Association of Regional Atrophy With Naming Decline in Primary Progressive Aphasia.

BACKGROUND AND OBJECTIVES: Primary progressive aphasia (PPA) is a neurodegenerative condition that predominantly impairs language. Most investigations of how focal atrophy affects language consider 1 time point compared with healthy controls. However, true atrophy quantification requires comparing individual brains over time. In this observational cohort study, we identified areas where focal atrophy was associated with contemporaneous decline in naming in the same individuals. METHODS: Cross-sectional analyses-related Boston Naming Test (BNT) performance and volume in 22 regions of interests (ROIs) at each time point using Least Absolute Shrinkage and Selection Operator (LASSO) regression. Longitudinal analysis evaluated changes in BNT performance and change in volume in the same ROIs. RESULTS: Participants (N = 62; 50% female; mean age = 66.8 ± 7.4 years) with PPA completed the BNT and MRI twice (mean = 343.9 ± 209.0 days apart). In cross-sectional left inferior frontal gyrus pars opercularis, superior temporal pole, middle temporal gyrus, and inferior temporal gyrus were identified as critical for naming at all time points. Longitudinal analysis revealed that increasing atrophy in the left supramarginal gyrus and middle temporal pole predicted greater naming decline, as did female sex and longer intervals between time points. DISCUSSION: Although cross-sectional analyses identified classic language areas that were consistently related to poor performance at multiple time points, it was not increasing atrophy in these areas that lead to further decline: longitudinal analysis of each person's atrophy over time instead identified nearby but distinct regions where increased atrophy was related to decreasing performance. The results demonstrate that directly examining atrophy (in each individual) over time furthers understanding of decline in PPA and reveal the importance of left supramarginal gyrus and middle temporal pole in maintaining naming when areas normally critical for language degenerate. The novel results provide insight into how the underlying disease progresses to result in the clinical decline in naming, the deficit most common among all 3 PPA variants.

Enhanced Imaging and Language Assessments for Primary Progressive Aphasia.

BACKGROUND AND OBJECTIVES: It is widely agreed that primary progressive aphasia (PPA) is a clinical syndrome with at least 3 distinct variants that differ in phenotype, areas of atrophy, and most common underlying neurodegenerative disease. The distinction between logopenic variant PPA (lvPPA) and other variants is important for prognosis and medical management. However, differentiating logopenic from nonfluent agrammatic variant can be difficult. We aimed to identify a novel behavioral assessment that (1) distinguishes logopenic from the other variants with a high degree of accuracy and (2) correlates with left superior temporal-inferior parietal atrophy (previously shown to be associated with this variant). The location of atrophy was measured using a novel, clinically useful imaging analysis. METHODS: In this observational cohort study of 227 individuals with PPA, participants were administered sentence reading and repetition subtests from a standard battery. A subset of 41 participants were administered enhanced reading and repetition subtests with 5 longer sentences, of which 13 had brain imaging. Ratios of word-level and sentence-level accuracy in reading:repetition were calculated. We used one-way analysis of variance (ANOVA) to determine whether either or both ratios of reading:repetition independently discriminated between variants and t test to determine whether the ratios distinguished between nonfluent and logopenic variants. We identified receiver operating characteristics and Pearson correlations between the reading:repetition ratios and ratio of left:right superior temporal-inferior parietal volume. RESULTS: The sentence reading:repetition ratio using the new stimuli significantly differed across the 3 variants (p < 0.00001) and differed between nonfluent and logopenic variants (t(27) = 4.1; p = 0.0003). The area under the curve for distinguishing logopenic from other variants was 0.85 (0.71-0.99), and the diagnostic accuracy was 87.5%. The sentence reading:repetition ratio correlated with left:right superior temporal-inferior parietal volume (r = 0.69; p = 0.0087), but not with left:right volume of regions of interest associated with other variants. DISCUSSION: Results provide an efficient and reliable clinical assessment, and a novel imaging analysis, to distinguish the clinical syndrome of logopenic variant from other variants of PPA. Results also support the proposal that lvPPA reflects a defect in phonological short-term memory caused by atrophy in the superior temporal-inferior parietal cortex. CLASSIFICATION OF EVIDENCE: This study provides Class III evidence that the sentence reading:repetition ratio distinguished logopenic PPA from other PPA variants.