Clonal hematopoiesis and cardiovascular disease: deciphering interconnections

In 2014 Jaiswal et al. were the first to describe a statistical association between atherosclerotic events and the presence of CHIP with a hazard ratio of 2.0 (95% confidence interval [CI] 1.1–1.8) for coronary artery disease (CAD) [55]. In a subsequent study, this group later compared 4726 CHIP-patients to a control group of 3529 individuals and confirmed a 1.9-fold overall risk of developing CAD for individuals affected by CHIP. They further measured the coronary artery calcification score based on computer tomography and detected a 3.3-fold increase in CHIP carriers. In line with this, the risk of developing early onset myocardial infarction, (age men < 40 years/woman < 50 years) was four times higher among CHIP-affected individuals. Closer analysis revealed two additional parameters to modulate the cardiovascular risk. First, risk was dependent on the clonal size (VAF). Second, it was dependent on the identity of the affected gene with mutations in JAK2 leading to a 12-fold increase, whereas mutations in the most frequently affected genes TET2 and DNMT3A conferred a two-fold increase in the risk of developing atherosclerosis [54]. The association between CHIP and CAD was replicated in a cohort of postmenopausal women where the likelihood was increased by a factor of 1.36 (95% CI 1.07–1.73) [49]. Since then, further clinical data have emerged from a range of CVDs (Table 1).

In line with the overall increased cardiovascular risk, the presence of a CHIP mutation was associated with a 2.6-fold increased likelihood for ischemic stroke in the cohort analyzed by Jaiswal et al. [55]. This increased risk of total strokes was replicated in a further cohort, following adjustment for age, sex and ethnicity, with a hazard ratio of 1.14 (95% CI 1.03–1.2). Importantly, the risk for hemorrhagic stroke was increased 1.24-fold, with TET2 being the strongest risk associated gene. Overall, CHIP-bearers were prone to hemorrhagic rather than to ischemic stroke. Among ischemic strokes, TET2 again showed the strongest association, with microvascular pathology being more frequent than large-vessel strokes [11].

The outcomes of patients with heart failure (HF) has repeatedly been associated with CHIP [9, 30, 35, 56, 70, 71], with correlations resembling those in atherosclerosis. Namely, the HF risk increases with clone VAF [9, 35, 56, 95] and with the number of the acquired mutations [30].

Atherosclerosis represents the major cause of congestive HF [12] and CHIP-carriers confer a significantly increased HF-associated mortality after having experienced myocardial infarction [30, 35]. Inflammation appears to lay an important role here [61]. In mouse models lacking TET2 or DNMT3A, blockage of Interleukin-1β (IL-1b) rescued the deficiency in heart function [78, 79]. Furthermore, transcriptome analysis of HF patients with DNMT3A mutations reveal elevated expression levels of inflammatory genes [4]. Hence, one possibility might be that CHIP is associated with HF through inflammation-mediated effects on atherosclerosis. Nevertheless, in the cohort of Dorsheimer et al., most of the deaths among TET2 or DNMT3A-mutation carriers were attributed to arrhythmic events or progression of HF rather than to myocardial infarction [35]. Furthermore, a large meta-analysis including over 55,000 participants failed to identify any difference between CHIP carrying HF patients regarding prior CAD [95]. Sano et al. investigated the pathogenesis of HF upon CHIP-mutations by establishing a mouse model system. For this purpose, they infused angiotensin II to mimic hypertensive HF and observed increased cardiac dysfunction associated with lack of TET2 or of DNMT3A. Interestingly, the phenotype induced by loss of TET2 or DNMT3A was accompanied by increased cytokine release and expansion of the hematopoietic compartment [79].

Cardiogenic shock represents the most severe clinical manifestation of acute heart failure. Böhme et al. recently studied the prognostic impact of CHIP in 446 patients with cardiogenic shock in acute myocardial infarction from the CULPRIT-SHOCK randomized clinical trial [15]. CHIP variants at ≥ 2% VAF were found in 29% of the patient population, most frequently in the DNMT3A and TET2 genes. Compared to non-CHIP patients, CHIP carriers experienced worse short-term clinical outcome with regard to a composite endpoint of all-cause mortality and the need for renal replacement therapy, even after adjustment for traditional risk factors and several established biomarkers such as arterial lactate, NT pro brain natriuretic peptide (NT-proBNP), cystatin C or IL-6. In line with this, Scolari et al. observed a 1.5-fold increased prevalence of CHIP-mutations, particularly TET2 and ASXL1, among patients with cardiogenic shock predominantly of non-ischemic origin, compared to HF patients (95% CI 1.0–2.1). At the same time, harboring a CHIP mutation was associated with a decrease of survival at multiple timepoints (30-days: HR 2.7; 95% CI 1.3–5.7, P = 0.006; 90-days: HR 2.2; 95% CI 1.3–3.9, P = 0.003; and 3-years: HR 1.7; 95% CI 1.1–2.8, P = 0.01), while individuals carrying a TET2-mutation displayed elevated serum levels of SCD40L, IFNγ, IL-4 and TNFα [81].

Sequencing a cohort of 279 patients with aortic valve stenosis (AS) having undergone transcatheter aortic valve intervention (TAVI) identified another association with CHIP. Mortality between one and eight months after TAVI was increased for carriers of DNMT3A or TET2 by a factor of 3.1 (95% CI 1.17–8.08) after adjustment for sex and age, and by 4.8-fold (95% CI 1.49–15.57) after further adjustment for NT-proBNP serum levels. Interestingly, basic clinical parameters such as concomitant atherosclerotic disease, blood cell count, inflammatory markers or procedural characteristics were comparable between CHIP and non-CHIP patients [66]. Another study analyzed monocytes of AS patients carrying CHIP-mutations and detected an increased expression of inflammatory genes and mediators, including IL-1b, IL-6-receptor, NLRP3 inflammasome complex and CD163 [4].

Compared to the other common CHIP mutations, those in JAK2 carry a specific risk profile. Wolach et al. identified a 12-fold increased risk for venous thrombosis among JAK2-mutation carriers, compared to a doubled risk among the other CHIP mutations. The authors offered mechanistical insight using Jak2^V617F knockin mouse model, in which increased formation of neutrophilic extracellular traps, as a mechanism of innate immunity, promoted thrombosis [91]. However, an independent study analyzed a cohort of 61 patients with unprovoked pulmonary embolism, 12 of whom carried CHIP-mutations, and failed to detect any difference between CHIP carriers and non-CHIP carriers. Possibly due to the small number of CHIP-positive individuals, no JAK2 mutation was found [82].

Another association with CHIP, and in particular with TET2 mutation, was recently demonstrated for pulmonary arterial hypertension (PAH) within a cohort of 1832 individuals. In this case, contradictory to the aforementioned data, the identification of a TET2 mutation predicted a favorable course of disease, with disease onset being delayed and pulmonary artery pressure being lower among TET2 mutation carriers. Nonetheless, the group found in accompanying mouse experiments that depletion of Tet2 did provoke PAH and this effect was reversible by blockage of IL-1b [75].

Atherosclerosis represents a severe vascular pathology, driven by the interplay between dyslipidemia and inflammation (Fig. 2). Given the evidence discussed above, it seems likely that CHIP promotes a pro-inflammatory environment, including changes of monocyte and macrophage biology, implying that chronic inflammation might embody the portentous link between cardiovascular risk and CHIP.

This conclusion was based initially on the observation of significantly increased cytokine levels in association with CHIP mutations. A direct causal association between CHIP and atherosclerosis via inflammation was subsequently demonstrated by Fuster et al. [41], who observed increased levels of IL-1b and an apparent atherosclerotic phenotype in Tet2 mutant mice. Blockage of IL-1b diminished the arteriosclerotic phenotype, identifying inflammation as the crucial connection [42]. Amelioration of atherosclerosis and HF in Tet2 depleted mice by pharmacological inhibition of the NLRP3 inflammasome (MCC950), followed by reduction of IL-1b further confirmed this functional concept [79].

These cytokines strongly regulate monocyte rolling, adhesion and transmigration through endothelium, followed by macrophage transformation, all of which are key processes contributing to atherosclerosis [59]. Global changes in cytokine levels and skewed monocytic and macrophagic functions therefore represent a compelling elicitor for the vascular pathology.

The majority of research in this area has analyzed the effects of variants in the two genes most frequently mutated in CH: TET2 and DNMT3A. As described above, TET2-depletion resulted in increased release of the cytokines IL-1b, IL-6 and tumor necrosis factor-alpha (TNF-α) in vitro [1, 42, 51, 78], as well as in mouse models [42]. Similar observations were made for DNMT3A. For instance, CRISPR/Cas-9 mediated DNMT3A knockdown provoked elevated levels of the cytokines CXCL1, CXCL2, IL-6 and CCL5 in macrophages [78]. Furthermore, monocytes bearing DNMT3A mutations showed an increased expression of pro-inflammatory genes, including the aforementioned cytokines [3, 14, 61]. In line with these in vitro data, CHIP-carriers also displayed elevated serum levels of IL-6 [14, 28], IL-8 [54], and IL1-b [14].

While modified global cytokine levels apparently represent a major hallmark in terms of CHIP mediated pathomechanism, there also appear to be changes in specific monocytic and macrophagic functions, although this effect seems to be context specific [27]. For instance, patients with germline variants of DNMT3A, showed reduced levels of monocyte secreted IL-10 [67]. In another experiment, DNMT3A stimulated antiviral immune response of macrophages through activating histone deacetylase 9 [64]. In contrast, monocyte-function appeared to remain mainly stable in the presence of TET2 mutations [56], with the exception of the promotion of macrophage migration inhibitory factor (MIF), a pivotal factor for monocytic differentiation, that is characteristically elevated in atherosclerosis [76]. With respect to Jak2-mutants, alterations in macrophage function also appear to contribute to atherosclerosis [89]. Here, it has been shown experimentally that Jak2 mutant macrophages induce DNA replication stress, activate the AIM2 inflammasome and thereby aggravated atherosclerosis [39].

Taking the pro-inflammatory state in atherosclerosis into account, it is not surprising that elevated levels of high-sensitive C-reactive protein (hs-CRP), a non-specific indicator of inflammation, are observed [59]. However, with regards to CHIP contradictory data exist. Some studies found no association of CHIP and hs-CRP [14, 59], whereas others state a significant elevation [21]. Altogether, CHIP appears to disturb the strongly cytokine regulated equilibrium of macrophage expansion and function, which further promotes the development of atherosclerotic lesions.

Lipoprotein homeostasis is a further factor contributing to the formation of atherosclerosis. Macrophages that arise from transmigrated circulating monocytes are able to ingest oxidized or native LDL inside the atherosclerotic lesion. If this process is dysfunctional or macrophages become overloaded, they can turn into foam cells and contribute to arteriosclerotic lesion progression [59]. Just recently, CHIP was found to have an impact in this process. Dotan et al. investigated mutant Jak2 mice and macrophages. They observed attenuated cholesterol efflux from macrophages, most likely due to dysfunctional ABCA1 cholesterol transporter, while cholesterol uptake was unaffected, leading to the acceleration of atherosclerosis. Consistent with this, systemic inhibition of JAK2 with ruxolitinib provoked the same effect [36]. In large CHIP cohorts, however, no alterations of serum cholesterol and its derivates were detected [14, 36, 54, 55].

One of the major hubs to sense and regulate cellular energy and substrate supply is the ubiquitous and highly conserved process of autophagy. Macroautophagic degradation is thought to accelerate atherosclerosis at several breaking points, such as efferocytosis, lipid metabolism and oxidative homeostasis of the endothelium [62]. THP1-derived macrophages, that were incubated with Ox-LDL, showed an attenuation of their macrophagic activity in a TET2 dependent manner, and effect that was reversible by mimicking de-methylation through treatment with azacytidine [60]. Immunohistochemical staining of aortic walls in mice revealed both reduced expression of Tet2 and reduced autophagic activity upon low shear stress. Changes in the expression of TET2 were shown to regulate autophagic activity in endothelial cell cultures, with the knockdown of Tet2 being accompanied by decreased expression of endothelial NO-synthase and increased expression of endothelin-1 as a typical hallmark of endothelial dysfunction [93]. Where it is of note that a recent report of endothelial cell differentiation form circulating monocytoid cells in a sheep model raises the possibility that CHIP myeloid cells may have the potential to transdifferentiate and contribute directly to the endothelial compartment [11]. In ApoE^−/− mice, overexpression of Tet2 was followed by upregulation of autophagy reduced atherosclerotic lesions and a decrease in the expression of ICAM-1 and VCAM-1, which are crucial for monocyte invasion. Despite this, cholesterol, tri-acyl-glyceride, and lipoprotein levels remained unchanged [72]. Together, these data indicate an involvement of TET2 in the regulation of autophagy in macrophages and endothelial cells. A potential mediator in this respect is Beclin-1, which is central to the assembly of autophagosomes and the expression of which is controlled by TET2-dependent de-methylation of promotor sequences [72]. It is worth noting in this context that changes in DNA methylation-patterns have widely been associated with increased atherosclerotic risk [6, 63, 73].

In the case of JAK2, further mechanisms have been discussed. As described above, since JAK2 mutations provoke neutrophils to form neutrophil extracellular traps, this may lead to thrombosis, which could further increase CVD risk [91]. Furthermore, Jak2^V617F mice showed increased atherosclerotic lesions with complex and comparatively large necrotic cores containing iron, erythrocyte and macrophagic depositions. Reduced erythrocyte expression of CD47 and reduced levels of c-Mer tyrosine kinase (MerTK), as a key modulator of efferocytosis, suggest that impaired efferocytosis and increased erythrophagocytosis may make substantial contributions to the formation of atherosclerotic lesions [89].