A shrinking brain surface, not just a thinning one, was linked to dementia risk
In brain scans of 5,430 older adults from three US studies, a smaller surface area of the brain's outer layer was linked to dementia separately from how thin that layer was, and people low on both measures had the highest risk. The study is observational and not yet peer-reviewed.
The cortex is the wrinkled outer layer of the brain, a sheet of nerve cells a few millimetres thick that does most of our thinking. When doctors look for signs of Alzheimer's disease on an MRI scan, they usually measure how thick that sheet is, or the volume of brain regions. But volume is really two things multiplied together: the thickness of the sheet and its surface area, which depends on how large and how folded it is.
Thickness and surface area seem to be shaped by different genes and different steps of brain development. A US team, working with data that include a University of Arizona study, asked whether surface area tells doctors something about dementia that thickness does not. In their data it did. The work is a preprint, not yet peer-reviewed, and it is observational: it shows links, not causes.
What they did
The researchers used brain scans from three studies of aging: two large US research programmes on memory and dementia, and a smaller study from Arizona of people with narrowed neck (carotid) arteries or other risks for blood-vessel disease. In all, 5,430 people over 50 were included. Of these, 3,124 had no memory problems, 1,634 had mild cognitive impairment (memory or thinking problems that don't yet interfere much with daily life) and 672 had dementia.
Standard software measured the surface area and thickness of the cortex in a set of regions known to be hit early by Alzheimer's disease. The team then compared these measures with people's diagnosis, taking into account age, sex, education, race, ethnicity and APOE4, the best-known risk gene for Alzheimer's. In 1,558 people without dementia at the start, they followed who developed dementia over an average of 4.3 years. In 694 people, they also looked at thousands of proteins in the fluid around the brain and spinal cord.
What they found
- Surface area and thickness were almost unrelated. In all three studies the correlation between the two was close to zero, so they really do seem to be separate measures.
- A smaller surface was linked to diagnosis on its own. Even after thickness was taken into account, each standard step down in surface area went with higher odds of mild cognitive impairment (about 1.5 times) and of dementia (about 2.1 to 2.5 times) in the two large studies. Thinner cortex had stronger links still.
- Low on both was worst. People in the lowest quarter for both surface area and thickness had by far the highest odds of dementia, more than those low on only one measure or low on volume.
- It predicted future dementia. Over the follow-up, 336 people developed dementia. Compared with people normal on both measures, those with low thickness alone had about 4 times the risk, those with low surface area alone about 2 times, and those low on both about 7.6 times. A prediction model using surface area and thickness did better than one using volume, and adding volume on top added nothing.
- Different proteins. Surface area and thickness were linked to largely different sets of proteins in the spinal fluid (67 and 169 proteins, only 12 shared), pointing to partly different biology. Signals involved in how nerve cells are wired and guided during development were linked to surface area only.
- In the Arizona study, narrowed carotid arteries went with a smaller surface area but not a thinner cortex.
What to keep in mind
- Not peer-reviewed yet, and the authors call for independent replication.
- Observational. The authors write that the design "precludes causal inference": a small brain surface may reflect a vulnerability present for a long time, or other disease-related differences, and hidden factors can't be ruled out.
- Each person's brain was measured once. The study doesn't show whether surface area shrinks as dementia develops or is low from early life; the authors list this as an open question.
- The cut-offs for "low" were set within each study and are practical summaries, not fixed medical thresholds.
- The measurements depend on the accuracy of the scan software, and the analysis didn't separate people by whether they had the amyloid deposits typical of Alzheimer's.
- The long-term part rests on one study; the other large one had too short a follow-up.
Why it matters
Brain scans are already part of dementia research and care, and the measurements in this study come from standard software that is widely used. If the results hold up, looking at the surface of the cortex as well as its thickness could help pick out people at the highest risk, for example for clinical trials. The authors suggest that a small surface area may mark a long-standing vulnerability, while thinning may reflect ongoing damage, but that idea still needs to be tested in studies that scan people repeatedly over the years.