Abstract 4337014: Association of Clonal Hematopoiesis with Incident Heart Failure in 360,000 Individuals

Background: Clonal hematopoiesis of indeterminate potential (CHIP) is the aging-related clonal expansion of hematopoietic cells with preleukemic mutations. CHIP is an independent predictor of coronary artery disease (CAD), chronic kidney disease (CKD), arrhythmias such as atrial fibrillation, and heart failure (HF). However, previous HF-focused analyses have lacked sufficient power to examine less common CHIP driver mutations and have not investigated mediators of the association between CHIP and HF. Hypotheses: Key gene-specific CHIP subtypes independently associate with incident HF. Coincident CAD, atrial fibrillation, and CKD collectively mediate only a minority of this association. Methods: UK Biobank participants without prevalent HF or hematologic malignancy and with whole exome sequencing of blood DNA were included. The primary outcome was incident HF ascertained by linkage to inpatient records, with follow-up censored in March 2020. Cox models tested the association of composite CHIP and of key specific variant CHIP subtypes with HF, adjusted for age, sex, race, education, income, BMI, systolic blood pressure, total cholesterol, HDL cholesterol, use of anti-hypertensive and lipid-lowering therapies, smoking, prevalent diabetes, CAD, stroke, and cancer. Mediation analyses examined CAD, atrial fibrillation, and CKD as mediators. Results: Among 362,058 participants, the mean (SD) age was 56.5 (8.1) years, 12,273 (3.4%) had CHIP at baseline, and 8548 (2.4%) developed incident HF over median 11.0 (IQR 10.3-11.7) years of follow-up. Participants with CHIP were older (60.7 [6.7] vs. 56.3 [8.1] years) and more likely to have prevalent CAD (5.3 vs. 3.8%), CKD (0.4 vs. 0.3%), and atrial fibrillation (1.9 vs. 1.5%). Incidence of HF was higher in those with CHIP, irrespective of prevalent CAD ( Figure 1 ). After multivariable adjustment, CHIP was independently associated with higher risk of HF, driven by associations with non- DNMT3A CHIP subtypes (aHR 1.44 [95% CI 1.27-1.63]; P <0.0001). By contrast, there was no association with DNMT3A CHIP ( Table 1 ). In mediation analyses, CAD, atrial fibrillation, CKD, individually or combined, modestly but significantly mediated the association between non- DNMT3A CHIP and HF, accounting for 10.2%, 15.7%, 4.9%, and 27.8% of the effect, respectively ( Figure 2 ). Conclusion: Non- DNMT3A CHIP is independently associated with incident HF. CHIP-associated comorbidities likely explain only a minority of this relationship.