The Box Task: A novel tool to differentiate the primary progressive aphasias

Objective Differentiating the primary progressive aphasia (PPA) variants in clinical settings remains complex and challenging, especially for the logopenic (lv-PPA) and non-fluent variants (nfv-PPA). Recent studies suggest that visuospatial memory is more compromised in lv-PPA than in nfv-PPA and is relatively spared in the semantic variant (sv-PPA). Accordingly, assessment of visuospatial memory performance may assist in the differential diagnosis of PPA variants. Here, we investigated the utility of a novel computerised visuospatial working memory test-the Box Task-to differentiate the three PPA variants and typical Alzheimer's disease (AD). Methods Eighteen lv-PPA, 14 nfv-PPA, 23 sv-PPA, 33 AD patients, and 32 healthy controls matched for age and education were recruited. All participants completed the computerised Box Task and WMS-III Spatial Span as measures of visuospatial working memory. Results The lv-PPA group made significantly more Box Task between-search errors than nfv-PPA, sv-PPA and control groups. The AD group, however, displayed the greatest impairments on this measure relative to the PPA variants. Logistic regression analyses in lv-PPA and nfv-PPA demonstrated that the combination of Box Task between-search error variables (i.e., 4- and 6-box levels) could correctly classify 72% of lv-PPA patients and nearly 79% of nfv-PPA patients. Area under the receiver operator characteristics curve (AUC) analyses revealed the Box Task was more sensitive than Spatial Span at differentiating lv-PPA from nfv-PPA. Conclusions Our findings suggest that a simple, computerised measure of visuospatial working memory-the Box Task-shows potential diagnostic utility in differentiating lv-PPA from the other PPA variants.

Automated summary

The computerised visuospatial working memory test, the Box Task, shows potential diagnostic utility in differentiating primary progressive aphasia (PPA) variants, with lv-PPA patients performing worse than nfv-PPA patients on this task. Logistic regression analyses demonstrated that the Box Task could correctly classify a significant proportion of lv-PPA and nfv-PPA patients. Area under the receiver operator characteristics curve (AUC) analyses revealed that the Box Task was more sensitive than Spatial Span at differentiating lv-PPA from nfv-PPA.