Primary Progressive Aphasia
The latest sourced reporting on Primary Progressive Aphasia from Dementias Today: dated, and plain about what the evidence does and does not show.
A narrative review published on October 20 inventories where generative AI already touches Alzheimer's work. The uses furthest along are the ones no family ever sees: synthetic notes, phenotype mining from visit records, and drafted summaries. None of it has been tested in a live clinic.
Twenty-five people with nonfluent variant primary progressive aphasia, a disease with no approved treatment, took an experimental pill for twelve weeks. A blood marker of nerve damage fell. The company that makes the drug reported it. The finding is real, and it is also the first rung of a very long ladder.
DemMA, published in the Findings of ACL 2026, puts a simulated person with dementia in the chair so trainees can practise the conversation: nine subtypes, nonverbal cues, 2,709 synthetic dialogues, and not one real patient in the data.
A data-driven cascade model in 489 patients with bvFTD, primary progressive aphasia, progressive supranuclear palsy, corticobasal syndrome and FTD-ALS places plasma GFAP first and steep clinical decline last. The work is observational and its authors present it as a framework for staging and trial design, not as a test for any individual.
An Amsterdam-led preprint finds polygenic scores for developmental dyslexia running high in semantic variant primary progressive aphasia, and only in the men. It is the first genetic hint behind the one risk factor PPA leaflets have always named, and it changes nothing anyone can do today.
An observational study of 58 people with semantic variant primary progressive aphasia found the cognitive reserve effect was confined to semantic tasks, and reversed over time
A 2025 meta-analysis pooled ten randomised trials of transcranial direct current stimulation in primary progressive aphasia. General naming and spelling improved. Everything else measured did not. Families asking about home devices deserve the pooled figure, not the best anecdote.
A Madrid trial gave 63 people with primary progressive aphasia half a year of theta-burst stimulation and word therapy. It is the longest sham-controlled test the syndrome has had. Its main result was a difference in brain metabolism so small the researchers themselves will not say what it means for a life.
A 184-patient records review published in Neurology on August 5 puts the exclusion rate at 70 to 85 percent, with brief cognitive screens the single biggest gate. A 13-patient Japanese series of primary progressive aphasia, worked up with a formal aphasia battery, reached a milder conclusion. Neither study treated anyone.
Northwestern researchers put 92 recorded fairy tales through machine learning to guess whether Alzheimer's or a 4R tauopathy is behind a person's primary progressive aphasia. The models worked. The paper's own discussion explains why that is not yet a test.
A prospective population study in the province of Lecce found 35 incident cases of primary progressive aphasia between 2021 and 2024, and every logopenic case tested carried Alzheimer's biomarkers.
Erasmus MC sequenced the temporal cortex of 18 people with FTLD-TDP type C, the pathology that almost always underlies semantic variant PPA, and found postsynaptic gene activity reduced after adjusting for tissue and disease overlap. The work is descriptive, not causal.