Researchers tracking a defined population of 767,356 residents in Salento, southern Italy, identified 35 new cases of primary progressive aphasia (PPA) over four calendar years, an overall crude incidence of 1.14 per 100,000 person-years (95% confidence interval 0.79 to 1.59). The study, published on 1 May 2026 in Alzheimer's Research & Therapy, ran prospectively from 1 January 2021 to 31 December 2024 across the province of Lecce, with referrals routed to a coordinating centre in Tricase where neurologists and a senior speech and language pathologist adjudicated each diagnosis against consensus criteria, using 3 Tesla MRI and language-focused testing.

The most consequential finding for families is the biomarker layer. Cerebrospinal fluid analysis, amyloid PET, or both were available for 30 of the 35 cases (85.7%), and 20 of those 30 (66.7%) were positive for Alzheimer's disease pathology. By subtype, positivity was 15 of 15 for the logopenic variant, 1 of 6 tested for the nonfluent/agrammatic variant, 1 of 5 tested for the semantic variant, and 3 of 4 unclassified cases. Logopenic PPA was the commonest subtype at 15 cases (0.49 per 100,000 person-years), ahead of nonfluent/agrammatic PPA at 10 cases (0.33), semantic variant PPA at 6 (0.20) and unclassified presentations at 4 (0.13). Incidence peaked at ages 60 to 64 in men and 75 to 79 in women.

PPA is usually described from specialist clinic series, which cannot say how often it arises in a general population. Only a handful of counts exist with a real denominator: a retrospective population-based study in Olmsted County, Minnesota, covering 2011 to 2022 and published in Neurology in 2024, reported a combined incidence of PPA and primary progressive apraxia of speech of 0.41 per 100,000 person-years. Differences in method, period, case-finding and whether apraxia of speech is counted separately make direct comparison between the two figures unreliable.

The counterpoint sits in the arithmetic. Thirty-five cases spread over four subtypes produce wide confidence intervals: the semantic variant estimate, for example, rests on six people. Case ascertainment depended on people with progressive language problems reaching a neurologist and being referred into the network, so any who did not are not in the count. Biomarker status was established by fluid or imaging measures, not by autopsy, and five cases had no biomarker data at all.