POND2018

POND2018_logo

19 February, Geneva

Second International Workshop on Modelling the Progression Of Neurological Disease (POND2018)

Date: 19 February 2018
Location: Hall H8-01, Campus Biotech, Geneva, Switzerland
Sponsors: EuroPOND consortium, the Centre for Medical Image Computing at UCL, and the UK’s EPSRC.
Program: below (PDF version).

Following on from POND2016, the EuroPOND consortium is pleased to host POND2018.

We are excited to have this follow-up opportunity to continue to share the latest methodological advances and applications in the field of computational modelling of neurological disease progression.

Format: Invited presentations and Contributed posters (see below).
Registration cost: Free!
Registration closed on 22 January 2018.

Program

(click on a title to read the talk abstract, where available)

08.30 Registration

09.00 Opening: Petra Huppi and Giovanni Frisoni (EuroPOND)

09.10-10.30 Session 1

09.10 Daniel Alexander and Neil Oxtoby (UCL, UK)
"Updates from the EuroPOND"

Abstract: Latest results emanating from the EuroPOND consortium.

09.30 Tammie Benzinger (Washington University in St Louis, USA)
"Autosomal Dominant Alzheimer Disease as a Model of Neurodegeneration"

Abstract:

10.00 Adam Schwarz (Eli Lilly & Company, Indianapolis, USA)
"Subtypes of Alzheimer disease and individualized disease trajectories"

Abstract: Recent research on heterogeneity in Alzheimer disease (AD) populations will be reviewed. Convergent findings from neuropathology, imaging and cognition point to subtypes of AD with differential biomarker and cognitive profiles. Models to predict individualized disease trajectories may improve clinical trial design and eventually have utility in clinical management.

10.30-10.50 Break

10.50-12.20 Session 2

10.50 Jean-Philippe Thiran (EPFL & CHUV, CH)
"Microstructure brain imaging by diffusion and multi-contrast MRI"

Abstract: Multi-modal MRI spectroscopy is used to unveil the actual contributions of nervous tissue microstructural compartments in a controlled ex vivo experimental setup. Diffusion-weighted and multi-echo spin-echo images are used in a synergetic manner to disentangle properties of myelin, intra/extra-axonal space, and cerebrospinal fluid. Quantitative and robust estimation of these microstructural properties of the brain tissue can lead to the definition of robust biomarkers to assess disease progression.

11.20 Thomas Yeo (NUS, Singapore)
"Latent Factors Underlying Atrophy, Cognitive and Tau Heterogeneity in Alzheimer’s Disease"

Abstract: I will present evidence for at least three factors underlying heterogeneity in Alzheimer&Rsquo;s disease (AD) dementia. The first factor is associated with medial temporal atrophy and deficits in hippocampal-related functions. The second factor is associated with posterior cortical atrophy and deficits in visuospatial executive function. The third factor is associated with left temporal atrophy and deficits in language processing. There might exist a fourth factor associated with subcortical atrophy and the least cognitive impairment. Associations between atrophy patterns and cognitive deficits persist in at-risk non-demented participants. Both atrophy patterns and cognitive deficits were also associated with the pattern of tau deposition early in the disease process.

11.50 Bruno Jedynak (Portland State University, USA)
"Quantitative templates for the progression of Alzheimer’s disease"

Abstract: I will present some recent methodological progress and discuss some opportunities and challenges for our community.
I will discuss and comment the following references:

  • Temporal order of Alzheimer’s disease-related cognitive marker changes in BLSA and WRAP longitudinal studies, by Murat Bilgel, Rebecca L. Koscik, Yang An, Jerry L. Prince, Susan M. Resnick, Sterling C. Johnson, and Bruno M. Jedynak, Journal of Alzheimer’s Disease 59.4 (2017): 1335-1347.
  • The TADPOLE Challenge: tadpole.grand-challenge.org/
  • The Alzheimer’s Disease Modelling Challenge: www.pi4cs.org/qt-pad-challenge
  • Multivariate Statistical Modelling for Capturing the Temporal Evolution of Alzheimer’s Disease, M. Bilgel and B. Jedynak, work in progress.

12.20-13.30 Lunch Break

13.30-15.00 Session 3

13.30 Dimitri Van De Ville (EPFL & UNIGE, CH)
"Dynamics of fMRI brain activity: perspective for disease diagnosis and prognosis"

Abstract: Over the past years, brain mapping techniques have become increasingly computational—largely inspired by approaches from signal processing and network theory—to overcome the shortcomings of voxel-wise detection of task-evoked activity.
Functional connectivity studies have largely contributed to the insight that spontaneous activity carries meaningful information, even during complete “resting state”; i.e., the subject is asked to relax and let his mind wander. Moreover, the intrinsic functional networks that can be identified are altered by almost any brain disease or disorder. Together with minimal patient compliance, these networks are promising candidates for non-invasive biomarkers for diagnosis and prognosis in a single patient, and for facilitating the translation of fMRI into the clinical realm.
Recently, the quest for better understanding of brain dynamics has triggered new ways to approach functional connectivity—conventionally measured as the correlation between time-courses over the whole run; i.e., using time-resolved rather that summarizing measures. In this talk we highlight a few of the recent advances in this promising research direction and put them in perspective for building better, more mechanistic, models of brain function and their potential for disease diagnosis and prognosis.

14.00 Jonas Richiardi (CHUV, CH)
"Network imaging genomics for disease progression modelling of dementia"

Abstract: The network degeneration model hypothesizes that neurodegenerative diseases propagate along trans-synaptic connections. If this holds true, then spatial patterns of disease spread in time can be used to model and predict disease evolution in individual patients.
The availability of high-resolution medical imaging in PET and MRI, together with large-scale availability of genotype information and post-mortem gene expression data means that it is possible to test specific hypotheses about the relationship between imaging findings in-vivo and the underlying putative abnormal molecular mechanisms.
In this talk, we will introduce the field of network imaging genomics, show recent work linking large-scale brain networks with genotype and gene expression in healthy controls and dementia patients, and discuss potential approaches to linking imaging and genomic data with clinical data for clinical outcome prognosis.

14.30 Rapid-fire Poster Power Pitches: Poster presenters to give a 1-minute pitch to advertise their poster.
Presenters

  • Neil Oxtoby (UCL, UK)
  • Djalel Meskaldji (EPFL and UNIGE, CH)
  • Sara Garbarino (UCL, UK)
  • Maura Bellio (UCL, UK)

15.00-15.20 Break

15.20-17.00 Session 4

15.20 Su Li (Cambridge, UK)
Relationship between tau and neuroinflammation in Alzheimer’s disease: reaction rate equation modelling of atrophy and neuropathology measured by &lsqb;<sup>18</sup>F&rsqb;AV1451, &lsqb;<sup>11</sup>C&rsqb;PK11195 and &lsqb;<sup>11</sup>C&rsqb;PIB

Abstract:
Background
In addition to beta-amyloid accumulation, misfolded tau and activated microglia are also present in Alzheimer’s disease (AD). However, the relationship between them is unclear, in particular whether tau is upstream or downstream of neuroinflammation. To investigate these relationships, we used PET imaging and a novel computational model based on reaction rate equation modelling in chemical kinetics.
Methods
Sixty-six subjects (15 AD, 24 Mild Cognitive Impairment (MCI) and 27 similarly aged healthy controls) underwent standardised clinical and neuropsychological assessments followed by dynamic PET using [18F]AV1451 (tau) and [11C]PK11195 (activated microglia) and multimodal 3T MRI. MCI patients also underwent [11C]PIB (beta-amyloid) PET. We compared regional PET binding and grey matter atrophy among AD, MCI and controls, as well as their spatial distribution across different brain areas. Then, we applied the reaction rate modelling to the cross-sectional data, to test the hypothesis that tau precedes neuroinflammation, across different degrees of dementia severity.
Findings
We found increased [18F]AV1451 and [11C]PK11195 binding as well as grey matter atrophy in AD, with a strong spatial overlap among these AD related biomarkers. We demonstrated that baseline [18F]AV1451 predicted the rate of increase in [11C]PK11195 binding (r=0.39, p=0.011) while the opposite (i.e., baseline [11C]PK11195 binding predicting the rate of increase in [18F]AV1451 binding) was not statistically significant (r=0.06, p=0.69).
Interpretation
In line with the current literature, we found significant increases in both tau and neuroinflammation in AD, with co-localisation. Computational modelling of our data is consistent with the hypothesis of tau as an upstream event of neuroinflammation, i.e. the presence and accumulation of tau is followed by an increase microglia activation, rather than vice versa. Our results, though requiring confirmation in longitudinal PET imaging studies, have significant implications for trials targeting tau or inflammation.

15.50 Nophar Geifman (Manchester, UK)
"Patient Stratification and Endophenotypes Discovery from Longitudinal Data"

Abstract: As the focus of medical research shifts towards a precision medicine approach for diagnosis and treatment, the identification of patient subgroups with biomedically distinct and actionable phenotype definitions is becoming increasingly important. Analysis of longitudinal data, where measurements of disease processes or response to treatment are repeatedly taken from individuals over time, can enable discovery of otherwise hidden patterns. Latent Class Mixed Modelling (LCMM), a type of latent class analysis which is gaining popularity in biomedical research, can facilitate the discovery of meaningful and differing patient subgroups with homogeneous patterns of change over time. Here we present the use, and consider the utility, of this unsupervised statistical learning approach for the identification of subgroups of patients demonstrating different patterns of disease progression or response to treatment. This is done in three disease areas: psoriasis, hypertension, and Alzheimer’s disease. The successful application of LCMMs in varied datasets demonstrates the plasticity and versatility of the approach. In all three studies, the LCMM not only enabled the identification of subgroups of patients but also allowed us to relate trajectories of disease progression or response to treatment to clinically relevant measures and characteristics, demonstrating the potential clinical utility of this approach.

16.20 Marco Lorenzi (INRIA, France)
"Probabilistic progression modeling of biological and clinical trials data"

Abstract: …coming soon

16.40 Stanley Durrleman (ICM, France)
"Digital models of Alzheimer’s disease progression"

Abstract: …coming soon

17.00 Closing remarks

18.00 Poster session (A0 portrait size), drinks, buffet

POND2018 Organising Committee:

University of Geneva — Djalel Meskaldji, Petra Hüppi, Giovanni Frisoni
University College London — Neil Oxtoby, Daniel Alexander, Sara Garbarino