Revisiting the Pace of Aging Biomarker – Fight Aging!

The Pace of Aging biomarker emerged from analysis of the Dunedin Study data. It is analogous to epigenetic clocks or phenotypic age, in that it is produced by a machine learning approach, working backwards from a large database of parameters and their changes with age in the study population. Instead of being an assessment of biological age, however, it is an assessment of the pace of biological aging.

In today’s open access paper, the Pace of Aging developers improve on their original design. They expand the study populations used and reduce the number of individual assays needed to construct the Pace of Aging marker. They also show correlations between Pace of Aging and aspects of aging such as life expectancy, mortality risk, and risk of developing age-related disease. Interestingly, Pace of Aging increases with advancing age, much as one might expect given the way in which age-related loss of function is observed to progress.

Pace of Aging in older adults matters for healthspan and lifespan

Our original Pace of Aging method was developed from analysis of health changes from young adulthood to midlife in the Dunedin Study 1972-73 birth cohort. To be most useful for comparative biodemographic analysis used by planners to evaluate efforts to promote healthy longevity, the Pace of Aging method needs to be adapted to a different context: samples of individuals representing a wide range of birth cohorts for whom follow-up begins later in the life course. In addition, whereas the Dunedin Study collected extensive biochemical and physical examination data from participants, the studies used by planners typically have access to much sparser measurement panels. Here, we introduce an adapted method for calculation of Pace of Aging in a sample composed of a wide range of birth cohorts with follow-up in midlife and older age and a sparse panel of biomarkers.

We compiled data from dried-blood spot, physical exam, and functional test protocols conducted by the US Health and Retirement Study (HRS) during 2006-2016 (six assessment waves). We identified nine parameters measured at all six waves that met criteria for inclusion in the Pace of Aging analysis: C-reactive protein (CRP), Cystatin-C, glycated hemoglobin (HbA1C), diastolic blood pressure, waist circumference, lung capacity (peak flow), tandem balance, grip strength, and gait speed. A total of 13,626 individuals provided data on at least six of these nine biomarkers across at least two of the follow-up assessments. We modeled longitudinal change in these biomarkers to estimate person-specific slopes for each of them. Then we combined slope information across biomarkers to compute each participants’ Pace of Aging.

The adapted Pace of Aging measure reveals stark differences in rates of aging between population subgroups and demonstrates strong and consistent prospective associations with incident morbidity, disability, and mortality. Pace of Aging accelerates at more advanced ages. HRS participants who were older at their baseline biomarker assessment showed more rapid change across subsequent follow-ups as compared to those who were younger. This observation is consistent with biodemographic data showing that mortality risk accelerates at older ages. Pace of Aging is faster in sociodemographic groups characterized by shorter lifespan. Men tended to experience faster Pace of Aging as compared with women. Those with less education tended to experience faster Pace of Aging as compared to those with more education, consistent with observations of a socioeconomic gradient in the pace of aging from the Dunedin Cohort and a Swiss cohort.

Midlife and older adults with faster Pace of Aging were at increased risk of incident chronic disease, disability, and mortality. Older adults with faster Pace of Aging more often developed new chronic diseases and disabilities and were at increased risk of death. Moreover, these associations were independent of smoking, obesity, and educational attainment.