A new study finds that scientists may be able to detect dementia sooner and faster

Dementia is a catch-all term for certain diseases affecting the memory of tens of millions of people. A new study finds that it might be possible now to detect dementia early and within minutes. The World’s host Carolyn Beeler speaks with Charles Marshall, the lead researcher for the study.

The World

More than 55 million people worldwide have some form of dementia: problems with memory, thinking and the ability to perform daily activities. It’s a leading cause of death for older adults around the world.

Now, scientists may have landed on a way to detect the most frequent cause of dementia, Alzheimer’s, earlier with a ten-minute brain scan and machine learning.

The study was published last week in the journal Nature Mental Health. The World’s host Carolyn Beeler spoke with Charles Marshall, a neurologist at Queen Mary University of London and the lead researcher for the study, to learn more.

Carolyn Beeler: How did you conduct this study?
Charles Marshall: We know that when people get symptoms of dementia, their brains have been changing for at least 10 to 20 years by that time, and so we were very keen to find better ways of identifying those brain changes early on. So, we used functional MRI brain scans from people in the UK Biobank cohort who had gone on to develop dementia in the years after their brain scan and compared them to people in the same cohort who’d had brain scans and not gone on to develop dementia.
And the UK Biobank is just a depository for brain scans, is that right?
So, the UK Biobank is a study of over half a million people from the UK. It’s a big, kind of medical database. And so, we looked at how strong the connections were between the brain regions in the default mode network. We then trained a machine learning tool to recognize what a dementia-like pattern of connectivity looked like. We showed that that pattern of connectivity was then able to differentiate people who would go on to develop dementia from people who weren’t going to develop dementia up to nine years in the future, with more than 80% accuracy.
What about the risk of false positives and also just being diagnosed with something when there might not be a lot that you could do to stave off the worst impacts of that diagnosis?
I agree, it’s a real worry, and I think it’s a worry with some of the other emerging tests — we’ve got the new blood tests for Alzheimer’s disease, for example — so I think we need to be very careful before we start suggesting that people without symptoms go and have a test like this to get a diagnosis. So, I think, at the moment, it’s a research tool for people who want to be involved in preventive studies, but I don’t think it’s ready for clinical use yet. And I don’t think that people should be going and having these tests to see if they’re going to get dementia in the future outside of a research study.
What kind of current tools already exist for early dementia diagnosis?
Well, there’s been a lot of interest in tests that can tell whether you have the proteins accumulating in your brain that cause Alzheimer’s disease. And we’re getting better and better at finding those proteins in people’s brains. The difficulty is that many people live with those proteins in their brain for decades without ever developing symptoms of Alzheimer’s disease. So, to be really precise about who’s going to get dementia in the future, we would ideally want to combine those tests with something that can give us an early indication that someone’s brain is starting to fail, and that’s what this type of functional MRI analysis could do. So, it could be used alongside current tests that we’ve got.
And right now, of course, there’s no cure for dementia. There are some treatments that help slow the onset of symptoms. Exactly how would this earlier identification of patients who might get dementia help?
Well, at the moment, we can see from the trials of the new treatments that are emerging for Alzheimer’s disease that they seem to be more effective the earlier they’re given. But that’s in people who’ve already got symptoms. So, there’s good reason to think that if we could go even earlier and treat people before they get symptoms when they’ve still got lots of brain cells to save, they’d be more effective still. The difficulty we have at the moment is that it’s very hard to run that type of study because you need so many people, and you need to follow them for such a long time that it becomes very expensive and difficult to do. So, with this type of test, we could find people who are at a high risk of getting dementia in the next five years, say, and we could then run a study without needing quite so many people and without it being such a barrier to testing treatments to find something that can prevent the development of dementia in the future.
So, do you see this as potentially being used as a diagnostic tool in a wider population, or is this just a step toward studying preventive drugs for dementia?
I think it could be both. So, I think one of the things that we’re very keen to do next is to see whether it would be helpful as part of the diagnostic process, when somebody at that later stage has symptoms and goes to a doctor to find out whether they’ve got Alzheimer’s disease or not. So, we would like to extend the work in that direction. But with what we’ve found already, it could be a very useful tool for selecting people to go into a trial.
How do you think about dementia and Alzheimer’s disease and its ability to be treated going into the future?
I think that we are behind where we are with other diseases. I think that cancer is decades ahead of us, but we’re in a very exciting phase because we’re seeing that there are treatments that can slow Alzheimer’s disease for the first time. These are just the first treatments emerging, and there are lots more being developed. So, I’m confident that we will continue to make progress.
Not asking you to look into your crystal ball here, but do you have any sort of estimate of when there might be significantly better treatments or preventive options for things like dementia?
If we take other diseases as an example, from the first effective treatments to having highly effective treatments, has been within 10 to 20 years, typically in other disease areas. So, I’d hopeful that we’d be on a similar trajectory.

This interview has been lightly edited and condensed for clarity.

Sign up for our daily newsletter

Sign up for The Top of the World, delivered to your inbox every weekday morning.