Where small brain scars sit may matter, not just how many there are

Brain scans of older people often show small bright spots in the white matter, the bundles of nerve fibres that connect different brain regions. These white matter hyperintensities are signs of damage to the fibres and their insulating coat, and they are more common in people with high blood pressure, diabetes and other blood-vessel problems. The more of them someone has, the higher the risk of memory decline and dementia, so doctors and researchers usually just measure their total volume.

A team led from Johns Hopkins University asked whether where the spots are adds anything. In a large American study, one particular pattern, deep in the brain, was linked to faster decline over about six years, even after taking the total amount into account. It is an observational study and a preprint, not yet peer-reviewed.

What they did

• Who: 1,667 Black and White adults without dementia, average age 76, from the Atherosclerosis Risk in Communities study, which has followed people from four US communities since the late 1980s. They had a brain MRI in 2011–2013.
• Mapping the spots: software split each person's white matter into 112 regions and measured how much of each region was affected.
• Finding patterns: a statistical method called principal component analysis looked for the ways in which affected regions tend to go together across people. The team kept only the eight patterns that came out the same when the analysis was repeated on random subsets of participants.
• Thinking tests: memory, language and executive function (planning, attention, switching between tasks) were tested at the time of the scan and again after about 5 and 6.5 years. The analysis took into account age, sex, race, education, the APOE ε4 Alzheimer's risk gene, weight, blood pressure, diabetes, cholesterol and smoking.

What they found

• Total amount mattered, as expected. People with more white-matter damage overall had lower scores at the start and declined faster in all areas tested.
• One deep pattern stood out. A pattern with damage concentrated in the corpus callosum (the thick bridge of fibres joining the two halves of the brain) and the front of the corona radiata (fibres fanning out to the outer layer of the brain), with the fibres just under the brain's surface relatively spared, was not linked to test scores at the start. But it was linked to faster decline in overall thinking, language and memory over the following years.
• It held up in checks. The link stayed when the team also took the total amount of damage into account, when they built the patterns from three of the four study sites and tested them on the fourth, and when they corrected for people who dropped out. In some of these checks, the memory result only came close to the usual threshold for being statistically significant.
• People with this pattern were more often women and more often had diabetes.
• A second pattern was linked to weaker executive function at the start, but this did not hold up in the checks, and the authors treat it as a tentative finding.

What to keep in mind

• Links in a population, not a test for individuals. The authors stress that the study shows associations across a group. Whether the pattern can help predict decline in one person still needs to be shown.
• Statistical patterns. The patterns come from a mathematical method, not from biology. Whether they reflect different underlying diseases is unknown.
• Healthier people stayed. About a quarter of participants had no follow-up visit. Those who left were older and had more blood-vessel risk factors and more white-matter damage. A statistical correction gave similar results, but it can only account for what was measured.
• Who was studied. Only Black and White participants took part, and all Black participants came from one site, so the effects of race and place can't be told apart. Automated measurement of the spots can also be thrown off by other things on a scan.
• Not yet peer-reviewed.

Why it matters

White-matter spots are usually summed up as a single amount. This study suggests that the location of the damage carries extra information: a pattern deep in the brain, in fibres that link widely spread brain networks, was linked to faster decline before it showed up in test scores. The authors call for confirmation in other groups of people before such patterns could be used to predict anything for individual patients.