Predicting longitudinal brain atrophy in Parkinson's disease using a Susceptible-Infected-Removed agent-based model.

Parkinson's disease is a progressive neurodegenerative disorder characterized by accumulation of abnormal isoforms of alpha-synuclein. Alpha-synuclein is proposed to act as a prion in Parkinson's disease: In its misfolded pathologic state, it favors the misfolding of normal alpha-synuclein molecules, spreads trans-neuronally, and causes neuronal damage as it accumulates. This theory remains controversial. We have previously developed a Susceptible-Infected-Removed (SIR) computational model that simulates the templating, propagation, and toxicity of alpha-synuclein molecules in the brain. In this study, we test this model with longitudinal MRI collected over 4 years from the Parkinson's Progression Markers Initiative (1,068 T1 MRI scans, 790 Parkinson's disease scans, and 278 matched control scans). We find that brain deformation progresses in subcortical and cortical regions. The SIR model recapitulates the spatiotemporal distribution of brain atrophy observed in Parkinson's disease. We show that connectome topology and geometry significantly contribute to model fit. We also show that the spatial expression of two genes implicated in alpha-synuclein synthesis and clearance, SNCA and GBA, also influences the atrophy pattern. We conclude that the progression of atrophy in Parkinson's disease is consistent with the prion-like hypothesis and that the SIR model is a promising tool to investigate multifactorial neurodegenerative diseases over time.

Supervised Phenotype Discovery From Multimodal Brain Imaging.

Data-driven discovery of image-derived phenotypes (IDPs) from large-scale multimodal brain imaging data has enormous potential for neuroscientific and clinical research by linking IDPs to subjects' demographic, behavioural, clinical and cognitive measures (i.e., non-imaging derived phenotypes or nIDPs). However, current approaches are primarily based on unsupervised approaches, without the use of information in nIDPs. In this paper, we proposed a semi-supervised, multimodal, and multi-task fusion approach, termed SuperBigFLICA, for IDP discovery, which simultaneously integrates information from multiple imaging modalities as well as multiple nIDPs. SuperBigFLICA is computationally efficient and largely avoids the need for parameter tuning. Using the UK Biobank brain imaging dataset with around 40,000 subjects and 47 modalities, along with more than 17,000 nIDPs, we showed that SuperBigFLICA enhances the prediction power of nIDPs, benchmarked against IDPs derived by conventional expert-knowledge and unsupervised-learning approaches (with average nIDP prediction accuracy improvements of up to 46%). It also enables the learning of generic imaging features that can predict new nIDPs. Further empirical analysis of the SuperBigFLICA algorithm demonstrates its robustness in different prediction tasks and the ability to derive biologically meaningful IDPs in predicting health outcomes and cognitive nIDPs, such as fluid intelligence and hypertension.

Identifying microstructural changes in diffusion MRI; How to circumvent parameter degeneracy.

Biophysical models that attempt to infer real-world quantities from data usually have many free parameters. This over-parameterisation can result in degeneracies in model inversion and render parameter estimation ill-posed. However, in many applications, we are not interested in quantifying the parameters per se, but rather in identifying changes in parameters between experimental conditions (e.g. patients vs controls). Here we present a Bayesian framework to make inference on changes in the parameters of biophysical models even when model inversion is degenerate, which we refer to as Bayesian EstimatioN of CHange (BENCH). We infer the parameter changes in two steps; First, we train models that can estimate the pattern of change in the measurements given any hypothetical direction of change in the parameters using simulations. Next, for any pair of real data sets, we use these pre-trained models to estimate the probability that an observed difference in the data can be explained by each model of change. BENCH is applicable to any type of data and models and particularly useful for biophysical models with parameter degeneracies, where we can assume the change is sparse. In this paper, we apply the approach in the context of microstructural modelling of diffusion MRI data, where the models are usually over-parameterised and not invertible without injecting strong assumptions. Using simulations, we show that in the context of the standard model of white matter our approach is able to identify changes in microstructural parameters from conventional multi-shell diffusion MRI data. We also apply our approach to a subset of subjects from the UK-Biobank Imaging to identify the dominant standard model parameter change in areas of white matter hyperintensities under the assumption that the standard model holds in white matter hyperintensities.

Accurate predictions of individual differences in task-evoked brain activity from resting-state fMRI using a sparse ensemble learner.

Modelling and predicting individual differences in task-fMRI activity can have a wide range of applications from basic to clinical neuroscience. It has been shown that models based on resting-state activity can have high predictive accuracy. Here we propose several improvements to such models. Using a sparse ensemble learner, we show that (i) features extracted using Stochastic Probabilistic Functional Modes (sPROFUMO) outperform the previously proposed dual-regression approach, (ii) that the shape and overall intensity of individualised task activations can be modelled separately and explicitly, (iii) training the model on predicting residual differences in brain activity further boosts individualised predictions. These results hold for both surface-based analyses of the Human Connectome Project data as well as volumetric analyses of UK-biobank data. Overall, our model achieves state of the art prediction accuracy on par with the test-retest reliability of task-fMRI scans, suggesting that it has potential to supplement traditional task localisers.

Brain atrophy in prodromal synucleinopathy is shaped by structural connectivity and gene expression.

Isolated REM sleep behaviour disorder (iRBD) is a synucleinopathy characterized by abnormal behaviours and vocalizations during REM sleep. Most iRBD patients develop dementia with Lewy bodies, Parkinson's disease or multiple system atrophy over time. Patients with iRBD exhibit brain atrophy patterns that are reminiscent of those observed in overt synucleinopathies. However, the mechanisms linking brain atrophy to the underlying alpha-synuclein pathophysiology are poorly understood. Our objective was to investigate how the prion-like and regional vulnerability hypotheses of alpha-synuclein might explain brain atrophy in iRBD. Using a multicentric cohort of 182 polysomnography-confirmed iRBD patients who underwent T1-weighted MRI, we performed vertex-based cortical surface and deformation-based morphometry analyses to quantify brain atrophy in patients (67.8 years, 84% male) and 261 healthy controls (66.2 years, 75%) and investigated the morphological correlates of motor and cognitive functioning in iRBD. Next, we applied the agent-based Susceptible-Infected-Removed model (i.e. a computational model that simulates in silico the spread of pathologic alpha-synuclein based on structural connectivity and gene expression) and tested if it recreated atrophy in iRBD by statistically comparing simulated regional brain atrophy to the atrophy observed in patients. The impact of SNCA and GBA gene expression and brain connectivity was then evaluated by comparing the model fit to the one obtained in null models where either gene expression or connectivity was randomized. The results showed that iRBD patients present with cortical thinning and tissue deformation, which correlated with motor and cognitive functioning. Next, we found that the computational model recreated cortical thinning (r = 0.51, P = 0.0007) and tissue deformation (r = 0.52, P = 0.0005) in patients, and that the connectome's architecture along with SNCA and GBA gene expression contributed to shaping atrophy in iRBD. We further demonstrated that the full agent-based model performed better than network measures or gene expression alone in recreating the atrophy pattern in iRBD. In summary, atrophy in iRBD is extensive, correlates with motor and cognitive function and can be recreated using the dynamics of agent-based modelling, structural connectivity and gene expression. These findings support the concepts that both prion-like spread and regional susceptibility account for the atrophy observed in prodromal synucleinopathies. Therefore, the agent-based Susceptible-Infected-Removed model may be a useful tool for testing hypotheses underlying neurodegenerative diseases and new therapies aimed at slowing or stopping the spread of alpha-synuclein pathology.

Differentially targeted seeding reveals unique pathological alpha-synuclein propagation patterns.

Parkinson's disease is a progressive neurodegenerative disorder characterized by the intracellular accumulation of insoluble alpha-synuclein aggregates into Lewy bodies and neurites. Increasing evidence indicates that Parkinson's disease progression results from the spread of pathologic alpha-synuclein through neuronal networks. However, the exact mechanisms underlying the propagation of abnormal proteins in the brain are only partially understood. The objective of this study was first to describe the long-term spatiotemporal distributions of Lewy-related pathology in mice injected with alpha-synuclein preformed fibrils and then to recreate these patterns using a computational model that simulates in silico the spread of pathologic alpha-synuclein. In this study, 87 2-3-month-old non-transgenic mice were injected with alpha-synuclein preformed fibrils to generate a comprehensive post-mortem dataset representing the long-term spatiotemporal distributions of hyperphosphorylated alpha-synuclein, an established marker of Lewy pathology, across the 426 regions of the Allen Mouse Brain Atlas. The mice were injected into either the caudoputamen, nucleus accumbens or hippocampus, and followed over 24 months with pathologic alpha-synuclein quantified at seven intermediate time points. The pathologic patterns observed at each time point in this high-resolution dataset were then compared to those generated using a Susceptible-Infected-Removed (SIR) computational model, an agent-based model that simulates the spread of pathologic alpha-synuclein for every brain region taking simultaneously into account the effect of regional brain connectivity and Snca gene expression. Our histopathological findings showed that differentially targeted seeding of pathological alpha-synuclein resulted in unique propagation patterns over 24 months and that most brain regions were permissive to pathology. We found that the SIR model recreated the observed distributions of pathology over 24 months for each injection site. Null models showed that both Snca gene expression and connectivity had a significant influence on model fit. In sum, our study demonstrates that the combination of normal alpha-synuclein concentration and brain connectomics contributes to making brain regions more vulnerable to the pathological process, providing support for a prion-like spread of pathologic alpha-synuclein. We propose that this rich dataset and the related computational model will help test new hypotheses regarding mechanisms that may alter the spread of pathologic alpha-synuclein in the brain.

Network topology of the marmoset connectome.

The brain is a complex network of interconnected and interacting neuronal populations. Global efforts to understand the emergence of behavior and the effect of perturbations depend on accurate reconstruction of white matter pathways, both in humans and in model organisms. An emerging animal model for next-generation applied neuroscience is the common marmoset (Callithrix jacchus). A recent open respository of retrograde and anterograde tract tracing presents an opportunity to systematically study the network architecture of the marmoset brain (Marmoset Brain Architecture Project; http://www.marmosetbrain.org). Here we comprehensively chart the topological organization of the mesoscale marmoset cortico-cortical connectome. The network possesses multiple nonrandom attributes that promote a balance between segregation and integration, including near-minimal path length, multiscale community structure, a connective core, a unique motif composition, and multiple cavities. Altogether, these structural attributes suggest a link between network architecture and function. Our findings are consistent with previous reports across a range of species, scales, and reconstruction technologies, suggesting a small set of organizational principles universal across phylogeny. Collectively, these results provide a foundation for future anatomical, functional, and behavioral studies in this model organism.

Local vulnerability and global connectivity jointly shape neurodegenerative disease propagation.

It is becoming increasingly clear that brain network organization shapes the course and expression of neurodegenerative diseases. Parkinson disease (PD) is marked by progressive spread of atrophy from the midbrain to subcortical structures and, eventually, to the cerebral cortex. Recent discoveries suggest that the neurodegenerative process involves the misfolding and prion-like propagation of endogenous α-synuclein via axonal projections. However, the mechanisms that translate local "synucleinopathy" to large-scale network dysfunction and atrophy remain unknown. Here, we use an agent-based epidemic spreading model to integrate structural connectivity, functional connectivity, and gene expression and to predict sequential volume loss due to neurodegeneration. The dynamic model replicates the spatial and temporal patterning of empirical atrophy in PD and implicates the substantia nigra as the disease epicenter. We reveal a significant role for both connectome topology and geometry in shaping the distribution of atrophy. The model also demonstrates that SNCA and GBA transcription influence α-synuclein concentration and local regional vulnerability. Functional coactivation further amplifies the course set by connectome architecture and gene expression. Altogether, these results support the theory that the progression of PD is a multifactorial process that depends on both cell-to-cell spreading of misfolded proteins and regional vulnerability.