Periodic Reporting for period 4 - MMAF (Novel multimodal approach to atrial fibrillation risk assessment and identification of targets for prevention by interdisciplinary exploitation of omics, advanced electrocardiography, and imaging)
Berichtszeitraum: 2020-07-01 bis 2021-06-30
Atrial fibrillation is an increasingly common disease in aging populations. Despite its public health impact, risk prediction is in its infancy. The main objective of the MMAF project was to establish a contemporary risk prediction algorithm for atrial fibrillation developed in international cohorts. Through the exploration of a multimodal approach, including electrocardiographic (ECG), imaging, and biomarkers/omics in blood and tissue we achieved novel pathophysiological insights into the genesis of atrial fibrillation and novel information for risk prediction in clinic and the general population. Our epidemiological research has improved the understanding of the prevalence, incidence and cost related to atrial fibrillation. Our omics work has shown pathways of inflammation and cardiac remodeling as central to atrial fibrillation. A clinical trial which demonstrates that the interference with these pathways can reduce disease progression is under evaluation. A simple version of the risk algorithm has been implemented in physician software and is currently being evaluated for opportunistic screening. Based on our ERC work the AFFECT-EU consortium (DIGITAL, RISK-BASED SCREENING FOR ATRIAL FIBRILLATION IN THE EUROPEAN COMMUNITY (affect-eu.eu) European Union’s Horizon 2020 Research and Innovation Programme, grant agreement N°847770) has been founded and will further explore the results of MMAF.
Along with the main objectives we have developed a multimodal approach to atrial fibrillation risk prediction combining prior knowledge and our own prior research using risk markers across a range of pathogenic factors related to atrial fibrillation. We could improve risk prediction beyond existing (partly our own prior) risk schemes. We have also been able to develop a simple, but precise risk prediction algorithm by machine learning application combining clinical information (manuscript under review at Heart Rhythm: Machine learning-based identification of risk-factor signatures for undiagnosed atrial fibrillation in primary prevention in clinical practice). We selected biologically and clinically plausible and applicable candidates representing clinical networks for integration into the novel risk algorithm, optimized discrimination and calibration of the new risk scheme including bootstrapping and prospective application in external cohorts, e.g. the Dutch PREVEND cohort.
We identified genomics candidates and replicated them in a polygenic risk score which compared well with a biomarker-based risk algorithm. Gene-set enrichment analyses and Multi-level ONtology Analysis (MONA) algorithms identified possible new pathways which currently are under work-up (Assum I. Nature Communications in press, Börschel CS, Europace 2021). Major pathways comprised remodelling/fibrosis and inflammation. In Mendelian randomization studies we have explored potential causal relations of biomarkers with atrial fibrillation (Geelhoed B, Europace 2020). The combination of electrocardiographic and imaging parameters using machine learning algorithms, largely artificial neural networks is ongoing in our local studies and in a German Center for Cardiovasvular Research (DZHK e.V.) project initiated by me. For the strongest biomarker that evolved (N-terminal pro B-type natriuretic peptide) we examined the differential relation with atrial fibrillation and its common comorbidity, heart failure (Schrage B, J Am Heart Assoc. 2020). In the AFFECT-EU (grant agreement N°847770) consortium we will test this top protein-based biomarker for optimized atrial fibrillation risk prediction.
Our data have entered international consortia (CHARGE Consortium) on intermediate phenotypes, e.g. genetic studies on electrocardiographic, vascular function traits and atrial fibrillation as an outcome (Dörr M, Circ Genom Precis Med. 2019, Christophersen I, Circ Cardiovasc Genet. 2017, Roselli C, Nat Genet. 2018).
We have explored autonomic tone and heart rate variabilityand its relation to atrial fibrillation and heart failure, but could not demonstrate strong associations and rather weak discriminatory ability. For sleep disordered breathing monitoring we applied screening methods in our cohorts (ECG monitoring, wrist worn device for sleep apnoea screening), but could not demonstrate strong associations that would merit implementation in risk algorithms either. We continue to examine sleep disordered breathing and sleep apnoea using more sophisticated measurement.
We have assessed the relation of standard and novel imaging parameters of ventricular and atrial structure and function in relation to remodelling, oxidative stress, inflammation and to the disease atrial fibrillation itself in the Hamburg City Health Study (Csengeri D, manuscript under preparation). Statistical analyses of modern echocardiographic and cardiac MRI imaging in relation to atrial fibrillation, atrial fibrillation subtypes and its sequelae, impairment in cognitive function and brain lesions/stroke in the Hamburg City Health Study are ongoing (Camen S, manuscript under preparation).
We have published important manuscripts on sex/gender differences in atrial fibrillation including epidemiology, classical AF risk factors, lifestyle, ECG, imaging, biomarkers and omics by stratified interrogation and interaction analyses, but also as a central topic of our publications.
Dissemination included
Invited and abstract talks at (inter)national cardiac societies (German Cardiac Society, AHA Scientific Sessions, European Society of Cardiology). The 2020 online European Society of Cardiology Conference had >100,000 attendees.
(Inter)national discussion of our results in the lay press.
Publications in high-ranked peer review journals.
(Inter)national teaching tutorials for cardiovascular nurses and allied professionals.
(Inter)national teaching tutorials for lay people with an interest in atrial fibrillation.
A simple machine-learning-base risk algorithm has been developed and implemented into physician computer software and will facilitate screening in high-risk individuals.
For researchers a comprehensive, unique OMICs data set has become available upon request. Different cardiovascular disease entities can be accessed and analyzed.
The AFFECT-EU consortium (European Union’s Horizon 2020 Research and Innovation Programme, grant agreement N°847770) has been made possible through our work on the ERC grant and will support the implementation of our research findings on atrial fibrillation risk prediction.
We identified genomics candidates and replicated them in a polygenic risk score which compared well with a biomarker-based risk algorithm. Gene-set enrichment analyses and Multi-level ONtology Analysis (MONA) algorithms identified possible new pathways which currently are under work-up (Assum I. Nature Communications in press, Börschel CS, Europace 2021). Major pathways comprised remodelling/fibrosis and inflammation. In Mendelian randomization studies we have explored potential causal relations of biomarkers with atrial fibrillation (Geelhoed B, Europace 2020). The combination of electrocardiographic and imaging parameters using machine learning algorithms, largely artificial neural networks is ongoing in our local studies and in a German Center for Cardiovasvular Research (DZHK e.V.) project initiated by me. For the strongest biomarker that evolved (N-terminal pro B-type natriuretic peptide) we examined the differential relation with atrial fibrillation and its common comorbidity, heart failure (Schrage B, J Am Heart Assoc. 2020). In the AFFECT-EU (grant agreement N°847770) consortium we will test this top protein-based biomarker for optimized atrial fibrillation risk prediction.
Our data have entered international consortia (CHARGE Consortium) on intermediate phenotypes, e.g. genetic studies on electrocardiographic, vascular function traits and atrial fibrillation as an outcome (Dörr M, Circ Genom Precis Med. 2019, Christophersen I, Circ Cardiovasc Genet. 2017, Roselli C, Nat Genet. 2018).
We have explored autonomic tone and heart rate variabilityand its relation to atrial fibrillation and heart failure, but could not demonstrate strong associations and rather weak discriminatory ability. For sleep disordered breathing monitoring we applied screening methods in our cohorts (ECG monitoring, wrist worn device for sleep apnoea screening), but could not demonstrate strong associations that would merit implementation in risk algorithms either. We continue to examine sleep disordered breathing and sleep apnoea using more sophisticated measurement.
We have assessed the relation of standard and novel imaging parameters of ventricular and atrial structure and function in relation to remodelling, oxidative stress, inflammation and to the disease atrial fibrillation itself in the Hamburg City Health Study (Csengeri D, manuscript under preparation). Statistical analyses of modern echocardiographic and cardiac MRI imaging in relation to atrial fibrillation, atrial fibrillation subtypes and its sequelae, impairment in cognitive function and brain lesions/stroke in the Hamburg City Health Study are ongoing (Camen S, manuscript under preparation).
We have published important manuscripts on sex/gender differences in atrial fibrillation including epidemiology, classical AF risk factors, lifestyle, ECG, imaging, biomarkers and omics by stratified interrogation and interaction analyses, but also as a central topic of our publications.
Dissemination included
Invited and abstract talks at (inter)national cardiac societies (German Cardiac Society, AHA Scientific Sessions, European Society of Cardiology). The 2020 online European Society of Cardiology Conference had >100,000 attendees.
(Inter)national discussion of our results in the lay press.
Publications in high-ranked peer review journals.
(Inter)national teaching tutorials for cardiovascular nurses and allied professionals.
(Inter)national teaching tutorials for lay people with an interest in atrial fibrillation.
A simple machine-learning-base risk algorithm has been developed and implemented into physician computer software and will facilitate screening in high-risk individuals.
For researchers a comprehensive, unique OMICs data set has become available upon request. Different cardiovascular disease entities can be accessed and analyzed.
The AFFECT-EU consortium (European Union’s Horizon 2020 Research and Innovation Programme, grant agreement N°847770) has been made possible through our work on the ERC grant and will support the implementation of our research findings on atrial fibrillation risk prediction.
In community-based and clinical cohorts we are collecting comprehensive biological and epidemiologic information on atrial fibrillation in more than 100,000 individuals. Deep-phenotyping including tissue and blood biomarkers provides insights that have not been possible before. Sex/gender differences have undergone thorough epidemiological work-up with clear differences and disparities that need to be addressed in prevention and treatment of the disease. Our findings, brought to a broad audience, will hopefully change the management of atrial fibrillation in practice.
Exploitation of omics has provided novel pathophysiological pathways that will be addressed in detailed work-up. Overall, we will be able to improve risk prediction of atrial fibrillation using a multimodal approach with the aim to help prevention of disease, thus reduce morbidity and improve wellbeing in aging societies.
Exploitation of omics has provided novel pathophysiological pathways that will be addressed in detailed work-up. Overall, we will be able to improve risk prediction of atrial fibrillation using a multimodal approach with the aim to help prevention of disease, thus reduce morbidity and improve wellbeing in aging societies.