Novel proteomic subtypes of frailty with distinct molecular patterns and prognosis

Objectives
The frailty phenotype has limitations in capturing the biological heterogeneity of the condition. Our study identified subtypes of frailty based on proteomics and examined their associations with several adverse outcomes.
Method
The study included 1513 frail individuals and 29,339 non-frail individuals from the UK Biobank and analyzed 2920 proteins. Unsupervised K-means clustering was applied to identify molecular subtypes of frailty and the Boruta algorithm was applied to identify the key proteins for distinguishing these subtypes.
Results
Four novel subtypes were identified among frail individuals: S1 (n = 403), S2 (n = 209), S3 (n = 587) and S4 (n = 314). In total, 567 key proteins for distinguishing subtypes were identified, in diverse biological pathways. Each subtype exhibited distinct molecular characteristics. S1 was characterized by elevated genomic instability, S2 by altered intercellular communication, S3 by broad upregulation of aging-related features, and S4 by loss of proteostasis and mitochondrial dysfunction. While the prognosis of S3 was similar to S1, S2 and S4 had a worse prognosis than S1. S2, in particular, presented a significantly increased risk of multiple adverse outcomes compared with S1, including all-cause mortality (hazard ratio 2.13; 95% confidence interval 1.60–2.85), cardiovascular disease (hazard ratio 1.78; 95% confidence interval 1.00–3.17), respiratory disease (hazard ratio 1.83; 95% confidence interval 1.24–2.70), kidney disease (hazard ratio 2.76; 95% confidence interval 1.57–4.85), liver disease (hazard ratio 6.19; 95% confidence interval 4.13–9.29), and cancer (hazard ratio 2.06; 95% confidence interval 1.43–2.96).
Conclusion
Our study identified four proteomic subtypes of frailty with distinct molecular signatures and differential prognostic implications, highlighting the biological heterogeneity of frailty and the need for personalized medicine and management strategies.