in normal colonic epithelium or in colorectal cancers, perhaps suggesting that these clones may be involved in the pathogenesis of IBD (90).

The clonal architecture of healthy and cirrhotic liver has also been examined by sequencing the whole genome of hepatocytes. In one study, it was found that structural variants and copy-number alterations occurred in moderate numbers across patients with liver cirrhosis, despite being rare in normal liver (67). Further, when the analysis was focused on select driver genes, mutations were found in the tumor suppressor genes, activin receptor type-2A (ACVR2A), AT-rich interactive domain (ARID) 2, ARID1A, and tuberous sclerosis complex 2 (TSC2), and these appeared to occur in both normal and cirrhotic liver (67). However, these mutations were not frequently observed in hepatic cell carcinoma samples (67). Moreover, candidate driver mutations were identified in only 1–5% of clones, which is substantially lower than that observed in normal skin and esophagus (67). It was also noted that samples from healthy controls were highly polyclonal and there was little genetic relatedness even between closely located regions of microdissection (67). In contrast, cirrhotic liver samples were usually monoclonal or oligoclonal, and mutations often extended in microdissection regions that were millimeters apart (67). It was also found that there was no clonal amplification of mutant cells across fibrotic bands of tissue, suggesting that the fibrotic tissue restricts the growth of mutant clones (67). A follow-up study by the same group identified several potentially “protective gene” mutations in some patients with alcoholic and nonalcoholic fatty liver disease, namely those in Forkhead box O1 (FOXO1), cell death inducing DFFA-like effector b (CIDEB), and glycerol-3-phosphate acyltransferase mitochondrial (GPAM) (69). These genes regulate lipid processing and storage, and thus mutations are thought to protect hepatocytes from lipotoxicity, which is common to the pathology of both these conditions. When comparing different sections of the liver within the same patient, it was observed that there were independent clones with mutations in the same gene suggesting that the environment of fatty liver disease may select for expansion of these clones. However, among different patients, there was considerable heterogeneity in the mutation frequency and specific gene affected, with many not harboring a putative protective gene mutation. Mutations in these genes were rarely found in hepatocellular carcinoma samples. From these findings, it is tempting to speculate that fatty liver disease patients with mutations in FOXO1, CIDEB, and GPAM may be protected from subsequently developing hepatocellular carcinoma, although this warrants further investigation. Another study that exclusively examined the mutational landscape of cirrhotic liver found that a different set of driver genes were mutated, specifically, polycystin 1, transient receptor potential channel interacting (PKD1), PPARG coactivator 1 beta (PPARGC1B), lysine methyltransferase 2D (KMT2D), and ARID1A (68). Further, it was also found that the number and size of the mutant clones increased with indexes of liver damage and fibrosis (68). Using an in vivo CRISPR screening approach in mice, the authors demonstrated that loss of Pkd1, Kmt2d, and Arid1a promoted clonal expansion (68). Interestingly, conditional heterozygous deletion of these genes in mice was also hepatoprotective in in vivo injury assays and promoted regeneration. Collectively, these findings highlight that not all somatic mutations are deleterious and can in some instances provide protection or promote regeneration.

2.3. Clonal Hematopoiesis

In comparison to many other tissues, substantial inroads have been made in understanding the impact of somatic mutations in the hematopoietic system on human disease. The reasons for this are manifold and likely, to some extent, reflect the recent discoveries that hematopoietic somatic mutations are associated with an increase in all-cause mortality and cardiovascular disease (92), attracting research interest in this area. First, it should be noted that in comparison to other tissues, the hematopoietic system is considerably easier to study. For instance, human blood can be sampled noninvasively, therefore avoiding the need for deceased organ donations, biopsies, or procedures involving surgical excision. Further, the blood is an admixture of cells, where cells can freely mix, unlike cells in solid tissues (93), allowing researchers to obtain a better idea of the clonal dynamics of mutant clones and their growth over time. The hematopoietic system is also a highly proliferative tissue, accounting for ∼86% of daily cellular turnover, thus making it easier for clones to propagate (94). Additionally, blood cells circulate to all organs of the body and therefore are likely to influence multiple organ systems beyond the hematopoietic system (95). More specifically, the immune system plays various physiological roles in tissues, such as regulating inflammation, tissue repair and regeneration, tissue growth, and neuroendocrine support; therefore, it is possible that mutations in blood cells may impact multiple of these physiological processes (96, 97). Furthermore, it is well established that human aging is associated with chronic low-grade inflammation, characterized by elevated levels of C-reactive protein and interleukin (IL)-6 (98). This is often referred to as “inflammaging,” and studies suggest it accelerates biological aging and age-related disease (98–100). Recent work has found that somatic mutations in HSPCs can affect the inflammatory profile of leukocyte progeny by augmenting the production of various proinflammatory cytokines (101–106). Indeed, studies have demonstrated that clonal hematopoiesis with driver mutations in certain genes [i.e., DNMT3A, TET2, Janus kinase 2 (JAK2), TP53, protein phosphatase Mn²⁺/Mg²⁺-dependent 1D (PPM1D)] can accelerate cardiovascular disease by augmenting the inflammatory profile of leukocytes (101–103, 105, 107). Given these recent findings, it raises the possibility that clonal hematopoiesis may contribute to the processes of inflammaging, which in turn may accelerate aging and age-related disease (108). The following sections of this review will discuss the recent evidence supporting this notion and aim to connect the dots between clonal hematopoiesis, aging, age-related disease, and inflammation. While this will focus primarily on clonal hematopoiesis associated with mutations in driver genes (i.e., CHIP), there will also be a discussion of clonal hematopoiesis associated with mCAs and mLOY. These types of clonal hematopoiesis are summarized in FIGURE 2.

FIGURE 2.

Simplistic summary of types of mutations observed in age-related clonal hematopoiesis. Created with BioRender.com, with permission. HSCs, hematopoietic stem cells; mCAs, mosaic chromosomal alterations; mLOY, mosaic loss of the Y chromosome. Diagram adapted from Ref. 229, with permission from Nature Reviews Genetics.

3. EPIDEMIOLOGY OF CLONAL HEMATOPOIESIS

3.1. Age-Related Clonal Hematopoiesis: Driver Gene Mutations

As discussed in the previous section, clonal hematopoiesis is strongly associated with advancing age (35–37). Some of the earliest indications of this originated from studies examining patterns of maternal to paternal X-chromosome inactivation in leukocytes of healthy females (109). This method was originally developed to assess the clonal nature of cancer, as each cell undergoes random inactivation of one of its X chromosomes during female embryonic development. Therefore, a skewing from the predicted 50:50 ratio of maternal:paternal X inactivation pattern is indicative of a clonal expansion. In these studies, it was observed that females older than 60 years of age were often observed to exhibit skewing of one of their inactive X chromosomes compared to younger females (109). A subsequent study over a decade later identified that females who exhibited inactive X chromosome skewing in their blood were more likely to also possess a mutation in ten eleven translocation 2 (TET2), a blood cancer driver gene, suggesting that this particular genetic mutation was playing a role in driving clonal expansion of blood cells within these women (110). These findings have been corroborated by more recent studies that have detected the presence of commonly occurring driver mutations in the peripheral blood of elderly individuals (35–37). As briefly discussed in the previous section, in 2014 three independent epidemiological studies involving a total of ∼32,000 individuals employed whole exome sequencing of peripheral blood to assess the frequencies of mutations in genes that are recurrently mutated in hematological malignancy (35–37). It was found that before the age of 40 mutations in driver genes were rare, occurring in less than 0.1% of individuals. However, by 70 years of age, ∼10% of individuals had a mutation in their white blood cells (35–37). The majority of individuals had mutations in the epigenetic regulators, DNMT3A, TET2, and ASXL1 (35, 37). However, it was noted that there were a smaller number of mutations in other driver genes, such as JAK2, TP53, PPM1D, and splicing factor 3b subunit 1 (SF3B1) (35, 37). The study by Genovese et al. (37) also performed nonbiased exome sequence analysis, which allowed for the detection of clonal hematopoiesis in individuals who lacked a putative driver gene. This study found that clonal hematopoiesis in the absence of known driver genes accounted for ∼40% of all detectable clonal events and these mutations also increased with age (37). Notably, two of these studies found that clonal hematopoiesis was associated with a substantial increase in all-cause mortality, which could not be accounted for by death due to hematological malignancy (35, 37). An unplanned secondary analysis by Jaiswal et al. (35) revealed that the large increase in mortality observed in individuals with driver gene-mediated clonal hematopoiesis could be attributed to an increase in death due to cardiovascular conditions. The association between clonal hematopoiesis, all-cause mortality, and cardiovascular disease will be discussed in detail in forthcoming sections of this review.

Since these initial reports, there have been several other studies examining the prevalence of driver gene-mediated clonal hematopoiesis across the life span. A study led by Zink et al. (30) employed barcoded nonbiased whole genome sequencing to detect clonal hematopoiesis in the peripheral blood of 11,262 Icelanders. Of note, this study did not select for candidate driver gene mutations and thus the prevalence of clonal hematopoiesis was considerably higher than previous estimates (30). More specifically, it was found that the frequency of clonal events increased from 0.5% in individuals younger than 35 years to more than 50% in individuals older than 85 years (30). Interestingly, when the frequency of mutations in candidate driver genes was examined, it was found that they only accounted for 12.6% of the total number of clonal events, suggesting there are other mechanisms or unidentified driver gene mutations that contribute to clonal hematopoiesis (30). In support of this notion, a recent study examining synonymous mutations (i.e., mutations that do not result in an amino acid change) in the blood found that the majority of these mutations reach a high VAF by hitchhiking with driver gene mutations. The authors noted that the majority of the high VAF synonymous variants did not cooccur with known driver genes, suggesting that there are numerous unidentified drivers of clonal expansion (46). Moreover, advances in sequencing technology have allowed the detection of mutations at an even greater depth, and data arising from this new technology further suggest that clonal hematopoiesis is more prevalent than previous estimates, even when the analysis only focuses on candidate driver genes. For instance, ultra-deep sequencing of mutational hotspots within select candidate driver genes has projected that the mean mutation frequency is >12% of individuals in their 60s, >19% in their 70s, >40% in their 80s, and >74% in their 90s (111). Similarly, a study using error-corrected sequencing, which allows detection of clonal events as low as 0.03% VAF, found that mutations in the candidate driver genes DNMT3A or TET2 occur in 95% of individuals aged 50–70 years. Notably, previous studies estimated that only 5% of individuals in this age group had detectable clonal hematopoiesis (39). This would suggest that clonal hematopoiesis characterized by small VAF clones is almost ubiquitous by middle age. From these studies, it is apparent that the prevalence of clonal hematopoiesis is largely dictated by the sequencing method used to identify mutant clones as well as the sequencing depth. These studies are summarized in FIGURE 3.

FIGURE 3.

The prevalence of clonal hematopoiesis depends on the sequencing approach and limit of detection. Less sensitive methods, such as whole exome sequencing restricted to candidate driver genes, estimate a lower prevalence of clonal hematopoiesis. Highly sensitive techniques, such as error-corrected sequencing, predict that clonal hematopoiesis is much more prevalent. DNMT3A, DNA methyltransferase 3A; TET2, ten eleven translocation 2; VAF, variant allele frequency.

More recent studies suggest that driver mutations can be acquired at a relatively young age and expand over the course of an individual’s lifetime. A study by Williams et al. (112) examined the time of acquisition of driver mutations in MPN patients by constructing phylogenetic trees of mutations. Using this approach, the study estimated that driver mutations in DNMT3A and JAK2 could be acquired in utero or during childhood (112). Another study by Fabre and coworkers (113) examined the longitudinal dynamics and natural history of clonal hematopoiesis in a cohort of 385 individuals older than 55 years. Using multi-time point sequencing, the authors determined that many clones expand at a stable exponential rate over time that appeared to be dependent on the identity of the mutated driver gene. When the authors used these estimated growth rates to trace back the time of mutation acquisition, the projected ages of some clones preceded conception. This was especially common for DNMT3A mutations; however, it was less common for TET2 and very uncommon for splicing factor gene mutations. While these age estimates are evidently inaccurate, it was hypothesized that certain clones are acquired early during life and perhaps expand at a faster rate during young age, accounting for an earlier than possible acquisition estimate. Interestingly, it was found that many clones showed a decelerated expansion rate in old age; however, for some clones (i.e., those with mutations in splicing factors), expansion was accelerated. A partnering study by Mitchell et al. (47) made similar observations using phylogenetic clone analysis. Using DNMT3A as a test gene, they observed that for 13 clonally expanded clades, 2 mutations were acquired before the age of 10, 3 before the age of 20, and the remaining were acquired before the age of 40. Nevertheless, the majority of mutations occurred in unknown driver genes and by 70 years of age, the prevalence of expanded clones (i.e., VAF >1%) was universal, with estimates of 10–20 expanded clones per individual. This contrasts with previous work focusing on CHIP, where it was implied that individuals typically only have one to two expanded clones (35, 37). From the studies above, it can be speculated that by adulthood, most individuals have small clones with driver mutations. Exactly what drives clone expansion in individuals remains speculative, although the findings from the work above would suggest that the specific driver mutation and time of acquisition play a part. Additional findings by Mitchell et al. (47) suggested that perhaps loss of clonal diversity during aging provides a fertile ground for a subset of clones to expand. Nevertheless, additional studies will be required to validate these theories. There also appears to be some level of individual variation, with clones growing in some individuals and not others (39, 114). A recent study suggested that once clones reach 2% VAF they are more likely to grow than clones below this threshold (114). Despite these theories, the mechanisms that enable clonal expansion remain incompletely understood and this topic will be discussed in greater detail in sect. 7.3.

With regard to other demographic characteristics of driver gene-mediated clonal hematopoiesis, it appears to be slightly more prevalent in males than females (35). Nevertheless, other studies have reported that after accounting for additional confounding factors, this sex bias is lost (115). In addition, it appears to be less prevalent in individuals of Hispanic or East Asian ancestry (35, 115). Further, it has been observed that specific germline variants can provide a small predisposition to clonal hematopoiesis. For instance, single nucleotide polymorphisms (SNPs) in TERT, which encodes the telomerase enzyme have been linked to an increased risk of clonal hematopoiesis (30, 115). Additionally, it has been documented that an intergenic region near TET2, which is specific to people of African ancestry, is strongly associated with clonal hematopoiesis (115). Given the ancestral differences that have been observed and the association with germline variants, it highlights the importance of studying clonal hematopoiesis in diverse groups of individuals. It also should be noted that smoking has also been associated with clonal hematopoiesis, and this appears to be driver gene specific, particularly for individuals with ASXL1 driver mutations (37, 115–118). However, whether the relationship between smoking and driver gene-mediated clonal hematopoiesis is causal remains uncertain (118). Finally, a recent report that clonal hematopoiesis is enriched in first responders of the World Trade Center terrorist attacks suggests that high-exposure particulate matter may lead to mutations in certain driver genes (119). Despite this interesting observation, additional studies will be required to validate these findings, particularly in large cohorts of diverse individuals.

3.2. Mosaic Chromosomal Alterations

As indicated in the section above, clonal hematopoiesis with mutations in unknown driver genes accounts for a considerable proportion of clonal events within the hematopoietic system (30, 37, 46, 47). Some of these missing drivers are likely to be mosaic chromosome alterations (mCAs) of either autosomes or the sex chromosomes. Like clonal hematopoiesis mediated by single nucleotide variants and indels in cancer driver genes, mCAs increase with age and appear to be relatively common within the human population (32, 33). The following section will detail the epidemiological studies describing mCAs and their prevalence.

3.2.1. Autosomes.

In 2012, two studies detected the presence of mCAs in the blood of otherwise healthy individuals and reported that the frequency of these alterations increased sharply with age. In particular, Jacobs et al. (120) performed genome-wide SNP arrays on blood or buccal samples (majority of samples from blood) from 31,717 cancer cases and 26,136 controls to examine the frequency of mCAs of autosomes (aneuploidy or copy-number loss of heterozygosity). In cancer-free individuals, it was found that the frequency of detectable alterations increased from 0.23% for individuals less than 50 years to 1.91% for individuals between 75 and 79 years of age (120). Moreover, the frequency of mCAs was higher in individuals with solid organ tumors, and for some types of cancers, it was predictive of later cancer diagnosis (120). Similarly, Laurie et al. (121) performed SNP microarrays on DNA samples (mostly peripheral blood) from 50,000 individuals from birth to old age to determine the frequency of large mosaic chromosome anomalies across the life span. It was found that the frequency of detectable clonal mosaicism was low (<0.5%) from birth until 50 years of age; however, it rapidly rose to 2–3% in the elderly (121). Findings from this study also suggested that mCAs were a risk factor for cancer (121).

It should be noted that the above-mentioned studies employed relatively insensitive techniques to measure chromosomal alterations, whereby at least 5–10% of cells had to harbor a mutation to be detected. Since these two studies, large biobanks have been established containing deep genetic, physical, and health data from large numbers of individuals, i.e., the UK Biobank (122) and Biobank Japan (123). This allows for larger populations to be more comprehensively analyzed for the detection of mCAs and their association with human disease. Furthermore, there have been advances made in computational methods, allowing for more sensitive detection of mCAs (32, 124). Thus, incorporating these two factors, Loh et al. (32) used SNP array data to identify 8,342 mCAs in 151,202 UK Biobank participants. Within the cohort, 71% of the detected alterations were classified as either gains, deletions, or copy-number-neutral loss-of-heterozygosity, and the remaining 29% of events could not be inferred definitively (32). The majority of detected mCAs (5,901 of 8,342) were present at cell fractions below 5%, which is below the detection limit for mCAs in previous studies. Similarly, it was found that mCAs increased with age, from ∼2% in individuals <45 year to ∼5% in individuals >65 years (32). A similar study by Terao et al. (33) aimed to determine the frequency of mCA in 179,417 participants of Biobank Japan. Participants in this population were, on average, older than those in the UK Biobank, with a median age of 62.8 years compared with 57 years for the UK Biobank (33). Given the difference in median age, it is probably not surprising that the estimated prevalence of mCAs in this study was higher than those of the UK Biobank, with ∼15% of all individuals harboring mCAs (33). As expected, the prevalence of mCAs rose sharply with age, with mCAs detected in ∼5% of individuals <30 years to 30–40% of individuals aged 90+ years (33). When comparing the results with that of the UK Biobank, it was found that Japanese and European individuals displayed some differences in the genomic location of mutations in their respective hematopoietic clones (33). As both these studies found differences in heritable loci, it has been suggested that perhaps some of the differences between these two cohorts can be explained by ancestral differences in heritable traits (32, 33). A more recent study analyzed a total of 768,762 individuals for mCAs using combined data from several biobanks and large population cohorts (34). While the primary focus of this study was to examine the relationship between all mCAs (including sex chromosomal loss) and infection, the authors also examined the prevalence of mCAs (34). It was found that the prevalence of expanded autosomal mCAs was 0.27% for individuals aged <40 years, 0.52% for those aged 40–60 years, 1.5% for those aged 60–80 years, and 4.6% for those aged >80 years (34). While these estimates are lower than what was observed for Loh et al. (32) and Terao et al. (33), the detection limit for this study was considerably higher at 10% of all cells (34).

3.2.2. Sex chromosomes: loss of Y chromosome.

As previously discussed, the most common mosaic chromosomal alteration is loss of the Y chromosome (25, 26). Hematopoietic mosaic loss of Y chromosome (mLOY) shows continuous mosaicism in the blood ranging from 0 to 100% of cells without a Y chromosome (125). This phenomenon was first reported in the hematopoietic system over 50 years ago from cytogenetic studies examining chromosomal abnormalities in blood cells (22); however, until relatively recently, very few studies had examined its phenotypic consequences and frequency in large cohorts of individuals (24, 126). In a study aimed at determining the frequency of structural variants in the peripheral blood of 1,141 elderly men (70.7–83.6 years of age at sampling), it was noted that mLOY was the most frequently occurring variant (25). In particular, it was found that 8.2% of individuals had >18% of blood cells with LOY and 2.3% of individuals had >35% of cells with LOY (25). Moreover, there was a significant positive association between LOY and age at genotyping, as would be expected (25). Interestingly, mLOY was also associated with all-cause mortality, particularly mortality due to nonhematological cancers (25). More recent studies have reported higher frequencies of mLOY within the blood, particularly those involving larger cohorts of individuals. For instance, a study led by Zink et al. (30) computed the frequency of peripheral blood mLOY in a sample of 5,039 individuals from the Icelandic DEcode study. This cohort included individuals from birth to old age, and therefore, the frequency of mLOY across the life span could be estimated. As would be expected, the frequency of mLOY rose sharply with age, being rare below 50 years of age (<2.5% of individuals) but rising to ∼40% in individuals >85 years of age (25). These data were corroborated by a later study examining mLOY frequency in 205,011 UK Biobank participants (26). Specifically, using a computational approach to estimate the mLOY frequency, the authors report that the prevalence of mLOY increased from 2.5% at age 40 to 43.6% at age 70 (26). From these data, it has become apparent that mLOY is the most frequently occurring somatic genetic aberration in human leukocytes.

Beyond aging, another significant correlate with LOY is smoking. This was first observed in a study that aimed to identify possible causes of hematopoietic mLOY. In this study, 6,014 men from three independent prospective cohorts were analyzed for mLOY and its association with various environmental, lifestyle, and clinical factors (127). It was found that there was a strong association between smoking and mLOY status in the three independent cohorts with current smokers having a significantly higher degree of mLOY mosaicism, compared with nonsmokers and previous smokers (127). It should be noted that the only other significant factor associated with mLOY was age, which is congruent with previous reports (127). It was also observed that there was a dose-dependent effect, with a higher percentage of cells with mLOY found in heavy smokers (127). These findings have since been supported by additional studies, which have also reported associations between smoking status and hematopoietic mLOY (27, 29, 128). Interestingly, however, this association appears to apply to current smokers and not those who smoked previously (27, 127). In particular, one study noted that the odds ratio of harboring hematopoietic mLOY progressively declined with increasing years after cessation (27). Given these associations with smoking status, it has been hypothesized that smoking may provide a selective advantage to cells with mLOY, facilitating their clonal expansion. Nevertheless, from the current literature, it is not clear whether mLOY represents a clonal event or is just a highly frequent somatic change that is tolerated by leukocytes. It is known that the Y chromosome possesses two tumor suppressor genes, namely ZFY and UTY, and thus it is possible that loss of these genes may facilitate the expansion of cells without a Y chromosome (129). Indeed, a longitudinal study indicated that the fraction of mLOY can increase over time in some individuals (130). However, in other individuals, the portion of LOY has been shown to decrease or stay the same over time, suggesting that changes in the fraction of mLOY cells are highly individualized (130). Further studies are thus required to determine if detectable mLOY is a result of a clonal event or whether it is a highly frequent somatic change, representing a barometer for widespread genomic instability.

In addition to smoking, other studies have associated mLOY with exposure to other environmental factors (131, 132). For instance, one study analyzed blood samples from 933 men ≥65 years of age from a prospective cohort that were analyzed for mLOY, and data were collected on exposure to various forms of outdoor air pollution. It was found that increased exposure to particulate matter (<10 um) was associated with mLOY, but no other association was found for any other air pollutant (131). Given the strong association that has been found between mLOY and smoking (27, 29, 127, 128), the authors postulated a common genotoxic component in both cigarettes and particulate matter may be driving LOY in the blood, such as polycyclic aromatic hydrocarbons (131). Indeed, a subsequent study examined the association between mLOY and exposure to polycyclic aromatic hydrocarbons (PAHs) in 1,005 male coke-oven workers (132). It was reported that mLOY increased linearly with exposure to PAHs, although there was a high proportion of cigarette smokers in this cohort of individuals, which may have confounded the findings (132). Nevertheless, studies using larger cohorts of individuals, ideally with multiple measurements over time, may provide a better insight into the relationship between mLOY and exposure to environmental pollutants.

Beyond exposure to environmental factors, studies suggest that there may be heritable traits that predispose to mLOY (26, 27, 29, 34). It has been reported that the frequency of mLOY is greater in individuals of European ancestry than those of African ancestry (128). Moreover, a substantial proportion of mLOY can be attributed to heritable loci. For instance, with the use of a sample of 67,034 men from the UK Biobank, it was identified that the most significantly associated SNP with mLOY was T-cell leukemia/lymphoma 1 A (TCL1A), which is located on chromosome 14 (29). This gene encodes the protein, which is known to be involved in B- and T-cell malignancies (29). Moreover, it was found that a further 18 autosomal genetic variants were associated with mLOY, many of which affected genes involved in cell cycle progression. Interestingly, the 19 variants identified in this study were also found to predict X chromosome loss in women and thus perhaps indicating that these variants may predispose individuals to generalized genomic instability (29). A subsequent study by Thompson et al. (26) used a larger sample of men from the UK Biobank and identified 137 novel loci associated with mLOY, which included the 19 that were previously identified. Similar to the previous study, the majority of the pathways affected by these variants appeared to affect genes involved in cell cycle regulation and the DNA damage response (26). The majority of these variants overlapped with cancer susceptibility loci and were also associated with health outcomes in women, such as breast cancer and later age of menopause. The authors also linked the variant in TCL1A to the growth advantage of mLOY cells, thus providing a potential explanation as to why clonal expansion of these cells may occur in some individuals. Given these findings, it has been suggested that mLOY may be a biomarker of genomic instability originating from heritable autosomal traits (26).

3.3. Therapy-Related Clonal Hematopoiesis: a Unique Case

Another type of clonal hematopoiesis, which has received less attention than other types, is therapy-related clonal hematopoiesis. This type of clonal hematopoiesis is prevalent in cancer survivors and is generally associated with mutations in genes involved in the DNA damage response pathway, such as TP53 and PPM1D (133, 134). Individuals with this form of clonal hematopoiesis are at increased risk of subsequently developing a therapy-related myeloid neoplasm, particularly for those with TP53 mutations (135, 136). This phenomenon was demonstrated by a study examining clonal hematopoiesis in 401 patients undergoing an autologous stem cell transplant for non-Hodgkin lymphoma (133). This process involves high-dose myeloablative chemotherapy and thus places the hematopoietic system under extreme stress. It was found that 30% of patients had driver gene-associated clonal hematopoiesis and there appeared to be a higher frequency of individuals with mutations in PPM1D and TP53 than that observed in the general population (133). Similar observations were reported for a cohort of 8,810 individuals with nonhematologic cancer (116). In this larger cohort, it was observed that 25% of patients exhibit driver gene-mediated clonal hematopoiesis and mutations in TP53 and PPM1D were significantly associated with prior exposure to chemotherapy (116). Taken together, these studies suggest that clonal hematopoiesis is prevalent in cancer patients, particularly involving mutations in PPM1D and TP53 (116, 133). This likely reflects the fitness advantage that these mutations confer on cells, allowing them to better survive cytotoxic stress (137, 138). This notion was further supported by a study examining driver gene clonal hematopoiesis in 119 lymphoma and myeloma patients comparing those that received chemotherapy to those that did not (139). It was found that PPM1D and TP53 mutations were significantly higher in patients that received chemotherapy compared to those that did not or compared to healthy controls (139). This was not observed for any other driver gene mutation, again supporting the idea that chemotherapy may be selectively facilitating the expansion of PPM1D and TP53 (140). These findings have also been supported by more recent studies, using other cohorts of individuals (137, 140).

While it has previously been posited that chemotherapy may directly cause mutations in PPM1D and/or TP53, studies suggest that this is not the case (135). Specifically, from retrospective analyses of patients that have developed therapy-related neoplasms, it has been shown that mutations are present as very small clones (i.e., below most sequencing detection limits), which subsequently expanded after exposure to chemotherapy (135). Furthermore, chemotherapy-based enrichment of Tp53 and Ppm1d mutations has been shown in experimental studies in mice (135, 137, 141). It also should be noted that only certain types of cancer treatment have been associated with therapy-related clonal hematopoiesis, specifically those that activate the DNA damage-response pathway (137, 140). In a study focusing on PPM1D mutations in patients with therapy-related myeloid neoplasms, it was found that there was a significant association with prior exposure to platinum agents (cisplatin, carboplatin, and oxaliplatin) and etoposide. There was no association observed between PPM1D mutations and other chemotherapeutic agents (i.e., methotrexate, 5-FU, vincristine/vinblastine) or radiation therapy (137). The authors then confirmed these findings using an experimental approach, revealing that PPM1D-mutated cells had a selective advantage over certain classes of chemotherapeutics (137). A later study by Bolton et al. (140) reported similar findings, using prospective targeted sequencing data from 24,146 patients with cancer. In this study, patients had a wide range of tumor types and as a consequence were exposed to many different agents, allowing the authors to examine associations with a wide range of treatment modalities. It was found that driver mutations in PPM1D and TP53 were most strongly associated with cancer therapy (140). In particular, mutations in PPM1D were associated with prior exposure to platinum, radionuclide therapy, topoisomerase I inhibitors, topoisomerase II inhibitors, taxanes, and radiation therapy but not antimetabolites or microtubule-damaging agents (140). Mutations in TP53 were associated with previous exposure to platinum, radiation therapy, and taxane but not topoisomerase I/II inhibitors, antimetabolites, or microtubule-damaging agents (140). In a similar manner, a systemic review of 416 patients with therapy-related myeloid neoplasms found that TP53 mutations were significantly associated with prior treatment with lenalidomide, a thalidomide analog (142). Collectively, these studies demonstrate that clonal hematopoiesis associated with mutations in the DNA-damage response genes, PPM1D and TP53, is common in individuals that have been previously treated for cancer. It also appears that cancer therapy based on the enrichment of PPM1D and TP53 clones is agent specific and likely reflects the ability of the agent to activate the DNA damage response pathway. This particular type of clonal hematopoiesis will be discussed at greater length in the forthcoming experimental sections and is summarized in FIGURE 4.

FIGURE 4.

Simplistic overview of therapy-related clonal hematopoiesis. Hematopoietic stem cells (HSCs) with mutations in DNA-damage response (DDR) genes, such as tumor suppressor protein 53 (TP53) or protein phosphatase Mn²⁺/Mg²+-dependent 1D (PPM1D), are present in very small numbers in normal bone marrow. Upon exposure to chemotherapy and/or radiation during treatment for cancer, clones with mutations in DDR genes have a survival advantage. Over time with each successive exposure, the mutant cells with the survival advantage become selected for within the bone marrow, thereby facilitating their clonal expansion. Created with BioRender.com, with permission.

4. CLONAL HEMATOPOIESIS AND ALL-CAUSE MORTALITY

Several studies have associated clonal hematopoiesis with an increase in all-cause mortality. The most prominent reports of this association, particularly for driver gene-mediated clonal hematopoiesis, originated from two large longitudinal studies that were concurrently published in the New England Journal of Medicine in 2014 (35, 37). As described above, these studies used whole exome sequencing of peripheral blood mononuclear cells from over 29,000 individuals that were unselected for cancer or hematological malignancies (35, 37). It was found that clonal hematopoiesis was associated with a 40–50% increase in all-cause mortality over the follow-up period (35, 37). This association could be correlated with age, as it was stronger in individuals beyond the age of 70 (35). Additionally, the study by Genovese et al. (37) employed a nonbiased exome sequencing analysis, allowing for the detection of clonal hematopoiesis with unknown driver mutations. They found that clonal hematopoiesis resulting from unknown mechanisms (e.g., mutations in unknown driver genes) accounted for ∼40% of all detectable clonal events. Furthermore, individuals in this category also exhibited increased all-cause mortality; however, the risk for mortality was slightly lower for individuals bearing known CHIP mutations (37). Both these studies reported a marked increase in the incidence of hematological cancer in individuals with clonal hematopoiesis (35, 37), which is to be expected, as mutations in driver genes represent early steps in the development of a hematologic malignancy. However, because hematologic malignancies are relatively infrequent in the general population, this observation could not fully explain the large increases in all-cause mortality observed in these two studies (35, 37). In particular, one of these studies found that only one individual died due to a hematological neoplasm, and a cause-specific analysis of 5,132 individuals further showed a lack of association with cancer death (35). Thus what was the underlying cause of all-cause mortality associated with clonal hematopoiesis? An unplanned secondary analysis revealed an association between clonal hematopoiesis and an increased incidence of coronary heart disease and ischemic stroke, suggesting that the increase in all-cause mortality might be attributed to an increased incidence of these conditions (35). Indeed, a cause-specific analysis provided further evidence for the association between clonal hematopoiesis and death due to cardiovascular disease (35). This surprising discovery has sparked increased interest in the relationship between somatic mutations in the hematopoietic system and age-associated diseases, and the findings of the resultant studies will be discussed in greater detail in the forthcoming sections.

Since these initial studies, there has been additional data published supporting an association between clonal hematopoiesis and all-cause mortality. The Icelandic study, also described sect. 3.2.2, that employed nonbiased whole genome sequencing of 11,262 individuals, reported an association between clonal hematopoiesis and all-cause mortality (30). In this study, it was found that the risk of mortality was similar for both clonal hematopoiesis associated with and without a putative candidate driver gene mutation. Furthermore, it was found that individuals with clonal hematopoiesis had a similar risk of all-cause mortality as that found with smoking (30). This study also reported significant associations between clonal hematopoiesis and smoking, smoking-related diseases, chronic pulmonary disease, and psychiatric disease, although cardiovascular outcomes were not examined (30).

Expanding this association to mCAs, Loh et al. (32) reported associations between large mCAs and all-cause mortality. In this study, 8,342 mosaic chromosomal alterations were detected within the blood of 151,202 participants in the UK Biobank using SNP-based computational techniques (32). It was observed that the presence of an mCA was associated with a doubling in the risk of all-cause mortality (32). This association could only partly be explained by cancer deaths indicating that some of these deaths could potentially be attributed to an increase in cardiovascular disease (32). Nevertheless, a subsequent study examining mCAs in the blood of 179,417 Biobank Japan participants found that the increase in all-cause mortality associated with mCAs could not be attributed to cardiovascular deaths (33). Using health outcome data from ∼70% of the cohort, the authors were only able to detect an increase in death due to leukemia (33). In terms of other specific causes of death, it has also been reported that mCAs are modestly associated with an increase in fatal infections, particularly sepsis, pneumonia, and neurological infections (34). It is perhaps tempting to speculate that it is more difficult to find associations between autosomal mCAs and specific causes of death as their effect on genomic loci is so varied between individuals.

It has also been documented that mLOY is associated with an increase in all-cause mortality. This association was first documented by a study analyzing the peripheral blood of 1,153 elderly men for chromosomal structural variants (25). The authors found that mLOY was associated with an increase in all-cause mortality, observing that men with mLOY had a 5.5-year shorter life span than controls (25). Moreover, it appeared that a dose-dependent relationship existed, with a greater level of mLOY being associated with a higher mortality rate (25). The authors also found that men with mLOY had an increase in death due to nonhematological malignancies. Given this finding, the authors postulated that mLOY may at least partially account for the higher death rate of cancer in men (25). The association between mLOY and all-cause mortality has since been corroborated by a study that used data from the UK Biobank. In this prospective study, it was found that mLOY was modestly associated with an increase in all-cause mortality (143). Further studies are required to better understand the relationship between mLOY and all-cause mortality, particularly those involving large cohorts of individuals.

5. CLONAL HEMATOPOIESIS AND AGE-RELATED DISEASE: EVIDENCE FROM THE CLINIC

5.1. Clonal Hematopoiesis and Cardiovascular Disease

5.1.1. Incidence.

The discovery that cardiovascular disease could substantially account for the increase in all-cause mortality associated with driver gene-mediated clonal hematopoiesis (35) has led to further investigations into the relationships between the two phenomena (42, 144–152). As noted above, the study by Jaiswal et al. (35) reported that individuals with CHIP had a significantly increased risk of coronary heart disease and ischemic stroke (hazard ratios of 2.0 and 2.6, respectively) after adjusting for age, sex, type 2 diabetes, systolic blood pressure, and body mass index (BMI). It was also found that individuals with large clones (VAF >10%) exhibit a higher risk of cardiovascular disease (35). Notably, the risk of coronary heart disease and ischemic stroke was higher in CHIP carriers than for individuals with traditional cardiovascular disease risk factors, including hypertension and high BMI (35). Jaiswal et al. (144) also conducted a follow-up investigation to directly test whether CHIP is associated with an increase in cardiovascular disease incidence. Whole exome sequence analysis was performed on blood samples obtained from 2 prospective and 2 retrospective case-control studies, consisting of 4,726 individuals with coronary heart disease and 3,529 controls. In the prospective studies, CHIP carriers had an increased risk of coronary heart disease (combined hazard ratio: 1.9). Similar to the previous studies, most mutations were detected in the DNMT3A, TET2, ASXL1, and JAK2 genes. To examine for gene mutation-specific differences in risk, a meta-analysis was performed by combining the data with the risks reported in their previous publication (35, 144). While the increased risk of coronary heart disease for individuals carrying mutations in DNMT3A (hazard ratio: 1.7), TET2 (hazard ratio: 1.9), or ASXL1 (hazard ratio: 2.0) was not statistically different, individuals who carried JAK2 mutations displayed a marked increase in disease risk compared to carriers of the other mutations (hazard ratio: 12.0) (144). However, this work should be viewed as preliminary as it should be noted that the sample size of JAK2 mutation carriers was very low. On the other hand, an analysis of individuals within the BioImage prospective study found that CHIP carriers were more likely to have coronary artery calcification, and this was dependent on clone size (144). Finally, analyses of the retrospective studies revealed an association between CHIP and early onset myocardial infarction (144).

In addition to the seminal studies described above, there have been several more recent reports of associations between clonal hematopoiesis and cardiovascular disease. First, a study by Yu et al. (152) used whole genome/exome sequencing data from 56,597 individuals to assess the association between CHIP and risk of heart failure. This study used data collated from the UK Biobank and four prospective population-based cohorts (TOPMed), which allowed the authors to achieve greater statistical power (152). It was found that CHIP was associated with a 25% increased risk of the first episode of hospitalized heart failure, independent of traditional cardiovascular disease risk factors (152). Similar to observations by Jaiswal et al. (144), the authors reported a dose-dependent relationship between CHIP and heart failure risk with larger clones being associated with greater risk. When examining specific CHIP driver genes, it was found that mutations in TET2, JAK2, and ASXL1 were associated with risk of heart failure. Individuals with DNMT3A mutations, which is the most commonly mutated gene, did not display an elevated risk for heart failure (152). Interestingly, only ASXL1 was associated with reduced left ventricular ejection fraction, although this finding warrants further investigation (152). Second, another study published concurrently reported an association between CHIP and increased incidence of coronary artery disease in postmenopausal women (151). Specifically, this study used a combined total of 19,606 postmenopausal from the UK Biobank with whole exome sequences and from the Women’s Health Initiative with whole genome sequences and was designed to examine the relationship between CHIP and premature menopause. While the authors found that CHIP was indeed associated with premature menopause in women, it was also noted that postmenopausal women with CHIP had an increased incidence of coronary artery disease (odd ratio: 1.36), and this association was stronger for women with larger clones (VAF >0.1), again indicating a dose-dependent relationship between CHIP and cardiovascular disease risk. Third, in a cohort of 200 patients undergoing hip replacement surgery, it was found that CHIP was significantly associated with cardiovascular disease, including hypertension, coronary heart disease, myocardial infarction, cardiac arrhythmia, and stroke (150). Fourth, in a study involving 50,122 individuals from 2 separate Biobanks, it was reported that CHIP was associated with incident peripheral artery disease (153). The authors also documented associations between CHIP and other atherosclerotic diseases, such as coronary artery disease, aortic aneurysm, and mesenteric ischemia, supporting the links between CHIP and atherosclerotic cardiovascular disease. A gene-specific analysis revealed that most of the commonly mutated CHIP genes were associated with peripheral artery disease, including TET2, ASXL1, PPM1D, TP53, and JAK2, but interestingly not DNMT3A. Finally, a study that used genome sequence data of blood DNA of 78,752 individuals from 8 prospective cohorts and biobanks found that CHIP was associated with an increased risk of stroke (154). When examining the major types of stroke, only the risk of hemorrhagic stroke was significantly increased in individuals with CHIP. Nevertheless, analysis of the subtypes within these major types revealed that CHIP was associated with an increased risk of all hemorrhagic types as well as small vessel ischemic stroke. In gene-specific association results, TET2 showed the strongest association with ischemic stroke, whereas DNMT3A was associated with an increased risk of hemorrhagic stroke. Collectively, these findings add weight to the initial reports that CHIP is associated with an increased risk of cardiovascular conditions.

The relationship between mCAs and risk of cardiovascular disease has received considerably less attention, and current data suggest that there may only be an association with mLOY. As mentioned previously, in a study examining mCAs in 179,417 participants of the BioBank Japan, it was reported that autosomal mCAs are associated with an increase in all-cause mortality; however, this could not be attributed to increased death due to cardiovascular conditions (33). A study by Haitjema et al. (155) examined the relationship between mLOY and atherosclerotic cardiovascular disease in 366 men undergoing carotid endarterectomy. It was found that mLOY was associated with major cardiovascular end points during follow-up (hazard ratio: 2.28). The authors also found that mLOY was associated with a larger atheroma size, although this association did not persist after correction for multiple testing. It has also been documented that mLOY is associated with self-reported prior myocardial infarction or stroke at baseline in the UK Biobank (128). More recently, mLOY has been associated with increased death of several cardiovascular conditions. With the use of the data from the UK Biobank, it was found that men with >40% mLOY in their blood cells had a 30% increased risk of dying from diseases of the circulatory system (143). A closer examination of diseases within this category found that this risk was particularly high for hypertensive heart disease (hazard ratio: 3.48), heart failure (hazard ratio: 1.76), congestive heart failure (hazard ratio: 2.42), and aortic aneurysms and dissection (hazard ratio: 2.76). Overall, these findings suggest a connection between mLOY and cardiovascular diseases, but possibly not mCAs, although further studies are required to validate these associations.

5.1.2. Prognosis: DNMT3A, TET2, and clone size.

The studies described above specifically examined the association between clonal hematopoiesis and cardiovascular risk or incidence; however, they provide limited information on the relationship between clonal hematopoiesis and disease prognosis or outcome. Thus to assess the potential prognostic significance of clonal hematopoiesis in patients with cardiovascular disease, Dorsheimer et al. (145) examined the long-term prognosis of CHIP carriers with chronic heart failure (CHF) owing to ischemic origin. The frequency of CHIP was assessed in bone marrow-derived mononuclear cells of 200 CHF patients by deep targeted sequencing of 56 genes that are recurrently mutated in CHIP or myeloid malignancies (145). They reported that 18.5% of patients exhibited a CHIP clone with a VAF of at least 2%. As expected from previous studies, most mutations were detected in DNMT3A and TET2. In addition, consistent with earlier reports, CHIP carriers were older compared to noncarriers; however, the CHIP carriers did not differ from noncarriers with respect to cardiac function or disease stage (145). However, patients harboring either DNMT3A or TET2 mutations had an overall worse clinical outcome for both death and death combined with rehospitalization for heart failure compared to noncarriers after adjusting for age (145). To assess whether this association was correlated with clone size, the VAF threshold of mutations was lowered to 0.5% (145). Thus an additional 66 and 53 patients with DNMT3A and TET2 mutations, respectively, were identified with clone sizes ranging from 0.5 to 2% VAF. When patient groups were examined based on VAF (no mutation in TET2 or DNMT3A, VAF <1%, VAF ≥1% and <2%, or VAF >2%), a dose-dependent relationship between clone size and clinical outcome was observed. Furthermore, this analysis indicated that clone sizes between 1 and 2% VAF, which is lower than the proposed CHIP threshold, were associated with a worse prognosis (145). While this study was small, it supports the hypothesis that clonal hematopoiesis is associated with progression and poorer prognosis in patients with CHF.

To meet the criteria for CHIP, individuals must have a VAF of at least 2% (38); however. this definition is largely based on the limit of mutation detection by sequencing methodology, rather than epidemiological or biological findings. Since the technology for variant calling has greatly improved allowing researchers to detect substantially smaller clones, it raises the question of whether smaller clones (i.e., VAF <2%) could have a clinical impact on cardiovascular disease. While preliminary findings from Dorsheimer et al. (145) suggested that clones between 1 and 2% VAF could affect the prognosis of CHF, this study was not designed to directly address the prognostic role of clone size. Thus, a study by Assmus et al. (42) aimed to define the threshold of VAF which clonal hematopoiesis could impact the prognosis of CHF. The authors focused their investigation on mutations in DNMT3A and TET2, the two most commonly mutated driver genes in both aged individuals and patients with cardiovascular disease. Using error-corrected deep-amplicon sequencing, which is able to reliably detect mutations at a VAF of <0.5%, the authors analyzed peripheral blood or bone marrow mononuclear cell samples from 419 chronic ischemic heart failure patients. A receiver operator curve (ROC) analysis revealed that the cut-off VAF to predict the 5-year survival of heart failure patients was 1.15% and 0.73% for mutations in DNMT3A and TET2, respectively (42). Given that these VAFs are significantly lower than those used to define CHIP, they challenge the quantitative definition of CHIP and its relationship to cardiovascular disease.

The association between CHIP and poorer prognosis of heart failure has been corroborated by other groups. A study by Pascual-Figal et al. (147) performed deep sequencing on the peripheral blood from 62 patients over 60 years of age with heart failure with reduced left ventricular ejection fraction owing to either ischemic or nonischemic causes. It was found that the prevalence of CHIP was 38.7%, which is higher than what has previously been observed for other cohorts of this age group (35, 37), further suggesting that CHIP is enriched in heart failure patients. Consistent with previous reports, it was found that patients with mutations in DNMT3A and TET2 mutations had accelerated heart failure progression in terms of all-cause death or all-cause death combined with heart failure-related hospitalization. This association was stronger for heart failure-related causes of death and individuals with DNMT3A mutations (147). A competing risk analysis, which entailed a composite outcome of heart failure-related death or heart failure-related hospitalization, confirmed that CHIP was specifically associated with worsening heart failure progression and this association was highly significant for individuals with DNMT3A and/or TET2 mutations, even after adjusting for age, sex, and other confounding variables (147). Importantly, this study suggests that CHIP also accelerates nonischemic heart failure, which may indicate that the connection between CHIP and adverse heart failure events is not driven solely by coronary artery disease (147). It should be noted that Pascual-Figal et al. were unable to observe an association between DNMT3A-/TET2-mediated clonal hematopoiesis and heart failure prognosis for VAFs between 1 and 2%. However, this study included substantially fewer individuals and thus may not have been sufficiently powered to detect an association with lower VAFs (147). Furthermore, both ischemic and nonischemic forms of heart failure were examined in this study, and it is currently unknown whether the new VAF thresholds described by Assmus et al. (42) also extend to nonischemic heart failure (147). Therefore, further work needs to be done, particularly using larger cohorts to further validate these cut-off VAFs.

A study by Wang et al. (156) reported that DNMT3A-/TET2-mediated clonal hematopoiesis is associated with a poor prognosis in patients presenting with ST-segment elevation myocardial infarction (STEMI). Specifically, in a cohort of 485 patients with STEMI, those with mutations in DNMT3A/TET2 were at a significantly greater risk of all-cause death, nonfatal myocardial infarction (MI) or stroke, or hospitalization due to heart failure compared to noncarriers over a 4-year follow-up period (156). Collectively, these studies suggest that mutations in the most common driver genes are associated with a poorer prognosis in heart failure owing to both ischemic and nonischemic origins. However, given the smaller number of individuals studied with nonischemic heart disease, future studies should focus on this subtype.

In addition to heart failure, a connection between mutations in DNMT3A and TET2 and prognosis in individuals treated for severe aortic valve stenosis by transcatheter aortic valve implantation (TAVI) was found (146). This study used deep amplicon sequencing to specifically detect DNMT3A and TET2 mutations within peripheral blood mononuclear cells. It was found that patients carrying DNMT3A or TET2 driver mutations experienced a significantly worse clinical outcome for death during the first 8 months following TAVI (146). Collectively, these findings support the hypothesis that CHIP may promote cardiovascular disease progression beyond heart failure.

5.1.3. Prognosis: additional driver genes and multiple mutations.

While the studies described above have focused primarily on individuals with DNMT3A and TET2 mutations, there is evidence to suggest that mutations in other candidate driver genes are also associated with a poorer disease prognosis in patients with heart failure even at low VAFs. For instance, a study by Cremer et al. (149) used error-corrected sequencing to examine the association between isolated and/or multiple CHIP mutations and mortality in a cohort of 419 ischemic heart failure patients. This study found a total of 223 mutations in 154 patients and 53 patients harbored more than 1 mutation. A ROC analysis revealed that beyond mutations in DNMT3A and TET2, an additional seven mutations were associated with mortality, namely PPM1D, PHD finger protein 6 (PHF6), structural maintenance of chromosomes 1A (SMC1A), enhancer of zeste 2 polycomb repressive complex 2 subunit (EZH2), CCAAT enhancer binding protein alpha (CEBPA), serine and arginine rich splicing factor 2 (SRSF2), and SET binding protein 1 (SETBP). Notably, these mutations were individually predictive of mortality at a lower VAF than the typical 2% cutoff for CHIP. Moreover, patients with two or more mutations had significantly higher mortality than those with one mutation or compared to non-CHIP carriers (149). This study suggests that mutations other than DNMT3A and TET2 have implications for the prognosis of heart failure and that multiple mutations may have a cumulative effect on prognosis. Nevertheless, given that relatively small numbers of patients with these seven additional mutations, findings from this study should be confirmed in other large cohorts of patients. A subsequent study from the same group also identified driver mutations other than DNMT3A and TET2 that correlate with heart failure prognosis (157). Similarly, the authors performed error-corrected sequencing on blood or bone marrow mononuclear cells from 399 patients with chronic ischemic heart failure. It was found that 87% of patients had a mutation with a VAF >0.5% and 10 additional driver genes were associated with an increased risk of death. Specifically, mutations in SETBP1, KMT2A, CBL, CEBPA, SRSF2, U2AF1, SMC1A, EZH2, G-protein subunit beta 1 (GNB1), and PHF6 resulted in a statistically significant increase in the hazard ratio in comparison to individuals without mutations. However, for three of these genes [SETBP1, U2 small nuclear RNA auxiliary factor 1 (U2AF1), and KMT2A], there was a very low number of individuals with mutations (n = 2–3) and thus this prevented further interpretation of these results. When the seven remaining genes were pooled together as a “risk gene set,” CHF patients carrying mutations in any one of these genes had a significantly lower survival when compared to individuals without mutations. Importantly, these effects were not confounded by cooccurring mutations in DNMT3A, TET2, or any other CHIP mutation with a VAF >2% (157). Overall, these studies reveal that numerous CHIP-related genes are associated with poorer prognosis in ischemic heart failure. It will be of interest to examine the functional consequences of these additional CHIP genes and how they may mechanistically contribute to heart failure.

5.2. Thrombosis

Clonal hematopoiesis has also been associated with thrombosis, the most common pathology underlying myocardial infarction and ischemic stroke. Specifically, a study by Wolach et al. (158) analyzed whole exome sequencing data from 11,527 individuals enrolled in a large case-control study that included healthy controls and patients with schizophrenia. It was found that CHIP was significantly associated with an increased incidence of major thrombotic events such as deep venous thrombosis or pulmonary thrombosis (2% non-CHIP vs. 5% CHIP carriers). This association was substantially stronger in carriers of the JAK2^V617F mutation, with 25% of carriers experiencing a thrombotic event. Further, it was documented that even individuals with JAK2^V617F clones as small as 2% VAF demonstrated an increased risk of thrombosis. The authors attributed this increase in thrombosis to an increased propensity of JAK2^V617F mutant neutrophils to form extracellular traps, which can promote thrombosis (158). It should be noted that individuals with JAK2^V617F-positive hematological disorders have a substantially increased risk of thrombosis and so this association is expected (159–161). The JAK2^V617F mutation has been linked to multiple pathogenic mechanisms including, abnormal function of erythrocytes and platelets as well as increased activation of integrins (162–164). Regardless, these findings may explain the substantially increased risk of cardiovascular disease observed in carriers of JAK2 mutations compared to carriers of other frequently occurring CHIP genes (144), although further studies will be warranted to more closely examine this relationship.

5.3. Chronic Kidney Disease

Individuals with chronic kidney disease (CKD) have an elevated risk of cardiovascular disease and often share similar risk factors, such as diabetes, smoking, hypertension, and dyslipidemia (165). Given the association between cardiovascular disease and clonal hematopoiesis, it has been of interest to also examine the relationship between CKD and clonal hematopoiesis. To do this, SNP array and whole exome sequencing data from 190,487 UK Biobank participants were analyzed (166). It was found that clonal hematopoiesis was associated with impaired kidney function, as indicated by a lower estimated glomerular filtration rate (eGFR) score and a higher urine albumin-to-creatinine ratio. In logistic models, clonal hematopoiesis was negatively associated with eGFR estimated from cystatin-C (eGFR-cys) score but not eGFR estimated from creatinine (eGFR-creat). A subtype-specific analysis revealed that myeloid-derived mutations were negatively associated with eGFR-cys, and this included both myeloid-specific mCAs and driver gene mutations. In terms of driver genes, mutations in TET2, JAK2, PPM1D, and CBL were negatively associated with eGFR-cys, but not mutations in other common drivers, such as DNMT3A or ASXL1, suggesting that mutations in a subset of drivers may be involved in the pathogenesis of CKD. Adverse outcomes, as defined as a composite end point of death, MI, or stroke, were 1.56-fold higher in individuals with myeloid-specific clonal hematopoiesis after excluding individuals with myeloid neoplasms and previous cardiovascular disease (166). Similarly, another study has also reported an association between clonal hematopoiesis and CKD (167). Specifically, in a cohort of 87 patients with CKD (defined as eGFR <60 mL/min/1.73 m²), it was found that CHIP was associated with lower baseline eGFR, and it was observed that CHIP carriers were more likely to exhibit kidney disease progression compared to noncarriers. Despite these findings, another study has reported that the association between CHIP and CKD may be less clear in the presence of type 2 diabetes (168). Individuals with type 2 diabetes are particularly susceptible to developing CKD, and when this occurs, it is often referred to as diabetic kidney disease. To explore the association between CHIP and diabetic kidney disease, deep-targeted amplicon sequence was performed on a nested case-control study cohort, consisting of 64 cases and 230 controls. It was found that CHIP was not associated with the incidence or progression of diabetic kidney disease, suggesting that other factors beyond clonal hematopoiesis contribute to the decline in kidney function in individuals with diabetes. Collectively, these studies suggest that clonal hematopoiesis may accelerate CKD progression in certain circumstances, although further studies are necessary using larger cohorts of CKD patients to more closely examine the relationship between clonal hematopoiesis and CKD progression.

5.4. Type 2 Diabetes and Obesity

There is some evidence to suggest that clonal hematopoiesis may be associated with type 2 diabetes, a chronic aged-related disease, and an independent risk factor for cardiovascular disease. This association was first reported by Bonnefond et al. (169), who analyzed blood DNA from a cohort of 7,659 individuals, of which 2,208 were type 2 diabetic. It was noted that there was a significant association between the presence of mCAs and type 2 diabetes. Remarkably, this association was only significant for nonobese individuals (i.e., BMI <30), suggesting that a higher BMI may be neutral or slightly protective (169). Moreover, it was demonstrated that carriers of mCAs with type 2 diabetes had a twofold higher prevalence of vascular complications compared to noncarriers, perhaps suggesting a worse clinical prognosis (169, 170). Findings from a more recent study also suggested that specific mCAs may be associated with type 2 diabetes (171). In this study, existing data from 482,396 UK Biobank participants were analyzed for associations with 1,290 International Classification of Diseases codes (171). There was no statistically significant association observed between autosomal mCAs overall and the incidence of type 2 diabetes; however, significant associations were observed for mCAs of chromosomes 10 and 17 (171). Interestingly, a significant association was reported for mLOY and lower incidence of type 2 diabetes and obesity, although this was lost after adjustment for BMI (171). Taken together, these data may suggest that autosomal mCAs are associated with diabetes, although further studies will be required to better understand this association and the exact types of mCAs that may be involved. Finally, there is also some evidence connecting CHIP with diabetes; Jaiswal et al. (35) found that CHIP was modestly associated with an increased risk of type 2 diabetes and that individuals with type 2 diabetes were more likely to have CHIP mutations than those without type 2 diabetes. Nevertheless, the intention of this study was not to examine the association between CHIP and diabetes, and further epidemiological studies are required to assess this putative association.

5.5. Infectious Diseases

Advancing age is a major risk factor for increased morbidity and mortality from a wide variety of infectious diseases. As clonal hematopoiesis is strongly associated with aging, it has been of interest to also examine the relationship between clonal hematopoiesis and infectious disease complications. It is known that hematopoietic mutations can affect immune cell function, and thus it is conceivable that specific mutations may affect the immune response to infectious agents. Thus, to determine whether a relationship exists between clonal hematopoiesis and infection, Zekavat et al. (34) analyzed data from 768,762 participants across 5 different Biobanks (UK Biobank, MGBB, FinnGen, Biobank Japan, and CUB COVID-19 cohort) for the presence of mCAs and their association with incident infection. In individuals without a history of cancer, it was found that mCAs were associated with an increased incidence of infection (hazard ratio: 1.12), and this association was stronger for expanded autosomal mCAs (hazard ratio: 1.25). More specifically, autosomal mCAs were significantly associated with sepsis, respiratory system infections, digestive system infections, and genitourinary infections. The association between mCAs and infection appeared to be stronger for individuals that subsequently developed cancer. For mCAs of the sex chromosomes, none of the infection types achieved statistical significance, although there were strong trends for respiratory infections and mLOX or mLOY (34). The relationship between clonal hematopoiesis with putative driver gene mutations and infection was not examined in this study. The authors also examined the relationship between mCAs and COVID-19 infections, given that numerous parallels have been observed in patients with severe COVID-19 and individuals with clonal hematopoiesis (34). For instance, both clonal hematopoiesis and severe COVID-19 are associated with advanced age, cardiovascular disease, and elevated levels of circulating IL-6 (148, 172, 173). Among 719 hospitalized cases within the UK Biobank, it was found that expanded mCAs were associated with COVID-19 hospitalization (34). Moreover, in 871 patients from the CUB cohort, it was found that the frequency of mCAs increased with COVID-19 severity, being detected in 5.8% of mild cases, 13.9% of moderate cases, and 16.9% of severe cases (34). It has been suggested that mCAs may affect the dosage of genes important for immunity and thus result in a compromised immune response to infection. Nevertheless, the underlying mechanisms for this are speculative, particularly given the wide range of mCAs observed in these cohorts. Similarly, in an independent study using only UK Biobank data, it has been documented that mCAs are associated with incident infection, particularly for sepsis and unspecified pneumonia (171). Collectively, findings from these studies suggest that mCAs may be a risk factor for certain types of infection, although further research is required to better understand this connection.

These findings have been more recently corroborated by a study by Bolton et al. (174) that sought to examine the relationship between clonal hematopoiesis and severe COVID-19. In this study, the blood from 525 COVID-19-positive patients across two separate cohorts was sequenced using a next-generation sequencing approach. To detect clonal hematopoiesis, the authors used a panel of putative driver genes, although this sequencing approach also allowed authors to detect nonputative driver mutations as well, such as silent clonal hematopoiesis (i.e., synonymous mutations). It was found that there was a significant association between clonal hematopoiesis associated with unknown driver gene mutations and severe COVID-19 (odds ratio: 2.01) as well as expansion of passenger mutations and severe COVID-19. A significant association was also observed for nonputative drivers of clonal hematopoiesis and risk of ventilation, again suggesting that clonal hematopoiesis due to unknown drivers is associated with COVID-19 severity (174). Surprisingly, no association was found between driver gene-mediated clonal hematopoiesis and severe COVID-19. These data support previous work that reported an association between COVID-19 severity and mCAs, which may fit into the category of “nonputative driver mutations.” However, it should be mentioned that one of the cohorts used in this study was a cancer patient cohort and thus may not truly reflect an association that may be otherwise found in the general population. Taken together, these studies may suggest that the presence of clonal hematopoiesis and resultant alterations in hematopoietic differentiation and not specific mutant alleles are predictive of COVID-19 disease severity.

Bolton et al. (174) also explored the relationship between clonal hematopoiesis and the risk of diverse infections in cancer patients. To do this, the billing codes were analyzed from 14,211 solid tumor patients who underwent blood sequencing as part of the MSK-IMPACT study. Clonal hematopoiesis was significantly associated with the onset of two infection types, specifically Clostridioides difficile infection and Streptococcus/Enterococcus infection. For C. difficile infections, significant associations were only found for nonputative driver mutations, in individuals with more than one clonal hematopoiesis mutation, and for mutations for a VAF >5%. On the other hand, for Streptococcus/Enterococcus infections, significant associations were found for putative driver gene mutations. These data suggest that clonal hematopoiesis may be a risk factor for developing infection, but the association is complex in that different drivers of clonal hematopoiesis may predispose to different types of infection. Nevertheless, additional studies will be required to directly explore this concept.

While clonal hematopoiesis may increase susceptibility to certain infections, it is possible that other infections increase the likelihood of developing clonal hematopoiesis. For instance, it has been hypothesized that chronic infections may facilitate the expansion of small clones by altering the bone marrow niche to favor their growth (175, 176). Human immunodeficiency viruses (HIV) are retroviruses that attack the body’s immune system, and once an individual is infected with this virus, the immune system is unable to eliminate it, leading to lifelong infection. As a consequence, there is sustained activation of the immune system, leading to chronic inflammation (177). Given the links between clonal hematopoiesis, chronic inflammation, and HIV, two independent studies recently examined the prevalence of clonal hematopoiesis in individuals living with HIV. First, a study by Dharan et al. (178) performed targeted amplicon sequencing of clonal hematopoiesis driver genes on the blood of 446 patients, which included 202 HIV-positive and 226 HIV-negative individuals. It was found that HIV-positive individuals had over twice the level of clonal hematopoiesis as those that were HIV negative. The majority of the mutations in HIV-positive individuals were generally similar to those commonly observed in aged individuals, such as DNMT3A, TET2, and ASXL1. Interestingly, ASXL1 mutations appeared to be overrepresented in HIV-positive patients compared to HIV-negative patients, perhaps suggesting that HIV infection may favor growth of ASXL1 mutant clones. Furthermore, it was observed that the VAF of mutant clones increased at a greater rate with age in HIV-positive individuals compared to those that were HIV negative. Collectively, these findings suggest that HIV infection may provide a selective advantage for growth of mutant clones (178). These findings have been corroborated by a study that also examined the prevalence of driver gene-mediated clonal hematopoiesis in individuals living with HIV (179). In this study, the prevalence of CHIP was evaluated within an HIV-positive cohort and the results were compared with individuals of a larger population-based cohort. After accounting for demographic differences between the cohorts, it was observed that HIV was associated with an increase in the prevalence of CHIP. The authors also noted that ASXL1 was the most commonly mutated gene in the HIV-positive cohort, which differed from that of the population-based cohort (179). This may again suggest that HIV infection alters the environment to favor growth of mutant cells, particularly those with ASXL1 mutations. However, a caveat of this study was that the HIV-positive cohort was a separate cohort from the controls and thus demographic balancing was required to perform meaningful comparisons. A more ideal scenario would be to compare HIV-positive with HIV-negative individuals in a population-based cohort unselected for disease or to perform a case-control study. Regardless, these studies should be replicated using larger cohorts of individuals to better understand the relationship between HIV and clonal hematopoiesis, particularly for ASXL1 mutations. Moreover, it would be of interest to examine other types of clonal hematopoiesis, such as mCAs, which may also be associated with HIV infection.

5.6. Autoimmune Conditions

While autoimmune conditions are not typically regarded as an age-related disease, there has been some recent interest in examining the relationship between clonal hematopoiesis and autoimmunity. This relationship is likely because, first, clonal hematopoiesis leads to alterations in immune cells, and thus it is conceivable that it may alter the pathogenesis of autoimmune conditions (172). Moreover, autoimmune diseases are characterized by chronic inflammation, and thus it has been proposed that this inflammatory environment may accelerate the expansion of certain mutant clones (180, 181). The association between CHIP and autoimmune disease was first examined in a small cohort of rheumatoid arthritis patients, whereby no connection was found (182). Nevertheless, other more recent studies have suggested that there is an association between CHIP and autoimmune disease. For instance, in a cohort of 200 patients undergoing hip replacement for osteoarthritis, it was observed that there was an association between CHIP and an increased incidence of autoimmune diseases (150). Moreover, a connection has been documented between CHIP and ulcerative colitis, which is often regarded as an autoimmune disorder (183). In this study, it was reported that CHIP was enriched in individuals with ulcerative colitis, particularly for those with DNMT3A and PPM1D clones (183). The authors speculated that the unique proinflammatory environment of individuals with ulcerative colitis may provide a selective advantage for DNMT3A and PPM1D clones to expand (183). These findings have since been corroborated by another study examining the prevalence of CHIP in inflammatory bowel disease, where it was also noted that CHIP was enriched in individuals with the disease (184). Another study reported that ∼30% of individuals with the autoimmune condition, antineutrophil cytoplasmic antibody-associated vasculitis have CHIP mutations, compared to ∼17% of healthy individuals (185). Collectively, these studies may suggest a connection between CHIP mutations and autoimmunity, although caution should be taken when interpreting these studies, particularly as data have been compared with historical controls from previously published studies rather than performing matched analysis on a new group of controls. Further, as different sequencing methodologies have been used between studies, this can influence the proportion of individuals that are identified as CHIP carriers. Thus it will be paramount for future studies examining the relationship between CHIP and autoimmune disease to use appropriate control cohorts to better understand this relationship. Moreover, it would be of interest to know whether CHIP correlates with a worse prognosis in individuals with these conditions and whether the autoimmune disease promotes clone expansion or both.

In addition to CHIP, mLOY has been associated with autoimmune disease. First, in a small cohort of patients with autoimmune thyroiditis, it was reported that the degree mLOY was significantly higher compared with healthy controls (186). A study from the same group later reported similar findings in a small cohort of patients with primary biliary cirrhosis (187). Given that the prevalence of these two autoimmune conditions is usually higher in females, it could be speculated that the presence of a Y chromosome is protective for these conditions. Nevertheless, more research is required to examine this association further, particularly as both these studies used relatively small disease cohorts.

5.7. Alzheimer’s Disease

Aging is a significant risk factor for the development of Alzheimer’s disease, the most common form of adult dementia. Many individuals with Alzheimer’s disease exhibit immune dysregulation, and it has been shown that the immune system is intricately involved in the pathogenesis of this disease (188, 189). Therefore, it is conceivable that clonal hematopoiesis could affect Alzheimer’s disease progression. The association between Alzheimer’s disease and clonal hematopoiesis has been examined for both driver gene- and mLOY-mediated clonal hematopoiesis; however, intriguingly, these two forms of clonal hematopoiesis appear to associate differently with the disease. To examine if an association exists between CHIP and Alzheimer’s disease, data from two cohorts within the TopMed study were analyzed (190). After exclusion of coronary heart disease and stroke, which themselves are risk factors for Alzheimer’s disease, it was found that the presence of CHIP was associated with a lower hazard ratio for incident Alzheimer’s disease. This association was subsequently validated in an independent case-control cohort, namely the Alzheimer’s Disease Sequencing Project. In this cohort, there also appeared to be a dose-dependent relationship between clone size and protection from disease. The authors also used brain autopsies from some deceased individuals within this cohort to determine if CHIP was associated with Alzheimer’s disease-like neuropathological changes. It was found that CHIP was associated with lower scores for neuritic plaques and neurofibrillary tangles, again suggestive of disease protection. Also, in these samples, it was noted that CHIP-associated mutations were present in microglia-like cells and the number of these cells with mutations appeared to be relatively abundant, suggestive of clone expansion within this population. The authors postulated that CHIP mutations may provide cells with a survival advantage following infiltration into the brain or may allow for bone-marrow cells to differentiate into microglia more readily. Nevertheless, this theory contradicts current dogma whereby bone marrow-derived hematopoietic cells are thought to remain distinct from microglia even under conditions of complete replacement (191). The mechanism by which CHIP protects from Alzheimer’s disease development remains to be determined; however, it has been speculated that the mutation may provide the microglia-like cells with an enhanced ability to clear accumulated beta-amyloid and/or tau. If validated by future work, interrogating the underlying mechanism behind this association will be of interest and could offer novel insights into therapies for individuals with Alzheimer’s disease.

These above-described findings contrast with mLOY, which has been associated with the development of Alzheimer’s disease in men (28). Specifically, analysis of three study cohorts found that men with Alzheimer’s disease had a significantly higher degree of mLOY in their blood compared to controls (28). Similarly, analysis of two prospective study cohorts found that individuals with mLOY had a higher risk for the development of Alzheimer’s disease (28). The divergent findings between mLOY and CHIP, possibly reflect the differing downstream consequences these types of mutations have on immune cells. However, whether mLOY is causally involved in the pathogenesis of Alzheimer’s disease remains to be determined.

5.8. Solid Organ Tumors

Aging is a major risk factor for the development of a diverse range of solid organ tumors (192); however, despite this association, considerably little attention has been paid to understanding the relationship between CHIP and nonhematological cancers. It has been shown that the prevalence of CHIP is higher in individuals with nonhematological cancers; however, it is possible that this may be a reflection of oncologic therapy-enriching mutant clones (i.e., therapy-related clonal hematopoiesis) rather than a direct association per se (116, 140). Nevertheless, it has been reported CHIP is associated with adverse outcomes in patients with nonhematological cancers, an effect independent of therapy-related myeloid neoplasms (116, 140). In addition, CHIP mutations have been found within solid-organ tumor biopsies (193–195), although whether these mutations represent contamination from the blood or true enrichment has been debated (193–195). It is well established that leukocytes are found within solid organ tumors and contribute to tumor growth (196), thus it is possible that the tumor microenvironment selects for or promotes clonal growth of leukocytes with driver mutations. In support of this notion, one study observed that leukocytes with driver mutations were enriched in untreated breast tumors, and in some instances, these mutations could not be detected in the blood (194). In the setting of lung cancer, it has been reported that risk factors, such as smoking and genetics, could largely explain the association between driver gene-mediated clonal hematopoiesis and lung cancer (197). From these studies, it appears that the relationship between CHIP and solid organ tumors is unclear. It is also conceivable that CHIP represents a biomarker for widespread genomic instability, whereby many cells throughout the body carry a substantial mutation burden and thus are more likely to become malignant. It is expected that large-scale studies will be required to improve our understanding of this relationship and to determine whether preexisting CHIP places individuals at a higher risk of developing solid organ cancers. Furthermore, experimental studies addressing causality will be of great importance in better understanding the relationship between clonal hematopoiesis and solid organ tumors.

Studies have found clearer associations between mLOY and solid organ tumors. First, in a cohort of 1,153 elderly men, it was reported that mLOY was associated with an increased risk of nonhematological cancer diagnosis and death (25). In support of this, another study found that mLOY was modestly associated with an increased risk of solid organ tumors (198). Of note, for some cancers, this risk appeared to increase with increasing levels of mLOY, perhaps indicative of a dose-dependent effect. It was also observed that the risk of lung cancer decreased after adjusting for smoking, which may suggest that smoking may be a confounder for cancer risk in men with mLOY, particularly given the association between mLOY and smoking (27, 29, 127, 128). In a small case-control study, it was found that mLOY was enriched in individuals with colorectal and prostate cancer (199). Interestingly, it was observed that mLOY was a more significant predictor of cancer presence than age in this cohort. Furthermore, it has been reported that individuals with autosomal mCAs have a modestly increased risk of solid organ tumors (120). Of note, this risk was highest for cancers associated with tobacco smoke, such as lung and kidney cancers. Given the connection between mCAs (including mLOY) and smoking, further studies will be required to better understand the directionality of the relationship between mCAs, cancer risk, and smoking.

5.9. Associations with Other Age-Related Conditions

There are several other age-related conditions that have been associated with clonal hematopoiesis. As indicated in sect. 3, clonal hematopoiesis was recently associated with natural premature menopause in two large cohorts of women (151). When examining common CHIP genes, a gene-specific analysis revealed that this association was only significant for mutations in DNMT3A. It also found that mCAs were enriched in women with premature menopause, possibly suggesting that premature menopause is associated with general genomic instability. It is worth noting that premature menopause has been genetically linked with SNPs in DNA damage response genes (200), and thus it is conceivable that individuals with these SNPs have greater amounts of genomic instability. CHIP has been also associated with incident osteoporosis in a large cohort of UK Biobank participants (201). Interestingly, it was found that mutations in DNMT3A and ASXL1 were most strongly associated with osteoporosis but not mutations in TET2 (201). Further, several independent studies have associated CHIP with the development of chronic obstructive pulmonary disease (COPD) (30, 202, 203). In one of these studies, it was found that having CHIP posed a similar risk of 20–40 pack years of exposure to cigarette smoke for the development of severe to very severe COPD (203). This would suggest that the presence of CHIP poses a similar risk as does a well-established risk factor for the development of severe COPD. Additionally, it has been documented that CHIP is enriched in individuals with hemophagocytic lymphohistiocytosis, a rare life-threatening disease associated with aberrant immune activation (204). Moreover, CHIP has been associated with an increased incidence of gout in two large Biobank cohorts, and this association appears to be most prominent for individuals with TET2 mutations (205). With regard to mLOY, other disease associations have been reported for macular degeneration (206, 207). Overall, from the studies described above, it is apparent that clonal hematopoiesis has a wide impact on human health and disease. This likely reflects the central role that the immune system plays in the pathogenesis of various diseases, particularly those that correlate with age. It is thus expected that over the coming years the list of associations with other diseases will grow as more diverse cohorts of individuals are analyzed. The reported disease associations for clonal hematopoiesis mediated by driver genes and mCAs are summarized in FIGURE 5 and FIGURE 6, respectively.

FIGURE 5.

Diagram summarizing documented associations between driver gene-mediated clonal hematopoiesis and human disease. Created with BioRender.com, with permission. COPD, chronic obstructive pulmonary disease.

FIGURE 6.

Diagram summarizing documented associations between mosaic chromosomal alterations (mCAs) of autosomes, including mosaic loss of the Y chromosome (mLOY), and human disease. Created with BioRender.com, with permission.

5.10. Clonal Hematopoiesis and Accelerated Biological Aging

In addition to the aforementioned disease associations, CHIP has been connected with accelerated aging. Recently, two studies have explored the relationship between CHIP and epigenetic aging by examining several methylation clocks, which have been shown to accurately correlate with chronological age (208, 209). The first of these studies used whole genome sequencing and methylation data from 1,136 elderly individuals in the Lothian Birth Cohort (209). The authors used a panel of six of the most prevalent driver gene mutations and found that individuals with hematopoietic mutations in these genes had accelerated epigenetic aging as measured by the Horvath clock, a measure of intrinsic age acceleration. The authors also found associations between clonal hematopoiesis and accelerated epigenetic aging using other methylation clocks. The second of these studies used whole genome sequencing data obtained from 5,522 individuals across several large cohorts, and similarly, it was found that CHIP was associated with epigenetic aging across all methylation clocks, particularly those correlating with intrinsic age acceleration (i.e., independent of the cell type) (208). When examining gene-specific associations, it was noted that mutations in DNMT3A and TET2 had similar contributions to age acceleration, although TET2 had significantly greater age acceleration for two of the clocks. Associations were also reported for other common driver gene mutations such as JAK2 and ASXL1, although surprisingly, there was no association found for individuals with mutations in DNA-damage response genes, such as TP53, PPM1D, and BRCA1-/BRCA2-containing complex subunit 3 (BRCC3). The authors also assessed whether the combination of CHIP and age acceleration could stratify CHIP carriers into high-risk and low-risk groups in terms of all-cause mortality and coronary heart disease. Indeed, individuals with CHIP and age acceleration, as defined by two methylation clocks, were at the highest risk of all-cause mortality and coronary artery disease. Most strikingly, the increased risk of CHIP on all-cause mortality and coronary artery disease was attenuated in individuals without age acceleration. This may suggest that the presence of age acceleration could be used to determine if CHIP carriers are at increased risk of adverse outcomes. In support of CHIP accelerating biological aging, another study reported an inverse association between CHIP and leukocyte telomere length (210). Telomere length is often used as a marker of cell biological age, with telomeres shortening with each successive cell division. It was noted that telomere length was inversely correlated with the VAF of the driver mutation. Interestingly, this inverse association was only observed for driver mutations in TET2, ASXL1, PPM1D, JAK2, and TP53 but not DNMT3A. This may reflect the effect of the specific mutation on HSC biology, where it is thought that DNMT3A mutations do not overtly provide HSCs with a proliferative advantage (180). Furthermore, it was also posited that CHIP-associated coronary artery disease risk may be at least partly explained by telomere attrition. Collectively, these findings suggest that CHIP mutations may drive HSC senescence contributing to biological aging. In addition to these studies, it has been suggested that individuals with conditions that cause age acceleration may have an increased risk of developing clonal hematopoiesis. For instance, individuals with Down syndrome exhibit features of accelerated aging, and typically succumb to age-related disease earlier than euploid individuals of the same chronological age. It has been reported that children and young adults with Down syndrome have a significantly higher risk of developing clonal hematopoiesis, particularly for mutations in TET2 (211). Although it was noted that Down syndrome individuals with clonal hematopoiesis had immune dysregulation, whether this was a direct cause of accelerated clonal expansion or occurred downstream of mutations in driver genes remains speculative. The authors of this study proposed that the Down syndrome environment may accelerate tissue changes that provide a more conducive environment for clones to expand. Finally, another study examined the prevalence of clonal hematopoiesis in 47 children with Hutchinson-Gilford progeria syndrome, an extremely rare genetic disease characterized by features of accelerated aging (212). In contrast to what was observed in Down syndrome, it was found that CHIP was not common in the population, with only one individual harboring a mutation in TET2. These findings highlight the importance of time for the expansion of clones harboring driver mutations, with many clones taking decades to expand. Collectively, many of these studies suggest that clonal hematopoiesis is associated with accelerated aging; however, the directionality of this relationship remains speculative. In other words, it is unknown whether clonal hematopoiesis directly contributes to the aging process or whether accelerated aging provides a more fertile environment for mutant clones to propagate. It is likely that this relationship is somewhat bidirectional; nevertheless, it is a research question that requires attention.

6. CLONAL HEMATOPOIESIS AND HEMATOPOIETIC STEM CELL TRANSPLANTS

Hematopoietic stem cell (HSC) transplants are sometimes used as a treatment for hematological malignancies, particularly when other treatment modalities have failed. These transplants involve a small portion of donor cells repopulating a recipient’s entire hematopoietic system where the donor cells may be autologous (i.e., derived from the patient’s own cells) or allogeneic (i.e., derived from a matched donor) in nature. Indeed, experimental studies have shown that certain clonal hematopoiesis driver mutations provide cells with a competitive advantage against transplantation stress, which could conceivably impact the extent to which clones evolve following transplantation (213–215). Also, patients usually have already received radiation/chemotherapy, and thus it is of concern that these treatments may have enriched preexisting mutant clones (i.e., therapy-related clonal hematopoiesis; see sect. 3.3), which will then either be transplanted back into the patient, in the case for autologous stem cell transplants, or may survive the preconditioning regime, which could impact on allogeneic transplants (116, 133, 137, 140) Therefore, it has been of interest to understand how clonal hematopoiesis may affect patient outcomes following HSC transplantation. The following section will discuss some of the work examining this; however, for a more detailed review on this topic, please see Ref. 216.

Clonal hematopoiesis in the setting of autologous stem cell transplantation (Auto-SCT) has been examined by several independent groups. It has been reported that CHIP is more prevalent in individuals before Auto-SCT, consistent with the notion that prior radiation/chemotherapy may be enriching mutant clones (133). In terms of expansion following transplantation, studies have reported different results. In one study, it was reported that individuals with preexisting CHIP were at increased risk of clone expansion following transplantation (217). However, another study found that the majority of clones did not expand following Auto-SCT (139). However, expansion risk may be driver gene dependent as it has been observed that DNMT3A mutant clones are likely to expand whereas PPM1D mutant clones are more likely to decrease following transplantation (139). Larger studies directed at examining the clonal dynamics of common driver mutations are thus warranted to understand this relationship more closely. In terms of other outcomes, it has been observed that Auto-SCT patients with preexisting CHIP require more days to collect an adequate number of stem cells, are likely to fail peripheral mobilization, and require a bone marrow harvest (133). Furthermore, it has been documented that patients with CHIP are more likely to exhibit poorer outcomes after Auto-SCT, such as an increased risk of therapy-related myeloid neoplasms and increased all-cause mortality (133).

For allogeneic stem cell transplantation (Allo-SCT), clonal hematopoiesis may be acquired in three ways. First, driver mutations may be present in the donor cells and these mutant cells engraft in the recipient following transplantation. Second, cells with driver mutations may already be present in the recipient and are able to survive the myeloablative preconditioning therapy that occurs before transplantation. Third, clonal hematopoiesis may arise de novo in the grafted cells after Allo-SCT (216). While little is known about this latter acquisition method, studies suggest that the first two of these acquisition methods have different effects on patient outcomes, and these will be discussed in the paragraphs below. First, for donor-derived clonal hematopoiesis, it has been observed that the incidence and mutational spectrum of driver mutations reflect that occurring within the general population (218, 219). To avoid donor-derived clonal hematopoiesis, it has been suggested that younger donors could be used. However, as HSC transplants often use siblings as donors, this is not always possible, particularly when both the sibling-paired donor and recipient are older in age. In addition, it has been shown that even young adult donors have small pathogenic clones that can engraft in the recipient (220). The effect of donor-derived CHIP on clinical outcomes following Allo-SCT has been recently examined by several independent groups (218, 219, 221, 222). Due to the differences in findings between these studies, we have summarized the main findings in TABLE 1. It is noteworthy that three out of four studies reported that donor-derived clonal hematopoiesis increased the rate of chronic graft versus host disease (GvHD), and this appeared to be particularly associated with mutations in DNMT3A (218, 221, 222). The reasons for this are not clear; however, it has been suggested that underlying inflammation associated with clonal hematopoiesis driver mutations may play a role in graft rejection. In support of this, one study found that Allo-SCT recipients with DNMT3A mutations had higher levels of IL-12, a cytokine that has been implicated in GvHD development (222). Interestingly, the largest of these studies found that patient outcomes in individuals with DNMT3A mutations were largely determined by the type of immunosuppressant therapy that they were prescribed (222). Specifically, it was found that Allo-SCT patients with DNMT3A mutations and taking calcineurin inhibitors had improved survival and a reduced rate of relapse, despite increased rates of chronic GvHD. If confirmed by additional studies, this could be important for determining the type of immunosuppressant individuals are prescribed, particularly in individuals whereby the matched donor has DNMT3A-mediated clonal hematopoiesis.

Table 1.

Summary of donor-derived clonal hematopoiesis and allogenic stem cell transplants

	Donor-Derived Clonal Hematopoiesis and Allogenic Stem Cell Transplants
Study	Oran et al. 2022 (219)	Newell et al. 2021 (221)	Frick et al. 2019 (218)	Gibson et al. 2022 (222)
Donor information	363 sibling-matched donors, aged >55 years	Related and unrelated donors, age not specified	500 related donors, aged >55 years	1,727 related and unrelated donors, aged >40 years
Recipient information	303 AML/MDS patients	209 patients with hematological malignancy	500 recipients	1,727 patients, majority with hematological malignancies
Clonal hematopoiesis prevalence in donors	18%	5.2%	16%	22.5%
Method of mutation detection	Targeted sequencing; VAF cutoff of 2%	Targeted next-generation sequencing; VAF cutoff of 2%	Targeted deep sequencing	Targeted-error corrected sequencing; analysis found >1% VAF to be clinically meaningful
Clinical outcomes
Engraftment	No effect	No effect	Superior engraftment	No effect
Relapse/disease  progression	No effect	No effect	Decreased incidence of relapse/disease progression	Reduced in individuals with DNMT3A mutations that received calcineurin based prophylaxis No effect for other driver genes
GvHD	Increase in acute GvHD	Increase in chronic GvHD	Increase in chronic GvHD, particularly for DNMT3A-mediated clonal hematopoiesis No effect on acute GvHD	Increased risk of chronic GvHD in individuals with DNMT3A-mediated clonal hematopoiesis who received calcineurin-based prophylaxis No effect on acute GvHD No effect for other driver genes
Donor cell  leukemia	Not reported	Not reported	Slightly increased rate	Increased rate
Survival	Not reported	No effect	No effect	Improved in individuals with individuals with DNMT3A mutations receiving calcineurin-based prophylaxis

AML, acute myeloid leukemia; DNMT3A, DNA methyltransferase 3A; GvHD, graft versus host disease; MDS, myelodysplastic syndrome; VAF, variant allele frequency.

Several studies have also examined the effect of preexisting driver mutations on outcomes after Allo-SCT. First, in a cohort of 113 patients that underwent Allo-SCT for acute myeloid leukemia (AML), it was found that ∼42% of them were carrying a driver gene mutation when in remission (i.e., after Allo-SCT) (223). In ∼28% of these patients, the mutant clone existed before Allo-SCT and most of the persisting mutations were in DNMT3A/TET2. Interestingly, individuals who harbored mutations in TET2 and ASXL1 were found to have superior survival in the years after Allo-SCT compared to individuals without mutations in these driver genes. Nevertheless, there appeared to be no difference in relapse or survival rates between patients with preexisting mutations and patients with newly acquired mutations within this small cohort. Given that these persisting mutations did not appear to affect outcomes, the authors concluded that these mutations were representative of CHIP rather than residual disease. In contrast, another study analyzed the effect of preexisting disease-associated mutations on outcomes after Allo-SCT in a cohort of patients with myelodysplastic syndrome (224). It was found that patients with disease-associated mutations that persisted after Allo-SCT experienced higher rates of disease progression than patients without mutations. A similar study assessed the clonal dynamics of mutations in AML patients from diagnosis, pre-Allo-SCT, and post-Allo-SCT and found that many patients still had detectable preexisting mutations (225). It was also noted that individuals with mutations after Allo-SCT had a higher rate of relapse and this appeared to correlate with a higher VAF. A higher allelic burden was also associated with inferior survival. This would suggest that a higher allelic burden of preexisting mutations after Allo-SCT is associated with poorer outcomes. Overall, these studies indicate that driver mutations can persist in Allo-SCT recipients; however, whether they represent CHIP or residual disease leading to poor outcomes remains to be fully determined.

7. EXPERIMENTAL STUDIES

The inherently descriptive nature of epidemiological studies makes it difficult to determine whether the relationship between clonal hematopoiesis and age-related disease is causal, as the association between these two phenomena could be secondary to confounding factors or simply reflect a shared consequence of the aging process. Additionally, as indicated above, in clinical studies it is difficult to assess directionality and understand factors that may drive clonal expansion of mutant cells. Understanding this is important, as it has been suggested that chronic inflammation and other stresses associated with cardiovascular disease may promote somatic mutagenesis and clonal expansion of mutant cells, rather than clonal hematopoiesis promoting cardiovascular disease (176, 226–228). Moreover, studies indicate that there are several genetic variants associated with clonal hematopoiesis, which could conceivably also increase risk of disease (26, 27, 30, 115, 229). Hence, animal models have recently proved to be powerful tools for better understanding the relationship between clonal hematopoiesis and age-related disease (101–103, 105, 141, 144, 230, 231). They have also played important roles in understanding the factors that promote the expansion of mutant clones (105, 141, 175). The following section of this review aims to describe the experimental progress in delineating clonal hematopoiesis and its association with age-related disease. This work is briefly summarized in FIGURE 7. Moreover, it will give a brief description of some of the animal models that have been used to study clonal hematopoiesis and their advantages and disadvantages. Finally, it will detail the experimental studies that have shed light on the potential factors that drive mutant clone expansion. It should be noted that the majority of experimental studies to date have examined the effect of specific driver gene-mediated clonal hematopoiesis on disease states and there have been substantially fewer studies examining the effects of mCAs, including mLOY on disease.

FIGURE 7.

Summary of experimental work causally linking clonal hematopoiesis with disease. Each disease is listed alongside the specific mutation that was investigated. Created with BioRender.com, with permission. BM, bone marrow; BMT, bone marrow transplant; Tet2, ten eleven translocation 2; Dnmt3a, DNA methyltransferase 3a; mLOY, mosaic loss of Y chromosome; Jak2, janus kinase 2; Ppm1d, protein phosphatase, Mg²⁺/Mn²⁺ dependent 1d; Trp53, transformation related protein 53.

7.1. Animal Models of Clonal Hematopoiesis

The development of experimental systems that closely model clonal hematopoiesis in humans is imperative for understanding the relationship between clonal hematopoiesis and age-related disease. Our group and others have played a crucial role in developing and refining these systems to study clonal hematopoiesis in disease settings (230, 232–234). The majority of these systems have used mice; however, there has been some recent interest in extending these studies to other species as well (233, 235). The paragraphs below briefly detail some of the experimental systems in mice and their application in disease models thus far; however, for a more extensive review of animal models of clonal hematopoiesis or detailed protocols, please see Refs. 102, 232, 234.

7.1.1. Bone marrow transplant approaches.

Mutations associated with clonal hematopoiesis are initially carried by a small number of HSPCs, which gradually expand over time (172). The simplest way to model this scenario is to use a competitive bone marrow transfer approach whereby mice are transplanted with a small proportion of mutant bone marrow cells mixed with wild-type cells (101, 236, 237). For these types of experiments, the use of congenic mouse strains or fluorescent reporter mice is useful in helping distinguish donor cells from competitor wild-type and/or recipient cells. This approach is ideal when studying mutations that confer a competitive advantage so that the expansion of the mutant cells can be tracked over time (101). Notably, the first study to report a causal connection between clonal hematopoiesis and cardiovascular disease used a competitive bone marrow approach to model Tet2-mediated clonal hematopoiesis in mice (101). For these studies, lethally irradiated, atherosclerosis-prone Ldlr-deficient mice were transplanted with a mixture of cells containing 10% Tet2-deficient or Tet2-sufficient bone marrow cells with 90% competitor wild-type bone marrow cells. To discriminate between Tet2-deficient/-sufficient cells and competitor cells, congenic mouse strains were used whereby Tet2-deficient/-sufficient cells originated from mice with the CD45.2 allele and competitor cells were from mice with the CD45.1 allele. Antibodies directed against CD45 variants were then used to distinguish CD45.2⁺ cells (Tet2 deficient/sufficient) from CD45.1⁺ cells via flow cytometry. Using this strategy, the authors found that Tet2-deficient cells expanded progressively in the bone marrow, spleen, and blood and exhibited a slight myeloid bias with preferential expansion into the Ly6C-high monocyte population (101). Importantly, there was no effect of HSPC Tet2 deficiency on the total numbers of white blood cells, which is analogous to what is observed in cancer-free individuals that exhibit clonal hematopoiesis associated with a TET2 mutation. This is an example of where mutant cells have a clear competitive advantage over wild-type cells in homeostatic conditions; however, this is not the case for all driver mutations. In circumstances where driver mutations do not show a competitive advantage over wild-type cells under homeostatic conditions or expansion is modest, mutant cells can be mixed with wild-type cells at a higher proportion to yield a stronger phenotype (141, 153, 238). Additionally, BMT approaches whereby all of the cells transplanted are mutant cells have also been used to model clonal hematopoiesis (144, 201, 239). While these approaches are of value, particularly when the mutant cells do not expand or show mild phenotypes, they do not fully recapitulate the human scenario whereby only a small fraction of leukocytes carry the mutation. On this point, sometimes a full transplant of mutant cells can lead to hematological issues or features of malignancy, which are not observed in clonal hematopoiesis associated with normal human aging (144, 239). Furthermore, it is thought that, in some instances, mutations can have non-cell-autonomous effects on wild-type HSPCs or immune cells, which may consequently impact disease pathology. Indeed, non-cell-autonomous effects have been reported for clonal hematopoiesis involving mutations in TET2, DNMT3A, PPM1D, and TP53 (101, 105, 201, 230, 240, 241). Therefore, when designing experiments to study clonal hematopoiesis in disease settings, these factors should all be taken into consideration.

A major drawback of using the described BMT approaches to study clonal hematopoiesis is the use of preconditioning regimes that completely ablate the host’s bone marrow cells. Preconditioning methods typically involve total body irradiation or myeloablative chemotherapy and are often necessary to achieve a high level of donor cell chimerism and can serve to prevent rejection by the host’s immune system (232). Nevertheless, these regimes can produce several undesirable systemic effects, which may influence the outcome of the study. First, irradiation has been shown to cause inflammation, fibrosis, and damage to multiple organs, including the skin, liver, kidneys, lungs, bone marrow, heart, brain, and intestines (242–248). Additionally, it has been observed that it alters the pathological processes that contribute to disease in experimental settings, such as atherosclerosis (249–251) and rupture of aortic aneurysms (252). Furthermore, preconditioning methods are known to deplete resident macrophage populations from organs such as the heart, brain, and lungs (141, 232, 253, 254). These resident populations colonize the tissues before birth from embryonic precursors and are maintained predominantly via self-renewal (255–257). Although depending on the tissue type, there may be some degree of replacement by bone marrow-derived macrophages (258). In the case of disease, these tissue-resident macrophages are usually early responders to injury and play numerous important roles in disease pathophysiology, from injury to repair (256). Thus, the depletion of resident macrophages, such as after preconditioning, can result in replacement by donor-derived bone marrow cells, which have a different functional and transcriptional profile to tissue-resident macrophages (191). Moreover, in the setting of clonal hematopoiesis, this would likely lead to the precocious colonization of the tissue by mutant cells (230). This could conceivably have a substantial impact on the outcome of a disease, particularly given the central role of tissue-resident macrophages in disease sequelae (256). To overcome these issues, shields can be used to protect the organ under study from radiation exposure (232). To do this, a lead shield is placed over the area which requires protection during the irradiation procedure. This method protects the organ under study and prevents depletion of resident immune cells and replacement with those derived from the bone marrow. Depending on the area shielded, ∼50–70% donor cell chimerism can be achieved using this approach (232). Thus, when using a competitive BMT strategy with shielding, it may be ideal to use a mixture of cells with a higher proportion of mutant cells.

Another pitfall of preconditioning is the detrimental effect it has on the bone marrow niche (259–261). While hematopoietic output can be somewhat restored following irradiation injury, it is unclear to what extent the bone marrow microenvironment recovers or whether the damage is irreversible (230, 261, 262). Furthermore, in the case of competitive transplantation, where mixed HSPCs are transplanted to an empty bone marrow niche, it can create an environment that allows for a proliferation race between wild-type and mutant clones rather than competition for a niche that is predominantly occupied by wild-type HSPCs (232). As a result, this alters the dynamics of mutant cell clonal expansion, which may result in misleading conclusions about clone fitness. To circumvent these issues, an adoptive transplant approach can be used to model clonal hematopoiesis mice (230, 231). To do this, mutant cells (or wild type for controls) can be transplanted directly into naïve mice without the need for preconditioning. While the level of donor cell chimerism is low using this method, it is still particularly useful for studying driver mutations that confer a competitive advantage under normal steady-state conditions. This is because the mutation will expand over time, eventually leading to a greater level of donor cell chimerism within the recipient. Our laboratory recently used this approach to study the effect of Tet2-mediated clonal hematopoiesis on heart failure and metabolic syndrome due to aging (230, 231). For these studies, 5 × 10⁶ bone marrow cells from Tet2-deficient or wild-type littermates were transplanted into unconditioned recipient mice over three consecutive days (a total of 1.5 × 10⁷ cells). Notably, congenic strains were used whereby donor mice expressed the CD45.2 allele and recipient mice expressed the CD45.1 allele. At 4 weeks following transplant, the level of CD45.2 chimerism in the peripheral blood was low for both mice transplanted with Tet2-deficient cells and those transplanted with wild-type cells. However, Tet2-deficient cells expanded progressively over the experimental time course, reaching ∼60% total blood chimerism by 80 weeks posttransplant. As would be expected, there was very little expansion of wild-type cells in peripheral blood compartments over the study period. Importantly, there was minimal replacement of embryonic-derived populations of cardiac macrophages by donor-derived cells using this approach (230). Collectively, these studies show that the adoptive transfer model is a viable and reliable tool for studying clonal hematopoiesis in the steady state and how it affects disease processes. The aforementioned mouse models of clonal hematopoiesis are summarized in FIGURE 8.

FIGURE 8.

Overview of mouse models of clonal hematopoiesis. Top: mouse models that have been previously used to study clonal hematopoiesis. Bottom: potential sources of mutant murine bone marrow cells that can be used to create the mouse models. Created with BioRender.com, with permission. Tet2, ten eleven translocation 2; Jak2, janus kinase 2; BM, bone marrow; BMT, bone marrow transplant.

7.1.2. Generation of mutant mouse lines.

To study clonal hematopoiesis using the approaches described above, there must be an availability of mutant cells to donate to recipient mice (summarized in FIGURE 8). For some driver genes, there are whole body gene knockouts or transgenic mice that are commercially available, from vendors such as the Jackson Laboratory (234). For instance, Tet2- and Trp53-deficient mice are available from the Jackson laboratory to study the effect of gene loss-of-function mutations. Additionally, there is a commercially available transgenic mouse line that is commercially available, which carries the common R270H missense mutation of TP53, namely, Trp53^tm3.1Tyj. In addition to these mouse lines, there are several commercially available floxed lines, whereby the gene of interest is flanked by LoxP. Some of these genes include Tet2, Trp53, and Dnmt3a-R878H (a DNMT3A point mutation). These mice can be bred with hematopoietic-cell specific Cre-recombinase-expressing mice, such as Mx1-Cre or Vav1-Cre mice, to create hematopoietic cell-specific mutants. While these mice could conceivably be used on their own to study the effect of a clonal hematopoiesis driver gene mutation on disease outcomes, studies have reported that both lines have leaky and/or off-target Cre expression, which may affect the outcome of a study (263–265). Moreover, as outlined above, mice that have a mixture of mutant and wild-type bone marrow cells better model clonal hematopoiesis observed in humans. Thus using the bone marrow from these animals and transplanting it to recipient animals is a much more rigorous approach to studying clonal hematopoiesis in disease settings. In addition, some laboratories have created their own mouse lines with the addition of LoxP sites or mutations in the driver gene of interest, which are currently not commercially available (137, 213, 266, 267). Nevertheless, the development of mouse lines is costly and time consuming, which considerably hinders research in this area. To overcome these obstacles, a CRISPR-based gene editing approach can be used to introduce mutations in a target gene within HSPCs (102, 104, 105, 268). This approach offers more flexibility and rapid gene manipulation, potentially facilitating a more cost-effective analysis of clonal hematopoiesis in disease settings (104, 269).

CRISPR-based genome editing works to induce double-stranded breaks in a target sequence of DNA. To do this, a single-guide RNA and a Cas nuclease, such as Cas-9, are required (269–271). To get these components into the target cell, lentiviruses with vectors encoding these two components can be used (272). Once transcribed, the single-guide RNA directs a Cas nuclease (Cas-9) to the sequence of interest, which generates a double-stranded break in the DNA (270). This triggers DNA repair pathways to fix the break, resulting in a series of random mutations within the sequence of interest (270, 271). Using a method adapted from Heckl et al. (268), our laboratory used a CRISPR-Cas9 approach to study the effect of Tet2- and Dnmt3a-mediated clonal hematopoiesis on nonischemic heart failure in mice (102, 105). To achieve this in the initial studies, lineage-negative bone marrow cells were isolated from wild-type mice and transduced with a lentivirus encoding Cas-9 with an eGFP tag and a single-guide RNA directed at the driver gene of interest (either Tet2 or Dnmt3a). These cells were then transplanted to lethally irradiated mice to create chimeric mice with the mutation. It was observed that only ∼10% of transplanted cells had mutations in the target gene, presumably due to limitations associated with the large size of vector needed to encode Cas9 protein. Nevertheless, this situation modeled clonal hematopoiesis whereby only a small number of cells have the mutations in the driver gene. Importantly, it was found that Tet2 mutant CRISPR-edited cells expanded in the bone marrow and blood and the expansion characteristics were similar to what was observed in prior studies using cells from Tet2-deficient mice. However, Dnmt3a-mutant cells did not show appreciable expansion under these conditions, which is analogous to what has been observed in other studies for this driver gene mutation that employed traditional transgenesis methodologies (213, 238, 273). Our laboratory has since used a CRISPR-Cas9-based approach to study hematopoietic mutations in Ppm1d and loss of the Y chromosome (105, 143). For studies, this approach was refined by using lineage-negative cells from Cas9 transgenic mice, which considerably improved the efficiency of target gene editing.

Despite the clear advantages of this approach, there are several limitations that should be noted. There is a possibility that the CRISPR-Cas9 approach can introduce off-target effects in an unintended part of the genome (274). These can potentially be avoided by careful design of the guide-RNA to make sure that it closely matches the target sequence and by corroborating results with a second guide-RNA that targets a different region in the target gene. Moreover, studies have reported that CRISPR-Cas9 can activate p53 and the DDR pathway in certain cell types (275), which could conceivably lead to confounding findings, especially when studying mutations in driver genes such as TP53 and PPM1D. Nevertheless, it has been reported that HSPCs are relatively resistant to CRISPR-induced double-stranded DNA breaks (276). Furthermore, our laboratory recently used a single-guide RNA targeting noncoding regions of the beta-actin gene within HSPCs and there were no observable effects on the DDR pathway at baseline, suggesting that this approach does not lead to activation of this pathway in these models of clonal hematopoiesis (105). Finally, this approach can only be used to study homozygous mutations (104), rather than hemizygous mutations that are commonly observed in humans with clonal hematopoiesis (35).

7.1.3. Naturally occurring age-related clonal hematopoiesis in animals?

Creating mouse models that harbor hematopoietic mutations in specific driver genes, as described above, is the most straightforward approach to examining the effect of clonal hematopoiesis on age-related disease. However, whether aged mice or other animals exhibit clonal hematopoiesis, and thus could serve as a tool to study age-related clonal hematopoiesis, is of considerable interest. This would more closely mimic the human scenario whereby individuals naturally acquire a mutation in an HSPC that gradually expands over time, presumably in response to exposure to different environmental factors (172). Moreover, it would conceivably avoid some of the caveats of the BMT models described in sect. 7.1.1. To examine whether aged mice also exhibit clonal hematopoiesis, Chin et al. (277) screened for the most common driver gene mutations (found in humans) in the bone marrow of 24-mo-old wild-type mice using error-corrected targeted DNA sequencing. The authors identified that only 2% of mice exhibited mutations in clonal hematopoiesis genes, namely Tet2 and Asxl1. An additional 3% of mice had mutations at a VAF that was below the range for accurate detection. Bone marrow cells from a subset of these aged mice were then transplanted to younger mice, and 10 months later the transplanted bone marrow cells were analyzed for driver mutations via error-corrected sequencing. It was found that 18% of mice had mutations in clonal hematopoiesis driver genes, of which at least 55% were not detected in the donor, even at a low VAF. While it is possible that these mutations occurred after transplantation in the recipient, it is also plausible that transplantation stress provided extremely small clones existing within the transplant with a competitive advantage leading to their expansion to a detectable level within the donor. By comparing the mutational rate per year of HSPCs from mice with humans, the authors suggest that despite mice having a higher yearly mutational rate, their relatively short life span prevents the accumulation of driver mutations and does not provide sufficient time for their expansion. While these data indicate that the frequency of murine clonal hematopoiesis associated with candidate human driver genes is low, it would appear that models employing the transfer of bone marrow harboring mutations in specific genes, to model the human condition of clonal hematopoiesis, would not be confounded by a high level of endogenous clonal hematopoiesis within laboratory mice.

Beyond the laboratory mouse, the presence of clonal hematopoiesis has been reported in aged rhesus macaques. Specifically, in a cohort of rhesus macaques aged 13–40 years (∼33–100 human years), it was observed that 20% of macaques harbored mutations in human driver genes (278). When this analysis was restricted to a VAF of >2%, it was found that the prevalence of clonal hematopoiesis in this population was ∼7%. Similar to what has been observed in humans with CHIP, the most frequently mutated driver gene in aged macaques was DNMT3A (35, 37, 278), although, interestingly, mutations in RUNX1 appeared to also be common, which is less frequently observed in humans with CHIP (35, 37, 278). Finally, the presence of clonal hematopoiesis has recently been reported in the companion dog (279). In a cohort of 93 dogs (median age = 12 years), it was documented that 4.3% carried mutations in genes commonly associated with clonal hematopoiesis. While this is higher than what was reported in aged mice, in comparison with what is observed in humans, this is relatively low (35, 277). It is conceivable that domesticated dogs could be used to uncover new associations between clonal hematopoiesis and disease as extensive veterinary medical records exist without stringent confidentiality regulations.

7.2. Mutation-Specific Studies

7.2.1. TET2.

The second most commonly mutated driver gene associated with age-related clonal hematopoiesis is ten eleven translocation 2 (TET2) (30, 35, 37). The protein encoded by the TET2 gene is an α-ketoglutarate- and Fe²⁺-dependent enzyme that catalyzes DNA demethylation through the conversion of 5-methylcytosine into 5-hydroxymethylcytosine and thereby promotes transcriptional activation (280, 281). Furthermore, depending on the cellular context, TET2 can also induce gene repression via the recruitment of histone deacetylases to gene promotors (101, 282). TET2 mutations are implicated in a wide range of hematological malignancies, including myeloproliferative neoplasms, acute myeloid leukemia, chronic myelomonocytic leukemia, and angioimmunoblastic T-cell lymphoma. The TET2 gene was the first to be reported to give rise to acquired mutations in the blood cells of individuals without hematological malignancies (110). Since then, more than 130 different TET2 mutations have been identified in the blood cells of humans (30, 35–37, 110, 283). Most of these mutations are predicted to lead to a loss of protein function because they are small insertions/deletions or nonsense or missense point mutations (35, 37). Furthermore, large chromosome alterations at 4q may lead to Tet2 loss of function in some circumstances (284). Since its discovery in 2009, studies in mice have improved our understanding of the function of Tet2 within the hematopoietic system and have documented that it plays a critical role in regulating HSPC self-renewal and myeloid cell differentiation (285–287). For instance, Tet2 knockout mice display aberrant patterns of hematopoiesis, and HSPCs deficient in Tet2 exhibit elevated self-renewal and proliferative capacities compared to their wild-type counterparts in competition assays (287, 288). Furthermore, Tet2-deficient mice also show myeloid lineage skewing and extramedullary hematopoiesis, with elevated monocyte and neutrophil content within the spleen (287, 289). Interestingly, it has been reported that Tet2 enzymatic activity is important for myelopoiesis whereas noncatalytic activity is associated with increased HSPC self-renewal (290). Notably, not all Tet2-deficient mice will develop a hematological malignancy, consistent with the concept that blood cancer develops from the stepwise accumulation of mutations in multiple driver genes (226). Recent studies have also shown that TET2 plays a key role in regulating inflammatory signaling, particularly in myeloid cells (101, 282, 291–293).

TET2 was the first clonal hematopoiesis driver gene to be causally connected to cardiovascular disease. To assess this connection, our laboratory and others examined the impact of HSPC Tet2 deficiency in several murine models of cardiovascular disease (101–103, 144, 230). For our initial study, we used a mouse model of atherosclerosis, as atherosclerosis is a primary cause of both coronary heart disease and ischemic stroke, the two cardiovascular conditions where an association was initially documented with CHIP (35). To model clonal hematopoiesis, a competitive bone marrow transplant (BMT) approach was taken, as described in 7.1.1, where a small number of Tet2-deficient hematopoietic cells were transplanted into atherosclerosis-prone Ldlr-/- mice (101). Over the 12-week time course, Tet2-deficient cells gradually expanded in the bone marrow, spleen, and blood, showing a slight myeloid skew (101). Consistent with the definition of CHIP, hematopoietic Tet2 deficiency had no impact on the total number of white blood cells (35–37). Of note, the clonal expansion of Tet2-deficient cells was associated with accelerated atherosclerosis, resulting in a considerable increase in overall plaque size (101). Similar experiments using hematopoietic cells that were heterozygous for Tet2 showed that hematopoietic Tet2 haploinsufficiency also accelerated atherosclerosis. However, the observed phenotype was milder and the kinetics of clonal expansion were slower, suggesting a gene dose-dependent effect. Moreover, it was found that myeloid-specific deletion of Tet2 accelerated atherosclerosis, indicating that myeloid cells and possibly macrophages were playing a key role in promoting atherosclerosis associated with the expansion of Tet2-deficient hematopoietic cells (101). These findings were strengthened when an ensuing independent study reported similar findings also using a mouse model of atherosclerosis (144). In this study, the authors used the conventional BMT method, whereby all transplanted bone marrow cells were Tet2 deficient. Consistently, full hematopoietic Tet2 deficiency led to a considerable increase in plaque size in this model of atherosclerosis (144). However, mice transplanted with Tet2-deficient bone marrow developed extensive xanthomas, splenomegaly, glomeruloslcerosis, and macrophage infiltration into various tissues, which are not typically observed in individuals with CHIP (144). It was also observed that myeloid-specific ablation of Tet2 resulted in an increased atherosclerotic lesion burden, supporting the notion that the atherogenic effects of hematopoietic Tet2 loss-of-function were mediated through myeloid cells (144).

To strengthen and extend this work, we subsequently examined the impact of Tet2-mediated clonal hematopoiesis on experimental heart failure (102, 103). In the first of these studies, two distinct mouse models of heart failure were used, specifically, the left anterior descending (LAD) coronary artery ligation model of myocardial infarction as well as transverse aortic constriction (TAC) model of pressure overload-induced cardiac hypertrophy. Using a competitive BMT approach, we found that hematopoietic Tet2-deficiency promoted adverse cardiac remodeling in both models, as shown by poorer left ventricular function and increased cardiac hypertrophy, fibrosis, and inflammation (103). Analogous to what was observed in the mouse model of atherosclerosis, transplantation of Tet2-hapoinsufficient bone marrow cells also augmented heart failure in these models (103). Moreover, it was noted that myeloid-specific ablation of Tet2 also accelerated heart failure, suggesting that cells of myeloid origin are likely to play a role in worsening the pathology (103). Credence to these findings was given by a follow-up study using the angiotensin II-induced pressure overload model of heart failure (102). For this study, hematopoietic Tet2 loss-of-function was achieved using the lentivirus-mediated, CRISPR/Cas9 system, as described in the section above. Using this approach, we found that mice transplanted with CRISPR-edited Tet2 mutant cells displayed poorer cardiac function and greater indexes of pathological cardiac remodeling after angiotensin II infusion, as indicated by increased levels of myocardial fibrosis and inflammation (102). Importantly, a similar profile of cardiac remodeling was observed following angiotensin II infusion using the CRISPR-mediated and conventional competitive BMT approach, suggesting that the CRISPR-Cas9 system is a reliable tool to study clonal hematopoiesis associated with known driver mutations (102). Collectively, the findings indicate that Tet2-mediated clonal hematopoiesis promotes adverse cardiac remodeling leading to heart failure in mice. Most notably, the initial findings that Tet2 (and Dnmt3a; see the following section)-mediated clonal hematopoiesis promoted heart failure proceeded the first report of an association between clonal hematopoiesis and heart failure (145) and subsequent analysis of heart failure cohorts have noted parallels with mechanistic findings in mice (240). Collectively, these observations serve to validate the utility of the murine models of clonal hematopoiesis.

The findings from the above-mentioned studies suggested that myeloid cells and possibly macrophages were playing a role in the exacerbated pathology observed during hematopoietic Tet2 loss-of-function (101–103). Experiments using cultured macrophages found that Tet2 deficiency did not alter proliferation, apoptosis, oxidized low-density lipoprotein (LDL) uptake, or the expression of regulators of cholesterol uptake, suggesting that other mechanisms may be playing a role (101). However, Tet2-deficient macrophages exhibited elevated inflammatory gene expression and produced greater amounts of interleukin 1β (IL-1β) when stimulated with oxidized LDL, TNF-α, and IFN-γ (101). Concomitantly, it was found that there were increased transcript and protein levels of IL-1β in the atherosclerotic plaques of mice transplanted with Tet2-deficient cells, perhaps indicating a causal role for IL-1β. Experiments in isolated macrophages revealed that Tet2 plays a role in inhibiting transcription of IL-1β via histone deacetylase-mediated histone deacetylation. While this differs from TET2s known role in promoting DNA demethylation (101), it is analogous to a previous study that reported that Tet2 represses transcriptional activation of inflammatory genes in macrophages by recruiting histone deacetylase 2 to the gene promotor (282). Taken together, these data suggested that Tet2-deficient hematopoietic cells may promote cardiovascular disease by generating a pool of macrophages with aberrant IL-1β signaling, a finding that is summarized in FIGURE 9, top.

FIGURE 9.

Summary of the work mechanistically linking ten eleven translocation 2 (TET2)-mediated clonal hematopoiesis to augmented IL-1β signaling. Top: Tet2 deficiency in macrophages leads to an increase in IL-1β transcription as well as increased expression of the NLRP3 inflammasome. As a result of these transcriptional changes, Tet2-deficient macrophages produce higher amounts of active IL-1β. Bottom left: blocking IL-1β production by inhibiting the NLRP3 inflammasome has been shown to reduce the burden of atherosclerosis, heart failure, and metabolic disease in mouse models of Tet2-mediated clonal hematopoiesis. Bottom right: in a subanalysis of the Canakinumab Antiinflammatory Thrombosis Outcome Study (CANTOS) it was found that individuals with hematopoietic mutations in TET2 experienced a substantially greater reduction in major adverse events than individuals without clonal hematopoiesis of indeterminate potential (CHIP) following treatment with the IL-1β neutralizing antibody Canakinumab. Created with BioRender.com, with permission. MACE, major adverse cardiovascular events.

It is known that IL-1β is synthesized as a precursor protein (pro-IL-1β) where it requires cleavage for activation, often by a complex called the NLR family pyrin domain containing 3 (NLRP3) inflammasome (294). In addition to regulating IL-1β, it was found that Tet2 regulates components of the NLRP3 inflammasome complex in macrophages. Moreover, it was found that inhibition of this complex with the small molecule inhibitor MCC950 reduced atherosclerotic lesion size specifically within animals with hematopoietic Tet2 deficiency, supporting a role for IL-1β and the NLRP3 inflammasome in the augmented pathology (101). In terms of the downstream actions of IL-1β and how it promotes atherogenesis, findings suggested that IL-1β released from Tet2-deficient plaque macrophages was promoting endothelial cell activation and P-selectin expression (101). Consequently, this resulted in an increase in the recruitment of monocytes, regardless of Tet2 status, to the plaque thus augmenting vascular inflammation. It can therefore be postulated that Tet2-deficient macrophages act in a catalytic manner within the plaque by augmenting and perpetuating vascular inflammation and therefore driving atherosclerotic plaque progression, which is consistent with a cell nonautonomous effect. Likewise, in studies of experimental heart failure, it was found that IL-1β was implicated in the worsened pathology associated with Tet2 clonal hematopoiesis (102, 103) and that inhibition of the NLRP3 inflammasome complex led to reduced cardiac remodeling and heart failure in the LAD and TAC models (103). Taken together, these studies indicate that Tet2 regulates IL-1β production at multiple levels in macrophages, and thus loss-of-function of Tet2 results in enhanced IL-1β signaling, which in turn drives cardiovascular disease progression (101–103). These findings are summarized in FIGURE 9, bottom left.

The above findings suggest that therapies directed against IL-1β may be particularly effective for the treatment of cardiovascular conditions in individuals with TET2 mutations. Therapies that target IL-1β signaling have recently been explored for the treatment and management of various cardiovascular conditions (295–300). Of particular importance was the CANTOS trial (Canakinumab Antiinflammatory Thrombosis Outcome Study) that examined escalating doses of Canakinumab (50, 150, and 300 mg every 3 months), a neutralizing IL-1β antibody, in post-MI patients with sustained levels of high sensitivity C-reactive protein (CRP), a measure of inflammation (295). This trial included over 10,000 individuals, and it was found that targeting IL-1β at the middle dose (150 mg) led to a 15% reduction in major adverse cardiovascular events (MACE) in the placebo group (295). It was noted that the reduction in MACE was highly dependent on the extent of hsCRP reduction. For instance, Canakinumab-treated patients who exhibited a reduction in hsCRP to less than 2 mg/L showed a 25% reduction in MACE and a 31% reduction in cardiovascular mortality. In contrast, patients that failed to exhibit a reduction in hsCRP also failed to exhibit a reduction in these end points. As CHIP has been associated with increased risk of atherosclerotic cardiovascular conditions and experimental studies in mice suggest that Tet2 loss-of-function may drive the progression of atherosclerosis via an IL-1β-dependent mechanism, an exploratory analysis of CHIP in the CANTOS trial was subsequently performed. In this exploratory analysis, deep-targeted sequencing was performed on nearly 4,000 individuals to determine if individuals with CHIP have a greater reduction in MACE in response to neutralization of IL-1β via the administration of Canakinumab. In this smaller subset of CANTOS participants, it was found that individuals without CHIP that received Canakinumab, at any dose, had a 7% reduction in MACE. However, individuals with driver mutations in TET2 displayed a 62% reduction in MACE in response to Canakinumab therapy (301). Notably, individuals with mutations in other driver genes did not show an enhanced benefit from Canakinumab treatment compared to patients without CHIP, and these findings may reflect the higher levels of IL-1β in TET2-mediated clonal hematopoiesis relative to other driver genes (115). These findings may suggest that individuals with TET2-mediated clonal hematopoiesis are superior responders to Canakinumab therapy (summarized in FIGURE 9). Interestingly, it was also found that TET2 was the most frequently mutated driver gene in this cohort. This finding contrasts with many previous reports, whereby DNMT3A is often the most mutated gene in individuals with clonal hematopoiesis. This observation may suggest that TET2-mediated clonal hematopoiesis is enriched in MI survivors with elevated levels of inflammation. It is also tempting to speculate that hematopoietic mutations in TET2 may be responsible for the persistently high levels of inflammation within some of the CANTOS participants.

The above-described findings are consistent with another study investigating the contribution of a common genetic polymorphism in IL-6 to cardiovascular risk in individuals with clonal hematopoiesis (106). In addition to IL-1β, it has been observed that IL-6 is elevated in cardiovascular tissues in mice carrying Tet2 mutant cells (101). Further, studies indicate that IL-6 is directly downstream of IL-1β signaling and that IL-1β can stimulate the release of IL-6 from the vascular smooth muscle (302, 303). Individuals with the polymorphism IL6R p.Asp358Ala display attenuated IL-6 signaling and thus it was hypothesized that this variant may serve as a genetic proxy to examine the effect of IL-6 inhibition on cardiovascular risk in individuals with DNMT3A-/TET2-mediated clonal hematopoiesis. Indeed, it was found that individuals with DNMT3A/TET2 mutations and this polymorphism displayed a reduced risk of cardiovascular events compared to individuals without this polymorphism. Interestingly, this risk attenuation was not observed in individuals without DNMT3A/TET2 mutations, again suggesting that individuals with clonal hematopoiesis may be superior responders to specific anticytokine therapies. While additional research is still required to understand the specific mechanisms that drive cardiovascular pathology in the setting of clonal hematopoiesis, these data alongside the experimental work suggest that targeting the IL-1β/IL-6 pathway may be particularly effective.

In addition to cardiovascular disease, the effect of Tet2-mediated clonal hematopoiesis on experimental obesity-induced insulin resistance has been assessed (231). In this study, a competitive BMT strategy was employed whereby recipient mice were transplanted with 10% Tet2-deficient or wild-type bone marrow cells mixed with 90% competitor cells. At ∼5 weeks posttransplant, recipient mice were placed on a high-fat/high-sucrose (HF/HS) diet for 14 weeks. It was found that hematopoietic Tet2 deficiency had no effect on weight gain or the composition of body fat over the experimental time course. Nevertheless, mice transplanted with Tet2-deficient cells exhibited exacerbated insulin resistance and higher fasting blood glucose levels compared to mice transplanted with wild-type cells after 14 weeks of an HS/HS diet. Moreover, it was observed that there was a lower pAkt:tAkt ratio in the epididymal white adipose tissue (eWAT) in mice transplanted with Tet2-deficient cells, suggestive of reduced insulin signaling within this tissue. In accordance with previous studies, hematopoietic Tet2-deficiency led to higher levels of IL-1β in the eWAT after HF/HS diet, suggesting a role for aberrant IL-1β signaling in the worsened pathology. Chronic treatment with the inflammasome inhibitor MCC950 abrogated the differences observed between the mice transplanted with the Tet2-deficient and wild-type cells, supporting a role for the NLRP3 inflammasome and IL-1β in the exacerbated insulin resistance observed in mice with Tet2 mutant cells (231).

Given that clonal hematopoiesis has been linked with accelerated biological aging (208, 209), the contribution of Tet2-mediated clonal hematopoiesis to features of aging has been experimentally assessed (230, 231). In these studies, the adoptive transfer approach was used whereby Tet2-deficient bone marrow cells were transplanted to nonconditioned mice. As outlined in sect. 6.1.1, this experimental strategy overcomes several drawbacks of standard BMT approaches, such as the systemic effects of irradiation on cardiovascular organs, destruction of the bone marrow niche, and altered competition dynamics between mutant and wild-type cells (232, 304–306). Moreover, standard BMT approaches are often not ideal for assessing natural biological aging, due to the confounding effects of irradiation on the aging process (307). Thus, with the use of the adoptive transfer approach, mice were transplanted with either Tet2-deficient or wild-type cells and cardiac and metabolic function was assessed over a ∼18-mo experimental timeframe (230, 231). It was observed that Tet2-mediated HSPC expansion promoted cardiac dysfunction and insulin resistance with age in the absence of external injury, suggesting that age-related pathologies that contribute to biological aging are accelerated in mice transplanted with Tet2-deficient bone marrow. While the exact mechanisms by which hematopoietic Tet2-deficiency may accelerate age and reduce health span remain speculative, it is likely that augmented inflammatory processes play a role. This is supported by the observations that transcript levels of inflammatory molecules were elevated in the eWAT and within cardiac macrophages from mice transplanted with Tet2-deficient bone marrow. These findings are consistent with the idea that clonal hematopoiesis may contribute to the processes involved in inflammaging, which in turn accelerates biological aging and age-related disease.

There have been several other experimental studies that have reported that Tet2-mediated clonal hematopoiesis can causally promote disease. First, it has been documented that hematopoietic Tet2 loss-of-function can worsen lung damage in a mouse model of cigarette smoke-induced COPD (203). In this model, it was noted that hematopoietic Tet2 loss-of-function led to increased type I and type II IFN signaling in the lungs, consistent with the notion that Tet2 clonal hematopoiesis can augment inflammation. However, the pulmonary inflammation observed in this model may be more representative of a COPD-associated viral infection, rather than COPD itself. In particular, the authors combined chronic cigarette smoke exposure with short-term administration of poly(I:C), a TLR-3 agonist, which mimics the response to viral infections. As Tet2-mediated clonal hematopoiesis had no effect on lung injury caused by cigarette smoke only, careful interpretation of these data is warranted. Second, a study by the same group reported that Tet2-mediated clonal hematopoiesis can exacerbate experimental gout (205). In vitro work in macrophages suggested that Tet2-inactivation led to enhanced IL-1β secretion following stimulation with monosodium urate crystals, an effect that was blunted with NLRP3 inflammasome inactivation. From this observation, the authors suggested that heightened IL-1β and NLRP3 inflammasome signaling was likely involved in the worsened gout pathology in vivo. Third, it has been reported that hematopoietic Tet2 loss-of-function can promote tumor progression in a mouse model of lung cancer (308). Using hematopoietic-specific Tet2-deficient mice, the authors found that the growth of subcutaneously implanted lung cancer cells was greater in Tet2-deficient mice. In terms of mechanism, they suggested that Tet2-deficient myeloid cells produce S100a8/S100a9, which promotes VEGF-A release by tumor cells, promoting angiogenesis and consequently tumor growth. In contrast, another group has reported that myeloid-specific Tet2 deficiency suppresses tumor growth in a mouse model of melanoma (309). It is possible that mechanistic differences in the way different tumors grow could explain these discrepant findings.

7.2.2. DNMT3A.

DNA methyltransferase 3 A (DNMT3A) is the most frequently mutated candidate driver gene associated with clonal hematopoiesis in aged individuals (30, 35–37). It belongs to a family of highly conserved cytosine methylases and plays an important role in epigenetic regulation by catalyzing the addition of methyl groups to the cytosine residue of CpG nucleotides (310). The DNMT3A protein consists of three major domains, namely the Pro-Trp-Trp-Pro domain, the ARTX-DNMT3-DNMT3L (ADD) domain, and the catalytic methyltransferase domain (311). The majority of mutations in DNMT3A occur in the region encoding the methyl-transferase domain, with a significant proportion of missense mutations occurring at the R882 codon (311–313). The R882 codon is a hot spot, where mutations frequently occur, with the R882H variant being the most frequent (314), although missense mutations outside the R882 codon as well as truncating mutations can be found within regions encoding the other domains (311–313). The R882H variant appears to be a unique mutation in that it has a dominant-negative effect on the wild-type protein, where it can dimerize with wild-type DNMT3A and reduce methyltransferase activity of the wild-type protein (314, 315). Furthermore, studies indicate that this point mutant protein can interact with Polycomb repressive complex 1 to inhibit HSPC differentiation, suggesting that this mutation may also have gain-of-function features (316). Experimental studies examining Dnmt3a loss-of-function within HSPCs have shown that mutant cells favor self-renewal over differentiation (213, 317). Effectively, this results in HSPC clones that can engraft and expand normally; however, they are impaired in their ability to produce blood cells (213, 214). Additionally, given the increased self-renewal of Dnmt3a-mutant HSPCs, studies indicate that they are more resistant to serial transplantation stress than wild-type HSPCs, showing little sign of exhaustion (213, 214). Further, mice with hematopoietic Dnmt3a ablation exhibit signs of primitive hematopoiesis, such as the upregulation of fetal hematopoietic gene expression, preferential myelopoiesis, as well as expansion of HSPCs within the liver (318). Moreover, studies suggest that DNMT3A plays an important role in regulating inflammatory pathways within several immune cell subsets, including mast cells (319), T-helper cells (320–322), and macrophages (322–324).

To explore the potential connection between DNMT3A-mediated clonal hematopoiesis and cardiovascular disease, our laboratory assessed the effect of hematopoietic Dnmt3a deficiency on experimental heart failure induced by angiotensin II infusion (102). Dnmt3a loss-of-function was achieved by using the same CRISPR-Cas9 approach described in the sections above, whereby mutations were induced in the Dnmt3a gene within lineage-negative cells from the bone marrow. Following genetic editing, the cells were transplanted to lethally irradiated mice. Unlike what was observed in mice transplanted with Tet2-deficient bone marrow cells, Dnmt3a CRISPR-edited mutant cells did not expand over the 4-mo time course of this study (102). This finding is similar to previous studies that have reported that Dnmt3a-deficient HSPCs only expand after sequential bone marrow transplants or during aging (213, 238, 273). Despite this, mice that had 5–10% Dnmt3a mutant cell chimerism displayed poorer cardiac function following angiotensin II infusion as well as greater indexes of adverse cardiac remodeling (102). In regards to the underlying mechanisms, data suggested that Dnmt3a mutant cells were potentially accelerating heart failure by augmenting inflammatory signaling pathways (102). In particular, it was observed that mice transplanted with Dnmt3a mutant cells had greater numbers of macrophages in the heart as well as increased expression of T-cell genes after angiotensin II infusion. Moreover, using a Dnmt3a CRISPR-edited macrophage cell line, it was found that there was increased expression of proinflammatory cytokines, Il-6, chemokine (C-X-C motif) ligand (Cxcl) 1, Cxcl2, and Ccl5 following stimulation with LPS. Of note, this inflammatory profile differed from that observed for Tet2 CRISPR-edited macrophages, which showed an increase in Il-1β, Il-6, and C-C motif chemokine ligand (Ccl) 5 but not Cxcl1 or Cxcl2 following LPS stimulation (102). These findings support the notion that mutations in different driver mutations may result in clones with divergent functions, which may ultimately promote disease progression through alternative mechanisms. Overall, the findings from this study suggest that Dnmt3a-mediated clonal hematopoiesis promotes heart failure, likely via augmenting inflammatory processes (102). Nevertheless, follow-up studies are required to better understand the immune cell subsets and the precise inflammatory pathways involved in accelerated cardiac pathology.

These findings were strengthened when an ensuing study analyzed the inflammatory signature of peripheral blood mononuclear cells from heart failure patients with hematopoietic DNMT3A mutations (240). In this study, it was found that monocytes from heart failure patients with DNMT3A mutations had an increased proinflammatory signature compared to heart failure patients without a DNMT3A mutation. In particular, it was observed that these monocytes had increased expression of various cytokines and immune molecules, including IL6, CXCL2, IL1β, TNF, and NLRP3. In addition to this, the authors found that monocytes from patients carrying DNMT3A mutations had higher levels of genes involved in T-cell stimulation. When examining T-cell-specific gene regulation, it was noticed that there was dysregulation of a myriad of genes in patients who harbored DNMT3A mutations, and these appeared to occur across all T-cell subsets in a subset-specific manner. Of note, the authors observed changes in expression of various T-cell receptor genes, perhaps indicative of skewing of the T-cell receptor repertoire. Interestingly, it was noted there were widespread genetic changes in monocytes and T cells, which lacked clustering, suggesting that all cells, regardless of whether they had a mutation or not, were affected. It thus could be speculated that DNMT3A mutant cells have effects on wild-type cells, making them more proinflammatory, consistent with a cell nonautonomous effect, which has been observed for other driver gene mutations (101, 105, 230). Overall, the findings add further weight to the notion that DNMT3A promotes heart failure via augmented inflammatory signaling.

Beyond heart failure, the effect of hematopoietic mutations in DNMT3A on the development of osteoporosis has been recently assessed (201). Using the whole bone marrow transplant approach, the study found that mice with Dnmt3a-deficient bone marrow had a decreased bone mineral density at 20 weeks posttransplant compared to animals with wild-type bone marrow cells. The authors of this study observed that transplantation of Dnmt3a-deficient bone marrow led to increased osteoclast number without affecting the number of osteoblasts. The adverse effect of Dnmt3a-deficient bone marrow on bone mineral density was reversed in mice treated with alendronate, an osteoclast inhibitor, providing further evidence that osteoclasts were playing a role in the exacerbated pathology. In terms of mechanism, it was noted that Dnmt3a-mutant macrophages had the same differentiation potential as wild-type macrophages to become osteoclasts. However, Dnmt3a-mutant macrophages appeared to secrete proinflammatory molecules that promoted the differentiation to osteoclasts, consistent with a cell nonautonomous effect, which has been observed previously for DNMT3A and other clonal hematopoiesis driver genes. Using a CRISPR-screening assay and a cytokine array, the authors identified IL-20 as a key molecule secreted by Dnmt3a-mutant macrophages that promotes their differentiation to osteoclasts (169). Further, it was found that at a molecular level, Dnmt3a-deficient macrophages had methylation erosion and open areas of chromatin, which allowed for the binding of interferon regulatory factor 3 (Irf3), an activator of NF-κB signaling. This increase in Irf3-NF-κB signaling was shown to be responsible for increasing the expression of IL-20, thus leading to osteoclastogenesis, in animals with Dnmt3a-mutant bone marrow. Overall, these data are supportive of the concept that DNMT3A-mediated clonal hematopoiesis can promote osteoporosis. However, it would be of interest to investigate this phenotype using bone marrow chimeras where there is a mixture of Dnmt3a- and wild-type bone marrow as this would not only better model clonal hematopoiesis in humans but also allow for better understanding of the cell nonautonomous effects of the Dnmt3a-mutant macrophages.

7.2.3. JAK2.

JAK2 is a member of the Janus family of cytoplasmic nonreceptor tyrosine kinases and plays an important role in immune signaling processes (325). Specifically, it is involved in many cytokine signal transduction pathways, such as those of interferons, interleukins, granulocyte colony-stimulating factor, erythropoietin, and thrombopoietin (325, 326). The most frequent type of JAK2 mutation associated with both clonal hematopoiesis and myeloproliferative disorders, such as polycythemia vera, is the V617F mutation, which is often denoted as JAK2^V617F (35, 37, 144, 327–329). This mutation is a guanine-to-thymidine substitution, resulting in an amino acid substitution of valine for phenylalanine at codon 617 of JAK2 (330). The mutation has been shown to cause changes to the autoinhibitory site of JAK2 kinase, resulting in a constitutively active protein. Thus, this type of mutation is considered to be a gain-of-function mutation (331). The underlying mechanisms by which JAK2^VF617 mutations confer a competitive advantage in the hematopoietic system remain elusive, although it has been shown that mutant HSCs have reduced self-renewal while lineage-restricted progenitors are skewed toward proliferation and differentiation (332). It has been suggested that JAK2 mutations may occur in lineage-restricted stem cells, such as myeloid-restricted stem cells (333–335); however, studies have challenged this and have observed that individuals that harbor the JAK2^V617F mutation possess it in all blood cell compartments, consistent with the idea that the mutation occurs in HSCs (336, 337). Nevertheless, it is possible that the mutation occurs in HSCs in some individuals whereas in others it occurs in lineage-restricted progenitor cells, and this leads to different clinical manifestations. Furthermore, it has been shown that JAK2^VF617 mutations result in myeloid lineage skewing, similar to what has been observed for other driver gene mutations (101, 107, 338). Moreover, it has been identified that carriers of certain germline variants have a higher risk of developing mutations in JAK2 (229, 327, 339); however, the mechanisms by which these germline variants predispose individuals to develop mutations in JAK2 remain incompletely understood. While the JAK2^V617F mutation was thought to be exclusively in individuals with MPNs and related hematological conditions, studies examining the prevalence of blood cancer driver gene mutations in healthy individuals suggested that this is not the case (35, 37, 340). Indeed, a study examining the prevalence of the JAK2^V617F mutation in the peripheral blood of 19,958 adult Danish citizens found that 3.1% of individuals harbor this mutation (341). Importantly, of these JAK2^V617F-positive individuals, only 2.3% were diagnosed with a myeloproliferative neoplasm (341). As discussed in sect. 5.1.1 of this review, a gene-specific analysis discovered that individuals with driver mutations in JAK2 were at high risk of developing coronary heart disease (144). However, interestingly an exome-wide study has associated the JAK2^V617F allele with lower triglyceride and LDL cholesterol levels (342), possibly suggesting that there are alternate mechanisms that factor into the increased risk of coronary heart disease in individuals with JAK2^V617F mutations. Given the relatively high prevalence of JAK2^V617F mutations and their association with cardiovascular disease risk, it has been of interest to understand the contribution of JAK2^V617F-mediated clonal hematopoiesis to cardiovascular conditions.

Our group assessed the relationship between clonal hematopoiesis associated with JAK2 gain-of-function and heart failure (107). To characterize the profile of mutant clone expansion, a competitive BMT approach was taken using cells from transgenic mice strain that expressed human JAK2^V617F. It was observed that JAK2^V617F mutant cells expanded exclusively within the myeloid lineage, and this expansion was accelerated following LAD ligation (107). However, at 16 weeks after BMT, all mice exhibited myeloproliferative disease, with elevated numbers of platelets and red blood cells; thus this approach was not appropriate for studying CHIP in the setting of cardiovascular disease (107). To avoid the myeloproliferative phenotype, which could conceivably confound the study, a lentiviral approach was adopted, whereby the exogenous JAK2 expression was under the control of a myeloid-specific promoter and enhancer. Specifically, lineage-negative bone marrow cells were transduced with a lentivirus encoding JAK2^WT (control) or JAK2^V617F (mutant) and transplanted into lethally irradiated recipients. Crucially, mice transplanted with JAK2^V617F cells in this manner did not display alterations in blood cells or platelets, suggesting it was a superior approach for studying clonal hematopoiesis (107). With the use of the LAD model of myocardial infarction and TAC model of pressure overload, mice transplanted with a small number of JAK2^V617F mutant cells exhibited poorer cardiac function and greater pathological cardiac remodeling compared to animals transplanted with JAK2^WT cells (107). Furthermore, there was elevated expression of proinflammatory cytokines, such as IL-1β and IL-6, within the hearts of animal-transplanted JAK2^V617F mutant bone marrow cells (107). In support of this, THP-1 macrophages that were transduced with a lentivirus expressing the human JAK2^V617F mutation had elevated expression of IL-1β, IL-6, TNF-α, CCl2, and absent in melanoma 2 (AIM2) following stimulation with LPS, consistent with a proinflammatory phenotype. Overall, findings from this study indicate that clonal hematopoiesis, associated with the JAK2^V617F gain-of-function, mutation can accelerate heart failure, likely through overactivation of inflammatory pathways. However, further mechanistic studies are required to better understand the precise inflammatory pathways by which JAK^V617F-mediated clonal hematopoiesis promotes heart failure.

In addition to these studies on heart failure, three studies from the same group have provided evidence that hematopoietic Jak2^V617F mutations accelerate experimental atherosclerosis (239, 343). In the first of these studies, the authors used Jak2^V617F-floxed mice, which have an inverted exon 14, carrying the valine to phenylalanine mutation, downstream of the endogenous exon 14 of the Jak2 gene. Both exons are surrounded by loxP and loxP511 sites, allowing for Cre-mediated removal of the endogenous exon 14 and inversion of the mutated exon 14. The Jak2^V617F-floxed mice were bred with Mx1-Cre mice to conditionally induce the Jak2^V617F mutation in hematopoietic cells (239). Bone marrow cells from mice carrying a Jak2^V617F mutation were subsequently transplanted to Ldlr^-/- atherosclerosis-prone mice, which were then fed a high-fat, high-cholesterol diet. It was found that mice transplanted with Jak2^V617F mutant cells had larger and more complex atherosclerotic lesions; however, with the use of this approach, recipient mice exhibited a myeloproliferative disease phenotype, which is not representative of CHIP. The second of these studies used a different approach to study JAK2^V617F-mediated clonal hematopoiesis in atherosclerosis, which avoided the MPN phenotype observed in the initial study (343). To do this, the authors bred Jak^V617F-floxed mice with Cre-recombinase-expressing mice, which were either under the Cx3cr1 or S100A8 promotor. This resulted in Jak2^V617F being conditionally transduced in macrophages and neutrophils, respectively. The bone marrow from these strains was then transplanted to lethally irradiated atherosclerosis-prone Ldlr^-/- mice, and 4 weeks after the transplant, mice were placed on a Western diet for 15 weeks. Importantly, the creation of these mouse strains did not lead to an observable myeloproliferative disease phenotype, thus allowing for a clearer assessment of atherosclerosis. It was found that mice transplanted with bone marrow from macrophage-specific Jak2^V617F mutant animals had larger atherosclerotic lesions than mice transplanted with control cells. This was also associated with greater indexes of plaque vulnerability, such as a larger necrotic core and greater proliferation of macrophages. In contrast, there was no difference in atherosclerotic lesion burden between mice transplanted with bone marrow cells from neutrophil-specific Jak2^V617F mutants versus those transplanted with bone marrow from control. These data suggest that Jak2^V617F mutations in macrophages may be playing a role in the atherogenic phenotype observed in these animals (343).

To interrogate the mechanism, the authors subsequently performed in vitro studies in Jak2^V617F mutant macrophages and found that they exhibited perturbations in metabolic activity, which resulted in increases in reactive oxygen species production, DNA damage, and activation of the AIM2 inflammasome (343). To examine this in vivo, a competitive BMT approach was taken, whereby Ldlr^-/- mice were transplanted with a mixture of 20% hematopoietic Jak2^V617F-mutant bone marrow cells (or control) and 80% wild-type bone marrow cells. It is noteworthy that the Jak2^V617F-mutant bone marrow cells were taken from the same mouse strain (i.e., the Jak2^V617F Mx1-Cre) as used in the first study, and this approach led to alterations in the numbers of blood cells across the 7-week experimental time course. Using this model, the authors found that genetic ablation of the AIM2 but not the NLRP3 inflammasome was sufficient to reduce lesion size and necrotic core area in mice transplanted with 20% Jak2^V617F mutant bone marrow, suggesting a role for the AIM2 inflammasome. Further, the authors found that inhibition of IL-1β signaling led to signs of improved plaque stability without affecting lesion size in mice transplanted with JAK2^V617F mutant cells. Overall, these findings suggest that the AIM2 inflammasome and IL-1β may, at least in part, play a role in the accelerated atherosclerosis observed with Jak2^V617F-mediated clonal hematopoiesis. Curiously, as blockade of IL-1β signaling had no effect on lesion size, it may be of interest to also examine the contribution of IL-18 to the phenotype. IL-18 is also produced by the AIM2 inflammasome, and previous studies have shown that IL-18 can accelerate lesion development in the setting of atherosclerosis (344–346). It is thus conceivable that IL-18 may play a role in the atherogenic phenotype observed in mice transplanted with JAK2^V617F mutant cells. Nevertheless, these findings may suggest that like carriers of TET2 hematopoietic mutations, individuals with cardiovascular disease and JAK2^VF617 mutations may at least benefit from therapies that inhibit IL-1β signaling, such as Canakinumab (101, 301).

A subsequent study by the same group reported that erythroid lineage Jak2^V617F expression can augment experimental atherosclerosis (347). To investigate this, the authors knocked down the Ldlr in erythroid-specific Jak2^V617F-expressing mice. It was found that erythroid-specific Jak2^V617F mutation led to increases in plaque necrosis, erythrophagocytosis, and ferroptosis without changes in hematological parameters. Analysis of red blood cells carrying the Jak2^V617F mutation indicated impaired antioxidant defenses and increased lipid hydroperoxides. Furthermore, erythrophagocytosis of Jak2^V617F mutant red blood cells by macrophages led to ferroptosis. As ferroptosis has been shown to contribute to the development of atherosclerosis, the authors postulated that heightened ferroptosis may be contributing to the accelerated atherosclerosis observed in erythroid-specific Jak2^V617F-expressing mice. Blockade of ferroptosis with Liproxstatin-1 reversed accelerated atherosclerosis in erythroid-specific Jak2^V617F-expressing mice, suggesting that targeting ferroptosis may be beneficial for individuals with atherosclerosis carrying JAK2^V617F mutations in their erythrocytes.

JAK2-mediated clonal hematopoiesis has recently been examined in the setting of lung disease (348). In collaboration with prospective clinical data that showed that hematopoietic mutations in JAK^VF617F are more common in patients with pulmonary hypertension, the authors assessed whether JAK2-mediated clonal hematopoiesis is causally linked to pulmonary hypertension. To do this, the authors performed competitive BMTs using Jak2 mutant cells, and 5 weeks later mice were subjected to chronic hypoxia to induce arterial remodeling within the right side of the heart and lungs, which is characteristic of pulmonary hypertension. To determine the threshold where Jak2^V617F clones become pathological, the authors performed competitive BMTs, whereby recipient mice received a different percentage of Jak2^V617F mutant cells mixed with wild-type cells. It was found that a chimerism of 1–19% of Jak2^VF617F mutant cells was sufficient to increase right ventricular pressure and hypertrophy of the right ventricle, compared to animals with wild-type cells following hypoxia. Importantly, these changes were not accompanied by differences in white blood cell, hemoglobin, or platelet counts, consistent with what is observed in CHIP. In mice that displayed 1–19% chimerism of Jak2^V617F mutant cells, it was observed that the neutrophil chimerism was significantly higher in the lungs than in the blood, perhaps indicative that neutrophils were playing a role. RNA sequencing of neutrophils from the bone marrow, blood, and lungs of Jak2^VF617F mice revealed enrichment of genes involved in the canonical IL-6-JAK-STAT3 signaling pathway. In particular, it was noticed that Activin A receptor like type 1 (Acvrl), which encodes ALK1, was the most upregulated gene in this pathway, suggesting a role for this molecule in worsened outcomes in mice transplanted with Jak2^VF617F mutant cells. Using JAK2^V617F knockin human cell lines, the study found that JAK2^V617F transcriptionally upregulates ACVRL1 by STAT3 binding, which in turn induces expression of ALK1 to upregulate the Smad1/5/8 signaling pathway, which likely contributes to the arterial remodeling observed in this model. Inhibition of ALK1/2 was found to prevent hypoxia-induced pulmonary hypertension in whole body Jak2^V617F mutant mice. This suggests that inhibition of ALK1/2 may be an effective therapy for individuals with pulmonary hypertension and JAK2^V617F mutations. Although, it should be mentioned that these experiments were performed in whole body Jak2^V617F transgenic mice, where every cell harbored the Jak2^V617F mutation and not a mouse model of JAK2^V617F-mediated clonal hematopoiesis. Therefore, whether ACVRL1 is involved in the worsened pathology under circumstances that more closely model clonal hematopoiesis remains to be tested.

As briefly discussed in sect. 5.2, JAK2^VF617-mediated clonal hematopoiesis has been associated with increased rates of thrombosis (158). It has been observed that mice heterozygous for the JAK2^V617F allele in their hematopoietic cells show an increased propensity for developing spontaneous pulmonary thrombosis (158). From a mechanistic level, it has been documented that the mutation results in an increase in extracellular trap formation by neutrophils (NETs). In particular, it has been found that JAK2^V617F human and mouse neutrophils have increased NET formation in response to ionomycin treatment in vitro (158). Moreover, it has been observed that JAK2^V617F mutant cells from humans have increased expression of protein arginine deiminase 4 (PAD4), an enzyme involved in the formation of NETs (158). It is known that NET formation can lead to thrombosis by the secretion of molecules such as histones and serine proteases, which cause coagulation and platelet aggregation (349). As thrombosis is central to the pathology of many ischemic cardiovascular conditions, such as myocardial infarction, it is conceivable that these mechanisms may also extend to these conditions. Nevertheless, the contribution of hematopoietic JAK2^VF617 mutations to thrombosis should ideally be examined in a mouse model of clonal hematopoiesis rather than mice carrying the JAK2^VF617 mutation in all hematopoietic cells as this is more representative of MPNs, which will confound the findings.

From the studies described above, it is apparent that hematopoietic mutations in JAK2^VF617F contribute to disease pathology via multiple different mechanisms. JAK2-mutant HSPCs give rise to multiple blood cell lineages, including myeloid cells, platelets, and red blood cells; thus it is conceivable that JAK2 mutations may have different effects on different lineages of blood cells (350). As mice that carry the JAK2^VF617 mutation in all hematopoietic cells typically exhibit signs of myeloproliferative disease, some of the models thus far have focused on using cell-specific mutants of JAK2 (343, 347, 350). These are advantageous for examining the effect of the mutation in a particular cell and for avoiding the confounding effects of myeloproliferative disease; however, they also discount the effect of the mutation in other cell types. Moreover, they can overlook potential interactions between mutant cell types, which may contribute to the pathogenesis of the disease. Thus care should be taken when interpreting findings from these models. It is also likely that as different mechanisms of pathogenesis underlie different diseases, and so it is possible that not all mutations will contribute to disease via the same mechanism. It is expected that our understanding of how hematopoietic JAK2^VF617 mutations contribute to disease pathology will deepen in the coming years, with growing research interest in this area.

7.2.4. TP53.

TP53, which encodes tumor suppressor protein 53, is one of the top 10 genes mutated in individuals with CHIP (30, 35, 37). As discussed in sect. 3.3, TP53 mutant clones are enriched in individuals that have been treated for certain cancers and this is phenomenon is referred to as therapy-related clonal hematopoiesis (116, 140, 351, 352). Approximately 90% of individuals with TP53-mediated clonal hematopoiesis have missense mutations, most of which occur in the DNA-binding domain of the p53 protein (35, 37, 139, 215). This disrupts the transcriptional activity of p53, and as a consequence, there is a loss of protein function. Furthermore, there are multiple hotspot mutations that have been identified, with the most common mutations involving codons 248, 273, and 220 (215). Interestingly, studies have shown that mutations in codons 248 and 273 (denoted as p53^R248W and p53^R273H) have new gain-of-function properties independent of wild-type p53 protein (215, 353). TP53 is a transcription factor that is activated by a myriad of cell stressors including oncogene signaling, DNA damage, oxidative stress, hypoxia, and inflammation (354, 355). Once activated, it regulates numerous genes, particularly those involved in DNA repair, cell cycle arrest, and apoptosis (354, 355). TP53 plays a critical role in the DNA-damage response pathway, and thus it is perhaps not surprising that TP53 mutations are found in approximately half of all human cancers (356). Furthermore, TP53 has been implicated in a wide range of age-related diseases beyond cancer, such as myocardial infarction (357, 358), stroke (359, 360), and atherosclerosis (361–364); however, considerably fewer studies have examined the contribution of TP53-mediated clonal hematopoiesis to age-related diseases.

TP53 plays an important role in the regulation of the hematopoietic system whereby mutations lead to alterations in HSPC function. Studies have shown that TP53 functions to maintain HSPC quiescence and that mice deficient in Trp53 have a larger HSPC pool, due to loss of quiescence (365, 366). Similarly, in competition assays, p53 mutant HSPCs outcompete their wild-type counterparts, again suggesting a greater proliferative capacity due to loss of stem cell quiescence (153, 365, 366). It has also been found that certain TP53 mutant proteins interact with the epigenetic regulator EZH2 in HSPCs to regulate their self-renewal and differentiation (215). Moreover, TP53 mutations can enhance the repopulating potential of HSPCs thus providing mutant HSPCs with a competitive advantage following transplantation stress (215). As detailed in sect. 3.3, TP53 mutant HSPCs are more resistant to radiation and specific chemotherapies, which provides the mutant clones with a competitive advantage following exposure to these agents (367, 368). It is also noteworthy that TP53 mutations in HSPCs and progeny immune cells result in widespread changes in inflammatory gene sets (141, 215, 369, 370). Collectively, these findings suggest that TP53 is critical to the maintenance of the hematopoietic system and mutations lead to a wide variety of alterations in HSPC and immune function.

The effect of TP53-mediated clonal hematopoiesis on cardiovascular disease has recently been examined by two independent studies (141, 153). The first of these studies examined the effect of TP53-mediated clonal hematopoiesis on anthracycline-induced cardiotoxicity (AIC) (141). AIC is a side effect of treatment with the anthracycline class of anticancer drugs (371). This class of cancer drugs is commonly used to treat many solid organ tumors as well as hematological malignancies and includes the chemotherapeutic agents doxorubicin, daunorubicin, epirubicin, and idarubicin (372). The risk of AIC increases as the cumulative dose administered increases with studies reporting a 3–5% risk for 400 mg/m² and 18–48% risk for 700 mg/m² (373–375). Moreover, this risk may higher in individuals that are either very young or old, for those receiving concurrent chest irradiation, or for individuals with preexisting cardiovascular disease (373). AIC has been previously classified into three distinct types based on when individuals present with symptoms (376). For instance, cardiotoxicity that occurs immediately or within 14 days of commencing treatment is classified as acute, whereas cardiotoxicity that manifests within 1 year of commencing therapy or many years after treatment cessation has been classified as early onset chronic and late-onset chronic, respectively. Chronic cardiotoxicity is traditionally characterized by a progressive decline of cardiac function and is generally irreversible, is resistant to standard heart failure treatment, and is associated with a poor prognosis (376). However, it should be mentioned that newer research suggests that AIC manifests as a spectrum rather than being three separate classifications (377). As discussed in sect. 3.3, TP53 mutant clones are often enriched in cancer survivors, as studies have shown that certain chemotherapies can provide these clones with a competitive advantage leading to their expansion (116, 133, 140). Considering these observations, our laboratory speculated that therapy-related clonal hematopoiesis associated with clonal expansion of hematopoietic TP53 mutations may contribute to the pathology of AIC (141).

To examine the relationship between TP53-mediated clonal hematopoiesis and AIC, the effect of doxorubicin on the expansion of Trp53 mutant clones was examined in mice (141). Using the adoptive transfer model described in the sections above, Sano et al. (141) found that cyclic administration of doxorubicin led to the expansion of both heterozygous Trp53-deficient cells and Trp53^R270H mutant cells over a period of 9 weeks. The doxorubicin-induced clonal expansion of Trp53 mutant cells was associated with significantly worse cardiotoxicity, characterized by poorer indexes cardiac function, myocardial wall thinning, reduced capillary density, and elevated fibrosis and inflammation when compared to mice that were transplanted with wild-type cells. From a mechanistic point of view, it was found that neutrophils played a key role in the accelerated doxorubicin-induced cardiotoxicity observed in mice transplanted with Trp53 mutant cells. In particular, neutrophil depletion ameliorated the detrimental effects of doxorubicin on several indexes of cardiac damage observed in mice with expanding Trp53-deficient cells. Moreover, neutrophils from heterozygous Trp53-deficient mice displayed elevated transcript levels of various proinflammatory mediators, consistent with the notion that neutrophils are involved in the poorer outcomes observed in mice transplanted with Trp53 mutant cells. Collectively, these findings may suggest that therapy-related clonal hematopoiesis associated with the expansion of TP53 mutant clones contributes to the risk of AIC in cancer survivors. Although, clinical validation of these findings will be required by studies on anthracycline-treated cancer survivors.

The contribution of TP53-mediated clonal hematopoiesis to the development of atherosclerosis has also been examined (153). To assess this relationship, Zekavat et al. (153) used the congenic competitive BMT approach that is described in sect. 7.1.1, to generate chimeric atherosclerosis prone Ldlr-/- mice carrying 20% Trp53^-/- hematopoietic cells. Control mice received a BMT of 20% wild-type Trp53^+/+ cells. Recipient mice were then fed an HFHC diet for 9 weeks to induce atherosclerosis. Over the study period, Trp53^-/- cells modestly expanded in both HSPCs and white blood cells without affecting overall white blood cell counts, consistent with the definition of CHIP. It was observed that mice transplanted with Trp53^-/- cells had ∼40% larger atherosclerotic lesions compared to mice that were transplanted with wild-type cells. In association with larger lesions, mice transplanted with Trp53^-/- bone marrow had increased numbers of macrophages within the plaques, although there were no changes in the expression of the proinflammatory cytokines IL-1β and IL-6. Interestingly it was noted that there was substantially higher chimerism of Trp53^-/- macrophages in the plaque compared with Trp53^-/- monocytes in the blood, raising the possibility that Trp53^-/- macrophages may be clonally expanding within the plaque. Indeed, it was found that Trp53^-/- plaque macrophages were more proliferative and in vitro work revealed that Trp53^-/- macrophages have accelerated mitotic cycle progression. Overall, these findings suggest that TP53-mediated clonal hematopoiesis accelerates atherosclerosis by generating a pool of macrophages with greater proliferation capacity.

7.2.5. PPM1D.

Clonal hematopoiesis has also been associated with mutations in protein phosphatase Mn²⁺/Mg²⁺-dependent 1D (PPM1D), a gene encoding a serine-threonine phosphatase that is also referred to as wild-type p53-induced phosphatase 1 (WIP1) (37). PPM1D is upregulated in response to DNA damage and negatively regulates several proteins involved in the DNA-damage response pathway, including ATM serine/threonine kinase (ATM), checkpoint kinase (CHK) 1, CHK2, and H2A histone family, member X (γH2AX) (378, 379). It functions as a regulator to return cells back to homeostasis after DNA damage, preventing cells from entering apoptosis (378, 380). Almost all individuals with mutations in PPM1D have truncating mutations in exon 6 of the gene, which result in a protein product that is highly stable thus increasing net PPM1D activation (116, 133, 139, 381, 382). Therefore, PPM1D mutations are often regarded as gain-of-function mutations. As mentioned in sect. 3.3, PPM1D mutations are commonly enriched in the blood of individuals treated with chemotherapy or radiation for cancer, which reflects the fitness advantage that PPM1D mutant cells have upon exposure to these agents (116, 137, 140). Specifically, studies indicate that PPM1D mutations reduce the stress response of cells and thus mutant cells show reduced rates of apoptosis when exposed to cell stressors, such as radiation or chemotherapy (137, 383). The overall result is increased survival of PPM1D mutant HSPCs compared to wild-type HSPCs and thus leading to an increase in the number of mutant leukocyte progeny with each successive stress exposure (137). However, it should be noted that not all cell stressors favor the expansion of Ppm1d mutant clones. It has been documented that Ppm1d mutant cells are only more fit in the context of certain chemotherapeutic agents, such as cisplatin, etoposide, doxorubicin, or cytarabine but not vincristine (137). Moreover, Ppm1d mutant cells have a selective disadvantage following serial transplantation stress (137). Overall, this suggests that cell stressors, specifically those that activate the DNA-damage-response pathway, favor the expansion of PPM1D mutant clones. These findings may explain why PPM1D mutations are rarely observed in patients with primary myeloid neoplasms compared to being relatively frequent (∼15–20%) in patients with therapy-related myeloid neoplasms (137).

As mentioned in the section above, cancer survivors have an increased medium- to long-term risk of heart failure (134, 373). This increased risk appears to only be for survivors of certain cancers, and mortality due to cardiovascular causes is often greater than cancer itself after ∼10-year follow-up (384). As stated in the section above, this type of heart failure is more frequently observed in patients treated with anthracyclines and can be classified as chronic late-onset cardiotoxicity (376). Given that it has been reported that hematopoietic PPM1D mutations in exon 6 are enriched in cancer survivors and the links between cancer and delayed onset heart failure, our laboratory hypothesized that PPM1D-mediated clonal hematopoiesis may accelerate nonischemic heart failure following cancer therapy (105). Thus an experimental study was performed to assess the putative connection between PPM1D-mediated clonal hematopoiesis and nonischemic heart failure. To assess this, a BMT approach was taken whereby Ppm1d-mutated HSPCs were transplanted to recipient mice. The Ppm1d-mutated HSPCs were generated using the CRISPR-cas9 approach described in sect. 7.1.2, whereby a guide RNA was designed to target regions within exon 6 of the Ppm1d gene. This mimics the human scenario whereby the majority of individuals harbor activating mutations in exon 6 of the PPM1D allele. To induce nonischemic heart failure, mice were subjected to a continuous infusion of a supraphysiological dose of angiotensin II. With the use of this model, it was found that mice with hematopoietic mutations in Ppm1d displayed worse cardiac function and exacerbated cardiac remodeling following infusion with angiotensin II. As monocytes and macrophages appeared to be playing a key role in the cardiac remodeling phenotype observed in this model, follow-up experiments were performed to examine the effect of Ppm1d mutations in macrophages. It was found that Ppm1d mutant macrophages had an impaired DNA-damage response pathway, which was associated with elevated ROS and cytokine production, particularly of IL-1β and IL-18. As both these cytokines are associated with inflammasome activation, it was speculated that the NLRP3 inflammasome may be involved in the augmented cardiac pathology observed in mice transplanted with Ppm1d mutant HSPCs. Indeed, it was found that inhibition of the NLRP3 inflammasome with MCC950 attenuated the worsened cardiac dysfunction and adverse cardiac remodeling observed in mice transplanted with Ppm1d mutant HSPCs. These data show that hematopoietic mutations in Ppm1d can exacerbate cardiac remodeling in a mouse model of nonischemic heart failure. Additionally, they shed light on the potential mechanisms by which activating Ppm1d mutations may exacerbate cardiovascular conditions such as nonischemic heart failure. Taken together, these findings suggest that PPM1D mutations associated with therapy-related clonal hematopoiesis may contribute to the delayed onset of heart failure observed in cancer survivors. Nevertheless, findings from this study await validation by studies in a clinical cohort of cancer survivors with delayed onset heart failure.

7.2.6. ASXl1.

Additional sex combs like 1 (ASXL1) is an epigenetic regulator that is frequently mutated in individuals with clonal hematopoiesis and myeloid malignancies (30, 35, 37). The majority of mutations in ASXL1 are frameshift or nonsense mutations occurring in the last exon and result in a premature stop codon (35–37, 385–387). Mutant ASXL1 transcripts are thought to escape nonsense-mediated decay, resulting in the production of a COOH terminally truncated protein (385). Studies suggest that truncating mutations in ASXL1 can lead to gain-of-function by enhancing the enzymatic activity of BRCA1-associated protein 1 (BAP1), a deubiquitinase that promotes H2AK119 deubiquitination (388, 389). It has also been suggested that ASXL1-truncating mutations possess dominant-negative effects and can inhibit the function of wild-type ASXL1 (385, 388). These observations highlight the importance of using mice with Asxl1 truncating mutations rather than Asxl1 deficiency in studies of clonal hematopoiesis. Several studies have shown that transgenic mice that express murine versions of Asxl1 truncating mutations display hematological abnormalities (390), some of which have reduced numbers and functions of HSPCs (391–393). This raises questions about how ASXL1 mutations confer a competitive advantage in the setting of clonal hematopoiesis. A recent study in mice observed that Asxl1 mutant cells have a competitive advantage only in settings of native, unperturbed clonal hematopoiesis (i.e., nontransplant conditions) and during aging through activation of the Akt/mammalian target of rapamycin (mTOR) pathway (266). This finding will be discussed in greater detail in sect. 8; however, it highlights that standard BMT approaches involving young mice may not be the most ideal approach for studying ASXL1-mediated clonal hematopoiesis. To date, very little is known about the effects of hematopoietic ASXL1 mutations on age-related diseases. Indeed, observations from clinical studies suggest that ASXL1 mutations are selectively enriched in individuals with HIV and smoking, perhaps suggesting that ASXL1 mutant clones may have a competitive advantage under these conditions (117, 178, 179). It is unknown whether ASXL1 mutations may participate in downstream pathologies observed in individuals who smoke and/or have HIV, and this will likely be an area for future investigation. Similarly, very little information exists on how ASXL1 truncating mutations affect progeny immune cells, and this is likely to be examined over the coming years.

7.2.7. mLOY.

The human Y chromosome is one of the two sex-determining chromosomes that contain a limited number of genes involved in regulating sex determination and spermatogenesis. As indicated in sect. 2.1, mLOY is the most prevalent somatic mutation in human leukocytes; however, to date has been very few studies interrogating its causal connection to human disease. The Y chromosome has long been viewed as a genetic wasteland, and it has thus been suggested that mLOY in leukocytes may be a marker for genomic instability, rather than play a causal role in disease (26, 29). However, emerging evidence suggests that genes on the Y chromosome can control the expression of at least 500 autosomal genes on leukocytes, suggesting that its loss could have downstream biological effects (394). Therefore, to examine its causal connection to age-related disease, our laboratory recently developed a mouse model of hematopoietic mLOY (143). To do this, a CRISPR-Cas9 approach was taken to ablate the Y chromosome within lineage-negative bone marrow cells from male mice. Following CRISPR-Cas9 editing, LOY cells were transplanted into lethally irradiated mice and the development of age-related pathologies was studied over the course of ∼15 months. It was found that mice transplanted with mLOY cells had significantly shorter survival than mice transplanted with control cells, which is analogous to what has been previously reported in men with mLOY (25). Serial echocardiography measurements revealed that mLOY mice developed greater levels of age-associated cardiac dysfunction, and this was associated with greater indexes of fibrosis. Consistent with the elevated cardiac fibrosis levels, the mLOY condition led to an increased left ventricular filling pressure, indicative of diastolic dysfunction, and this seemed to occur independently from any changes in blood pressure. In addition to the cardiac changes, it was found that aged mLOY mice exhibited elevated levels of pulmonary and renal fibrosis and also showed worse cognitive function than control mice. Importantly, there were few changes in any hematological parameters, consistent with what is observed in humans who harbor mLOY. Collectively, these findings suggest that mLOY may accelerate age-related conditions and that augmented tissue fibrosis could be a mechanistic hallmark of this condition.

To more closely study the relationship between mLOY and age-related fibrotic disease, we chose to follow up our investigations using a murine model of TAC (143). This model was chosen as following TAC the heart undergoes profound remodeling, characterized by hypertrophy and significant fibrosis. Moreover, it models hypertensive heart failure, which has been associated with mLOY in humans. With the use of the CRISPR-Cas9 mLOY model described above, mice were subjected to TAC at 4 weeks after BMT. In accordance with the findings in aged mice, the mLOY condition led to a greater progressive decline in cardiac function over 28 days following TAC surgery. This was associated with greater levels of cardiac interstitial and perivascular fibrosis and elevated numbers of fibroblasts within the heart.

To determine the potential cellular mechanism by which the mLOY condition was promoting a worse cardiac phenotype in the setting of TAC, mice were treated with the anti-granulocyte receptor-1 antibody (anti-Gr1) to block neutrophil and monocyte recruitment to the injured heart (143). It was found that administration of anti-Gr1 reduced the level of cardiac dysfunction and fibrosis in mLOY mice, suggesting that myeloid cells were playing a role in the pathology. To interrogate this further, single-cell sequencing was performed on FACS-isolated mLOY (or control) immune cells within the heart at 7 days following TAC. When examining the cardiac macrophage populations, PHATE analysis revealed there was a trajectory from nonactivated macrophages to a continuum of inflammatory and fibrotic macrophages. Interestingly, the fibrotic macrophage subpopulation appeared to contain a greater proportion of mLOY cells whereas the inflammatory macrophage subpopulation contained a greater proportion of control cells. Further, it was noted that mLOY promoted the enrichment of regulons specifically associated with transforming growth factor-1β (TGF-β1) signaling within the fibrotic macrophage subpopulation, while regulons associated with IL-1β signaling were downregulated. These findings were congruent with bulk RNA sequencing analysis of cardiac macrophages where gene set enrichment analysis showed that differentially expressed transcripts related to TGF-β1 signaling and binding were enriched in mLOY macrophages, again indicative of a more fibrotic phenotype. On the contrary, bulk and single-cell analysis revealed little differences in the transcriptomes of cardiac monocytes, cardiac neutrophils, or blood neutrophils between LOY and control conditions, suggesting macrophages are likely the key players involved in elevated fibrotic signaling observed in mLOY mice.

Consistent with the transcriptome analysis of LOY macrophages, it was found that there were higher levels of TGF-β1 protein and downstream SMAD2-phosphorylation in the myocardium from mLOY mice (143). To assess whether elevated TGF-β1 signaling was responsible for driving the profibrotic phenotype observed in the mLOY condition, mice that had undergone TAC surgery were treated with either a control or neutralizing antibody directed against TGF-β1. It was found that treatment anti-TGF-β1 partially blunted the accelerated cardiac dysfunction observed in mLOY mice following TAC. Concomitantly, administration of anti- TGF-β1 reversed the elevated indexes of fibrosis associated with the mLOY condition, such as the elevated fibroblast number and matrix deposition. This may suggest that therapies that target TGF-β signaling may be particularly effective for mLOY carriers experiencing diseases in which fibrosis plays a central role. Collectively, the data support the notion that the mLOY condition generates a population of macrophages with an overactive profibrotic signaling network. After infiltration into tissue, such as under conditions of aging or injury, these macrophages drive a greater fibrotic response, leading to a decline in tissue function (FIGURE 10). It is noteworthy that this is mechanistically different from CHIP, which appears to drive disease progression via proinflammatory mechanisms (101, 103, 105, 106, 195, 240). This would support the idea that hematopoietic somatic mutations may augment pro- and anti-inflammatory processes within immune cells thereby differentially affecting disease development.

FIGURE 10.

Summary of experimental work linking hematopoietic mosaic Y chromosome loss (mLOY) to cardiac fibrosis. Following injury, leukocytes migrate from the bone marrow to the heart and are involved in the pathogenic pathways leading to heart failure. Under mLOY conditions, blood-derived macrophages adopt a more fibrotic phenotype characterized by an enrichment of genes involved in transforming growth factor-1β (TGF-β1) signaling. Consequently, this leads to an elevation in cardiac fibroblasts and matrix deposition thereby augmenting cardiac fibrosis. Created with BioRender.com, with permission. HSC, hematopoietic stem cells; TAC, transverse aortic constriction.

7.3. Factors Promoting Clonal Expansion of Mutant Cells

As evident from this review, clonal hematopoiesis is associated with a wide range of age-related diseases. Moreover, it is apparent from recent experimental studies that mutations in several different driver genes play a causal role in accelerating these age-related diseases, particularly cardiovascular disease (101–103, 105, 107, 141, 153, 201, 230, 231, 343). Given the detrimental impact that clonal hematopoiesis has on numerous age-related diseases, understanding how we can prevent or minimize this is of paramount importance. As indicated above, one strategy is to target specific molecules released by the mutant immune cells that directly contribute to the disease processes. For instance, therapies targeting IL-1β may be particularly effective in reducing the effects of TET2 mutations on cardiovascular disease (101, 103, 231, 301). Another strategy could be to target the expansion of mutant clones and prevent them from reaching a pathological level. As mentioned in sect. 3.1, studies using deep error-corrected sequencing suggest that clonal hematopoiesis is almost ubiquitous by middle age (39). However, these clones only appear to expand in a subset of individuals, suggesting that there may be specific factors that promote the expansion of mutant clones. Identifying factors that promote clonal expansion involves an understanding of how the mutation affects the function of HSCs and the environmental cues to which mutant HSCs respond to. This is likely to be different between the driver mutations, as they confer different effects on HSCs. For instance, studies suggest that TET2 loss-of-function in HSCs leads to increased self-renewal and myeloid differentiation (287, 338), whereas mutations in DNMT3A result in increased HSC self-renewal at the expense of differentiation (213, 317). Furthermore, mutations in TP53 and PPM1D appear to provide mutant clones with intrinsic resistance to cell stressors that trigger DNA damage response pathways, such as radiation and certain chemotherapies (135, 137, 139, 141, 215). While epidemiological studies have provided some clues as to what may promote the expansion of some clones, the descriptive nature of these studies makes it difficult to discern causality and directionality. Recent experimental studies have attempted to identify some of the factors that promote the expansion of clones with driver mutations, although progress in this area has been limited, except for the unique case of therapy-related clonal hematopoiesis, which is detailed in sect. 3.3, sect. 7.2.4, and sect. 7.2.5.

To understand what drives age-related clonal hematopoiesis, it is imperative to understand the changes that occur to HSCs and the hematopoietic system during aging. These changes can broadly be classified as HSC intrinsic (i.e., those directly affecting the HSC) and HSC extrinsic (i.e., those affecting the surrounding environment), although these are not mutually exclusive and generally feed into one another (395, 396). Regarding HSC intrinsic age-related changes, studies indicate that aged HSCs have a diminished capacity for self-renewal, impaired homing, and skewed differentiation toward the myeloid lineage (397–400). Moreover, aged HSCs exhibit numerous changes at a molecular level, such as altered signaling pathways, increases in oxidative stress, DNA damage, epigenetic drift, telomere shortening, and mitochondrial dysfunction, which all contribute to the demise of HSC function (395, 396, 401). Extrinsic changes generally involve the bone marrow niche and include vascular remodeling, endosteal degeneration, alterations in sympathetic innervation, inflammation, and changes to stromal support cells (395, 402–404). It is also noteworthy that age-related diseases and exposure to disease risk factors can affect HSC function and the bone marrow microenvironment (405–409), indicating the complex interplay between disease and hematopoietic function. It is likely that some driver mutations provide HSCs with a competitive advantage by rendering them more resistant to intrinsic age-related changes compared to wild-type cells, whereas other mutations provide the HSC with superior survival, proliferation, and/or differentiation in the aged bone marrow niche or some combination of these two scenarios. However, there is a paucity of mechanistic studies directly supporting these concepts.

Given the connection between aging, clonal hematopoiesis, and inflammation, it has been hypothesized that the expansion of mutant clones may be promoted by inflammatory conditions. Indeed, inflammatory signals have numerous effects on HSC function and can produce changes to the bone marrow niche, which could conceivably favor growth of mutant clones (410–413). For more extensive reviews of the impact of inflammation on the HSC behavior and the bone marrow niche, see Refs. 410, 411, 413. Recent experimental studies have suggested that an inflammatory environment may favor the expansion of Tet2 mutant clones. In particular, it has been reported that Tet2-deficient HSPCs have a survival and proliferative advantage in response to inflammatory stress, after exposure to TNF-α or IL-6 (176, 414). It has also been documented that Tet2-deficient hematopoietic cells produce increased levels of IL-6 and that this increased IL-6 production leads to the upregulation of the long noncoding RNA Morrbid in Tet2-deficient HSPCs to increase their survival (414). This finding perhaps indicates that Tet2 mutant cells promote their own expansion by IL-6 signaling pathways. Moreover, in the setting of myeloid malignancies, it has been suggested that inflammatory signaling can drive malignant transformation of Tet2-mutant clones (226). In particular, it has been documented that bacterial translocation from the gut to lymphoid tissues and the consequent activation of IL-6 signaling pathways promotes the development of myeloproliferative disease in Tet2-deficient animals. Nevertheless, this study was not performed in a model of clonal hematopoiesis, and the majority of individuals with clonal hematopoiesis do not develop myeloproliferative disease, raising questions about whether this finding could apply to age-related clonal hematopoiesis.

In addition to mutations in TET2, studies suggest that inflammation may promote DNMT3A mutant clone expansion. Of note, a study by Liao et al. (415) used a competitive bone marrow transplant approach to examine the dynamics of Dnmt3a^R878H mutant cell expansion in young and aged recipient mice. It was noted that Dnmt3a^R878H hemizygous mutant cells had a greater propensity to expand and differentiate into blood cells when transplanted to aged recipient mice, suggesting that the aged environment may accelerate clone growth. A cytokine array comparing young and aged bone marrow found that TNF-α was elevated in the aged bone marrow, perhaps suggesting that augmented TNF-α contributes to mutant cell expansion. Subsequent analyses found that Dnmt3a^R878H mutant cells were resistant to TNF-α-induced cell death and mutant HSPCs maintained their self-renewal capacity following exposure to TNF-α, unlike wild-type cells. Furthermore, Dnmt3a^R878H mutant HSPCs had reduced expression of genes involved in necroptosis, such as receptor interacting serine/threonine kinase (RIPK) 1 and RIPK2, and treatment with a necrosis inhibitor negated the difference in survival between mutant and wildtype HSPCs. Collectively, these findings suggest that the Dnmt3a^R878H mutation provides HSPCs with a survival and self-renewal advantage to inflammatory stress, by downregulating molecules involved in necroptosis. Consequently, Dnmt3a^R878H mutant cells may have a competitive advantage in the aged bone marrow niche, whereby levels of inflammation are higher (415, 416). Similarly, a recent study found that inflammation generated as a result of chronic infection can promote the expansion of Dnmt3a-deficient HSCs (175). With the use of a competitive BMT model in mice, it was observed that chronic Mycobacterium avium infection promoted HSPC expansion over an 8-week time period. Previous work by that same group had shown that Mycobacterium avium infection upregulates IFN-γ signaling pathways, promoting HSC terminal differentiation at the expense of self-renewal, ultimately leading to HSC depletion (417). Treatment with IFN-γ phenocopied the HSC expansion observed during infection with Mycobacterium avium in mice transplanted with Dnmt3a-deficient bone marrow, perhaps indicating a role for IFN-γ (175). Importantly, mice transplanted with Dnmt3a and IFN-γ receptor double-knockout bone marrow showed no expansion following infection, supporting the notion that IFN-γ signaling plays a role in the observed phenotype. Mechanistically, it was found that Dnmt3a-deficient HSCs had an improved survival to secondary stress but reduced differentiation compared to wild-type HSCs following exposure to IFN-γ. Moreover, serial replating assays suggested that IFN-γ reinforced self-renewal of Dnmt3a-deficient HSCs. Epigenetic profiling found that infection leads to considerable differences in methylation between wild-type and Dnmt3a-deficient HSCs. Of note, Batf2, Jun, and Fos, which are involved in HSC differentiation, were hypermethylated in Dnmt3a-deficient cells, resulting in transcriptional repression of these genes. Overall, this study suggests that IFN-γ signaling, as a consequence of chronic infection, provides Dnmt3a mutant cells with a competitive advantage due to their defects in differentiation, resistance to secondary stress, and increased self-renewal capacity, allowing for their clonal expansion overtime. Collectively, the studies described above suggest that mutations in DNMT3A may provide HSCs with a competitive advantage in an inflammatory environment thereby promoting their expansion. These findings may, at least in part, explain the large frequency of hematopoietic DNMT3A mutations observed in older individuals (35, 37), who likely have increased levels of inflammation. Although further studies are needed to confirm these findings, particularly in human cohorts.

In addition to TET2 and DNMT3A, other studies have attempted to understand what promotes clone growth of other driver mutations. A study using a zebrafish model of clonal hematopoiesis suggested that asxl1 mutant clones may also have enhanced fitness during inflammatory stress (233). The authors of this study documented that mutant myeloid cells have enhanced expression of proinflammatory genes and thereby promote inflammation. The authors suggested that asxl1 mutant progenitor cells are able to resist this inflammatory stress by the upregulation of genes that suppress inflammation, such as nuclear receptor subfamily 4 group A member (nr4a1). However, the precise inflammatory signals that drive asxl1 clone dominance in this setting remain speculative. Moreover, an independent study in mice suggested that Asxl1 mutant cells, slowly expand over time only in the setting of native clonal hematopoiesis (266). More specifically, it was observed that hematopoietic Asxl1-truncating mutations had a competitive disadvantage following transplantation, suggesting that the mutation may confer a reduced repopulation potential. As noted previously, the knockin of an Asxl1 truncating mutation resulted in a dysfunctional HSPC with overactive Akt/mTOR signaling. This resulted in augmented cell cycling and the accumulation of DNA damage, contributing to mutant cell expansion over time, an effect that could be abrogated with the mTOR inhibitor rapamycin. Exactly how Asxl1 mutant cells can evade eventual exhaustion in these circumstances remains unknown and should be an area for future investigation. Finally, findings from our laboratory suggest that myocardial injury can promote JAK2^V617F expansion, although the precise mechanisms by which this disease setting promotes clone expansion remain largely unexplored (107).

8. CONCLUDING REMARKS

We are only beginning to understand that organs within the human body are composed of a patchwork of cells with different genotypes. This appears to be an inevitable consequence of human aging where mutations accumulate over the course of an individual’s life. We can also appreciate that within each tissue there is a unique mutational landscape, which is shaped by numerous factors such as cell proliferation rate, exposure to endogenous and exogenous mutagens, organ structure, cellular injury processes, and likely many unidentified factors (17, 18, 58, 67, 71, 81). We have a limited understanding of why some tissues are prone to malignancy despite a low percentage of driver mutations, while others are more resistant despite higher percentages of mutations (35, 64, 71). This may, in part, reflect the nature of the driver mutations within the tissue or perhaps be due to differences in the sampling/sequencing methodology used to characterize different tissues. Recent findings from esophageal tissue provide some clues, suggesting that highly fit mutant cells can outcompete malignant cells (86); nevertheless, further research is needed to better understand this paradox. While driver mutations are often viewed as deleterious, in some instances they may be protective, such as in liver cirrhosis (68). In many organ systems, we have very limited knowledge of the effect of driver mutations on age-related disease and how they contribute to disease pathogenesis. This will undoubtedly be an area for future research. However, the hematopoietic system is an exception to this, and research over the past 5 years has improved our understanding of how clones with driver mutations, i.e., clonal hematopoiesis, can contribute to disease. This understanding has likely been driven by advancements in sequencing technology, the availability of large biobanks with extensive medical histories, refinement of computer-based algorithms for analysis, and the development of animal models that faithfully recapitulate the human condition of clonal hematopoiesis (32, 33, 39, 40, 101, 103, 111, 122, 123, 230). Moreover, the notion that clonal hematopoiesis likely contributes to “inflammaging,” as well as acceleration of human aging and disease, has attracted research attention to this area (101, 103, 107, 208, 209, 230, 231). Despite the progress in this area, much additional work is still required to understand the contribution of clonal hematopoiesis to human health and disease. For example, a large portion of clonal events within the hematopoietic system remain unaccounted for, suggesting that there may be additional driver mutations that contribute to this phenomenon (30, 37, 46). While clonal hematopoiesis has been associated with a wide range of human diseases, the causality of these relationships, in many cases, remains uncertain. Thus follow-up studies using animal models will be useful for improving our understanding of these relationships. On this point, we still have an incomplete understanding of how clones with different driver mutations mechanistically contribute to disease pathogenesis, and recent research suggests that this may vary depending on the driver mutation and the disease under study (101, 102, 107, 343, 348). A precise understanding of how clones promote disease may pave the way for the development of personalized therapies for patients with specific driver mutations in their blood cells (101, 301). Moreover, a better definition of the level at which clones start to become pathological in disease settings will be important to know, particularly for understanding an individual’s risk of disease (43). While it may be possible to prevent the expansion of clones, we still have an extremely limited understanding of what dictates clone growth, and this is evidently an area for future investigation. It is the hope that research directed at these outstanding questions will not only provide us with a deeper understanding of the consequences of clonal hematopoiesis and other forms of somatic mosaicism on age-related disease but also help us to develop strategies to clinically intervene and limit disease.

GRANTS

This study was supported by National Heart, Lung, and Blood Institute Grants HL139819 and HL141256, National Institute on Aging Grant AG072095, and American Heart Association Postdoctoral Fellowship 20POST35210098.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.A.E. prepared figures; M.A.E. and K.W. drafted manuscript; M.A.E. and K.W. edited and revised manuscript; M.A.E. and K.W. approved final version of manuscript.