Adding metabolic biomarkers to clinical features improved 10-year mortality prediction in older heart failure patients, increasing the area under the curve from 0.662 to 0.691.
Does integrating metabolic biomarkers with standard clinical characteristics improve 10-year mortality risk prediction compared to a clinical-only model in older patients with heart failure?
Integrating metabolic biomarkers such as citrate, omega-3 fatty acids, and HDL subclasses with clinical features significantly improves the calibration and clinical utility of 10-year mortality prediction in older heart failure patients.
Absolute Event Rate: 0% vs 0%
Background and Aims: While current heart failure models focus on short-term outcomes, long-term risk prediction is essential for an aging population. Given that heart failure is often the terminal manifestation of systemic metabolic disorders, metabolic assessments are crucial for early risk identification. This study aimed to identify key metabolites linked to mortality in older patients with heart failure and to develop and compare two 10-year mortality prediction models: a clinical model and a clinical-plus-metabolites model. Methods and Results: We analyzed the UK Biobank data from 1104 patients aged 60 years or older with a previous diagnosis of heart failure. Over a median follow-up of 13.37 years, 530 deaths occurred. Multivariable Cox models identified 19 metabolites significantly associated with all-cause mortality. Using Least Absolute Shrinkage and Selection Operator regression and Recursive Feature Elimination, we constructed Model 1 (clinical features) and Model 2 (clinical features plus metabolites, including citrate, omega-3 fatty acids, and specific high-density lipoprotein fractions). In the geographical set, Model 2 achieved an area under the curve of 0.691 compared to 0.662 for Model 1. Notably, Model 2 demonstrated superior calibration (Hosmer-Lemeshow p = 0.444 vs. p = 0.004) and provided a consistently higher clinical net benefit in decision curve analysis. Conclusion: Integrating metabolic biomarkers with standard clinical characteristics significantly improves model calibration and clinical net benefit for predicting 10-year mortality in older adults with heart failure. This approach highlights underlying mechanisms, such as mitochondrial energy dysfunction and chronic systemic inflammation, providing a practical tool for personalized care planning and targeted interventions. • Citrate, omega-3 fatty acids, and HDL subclasses predict long-term survival. • Metabolite integration enhances model calibration and clinical net benefit. • Metabolites reveal mitochondrial and inflammatory mechanisms.
Zeng et al. (Sun,) reported a other. Adding metabolic biomarkers to clinical features improved 10-year mortality prediction in older heart failure patients, increasing the area under the curve from 0.662 to 0.691.