Anatomy of the developing brain predicts age

Whilst looks may be deceiving, it appears that the anatomy of our brain can give the game away when it comes to age. Researchers from ten universities and institutions in the United States have developed a technique, reported in the journal Current Biology, that is apparently able to predict an individual's age to within a year.

The team identified 231 different biomarkers of brain anatomy using structural magnetic resonance imaging (MRI) and combined the data to assess an individual’s age with 92 per cent accuracy. The study examined the brains of 885 participants, all aged between three and 20 years old. Although it is far from clear how brain anatomy and structure relate to behavioural maturity, which cannot be chronologically sequenced with such ease, there are many exciting implications of the findings.

Neuroscientist Timothy Brown from the University of California, San Diego (UCSD) School of Medicine is lead author of the paper, and I questioned him about the basis of its findings and the potential implications and applications of such accurate biological age estimation…

Broadly speaking, in what measurable ways does the brain change as it matures?
The human brain changes significantly in its anatomy and physiology over the course of maturation – that is to say, in its structural and architectural characteristics, as well as in its functional characteristics (such as how different brain regions show activity for certain kinds of information processing and how these activity patterns change). These two broad types of changes (anatomical/physiological) interact and are mutually influential, and at any point in brain development they reflect a complex combination of both intrinsic and extrinsic factors, such as from genes and experience. Using noninvasive neuroimaging measures, such as in our MRI study, these changes can be measured in several ways. In this paper, we have focused on developmental changes in brain morphology (volume, surface area, thickness) and in various tissue properties within cerebral and cerebellar structures and tissue types (in this case diffusivity and signal intensity).

Why is the combination of different biomarkers so important to the approach you have developed?
Previous studies have looked at the developmental time-courses of different brain features in isolation, as a list of separate characteristics, and these show great variability even among children of exactly the same age. No single feature closely tracks a person’s age across the years, because changes in different brain characteristics cascade dynamically over development. But by looking at brain development within a multidimensional analysis, simultaneously comparing many different kinds of features, we’ve uncovered a timing signal within the brain – a developmental clock you might say – that closely tracks a child’s chronological age regardless of other differences.

Is such accurate age estimation only possible during the transition from infanthood to adulthood, or is it possible to predict the age of older brains too?
There’s no reason this approach can’t be tried with brain measures taken across any age range, but it’s possible that it might not be as accurate at predicting age at older ages. It all depends on how dynamic and detectable the brain changes are that occur across the particular years of development and ageing that are of interest. This is why our predictive power was especially high at the youngest ages we studied, because that’s when the largest changes, year-by-year, are occurring in most of the measures we looked at. But the accuracy of age prediction can certainly be addressed for any age group using these same methods.

What are the implications of the finding that brain development is so tightly controlled and therefore predictable?
Most fundamentally, this discovery addresses a basic science question. It tells us that there are aspects of human brain development that have much more regularity in timing than we previously knew. Our impression of brain development up until now has been dominated by data emphasizing the great variability that exists across children and across maturation. These new multidimensional findings that reveal tight timing control in the developing anatomy will have to be reconciled with this existing impression, and more studies using similar techniques will be required to inform this new multidimensional view of the developing brain.

What are the potential medical applications of the technique? Are there potential legal applications also?
The fact that there is such timing regularity in the developing neuroanatomy (as measured in a multidimensional way, as we did) among healthy, typically developing children means that we now have a fairly complete and accurate characterization of brain maturity at all points across these ages (three to 20 years old), which we can use as a frame of reference. In theory, some developmental disorders may affect the brain in ways that we can detect using this method – ways that show up as a significantly younger or older estimated brain age than we would expect for that person’s real age. We will explore the application of our maturational phase metric toward exactly this kind of detection of abnormal brain development, testing several different clinical diagnostic groups.

Legally, if there are situations where a young individual’s age cannot be known reliably from records or by self-report, this method could potentially be used as an objective, biologically based way to estimate a person’s age. I would be quick to add, however, that structural and functional brain scans cannot and should not be used to try to make inferences about an individual’s psychological maturity.