The Latest Heart Research using UK Biobank data
According to the World Health Organisation (WHO), cardiovascular disease (CVD) is the biggest killer globally. CVD is an overarching term for diseases of the heart (heart disease) and vascular diseases (concerning blood vessels). There are number of factors which increase the risk of CVD, from lifestyle habits such as smoking or physical inactivity to genetics.
UK Biobank provides an open-access resource for global scientists to further our understanding of the most common and life-threatening diseases including cardiovascular disease. The imaging, health, genetic and lifestyle data generously donated by our 500,000 participants is enabling better predictive models of disease for earlier diagnosis, and developing therapeutic interventions.
What is some of the research into CVD being conducted using UK Biobank data?
Exploring genetic risk
Underpinning the genetic basis of a disease is crucial if preventative measures and treatments are to be developed.
It is important to identify areas of genetic variation in the genome to determine which genes may play a role in disease development, or increase the risk of disease. Polygenic risk scores (PRS) are used by scientists to quantify inherited genetic risk by integrating information from many sites of variation across one individual’s DNA.
Last year, a new tool for the identification of patients at-risk of cardiovascular disease (CVD) was incorporated into 12 GP practices across the North-East and North Cumbria. This ‘Healthcare Evaluation of Absolute Risk Testing’ or ‘HEART’ pilot study was developed by Genomics Plc to assess the effectiveness of adding extra genetic information to an existing GP platform ‘QRISK’ which aims to prevent and manage CVD in patients. This integrated risk tool (IRT) was developed as a result of research through the use of UK Biobank data.
Through QRISK, individuals receive a percentage score representing their chance of heart attack or stroke over the next 10 years. A score of 10% is indicative of high risk.
Using the novel IRT tool, these risk scores were reclassified to a clinically significant level of 50% or more in 24% of patients.
The study was highly successful, with over 90% of GPs in agreement that the tool could be used in routine primary care for managing CVD in patients. It is incredibly exciting that UK Biobank has contributed to what looks to be a significant change in real-life clinical settings.
Revealing Novel insights
It is known that the health of retinal blood vessels can indicate issues concerning the heart.
Recently, deep learning methods (complex algorithms which aid computer predictions and which map patterns in data) were used to ‘train’ an artificial intelligence system (AI). In the research, led by the University of Leeds, the system was able to ‘read’ retinal scans and predict who may have a heart attack in the next 12 months.
The retinal scans of over 5000 participants in UK Biobank were used in the study. The AI system estimated the size of the left ventricular of the heart and its pumping efficiency. A larger left ventricle, usually observed through expensive tests such as echocardiography (ECG) or MRI, is associated with increased risk of heart disease. When the AI combined these left ventricular characteristics with demographic data such as participant age, it exhibited an accuracy of 70-80% in its predictions of heart attack risk.
Such AI methods could provide a comparatively inexpensive way to support a clinician’s assessment of identifying those at risk of heart disease through early signs in the eye’s blood vessels. This could ensure that preventative treatments against cardiovascular disease can begin earlier too.
Advancing understanding through large-scale imaging analyses
Imaging our major organs in vast numbers through magnetic resonance imaging (MRI) and ultrasound, and combining these images with genetic and lifestyle data, is pivotal to the diagnosis, prevention and potential treatment of diseases such as dementia or heart disease.
In 2019, AMRA Medical conducted analyses on imaging scans from 10, 000 UK Biobank participants, to map distributions of fat under our skin (subcutaneous adipose tissue), around our waist and organs (visceral fat), alongside measurements of liver fat. This allowed researchers to calculate an individual’s predisposition to Coronary Heart Disease (CHD) and Type 2 Diabetes.
Since 2021, a collaboration between AMRA Medical, Pfizer Inc. and UK Biobank, has been expanding the dataset, with data on 25, 000 UK Biobank participants now under analysis. New measurements have been conducted for more detailed findings, such as fat values in different muscles like the thigh.
UK Biobank’s existing genetic and health data in combination with highly effective imaging analysis techniques for profiling the body composition of individuals, presents an exciting avenue ahead for greater understanding of metabolic diseases such as diabetes or obesity, both of which impact the heart.
Transforming disease prediction
Researchers at the Icahn School of Medicine at Mount Sinai (New York) have trained a computer-based machine learning model, named ‘in silco score for coronary artery disease’ (ISCAD). It allows researchers to measure coronary artery disease (CAD) risk, the most common type of heart disease.
To make its predictions, this ‘digital marker’ integrated several clinical features from electronic health records, such as medications and past diagnoses. Over 90, 000 records were used, 60, 000 of which were from the UK Biobank cohort. The model was able to follow the extent to which the coronary arteries narrowed, as well as complications of CAD such as heart attack.
It is hoped that the findings will help to develop more targeted diagnoses and improve management of CAD for those living with the disease, with further large-scale studies planned to validate findings.