Longevity Briefs: How Reliable Are ‘Ageing Clocks’ Really?

Longevity briefs provides a short summary of novel research in biology, medicine, or biotechnology that caught the attention of our researchers in Oxford, due to its potential to improve our health, wellbeing, and longevity.

The problem:

In science, a reliable measurement is one that will consistently produce the same result when the same measurement is repeated under identical conditions. For example, if you were trying to measure the length of a table with a tape measure, a perfectly reliable measurement would yield exactly the same length each time. In practice, slight variations in how you make the measurement will probably lead to small variations in measured length. When making measurements in scientific studies, it is obviously important to know how reliable those measurements are. But when it comes to measuring the biology of ageing, reliability may leave something to be desired.

The most common approach to measuring biological age is to use epigenetic clocks. These are algorithms that use epigenetic alterations (modifications to the DNA molecule that change how DNA is read, without changing the genetic code itself) in order to estimate someone’s biological age – the ‘true age’ of their cells and tissues, as opposed to chronological age, the number of years since they were born. Whether these clocks are actually measuring biological age is somewhat uncertain, but that’s a topic for another day. What is known is that higher epigenetic age correlates with poorer health outcomes in old age, but how useful this is for an individual seeking to measure their own biological age depends greatly on the reliability of these measurements. So, how reliable are they? In this study, researchers comb through existing data for 18 different epigenetic clocks in order to answer that question.

The discovery:

Researchers investigated reliability at two levels: technical reliability and biological reliability. Technical reliability referred to the reliability of the measurement technique itself – that is to say, if multiple different labs were to be sent the same blood sample for testing, would they produce the same biological age estimate? In this domain, researchers found that epigenetic clocks were generally quite reliable, with ICC (Intraclass Correlation Coefficient, a statistical measure of reliability) in the ‘good’ to ‘excellent’ range.

Biological reliability was a different story. Biological reliability referred to the ability of clocks to produce consistent measurements from the same individual, while resisting short term perturbations. Epigenetic alterations can occur throughout the day and in response to environmental conditions, so if an epigenetic clock is to measure ageing (a long-term process), it needs to be able to see through these short term changes. Unfortunately, most clocks struggled to do so: for example, estimates for the same individual varied significantly depending on whether they were made before or after they had eaten, with only three clocks achieving ‘moderate’ reliability or better. Short-term stress and exposure to pollution also greatly reduced reliability.

Researchers also wanted to see if they could retroactively improve reliability by controlling for white blood cell counts. This is because epigenetic data used by clocks come from white blood cells, and counts of different white cell types can vary throughout the day and in response to environment, so researchers hypothesised that adjusting for this would improve reliability. Somewhat surprisingly, they found that doing this actually worsened reliability even further.

The implications:

The finding that epigenetic clocks are susceptible to short term changes isn’t new – for example, one study found that measured epigenetic age can be 3 years older at noon than at midnight in the same person. This doesn’t mean that epigenetic clocks are useless, but it does highlight the need to control the conditions under which measurements are taken, something which many studies unfortunately seem to ignore. For example, blood samples should really be collected at the same time of day and in a fasted state.

This study also highlights yet another reason why getting your epigenetic age measured as a way to inform health decisions isn’t a great idea – at least not yet. On an individual basis, finding out that your epigenetic age is 3 years higher than it should be or that it decreased over the last month cannot reasonably be interpreted when something like short-term stress can swing the measurement.