Aspirin and dementia: a genetic subgroup in a big trial hints at benefit, with more bleeding

No medicine is approved to prevent dementia. Aspirin is cheap and affects blood clotting and possibly inflammation, so it has long been a candidate. But the large ASPREE trial, which gave more than 19,000 healthy older adults either a daily low-dose aspirin or a placebo, found no overall effect on dementia.

An average, though, can hide differences between people. A team at Monash University in Australia went back to the trial's genetic data to ask whether some people might respond differently. They found one genetic subgroup in which aspirin was linked to far fewer dementia cases, and more serious bleeding. The team stresses that the result is exploratory and has not been confirmed anywhere else.

Aspirin works partly on platelets, the small cell fragments in the blood that clump together to form clots. How many platelets a person makes is strongly inherited. A polygenic score sums up thousands of gene variants into one number that estimates such an inherited tendency.

What they did

• The trial: ASPREE randomly gave 100 mg of aspirin a day or a dummy pill to healthy people aged 70 and over (65 and over for US minority groups) in Australia and the United States. Nobody had heart disease, dementia or a disability at the start. Dementia was diagnosed by an expert committee that did not know who took what.
• Who was analysed: 13,541 participants whose DNA could be tested. Their average age was 75, 54.8% were women and 96.6% were of European descent. Over a median of 4.6 years, 345 were diagnosed with dementia.
• The screen: the team took every suitable polygenic score from a public catalogue (1,848 after quality checks) and tested, one by one, whether each changed the effect of aspirin on dementia. Because testing so many scores makes chance hits likely, they used a strict correction for multiple testing (Bonferroni). They also looked at major bleeding, the best-known harm of aspirin.

What they found

• Platelet scores stood out. Three closely related scores for platelets passed the strict test. All three essentially carry the same genetic signal. Platelet scores were also heavily overrepresented among the weaker hits: 18 of 112, when only about 1.5 would be expected by chance.
• Fewer dementia cases in the top fifth. Among the 2,708 people in the top fifth of the main platelet score, those on aspirin had about a 3.5-fold lower risk of dementia than those on placebo (hazard ratio 0.29). In numbers: 13 dementia cases among 1,310 people on aspirin, against 46 among 1,398 on placebo.
• More bleeding in the same group. Major bleeding was about twice as common on aspirin (hazard ratio 2.08). In absolute terms, dementia went down by about 2.3 percentage points (3.3% to 1.0%) and major bleeding went up by about the same amount (2.1% to 4.4%): roughly five fewer dementia cases and five more major bleeds per 1,000 people per year. Deaths did not differ.
• No benefit elsewhere. In the other four fifths, aspirin showed no significant effect, and the authors write that they "cannot rule out" harm in people with lower scores.
• Not the usual dementia genes. The best-known genetic risk factor for Alzheimer's, APOE, did not change aspirin's effect, and neither did any of 18 scores for dementia, Alzheimer's or Parkinson's disease.

What to keep in mind

• Exploratory, after the fact. The authors call their study "hypothesis-generating". It was not planned when the trial began, and the result has not been tested in any other group of people. The team notes that ASPREE may be the only trial large enough to check it in, which makes confirmation hard.
• The winner's effect may be too big. Picking the strongest signal out of 1,848 tends to overstate its size, the authors write. The number of dementia cases per subgroup was small, so the uncertainty is wide.
• Short follow-up. Under five years is short for a disease that develops over decades.
• Genes, not measured platelets. The trial did not measure platelet counts or how well platelets clump, so the biological reason behind the link is unclear.
• One ancestry. Almost all participants were of European descent, so the result may not apply to others.
• Bleeding is a real harm. In the group that seemed to benefit, the extra bleeding was as large as the drop in dementia. Low-dose aspirin is not recommended for preventing dementia, and nobody should start it on the strength of this study.

Why it matters

Large trials often report a single average answer: "aspirin doesn't prevent dementia". This study shows how genetic data collected in such a trial can be used to look for people who respond differently, while still keeping the protection of randomization within each subgroup. If the platelet link holds up in other studies, it could point to a role for platelets and blood vessels in how dementia develops. For now it is a lead to test, not a reason to change what anyone takes.