What is biological age?

Making sense of a term that scientists have quibbled over and has confused the public

When we first published papers during my PhD on what are now known as “aging clocks” in 2007, we referred to them as a “gene expression signature of aging” or “brain molecular age”. While scientifically accurate, which meant fewer scientists would chastise us at conferences and during the paper review process, using this terminology unsurprisingly resulted in almost no press.

Our “accurate jargon” didn’t explain why biological aging clocks are important in a way that people could understand.

If a scientific breakthrough happens in the forest and no one hears about it, did it really matter?

It wasn’t until Dr. Steve Horvath, now a professor at Altos Labs, came along a few years later and coined the term “aging clock” that people started to pay attention.

Oh, “a clock”, they said, now that makes sense. It’s your “real age”.

Except this (over)simplification created new problems. People now quibble over whether what we are measuring is actually accurately measuring biological age especially with the rise of commercially available tests.

The question of whether we are measuring biological age is impossible to answer because we haven’t agreed upon a definition. One scientist may measure skin wrinkles, another grey hair, and yet another muscle strength. Other scientists look for changes in epigenetics, which are modifications to your DNA that change with age, and ignore the aforementioned more outward signs of biological aging. All of these approaches (and many others) have validity, but all will give you a somewhat different answer to how old someone is biologically.

When you see articles in the press with people complaining that they took three different biological aging clock tests and got three different answers as to how old they are, I say, “well of course”. They are measuring three different sets of parameters.

There are silver foxes with hardly any wrinkles. There are very fit people who look older and people who look young who might be biologically aging pretty fast on the inside. Depending on how you choose to measure biological age, you will get different answers to how old someone is.

To add to the complexity, your organs are aging at different rates, which means that we need different clocks for different organs. Imagine a fit 40 year old runner who lives in Arizona and doesn’t use sunscreen. Their skin age may be 65 years old, while their heart age may be 25 years old.

How old would you say that they are biologically? Do you take the average of heart age and skin age? Report two different numbers?

I think it is less relevant what biological age is, and maybe impossibly tedious to answer, but instead we should focus on what criteria of a biological aging clock would make it most useful to the public and clinicians. If we can agree upon that, then we can design “the best aging clocks” and maybe get to a satisfying answer for people at home to track and as a measurement to add to clinical trials that assess anti-aging interventions.

My two cents on a useful biological aging clock:

1. Should have inputs that change with age (otherwise it’s not measuring aging)
2. Should predict function (does it tell you whether you will be able to carry your grandkids up the stairs when you are 80?)
3. Should predict risk of multiple age-related diseases (if we are measuring a root cause of age-related disease, then it should predict many diseases with aging as a risk factor)
4. Should be technically reproducible, ie. if you test the same person on the same day three times, you should get pretty close to identical results.
5. Should be reversible or slowed with anti-aging interventions (this makes it useful to track)
6. Should be organ-specific (you lose a lot of predictive power when you lump brain aging in with skin aging)

Since I study the brain, I will take brain aging as a for instance.

What most people care about when it comes to brain aging is: 1. they are cognitively sharp now and for their lifetime (function) 2. they are free of neurodegenerative diseases such as Alzheimer’s and Parkinson’s now and for their lifetime (multiple age-related diseases).

So far, it appears that organ-aging clocks, like NeuroAge, are more accurate and better fulfill the criteria of a “useful biological aging clock” when they combine multiple ways of measuring aging.

If you take the analogy of how we test for heart disease risk, this makes sense.

If you test cholesterol, you are testing an input that changes with age, predicts function, predicts risk of heart disease (and stroke and Alzheimer’s), is technically reproducible, is reversible (with statins or lifestyle), and is organ specific. Cholesterol testing is, therefore, an aging clock.

However, on its own, its not great at predicting who will have a heart attack in the future and who won’t. If you want to have a good predictor of heart health, you would commonly also test blood pressure, triglycerides, APOB, and calcium score. If you want a really accurate predictor, you would add imaging and functional testing- stress test, EKG, VO2 max, Cardiac CT, echocardiogram, MRA etc. All of these components are also aging clocks that could be combined to give a more accurate overall “heart age” score.

We are basically doing the same thing for the brain.

We combine blood biomarkers that track the “hallmarks of brain aging”, cognitive testing for memory and reaction time, and brain MRI for brain volume together. Each of these measurements on its own is a brain aging clock. Indeed each cognitive test within NeuroGames is a brain aging clock and each of the 52 RNA levels of genes that we measure in the blood is a brain aging clock. Each piece of NeuroAge testing can on its own predict function and multi-disease risk to some extent. In combination, though, they are much more accurate.

All of the NeuroAge inputs meet the criteria for a “useful aging clock”:

1. They change with age

   1. brain MRI: brains shrink in volume with age on MRI. Below is the average of 1,000 brains of people 18 years old through 70 years old. You can see that older brains are on average smaller due to the loss of neurons and the loss of connections between neurons.

      

2. Blood biomarkers: RNA levels of genes that underlie the hallmarks of brain aging change with age

   

   

   c. NeuroGames: the cognitive tests that we measure change with age significantly in large cohorts starting in people’s 20s.

   

   

   2. Each of these tests predict neurological function. NeuroGames directly tests function.

   3. Each of these tests predict risk of multiple age-related neuro diseases- Alzheimer’s, Parkinsonian signs, and vascular dementia. You can read more about our RNA biomarkers prediction of Alzheimer’s, Parkinsonian signs, and normal cognitive function in my publication from my time at MIT.
   4. Each are technically reproducible. The same person undergoing MRI close in time together has resulted in very similar results. The blood biomarkers are technically reproducible within 1-2 years or so from the same person in triplicate on multiple days. We do see slightly younger blood biomarker ages in morning samples than in evening samples- we suggest people complete testing in the morning for this reason.

   5. The combination of measurements is reversible or slowed with anti-aging interventions (this makes it useful to track). We know that people are able to increase their grey matter volume on MRI with just 25 minutes of extra exercise per week. We also know that cognition can be improved with lifestyle modification, even in people with early Alzheimer’s.
   6. These measurements are all specific to the brain. Having slower brain aging does have the additional perk of predicting overall healthspan and lifespan.

The tradeoff for the components of our testing is between accuracy and cost and time. Unfortunately, for now, it appears that the most expensive and time consuming test, brain MRI, is also the most predictive. Blood biomarkers are a close second and do seem to be predicting brain MRI results pretty accurately so far.

The order of predictive power for both chronological age and for disease risk is brain MRI> blood biomarkers> cognitive testing.

Cognitive testing is the least accurate of the tests, which makes sense since people have different baselines. Imagine a professional ping pong player who at age 80 may still have reaction times as fast as the average 25 year old.

Cognitive testing does become a much better brain aging predictor if you track it longitudinally in the same person over time. This removes the noise from different baseline abilities between people.

We are still learning about the best combination of our testing that retains >95% accuracy while being the least expensive and time consuming. We think eventually we will have a simple blood or saliva test at home that is <$100 and is as accurate as the combined NeuroAge Test. This would be much easier and more affordable for people.

Our clinical study funded by the Alzheimer’s Drug Discovery Foundation (ADDF) should help us narrow down on the optimal combination of tests. We may always lose some information by only blood testing. However, we may be able to suggest testing in series rather than in combination. Perhaps people will start with the blood test and move onto a MRI if warranted.

Having people complete all the testing, on the other hand, does allow us to use AI to suggest personalized anti-brain aging medications and lifestyle recommendations. For this reason, we may have both the complete testing and the cheaper easier testing available as options going forward.

In Summary

I hope this post provided some clarity on biological aging without either oversimplifying or becoming so technical that it wasn’t useful.

In the next five years I expect that biological aging tests will become mainstream for the public, clinicians, and researchers.

Imagine if every clinical trial measured their therapeutic for anti-aging potential using clocks? We could have youthful elixirs already available to us that we don’t know about. We just need to agree on a way to measure their potential.