Science1 publisher2 min readPublished
Karolinska registry data links newer blood thinners to slower cognitive decline in Alzheimer's
In 7,308 Swedish patients who had both atrial fibrillation and Alzheimer's, the group on NOACs lost slightly more than 0.2 Mini-Mental State Examination points a year less than warfarin users or untreated patients. The study is observational.
The Scientist · Science desk
What happened
- Karolinska Institutet researchers pulled records for 7,308 people diagnosed with both atrial fibrillation and Alzheimer's disease from SveDem, Sweden's national dementia quality registry.
- Patients were sorted into three matched groups, one on newer anticoagulants called NOACs, one on the older drug warfarin, and one on no anticoagulant, with cognition tracked by repeated Mini-Mental State Examinations.
- The NOAC group declined significantly more slowly than either comparison group, by slightly more than 0.2 MMSE points per year.
- Warfarin was associated with lower risks of death, stroke and blood clots, and with a higher risk of major bleeding.
- The authors state that the observational design cannot prove NOACs caused the slower cognitive decline or the other health differences in the data.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- decision A clinician already choosing between two anticoagulant classes for a patient with both diagnoses now has a cognitive column to weigh next to stroke and bleeding risk, on matched registry evidence and not a randomised comparison.
- constraint A fifth of an MMSE point will not be visible in one year's clinic score, so prescribing on this finding means betting the same gap keeps accruing across the years a patient has left.
- precedent Attaching a cognitive endpoint to a cardiology drug comparison opens the same question for other classes prescribed to dementia patients, using MMSE series that national registries already hold.
NOAC against warfarin is the informative comparison. Both groups hold patients a doctor judged fit to anticoagulate, so they resemble each other in ways neither resembles the group given no anticoagulant at all [3]. Matching inside a registry handles the variables the registry records [2]. The authors write that some factors capable of affecting both which treatment a patient received and how their health changed over time could not be fully evaluated [9].
Earlier work had suggested that anticoagulation may lower the risk of developing dementia, and much less was known about cognition after an Alzheimer's diagnosis is already in place [16]. Maria Eriksdotter, who led the study and is a senior consultant in geriatric medicine at Karolinska University Hospital, gave a reason to expect an effect. "There are reasons to believe that the treatment could have a positive effect on cognition, for example by improving blood flow and reducing small-scale damage in the brain," she said [10][12].
That mechanism concerns blood flow in the brain. It has no bearing on a broken bone, and the NOAC group had fewer fractures than the untreated group as well as fewer deaths, strokes and clots [6]. In my view the untreated arm carries the frailty of patients a doctor chose not to anticoagulate, and the fracture result is where that shows.
The cognitive difference amounts to slightly more than 0.2 MMSE points a year [5]. Hold that same gap for five years and it comes to slightly more than a single point [17]. "The difference is modest for an individual patient from one year to the next, but over a longer period even such an effect could influence how cognitive function develops," said Nanbo Zhu, a researcher in the same department at Karolinska Institutet [11][13].
The Karolinska summary states the death, stroke, clot, fracture and bleeding results as directions of risk and does not give hazard ratios or confidence intervals [19]. Funding is listed as coming from, among others, the Swedish Research Council, the Swedish Brain Foundation, CIMED, ALF project funding and Karolinska Institutet funds [14]. Eriksdotter has been a consultant in one-off meetings with BioArctic AB, Roche, Eli Lilly, Biogen/Eisai and Novo Nordisk, and has given lectures at symposia sponsored by Roche and BioArctic/Eisai [15].
What to watch
- Group sizes, confidence intervals and hazard ratios in the full European Heart Journal paper.
- Whether the cognitive gap survives an on-treatment analysis, given that some patients changed anticoagulant during follow-up.
- Whether another national dementia registry, or a trial with a cognitive endpoint, reproduces the NOAC-warfarin difference.