A Novel Aging Clock Built on Seven Clinical Biomarkers
The big advantage of aging clocks based on clinical biomarkers, such as the results of a complete blood count, or LDL cholesterol level, and so forth, is that one can at least theorize a little about what is going on under the hood when the clock output changes to indicate a higher or lower biological age. Each of the underlying biomarkers has meaning and a body of work attached to it, which is not the case for epigenetic clocks and barely the case for proteomic or transcriptomic clocks. Phenotypic age is the prototype of a widely used clinical biomarker clock. Others have been developed in recent years, and here find yet another recently published novel clinical biomarker clock.
Biological aging clocks offer valuable insights into age acceleration and disease development, making them a very powerful clinical tool for preventive medicine. However, the applicability of biological aging clocks in preventive clinical settings is closely linked to the effectiveness and efficiency of biomarker screening protocols, as well as their economic feasibility. To address this, we investigated the relationship between the performance of the regression model and the number of biomarkers utilized. Our aim was to unlock the full preventive potential of our biological aging clock.
We used a clinical cohort dataset from the Bumrungrad International Hospital in Bangkok, Thailand, encompassing 184,833 individuals and comprising 597,781 samples from 2000 to 2022. The total of 597,781 samples contained data on 174 clinical biochemistry biomarkers. Through expert consensus and iterative refinement, the biomarker set was refined to 51. Using an iterative approach, we systematically removed biomarkers with the least impact on predictive performance, ultimately narrowing the model down to six clinical biochemistry markers plus sex. These six biomarkers were creatinine, hemoglobin A1c (HbA1c), alanine aminotransferase (ALT), high-density lipoprotein (HDL), triglycerides, and albumin.
Based on only seven biomarkers, our clock accurately predicts both self-reported and physician-annotated ICD health data, indicating an increased hazard ratio. Importantly, the clock is robust even in the presence of acute infections or transient immune activation. To demonstrate the multi-ethnic generalizability of our biological age clock, we validated our approach using data from both the NHANES and UK Biobank cohorts. Our approach demonstrates the feasibility of a simple, robust, and interpretable clinical aging clock with potential for real-world implementation in personalized health monitoring and preventive care.