A Golden Age of Aging Biomarker Discovery

Frailty is a geriatric syndrome defined as an increase state of vulnerability to adverse outcomes resulting from age-related decline in multiple physiological domains (1). With the aging population increasing worldwide that maximally affects the oldest old, efforts to prevent or delay the onset of frailty has become a critical public health goal. In the era of big data science, there is increasing interest in identifying biomarkers of frailty to understand the mechanisms underlying pathogenesis of frailty as well as discovery of tools to identify those at risk for developing frailty at an early stage. Traditionally, biomarker discovery was done using finite number of candidate biomarkers or large number of single layer of data (ie genetic, metabolomic, proteomic). However, the rapid development of new technology that allows measuring thousands of biomarkers in small specimens with unprecedent sensitivity and specificity has opened a new opportunity to describe the comprehensive molecular landscape of frailty by combining large number of multilayered data.

One of the challenges in the field, and still is a source of academic discussion, is a unified definition and operationalization of frailty (2, 3). Frailty is a heterogenous state, where individuals can become frail through different paths and manifest itself in multiple clinical forms. For example, an individual can become frail through one catastrophic event, such as a stroke, and never recover while another may become frail gradually through poor lifestyle choices and accumulation of different metabolic diseases in addition to losing fitness because of inactivity. Regardless of the paths to frailty, the concept of geroscience hypothesis postulates that the biology of aging is at the root of frailty, and that there are core mechanisms that contributes to accelerated aging observed in frailty (3, 4, 5). While many different definitions have been developed, there are two most frequently used concepts which are the frailty phenotype and accumulation of deficit model. Frailty phenotype, developed by Fried and colleagues, is based on presence of three or more of the five criteria of weight loss, exhaustion, weakness, slow walking speed, and decreased physical activity (6). On the other hand, frailty index, or accumulation of deficits is based on a sum score of varying number of aging traits such as chronic diseases, functional disability, cognitive function, and depressive symptoms (7, 8). While there are overlaps in these definitions, the prevalence of frailty differs depending on which definition is used. Frailty biomarker discovery studies have implemented different working definition of frailty making replication of initial reports a challenge, however, there have been several biomarkers that have consistently been associated with frailty across different definitions. Perhaps the use of different definition is a strength as validation across different definition would be a support of the geroscience concept of a common aging mechanisms that leads to frail state.

In this issue, Angioni and colleagues addressed the heterogeneous nature of frailty and evaluated whether biomarkers could differentiate between frailty subtypes (age-related vs disease-based) observed in the data (9). In a population of 1394 older adults, frailty was assessed using the frailty phenotype model. In 1199 participants who were either robust or prefrail at baseline, they evaluated the associations of nutritional, inflammatory-related, neurodegenerative and neuroimaging markers biomarkers with incident frailty over a 5-year follow up period. Participants who developed disease-related frailty had lower plasma DHA, DHA+EPA, and higher plasma GDF-15 and TNFR1 while no differences were observed in participants who developed age-related frailty. In addition, baseline plasma GDF-15 abundance was higher in those who developed disease-related frailty compared to those that developed age-related frailty. On average, participants who developed disease-based frailty were older than those who developed age-related frailty. As GDF-15 is strongly associated with age (10, 11, 12), the differences in GDF-15 observed may be driven by this age difference. As concluded by the authors, disease-based frailty could be viewed as a more severe state of frailty compared to age-based frailty. It follows that GDF-15 may serve as a marker to monitor the progression of frailty.

GDF-15 is a member of the transforming growth factor-b cytokine superfamily and is a protein that is upregulated in response to oxidative stress and inflammation (13). Circulating levels of GDF-15 has been positive associated with advance age (10, 11, 12) and shown to be predictive of many age-related outcomes including cardiovascular disease, mobility disability, and mortality (14, 15, 16, 17, 18, 19). The positive association of GDF-15 with frailty have been observed in other proteomic studies of frailty (20). The example of GDF-15 raises an important consideration for future research. A valid biomarkers and risk factor for frailty is not equivalent to a causal factor. It is quite possible that the condition that eventually cause frailty triggers the production of GDF-15 as a resilience response. Regardless of considerations concerning a causal role, the findings of this study confirm that GDF-15 is a promising biomarker for aging and could be useful clinical tool to identify individuals that are at risk for rapid health decline. It is important to keep in mind that frailty a complex trait that can progress over time, thus there likely require different signatures at different stages of life. For example, frailty in supercentenarians is highly prevalent while good health is the exception. Further, these signatures will likely include multiple biomarkers. In fact, recent proteomic analyses of frailty identified many frailty-associated proteins representing proteins in lipid metabolism, musculoskeletal development, and growth factor pathways (20, 21). The challenge of future studies will be to fine tune the best combination of biomarkers tailored for different goals (such as risk stratification, prediction, and monitoring). With increasing number of studies screening greater number of candidate molecules, moving beyond linear models and applying more complex machine learning and artificial intelligence algorithms, we are living the golden age of aging biomarker discovery, and we can expect that a large catalogue of biomarkers will be fast developed and used to derive specific signatures as tools for clinical evaluation. Perhaps in the not so far away future, geriatrician may be able to accurately predict health trajectory of their patients from a drop of blood, allowing them to implement intervention before health decline begins and prevent or delay frailty until very close to the end of life.

Conflict of interest

None.