#5
Mario Tranfa

A Connectome in Motion: Capturing Structural Changes Across the Alzheimer’s Continuum

White matter is increasingly recognized as an active contributor to the pathophysiology of Alzheimer’s disease (AD), rather than simply a passive bystander. It constitutes a dynamic substrate that both shapes disease progression and reflects the underlying
pathophysiological changes occurring throughout the AD continuum. In this talk, I will discuss how advanced diffusion MRI provides a unique window into these evolving processes, from preclinical stages to post-mortem validation.

I will argue that structural connectivity is continuously reconfigured through the interplay of genetic susceptibility, proteinopathy, vascular injury, and neurodegeneration. White matter alterations emerge early in the disease course and exhibit distinct biological signatures, allowing AD-related changes to be disentangled from those driven by cerebrovascular pathology. At the same time, the impact of genetic risk is not uniform across the brain: biological pathways linked to lipid metabolism, immune activation, and protein clearance differentially influence regional white matter vulnerability and modulate the brain’s response to
amyloid and tau pathology.

Beyond reflecting disease burden, the connectome provides the structural scaffold that supports disease evolution. Brain network architecture influences the spatial progression of pathological protein accumulation, contributing to the emergence of distinct
disease trajectories. In parallel, alterations in neuromodulatory systems, such as the noradrenergic network, moderate the
relationship between structural connectivity and cortical neurodegeneration, influencing mechanisms of trophic support and
neuronal resilience.

Finally, I will show how post-mortem MRI-histopathology studies provide biological validation for diffusion MRI biomarkers, demonstrating that changes in tissue microstructure correspond to demyelination and neuroaxonal injury in both white and grey matter. These observations refine the interpretation of diffusion-derived measures and strengthen their translational value.

Together, these findings support a unified view of the structural connectome as a system in motion: genetically shaped, biologically responsive, and continuously reorganized across the Alzheimer’s continuum. Understanding this dynamic interplay offers new opportunities to improve disease characterization, identify biologically meaningful subtypes, and develop imaging biomarkers that bridge molecular pathology with large-scale brain network organization.

#6
Elise Saul

A unified connectome framework reveals divergent mechanisms of atrophy across dementia syndromes

Introduction: Neurodegeneration is increasingly conceptualized as a network-constrained process shaped by the brain’s structural and functional architecture. However, it remains unknown whether a universal “spreading rule” applies across different
pathologies. We tested a unified framework to determine how connectome-based mechanisms (i.e., epicenter proximity, hub vulnerability, and diffusion-like spread) differentially explain regional atrophy across five clinically and molecularly distinct syndromes: early-onset Alzheimer’s disease (EOAD), behavioral variant of frontotemporal dementia (bvFTD), posterior cortical
atrophy (PCA), dementia with Lewy bodies (DLB) and vascular dementia (VaD).

Methods: We analyzed T1-weighted MRI from patients with EOAD (n=105), bvFTD (n=36), PCA (n=27), DLB (n=40), VaD (n=20), and amyloid–tau-negative controls (n=406) from the Alzheimer Centrum Amsterdam. Demographics are summarized in Table 1. Control-referenced atrophy maps were computed and modeled using connectivity-derived predictors capturing epicenter proximity, hub vulnerability, and diffusion-like propagation based on structural and functional connectivity, with intracortical connectivity included as a control covariate to account for non-network spatial/geometric effects. Cross-validated regression quantified explained variance. FDR-corrected pairwise contrasts were used for between-syndrome predictor profile comparisons.

Results: Connectome-informed models revealed non-uniform mechanistic profiles. EOAD showed a heterogeneous and strongly network-dependent profile dominated by functional epicenter proximity and structural nodal strength. bvFTD was also strongly network-dependent and mixed, yet dominated by structural and functional diffusion. PCA displayed the most restricted pattern,
primarily driven by structural diffusion. In contrast, DLB was distinguished by structural and functional nodal strength. Structural epicenter proximity and diffusion-like spread dominated the VaD profile. Pairwise group comparisons confirmed these dissociations, with significant differences across diffusion-, epicenter-, and nodal-strength predictors, supporting syndrome-level
separation of mechanistic profiles.

Discussion: Our findings demonstrate that neurodegeneration does not follow a single generic propagation rule across
neurodegenerative diseases. Instead, each syndrome engages the connectome through distinct “mechanistic signatures”: EOAD as
a mixed profile dominated by combined hub susceptibility and propagation from vulnerable anchors, bvFTD as distributed network spread dominated by diffusion, PCA as focal and anatomically constrained progression, DLB as hub-centered vulnerability, and VaD as predominantly anatomically constrained structural spread. This highlights that while the connectome provides the pathways for progression, the underlying pathology dictates the dominant mechanism of traversal.

#7
Gerard Temprano-Sagrera

White matter hyperintensities mediate the association between CSF proteomics and cognitive decline in
Alzheimer’s disease

Background: White matter hyperintensities (WMH) are neuroimaging markers of cerebrovascular damage linked to cognitive decline in Alzheimer’s disease (AD). However, the molecular mechanisms underlying this relationship are not fully understood. We investigated diagnosis-dependent cerebrospinal fluid (CSF) proteomic associations with WMH burden and tested whether WMH mediate the relationship between specific proteomic signatures and longitudinal cognitive decline.

Methods: We studied 699 ADNI participants with baseline WMH quantification and baseline SomaScan 7k platform CSF proteomics (Fig. 1). Cross-sectional protein–WMH associations were evaluated in diagnosis-stratified models and whole-sample models including protein×diagnosis interaction terms, adjusted for age, sex, intracranial volume, years of education, and APOE- ε4 status. Functional enrichment analyses were performed on the top-100 associated proteins in the whole sample and within each diagnostic group. Candidate proteins for mediation analyses were selected based on consistent evidence across multivariable models and stringent multiple-testing correction. We conducted mediation analyses with WMH as the mediator between each candidate protein and longitudinal cognitive performance measured by the Mini-Mental State Examination (MMSE).

Results: WMH–proteomic associations showed clear diagnosis-dependence. The mild cognitive impairment (MCI) group exhibited the strongest and most biologically coherent signal, with significant functional enrichment in pathways related to neuronal projection, nervous system development, and synaptic organisation (Fig. 2). Mediation analyses provided exploratory evidence that WMH partially mediates the association between baseline levels of RTN4RL1, CD274, and NPTXR and longitudinal cognitive decline, explaining 13.5–24% of the total effect (Fig. 3). For all three proteins, lower levels were associated with higher WMH burden, which in turn predicted greater cognitive decline.

Conclusions: CSF proteomic associations with WMH are diagnosis-dependent, with MCI showing the strongest links to WMH burden. Although this may partly reflect sample size, consistent results support the robustness of the associations and their biological meaning. These findings support a stage-specific interplay between CSF proteomic signatures and WMH, highlighting MCI as a critical phase where vascular and neurodegenerative processes converge to drive cognitive decline.

#8
Alessandra Griffa

Brain Connectivity to Understanding Inter-Individual Variability in Tau Pathology and Disease Progression in
Alzheimer’s Disease

Background: Network models of Alzheimer’s disease propose that tau pathology propagates through large-scale brain
connectivity. However, existing studies primarily evaluate spatial correspondence between tau burden and connectome organization, providing limited insight into the mechanisms through which network architecture shapes inter-individual patterns of
tau accumulation.

Methods: We investigated the relationship between tau PET patterns and brain connectivity using two complementary connectomics frameworks. First, individual tau SUVR maps were projected onto connectome harmonics derived from structural
connectivity, functional connectivity, partial functional connectivity, or Euclidean distance. This yielded subject-specific tau spectra describing the connectivity-frequency content of tau pathology. Spectral features included spectral exponential decay fitting (apparent diffusion coefficient) and energy within low-, intermediate-, and high-frequency bands. Second, we evaluated whether communication models operating on the structural connectome explain individual whole-brain tau distributions. Communication
efficiency from disease epicenters was computed using multiple communication strategies, including shortest paths, navigation, search information, and communicability. Subject-level explanatory power was quantified as the association between
communication efficiency and regional tau burden.

Results: Across the Alzheimer’s disease continuum, tau spectra progressively shifted toward low-frequency connectome harmonics
and increasingly exhibited diffusion-like spectral organization. Spectral biomarkers were associated with cognitive impairment,
Braak staging, and future cognitive decline, outperforming global tau burden measures. In parallel, communicability consistently provided the strongest explanation of individual tau accumulation patterns, exceeding alternative communication models,
monosynaptic structural connectivity, and functional connectivity. Communication-based explainability increased from cognitively
unimpaired amyloid-negative individuals to amyloid-positive cognitively unimpaired participants, patients with mild cognitive impairment, and dementia. Females exhibited stronger tau-communicability coupling than males.

Conclusions: Convergent evidence from spectral and communication-based analyses indicates that tau accumulation is progressively constrained by the communication architecture of the human connectome. Polysynaptic communication appears to
govern propagation, while functional network modes best characterize the resulting spatial organization. Together, these findings support diffusion-like mechanisms of tau propagation and demonstrate that connectivity-informed representations of tau
PET provide clinically meaningful biomarkers that capture inter-individual variability and heterogeneous tau-connectivity coupling
levels.