Razvan Valentin Marinescu

Multiple Orderings of Events in Disease Progression

The event-based model constructs a discrete picture of disease progression from cross-sectional data sets, with each event corresponding to a new biomarker becoming abnormal. However, it relies on the assumption that all subjects follow a single event sequence. This is a major simplification for sporadic disease data sets, which are highly heterogeneous, include distinct subgroups, and contain significant proportions of outliers. In this work we relax this assumption by considering two extensions to the event-based model: a generalised Mallows model, which allows subjects to deviate from the main event sequence, and a Dirichlet process mixture of generalised Mallows models, which models clusters of subjects that follow different event sequences, each of which has a corresponding variance. We develop a Gibbs sampling technique to infer the parameters of the two models from multi-modal biomarker data sets. We apply our technique to data from the Alzheimers Disease Neuroimaging Initiative to determine the sequence in which brain regions become abnormal in sporadic Alzheimers disease, as well as the heterogeneity of that sequence in the cohort. We find that the generalised Mallows model estimates a larger variation in the event sequence across subjects than the original event-based model. Fitting a Dirichlet process model detects three subgroups of the population with different event sequences. The Gibbs sampler additionally provides an estimate of the uncertainty in each of the model parameters, for example an individuals latent disease stage and cluster assignment. The distributions and mixtures of sequences that this new family of models introduces offer better characterisation of disease progression of heterogeneous populations, new insight into disease mechanisms, and have the potential for enhanced disease stratification and differential diagnosis.

Progression of regional grey matter atrophy in multiple sclerosis

See Stankoff and Louapre (doi:10.1093/brain/awy114) for a scientific commentary on this article. Grey matter atrophy in multiple sclerosis affects certain areas preferentially. Eshaghi et al. use a data-driven computational model to predict the order in which regions atrophy, and use this sequence to stage patients. Atrophy begins in deep grey matter nuclei and posterior cortical regions, before spreading to other cortical areas.

An image‐based model of brain volume biomarker changes in Huntington's disease

Determining the sequence in which Huntingtons disease biomarkers become abnormal can provide important insights into the disease progression and a quantitative tool for patient stratification. Here, we construct and present a uniquely fine‐grained model of temporal progression of Huntingtons disease from premanifest through to manifest stages.

Uncovering the heterogeneity and temporal complexity of neurodegenerative diseases with Subtype and Stage Inference

The heterogeneity of neurodegenerative diseases is a key confound to disease understanding and treatment development, as study cohorts typically include multiple phenotypes on distinct disease trajectories. Here we introduce a machine-learning technique—Subtype and Stage Inference (SuStaIn)—able to uncover data-driven disease phenotypes with distinct temporal progression patterns, from widely available cross-sectional patient studies. Results from imaging studies in two neurodegenerative diseases reveal subgroups and their distinct trajectories of regional neurodegeneration. In genetic frontotemporal dementia, SuStaIn identifies genotypes from imaging alone, validating its ability to identify subtypes; further the technique reveals within-genotype heterogeneity. In Alzheimer’s disease, SuStaIn uncovers three subtypes, uniquely characterising their temporal complexity. SuStaIn provides fine-grained patient stratification, which substantially enhances the ability to predict conversion between diagnostic categories over standard models that ignore subtype (p = 7.18 × 10−4) or temporal stage (p = 3.96 × 10−5). SuStaIn offers new promise for enabling disease subtype discovery and precision medicine.Progressive diseases tend to be heterogeneous in their underlying aetiology mechanism, disease manifestation, and disease time course. Here, Young and colleagues devise a computational method to account for both phenotypic heterogeneity and temporal heterogeneity, and demonstrate it using two neurodegenerative disease cohorts.

MECHANISTIC PROFILES OF NEURODEGENERATION: A STUDY IN ALZHEIMER’S DISEASE, HEALTHY AGEING AND PRIMARY PROGRESSIVE MULTIPLE SCLEROSIS

Figure 1. Study design schematic. (a): we construct the average structural connectome and the metrics to assess the hypothetical mechanisms, based on Human Connectome Project data; (b): we estimate the temporal progression of neurodegeneration, using a disease progression model (DPM, Lorenzi et al, 2017) on Tl-MRI volumes from our cohorts (6670 longitudinal measurements of scans from 1713 ADNI subjects (415 controls, 960 MCI, 338 AD); 11593 longitudinal scans from 5389 ageing subjects from the Rotterdam Study Scan data (Ikram et al, 2015); 244 longitudinal scans from 64 subjects acquired at the Institute of Neurology, UCL (20 controls, 44 P-MS); (c): the model estimates the mechanistic profiles b. Poster Presentations: Tuesday July 24, 2018 P1280

CHARACTERISING THE PROGRESSION OF ALZHEIMER’S DISEASE SUBTYPES USING SUBTYPE AND STAGE INFERENCE (SUSTAIN)

ures (A)-(D) show the progression pattern of each of the four subtypes estimated by SuStaIn. The cumulative probability each region has reached a particular z-score is shown for different stages along the progression; the cumulative probability of a region going from a z-score of 0-sigma to 1-sigma ranges from 0 in white to 1 in red, the cumulative probability of a region going from a z-score of 1-sigma to 2sigma ranges from 0 in red to 1 in magenta, and the cumulative probability of a region going from a z-score of 2-sigma to 3-sigma ranges from 0 in magenta to 1 in blue, f is the proportion of subjects assigned to each subtype. CVS is the model cross-validation similarity: the average similarity of the subtype progression patterns across cross-validation folds, measured using the Bhattacharyya coefficient. The CVS ranges from 0 (no similarity) to 1 (maximum similarity). Poster Presentations: Monday, July 17, 2017 P791

DATA-DRIVEN MODELS OF DISEASE PROGRESSION AND APPLICATIONS TO ALZHEIMER’S DISEASE: EVENT-BASED MODEL AND DIFFERENTIAL EQUATION MODELS OF BIOMARKER CHANGES IN ADNI

improve statistical power. Conclusions:Harmonizing datasets from different sites is vital in increasing the power in detecting early onset of AD. Our framework estimates the transformation needed to match the two sites’ distributions, and then assigns a p-value via a new hypothesis test. This procedure can be easily deployed in practice (open-source code). We show one setting where clinical thresholds for detecting disease groups defined on ADNI can be adapted to other datasets.

MULTIPLE DISTINCT ATROPHY PATTERNS FOUND IN GENETIC FRONTOTEMPORAL DEMENTIA USING SUBTYPE AND STAGE INFERENCE (SUSTAIN)

Test), motor function (UPDRS) and disease duration. Results:There was significantly increased BPND in the basal ganglia (putamen) and the occipital lobe (cuneus) (p1⁄40.004 and p1⁄40.007, respectively: Mann-Whitney U) in DLB compared to the control group. The BPND within the cuneus correlated positively with ACE-R scores (p1⁄40.017, Spearman R, r1⁄40.59) and negatively with disease duration (p1⁄40.016, Spearman R, r1⁄4-0.59). Therewas no significant correlation with motor scores. Conclusions: Our results show increased [C]PK11195 binding consistent with microglial activation in brain regions known to be affected in DLB, suggesting that neuroinflammation can be demonstrated in vivo. Although its significance remains unclear, the positive correlation with ACE-R scores and negative correlation with disease duration suggests that microglial activation is elevated in early disease and diminishes in the later stages. This suggests that either inflammation is protective in DLB, or that the potential immunotherapeutic window is narrow and early in disease. References: (1) Fan Z, et al. Influence of microglial activation on neuronal function in Alzheimer’s and Parkinson’s disease dementia. Alzheimers Dement 2014;:1–14. http://dx.doi.org/10.1016/j.jalz.2014.06.016; (2) Mackenzie IR. Activated microglia in dementia with Lewy bodies. Neurology 2000;55:132–4. IC-P-079 MULTIPLE DISTINCTATROPHY PATTERNS FOUND IN GENETIC FRONTOTEMPORAL DEMENTIA USING SUBTYPE AND STAGE INFERENCE (SUSTAIN) Alexandra L. Young, Razvan Valentin Marinescu, Neil P. Oxtoby, Martina Bocchetta, David M. Cash, David L. Thomas, Katrina M. Dick, M. Jorge Cardoso, Sebastien Ourselin, John C. van Swieten, Barbara Borroni, Daniela Galimberti, Mario Masellis, Maria Carmela Tartaglia, James B. Rowe, Caroline Graff, Fabrizio Tagliavini, Giovanni B. Frisoni, Robert Laforce Jr., Elizabeth Finger, Alexandre Mendonca, Sandro Sorbi, Nick C. Fox, Jonathan M. Schott, Jonathan D. Rohrer, Daniel C. Alexander and Genetic FTD Initiative (GENFI), University College London, London, United Kingdom; Department of Computer Science and Centre forMedical Image Computing, UCL, London, United Kingdom; UCL Institute of Neurology, London, United Kingdom; Translational Imaging Group, Centre for Medical Image Computing, Department of Medical Physics and Biomedical Engineering, UCL, London, United Kingdom; Neuroradiological Academic Unit, Department of Brain Repair and Rehabilitation, UCL Institute of Neurology, London, United Kingdom; Erasmus MC, Rotterdam, Netherlands; University of Brescia, Brescia, Italy; University of Milan, Milan, Italy; University of Toronto, Toronto, ON, Canada; University of Cambridge, Cambridge, United Kingdom; Karolinska Institutet, Stockholm, Sweden; Fondazione IRCSS Istituto Neurologico Carlo Besta, Milano, Italy; Lab Alzheimer’s Neuroimaging and Epidemiology, IRCCS Fatebenefratelli, Brescia, Italy; Universite Laval, Quebec, QC, Canada; University of Western Ontario, London, ON, Canada; University of Lisbon, Lisbon, Portugal; University of Florence, Florence, Italy; Dementia Research Centre, Institute of Neurology, University College London, London, United Kingdom. Contact e-mail: alexandra.young@ucl.ac.uk

ANALYSIS OF THE HETEROGENEITY OF POSTERIOR CORTICAL ATROPHY: DATA-DRIVEN MODEL PREDICTS DISTINCT ATROPHY PATTERNS FOR THREE DIFFERENT COGNITIVE SUBGROUPS

Figure 1b. Positional variance diagram for the vision subgroup showing the uncertainty in the progression of volume loss (biomarkers on y axis) across different stages (positions on x axis). Each entry (x,y) represents the probability of region y becoming abnormal at position x in the sequence, ranging from 0 in white to 1 in black. Ram J. Bishnoi, Sandarsh Surya, W. Vaughn McCall, Augusta University, Augusta, GA, USA. Contact e-mail: rbishnoi@augusta.edu

LONGITUDINAL EVALUATION OF NEUROPSYCHOLOGICAL AND NEUROIMAGING PROGRESSION IN POSTERIOR CORTICAL ATROPHY

Background: fNIRS is expected as a non-invasive and unconstrained neuroimaging modality used to measure activationinduced changes in cerebral hemoglobin concentration. By this technique, changes in the optical absorption of light over the region of interest on the brain during the task are measured to estimate the local cerebral vascular and oxygen metabolic effects so that the task design is most important factor for successful evaluation of brain functions. The objective of this study is to establish the most suitable fNIRS task which is easily to be executed in elderly subjects and indicates significant oxygenated Hb (Oxy-Hb) changes in Alzheimer’s disease (AD) patients with less cognitive ability. Methods:We designed eight kinds of tasks for fNIRS measurement for the elderly. The task is related to detect memory and sensate function, such as touching, hearing, and smelling. We investigated the adaptable task for AD subject that is easily to execute, to understand, and to obtain enough Oxy-Hb changes during task duration. fNIRS measurement was performed employing 54 channels that covered bilaterally over the frontal and parietal area from 22 healthy control subjects (between ages of 60 to 82) and 17 AD subjects (between ages of 60 to 78). A five trials of block design was applied for data acquisition. Results: Among eight tasks, cold stimulation and calculation tasks were simple and understandable tasks and most suitable for fNIRS measurement in AD. From the fNIRS recording data, while, in healthy elder subjects, the typical oxy-Hb concentration change patterns showing significant increase in the local area of the left prefrontal cortex during cold stimulation task in the right hand, AD patients showed significant decrease compared to the healthy control. Conclusions: We have investigated most suitable fNIRS tasks which is easily to be executed for eldered AD subjects, and differences in OxyHb changes by task activation between normal control and AD. fNIRS might have the potential to become either a disease or syndrome-specific diagnostic tool in the future.