Serially Measured Metabolic Plasma Biomarkers and Adverse Outcomes in Chronic Heart Failure: A Longitudinal NMR Spectroscopy Study in Two Cohorts

BACKGROUND: Heart failure (HF) involves complex metabolic and lipid-related disturbances. While prior studies have shown that static lipid and metabolic profiles are associated with poor prognosis in chronic stable HF, it remains unclear whether serially measured lipid and metabolic biomarkers are associated with adverse outcomes. METHODS: We analyzed longitudinal data from 2 independent HF cohorts with serial blood sampling (Bio-SHiFT and TRIUMPH [Translational Initiative on Unique and Novel Strategies for Management of Patients With Heart Failure]), using nuclear magnetic resonance spectroscopy to quantify 45 circulating metabolic biomarkers. Joint models, combining linear mixed-effects models for repeated biomarker measurements with Cox proportional hazards models for time-to-event outcomes, were used to assess associations between current levels of serially measured biomarkers and the cohort-specific primary end point, defined as a composite of HF hospitalization, advanced HF therapy, or cardiovascular death. Models were adjusted for relevant clinical covariates. RESULTS: The Bio-SHiFT cohort (n=382; mean age, 63±13 years; 73% men; 72% New York Heart Association class I–II) contributed 2956 serial nuclear magnetic resonance measurements, while the TRIUMPH cohort (n=233; mean age 69±13 years; 69% men; 21% New York Heart Association class I–II) contributed 666. Multiple lipid-related biomarkers, including lipoprotein particle concentration, size, and standard lipid components, were associated with the primary end point. Findings were consistent between the 2 cohorts. Among high-density lipoprotein-related measures, higher current levels of S-high-density lipoprotein particle (hazard ratio [95% CI] for Bio-SHiFT per Z score of natural log-transformed biomarker, 0.59 [0.45–0.78]), high-density lipoprotein subclass 2 particle (0.57[0.42–0.80]), and total high-density lipoprotein particle (0.54 [0.41–0.70]) were consistently associated with lower risk, whereas greater mean high-density lipoprotein particle size (1.52 [1.23–1.87]) was associated with higher risk. In addition, higher current levels of medium-triglyceride-rich lipoprotein particle, total triglycerides, triglyceride-rich lipoprotein cholesterol, triglyceride-rich lipoprotein total triglycerides, and ApoA-I (apolipoprotein AI) were associated with lower risk. Beyond lipid-related biomarkers, higher current levels of the inflammation and metabolic dysregulation markers inflammation vulnerability index (3.11 [2.07–4.88]), metabolic malnutrition index (1.98 [1.38–2.96]), and metabolic vulnerability index (3.79 [2.43–5.98]) were associated with higher risk. CONCLUSIONS: Serially measured nuclear magnetic resonance-derived lipoprotein and metabolic biomarkers were associated with the primary end point in chronic stable HF. These findings support further evaluation of serial metabolic profiling as a complementary approach to risk assessment in HF.