Scientists Use Machine Learning to Develop an Epigenetic Clock for Predicting Biological Age Better

In the quest to unravel the mysteries of aging, scientists have long turned to our genetic code for answers. While machine learning models have offered insights into predicting biological age, understanding the causal factors behind aging has yet to be discovered. However, a groundbreaking study has now shed light on the hidden workings of aging by delving deep into the intricate realm of epigenetics.

Scientists at Brigham and Women’s Hospital, part of Mass General Brigham, have developed an innovative “epigenetic clock” to help understand on what drives aging. Based on analyzing DNA methylation patterns, the new approach more accurately predicts biological age compared to previous models. Published in Nature Aging, this study unveils how the new model distinguishes between genetic factors that speed up aging and those that decelerate it.

Previously developed epigenetic clocks analyze the relationship between methylation patterns and aging-related features to predict biological age, the real age of our cells (not chronological age). However, they do not provide any insight into the factors that cause or slow down aging. The researchers, led by principal investigator Vadim Gladyshev, PhD, have created the first clock that distinguishes between cause and effect in aging.

“Our clocks distinguish between changes that accelerate and counteract aging to predict biological age and assess the efficacy of aging interventions,” says Gladyshev.

The investigation was primarily focused on a specific type of DNA region called CpG sites, which have been found to be strongly linked to the aging process. Lifestyle choices like smoking and diet affect DNA methylation, but so does genetic inheritance. This explains why some people seem to age faster or slower than others, even if they have similar habits and behaviors.

Using a technique called epigenome-wide Mendelian Randomization, the team identified 20,509 CpG sites linked to eight aging-related traits, such as lifespan and health span. They then created three clock models: CausAge, which predicts biological age based on causal DNA factors; DamAge and AdaptAge, which assess damaging or protective changes, respectively. DamAge is linked to adverse outcomes, such as mortality, while AdaptAge is associated with favorable adjustments.

Testing the models on over 7,000 individuals’ blood samples from the Generation Scotland Cohort, the researchers developed a map pinpointing the CpG sites responsible for biological aging. These biomarkers help evaluate interventions aimed at promoting longevity or slowing aging.

Further validation of data from the Framingham Heart and the Normative Aging Studies confirmed DamAge’s correlation with adverse outcomes like mortality, while AdaptAge was associated with longevity. Additionally, the clocks accurately assessed biological age in reprogrammed stem cells and samples from patients with chronic conditions or lifestyle-related damage.

Gladyshev emphasizes that aging research is challenging and that we are still learning which interventions truly combat it. “Our findings present a step forward for aging research,” he said, “allowing us to more accurately quantify biological age and evaluate the ability of novel aging interventions to increase longevity.”

While further testing is needed to refine age measurement accuracy, these new clocks excel at recognizing the impact of short-term actions. The findings presented here highlight specific DNA regions that might significantly affect how long we live and how healthy we remain. Additionally, the study suggests potential avenues for exploring methods to slow down aging and potentially reverse age-related changes.

Source: Kejun Ying et al. Causality-enriched epigenetic age uncouples damage and adaptation. Nature Aging. January 19, 2024.

Reference: New Epigenetic Clocks Reinvent How We Measure Age. Brigham and Women’s Hospital. February 13, 2024.