Proteomic mapping identifies biomarkers driving healthy aging and preventing chronic diseases

A groundbreaking study reveals how blood proteins can predict disease risk, track health status, and offer new strategies to combat aging-related conditions.

Study: Longitudinal serum proteome mapping reveals biomarkers for healthy ageing and related cardiometabolic diseases. Image Credit: ArtemisDiana / Shutterstock

A recent study published in the journal Nature Metabolism identified biomarkers of healthy aging and cardiometabolic diseases by mapping longitudinal serum proteome.

Aging is characterized by progressive organic, cellular, and molecular-level degeneration, leading to elevated disease susceptibility and functional decline. While various methods exist to predict chronological age, the biology of aging remains elusive. Proteostasis, the ability to maintain stable and functional protein systems, is increasingly recognized as a cornerstone of healthy aging, with its disruption linked to numerous age-related diseases.

Loss of proteostasis is a hallmark of aging. Blood, as a major reservoir of proteins, is an ideal platform to explore proteomic biomarkers. Indeed, studies using various technologies have identified multiple proteins in the blood that correlate with age. Nevertheless, they have been limited by shorter follow-up periods and smaller sample sizes.

The Study and Findings

In the present study, researchers performed longitudinal proteome profiling in a cohort of middle-aged and older adults. They used data from the Guangzhou Nutrition and Health Study (GNHS), which included 3,796 participants and over 7,500 serum samples. Participants were split between discovery and validation cohorts, comprising 1,939 and 1,857 individuals.

Besides, an external validation cohort of 124 individuals was included. The serum proteome was measured using a mass spectrometry (MS)-based method. Overall, 438, 413, and 432 proteins were quantified in the GNHS discovery, GNHS validation, and external validation cohorts, respectively. The team analyzed data from over 1,000 discovery cohort participants with proteomic measurements at three time points.

They used k-means clustering to classify 438 proteins across time points. This revealed four trajectory clusters: cluster 1 (32 proteins) showed a marked increase over time; cluster 2 (124 proteins) exhibited a slight increase; cluster 3 (179 proteins) remained constant; cluster 4 (103 proteins) showed a decline. These trajectories reflect distinct biological trends, such as muscle protein synthesis imbalance or immune response alterations, which are critical for understanding aging processes.

Next, the researchers used linear mixed models to assess correlations between serially measured protein levels and participants’ chronological ages over follow-up. In total, 148 proteins were significantly associated with age in the discovery cohort. Of these, 86 proteins showed similar significant associations with age in both validation cohorts.

Further, 83 proteins showed high accuracy in predicting age. Forty-one proteins were associated with sex and age, with significant interactions between sex and age for seven proteins. Notably, three proteins showed robust associations with age in females only, and four proteins were strongly associated with age in males only.

Connections to Health and Disease

Further, linear mixed models examined the longitudinal relationship between aging-related proteins and 32 clinical traits, including renal, hepatic, inflammatory, metabolic, anthropometric, and cognitive parameters. This identified 320 significant protein-trait associations in both GNHS cohorts. Next, the associations between the baseline levels of 86 aging-related proteins and 14 chronic diseases were investigated using Cox proportional hazards models.

Overall, 131 nominally significant associations were observed for 67 proteins; only 35 remained significant after correction for multiple tests. Of these, 13 were associated with incident type 2 diabetes, 11 with fatty liver, five with hepatitis, three with dyslipidemia, and one each with rheumatoid arthritis, renal disease, and hypertension. Among these, several proteins, such as alpha-1-antitrypsin, have established roles in regulating metabolic and inflammatory pathways, offering potential therapeutic targets. Notably, 16 proteins were drug-targetable, with zinc and its compounds targeting 10.

Developing a Health Indicator

Further, using a random forest model, the team examined whether the 86 aging-related proteins could serve as classifiers for healthy and unhealthy status. The model achieved an area under the curve (AUC) of 0.7 in distinguishing between unhealthy and healthy participants. After a ten-fold cross-validation, a more concise model with the top 22 proteins emerged with an equivalent accuracy.

Next, a proteomic healthy aging score (PHAS) was developed using the model with 22 proteins to serve as a health status indicator. Higher PHASs were longitudinally associated with improved anthropometric parameters and hepatic, renal, lipid, and glucose metabolic biomarkers. In addition, the researchers noted that one standard deviation increment in PHAS reduced the risk of chronic diseases by 72% in the entire GNHS cohort.

Exploring Determinants

Further, the team assessed the determinants of the 22 proteins and PHAS by examining the proportion of variance explained by diet, intrinsic factors, host genetics, and gut microbiota within the discovery cohort. Host genetics explained about 8% of the variance for all proteins, intrinsic factors explained 4%, and gut microbiota accounted for 3.8%. By contrast, inherent factors explained 7% of the variance for PHAS, while host genetics and gut microbiota accounted for 4.1% and 6.3%. The study identified specific microbial species contributing to PHAS variance, suggesting their importance in modifying aging-related health outcomes.

Finally, exploratory analyses were performed to ascertain associations of PHAS with 18 gut microbial species that contributed to the observed variance. Fifteen species were significantly associated with PHAS. The team developed a gut microbial score based on the 18 gut microbial species. It was noted it was positively associated with PHAS in the discovery cohort, which remained consistent with the external cohort.

Conclusions

Together, this longitudinal analysis identified multiple proteomic biomarkers associated with aging. These aging-related proteins were closely associated with disease risk and health status. PHAS developed based on these proteins showed associations with the long-term risk of various cardiometabolic diseases. This highlights its potential as a clinical tool for monitoring health and targeting interventions to reduce aging-related morbidities.

Overall, these proteomic biomarkers hold considerable clinical relevance, providing potential targets for therapeutic interventions to address aging-related morbidities.