Circulating Metabolomic Profile Predicts Change In Ejection Fraction In Heart Failure Patients

Document Type

Conference Proceeding

Publication Date


Publication Title

J Card Fail


Introduction: Heart failure (HF) with reduced ejection fraction remains a major public health concern, despite multiple effective pharmacotherapies, and disease progression can be highly variable. Accumulating data indicates that HF disease severity and progression are likely influenced by impaired structure and function of the myocardial energetic apparatus. Growing interest in circulating metabolites in HF has revealed several metabolite derangements as well as the association of certain metabolomic profiles with survival. Still unclear is whether specific metabolite levels can predict disease course, and once identified, what is the mechanism of this association. This study aimed to test whether key metabolites can predict changes in ejection fraction (EF) over time. If true, this could aid in understanding the course of disease and identify the precise metabolite targets for further mechanistic study.

Methods: We conducted a prospective registry of patients with HF and reduced ejection fraction, enrolling a total of 1122 patients who had a clinical EF assessed at <50%. Baseline fasting plasma samples underwent targeted metabolomic profiling for organic acids, amino acids, and acyl carnitines (108 species in all). Roughly five years later registry participants were re-contacted to undergo repeat echocardiogram and metabolite profiling. The association of 1) baseline metabolite levels and 2) change in metabolite level, with the change in EF was tested using linear regression adjusted for baseline EF. We also tested an additional model with adjustment for age, sex, race, diabetes, and creatinine. False discovery rate (FDR) was used to adjust for multiple comparisons with FDR<0.05 being considered statistically significant.

Results: A total of 268 patients underwent EF and metabolite reassessment. Of these patients 97 were female (36.2%), 140 were African American (52.5%), the mean age was 63.6 (SD 10.2), and baseline EF was 36.5% (SD 10.8). The median follow-up time was 5.2 years (IQR 4.4, 6.0) Five metabolites were statistically significantly associated with change in EF (Table 1). In the fully adjusted model the same 5 metabolites remained the top associations but narrowly missed statistical significance. Changes in malate (FDR=0.015) and alpha-ketoglutarate (FDR=0.022) over time were also associated with corresponding changes in EF.

Conclusions: Plasma metabolomic profiling in patients with HF reveals that organic acids, including several citric acid cycle intermediates, are predictive of future EF changes over 5 years follow-up. These metabolite levels may be a new tool to predict disease course, and further investigation is needed to illuminate the biologic mechanisms underlying this association.





First Page