Too early to claim MRI-based screening test for dementia is beneficial


Our Review Summary

This news release announced findings of a study that looked at differences in brain MRI scans between older people who went on to develop Alzheimer’s disease symptoms and those who did not get the disease.

The release said the findings, presented at the Radiological Society of North America meeting in Chicago, suggest MRI scans may one day help predict whether people will develop dementia.

The news release mentioned alternative ways to predict Alzheimer’s, but it didn’t say anything about potential harms or costs of using MRIs, and it didn’t explain that the method has not been tested in a clinical trial.

Why This Matters

A test to predict Alzheimer’s disease might be useful, but anytime such a new test is discussed, one should be reminded of the limited treatment options for anyone who finds they are at higher risk.  Costs and potential harms need to be weighed against potential benefits.

One of the potential benefits pointed out by the authors is that people who are told they are at risk would have time to make financial and living arrangements in advance of health problems. That’s something every aging person would ideally have on their to-do list, not just those at risk.


Does the news release adequately quantify the benefits of the treatment/test/product/procedure?

Not Satisfactory

The news release said MRIs “predicted with 89 percent accuracy who would go on to develop dementia within three years” when looking at the whole brain, and accuracy rose to 95% when researchers focused on “specific parts of the brain most likely to show damage.”

Those percentages might sound impressive. However, the news release didn’t say how many people might benefit from a scan to detect Alzheimer’s early, and how many might be subjected to testing that doesn’t help them but could cause harm (see “Explain Harms,” below).

For example, the lifetime risk of developing Alzheimer’s has been estimated at around 10-20%. According to our back-of-the-napkin calculations, if 100 older people were regularly screened using MRIs, as many as 19 might benefit from an early warning. The other 81 or so would not benefit or might be harmed.

Does the news release seem to grasp the quality of the evidence?


To its credit, the news release rightly explained in the lead that this is a “small study.”

It also gave details about how the study was conducted — first by identifying disproportionate white matter damage that had occurred in 10 people in their 70s with significant cognitive decline compared with those who did not decline, and then measuring “white matter integrity” of a separate sample of 61 people.

Finally, it quoted a researcher explaining that what’s needed is to “get more control subjects and develop computerized tools that can more reliably compare individual patients’ scans to a baseline normal standard.”

However, the news release didn’t explain that this was not a randomized trial that could measure how well MRI brain scans actually do work at predicting Alzheimer’s.

Does the news release identify funding sources & disclose conflicts of interest?


The funder — the Boerger Research Fund for Alzheimer’s Disease and Neurocognitive Disorders from the Foundation of the American Society of Neuroradiology — is noted on the sidebar of the release hosting site, EurekAlert! We encourage news release writers to include funders in the body of the news release text as well.

We didn’t observe any conflicts of interest among the researchers.

Does the news release establish the true novelty of the approach?

Not Satisfactory

The release said the study showed that by using MRIs doctors “might soon be able to tell people whether they are likely to have Alzheimer’s develop in the next few years.”

The novelty suggested here is that the changes detected were in the white matter, rather than the gray cerebral cortex, where we traditionally think plaques, tangles and nerve cell loss of Alzheimer’s are occurring.

However, the concept of using MRIs and diffusion tensor imaging to predict Alzheimer’s isn’t new, and the news release didn’t explain how these findings add to the body of evidence about Alzheimer’s detection.

Total Score: 5 of 10 Satisfactory

Source :