Today, diseases are still defined largely based on signs and symptoms, yet while two patients may share the same diagnosis, the underlying causes of their symptoms may be very different. Naturally, this means that a treatment that works in one patient may prove ineffective in another. There is now broad acceptance that a new direction for disease classification is needed, built not on symptomatology, but derived from the mechanisms, or causes, that drive the disease.
Prototype taxonomy and a knowledge base
The AETIONOMY team started by tackling the problem of how to obtain, organise, structure, integrate and interpret a broad range of data (ranging from molecular data, to information on symptoms) from the neurodegeneration community. They then ‘dissected’ the underlying mechanistic causes in order to bring structure to their classification; linking the mechanistic causes to clinical evidence in an attempt to validate them.
The consortium demonstrated that their prototype taxonomy can be used to identify patient subgroups in PD. They were able to perform a partial validation of a selection of candidate mechanisms for both diseases. The validation was done in-silico and by means of dedicated wet lab experiments performed by clinical research, biotech and pharmaceutical partners.
AETIONOMY resulted in an open-access knowledge base with inventories of mechanistic hypotheses; these form the basis for the prototype taxonomies for both AD and PD. The knowledge base also contains curated clinical and relevant OMICS-data (analysis of complete genetic or molecular profiles), disease models for AD and PD, analysis and visualisation, and data from participants of a cross-sectional clinical study and additional integrated cohorts. The neurodegenerative research community now has access to the inventory of computable models, as well as a dedicated set of algorithms that can facilitate the interrogation of the mechanistic hypotheses against patient data.
Virtual data cohorts
AETIONOMY recruited a significant cohort of patients and controls, and coordinated the logistics of sample transfers and execution of a range of biomarker assays. The patient cohorts were profiled and characterised to a degree that allowed the consortium to perform ‘mechanism-enrichment’ procedures on patient-level biomarker data.
To overcome hurdles related to accessing patient data to carry out in-silico validation of disease mechanisms, the consortium developed the concept of virtual data cohorts (VDCs), which are artificial data sets that share features and characteristics of real-world study cohorts. VDCs have the potential to overcome some of the challenges inherent to translational neurodegeneration research, namely the sharing of patient-level data without compromising patient data privacy (but also other challenges, like merging heterogeneous, complex clinical data sets and more).
Through this conceptualisation and implementation of VDCs, the AETIONOMY project may pave the way for future data sharing and data integration of patient-level data without compromising patient privacy. The virtual patient topic was not a novel concept invented by the AETIONOMY consortium but is in line with globally emergent trends.
Collaboration and synergies
The project brought together 18 partners including pharmaceutical companies, universities, and patient groups, and combined substantial expertise in neurodegenerative diseases, molecular biology, clinical research, research ethics and law, neuroimaging, data modelling and simulation, data standards, and patient engagement in research among others.
Collaborations, beyond the original scope of the project but initiated through the identification of common goals, were key for the project’s success; in particular, collaborations with the University of Oxford and the Human Brain Project. Working with Alzheimer Europe, AETIONOMY had the valuable opportunity to network with leading European researchers in the field and to help ensure that the perspectives of people with dementia were taken into consideration.
The consortium is still (partially) alive
Part of the AETIONOMY consortium is still working together; Fraunhofer SCAI (Institute for Algorithms and Scientific Computing) in Germany and the ICM (Institute for Brain and Spinal Cord) in France have collaborated since the official end of the funded period. Based on an initial workshop in Bonn on longitudinal modelling of disease progression, the collaborators decided to make the workshop a regular event that reaches out to other projects at EU level, including the IMI project RADAR-AD and the Horizon 2020 initiative, the Virtual Brain Cloud. Work in AETIONOMY has thus led to a major collaborative effort to understand the dynamics of neurodegenerative disease progression. Other collaborative efforts that have been sustained concern the topic of information extraction for model generation, where Frauenhofer SCAI have partnered with the University of Luxembourg (currently applied to the context of COVID-19).