Epigenetic clocks are currently the best method to measure your biological age.
They can measure how old (or young!) you really are.
Everyone has a chronological age (the age according to one’s birth certificate), but everyone ages at a different rate.
This means that some people are biologically younger than their chronological age, while others are (much) older, biologically speaking.
Before we talk about the best clocks to assess your aging process, let’s give a quick primer on how these clocks work.
A Quick Introduction to Epigenetic Clocks
Epigenetic clocks look at methylation patterns on our DNA. A large part of our DNA is covered by little molecules called methylgroups, which prevent the DNA from being accessed, silencing specific regions of the DNA.
By putting or removing methyl groups on DNA, the epigenome regulates which genes are active or not. This is very important for proper cell functioning. After all, you don’t want liver genes to be activated in heart or eye cells!
However, during aging, this methylation process goes awry, leading normally silent genes to become active (like pro-inflammatory or tumor-promoting genes), while healthy genes become silenced. We explain more about epigenetic clocks here.
Epigenetic clocks look at these methylation patterns to determine how you are aging and how old you really are biologically.
Currently, many different epigenetic clocks have been developed, are in development, or are continuously improved upon.
So What’s the Best Epigenetic Clock?
To answer this question properly, one first needs to know there are two main types of epigenetic clocks:
1. Rate of aging clocks: these clocks measure how fast you are aging; they measure how fast the clock is currently ticking.
2. Biological age clocks: these clocks measure your biological age; they measure the time on the clock (how much time has passed).
Both clocks are important to get a better view of your aging process: you want to know how fast you are aging at the moment (via a “rate of aging clock”) and you want to know how old you biologically are (via a “biological age clock”).
Let’s explain these two clocks a bit more!
Rate of Aging Clocks
Most epigenetic clocks are not rate of aging clocks, but biological age clocks. So they do not provide insights into how fast you are aging at the moment.
However, you also want to know how fast you are currently aging.
For example, over the course of your life you perhaps aged less fast (you are biologically younger according to your biological age clock), but currently you could be aging much faster.
Also, rate of aging clocks are generally more accurate and precise and are better for finding out how an intervention (like improving your lifestyle) impacts your age process.
This makes sense, given biological age clocks measure all the epigenetic damage and changes accumulated over a lifetime. Living healthier for a few months is unlikely to make a big impact on an epigenome that has been accumulating changes for many decades.
Rate of aging clocks are better in looking at recent changes.
Currently, the best rate of aging clock is the DunedinPACE clock (R). It’s an improved version of the DunedinPoAm clock (R) (the clock is continuously being updated).
The DunedinPACE clock is a unique clock. For decades, it has been following more than a thousand people in a small city in New Zealand called Dunedin. Scientists have been testing these people regularly, measuring all kinds of aging biomarkers, like grip strength, blood biomarkers (e.g., cholesterol, inflammation), dental health, brain shrinkage, and so on, and correlating these values to the epigenetic data of each patient.
This enables scientists to have a very good view on the general health and aging process of these participants and see how this is reflected in the epigenetic patterns of each participant.
This way, they created a clock that can just look at specific parts of the epigenome of people to see how healthy they really are.
Also very interesting is that the DunedinPACE clock is being developed by scientists collaborating all over the world and has been published in peer-reviewed scientific journals, enabling other independent researchers to verify and check these clocks and the methods used.
This is contrary to various other “proprietary” epigenetic clocks developed by specific companies, which have not been published in peer-reviewed scientific journals. That way, these clocks are less able to be properly checked and verified by other scientists.
Biological Age Clocks
Besides rate of aging clocks, there are also biological age clocks. These clocks try to determine how old you really are.
For example, if people have been living unhealthy lifestyles or had periods of severe stress or disease, they likely have aged faster, so they would be older than their chronological age according to their biological age clock.
Biological age clocks look at the total “damage” or aging-related changes that have accumulated in the epigenome during your life to determine your biological age.
So while rate of aging clocks tell you how fast your biological clock is currently ticking, the biological age clocks tell you how much time has already passed.
To get the best view on your aging process, it’s important to use both a rate of aging clock and a biological age clock.
Not All Epigenetic Clocks Are the Same
There exist many different epigenetic clocks according to the way they are trained and calibrated and which data they used for this.
In general, there are three kinds of epigenetic clocks:
– Clocks measuring chronological age (via the epigenome) to figure out your “biological age”
– Clocks looking at indirect biomarkers (via the epigenome) of health and mortality to determine your “biological age”
– Clocks that look at actual biomarkers of health and mortality to determine your “biological age”
Let’s delve a bit deeper into each one of these.
Chronological Age-Based Clocks
These clocks are the first-generation epigenetic clocks. They look at epigenetic patterns to estimate one’s chronological age.
But why should they do that, given you can easily find out your chronological age by just looking at your birth date?
These clocks do this because chronological age is (loosely) correlated with your risk of dying. The older you are chronologically, the greater your risk of dying of course. After all, an 80-year-old has much more risk of dying than a 50-year-old.
Given chronological age is very easy to know (everyone knows their age), this “biomarker” is thus very easy to come by, and thus scientists could easily train their clocks on epigenetic data, trying to estimate as accurately as possible the real chronological age of people based on epigenetic patterns.
This epigenetic chronological age is then of course correlated to one’s risk of disease and mortality.
An example: if a 40-year-old person takes such a test and finds out he is 48 years old instead of 40 years old, this would imply he is epigenetically older than expected for his age, meaning he has a higher risk of getting aging-related diseases and has an increased risk of dying.
More specifically, it would mean this person has double the risk of dying, given we know that for each 8 years one is older, the risk of dying doubles.
Examples of chronological-age-focused epigenetic clocks are the 2013 Horvath aging clock and the Hannum clock.
However, chronological age is a crude estimate of your risk of dying. Some people die at age 65 while others are 90 years old and still in good health. This brings us to a second type of epigenetic clock:
Clocks Trained on Indirect Biomarkers of Health and Mortality
These epigenetic clocks will not just base themselves on chronological age to indirectly estimate your risk of dying; they will also look at specific biomarkers that are associated with health and mortality.
For example, scientists will look at the concentrations of specific proteins in the blood of people that are associated with disease and mortality, and then correlate these levels with epigenetic patterns.
Given the concentrations of these specific blood proteins are correlated with disease and mortality, it’s then also possible to correlate epigenetic patterns with disease and mortality.
An example is the Levine DNAm PhenoAge clock. To create this clock, it looked at various blood biomarkers like creatine, red blood cell distribution width, total leukocytes, lymphocytes, CRP (a protein involved in inflammation), albumin, glucose, mean cell volume, and alkaline phosphatase, and correlated these and some other data with specific epigenetic patterns (513 specific methylation marks to be exact).
This way, one could infer your risk of disease and mortality by just looking at your epigenetic patterns.
Another example is the DNAm GrimAge clock that looks at likely even better blood biomarkers, such as adrenomedullin, beta‐2 microglobulin, cystatin C, growth differentiation factor 15, leptin, plasminogen activation inhibitor 1 (PAI-1), and tissue inhibitor metalloproteinase 1 (TIMP1), and then correlates these biomarkers (and some others) with epigenetic patterns.
Likely, these clocks are better than the first-generation clocks that just look at one’s chronological age.
However, blood biomarkers like creatine or GDF15 levels are still loosely correlated with health, disease, and mortality risk. Why not look at actual health and disease outcomes in patients?
This leads us to the third kind of clocks:
Clocks Trained on Direct Biomarkers of Health and Disease
These clocks are often called third-generation epigenetic clocks.
Instead of basing themselves on indirect markers of health and disease like your age or specific blood proteins, they actually directly try to measure how healthy people are.
For example, these clocks look at grip strength in people, aging-related brain shrinkage, dental health, lung function, and so on.
Of course, these clocks are much rarer, because it’s very time-consuming and expensive to measure all these biomarkers in large groups of people.
Clocks like DunedinPoAm (DunedinPACE), however, is such a clock.
For decades, scientists measured many biomarkers of health and disease in more than 1,000 people, such as body mass index, lung function, cognition, cardiovascular fitness, facial aging, and so on (among various blood biomarkers like leptin, cholesterol, triglycerides, and so on). Then they correlated these biomarkers to epigenetic patterns to train their epigenetic clocks.
This way, one can then just look at one’s epigenetic patterns to infer someone’s health, disease, and mortality risk.
These are the biomarkers that the Dunedin clock is trained on:
|BIOMARKER||HOW IT WAS MEASURED IN THE DUNEDINPACE STUDY|
|Body mass index|
Height was measured to the nearest millimeter using a Seca 264 Wireless Stadiometer. Weight was measured to the nearest 0.1 kg using calibrated scales. Individuals were weighed in light clothing. Body mass index (BMI) was calculated using the standard formula weight in kilograms divided by height in meters squared.
Waist girth was the perimeter at the level of the noticeable waist narrowing located between the costal border and the iliac crest. Hip girth was taken as the perimeter at the level of the greatest protuberance and at about the symphysion pubic level anteriorly. Measurements were repeated and the average used to calculate waist-hip ratio.
|Glycated hemoglobin (HbA1c)|
Whole blood glycated hemoglobin concentration (expressed as a percentage of total hemoglobin) was measured by ion exchange high-performance liquid chromatography (BioRad D-100, Hercules, Calif.), a method certified by the US National Glycohemoglobin Standardization Program (http://www.ngsp.org).
Serum leptin (μg/L) was measured using Human Leptin RIA kit (Cat# HL-81K, Linco Research, Missouri, USA) (ages 32 & 38) and the Quantikine ELISA Human Leptin Immunoassay (Cat# SLP00, R&D Systems Inc, Minneapolis, MN) (age 45) according to the manufacturer’s instructions.
|Blood pressure (mean arterial pressure)|
Systolic and diastolic blood pressure were assessed according to standard protocols with a Hawksley random-zero sphygmomanometer with a constant deflation valve (age 32 & 38) and with a BpTRU™ Vital Signs Monitor BPM 200 (age 45). Mean arterial pressure (MAP) was calculated using the formula Diastolic Pressure+1/3(Systolic Pressure – Diastolic Pressure).
|Cardiorespiratory fitness (VO2Max)|
Cardiorespiratory fitness was assessed by measuring heart rate in response to a submaximal exercise test on a friction-braked cycle ergometer. Depending on the extent to which heart rate increased during a 2-min. 50 W warm-up, the workload was adjusted to elicit a steady heart rate in the range 130–170 beats per minute. After a further 6-min. constant power output stage, the maximum heart rate was recorded and used to calculate predicted maximum oxygen uptake adjusted for body weight in milliliters per minute per kilogram (VO2max) according to standard protocols.
|Lung function (FEV1 and FEV1/FVC)|
We calculated post-albuterol forced expiratory volume in one second (FEV1) and the ratio of FEV1 to forced vital capacity (FVC; FEV1/FVC) using measurements from spirometry conducted with a Sensormedics body plethysmograph (Sensormedics Corporation, Yorba Linda, CA, USA).
|Total cholesterol, triglycerides, high-density lipoprotein (HDL) cholesterol|
Serum non-fasting total cholesterol, triglycerides, and high-density lipoprotein (HDL) cholesterol levels (mmol/L) were measured by colorimetric assay on a Hitachi 917 analyzer (ages 26-32), a Modular P analyzer (age 38), and a Cobas c702 analyzer (age 45).
Serum lipoprotein(a) (nmol/L) was measured by a particle-enhanced immunoturbidimetric assay on a Hitachi 917 analyzer (ages 26-32), a Modular P analyzer (age 38), and a Cobas c502 analyzer (age 45).
|Apolipoprotein B100/A1 ratio|
Serum apolipoprotein A1 (g/L) and apolipoprotein B100 (g/L) were measured by immunoturbidimetric assay on a Hitachi 917 analyzer (ages 26-32), a Modular P analyzer (age 38), and a Cobas c502 (age 45), and the ratio between the two was calculated.
|eGFR (estimated glomerular filtration rate)|
Serum creatinine (umol/L) was measured by kinetic colorimetric assay on a Hitachi 917 analyzer (age 32), Modular P analyzer (age 38), and Cobas c702 (age 45) (Roche Diagnostics, Mannheim, Germany). eGFR was estimated utilizing the CKD-Epi formula calculated from serum creatinine.
|Blood urea nitrogen (BUN)|
Blood urea nitrogen (mmol/L) was measured by kinetic UV assay at ages 26 (Hitachi 917 analyzer) and 45 (Cobas c702 analyzer), and by kinetic colorimetric assay at ages 32 (Hitachi 917 analyzer) and 38 (Modular P analyzer), and Cobas c702 analyzer at age 45.
|High sensitivity C-reactive protein (hsCRP)|
Serum C-reactive protein (mg/L) was measured by high sensitivity immunoturbidimetric assay on a Hitachi 917 analyzer (age 32), a Modular P analyzer (age 38), and a Cobas c702 (age 45). Values were log-transformed for analysis.
|White blood cell count|
Whole blood white blood cell counts (x109/L) were measured by flow cytometry with a Coulter STKS (Coulter Corporation, Miami, FL) (age 26), a Sysmex XE2100 (Sysmex Corporation, Japan) (age 32), and a Sysmex XE5000 (Sysmex Corporation, Japan) (ages 38 and 45). Counts were log-transformed for analysis.
|Mean periodontal attachment loss (AL)|
Calibrated dentists used a PCP-2 periodontal probe (Hu-Friedy; Chicago) to measure gingival recession (the distance from the cementoenamel junction to the gingival margin) and probing depth (the distance from the probe tip to the gingival margin) in millimeters at three sites (mesiobuccal, buccal, and distolingual) per tooth, excluding third molar teeth. Periodontal attachment loss for each site was computed by summing gingival recession and probing depth and then averaged across all periodontally examined teeth. Periodontal examinations were conducted with half-mouth examinations at age 26 and full-mouth examinations at ages 32, 38, and 45 years.
|Caries-affected tooth surfaces|
Calibrated dentists examined the teeth for caries and restorations following the World Health Organization Oral Health Surveys methodology. Four surfaces were considered for anterior teeth (canines and incisors): buccal, lingual, distal, and mesial; a fifth surface, occlusal, was considered for premolar and molar teeth. Tooth surfaces were classified as having untreated caries (DS) if a cavitated carious lesion was present, as filled (FS) if a dental restoration was present (including crowns), and missing due to caries (MS) if the participant indicated that a given tooth had been removed due to decay or failed dental restorative work. DS, MS, and FS counts were summed to obtain a DMFS score (ranging from 0 to 148 surfaces). Caries experience was expressed as the % of tooth surfaces that had been caries-affected, excluding surfaces of teeth that were unerupted, lost due to trauma, extracted for reasons other than caries (impaction, orthodontic treatment, or periodontal disease), or could not be visualized by the examiner.
Balance was measured using the Unipedal Stance Test as the maximum time achieved across three trials of the test with eyes closed (Bohannon et al., 1984; Springer et al., 2007; Vereeck et al., 2008).
Gait speed (meters per second) was assessed with the 6-m-long GAITRite Electronic Walkway (CIR Systems, Inc) with 2-m acceleration and 2-m deceleration before and after the walkway, respectively. Gait speed was assessed under 3 walking conditions: usual gait speed (walk at normal pace from a standing start, measured as a mean of 2 walks) and 2 challenge paradigms, dual-task gait speed (walk at normal pace while reciting alternate letters of the alphabet out loud, starting with the letter “A,” measured as a mean of 2 walks) and maximum gait speed (walk as fast as safely possible, measured as a mean of 3 walks). We calculated the mean of the 3 individual walk conditions to generate our primary measure of composite gait speed (Rasmussen et al., 2019).
|Steps in place|
The 2-min. step test was measured as the number of times a participant lifted their right knee to mid-thigh height (measured as the height halfway between the kneecap and the iliac crest) in 2 minutes at a self-directed pace (Jones and Rikli, 2002; Rikli and Jones, 1999).
Chair rises were measured as the number of stands a participant completed in 30 seconds from a seated position (Jones et al., 1999; Jones and Rikli, 2002).
Handgrip strength was measured for the dominant hand (elbow held at 90°, upper arm held tight against the trunk) as the maximum value achieved across three trials using a Jamar digital dynamometer (Mathiowetz et al., 1985; Rantanen T et al., 1999).
At ages 38 and 45, we measured motor functioning as the time to completion of the Grooved Pegboard Test with the dominant hand.
Physical limitations were measured with the 10-item RAND 36-Item Health Survey 1.0 physical functioning scale (Ware and Sherbourne, 1992). Participant responses (“limited a lot,” “limited a little,” “not limited at all”) assessed their difficulty with completing various activities, e.g., climbing several flights of stairs, walking more than 1 km, participating in strenuous sports, etc. Scores were reversed to reflect physical limitations so that a high score indicates more limitations.
|Decline in physical functioning|
Tests of balance and grip strength and interviews about physical limitations were completed at both the age-38 and age-45 Dunedin Study assessments. We measured decline across the 7-year measurement interval by subtracting the age-38 test score from the age-45 test score.
The Wechsler Adult Intelligence Scale-IV (WAIS-IV) (Wechsler, 2008) was administered to the participants at age 45 years, yielding the IQ. In addition to full-scale IQ, the WAIS-IV measures four specific domains of cognitive function: Processing Speed, Working Memory, Perceptual Reasoning, and Verbal Comprehension.
IQ is a highly reliable measure of general intellectual functioning that captures overall ability across differentiable cognitive functions. We measured IQ from the individually administered Wechsler Intelligence Scale for Children-Revised (WISC-R; averaged across ages 7, 9, 11, and 13) (Wechsler, 2003) and the Wechsler Adult Intelligence Scale-IV (WAIS-IV; age 45) (Wechsler, 2008). We measured IQ decline by comparing scores from the WISC-R and the WAIS-IV.
Study members rated their health on a scale of 1-5 (poor, fair, good, very good, or excellent).
Facial aging is the subjective perception of aged appearance based on a facial photograph and is proposed as a clinically useful marker of mortality risk. (Christensen et al., 2009). Facial aging measurement in the Dunedin Study was based on ratings by an independent panel of 8 raters of each participant’s facial photograph (Belsky et al., 2015; Shalev et al., 2014). Facial aging was based on two measurements of perceived age. First, Age Range was assessed by an independent panel of 4 raters, who were presented with standardized (non-smiling) facial photographs of participants and were kept blind to their actual age. Raters used a Likert scale to categorize each participant into a 5-year age range (i.e., from 20-24 years old up to 70+ years old) (interrater reliability = 0.77). Scores for each participant were averaged across all raters. Second, Relative Age was assessed by a different panel of 4 raters, who were told that all photos were of people aged 45 years old. Raters then used a 7-item Likert scale to assign a “relative age” to each participant (1=“young looking”, 7=“old looking”) (interrater reliability = .79). The measure of perceived age at 45 years, Facial Age, was derived by standardizing and averaging Age Range and Relative Age scores.
Self-rated Health and Facial Aging were measured at both the age-38 and age-45 assessments. We measured decline in self-rated health as incident fair/poor health reported at the age-45 assessment. We measured acceleration in Facial Aging by computing the difference in Facial Aging Z-scores between the age-45 and age-38 assessments.
Given the large number of biomarkers the DunedinPACE clock is trained on, it’s currently the most accurate epigenetic clock available.
It’s unique compared to most other clocks, which are often much simpler clocks, trained on only one “biomarker” of aging, namely your chronological age, or mainly on blood biomarkers that are loosely correlated with disease and mortality risk.
The Importance of Lots of Data and Proper Analysis
Two other aspects that are very important for epigenetic clocks are the following:
- The clocks need to be trained on very large datasets
- The clocks use proper mathematical, statistical, and machine learning methods
Ideally, clocks are trained on large data sets, meaning lots of epigenetic data from lots of people, like thousands of them.
Ideally, they look at a big part of the epigenome. The first epigenetic clocks often only looked at 27,000 different regions in the DNA (to see if these places were methylated or not), while newer clocks are trained on data looking at 450,000 regions or even 850,000 regions (CpGs).
Of course, this leads to huge amounts of data, which need to be properly analyzed. For this, you need proper mathematical and statistical methods.
The first-generation epigenetic clocks mainly used more “simple” mathematical methods like linear regression to correlate epigenetic patterns with a specific biomarker, like chronological age or blood protein levels.
However, newer clocks use more sophisticated algorithms, such as elastic net regression, or use machine learning (artificial intelligence) to make more sense of these thousands of epigenetic patterns.
Newer clocks also use better statistical methods, such as principal component analysis (PCA). This method significantly improves the precision of epigenetic clocks.
Not all epigenetic clocks are the same. Currently, the best epigenetic clocks are the ones that are trained on direct markers of health and mortality (such as lung function, blood pressure, grip strength, and brain size), while also looking at indirect markers (blood biomarkers like CRP, triglycerides, and cholesterol) and chronological age. There is only one clock that has been so extensively trained, and that is the DunedinPACE clock.
The DunedinPACE clock measures the rate of aging (how fast the clock is currently ticking). Ideally, you also use a biological age clock (telling you how much time has passed already).
It’s important that epigenetic clocks are trained on large datasets (looking at hundreds of thousands of epigenetic marks in thousands of people) and use good mathematical and statistical methods.
It’s also important that these clocks are developed by scientists who collaborate and publish their methods and results in peer-reviewed scientific journals, so these clocks can be verified and checked by other, independent scientists.
Unfortunately, most commercially available epigenetic clocks do not fall in these categories. NOVOS, however, does fall into each of these categories and offers the best-in-class NOVOS Age.