Discourse in primary progressive aphasia: Longitudinal changes and neural correlates

BackgroundPrimary progressive aphasia (PPA) is a neurodegenerative brain condition with three variants - nonfluent (nfvPPA), logopenic (lvPPA), semantic (svPPA). At onset, nfvPPA shows atrophy in left inferior frontal gyrus with grammatical impairment; lvPPA involves left temporo-parietal junction, affecting phonological processing; and svPPA impacts left anterior temporal pole, impairing semantic processing. Agrammatism is a hallmark of nfvPPA, but grammatical symptoms have been reported across variants in cross-sectional studies. Here, discourse skills and cortical integrity are evaluated longitudinally for a large sample of cases to understand variant-specific mechanisms of change. MethodParticipants were 27 individuals with nfvPPA (M = 66.6 years {+/-} 8.3), 30 lvPPA (M = 66.7 {+/-} 7.3), 33 svPPA (M = 64.8 {+/-} 6.7), and 36 healthy controls (HC; M = 65.5 {+/-} 6.8). Discourse language samples from descriptions of the Cookie Theft picture were analysed for word density and diversity, sentence complexity, well-formedness, and fluency annually for up to three time points. Structural MRI brain scans were collected for analysis of cortical thickness. ResultsAt baseline, nfvPPA performed more poorly than other groups on most measures. lvPPA were differentiated from svPPA on fluency measures only. Longitudinally, utterance length declined in all variants. For nfvPPA, this was linked with reduced sentence complexity and cortical thickness in regions engaged by tasks of higher attentional demand. For lvPPA, it was linked with increasing grammatical errors and atrophy extending broadly into perisylvian language network. No associations were identified in svPPA. ConclusionsThis study is the largest to date to explore longitudinal changes in discourse and associated changes in cortical thickness, across the three PPA variants. In nfvPPA and lvPPA, grammatical changes reflect extension of the pathology both beyond and within the left perisylvian language network, respectively. Findings inform assessment, prognosis, and intervention to support communication throughout the disease course.