We will use existing tools complementarily to demonstrate the utility of a 4-tiered systems medicine approach in AF therapy and treatment planning. The core clinical concept of the SysAFib project relies on:
I. Retrospective validation
of four predictive scores underpinned by various datasets resultant from:
- ECG analysis: forward and inverse modeling and data assimilation,
- Medical image analysis: statistical shape-modeling and analysis,
- Biophysically-based modeling and simulation, and
- Genetic profiling for patient stratification: genetic screening data combined with sophisticated data analysis tools.
As the feasibility of obtaining existent combined training data from identical patient cohorts (via published sources or otherwise) for validation of each of these individual scores is low, each individual retrospective validation study above will rely on its own dataset with the aim of intersection as possible (the clinical datasets are mainly provided by three clinical partners from OUH, MU, and CCT (associate partner); genetic datasets will be mainly provided by an associate partner from University Heart Clinics at UH).
Available toolsets and datasets to be utilized already include:
- Clinical electrophysiological data: (multipolar) ECG (from pre- and post-ablation), including loop recorders, high resolution 12-lead ECG in AF (persistent / paroxysmal) and no AF individuals, Holter monitoring, and intermittent ECG monitoring; electro-anatomical (EA) mapping data from patient ablations (ca. 300 patients);
- Clinical imaging data: atrial strain data and atrial functional measurements; echo, CT, MRI, LGE MRI; trans esophageal echo before ablation procedure and trans thoracic echo before ablation.
- Clinical history data: patient specific-data on lifestyle; traditional risk-assessment scores; outcome data of the treatment (also with long-term follow up, around 5 years).
- Genetic data: population-based genetic biomarkers for AF risk assessment (at least 5-10 SNPs) from a genome wide association study (GWAS) with 15k people (ca. 450 with AF).
- Simulation and modeling: Structural models; forward electrophysiological models of activation and arrhythmia; ablation/line simulation; and fibrosis modeling.
- Advanced data analysis: ECG signal processing/analysis; image based analysis of atrial structure and shape; advanced data analysis/machine learning tools for robust and accurate integration of heterogeneous data with uncertainty quantification.
II. Proof-of-concept demonstrator study
of a combined risk score for the four modalities outlined in I above on ~30 patients in 3 clinical centers (OUH, MU, and CCT). The patient cohort will be those in early persistent AF, and ablation strategy will follow aligned protocols in all three centers, beginning with pulmonary vein ablation (PVI) procedure as a well-proscribed therapeutic option for these patients and a commonly used technique with reasonable associated standards.
III. Design of a clinical trial for final combined risk score validation.
Although a full clinical trial is out-of-scope for the SysAFib demonstrator project, our focus will be on a clinical trial design for involved and additional centers which can start enrollment following project close.
Given this targeted approach and rich datasets, to be complemented by the final pilot study, our specific scientific and technological objectives are to:
- Extend the extant pipeline for patient-specific model of human ventricles to human atria as based on local technologies and existing research.
- Define strategies for patient stratification via population-based advanced genetic data analysis.
- Utilize expertise developed within the consortium in the field of advanced image-based and signal-based (ECG) analysis
- Validate electrophysiological and mechanical 3D models of the human atria via an in-depth retrospective clinical study.
- Translate findings and methodologies in patient-specific atrial modeling to the clinic via utilization of developed models to simulate patient ablation scenarios.
- Integrate these diverse sources of existing clinical data with emerging biomarkers developed elsewhere together with the validated patient-specific atrial modeling tools in a software environment for clinical decision support using the advances from data analysis. The core of SysAFib will be a retrospective study of a well-phenotyped patient cohort, to serve as a validation and proof-of-concept for our systems medicine approach.
- Demonstrate the utility and expected impact of the developed system via real-life proof-of-concept clinical system testing and deployment in at least 3 clinical centers across Europe.
Clinical concept. Asking whether ablation is (a) the right treatment for a patient and (b) if AF is likely to recur, is essentially the same inquiry posed pre- and post-ablation. Both questions aim to assess risk for AF based on slightly different pools of evidence. Clinical partners will draw on well-documented archival enrolled patient data:
- Pre-ablation history and patient data, including ECG recordings and image data suitable for atrial model construction, and genetic screening data.
- Data recorded during ablation, recorded electrograms, and the actual ablation sites.
- Data from patient follow-up, with information on AF recurrence (i.e., data on rhythm outcome over the next 5 years).
Addressing the core objectives
To address Objective 1, we will build advanced computational models of the atria using A, including tissue fibrosis and structural information obtained from images and electrophysiological information from ECG (Objective 3). Prediction of individual patient outcomes and stratification is based on an risk score combined with Bayesian feature selection, estimated from genetic data from A (Objective 2), which will additionally inform the pre-ablation assessment (Objective 5). Data from B will then be provided for further analysis and modeling post-ablation. The base model may be adapted to incorporate advanced electrogram mapping, actual ablation sites inserted, and advanced data analysis extended. The improved biophysical 3D, image-based and population cohort models will be used to update the assessment of the clinical outcome (Objective 4). Addressing Objective 5, advanced simulation of electrical activation will be performed, and verified with recorded ECG. Optimization algorithms will be used to select likely ablation sites, virtual lesions will be inserted into the patient model, and the likelihood of AF termination/recurrence determined. To address Objective 6, data streams C will be provided to the teams working on analysis and modeling, and results will be directly compared with actual outcomes to complete the retrospective study.
This framework, relying upon both validated biophysical simulation models and advanced data analysis will then be integrated into a software tool to support decision-making for the individual patient for atrial ablation (Objectives 6 and 7). This integrated, interdisciplinary approach will be built/learned on data from well-phenotyped patient cohorts, extend robust multi-scale systems models of human atrial function, and create a unifying clinical tool integrating all potential knowledge in order to support prognosis and treatment for the individual patient.
This consortium represents the unique possibility of creating an innovative environment resulting in important new tools for treatment planning for AF. By combining actors from different sectors and disciplines we will, by utilizing diverse research models and tools, be able to span all the way from idea to clinical feasibility and gain immense knowledge about a complex disease. The ambitious, but feasible objectives, together with the interdisciplinary expertise embodied in our consortium, ensures freely moving and extensive exchange of ideas and knowledge, promoting high credibility of the research program. Furthermore, the inclusion of top-level European researchers and research institutions guarantees a high standard of scientific research in this project and rapid, agile delivery of quality outcomes. Medical need is the driver of our work, and we aim to develop tools, methodologies, and medicines to be brought to everyday clinical practice.