Longitudinal evaluation of common and unique brain-networks in variants of primary progressive aphasia

The two variants of primary progressive aphasia (PPA) with Frontotemporal dementia pathology- semantic (svPPA) and non-fluent (nfvPPA) exhibit both shared and distinct features. The phenotypic heterogeneity stems from alterations in underlying brain networks. Investigating the common network (CN) and variant-specific unique network (UN) is critical to understanding the pathology of these conditions. Additionally, examining the evolution of these networks is key to elucidating disease progression.

We applied algebraic-topology to explore the CN and UN using individualised resting-state fMRI data from 31 patients with nfvPPA, 32 with svPPA, and 38 age- and sex-matched controls (scanned at first visit). Using persistent homology, we identified the networks that differentiated (p < 0.01) each patient group from healthy controls. Since the subset of these subjects was scanned at the 6th and 12th months, the longitudinal changes in the rsfMRI networks were evaluated at each interval. Network features were correlated with clinical behaviours, and the longitudinal impact of the changes in these networks on behaviours was evaluated over the 12-month period. To validate the rsfMRI networks and the longitudinal changes, we evaluated the grey matter (GM) volume, GM atrophy and the rate-of-atrophy of the brain areas. To corroborate the findings, we applied persistent homology on the structural networks derived from diffusion tensor images.

We found the existence of a left lateralised functional network identical in both PPA groups. This CN, comprising regions associated with language and cognition, remained stable over time (12 months period) and was associated with the severity of dementia. Conversely, the right-dominant UN in both variants showed progressive disintegration annually. In svPPA, cerebellar disassociation led to a decline in daily life activities, while parietal lobe degradation in nfvPPA impaired naming abilities. The CN and UN with similar regions were also found in the structural connectivity and the longitudinal changes in UN aligned with accelerated GM atrophy in the affected regions.

Given the limited availability of pharmacological treatments, rehabilitation in PPA has primarily focused on modulating the left hemisphere using brain stimulation techniques. However, our findings indicate that while the disintegrated left hemispheric CN remained relatively stable, dysconnectivity progressed in the right hemisphere. These observations, along with the phylogenetic organization of brain networks and the variant-specific patterns of progression, highlight the need to incorporate right hemispheric rehabilitation strategies alongside the conventional left-hemispheric approaches in PPA.