Clonal Hematopoiesis, The Emergent CVD Risk Factor

In 1948, the Framingham Heart Study (FHS) was commissioned to elucidate commonalities that may contribute to cardiovascular disease (CVD).¹ Through this work, many of the core risk factors for CVD were established by the late 1970s. Despite these major advances, it has long been recognized that the conventional modifiable risk factors, incompletely account for the incidence of CVD.² In view of this, advanced age is the greatest CVD risk factor, but it is an elusive therapeutic target and has traditionally been viewed as an unmodifiable risk factor. Since the pioneering findings by the FHS, a great deal of work has been performed to better understand the molecular pathogenesis of CVD, spurred in part by the completion of the human genome project. Despite these important advances, little insight has been gained on novel and prevalent causes of age-related CVD until recent work uncovered age-related clonal hematopoiesis (CH) as a causal risk factor for CVD.

On a daily basis, billions of cells of the body turnover in order to generate new cells necessary to maintain proper homeostatic function. As these cells replicate, they can incur mutations due to the imperfect fidelity of DNA polymerase and repair processes. Frequently, these mutations are synonymous, intronic, or dispensable and are, as a consequence, well-tolerated. However, as individuals age, this mutational burden accumulates, particularly in tissues with high turnover rates such as hematopoietic cells. In the process of clonal hematopoiesis (CH), hematopoietic stem cells incur these mutations in particular loci denoted “driver” genes, which when mutated, impart a competitive growth advantage that allows mutant cells to outcompete wildtype cells and clonally expand.³ Unlike other forms of somatic mosaicism such as cancer, CH is a premalignant state that becomes a near ubiquitous phenomenon with advanced age. Because this condition is not associated with overt hematological changes, it generally has been referred to as CH of indeterminate potential or CHIP.

Prior to the seminal reports on CH, age was a well-known risk factor of all-cause mortality and cardiovascular disease.⁴ However, the underpinnings of this additive risk remained unknown. In 2014, two pioneering studies on CH were published by Genovese et. al.⁵ and Jaiswal et. al.⁶ In both reports, peripheral blood samples from over 10,000 patients underwent whole-exome sequencing to identify and quantify CH by conventional Next Generation Sequencing (NGS) methodology. Both studies found that CH prevalence increased with age, and that this condition was associated with all-cause mortality. Expectedly, the premalignant state of CH was associated with increased risk of hematologic cancer. However, Jaiswal et. al. also noted an increased risk of incident coronary heart disease and ischemic stroke.⁶ In response, Fuster et. al. investigated whether deficiency of the clonal hematopoiesis driver gene Tet2 modified atherosclerosis progression.⁷ Indeed, in a murine model of atherosclerosis, the competitive bone marrow transplantation of Tet2-deficient cells led to exacerbated atherosclerosis progression through a myeloid-driven and Nlrp3 inflammasome/IL-1β-mediated mechanism. Following the establishment of this causal association, Jaiswal et. al. corroborated Fuster’s findings, verified the association between CH and coronary heart disease, and found that CH was associated with increased coronary artery calcification and increased risk of myocardial infarction (MI).⁸ Furthermore, after driver gene stratification, mutations in DNMT3A, TET2, ASXL1, and JAK2 were individually associated with increased incidence of coronary heart disease. Since these groundbreaking studies, additional work has demonstrated CH is associated with increased atherosclerotic cardiovascular disease (ASCVD) event rate and increased all-cause mortality in a cohort of 13,129 individuals with established ASCVD.⁹ Given the pronounced and well-established association of CH with coronary macrovascular disease, Akhiyat et al. recently reported that CH was associated with worse coronary flow reserve and increased major adverse cardiovascular event (MACE) rate in patients with coronary microvascular disease.¹⁰

In light of the initial association of CH with ischemic stroke and atherosclerotic disease, Bhattacharya et. al. analyzed 86,178 patients from 8 prospective studies for CH.¹¹ In this report, CH was associated with increased risk of total stroke. Moreover, DNMT3A- and TET2-medidated CH were associated with increased incidence of hemorrhagic stroke while TET2-mediated CH was also associated with increased incidence of ischemic stroke. Extending upon these findings, Arends et. al. analyzed peripheral blood samples from 581 patients with first-ever ischemic stroke and a 3-year follow-up via error-corrected, targeted DNA sequencing, permitting greater sensitivity in CH clone detection.¹² In this cohort, CH was associated with large-artery atherosclerosis and increased white matter lesion load. Furthermore, patients with CH had a greater risk of major adverse cardiovascular events. Of the different diver gene mutations, TET2- and PPM1D-mediated CH exhibited the greatest risk of a vascular event or death. Despite the wide accessibility and well-characterized models of ischemic and hemorrhagic stroke in mice, there has been no published causal evidence connecting CH to stroke, and the mechanistic role of CH in modifying outcomes after stroke remains outstanding.

The pathologic role of CH has been well-documented in the context of heart failure. In a cohort of 56,597 individuals from 5 prospective studies, CH was associated with a 25% increased risk of HF incidence.¹³ Clonal mutations in ASXL1, TET2, and JAK2 were each separately associated with increased HF incidence. These findings were further corroborated by Shi et. al., who found that CH was associated with heart failure incidence and specifically, heart failure with preserved ejection fraction (HFpEF) incidence.¹⁴ CH has also been robustly associated clinically and experimentally with worse prognosis in heart failure with reduced ejection fraction (HFrEF). In patients with HFrEF irrespective of etiology, clonal hematopoiesis, specifically DNMT3A- and TET2-driven CH, was associated with increased risk of HF-related death or HF-related hospitalization.¹⁵ Furthermore, Assmus et. al. performed targeted error-corrected DNA sequencing on bone marrow-derived mononuclear cells or peripheral blood mononuclear cells from 419 patients with chronic ischemic heart failure that was sufficient to discern mutations in driver genes at a variant allele fraction (VAF) greater than or equal 0.5%, (i.e. 1% of cells harboring a mutation).¹⁶ At optimized VAF cutoffs of 0.73% and 1.15%, patients with TET2- or DNMT3A-mediated CH were at a 77% increased risk of death. Expanding upon these findings, Kiefer et. al. demonstrated that mutations in other CH driver genes, which included CBL, CEBPA, EZH2, GNB1, PHF6, SMC1A, and SRSF2, were associated with increased mortality in patients with chronic ischemic heart failure.¹⁷ Finally, in patients who developed cardiogenic shock following an acute MI, TET2- and ASXL1-mediated CH were associated with decreased 30-day survival.¹⁸ Translating these findings, mutations in Tet2, Dnmt3a, Jak2, Ppm1d, Trp53, and Asxl1 exacerbated heart failure in experimental murine models of ischemic and non-ischemic HFrEF.^3,19 Additionally, in some of these models, the cardiac pathology was a consequence of a deficiency of the CH driver gene within the myeloid populations and was able to be ameliorated with Nlrp3 inflammasome inhibition.^7,19 Recently, Cochran et. al. performed ultradeep error-corrected sequencing on two cohorts of patients with HFpEF and uncovered that CH was associated with worse diastolic heart function and increased risk of CV-related hospitalization in patients with HFpEF.²⁰ Notably, a VAF of 0.5% was the most predictive of adverse outcomes, suggesting that these small clones, which are typically neglected by traditional sequencing methodologies, may harbor important prognostic significance. Additionally, TET2-mediated CH was found to be enriched in the HFpEF population compared to a control population without HFpEF. Translating these findings, adoptive transfer of Tet2-deficient bone marrow exacerbated diastolic heart function and cardiac hypertrophy in a murine model of HFpEF.²⁰

Despite the rigorous and reproducible association of CH with worse CVD prognosis, and the relatively conserved pathogenesis of CH-exacerbated disease, there exists a paucity of work exploring how CH can aid in clinical management. To date, the Canakinumab Anti-Inflammatory Thrombosis Outcomes Study (CANTOS) remains the only trial to examine CH as a potential biomarker for treatment response.²¹ In CANTOS, 10,061 patients with atherosclerotic disease and signs of systemic inflammation were randomized to either placebo or the IL-1β neutralizing antibody Canakinumab.²² Despite a 15% reduction in MACE rate, Canakinumab failed FDA approval, citing a modest efficacy relative to the side effect profile. Since the initial trial, several retrospective analyses have been published examining the utility of different biomarkers in predicting treatment response to Canakinumab. In two of these studies, on-treatment levels of the systemic inflammatory biomarkers C-reactive protein (CRP) and IL-6 improved treatment response stratification.^23,24 Specifically, patients who achieved lower levels of CRP on treatment exhibited 25% reduction in MACE rate, whereas patients who achieved lower levels of IL-6 on treatment exhibited a 32% reduction in MACE rate. Recently, Svensson et al. characterized CH via targeted DNA sequencing of peripheral blood samples sourced from baseline visits for CANTOS participants.²¹ Notably, patients with mutations in TET2 observed a 62% reduction in MACE rate with Canakinumab treatment (P = 0.04), whereas individuals without detectable CH exhibited a non-significant reduction in MACE rate with Canakinumab treatment (HR = 0.93, P = 0.38). Presently, TET2-mediated CH is the only candidate prospective biomarker and displays the greatest power in predicting treatment response. Given the large number of studies ongoing or completed using anti-inflammatory therapies for CVD,²⁵ it will be interesting to determine whether CH can enhance prediction of treatment response in other disease states.

Conclusions and Perspectives

In closing, an exhaustive amount of work has accumulated on the association between CH and increased CVD incidence, burden, and prognosis, and mechanistic studies have demonstrated this relationship to be causal and frequently mediated through inflammatory pathways (Figure). In particular, TET2-mediated CH has emerged as a potential, prospective biomarker for response to IL-1β antagonism in the context of atherosclerotic CVD. However, as patients with non-TET2-driven CH did not observe a benefit from treatment, it will be important to more thoroughly characterize the distinct pathogenesis of other CH driver genes in CVD and exploit this understanding to better tailor CVD treatment for patients. In future studies, it will be interesting to determine how the more definitive analysis of CH through ultradeep error-corrected sequencing can be used to enhance risk stratification and inform treatment responses for various CVDs.

Figure:

Summary of hallmark clinical and experimental findings connecting clonal hematopoiesis with cardiovascular disease. More details can be found in an extensive review.³ HFrEF: heart failure with reduced ejection fraction; DCM: dilated cardiomyopathy; HFpEF: heart failure with preserved ejection fraction; MACE: major adverse cardiovascular event; Ldlr: low-density lipoprotein receptor; HFD: high fat diet; L-NAME: N[w]-nitro-l-arginine methyl ester.

Nonstandard Abbreviations and Acronyms

CANTOS

Canakinumab Anti-Inflammatory Thrombosis Outcomes Study

CH

clonal hematopoiesis

CRP

C-reactive protein

CVD

cardiovascular disease

FHS

Framingham Heart Study

MACE

major adverse cardiovascular event