Association of Clonal Hematopoiesis with Incident Heart Failure

Introduction

Heart failure (HF) is a leading cause of death in the elderly (1). Lifetime risk for HF is 1 in 5, and HF is associated with short-term mortality rates exceeding those of many cancers in western countries (2,3). Coronary heart disease (CHD), along with hypertension, atrial fibrillation, and chronic kidney disease, are all risk factors for incident HF and strongly associated with aging. Age remains the strongest independent predictor for HF, but the age-related factors promoting HF development are incompletely understood.

Recent genetic analyses of large asymptomatic populations have revealed that somatic mutations (most often in DMNT3A, TET2, ASXL1, and JAK2) in hematopoietic cells leading to clonal expansion are commonly acquired during human aging (4,5). Clonally restricted hematopoiesis with cytopenias and dysplastic morphology of blood and marrow cells are associated with subsequent diagnosis of hematologic malignancies and increased risk of all-cause mortality (6). However, most individuals with clonal hematopoiesis detected in peripheral blood do not have cytopenia, dysplasia, or neoplasia, a phenomenon termed clonal hematopoiesis of indeterminate potential (CHIP) (6,7). The prevalence of CHIP increases with age; the frequency is low (<0.5%) from birth until 50 years of age after which it rises rapidly, affecting 10–20% of persons aged 70 to 80 years (4,5). There is increasing evidence that individuals with CHIP are at increased risk of incident atherosclerotic cardiovascular disease (CVD) events, including CHD, stroke and CVD-related mortality (8–13). Data from humans and experimental models relating CHIP somatic mutations to CVD have implicated altered immune cell function and pro-inflammatory cytokines (11,14–18).

Recently in a cohort of patients with HF, Dorsheimer et al found during 4.4 years of median follow-up, those with either TET2 or DNMT3A mutations had increased risk of death or HF hospitalization (HR=2.1, 95% CI 1.1–4.0) (19). Murine models with hematopoietic or myeloid-specific deficiency of Tet2 or with myeloid-specific transgenic Jak2^V617F are more prone to cardiac dysfunction after coronary artery ligation-induced myocardial infarction or aortic constriction-induced pressure overload (20–22). Therefore, we tested the hypothesis that CHIP driver mutations are associated with incident HF in four cohorts from the NHLBI Trans-Omics for Precision Medicine (TOPMed) Program and the United Kingdom Biobank (UKBB) study.

Methods

Results

A total of 56,597 study participants were analyzed in the present study to assess the association between CHIP and incident HF. 4,694 of them developed HF with up to 20 years follow-up. The mean age of each study ranged from 54.5 to 74.6 (SD between 5.4 and 13.0) years old, 6% of the participants had CHIP, and 3.3% of the participants had high-VAF CHIP. Table 1 shows baseline characteristics for those participants with CHIP compared to those without CHIP, and Supplemental Table 2 provide HF events follow-up time. In brief, CHIP carriers were older and more likely to have comorbidities. Prevalent CHIP did not appear to be related to BMI or lipid profiles. Consistent with prior observations, the most common CHIP genes were DNMT3A, TET2, ASXL1 and JAK2, as shown in Table 2. The numbers for somatic mutation carriers for each ascertained CHIP gene are provided in Supplemental Table 3.

In the fixed-effect meta-analysis, we observed that any CHIP mutation aggregately was associated with a 25% increased risk of HF (HR= 1.25, 95% CI 1.13, 1.38), with consistent direction of effect in four of the five studies (Central illustration). TET2 (HR=1.59, 95%CI 1.18, 2.14), JAK2 (HR=2.50, 95%CI 1.35, 4.64) and ASXL1 (HR=1.58, 95%CI 1.20, 2.08) somatic mutations were strongly associated with an increased risk of HF, while DNMT3A mutations were not associated with HF (Figure 1). In secondary analyses, we observed a slightly stronger association between high-VAF CHIP and the risk of HF (HR=1.29, 95% CI 1.15, 1.44). The associations for CHIP mutations on HF without prior CHD (HR=1.21, 95%CI 1.07, 1.36) and HF with prior CHD (HR=1.26, 95% CI 0.97, 1.64, Figure 2) were homogeneous (p=0.78 for test of homogeneity,).

Follow-up analyses in UKBB were conducted to further investigate the association between CHIP and LVEF. Demographics of the 4,122 individuals in the LVEF analyses are provided in Supplemental Table 4. We found that any CHIP was not significantly associated with reduced LVEF (p = 0.07). However, ASXL1 somatic mutations were significantly associated with reduced LVEF (beta −4.02%, 95% CI −6.97, −1.06, p=0.008). We did not observe significant associations across DNMT3A, TET2, JAK2 specific somatic mutations (Figure 3).

Discussion

In our meta-analysis from five prospective population-based studies, CHIP (as defined by somatic mutations in leukemia-related genes in the absence of hematologic malignancy) was associated prospectively with increased risk of a first episode of hospitalized HF, independently of traditional CVD risk factors. In analyses of specific CHIP driver mutations, TET2, JAK2 and ASXL1 were most strongly associated with risk of incident HF. ASXL1 somatic mutations were significantly associated with reduced LVEF. In secondary analyses, greater levels of clonal expansion (VAF>10%) were associated with higher risk of incident HF among individuals and there was no significant differences between those with and without prior CHD.

We included large, prospective NHLBI-sponsored cohort studies with available CHIP data, extensive baseline CVD risk factor data, and relatively long follow up on HF, as well as all available UKBB data. The large-scale study population enables us to have adequate statistical power for CHIP and HF association detection. Our findings are consistent with and extend recent observations from clinical and murine studies of CHIP and HF, as well as studies of CHIP and age-related CVD in general. CHIP has been associated with increased risk of subclinical atherosclerosis, myocardial infarction, ischemic stroke, and all-cause mortality, independently of traditional CVD risk factors (11,12,14). These associations were dependent on clone size with the greatest risk of CVD in those with VAF>10%. In a cohort of 200 patients with pre-existing ischemic HF at baseline, the presence of CHIP and magnitude of clonal expansion were prospectively associated with HF outcome severity (19). This association was confirmed in a larger cohort of 419 stable chronic HF patients with previous MI, where CHIP was associated with higher mortality independently of other risk factors (27). Moreover, individuals carrying multiple CHIP mutations had higher mortality compared to those carrying a single CHIP mutation or non-CHIP carriers (27). We analyzed HF with and without prior CHD to delineate the association in ischemic vs. non-ischemic forms of HF, and homogeneous effect of CHIP on the two forms of HF was observed. Previous animal study showed that Tet2, Dnmt3a, and transgenic Jak2^V617F mutations in mice induced both ischemic and hypertensive models of HF (18,21,22), which supports our finding that CHIP may influence both ischemic and non-ischemic forms of HF.

While we observe an association with overall CHIP presence and incident HF, gene-based analyses demonstrate significant associations specifically for TET2, JAK2, and ASXL1. The protein encoded by TET2 is an epigenetic regulatory enzyme that modulates hematopoietic stem cell self-renewal but is also involved in inflammatory pathway regulation (11). The mice experiments demonstrated that inactivation of Tet2 and Dnmt3 promoted Ang II-induced cardiac dysfunction and renal fibrosis (18), and in mice transplanted with Tet2-deficient bone marrow or conditional Tet2 deletion in myeloid cells, there was clonal expansion of mutant cells, along with worse cardiac remodeling, hypertrophy, and fibrosis following myocardial ischemia or pressure overload (22). A recent study also showed that bone marrow transplantation of Tet2-deficient cells was sufficient to induce HF in otherwise unchallenged mice (28). Given the observed effect estimate, CHIP is likely an important additive risk factor for HF. Whether distinct genes yield differential risk in the context of established HF risk factors merits further study.

Similar to TET2, the protein encoded by ASXL1 is an epigenetic regulator of gene expression that has been frequently mutated in myeloid malignant myeloid diseases (29–31). Somatic mutations in ASXL1 have been associated with CHD (11), however, little is known about the effect of ASXL1 somatic mutations and its potential role on the development of HF remains unexplored. We showed for the first time that ASXL1 somatic mutations were associated with LVEF, which is exploratory and hypothesis-generating for future research.

The JAK2^V617F driver mutation associated with myeloproliferative neoplasms and atherothrombotic disease is mechanistically distinct from other CHIP driver mutations. In our study population, JAK2^V617F constitutes the majority of JAK2 mutations. In a mouse transgenic model of myeloid-restricted expression of mutant JAK2 that more closely resembles the human CHIP phenotype, transgenic mice developed worsening HF following ischemia or pressure overload by promoting macrophage inflammation of the myocardium and cardiac remodeling and infarct size in an IFNGR1 and STAT1-dependent manner (21). The experimental data described above further support the role of inflammatory cytokine production and signaling by immune cells in the pathogenesis of HF (32). While HF in humans has diverse causes, clinical manifestations, and pathophysiologic mechanisms, inflammation represents a common mechanism involved in different HF subtypes (33). In both hematopoietically deficient Tet2 and Jak2 murine HF models, IL-1beta, IL-6, TNF-alpha, and CCL2 expression were increased (21,22). Thus, CHIP has diverse effects on various immune cell types and inflammatory mediators which play key roles in HF and atherosclerosis (14). These observations are also consistent with subgroup analyses from the CANTOS cohort that the IL-1beta inhibitor canakinumab was associated with a dose-dependent reduction in HF hospitalization and mortality and improvement in left ventricular function (34,35). Whether these effects for HF events are greater among individuals with CHIP as observed for MACE in a CANTOS exploratory analysis requires further study (36).

Conclusion

Our findings identify CHIP as a potentially important novel age-related risk factor for HF, consistent with previous findings of the role of CHIP as a risk factor for age-related atherosclerotic CVD more broadly. If confirmed, these findings ultimately may have potential implications for development or targeting of anti-inflammatory therapies such IL-1beta or NLRP3 inflammasome inhibitors in HF patients.

Supplementary Material