Modeling pathological spread through the structural connectome in the FTLD spectrum

Abstract Background To explore the relationship between network vulnerability and longitudinal atrophy progression in FTLD spectrum, using Network Diffusion Model (NDM) of pathology spread. Method Thirty‐four behavioural‐variant frontotemporal dementia (bvFTD), 11 semantic‐variant primary progressive aphasia (svPPA) and 11 nonfluent/agrammatic‐variant primary progressive aphasia (nfvPPA) patients underwent longitudinal T1‐weighted MRI. Forty‐eight young controls underwent multi‐shell diffusion MRI scan. NDM was developed to assess the progression of FTLD pathology originating from a seed (right orbitofrontal cortex [bvFTD], left inferior temporal gyrus [svPPA], and left supplementary motor area [nfvPPA]) and proceeding through the healthy connectome. The connectivity measures were fractional anisotropy (FA) and intra‐cellular volume fraction (ICVF). Correlations were tested between atrophy estimated by NDM and those empirically obtained in FTLD patients over a follow‐up of 24 months. Result In bvFTD, NDM showed an early spread to frontal lobe and basal‐ganglia and to right sensorimotor, parietal, temporal and occipital lobes, with a subsequent involvement of the left hemisphere. In svPPA, NDM suggested an early spread of pathology to the left occipital and frontal lobes, while parietal lobe was reached later. In nfvPPA, NDM‐predicted spread through all brain regions, except for occipital lobe. NDM‐predicted atrophy of each region was positively correlated to longitudinal atrophy empirically observed in all three FTLD variants. Overall, NDM with ICVF connectome provided higher correlation values relative to NDM with FA maps. Conclusion The NDM implementation to cross‐sectional structural connectome is a valuable tool to predict atrophy patterns and pathology spreading in FTLD variants. Supported‐by: European‐Research‐Council (StG‐2016_714388_NeuroTRACK); Foundation Research on Alzheimer’s Disease.