Atrophy network mapping of clinical subtypes and main symptoms in frontotemporal dementia

Abstract

Frontotemporal dementia (FTD) is a disease of high heterogeneity, apathy and disinhibition present in all subtypes of FTD and imposes a significant burden on families/society. Traditional neuroimaging analysis has limitations in elucidating the network localization due to individual clinical and neuroanatomical variability. The study aims to identify the atrophy network map associated with different FTD clinical subtypes and determine the specific localization of the network for apathy and disinhibition. Eighty FTD patients [45 behavioural variant FTD (bvFTD) and 35 semantic variant progressive primary aphasia (svPPA)] and 58 healthy controls at Xuanwu Hospital were enrolled as Dataset 1; 112 FTD patients including 50 bvFTD, 32 svPPA and 30 non-fluent variant PPA (nfvPPA) cases, and 110 healthy controls from the Frontotemporal Lobar Degeneration Neuroimaging Initiative (FTLDNI) dataset were included as Dataset 2. Initially, single-subject atrophy maps were defined by comparing cortical thickness in each FTD patient versus healthy controls. Next, the network of brain regions functionally connected to each FTD patient's location of atrophy was determined using seed-based functional connectivity in a large (n = 1000) normative connectome. Finally, we used atrophy network mapping to define clinical subtype-specific network (45 bvFTD, 35 svPPA and 58 healthy controls in Dataset 1; 50 bvFTD, 32 svPPA, 30 nfvPPA and 110 healthy controls in Dataset 2) and symptom-specific networks [combined Datasets 1 and 2, apathy without depression versus non-apathy without depression (80:26), disinhibition versus non-disinhibition (88:68)]. We compare the result with matched symptom networks derived from patients with focal brain lesions or conjunction analysis. Through the analysis of two datasets, we identified heterogeneity in atrophy patterns among FTD patients. However, these atrophy patterns are connected to a common brain network. The primary regions affected by atrophy in FTD included the frontal and temporal lobes, particularly the anterior temporal lobe. bvFTD connects to frontal and temporal cortical areas, svPPA mainly impacts the anterior temporal region and nfvPPA targets the inferior frontal gyrus and precentral cortex regions. The apathy-specific network was localized in the orbital frontal cortex and ventral striatum, while the disinhibition-specific network was localized in the bilateral orbital frontal gyrus and right temporal lobe. Apathy and disinhibition atrophy networks resemble known motivational and criminal lesion networks, respectively. A significant correlation was found between the apathy/disinhibition scores and functional connectivity between atrophy maps and the peak of the networks. This study localizes the common network of clinical subtypes and main symptoms in FTD, guiding future FTD neuromodulation interventions.

Keywords: apathy; atrophy network map; disinhibition; frontotemporal dementia.

Figure 1

Workflow of individual brain atrophy mapping and network mapping. (A) Using Freesurfer, the cortical thickness of healthy controls is computed. A standard general linear model (GLM) for cortical thickness is generated based on age and sex. Each patient’s cortical thickness (CT) is compared to the standard model to generate a vertex-wise map of cortical atrophy, referred to as the individual brain atrophy map. (B) The individual brain atrophy map is mapped onto a functional MRI dataset of 1000 healthy individuals. This generates an individual atrophy network map, indicating the functional regions connected with the patient’s atrophy map. The network map overlap among patients indicates the percentage of patients whose atrophy is functionally connected to the same regions. HC = healthy control.

Figure 2

Atrophy network mapping results in FTD patients. (A) The percentage of overlap (w-score < −2) in the brain atrophy maps for each participant. (B) The percentage of overlap in the brain atrophy network maps for patients in the same locations. (C) The voxel-wise t-test comparisons of brain atrophy network maps between patients and healthy controls (10 000 simulations, voxel-wise FWE-corrected P < 0.05). Dataset 1: [A(i)–C(i)] frontotemporal dementia (FTD); [A(ii)–C(ii)] behaviour variant FTD (bvFTD); [A(iii)–C(iii)] semantic variant primary progressive aphasia (svPPA). Dataset 2: [A(iv)–C(iv)] FTD; [A(v)–C(v)] bvFTD; [A(vi)–C(vi)] svPPA; [A(vii)–C(vii)] non-fluent variant PPA (nfvPPA).

Figure 3

Brain atrophy network mapping of apathy/disinhibition in FTD. The atrophy network map for apathy (10 000 simulations, voxel-wise FWE-corrected P < 0.05), [A(i)] was similar to the network map obtained using peak coordinates from a conjunction analysis of motivation-related regions [A(ii)], with a spatial correlation of R = 0.83 [A(iii)]. The atrophy network map for disinhibition (10 000 simulations, voxel-wise FWE-corrected P < 0.05) [B(i)] was similar to the criminal lesion network map [B(ii)] with a spatial correlation of R = 0.71 [B(iii)]. FTD = frontotemporal dementia; FWE = family-wise error.

Figure 4

Differences in functional connectivity strength between peak location of symptom-related brain network mapping and brain atrophy maps. Comparison of functional connectivity between the frontotemporal dementia (FTD) brain atrophy map and peak location of apathy brain network mapping [A(i)] or motivation-related brain network mapping [A(ii)]. The bar graph represents the differences between the apathy and non-apathy groups. Regression analysis of the functional connectivity mentioned above and Neuropsychiatric Inventory (NPI) apathy severity scores [A(iii)] or Frontal Behavioral Inventory (FBI) apathy scores [A(iv)]. Comparison of functional connectivity between the FTD brain atrophy map and peak location of disinhibition brain network mapping [B(i)] or criminal lesion network mapping [B(ii)]. The bar graph represents the differences between the disinhibition and non-disinhibition groups. Regression analysis of the functional connectivity mentioned above and NPI disinhibition severity scores [B(iii)] or FBI disinhibition scores [B(iv)].