2022 Jeffrey M. Hoeg Award Lecture: Genomic Aging, Clonal Hematopoiesis, and Cardiovascular Disease

Abstract

Chronologic age is the dominant risk factor for coronary artery disease but the features of aging promoting coronary artery disease are poorly understood. Advances in human genetics and population-based genetic profiling of blood cells have uncovered the surprising role of age-related subclinical leukemogenic mutations in blood cells, termed ‘clonal hematopoiesis of indeterminate potential,’ in coronary artery disease. Such mutations typically occur in DNMT3A, TET2, ASXL1, and JAK2. Murine and human studies prioritize the role of key inflammatory pathways linking clonal hematopoiesis with coronary artery disease. Increasingly larger, longitudinal, multi-omics analyses are enabling further dissection into mechanistic insights. These observations expand the genetic architecture of coronary artery disease, now linking hallmark features of hematologic neoplasia with a much more common cardiovascular condition. Implications of these studies include the prospect of novel precision medicine paradigms for coronary artery disease.

Graphical Abstract

For at least a century, cardiovascular disease, primarily due to coronary artery disease (CAD), remains the leading cause of death in the United States.¹ While significant public health and biomedical advances curbed alarming trends in the early 20^th century in the United States, CAD complications overtook communicable diseases as the leading cause of premature deaths among adults worldwide within the last two decades.² Modern paradigms in cardiovascular disease prevention include risk factor modification, with a particular focus on blood pressure, cholesterol, and glycemic management as well as lifestyle modification including tobacco cessation, physical activity, a healthy diet, and sleep. Furthermore, individuals estimated to be at least elevated intermediate risk for CAD are candidates for pharmacologic cholesterol-lowering. Despite the overall benefits of this approach, CAD-attributable deaths remain unacceptably high, and medical costs are rising.³

Aging and Coronary Artery Disease

While age is the most important risk factor in CAD risk prediction,⁴ the mechanisms of aging promoting CAD risk are not well understood.^5,6 The demographic shifts toward an older mean age in addition to larger overall population sizes are anticipated to yield substantial effects on overall healthcare costs.³ A sizeable fraction of this cost is due to a corresponding increase in the prevalence of cardiovascular diseases and related risk factors.

Much work has focused on molecular features often observed with advanced age – cellular senescence, autophagy, oxidative stress, and epigenetic changes. Vascular remodeling, as a consequence of these features, is well documented leading to endothelial dysfunction and arterial stiffness. While such features are also invoked in other conditions such as heart failure with preserved ejection fraction and valvular calcification, disentangling the key causal features suitable for therapeutic modulation remains elusive.

A principal genomic feature of aging is telomere shortening. Each subsequent round of somatic cellular division leads to the progressive removal of telomere sequence.⁷ Critical replicative shortening leads to chromosomal instability and cellular senescence. Hutchinson-Gilford Progeria Syndrome (HGPS), caused by mutations in LMNA, is a premature aging syndrome characterized by premature cellular senescence and accelerated telomere shortening. In a transgenic HGPS mouse model, addressing telomeric dysfunction improved lifespan consistent with an important role in cellular senescence and longevity.⁸ In the population, shorter leukocyte telomere lengths are associated with a wide range of cancer and cardiovascular outcomes.⁹ The causal relationships across these various outcomes remain incompletely known, however.

Accumulation of mutations is another common feature of aging and is correlated with accelerated telomere shortening. A prior study showed that individuals with a greater burden of coronary atherosclerosis had more evidence of DNA damage in circulating blood cells.¹⁰ Furthermore, mice genetically predisposed to greater dysfunction of DNA repair genes have early-onset aging features including premature vascular stiffness.¹¹ However, the extent to which mutational burden reflects a cause or consequence of other aging processes is not fully understood.

Circulating leukocytes derived from hematopoietic stem cells are necessary features of atherosclerosis and related events.¹² Like other stem cells, hematopoietic stem cells undergo serial mitotic divisions across the lifespan. Cellular machinery to repair the consequences of genotoxic exposures and stochastic influences resulting in mutations becomes less efficient with age.¹³ Hematopoietic stem cells have restricted self-renewal potential progressing ultimately toward exhaustion largely driven by intrinsic factors.^14-17 Whether age-related genomic alterations to hematopoietic stem cells influence risk for, or are merely correlated with, CAD is not currently clear.¹⁰

Identifying Clonal Hematopoiesis

The presence of a mutation influencing the fitness of a hematopoietic stem cell clone leading to a survival advantage is an established cause of hematologic malignancies, and a hallmark of cancers generally.^18-20

Initial applications of X chromosome inactivation (XCI) assays provided initial evidence for the existence of clonal hematopoiesis, or hematopoietic stem cell clones with a favorable survival advantage in the absence of other clinical hematologic conditions. As only one X chromosome is randomly expressed in the somatic tissues of women, patterns of XCI of X chromosome-expressed genes may identify potential mosaicism. Initial studies showed that XCI skewing occurred more commonly in blood compared to other tissues^21,22 and was also more common in older versus younger women.^23,24 Whole exome sequencing of older women with XCI skewing identified recurrent mutations in TET2.²⁵ TET2 is an established myeloid cancer-causing gene and the observed large variant allele fractions among the women studied demonstrated that a survival advantage had been conferred but neoplasia was absent.

Cross-sectional next-generation sequencing of acute myeloid leukemia (AML) blasts, other leukocytes, and hematopoietic stem cells described how single and cooperating initiating ‘driver’ mutations, often occurring randomly and sometimes in response to clear exogeneous factors, are necessary features for a founding clone to ultimately become a blast.^19,26-28 Given the presence of initiating mutations among pre-leukemic hematopoietic stem cells among those with leukemia as well as older women with clonal hematopoiesis, a proposed sequential somatic mutational model of acute myeloid leukemia emerged.²⁹

Large-scale population-based whole exome sequence analysis of blood DNA over the past decade permitted the estimation of the prevalence of clonal hematopoiesis.^30,31 Repurposing exomes generated principally to discover the heritable bases of other conditions, investigators in 2014 uncovered the presence of mutations recurrently observed in hematologic malignancies in a detectable fraction of the blood indicative of clonal hematopoiesis.^32-34 Germline sequence variants, compared to the reference genome, are expected to have variant allele fractions (VAFs) of approximately 50% or 100% for heterozygous or homozygous genotypes, respectively. Sites indicative of a somatic variant will typically have VAFs between 0 and 50% of the overlapping short-reads or greater in the setting of copy-neutral loss of heterozygosity. The technical sensitivity of standard-coverage whole exome sequencing is VAF >2%, meaning that >4% of blood cells harbor a heterozygous somatic variant indicative of mosaicism.²⁹ Because CHIP variants are not inherited and recurrently observed in hematologic malignancy and are believed to be ‘drivers,’ we infer the observed mosaicism is due to clonal proliferation.

Across all three 2014 studies, there was a similar age-related prevalence of mutations, most commonly in DNMT3A, TET2, and ASXL1.^32-34 This pattern has generally been observed in subsequent studies in unselected populations.^35,36 Approximately 1 in 10 asymptomatic adults aged greater than 70 years old have CHIP. DNMT3A encodes DNA methyltransferase 3 alpha, which catalyzes de novo 5-methylcytosine methylation. DNMT3A plays an important role in embryonic and hematopoietic stem cell differentiation. Mutations in DNMT3A yielding CHIP representing up to 40% of CHIP and are generally believed to be loss-of-function. However, the p.Arg882 hotspot mutation is known to be dominant-negative.³⁷ TET2 encodes ten-eleven translocation 2, which converts 5-methylcytosine to 5-hydroxymethylcytosine indirectly promoting DNA demethylation.³⁸ TET2 plays an important role in hematopoietic stem cell self-renewal and lineage commitment as well as monocyte terminal differentiation.³⁹ In cancer cell lines, 5-hydroxymethylcytosine sites are enriched for DNA mutations indicating a potential role of TET2 in DNA damage response as well.⁴⁰ As a classic tumor suppressor gene like DNMT3A, pathogenic TET2 mutations are generally believed to be loss-of-function. ASXL1 encodes additional sex combs like transcriptional regulator 1. ASXL1 is similar to Drosophila Asx, which encodes a chromatin-binding protein important in determining segment identity. ASXL1 belongs to the enhancer of trithorax and polycomb family, and binds to and modifies chromatin via H3K27 methylation, with dual roles in transcriptional activation and repression.^41-43 Pathogenic mutations in ASXL1 are almost all in the 12^th exon. Approximately half of the pathogenic ASXL1 mutations are from a guanine duplication (c.1934dupG) leading to a frameshift (p.Gly646TrpfsX12). The resultant C-terminal truncated ASXL1 protein is expressed in leukemia cell lines with evidence indicating dominant-negative effects, with decreased H3K27 methylation,⁴⁴ and gain-of-function effects, with enhanced BRCA1 associated protein (BAP1) activity and recruitment of BET bromodomain-containing protein 4 (BRD4) to promote transcriptional activation.^45,46 Other genes implicated in CHIP have important roles in cell cycle regulation, splicing, and DNA damage response.³⁰

Unsurprisingly, CHIP mutations are associated with an increased risk for incident hematological malignancy. While the relative risk is ~11-fold, this translates to an absolute risk of 0.5% incidence/year since hematologic malignancy is relatively uncommon in the general population.³³ Individuals with larger CHIP clones, indicated by VAF >10%, carried roughly double this risk. The relatively low prevalence of CHIP, very low incidence rate of hematologic malignancy, and current sample sizes in unselected cohorts currently limit gene-specific and variant-specific penetrance estimates for cancer risk.

Relationship Between Clonal Hematopoiesis and Coronary Artery Disease

One study in 2014 first provided the hypothesis that CHIP may be associated with increased CAD risk.³³ CHIP is associated with a 40% relative increase risk of all-cause mortality but insufficiently explained by the modest absolute cancer risk.^32,33 Since cardiovascular disease is the leading cause of death, investigators tested whether CHIP was associated with an increased risk for cardiovascular disease. Indeed, exploratory analyses among a subset without prevalent cardiovascular disease, showed an association of CHIP with increased risk for incident CAD and incident ischemic stroke.³³

Additional human studies subsequently supported the association between CHIP and increased risk for incident CAD (Figure 1).^35,47 In a nested case-control study of incident CAD cases after exome sequencing and matched controls, older individuals with coronary artery disease were nearly two-fold enriched for CHIP variants compared to matched controls. The association was confirmed in prospective age- and comorbidity-adjusted association analyses with incident CAD in larger datasets.^35,48,49 Individuals with prevalent early-onset myocardial infarction were four-fold enriched for CHIP variants compared to matched controls, including older co-morbidity-matched controls in one cohort.⁴⁷ Furthermore, older individuals with or without known CAD had a greater burden of subclinical coronary atherosclerosis as estimated by coronary artery calcium scoring from non-contrast cardiac CT.⁴⁷

Figure 1. Dynamic genomic model of coronary artery disease.

Large-effect single rare disruptive coding variants, such as LDLR putative loss-of-function variants, are a well-recognized monogenic factors for coronary artery disease. Over the last decade, advances in common trait / common variant genetics continue to refine the polygenic contribution of coronary artery disease, as estimated through coronary artery disease polygenic risk scores. More recently, population-based next-generation sequencing studies that use blood DNA estimated the age-related prevalence of clonal hematopoiesis of indeterminate potential, as indicated by the presence of clonally expanded leukemogenic variants among asymptomatic adults. Subsequent studies then demonstrated the contribution of clonal hematopoiesis of indeterminate potential to coronary artery disease. Figure made with BioRender.com.

Genes implicated in CHIP have diverse functions, and emerging evidence unsurprisingly indicates that CAD effects are varied. Given the overall relatively lower prevalence of CHIP and current sample sizes, disentangling gene-specific prognosis has been challenging. Initial reports described the strikingly greater CAD risks of JAK2 p.V617F, one of the most common CHIP mutations but comprising a very small fraction of CHIP.⁴⁷ While DNMT3A mutations are overall most common, their CAD effects may be more moderate compared to TET2 mutations.^35,48-50 Larger datasets are necessary to resolve gene- and mutation-specific effect estimates. Nevertheless, larger VAF is a key stratifier of CAD risk beyond the gene and mutation as it is for hematologic cancer risk as well.

Individuals with lower-risk myelodysplastic syndrome (MDS), with generally similar features as CHIP, also carry greater CAD risk. As individuals with CHIP do not otherwise have clinically detectable cytopenic abnormalities, they are generally clinically unrecognized by current clinical practice.²⁹ MDS reflects a clinically significant intermediate condition between CHIP and acute myeloid leukemia (AML). Many of the mutations observed in CHIP are thus observed among patients evaluated for MDS. Recent studies extended CHIP studies demonstrating that individuals with MDS at lower risk for AML transformation also carry a greater risk for cardiovascular disease.^51,52 Generally, cytopenias are correlated with increased cardiovascular disease risk⁵³ but the extent to which clonal hematopoiesis driver mutations may further stratify this risk isn’t well understood. Furthermore, a recent study observed that among patients intensively treated for AML, those with driver mutations typically seen in CHIP versus those without had a significantly greater risk for cardiovascular events.⁵⁴

Mosaic chromosomal alterations (mCAs) are not associated with increased CAD risk. They are large structural variants detectable at appreciable VAFs indicative of somatic mosaicism are increasingly prevalent among older adults and are similarly associated with excess hematologic malignancy and mortality risks.^55-57 Excess mortality risks attributed to mCAs may be related to excess severe infection risks and not related to greater CAD risks.^58,59 Generally, mCAs are associated with incident lymphoid leukemias and CHIP mutations are associated with incident myeloid leukemias providing insights into reasons why non-oncologic prognoses may be distinct.^58,60 Nevertheless, the co-occurrence of both CHIP mutations and mCAs yields greater leukemia and cardiovascular mortality risks.^61,62 Since the mCAs observed with CHIP mutations yielding greater cardiovascular risk are generally observed in myeloid leukemias, they could reflect synergistic or local mutational processes.⁶¹

Mechanisms Linking Clonal Hematopoiesis with Coronary Artery Disease

As the links between CHIP and CAD was first suggested in 2014 in humans with confirmation and recapitulation in experimental models in 2017, the mechanistic links between CHIP and CAD are recently increasingly becoming understood.

Tet methylcytosine dioxygenase 2 (TET2)

Although TET2 is the second most commonly implicated gene in CHIP (second to DNMT3A), initial studies focused on TET2 given its stronger age-independent association with CAD.

Murine studies of hematopoietic Tet2 deficiency corroborate human associations with CAD. Hematopoietic Tet2 disruption in mice leads to enhanced hematopoiesis with myeloid skewing along with a competitive advantage over wild-type hematopoietic stem cells.^38,63,64 Atherogenic mice with full or partial bone marrow-specific deficiency of Tet2 have a greater burden of atherosclerosis.^47,65 Furthermore, plaque sizes remained greater when Tet2 deficiency was restricted to myeloid cells.⁶⁵

Experimental studies of hematopoietic Tet2 deficiency strongly implicate NLRP3 inflammasome activation in CAD pathogenesis (Figure 2). A major driver of Tet2-driven leukemic transformation is through widespread aberrant DNA methylation.^66,67 However, much focus on CAD mechanistic investigation has stemmed from the observation that Tet2-deficient murine macrophages also have greater secretion of IL-1B and IL-6.^47,65 CRISPR-mediated inactivation of Tet2 in hematopoietic stem cells similarly yielded greater expression of IL-1B and IL-6.⁶⁸ Murine studies show that Tet2 endogenously represses IL-6 transcription during inflammation via Hdac2 recruitment and resultant histone deacetylation independent of methylation effects.⁶⁹ IL-1B and IL-6 are well-described downstream mediators of the NLRP3 cascade and a large body of evidence links them to atherosclerosis prior to the description of CHIP’s connection to CAD.¹² Ldlr−/− mice with transplanted Tet2-deficient bone marrow compared to atherogenic compared to Ldlr−/− mice with transplanted wild-type bone marrow exhibit a greater reduction in atherosclerotic plaque when administered an NLRP3 inhibitor.⁶⁵

Figure 2. The NLRP3 inflammasome connects TET2 clonal hematopoiesis of indeterminate potential with coronary artery disease.

Multiple lines of evidence support this connection. (1) Atherogenic mice with a subsequent of hematopoietic stem cells deficient of Tet2 versus wild-type have greater reduction of supravalvular aortic atherosclerosis from NLRP3 inflammasome inhibition versus placebo. The activated NLRP3 inflammasome leads to a cascade of enhanced IL-1B and IL-6 signaling, and was connected to atherosclerosis previously but only recently with greater relevance in this context. (2) The common IL6R p.Asp358Ala variant was previously associated with a modest reduction of coronary artery disease odds in the general population. In the UK Biobank, individuals with TET2 or DNMT3A clonal hematopoiesis of indeterminate potential (CHIP) compared to those without had a great reduction in cardiovascular disease event rates with IL6R p.Asp358Ala versus wild-type. (3) The CANTOS trial previously showed that individuals with coronary artery disease and increased high-sensitivity C-reactive protein had a moderate reduction in major adverse cardiovascular disease events when given canakinumab (monoclonal antibody directed at IL-1B) versus placebo. In a non-prespecified post hoc analysis of CANTOS, individuals with TET2 CHIP appeared to have a much larger reduction in major adverse cardiovascular disease events from canakinumab vs placebo compared to others. Figure made with BioRender.com.

Observations in humans are consistent with the hypothesis that the NRLP3 inflammasome inhibition may yield greater clinical cardiovascular benefits among individuals with TET2 CHIP. C-reactive protein (CRP), the acute phase reactant, is modestly greater among those with CHIP although not a reliable biomarker to prioritize individuals with CHIP, including TET2 CHIP.^35,36,47 However, individuals with TET2 CHIP indeed have greater concentrations of circulating IL-1B and IL-6.³⁶ Among individuals with CAD and CRP > 2mg/L, canakinumab (IL-1B targeting monoclonal antibody) versus placebo led to 15% relative reduction in cardiovascular disease risk, but there was a greater risk of fatal infection. In a non-prespecified post hoc analysis of this trial, individuals with TET2 CHIP had a 62% relative reduction in cardiovascular disease risk from canakinumab versus placebo.⁵⁰ Analogously, the common germline variant IL6R p.Asp358Ala is associated with a 5% reduction in CAD risk in the population.⁷⁰ Among individuals with CHIP, the presence of IL6R p.Asp358Ala was associated with a 54% reduction in cardiovascular event risk.³⁵ Such preclinical data are supportive of prospective trials estimating the cardiovascular effect of NLRP3 inflammasome inhibition among individuals with TET2 CHIP.

Hypermethylation from TET2 CHIP may also play an important role in CAD risk as well. “Epigenetic age acceleration” occurs when methylation-based predicted age exceeds chronological age, and is associated with increased CAD and all-cause mortality risks.⁷¹ CHIP is associated with greater epigenetic age acceleration, with even greater acceleration for TET2 compared to DNMT3A CHIP.⁴⁸ Interestingly, epigenetic age acceleration stratified CAD and all-cause mortality risk greater among individuals with CHIP versus without. Furthermore, the aforementioned IL6R p.D358A interaction (i.e., greater CAD risk reduction) was particularly notable among individuals with TET2 CHIP and heightened epigenetic age acceleration. Widespread hypermethylation from Tet2 deficiency of hematopoietic stem cells was observed in mice and among humans with TET2 CHIP.^72,73 In the latter study, increased methylation at three CpG (near RGS12, DOCK9, and LRP1) supported potentially new causal mechanisms between TET2 CHIP and CAD using Mendelian randomization procedures.⁷³

Vitamin C treatment was proposed as a strategy to ameliorate hypermethylation from TET2 loss-of-function mutations. Exogenous vitamin C promotes demethylation in stem cells, particularly in a TET-dependent manner.^74,75 In cancer cell lines, vitamin C treatment mimics Tet2 restoration and the resultant transient DNA damage response may sensitize the cells to specific chemotherapeutic regimens.⁷⁶ Small trials assessing the role of vitamin C in the hematologic management of patients with TET2 myelodysplastic syndrome or myeloid malignancies are currently ongoing. Whether this has implications for TET2-associated CAD risk is currently not well understood.

DNA methyltransferase 3 alpha (DNMT3A)

While DNMT3A is the most common gene implicated in CHIP, the association of DNMT3A mutations with CAD is less robust. As such, functional studies investigating their connection to CAD are less clear as well.

DNMT3A plays an important role in the biology of myeloid cells. DNMT3A is highly expressed in macrophages.⁷⁷ Murine hematopoietic stem cells with Dnmt3a deficiency exhibit myeloid skewing.⁷⁸ Murine hematopoietic stem cells with Dnmt3a deficiency have impaired production of type I interferons via deficiency in Hdac9 expression.⁷⁷ However, type I interferons have been implicated in promoting atherosclerosis.⁷⁹

Inflammatory cytokines are also implicated in DNMT3A CHIP. CRISPR-mediated disruption of Dnmt3a in murine macrophages promoted the expression of IL6, but also Cxcl1, Cxcl2, and Ccl5.⁶⁸ Among patients with DNMT3A CHIP and heart failure or aortic stenosis, transcriptomics analysis of peripheral mononuclear cells showed greater gene expression of IL6, IL1B, and NRLP3 as well as macrophage inflammatory proteins. Furthermore, interferon-gamma promotes the expansion of Dnmt3a−/− hematopoietic stem cells in mice.⁸⁰

While DNMT3A CHIP is associated with widespread hypomethylation, it leads to a similar phenotype of hematopoietic stem cell self-renewal and proliferation as TET2 CHIP.^72,73 Mendelian randomization analyses suggest that DNMT3A CHIP-promoted hypomethylation at some sites may promote the development of CAD.⁷³ More recent single cell multi-omics analysis of individuals with the DNMT3A p.R882 hotspot mutation more specifically characterizes methylation changes attributed to this mutation.⁸¹

Janus Kinase 2 (JAK2) p.V617F

JAK2 p.V617F is identified in the vast majority of patients with myeloproliferative neoplasms.^82-84 Polycythemia vera, a subset of myeloproliferative neoplasms where JAK2 p.V617F is virtually always present, is associated with a greater risk for atherothrombotic complications.⁸⁵ In the context of CHIP, JAK2 p.V617F is the most common recurrently observed mutation and is associated with the greatest CAD risk.⁴⁷

Mice with conditional knock-in of Jak2 p.V617F have a greater prevalence of spontaneous thrombosis compared to Jak2 wild-type.⁸⁶ Neutrophils of patients with myeloproliferative neoplasms had greater neutrophil extracellular trap (NET) formation which appeared to drive thrombotic phenomena in Jak2 p.V617F transgenic mice. The thrombotic tendencies were ameliorated with ruxolitinib, a JAK inhibitor. LNK (also called SH2B3) is an adapter protein primarily expressed in hematopoietic and endothelial cells that attenuates JAK/STAT signaling via negative regulation from thrombopoietin and erythropoietin stimuli.^87-89 Lnk-deficienct mice have greater NETosis via greater exposure and release of oxidized phospholipids.⁹⁰ LNK p.R262W reduces LNK function and is associated with greater CAD risk among individuals carrying JAK2 p.V617F.⁹⁰

In addition to heightened thrombosis, atherogenic transgenic mice with Jak2 p.V617F had increased atherosclerosis compared to atherogenic transgenic mice with Jak2 wild-type.⁹¹ Jak2 p.V617F macrophages had strikingly increased IL-6 secretion as well as IL-1B secretion.⁹¹ IL-1B inhibition in atherogenic Jak2 p.V617F transgenic mice led to greater plaque stabilization features.⁹² And while ruxlotinib reduced the concentrations of the closely related IL-18, it led to increased cholesterol concentrations in the atherogenic Jak2 p.V617F transgenic mice.⁹² Motivated by observations that atherogenic Jak2 p.V617F transgenic mice had complex plaque features, investigators showed that such mice had defects in efferocytosis of apoptotic cells, which is a role of lesional macrophages to limit necrotic core formation.⁹¹ In addition to NLRP3 inflammasome activation, Jak2 p.V617F was recently shown to activate the AIM2 inflammasome further accelerating atherosclerosis development.⁹² In atherogenic Jak2 p.V617F transgenic mice, Aim2−/− but not Nlrp3−/− bone marrow led to reductions in lesion and necrotic core areas.⁹²

Risk Factors for Clonal Hematopoiesis

Several factors have been correlated with the presence of CHIP. However, the cross-sectional nature of these association analyses limits robust inferences regarding temporal associations with putative risk factors or consequences.

Several clinical cardiovascular risk factors are enriched among individuals with CHIP.³⁰ Unsurprisingly, chronological age consistently shows the most robust and strongest association with CHIP. Across studies, CHIP prevalence has been slightly but significantly greater among men compared to women.^36,47 However, women are more likely to have DNMT3A CHIP than men. Individuals with higher BMI levels generally have a greater prevalence of CHIP.^93,94 Individuals self-identifying as being of European descent are modestly enriched for CHIP, but the mechanisms underlying this observation are not well understood.^36,47 Individuals with type 2 diabetes status are strongly enriched for CHIP,^35,36,47 and murine studies suggest that Tet2 bone marrow deficiency may exacerbate insulin resistance thereby also promoting type 2 diabetes risk.⁹⁵

Exogenous exposures also influence the likelihood of observing CHIP. Independent of these factors, individuals with poor diet quality have a greater prevalence of CHIP.⁹⁶ While smoking is a strong risk factor for CHIP, smoking is strikingly strongest for ASXL1 CHIP.^97,98 Patients exposed to cytotoxic chemotherapy are enriched for mutations in DNA damage response genes, such as TP53 and PPM1D.⁹⁷ Greater sleep quality promotes clonal diversity and promotes hematopoietic stem cell epigenomic control,⁹⁹ and sleep fragmentation in atherogenic mice with partial Tet2-deficient bone marrow promotes expansion of Tet2-deficient cells.¹⁰⁶

Women with prior premature natural menopause, a previously described risk factor of cardiovascular disease, are enriched for the presence of CHIP mutations.^100,101 Since they are not enriched among women who underwent premature surgical menopause, whether the mutations prematurely occur and hasten natural menopause, are a consequence of premature ovarian failure, or reflective of organism-wide accelerated aging is not well understood.

Persons living with HIV (PWH), who are known to have heightened cardiovascular disease risk, are enriched for CHIP mutations.^102,103 Early studies indicate a greater relative enrichment of non-DNMT3A CHIP compared to unselected populations indicating the likelihood of unique selective pressures. Among PHW on stable antiretroviral therapy, CHIP was associated with low CD4 nadir and increased residual HIV-1 transcriptional activity.¹⁰⁴ Similar to non-HIV studies, CHIP among PWH was associated with a greater burden of subclinical coronary atherosclerosis.¹⁰⁵

One study recently advanced the hypothesis that atherosclerosis itself may influence CHIP fitness.¹⁰⁶ Prior genetic and diet-induced hypercholesterolemia murine models demonstrated that the hematopoietic system is activated in the context of hypercholesterolemia,¹⁰⁷ with consistent prior observations for diet-induced hypercholesterolemia in rabbits¹⁰⁸ and swine.¹⁰⁹ Furthermore, sleep fragmentation among hypercholesterolemic mice further leads to both hematopoietic system activation and atherosclerosis.¹¹⁰ Similarly, the authors of the aforementioned study showed that diet-induced hypercholesterolemia versus control diet among Ldlr−/− mice transplanted with Tet2−/− bone marrow yielded a greater Tet2−/− cell fraction.¹⁰⁶ Sleep fragmentation of these Ldlr−/− mice transplanted with Tet2−/− bone marrow also showed a greater Tet2−/− cell fraction. Tet2−/− fitness was not altered without these atherogenic stimuli. However, disentangling the Tet2−/− survival fitness from the hypercholesterolemia stimuli critically required for murine atherosclerosis versus the resultant atherosclerosis is a key limitation. Prior studies did not show Tet2−/− competitive advance hypercholesterolemia-induced atherosclerosis or ischemic heart disease murine models.^65,111,112 The authors show that humans with CAD have larger VAFs for CHIP driver mutations as well as neutral mutations consistent with greater hematopoiesis. However, this does not consider atherogenic stimuli (i.e., hypercholesterolemia, diabetes, and smoking) enriched among patients with CAD likely to influence fitness. Furthermore, the cross-sectional analyses are not inconsistent with CHIP being a risk factor for CAD, which would be consistent with the several murine and prospective human studies mentioned earlier. Additionally, varied CAD prognosis by CH mutation (i.e., CHIP drivers and mCAs) is not consistent with reverse causality. Like all other putative CHIP risk factors, longitudinal studies of CHIP profiling accounting for correlated factors influencing fitness are needed.

Germline Genetic Factors and Clonal Hematopoiesis

Germline genetic association studies offer the prospect of prioritizing causal features that may be driving CHIP. However, given the relatively lower prevalence of CHIP in the population and age-dependency indicating the strong influence of exogenous influences, germline genetic factors are not expected to be principal influencers of CHIP.

Early genome-wide association studies of CHIP identified the TERT (telomerase reverse transcriptase) locus as the lead locus.^36,113 Paradoxically, the CHIP risk allele is associated with longer telomere length although individuals with CHIP have shorter telomeres.¹¹³ A bidirectional Mendelian randomization analysis indicated that processes promoting telomere lengthening promote CHIP presence whereas CHIP presence in turn promotes telomere shortening presumably in affected cells.⁴⁹ While this exemplifies new potential pathways for CHIP-associated consequences, it also highlights important considerations for future Mendelian randomization analyses of CHIP.

Genetic association analyses across more diverse individuals identified an African ancestry-specific low-frequency variant at the germline TET2 locus associated with the increased prevalence of both TET2 and non-TET2 CHIP.³⁶ In silico-informed in vitro analyses indicated that the risk allele disrupted the TET2 promoter leading to decreased gene expression and resultant increased hematopoietic stem cell self-renewal and proliferation.

Larger genetic association studies are beginning to resolve gene-specific germline influences.⁹⁴ For example, a germline allele at TCL1A had opposing effects on TET2 and DNMT3A CHIP prevalence. In these latest studies, the heritability of CHIP is estimated to be nearly 4%.

Insights from Longitudinal Analyses

Increasing evidence supports earlier CHIP mutation acquisition with life-course experiences influencing the fitness of these mutations. Ultra-deep targeted sequencing studies indicate that CHIP mutations are virtually ubiquitous yet generally quiescent at very low VAFs among middle-aged adults.¹¹⁴ Phylogenetic analyses of individuals with myeloproliferative neoplasms indicate that mutations often decades earlier and sometimes in utero.¹¹⁵ Furthermore, single cell genomics analyses of patients with leukemias, demonstrate the subclonal diversity well beyond a hallmark driver mutation charting paths toward neoplasia.^116,117

Individuals with larger clones have greater risks for incident CAD highlighting the significance of understanding temporal trajectories.³⁵ Generally, this is indicative of greater mutational fitness as recently characterized through recent longitudinal studies of CHIP.^118-122 Given the commonality at advanced age, increasingly limited clonal diversity was posited to be a characteristic feature of normal aging.¹²⁰ Despite mutations in DNMT3A being more common among individuals with CHIP, DNMT3A clones generally grow slower compared to other driver mutations.¹¹⁸ Driver mutations in TET2 yield greater positive selection. However, interindividual variability of clonal trajectories was high among individuals with JAK2 mutations.

Refining our understanding of mutational fitness may better refine surveillance patterns as well as the timing of therapeutics. Since longitudinal sample sizes to date have been limited, the factors that alter longitudinal clonal trajectories are currently not well understood.

Links with Other Cardiovascular and Non-Cardiovascular Outcomes

Given the key role of leukocytes in human health and disease, it is not surprising that CHIP is increasingly linked to a wide range of phenotypic outcomes of varying importance. With respect to cardiovascular disease, published links supported by human incident associations with experimental corroboration include venous thromboembolic disease,⁸⁶ and ischemic and non-ischemic heart failure^112,123-125. For non-cardiovascular non-oncologic outcomes, similar links have been demonstrated for chronic obstructive pulmonary disease,¹²⁶ gout,¹²⁷ and osteoporosis.¹²⁸

CHIP has been described as identifying very high-risk subgroups among patients with severe cardiovascular disease. Prognosis is worse for patients with CHIP who have chronic ischemic heart failure,^129,130 severe aortic stenosis undergoing transcatheter aortic valve implantation,¹³¹ or cardiogenic shock.^132,133

Conclusions

Technological, computational, and infrastructural advances have uncovered the rich evolving genomic diversity across blood cells for a given individual. Not only has this led to new insights for hematologic malignancy, but such studies have led to know paradigms for CAD and increasingly other conditions. Somatic mosaicism represents an exciting new frontier for precision cardiovascular medicine.

Highlights.

• Population-based next-generation sequencing of blood DNA uncovered the age-related presence of expanded leukemogenic driver mutations indication of clonal hematopoiesis, called clonal hematopoiesis of indeterminate potential (CHIP).

• Both human and murine studies support CHIP as a new independent risk factor for coronary artery disease, with varying prognosis and mechanism by gene implicated.

• Multiple lines of evidence in human and murine studies support a critical role of NLRP3 inflammasome activation in the development of TET2 CHIP-associated CAD.

• Several questions exist regarding risk factors, modifiers of fitness, and clonal dynamics as they related to CAD outcomes. Multi-omic, single cell, and longitudinal studies are beginning to unpack these important knowledge gaps and others.