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

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.

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.

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.

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.

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.

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.

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.

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.