Aging Clocks
Looking good and feeling young! How should we actually measure age?
How often do we remember looking at someone after many years and wondering why they have not aged or expressing surprise that someone looks younger than their age — or feeling sad for people who seem older than their years? No? I do understand that it is not all about how one looks but more about how one feels. Often, though, how one looks and feels are highly correlated.
Aging seems to be a highly personalized phenomenon and all of us seem to age at different rates. This write-up focuses on this phenomenon and the emerging field of evolving ways to convert this observation into actual measurements referred to as the “Aging Clocks.” These metrics, increasingly, can be relied upon to predict health, lifespan, and mortality, as well as how those predictions can effectively measure the impact of interventions such as exercise, hormone replacement therapy, or calorie restriction. There are three different aspects that these clocks need to inform us on. The first one relates to the ultimate outcome of aging, namely death, entailing predictions of mortality and survival. The second one is measuring function, especially with a focus on the rate of decline, and brings in the concept of healthy aging and healthspan. The third aspect is to measure how medical therapies (current or potentially upcoming) or lifestyle interventions impact healthspan. But first, let’s get some terminology in place.
The working definition of aging is the accumulation of damage, i.e. molecular disruptions in the structure of tissues and organs, leading eventually to compromise at the level of function or physiology.
The concept of aging is associated with two specific terms relating to the passage of time: heathspan and lifespan. Lifespan is easy to define, quantify, and study using several methods such as survival curves, whereas healthspan is a much more complicated concept and needs biomarkers to quantify. These range from metabolites such as NAD, fasting glucose or cholesterol, cellular functions such as autophagy or inflammation, cellular states such as senescence, physiological features such as cognition, frailty, blood pressure, cardiovascular health, visual acuity, auditory capacity, etc. This gives rise to the need for universal measures — clocks — that distinguish and measure biological age in comparison to chronological age. So, what are these two “ages”?
Chronological age is simply the number of years since you were born. Biological age is a much more complicated index that relates to how much decline, wear and tear has accumulated in the body at the molecular, cellular, or organ level or in the body as a whole. Therefore, it can be a window to an individual’s physiological or biological state. A 28-year-old male who smokes a few packs of cigarettes per day, eats a lot of sugars, and seldom exercises will definitely have a higher biological age than his chronological age.
Now that we have defined the different “ageotypes” (a term coined by Mike Snyder, Stanford- Link), let’s understand the concept of “delta Age” (ΔAge). This refers to the difference between the predicted age (either chronological or biological) and the actual chronological age. For these clocks to be useful predictors of mortality or healthspan or the effects of an intervention, ΔAge needs to show a strong association, i.e. to be small for most people at most ages. If so, a higher biological age or predicted chronological age is equivalent to a prediction of a shorter remaining healthspan and lifespan. In addition, one necessary aspect is to understand how the measurements offered by these clocks correlate to other clocks out there and also some of the known biomarkers of aging such as frailty, fasting glucose, gait speed, grip strength, and several others that fit the criteria defined by AFAR (The American Federation of Aging Research).
ΔAge is calculated as the residuals of regression of predicted biological age against chronological age. A positive ΔAge means that the biological age is larger than the chronological age and is associated with poor health, shorter healthspan, and higher incidence of age-related diseases such as diabetes, cardiovascular indications, and cognitive decline. Stress, inflammation, childhood trauma, psychiatric diseases, and some of the genetic diseases also show up as increased biological age in some cases.
Some examples of the clocks that are being studied and on the way to becoming industry standard (or far from it!) are illustrated below.
Telomere length
The repetitive sequences that cap the ends of the chromosomes are integral to maintenance of cell function and structural integrity. These ends shorten with each cell division and are used to predict the number of divisions a cell can undergo in the future, or what can be loosely termed as the age of the cell.
It has been established for some time that shorter telomeres are associated with increased mortality risk. Despite being one of the oldest and best-studied measures, little or no correlation is shown between this measure and the other clocks such as the epigenetic clocks that I will discuss below. It is interesting to note that this clock is able to predict the incidence of some types of cancers such as gastrointestinal and head and neck cancers. Other disease conditions this measure is associated with include Alzheimer’s disease and cardiovascular indications. Though this represents an interesting measure of aging and shows association phenotypes such as cognition, frailty and physical decline, its correlation with other clocks and standardization of measurement technique needs to be improved before it can be considered mainstream.
Transcriptomic age
Gene expression profile, or the transcriptome, is also used to measure biological age. One of the examples here is the blood transcriptome where a defined number of selected transcripts was used to classify young and old groups of subjects. In one of the studies that used five transcript predictors (PMID: 23311345), several markers such as interleukin 6, blood urea, serum albumin, and muscle strength were all found to vary significantly between the young and the old group. Some other well-known markers such as C-reactive protein, blood pressure, and hematocrit did not show any such correlation.
Some studies have looked at the expression of a large number of transcripts and trained on very large sample populations. These studies found that higher biological age, by the transcriptomic age measure used here, correlated with higher blood pressure, cholesterol, and fasting glucose levels. Smokers also exhibited higher biological age. Unfortunately, though, several of these studies did not show any correlation with epigenetic clocks, clearly indicating the need to expand to other tissues and the need for platform-independent signatures.
Proteome-based clocks
The underlying principle here is to look at the modifications that a protein goes through or the protein fingerprints that correlate with age.
Different proteins in the blood have specific half-lives (how long they remain in circulation) during which they undergo specific chemical modifications. One such modification is glycosylation. The degree of glycosylation on specific proteins has been studied in large cohorts and has been found to reveal some interesting correlations. Research has shown that glycosylation of proteins is highly responsive to inflammation and aging.
GlycanAge is one such measurement currently available in the market. This measures three different glycans on IgG (the IgG half-life is 25.8 days). This has shown high reproducibility. The standard deviation for the chronological age measurement is about 9 years. Correlation was shown with the following variables: fibrinogen, HbA1c, uric acid, body mass index, and also triglycerides when sexual dimorphism was factored in. Calorie restriction, weight loss, and decrease in BMI were all shown to decrease the glycosylation and therefore a decrease in the predicted biological age. Specific glycoforms of IgG have been shown to be associated with CVD pathogenesis, inflammation, and other disease conditions.
The most interesting aspect of this measurement is that lifestyle changes or drug-based interventions were shown to impact the biological age measurements, providing a window to select and implement such changes in a personalized manner. For example, in the case of ulcerative colitis, a polymorphism in the glycosylation enzymes was correlated to T cell hyperactivity and therefore presents the possibility that IgG glycosylation signatures can be used for patient stratification. Applicability of these measurements to clinical trials and other aging studies is actively being pursued in large studies.
Metabolomic age
The metabolome is the sum total of all the metabolites that exist in a biological sample such as blood or urine. Often, mass spectrometry is used to measure selected metabolite biomarkers or analyze a fingerprint without the targeted approach. For example, the tripeptide CGT was shown to be a biomarker for age-associated decline in lung function and hip bone mineral density. Urine samples have also been used to predict age with good correlation to some of the known age-related biomarkers such as kidney function, hyperglycemia, and HbA1c levels. Metabolite-based clocks were found to effectively predict all-cause mortality as well. Nonetheless, a lot more work needs to be done especially using larger sample cohorts to establish these clocks.
Epigenetic age or DNAmAge
One of the best-studied types of clock is the epigenetic clock. The human genome is a compilation of nucleotides (A/T/G/C) occurring in a particular order, but it is essentially the same in all cells. The epigenome, by contrast, is what makes different cells do different things. One key component of the epigenome is the methylome, which refers to a chemical modification that these basic nucleotide sequences undergo: methylation of cytosine residues that are followed immediately by a guanine (this dinucleotide is termed CpG). The first two such clocks were published in 2013 (Hannum et al, PMID: 23177740 and Horvath, PMID: 24138928).
So what does this mean? What does the way the genes are packaged and regulated have to do with aging? Observations on the linkage between age-related diseases and the epigenome have been ongoing for the last 50 years. High throughput methods, combined with the high stability of CpG methylation in biological samples, has made this method possible and reliable. A DNAm clock (DNA methylation clock) is built on CpG loci that are strongly correlated to chronological age using a supervised machine learning method. These loci need to be sparse and informative. A high degree of correlation was observed in both the epigenetic clocks mentioned above, with small mean deviations from chronological age: 3.6 years for Horvath and 4.9 years for Hannum’s clock. Horvath’s clock is a multi-tissue predictor based on 353 CpG sites while the Hannum clock is built on 71 CpG sites from blood. Horvath’s and Hannum’s clocks represent the first generation epigenetic clocks. Though they measured the chronological age with high accuracy, they showed only a weak correlation with clinical measures of age-related physiological dysregulation.
Epigenetic clocks have captured the imagination of aging researchers, policymakers, and industry sectors as a lot of work has gone into studying these in the past eight years. These clocks can predict all-cause mortality independent of risk factors such as physical activity, smoking, and body mass index to name a few. A one-year increase in ΔAge was associated with a 6% increase in developing cancer in the next three years and a 17% increase in the risk of dying of cancer in the next five years.
A high degree of correlation has been shown with the diseases of aging. For example, an acceleration of Horvath’s clock was shown to correlate with increased plaques, global cognitive decline, and episodic and working memory decline in Alzheimer’s patients.
Various versions of these epigenetic clocks are being studied and they are being referred to as the second generation clocks. Examples are PhenoAge and GrimAge. These overcome the limitations of the first-generation clocks by including methylation measures with mortality and morbidity . For example, PhenoAge was developed by Levine et al (PMID: 29676998) in two stages. The first stage used a weighted composite of 10 clinical characteristics to develop a phenotypic age estimator and in the second stage, this estimator was regressed with methylation levels leading to the identification of unique CpG sites that exhibited marked differences in disease and mortality amongst individuals of the same calendar age. This clock was shown to predict several diseases associated with age and measure lifespan.
The other second-generation epigenetic aging clock just mentioned, GrimAge, developed also by Horvath’s team (PMID: 30669119), adopted a slightly different strategy. The first stage involved identifying 12 plasma proteins as well as smoking pack years followed by regression with time to death due to all-cause mortality on the methylation-based markers of plasma protein levels and smoking pack years in essence, identifying CpG sites that jointly predicted mortality risk.
Let’s now talk a bit about what the future holds for these clocks in terms of what needs to change to make them more accurate and useful. The first and foremost is the correlation with other clocks. Aging is a complex phenomenon and it is possible that each clock looks at a different underlying mechanism that reacts differently from other mechanisms and therefore shows a higher degree of variance. Determining the cross-correlation or lack of it can help not only in creating a baseline but also in understanding the mechanisms and how they affect aging as a whole at differential rates.
How can these biological age measurements impact our lives and the societies we live in? As always, I love wearing the futurist hat and imagine the changes these measurements can bring in if they become a part of our lives. The first question is: what becomes our age? Chronological or biological? Will it lead to legal age changes? What would be represented in our passports? Recently (well, almost!) the Dutchman Emilie Ratelband went to court as he wanted to change his age legally to his biological age. Though the court dismissed his case, the seeds are sown.
I can almost imagine how age moves from a number to become a health state. Conversations between people could/would shift from “how old are you?’ to “hey I changed my age from 40 to 32 because of — — — interventions!” Or “oh! I am five years younger than when we last met!” Is that not a fabulous world to live in? To live longer and healthier and continue to contribute beyond years that are known to be possible. We are not there yet but definitely moving in that direction and will surely and definitely get there. I have not even for a minute forgotten the chronic diseases or the drug discovery industry and all the non-proactive drugging approaches we are all used to or the insurance agency where chronological age is used as a major predictor, in the above futuristic predictions. My humble submission is that all these exist today because of the perceived need and will readjust as the needs change to a more proactive health-conscious approach.
Prime Movers Lab invests in breakthrough scientific startups founded by Prime Movers, the inventors who transform billions of lives. We invest in companies reinventing energy, transportation, infrastructure, manufacturing, human augmentation, and agriculture.
Sign up here if you are not already subscribed to our blog.