|
SUMMARY
The three-year preDiCT project commenced in June 2008 and over the first 12 months has made excellent progress both scientifically and administratively in establishing itself as a collaborative community.
The project has had extensive scientific outputs at major conferences and in well respected journals, and has disseminated its work through conference and workshop presentations. The project has made major progress in solidifying its links with the pharmaceutical industry through close interaction both with original industrial project partners and active engagement with new representatives who have subsequently joined our Scientific Advisory Board. We will continue this engagement through our specialist Pharma workshop in October 2009.
Taking each work-package in turn, we can highlight the key activities and challenges over the first year:
WP1: Management
Despite initial delays, we have executed a mutually acceptable signed consortium agreement between the project partners. Recruitment was slower than expected across the board but it was imperative that we recruited the very best people to undertake the proposed work. We are confident that we have an excellent team to take forward this project and are grateful to those who supported and provided input without charge to achieving the early goals. Whilst finance and administration are running smoothly, we have had to recruit cover (Alexandra Masindova) to continue the work of Katherine Fletcher whilst she is on maternity leave to ensure continuity. Some institutions are currently below forecasted budgets but this has had no impact on scientific progress and several partners have contributed (very substantial) additional effort at no cost to the project.
WP2: Dissemination
The project website was set up during the first month of the project (http://www.vph-predict.eu) and this enables the public to review the project and its deliverables, as well as allowing project members to share information between themselves. Initiatives like the ‘science blog’ and the ‘pharma log’ have been used to collate details of interactions and scientific activity. In addition, we have produced a newsletter in electronic and printed form and this will be repeated this year.
Wider scientific dissemination has been achieved through scientific conferences, invited talks, a Chaste (our open source cardiac modelling software) dissemination and training workshop and presentations at the NOE Industry Event.
WP3: Ion channel models
The aim of this work package (WP3) is to validate and improve ion channel models that can reproduce with a high degree of fidelity the experimental measurements of drug effects in healthy and diseased heart.
Ion channel models for IKr and INa for rabbit and human have been validated against the latest experimental data. The existing models of these currents were analysed; specifically, the most recent Hodgkin–Huxley formulations for the INa, i.e. the Shannon–Mahajan model for rabbit, and the ten Tusscher model for human, were considered. An important improvement to address was the formulation of the late sodium current (INaL), which was lacking in these two models. Regarding the validation of IKr models for rabbit and human, a similar procedure was followed. In this case, the following models were addressed: Shannon–Mahajan and ten Tusscher for the HH formulation and several Markov models following the Kiehn and Wang formulation.
We examined the parameter identifiability and analysed the existing Ca2+ model structures (comparing phenomenological non-common pool deterministic and local control approaches) with the aim to develop an optimized Ca2+ subsystem model. These studies will provide pharma companies with additional insight into Ca2+ homeostasis and hence help to choose which kinetics and/or mechanisms the drug should target to achieve a desired effect.
During the first year, we have developed two models of drug action on ionic channels. One of them, lidocaine, affects mainly the INa current. The second one, dofetilide, is a potent and specific blocker of IKr.
WP4: Inter-species comparison
The developed software tools provide a framework for model development, i.e.,a) analysis of the relation of model parameters, structure and experimental data; b) fitting of the model parameters; and c) validation of the model. These steps are necessary to obtain models with predictive power. The software has already been used to show that currently used models are heavily over-parameterised and to improve the experimental protocol for voltage-gated ion channels.
Further model analysis has helped to identify the shortcomings of a human ventricular cell model in predicting relevant biomarkers, indicating that model refinement is essential. The software is also used for investigating the calcium subsystem which is hard to access experimentally as many processes are interacting with each other via feedback loops and therefore it is even more difficult to assess the necessary experimental data for fitting the models. Preliminary results again indicate overparameterisation of models, but as the tools also provide information on possible model simplifications the development of an improved model for the calcium subsystem is underway.
WP5: Computational tools and methodologies
We successfully delivered the Requirements and Technology Assessment report in August 2008. This describes both the VRE and whole-heart simulator. Following this, a benchmarking suite was developed enabling progress to be evaluated in the development of a modelling environment suitable for high-performance computing.
We have seen substantial improvements in performance of the whole-heart simulator Chaste (June 2008 – March 2009). preDiCT benefited from input from a UK EPSRC project which provided resources to improve parallel scaling of Chaste. Improvements were made to assembly, solvers, output, parallel load balancing, as well as optimisation of CellML. In collaboration with Imperial College London, we have performed an initial incorporation of adaptive meshing (March–May 2009). We have also commenced a study into preconditioners and solvers.
WP6: Data and knowledge management integration
By working with the scientific work packages in preDiCT, we have developed a database schema for publications, protocols (biological material, population, target additional elements) and tested compounds. The description includes 18 tables and 29 glossaries/thesaurus. We have also validated the initial content with WP7.
We currently have an implementation of the database on a test server in Aureus Pharma and we are now developing the data capture application. We have also developed a prototype of the glossary management application.
We have populated the database with content with the list of drugs and selected publications complete, and data acquisition in progress.
WP7: New biomarkers for drug cardiac toxicity
During the first year of the project, we have developed computationally efficient models of rabbit and human cardiac electrophysiology from ion channel to ECG (in collaboration with Oxford BBSRC-3D heart project and Chaste project). We have performed a simulation study of the effect of HERG block on the ECG using rabbit and human ventricular models and developed signal processing techniques to investigate the ability of T wave morphology biomarkers to characterise sotalol effects on the clinical ECG.
Progress has been made in the characterisation of the role of ion channel stochastic gating properties in LQT2 (HERG block)-induced arrhythmic risk and, in collaboration with WP3, we have investigated the impact of biological variability on cellular biomarkers of arrhythmic risk.
In collaboration with USZ, we have evaluated ventricular heart rate adaptation as a biomarker of arrhythmic risk and we have reviewed ionic and cellular biomarkers of arrhythmic risk. Biomarkers based on ECG are currently under review.
In progressing the science in WP7, the team have also discussed the use of tools from other projects in the VPH community including the use of GIMIAS for image analysis and visualisation.
Conclusions
Whilst progress to date has been good, we have also faced challenges which are typical of collaborative multi-institution projects. Communication has been vitally important, both regular and considered to ensure inclusivity for all partners. We have recognised the importance of face to face meetings in making progress, and are looking forward to the second year of the project where we will also increase our outward facing activity.
We are able to simulate different electrical stimulation patterns, both apical and endocardial. Electrical stimulation of the ventricular muscle in vivo occurs, for the most part, within the apical portion of the endocardial surface, where the Purkinje fibres enter the myocardium.
Top panel (right) shows the simulation of the initiation of an activation sequence from the endocardial surface whereas the top left figure shows an activation sequence initiated from the apex of the heart.
The bottom panel shows the ability to capture cellular inhomogeneities of the ventricular wall. Cells in different layers posses different intrinsic properties in terms of ion channels conductance and action potential duration. We are able to characterize different cell properties in different layers of the cardiac wall and prescribe, for each layer, distinct ion channels properties. This allows for a more reliable simulation of the repolarisation sequence that is of primary importance for clinical evaluation through the ECG. |
