Why Biomarkers of Biological Age Could Be Misleading
How old are you really?
Birthdays can be met with mixed feelings. For some, it’s a celebration of yet another year on the planet – an opportunity to reflect on achievements or memories made. For others, the idea of being another year older can spark fear or anguish; we’re only here for a short time, and another birthday forces us to confront this reality. Which of the two categories you fall into might depend on your perspective of what “age” means.
For centuries, chronological age – the amount of time that has lapsed since the day you were born – was used to define how old an individual is. The truth is that our cells, and the complex molecular processes that occur within them, are completely oblivious to the passing of time measured in this way.
Enter “biological age”, which centers around how fast our cells deteriorate and the damage they incur as a measure of aging. While chronological age increases at exactly same rate for everyone, the biological aging rate differs from one person to the next, and can be influenced by health and lifestyle factors.
Calculating biological age relies on a comprehensive understanding of the aging process at the molecular level and finding valid ways to measure it. Despite major advances in aging research over recent decades, such comprehensive understanding remains to be achieved. However, some milestones have been reached in our pursuit, namely an increased appreciation of how epigenetics contributes to our phenotype, including the health of our cells.
What is epigenetics?
Epigenetics – epi meaning “on” or “above” in Greek – refers to the chemical modifications that are made to DNA to control the activity of a specific gene. DNA changes, such as mutations, impact which protein is produced from a DNA sequence. Epigenome changes, however, affect whether a gene is “turned on” or “turned off”. The most studied epigenetic change is DNA methylation, where a chemical group is added to a specific location on DNA, blocking the proteins that are required to read the DNA and transcribe it to mRNA.
Consequently, epigenetic changes are the most common biomarker adopted for calculating biological age. In 2018, a University of California Los Angeles professor, Dr. Steve Horvath, created the first “clock” of biological aging, called DNAm PhenoAge. Outside of epigenetics, the most cited biomarkers in the literature are telomere length, amount of DNA-damage and mitochondrial dysfunction.
A dynamic shift in how we think about aging
In principle, understanding the molecular underpinnings of biological aging could help us to manipulate it, either via making conscious lifestyle changes or by pharmacological means, extending human lifespan and preventing diseases associated with aging. This is the ultimate goal of “longevity research”. More recently, aging is being recognized as a disease in itself, rather than a symptom of being alive, a paradigm shift that makes this area of research all the more interesting.
“This paradigm shift builds on 25 years of fundamental research into the basic biology of aging,” says Dr. Nicholas Stroustruproup leader at the Centre for Genomic Regulation (CRG) in Barcelona. “In the late eighties and early nineties, basic scientists like Professors Cynthia Kenyon and Tom Johnson showed that mutations in single genes were sufficient to extend the lifespan of laboratory animals, suggesting that targeting a small number of components in an individual could produce huge longevity benefits.”
Stroustrup says that, for a long period of time, this work wasn’t taken seriously by the clinical community: “I think to some extent, that remains true. There have been several highly active evangelists recently that play a role, but essentially what has happened is that a sustained effort in basic science has steadily built up a body of evidence that is increasingly hard to dismiss.”
Discovering and validating biomarkers of aging
Valid and accurate measures of biological age – such as biomarkers – are lacking. The core issue here, says Stroustrup, is the “random” nature of aging: “Identical twins vary almost as much in lifespan as two complete strangers. There’s a low heritability, so we can’t simply sequence people’s genomes to give them useful information about their aging. It’s an open question as to how early in life the timing of an individual’s age-related outcomes (disease, loss of mobility and death) are determined.”
The discovery and validation of biomarkers requires humans to be studied throughout their lifetime – a logistical challenge. Consequently, research in this field relies heavily on laboratory models, like the nematode Caenorhabditis elegans (C. elegans), which live for approximately two weeks.
“We work with C. elegans because they are made from most of the same parts we are – evolutionary conserved genes, cell types and tissue types. C. elegans are also similar in how they age; they get old, wrinkly and build up lipid deposits in unbeautiful ways,” describes Stroustrup.
The nematodes are active and curious creatures, exploring their environment from an early age. However, over time, they slow down and eventually stop crawling. “They gradually lose muscle tone and function, which results in them spending a substantial fraction of their life immobile – just like humans. In worms, we call the transition from youthful vigorous movement to old age “vigorous movement cessation”, or VMC for short.”
In laboratory research, VMC is often used as a biomarker of aging and an indicator of C. elegans’ health. “Some people have called the time between birth and VMC a “healthspan” to be contrasted with “lifespan”,” explains Stroustrup. Interestingly, interventions designed to expand lifespan disproportionately affect VMC. “If you mutate a gene that grants longer life, it doesn’t necessarily give you a proportional amount of youthful vigor,” says Stroustrup. “This is concerning for human clinical work, because youthful vigor is the “big product” people want to deliver.” Effectively, using VMC as a measurable outcome when testing anti-aging interventions for humans might not produce valid research that can be translated.
Stroustrup and colleagues wanted to know why this occurs, and what it could mean for the aging process in humans. In their latest study, published in PLoS Computational Biology, they present the “LifeSpan Machine”, a tool that automates lifespan assays, following thousands of nematodes and imaging them once per hour to capture their life and death.
C. elegans has two partially independent aging processes
“The Lifespan Machine, in its original incarnation, was designed to measure the timing of only one outcome of aging – death,” explains Stroustrup. “It did a pretty good job at this, but aging is obviously multi-faceted. Life has qualities in addition to how long it lasts. When I set up my own group, we realized that the lifespan machine could probably be developed to measure the time of VMC and death precisely in each individual nematode, giving us twice the number of outcomes to study.”
The team questioned whether these two outcomes shared causes, i.e., were they products of the same underlying aging process? They used genetic tools that enable fine-tuning of the nematodes’ lifespan from a few days to two weeks to test this.
The scientists uncovered two partially independent aging processes in C. elegans – one that is responsible for determining VMC, and a second that determines time of death. “The statistical relationship between VMC timing and lifespan is inconsistent with them being determined by a single process, so we conclude there must be at least two aging processes,” says Stroustrup.
Regardless of what lifespan-altering mutation was inserted in the nematodes, this statistical correlation remained constant, bar one exception: starvation. In nematodes that were starved, the relationship between VMC and lifespan changed quantitatively.
“Most interventions left the correlation structure intact, even interventions that made individuals live on average 10 times shorter or 50% longer,” – Stroustrup.
What are the underlying mechanisms that control the maintenance of this correlation? “Open up a textbook and you’ll find hundreds of mechanisms, from oxidative stress, DNA damage to protein homeostasis,” says Stroustrup.
His own hypothesis is that the mechanisms mediating the hierarchical process structure are among those already discovered: “We know something about the mechanistic reasons why molecules, cells and organs fail with age. Our work describes the organismal dynamics of the consequences of such failure. I think it will be fruitful to now go and specifically try to map the dynamics we describe onto specific mechanisms that perhaps other people have already discovered,” he explains.
What does this data mean for aging research and associated commercial products?
Over the last decade there has been a prolific “boom” in aging and anti-aging research, extending beyond academic work into biotech and big pharma. Some companies are already marketing “epigenetic wellness tests” that offer to calculate methylation levels across the genome. As a self-indulgent experiment, I undertook such a test several years ago; the results were somewhat disappointing, estimating my biological age at 26 when I was in fact 23. As living conditions and life expectancy continue to improve in most regions of the world, interest and investment in anti-aging therapeutics will no doubt increase – the market is expected to be worth £20,135 million by 2030.
Technology Networks asked Stroustrup what the findings of this study might mean for the utility of commercial products in the anti-aging space. “There’s no substitute for functional validation. Functional validation – giving someone a treatment and then waiting for years or decades to measure that treatment’s effect on aging – is slow and expensive,” he says. “The most definitive experiments will certainly take longer than current scientists’ remaining lifespans. There is a huge and understandable pressure to develop quick proxies for the long-term outcomes that people care about, both as research tools and in commercial products. I think that our study is a warning that many simple assumptions one could make to develop such proxies rapidly are likely to be wrong.”
The research team looked at VMC and lifespan and found you cannot just assume that because an early life measurement correlates with lifespan, it can be used as a biomarker. “Everyone knows we need to test these assumptions before we trust commercial aging tests, but I think there’s a lot of hope right now that untested assumptions will eventually turn out to be correct,” Stroustrup adds.
He continues, “If so, then everything we use those assumptions for in the meanwhile will also turn out to be correct. But if the assumptions are wrong, then the commercial produces will err in complex and unpredictable ways. We won’t know for sure until we do functional validation of those tests – which of course will take years and years.”
In their landmark paper sharing DNAm PhenoAge with the world, Horvath and colleagues themselves acknowledged that, should measures of biological age like DNAm PhenoAge be used to assess the efficacy of aging interventions, more will be required to explore the dynamics of the clock following treatments.
Stroustrup and colleagues’ next steps will be to improve understanding of how multiple aging processes can co-exist in a single individual, and how those processes can be coupled together by systematic factors. This is no easy feat, as aging processes aren’t physical objects that can be directly measured.
“One exciting project in the lab right now involves trying to extend the principles we apply to VMC and lifespan to study gene regulation. These days, sequencing technology has progressed to the point where its practical to measure ten or even fifteen thousand mRNAs in a single individual, across large populations, and study how each gene changes with age,” says Stroustrup. “We can potentially go from a study of two outcomes of aging to ten-thousand little mini-outcomes of aging – changes in mRNA abundance – and ask how those outcomes are related,” he concludes.
Reference: Stroustrup et al. A hierarchical process model links behavioral aging and lifespan in C. elegans. 2022. PLoS Comput Biol. doi: 10.1371/journal.pcbi.1010415.
Dr. Nicholas Stroustrup was speaking to Molly Campbell, Senior Science Writer at Technology Networks.
Complete the form below to unlock access to this Audio Article: "Why Biomarkers of Biological Age Could Be Misleading"