Narrative discourse in logopenic variant of PPA: a multi-level approach

Objectives: Ipofluency, phonological errors and mild agrammatism are the microlinguistic markers of the logopenic variant-Primary Progressive Aphasia (lvPPA); it is less clear if patients may also experience pragmatic difficulties in the construction of coherent narratives. The present study aimed to describe narrative discourse impairment in patients with lvPPA. Materials: 18 lvPPA patients (mean age 74,56 \ub1 7,91; 11 males, 7 females) and 18 matched healthy controls (HCs) were enrolled. Speech samples were recorded and transcribed. Method: Connected speech was analyzed according to Marini et al.\u2019s criteria (2011) focusing on microlinguistic (speech rate; well-formed sentence ratio; principal/subordinate clause ratio; closed/open class word ratio; % phonemic and % semantic paraphasias; % fragments) and macrolinguistic (mean length of utterance (MLU); % cohesive errors; % local and global coherence errors, informativeness) measures. A preliminary two-step cluster analysis was conducted to confirm the clustering of HCs and lvPPA patients on the whole performance. A Principal Component Analysis (PCA) was then performed to identify factors explaining measures of micro and macrolinguistic deficits on patients only; factor scores were entered into a cluster analysis to evaluate the presence of different subgroups. Differences between subgroups were tested by one-way ANOVAs. Results: lvPPA patients were significantly different from HCs in both micro and macrolinguistic measures. The PCA performed on patients only (Bartlett\u2019s \u3c72 = 111.83, p < .0001) showed a first factor accounting for production of % semantic paraphasias, well-formed sentence ratio, MLU, closed/open class words ratio; a second factor accounting for production of % cohesive errors, principal/subordinate clause ratio, informativeness and a third factor for the production of % global coherence errors. The two-step cluster analysis on factorial scores showed a 3-cluster solution. A first cluster (2 cases) had lower speech rate, higher % phonemic and semantic paraphasias, lower % global coherence errors and lower informativeness compared with the second (7 cases); a third cluster (9 cases) had higher % global coherence errors and lower informativeness than the second. Discussion: The cluster analysis displayed subgroups of patients with both micro and macrolinguistic distinctive profiles. Disorders of macrolinguistic components in lvPPA are consistent with a more diffuse cognitive impairment (extending to short-term memory, semantic memory and episodic memory) due to the subtending Alzheimer\u2019s type pathology. Conclusion: Our findings confirm that, at variance with semantic dementia and primary nonfluent aphasia, lvPPA does not correspond to a specific aphasic variant and that also pragmatic features may differentiate subgroups