Background and objectives: Current formulations of primary progressive aphasia (PPA) derive largely from English-speaking patients. We hypothesized that language-specific characteristics influence PPA phenotypes in 2 contrasting languages: Italian and English.
Methods: We undertook a retrospective, cross-sectional, observational comparison of 2 patient cohorts representing all major PPA syndromes, in London and Florence. Neuropsychological scores in a range of linguistic and general cognitive domains were normalized to native speaker controls and dichotomized as impaired/unimpaired. Proportions were compared using χ2 tests and adjusted for symptom duration and severity of cognitive impairment using logistic regression.
Results: The cohorts comprised 106 (48.1% female) Italian speakers (14 nonfluent/agrammatic [nfvPPA], 20 semantic [svPPA], 41 logopenic variant [lvPPA], 31 mixed PPA [mPPA]) and 166 (45.2% female) English speakers (70 nfvPPA, 45 svPPA, 42 lvPPA, 9 mPPA). Comparing cohorts, the English cohort was younger (mean 62.7 [SD = 8.4] vs 65.9 [7.8] years; p = 0.003, Cohen d = 0.39), with longer symptom duration (4.6 [4.3] vs 3.1 [2.5] years; p = 0.048, d = 1.08), a higher proportion of nfvPPA cases (42% vs 13%, χ2 = 25.4, p < 0.001), and lower proportions of lvPPA (25% vs 38%, χ2 = 5.46, p = 0.019) and mPPA (5% vs 29%, χ2 = 29.3, p < 0.001). English-speaking nfvPPA patients had less frequent expressive agrammatism (46% vs 93%, p = 0.015, odds ratio [OR] 16.05, 95% CI 1.70-151.13) but more frequently impaired single-word comprehension (60% vs 8%, p = 0.013, OR 0.06, 95% CI 0.00-0.56). English svPPA patients had more frequent surface dyslexia (68% vs 30%, p = 0.046, OR 0.24, 95% CI 0.06-0.97) and dysgraphia (38% vs 10%, p = 0.021, OR 0.09, 95% CI 0.01-0.70) while English lvPPA patients had more frequently impaired single-word comprehension (89% vs 29%, p < 0.001, OR 0.05, 95% CI 0.01-0.28), word repetition (61% vs 26%, p = 0.020, OR 0.24, 95% CI 0.07-0.80), nonword repetition (78% vs 30%, p = 0.010, OR 0.18, 95% CI 0.05-0.06), nonverbal working memory (69% vs 36%, p = 0.005, OR 0.10, 95% CI 0.02-0.51), and visuomotor function (89% vs 25%, p < 0.001, OR 0.02, 95% CI 0.01-0.20).
Discussion: Language-specific characteristics influenced PPA phenotypes, with more frequent expressive agrammatism in Italian (reflecting its morphologic complexity), more frequently impaired word processing in English (reflecting its articulatory, acoustic, and orthographic complexity), and increased prevalence of mPPA in Italian. These findings have implications for PPA diagnosis and management. Limitations of test heterogeneity and cohort size should be addressed in future, prospective, multicenter initiatives using cross-linguistic tools.