A protein in the blood of older women is associated with an increased risk of cognitive decline and dementia

The headline of this press release is misleading, in my view.

The blood test does not ‘predict’ dementia early as one would commonly understand it; if you get a result above a certain level, it does not tell you if you are, or not, going to develop dementia with any degree of accuracy.

We know that there are changes in the brain for decades prior to the appearance of visible dementia symptoms, and one of those changes is the build-up of the molecule pTau217. The researchers in this study have measured the levels of pTau217 in the blood of a large number of women and then tracked them for an average of 15 years (longest 25 years) to see who developed either dementia or a condition called mild cognitive impairment.

They show that if you have substantially above-average levels of pTau217 then you are more likely to develop dementia than someone with lower levels. It does not predict that you will get dementia. This increase in risk is probably slightly higher than the dementia risk associated with smoking or heavy alcohol consumption and this probably reflects these early pre-symptomatic changes in the brain.

This is useful information – although it is not a prediction of a future dementia diagnosis, it would give an individual the opportunity to manage risk factors under their control, for example give up smoking and drinking alcohol, exercise, and maintain a healthy weight.

This is a well conducted study that adds to growing evidence that the blood biomarker p-tau217 is linked to future Alzheimer’s disease–related changes in the brain. However, the claim that a blood test could predict dementia 25 years before symptoms should be treated cautiously. The study shows that higher p-tau217 levels in older women were associated with a higher risk of developing mild cognitive impairment or dementia during up to 25 years of follow up, but this reflects the length of observation in the cohort rather than a test that can reliably predict the disorder decades in advance for an individual. It is also important to note that mild cognitive impairment (MCI) is not dementia, and many people with MCI do not go on to develop dementia. In studies like this, combining MCI and dementia as a single outcome can make the association appear stronger, but these are clinically distinct stages with different implications for patients and healthcare systems. MCI represents a risk state rather than established disease, so interpreting results as prediction of dementia needs caution, particularly when considering how such biomarkers might be used in real world clinical decision making. The bigger question now is how these biomarkers translate into real world clinical practice. Dementia in older adults is often caused by several overlapping brain diseases, and research cohorts are not always representative of routine healthcare populations; p-tau217 also only tells you about Alzheimer's disease specifically. Understanding how tests like p-tau217 perform in everyday clinical settings, and whether they meaningfully improve diagnosis and care in a cost effective way, will be critical before they can be widely used.