When you try to prognosticate stroke survivors, basically what you find is that patients that are older, or suffer from more severe stroke, have unfavorable post-stroke outcomes. But as we all know, some people just age faster than others, right? And so that’s something that translates into neuroimaging. Because when you look at the brain scan, you can basically assess brain health by looking at the scans and say this brain scan… this brain belongs to someone that ages a bit faster...
When you try to prognosticate stroke survivors, basically what you find is that patients that are older, or suffer from more severe stroke, have unfavorable post-stroke outcomes. But as we all know, some people just age faster than others, right? And so that’s something that translates into neuroimaging. Because when you look at the brain scan, you can basically assess brain health by looking at the scans and say this brain scan… this brain belongs to someone that ages a bit faster. This is an 80-year-old man’s brain in a 60-year-old body. And so maybe, should we trust more the age of the body? Or should we trust more what we are seeing on the brain scans?
And so that’s basically the observation that underlies the work we’ve been doing, because we try to establish whether we could define some personalized brain health biomarker applicable to the stroke population and say, well, maybe that brain is a bit more frail, maybe it needs more help. And we try to understand what was causing this accelerated brain aging, and did it lead to worse post-stroke outcomes, and that’s what we saw, indeed. We saw that people that were presenting more cardiovascular risk factors, and especially history of strokes, had a brain that looked older than the other people. And that if a person had a brain that looked older and was to suffer from a stroke, he had a lesser likelihood of reaching a favorable functional outcome after stroke.
We leveraged a very, very large international cohort called MRI-GENIE, who included more than 6000 patients across the US and Europe for performing a GWAS, a genome-wide association study, to find genetic variation associated with some brain health neuroimaging biomarkers. And so we had this very large cohort of people that had a brain scans, and we tried to find whether we could perform this brain age study in this population.
So we analyzed more than 4000 MRI scans and performed some textural analysis on the MRI scans, and analyzed it with machine learning algorithm to produce those brain age estimates. And then we looked at the difference that lies between the prediction of the brain age and the chronological age. So let’s say you have a brain that looks 10 years younger, then you’re going to have a -10 score, and we tried to find whether some risk factors were associated with that difference, or if some diseases were affected with that difference. So that’s what we did. And so what we think it could be helpful for, is that it could help phenotyping patients and especially identifying patients that need a bit more help at the hyperacute stage for prevention, or for follow up. Because maybe if we think that someone is carrying a certain type of frailty, or suffer from lesser brain health, maybe we should watch these patients more carefully, and be more thoughtful about prevention, for example, or relapse of the stroke.