Changing Aging — 11

That is an alarming finding. This observation of cells aging faster across the population has been offered as a potential explanation for the increased incidence of cancer in young people. PhenoAge has also been shown to predict cardiometabolic diseases and mortality. The DNAm GrimAge publication entitled “GrimAge strongly predicts lifespan and healthspan” uses a two-stage approach, multiple blood biomarkers, and added data inputs such as smoking pack-years to differentiate slow versus fast aging. Subsequently, the DunedinPACE clock was derived from DNA methylation and nineteen biomarkers in blood for determining pace of aging. It uniquely uses longitudinal data from a large birth cohort unlike the others that rely on cross-sectional data from different cohorts. The methylation markers in epigenetic clocks can be used to assess population-wide impact such as the COVID pandemic and smoking, with patterns indicative of accelerated aging. The loss of a family member’s life has been associated with acceleration of epigenetic clocks; exercise and a plant-based diet have been associated with slowing the pace. But the different epigenetic clocks differ in their predictive accuracy and response to various interventions, which has resulted in the need for benchmarking studies, currently a work in progress.

Many other clocks have been built from other omic layers, including metabolomics, transcriptomics, proteomics, lipidomics, and glycomics (measuring the sugar attachments to proteins). Of these, a proteomic aging clock has had extensive validation across three country biobanks (UK, Finland, and China) for its association with the risk of eighteen chronic diseases, mortality, and multiple age-related traits such as cognitive decline. This landmark study emphasized how important blood proteins will be in the future for predicting age-related functional status.

An inflammation marker clock (iAge) has been shown to reflect biologic age, as has the immune system response (IMM-AGE). There are also “mitotic cell” division clocks, which include both global DNA hypomethylation and focal hypermethylation specifically related to cancer risk. A multimodal AI model that integrated images of face, tongue, and retina from more than eleven thousand healthy participants, without any blood tests, also purported to provide an accurate estimation of biological age. ExplaiNAble BioLogical (ENABL) Age is another AI multimodal model that used genomics, protein markers, and clinical history in three large and diverse populations, with improvement for interpretation of biological age and accuracy of mortality prediction compared with PhenoAge and GrimAge. EyeAge uses deep-learning AI of the retinal image to determine chronological age and was more accurate than epigenetic clocks in three different population cohorts. A problem with most of these models is that chronic diseases are in the mix, so we don’t see age-related illness versus the healthy aging process. One model with the specific objective to disentangle these two strands found that a low neutrophil count and low alkaline phosphatase on lab tests at ages fifty to sixty years were predictive of healthy aging, well beyond seventy years.