Plasma Proteins Reveal Three Metabolic Aging Surge Points at Ages 44, 51, and 63

As the global population ages, identifying measurable, predictive biomarkers of metabolic decline has become a public health priority. Chronological age is an imperfect proxy for biological aging, motivating the search for molecular clocks that better capture individual aging trajectories. This study addressed that gap using one of the largest plasma proteomic datasets ever assembled for aging research.

The team developed a Metabolic Age (MA) score using a LASSO Cox regression model applied to 203,491 UK Biobank participants with nuclear magnetic resonance (NMR) metabolomic data. Eighteen metabolomic indicators were selected, with glycoprotein acetyls, fatty acid unsaturation degree, and VLDL particle diameter ranking as the most influential. MA correlated strongly with chronological age (Spearman's r = 0.876) but added meaningful predictive value beyond traditional risk factors, improving C-indices for mortality, CVD, and T2D up to 0.786. Accelerated metabolic aging was associated with 30–66% higher risks of death and cardiometabolic disease, shorter telomere length, and higher frailty index scores.

In 24,920 participants with both metabolomic and proteomic data (Olink Explore 3072 platform measuring 2,923 proteins), proteome-wide analyses identified proteins significantly associated with each aging phenotype: 535 with mortality, 626 with frailty, 1,924 with MA, and smaller numbers with CVD, T2D, and telomere length. Sixty proteins were associated with all six phenotypes simultaneously, constituting a core metabolic-aging proteome. Eleven proteins were especially prominent: GDF15 and PLAUR ranked in the top 20 across all six phenotypes; TNFRSF10B, IFI30, HGF, and WFDC2 across five; and TNFRSF10A, COL6A3, PIGR, IGFBP4, and EDA2R across four. These proteins cluster around cytokine signaling, immune regulation, and extracellular matrix remodeling pathways.

The study's most novel contribution is the differential expression–sliding window analysis applied to 7,092 participants, which revealed that plasma protein changes during metabolic aging are not linear but wave-like. Three distinct peaks of differentially expressed proteins occurred at metabolic ages 44, 51, and 63 years, each with partially unique protein signatures. Nine proteins—including IL6, HAVCR2, LGALS9, PLAUR, TNFRSF10B, and WFDC2—were significant at all three peaks. Pathway enrichment showed that the age-44 wave involved defense responses to bacteria and cytokine–receptor interaction; the age-51 wave centered on inflammatory response; and the age-63 wave highlighted cellular response to lipopolysaccharide and cytokine–cytokine receptor interaction. Protein trajectory clustering identified three groups: two with linear MA-associated increases and one with nonlinear (accelerating) increases, suggesting heterogeneous aging dynamics across different biological systems.

The findings are robust across training and validation sets, with 99.1–99.9% directional consistency. Key limitations include the UK Biobank's predominantly White, relatively healthy, and middle-to-older-aged population, limiting generalizability. The cross-sectional nature of the proteomic measurements precludes definitive causal inference. Nevertheless, the identification of discrete surge points at ages 44, 51, and 63 opens a compelling framework for timing preventive interventions to counter accelerated metabolic aging.