Serially measured lipid and metabolic plasma biomarkers and adverse outcomes in chronic heart failure: a longitudinal NMR study in two cohorts
L Lu, S Abou Kamar, C L Palm, N T B Scholte, N Suthahar, M A Connelly, H Hillege, K Martijn Akkerhuis, V A Umans, R A De Boer, E Boersma, D Westenbrink, I KardysAbstract
Background/ Introduction
Chronic heart failure (HF) is associated with frequent hospitalisations and high mortality despite contemporary therapy. Circulating lipid and metabolic biomarkers may reflect systemic manifestations of cardio-metabolic disturbances. While static lipid and metabolic profiles have been linked to adverse prognosis, it remains unclear whether longitudinal changes reflect disease dynamics and predict clinical deterioration.
Purpose
To evaluate longitudinal associations between serially measured lipid and metabolic plasma biomarkers and adverse outcomes in chronic HF, using two independent cohorts.
Methods
We analysed longitudinal data from the Bio-SHiFT and TRIUMPH cohorts with serial blood sampling and nuclear magnetic resonance (NMR) spectroscopy to quantify 45 circulating lipid and metabolic biomarkers. Longitudinal associations with the composite endpoint of death or HF hospitalisation were assessed using joint models, combining linear mixed-effects models for biomarker trajectories with Cox proportional hazards models for time-to-event outcomes. Models were adjusted for clinical covariates and corrected for multiple testing using the false discovery rate.
Results
The results of Bio-SHiFT (n=382; 120 endpoints during median 2.2 years follow-up; 3,202 repeated samples; median age 64.4 years; 73% men) and TRIUMPH (n=250; 77 endpoints during median 1.0 years; 749 repeated samples; median age 71.0 years; 32% women) were highly consistent: multiple serially measured lipid-related biomarkers, including lipoprotein particle concentration, size, and standard lipid components, were predictive of the study endpoint. A size-dependent gradient was observed for high-density lipoprotein (HDL) particles: smaller HDL particles were associated with lower risk (Bio-SHiFT: hazard ratio [HR] per Z-score of log-transformed biomarker [95% confidence interval]: S-HDLP, 0.50 [0.40–0.64]; H2P, 0.49 [0.37–0.64]; H1P, 0.70 [0.53–0.92]), whereas large HDL particles entailed higher risk (L-HDLP, 1.42 [1.07–1.93]). Additionally, medium triglyceride-rich lipoprotein particles (M-TRLP), triglycerides, triglyceride-rich lipoprotein cholesterol (TRL-C), triglyceride-rich lipoprotein triglycerides (TRL-TG) and apolipoprotein A-I were inversely associated with the endpoint. Beyond lipid metabolism, inflammatory and metabolic dysregulation indices were positively associated with the endpoint (inflammatory vulnerability index [IVX], 3.71 [2.47–5.82]; metabolic malnutrition index [MMX], 2.16 [1.51–3.12]; metabolic vulnerability index [MVX], 4.38 [2.88–6.76]).
Conclusion(s)
Dynamic changes in NMR-based lipid and metabolic biomarker levels provide prognostic information in chronic HF. Subclass-specific lipoprotein trajectories and broader markers of inflammation and metabolic stress identify patients at higher risk of death or HF hospitalisation and support the use of serial metabolic profiling for personalised risk stratification in chronic HF.Forest plots of joint-model hazard ratioFor image description, please refer to the figure legend and surrounding text.