DunedinPACE is an epigenetic biomarker that measures the human pace of aging. It is designed to function as a speedometer for aging, offering a single-timepoint measurement of how fast a person is now aging.
The details of DunedinPACE are published in a scientific article (Belsky et al. 2022 eLife, link here.).
Code is available on GitHub, here to compute DunedinPACE from Illumina 450k or EPIC1 or EPIC2 array DNA methylation data. The code is also included in BioLearn.
Dunedin PACE has now been validated in more than 65 large cohorts, including older adults and children, in more than 17 countries, and more than 6 ethnic ancestry groups (see details here). More than 300 publications have reported DunedinPACE, including clinical studies and experimental intervention studies, as well as cohort studies.
Six design advantages distinguish DunedinPACE from other aging measures:
First, DunedinPACE was derived from 19 biomarkers ascertained longitudinally at ages 26, 32, 38, and 45, comprising 69,715 data points; Participants were tested four times, each 5 years apart. This spacing prevents short-term illness, like an infected tooth or flu, from skewing aging measurements. Unlike single-point aging measures that can be distorted by brief sickness, DunedinPACE captures the gradual, long-term changes that truly reflect aging.
Second, DunedinPACE was developed using data from healthy adults tracked through midlife, before they developed chronic diseases like diabetes or heart disease, and before they began taking medications for age-related conditions such as hypertension or elevated cholesterol. This approach avoids the noise that comes from deriving aging measures in older adults who may already have diseases, which can skew measures toward reflecting disease more than aging itself.
Third, DunedinPACE was developed using data from people all born in the same year, avoiding the problems faced by other aging measure that were developed by comparing different generations. For example, older generations had more exposure to toxins like cigarette smoke and leaded gasoline, while younger ones had childhood vaccinations and antibiotics. These factors alter the body and the epigenome but don’t reflect aging.
Fourth, DunedinPACE was built in participants tracked through midlife. Other studies tested older-adult participants, and in those studies the people aging fastest have already died, which introduces bias and makes their aging measures less precise. DunedinPACE includes data from the population who are aging slowest, and who are aging fastest.
Fifth, DunedinPACE was derived from a pool of epigenetic DNA methylation probe sites that we pre-selected to have the strongest test-retest reliability, yielding reliability over .90 for DunedinPACE.
Sixth, because we intended it to be responsive to anti-aging interventions, unlike other aging clocks DunedinPACE was trained on change. There is evidence that among aging clocks, DunedinPACE is the most sensitive to intervention to date, slows most in intervention trials, and is the most consistently responsive across studies.
The third generation: Longitudinal. DunedinPACE is sometimes referred to as a “third generation” among clocks, because earlier generations of clocks were not derived in longitudinal data to assess actual physiological decline in healthy people who are all the same chronological age.
