Image-driven knowledge graphs reveal new targets for treating heart disease

Image-driven knowledge graphs reveal new targets for treating heart disease

Knowledge graphs are a powerful tool for bringing together information from biological databases and linking already known knowledge about genes, diseases, treatments, molecular pathways and symptoms in a structured network. Until now, there has been a lack of detailed, individual information about what the affected organ actually looks like and functions.

The latest research led by postdoctoral researcher Dr. Khaled Rjoob and group leader Professor Declan O’Regan from the Computational Cardiac Imaging Group at the MRC Laboratory of Medical Sciences have further developed this technology by adding image data to a knowledge graph for the first time. CardioKG provides a detailed overview of the heart's structure and function, greatly improving the accuracy of predicting which genes are linked to diseases and whether existing drugs could treat them.

Capture cardiac variation

To develop CardioKG, the team used cardiac imaging data from 4,280 UK Biobank participants with atrial fibrillation, heart failure or heart attack, as well as 5,304 healthy participants, and captured variations in the heart's structure and function. In total, over 200,000 image-based features were generated and used to train the model. The team integrated this with data from 18 different biological databases and used artificial intelligence (AI) to predict links between genes and diseases, as well as opportunities for drug repurposing.

One of the advantages of knowledge graphs is that they integrate information about genes, drugs and diseases. This means you have more opportunities to make discoveries about new therapies. We found that incorporating cardiac imaging into the graph changed the way new genes and drugs could be identified.”

Professor Declan O’Regan, Computational Cardiac Imaging Group, MRC Laboratory of Medical Sciences

Predicting new drug opportunities

The model identified a list of new disease-associated genes and predicted two drugs to treat heart disease; Methotrexate, a drug for rheumatoid arthritis, could improve heart failure, and gliptins, used to treat diabetes, could be helpful for atrial fibrillation. The team also made a surprising discovery that caffeine, which makes the heart more excitable, has a protective effect in patients with atrial fibrillation, who have an irregular and rapid pulse.

“The exciting thing is that there are other recent studies in this area that support our preliminary results,” says Declan. “This highlights the enormous potential of knowledge graphs in uncovering existing drugs that could be used as new treatments.”

Extending the technology to other organs

CardioKG offers proof-of-concept technology that can go far beyond the heart. Researchers could now develop knowledge graphs that integrate imaging data wherever organ imaging exists, meaning the same approach could be applied to brain scans, to body fat imaging, or to other organs and tissues to explore new therapeutic possibilities in areas such as dementia or obesity.

The ability of these knowledge graphs to generate accurate and rapid lists of high-priority genes across a range of diseases would provide pharmaceutical companies with a valuable starting point, highlighting biological targets that they can explore, validate, and potentially develop into new therapies far more efficiently than traditional discovery methods.

“Building on this work, we will expand the knowledge graph into a dynamic, patient-centered framework that captures real-world disease progression,” says Khaled. “This will open up new opportunities for personalized treatment and predicting when disease is likely to occur.”

This study was supported by the Medical Research Council, the British Heart Foundation, Bayer AG and the Imperial College Biomedical Research Center of the National Institute for Health and Care Research (NIHR).

In addition to his role at the LMS, Declan is Professor of Cardiovascular AI at the British Heart Foundation and Clinical Topic Lead for the British Heart Foundation Center of Research Excellence at Imperial's National Heart and Lung Institute.

Sources: