Aging biomarkers and the brain

Abstract

Quantifying biological aging is critical for understanding why aging is the primary driver of morbidity and mortality and for assessing novel therapies to counter pathological aging. In the past decade, many biomarkers relevant to brain aging have been developed using various data types and modeling techniques. Aging involves numerous interconnected processes, and thus many complementary biomarkers are needed, each capturing a different slice of aging biology. Here we present a hierarchical framework highlighting how these biomarkers are related to each other and the underlying biological processes. We review those measures most studied in the context of brain aging: epigenetic clocks, proteomic clocks, and neuroimaging age predictors. Many studies have linked these biomarkers to cognition, mental health, brain structure, and pathology during aging. We also delve into the challenges and complexities in interpreting these biomarkers and suggest areas for further innovation. Ultimately, a robust mechanistic understanding of these biomarkers will be needed to effectively intervene in the aging process to prevent and treat age-related disease.

Keywords: Aging; Biomarker; Epigenetic clock; Neurodegeneration; Neuroimaging; Proteomics.

Figure 1.. A hierarchical framework for aging biomarkers.

Aging involves a series of changes that propagate upwards across molecular, cellular, physiological, and functional levels. These latent aging processes can be detected using data on observable variables collected at each of these levels and modeled using aging biomarkers. Many types of variables are needed due to the vast and encompassing effects of aging. Brain aging can be captured by data from both the CNS and periphery due to interactions between organ systems, regulated by the blood-brain barrier. We list here the types of observable variables that have been used to create aging biomarkers that are relevant to the brain, but this list is non-exhaustive and will expand in coming years. The bolded variables are those that we discuss in detail, selected because they are the best-studied with respect to the brain. These biomarkers are highly interconnected and can capture overlapping sets of latent aging processes. Created with BioRender.com.

Figure 2.. Framework for developing and interpreting aging biomarkers.

See Section 3 text for details. Many different aging biomarkers can be trained that may incorporate completely different sets of variables in their models, but actually capture overlapping aging signals. Testing and validation are critical to ensure the models generalize to other datasets and capture meaningful information. Studying the underlying latent aging processes may lead to a mechanistic understanding of these biomarkers. ML = Machine learning; fn = function. Created with BioRender.com.