"Tissue inflammation blood test points to dementia risk," is the headline in The Times.
Researchers in the US say people who have higher measures of inflammation in middle age are likely to have less brain tissue in some parts of their brain in older age.
The differences in brain volume, seen on MRI scans, were also accompanied by small differences in performance on memory tests.
But the study didn't find that people with raised inflammatory measures in middle age were more likely to get dementia, as it wasn't set up to directly measure dementia risk.
Previous research found people with dementia and a smaller brain volume are likely to have higher measures of substances linked to inflammation in their blood.
But it wasn't clear whether the inflammation happened before the dementia, or afterwards.
The association is further complicated by the fact it's normal for people's brains to experience some shrinkage as they get older. And, obviously, not everyone gets dementia as they get older.
While the study is certainly interesting, it doesn't provide any concrete answers.
For example, we don't know how people's inflammatory measures changed over time, or what role factors other than inflammation may have had.
There are steps you can take to reduce your risk of dementia, although these aren't guarantees.
This includes eating a healthy diet, maintaining a healthy weight, exercising regularly, moderating how much alcohol you drink, and quitting smoking if you smoke.
The researchers came from the Johns Hopkins School of Medicine, the Baylor College of Medicine, the University of Minnesota, the Mayo Clinic, and the University of Mississippi Medical Centre, all in the US.
The study was funded by the US National Heart, Lung and Blood Institute, and was published in the peer-reviewed journal Neurology.
The Times and the Mail Online covered the study in reasonably balanced and accurate stories.
Both made it clear in the article (although not in The Times' headline) that the study didn't show a cause and effect relationship between inflammation and dementia.
This was a prospective cohort study.
These types of observational study are good for spotting links between factors – in this case, inflammation and brain volume – but can't prove that one factor causes another.
Researchers recruited more than 15,000 people aged 45 to 65 for an ongoing study principally intended to look at heart disease risk.
As part of the study, they measured 5 substances linked to inflammation in the participants' blood when they were aged 53 on average.
Twenty-four years later, they selected 1,978 participants to have their brain volume measured by MRI scan and take a word recall memory test.
They then looked at whether higher inflammatory measures were linked to brain volume and memory test performance.
The researchers specifically sought to find out whether age, sex or race might have affected the results, as these have already been linked to dementia risk.
The 5 substances chosen as markers of inflammation were:
Most of these are linked to blood clotting or the body's response to infection.
The researchers combined people's scores to give an overall inflammatory marker score.
The memory test involved listening to a list of 10 words and recalling as many as possible after a short delay.
The MRI scans looked at total brain volume, as well as analysing specific areas of the brain known to be affected by Alzheimer's disease (AD), such as the hippocampus.
People who had higher total inflammatory marker scores in middle age (the average age was 53 at the start of the study) were more likely to have a smaller brain volume in certain areas at the end of the study.
But the people involved in the study did have larger volumes in ventricular parts of the brain (these are cavities in the brain filled with fluid).
Compared with people who didn't have raised levels of any inflammatory markers at the start of the study, those with raised levels on 3 or more markers had smaller hippocampal (4.6% smaller), occipital lobe (5.7% smaller) and AD signature region (5.3% smaller) volumes.
They also did very slightly worse on the memory test, remembering on average 5 words out of 10, compared with 5.5 words for those without inflammatory markers.
The researchers didn't see any link between total brain volume and inflammatory markers.
The association between inflammatory markers and brain volume was stronger in people who had higher markers of inflammation at a younger age, and was weaker in African American participants. There were no differences between the sexes.
The researchers said their findings "provide support" for an early role for inflammation "in the development of neurodegenerative brain changes associated with late-life cognitive decline, AD [Alzheimer's disease] and other forms of dementia".
Inflammation in the body is a response to injury or disease. But if the body is constantly in an inflammatory state, it can harm blood vessels and lead to heart disease.
This study suggests high levels of inflammation over the long term might also damage the brain.
That's not surprising – what's good for the heart is usually good for the brain, and we already know exercising, avoiding high blood pressure and eating healthily may help protect the brain.
Studies like this will help researchers work out more precisely what's happening in the brain when people experience memory loss or dementia.
But this study has some limitations.
The first and most important is that researchers didn't measure people's brain volume at the start of the study.
This means we don't know whether the results at the end of the study end represent brain shrinkage, or whether some people had always had smaller brain volume in certain areas.
This makes it harder to be sure that differences in inflammatory markers predated the differences in brain volume. This type of study design can't prove cause and effect – and in this case, it can't prove that one situation predated another.
Also, the substances measured may not be very precise measures of inflammation – they're also involved in other physiological processes.
And the study didn't look at whether people with higher inflammatory markers were more likely to get dementia, only at their brain volume and performance in one type of memory test.
We don't know the effect of the smaller brain volume in some areas on those people. The different performance on the memory test was also pretty small.
All in all, it's far too early to say we could ever have a blood test that accurately predicts dementia risk.