Mapping the neural correlates of the effect of psycholinguistic variables on picture naming performance: a FDG-PET study across neurodegenerative diseases

Background Picture naming performance is influenced by the properties of the stimuli and of the words to be retrieved, such as word length and lexical frequency. Significant inconsistencies, however, remain regarding the brain regions mediating these effects in neurodegenerative patients. In the present study, we addressed this issue by correlating regional cerebral metabolism with several naming-related variables in a large cohort of neurodegenerative patients, who are likely to exhibit naming impairments due to different mechanisms of cognitive dysfunction. Methods A total of 178 patients classified within the Frontotemporal (FTD) and Alzheimer’s Disease (AD) spectra were administered a picture naming test validated for the Italian language (CaGi) and underwent a FDG-PET scan. Principal Component Analysis on 10 psycholinguistic variables resulted in the extraction of four components, labelled as word-form, visual, lexical, and semantic, according to the variables populating each of them. Using an item-level approach, the influence of each component on patients' performance was assessed and correlated with brain metabolism data from 11 left hemispheric Regions of Interest. Results A simple word form and lexical structure were associated with better naming performance. The imaging findings reveal a distributed neural network, with fusiform gyrus supporting both visual and semantic features. Inferior frontal and posterior temporal/parietal gyri represented an interface between lexico-semantic and phonological properties. The anterior temporal lobe contributed to all the stages of picture naming. The two dementia spectra activated different areas in response to the same variables, in particular for the visual and semantic components, suggesting the presence of disease-specific compensatory mechanisms. Conclusions Our results suggest a distributed neural network showing both commonalities and specificities in how picture and word properties influence naming performance. The network also seems capable of compensatory changes in the face of the extension of neurodegenerative processes.