Unleashing the potential of big data in ophthalmology
Q&A with Professor Paul Foster, Moorfields Lead for the UK Biobank Eye and Vision Consortium
UK Biobank is an internationally recognised research platform that provides the global scientific community with the tools to identify previously unrecognised causes of a broad range of life-threatening and disabling illnesses. To create the Biobank, researchers have collected a wide array of health and wellbeing information from participants across the United Kingdom, storing it on an anonymised database.
Since the UK Biobank was first set up, it has broadened its scope to capture more data from the original participants, including eye and vision examinations. This led to the establishment of the UK Biobank Eye and Vision Consortium, facilitated by a range of clinical and academic talent from across the UK.
As a world-leading ophthalmic research partnership, Moorfields Eye Hospital and the UCL Institute of Ophthalmology have provided the infrastructural and administrative support for the eye and vision module of the UK Biobank. Here, Professor Paul Foster, the Moorfields Lead for the UK Biobank Eye and Vision Consortium, provides further insight into why collecting eye data from this large cohort of participants will help unleash a wave of new discoveries in ophthalmology (eye care) and beyond.
What is the UK Biobank and how did it come to be?
UK Biobank was set up with the aim of recruiting 500,000 people and capturing data from these individuals throughout their lives – so-called ‘longitudinal data’. Most of the data was collected at the start of the project, when each participant had a comprehensive health check, blood and other biological samples were taken, and records were made of their physical characteristics and lifestyle. Each individual agreed to be part of the Biobank for 20 years in the first instance, with the aim being to study what external factors or personal characteristics may put them at risk of developing a particular condition or disease.
As one of the largest healthcare data collection exercises in the world, UK Biobank presented several logistical challenges. To overcome these, the Biobank brought together many leading experts and nurtured collaborations with universities, NHS trusts across the UK and global bioscience enterprises.
UK Biobank was one of the first initiatives to take a standard approach to medical discovery and turn it on its head. Hypotheses-driven methodology works well in the era of small to medium-size studies. However, the growth of computing power and the capacity to store and process more information has allowed researchers to move away from hypothesis-driven studies towards ‘data mining’: looking for novel associations within enormous data sets.
UK Biobank is at the forefront of this new ‘big data’ approach. Through the vision and ambition of the people who set it up, UK Biobank has been at the cutting edge of scientific investigation, and is regarded as an internationally valuable biomedical research resource.
What is the Eye and Vision Consortium and what led to your involvement?
In 2008, two years after it was initially set up, the Wellcome Trust, Medical Research Council and other funding bodies realised the potential of UK Biobank, and decided to develop and fund it to a higher level. They recognised its potential to be able to provide new insight and a better understanding of some of the most common causes of death and major illness, including cardiovascular disease, cancer and diabetes. This support provided the opportunity for UK Biobank to expand its remit and capture specific data relevant to other areas of health – including eyes and vision.
The Eye and Vision Consortium was set up to manage the analysis of new ophthalmic data. The aim was to use this new data, alongside existing data from the UK Biobank, to study the causes of eye disease and vision loss, and to investigate how they can be controlled and prevented. In turn, this should help facilitate research into the causes of disease, potentially helping to develop new, more effective treatments.
My involvement came about for two reasons. Firstly, we’re very fortunate that the Consortium's administration is based at Moorfields Eye Hospital and the UCL Institute of Ophthalmology in London. Although I am a clinical ophthalmologist specialising in glaucoma, I was approached to be involved and run the administration as I had 15 years’ experience in epidemiological research. In fact, I was previously involved in a large study in East Anglia (EPIC Norfolk), which was used as a template for the UK Biobank eye and vision study.
What type of ophthalmic data was collected from patients?
We started eye and vision data collection three years in to the main study, so we were only able to gather baseline data on 117,649 of the 500,000 people involved. However, this is still the largest comprehensive eye study undertaken anywhere in the world to date.
With the consortium’s advice, we conducted tests using four machines, which effectively allowed us to collect seven different bits of information:
Visual acuity was measured through testing eye-chart reading ability
Auto-refraction was measured, capturing an individual’s spectacle prescription and curvature of the cornea
Intraocular pressure (pressure within the eye) was also measured as it’s a risk factor for glaucoma
Corneal biomechanics, which is the measurement of the physical properties of collagen (the main body tissue in the cornea)
Over 68,000 volunteers also had a simultaneous retinal photograph and an OCT scan of the macula (part of the retina at the back of the eye). An OCT, or Optical Coherence Tomography, uses light waves to capture a detailed, 3D image of the back of eye.
The range of this data, linked with accompanying data that already exists for these volunteers, makes the UK Biobank eye and vision dataset unique.
Arguably, the most exciting data from the examination, from a research standpoint, is the retinal photography and OCT scans . These images were graded at the Image Reading Centre at Moorfields – a crucial service that was able to provide a rapid turnaround and quality assurance. These scans are the key to seeing the early features of some of the most common eye conditions in the world affecting the retina, including glaucoma, macular degeneration and diabetic retinopathy. Further to this, these scans could potentially be relevant to the detection of other cognitive and neurodegenerative diseases such as Alzheimer’s or Parkinson’s. Similarly, while we know diabetic retinopathy affects sight, it is also widely thought that it can also cause neuro-degeneration, which could potentially be measured from the images we capture of the retina. This work is in its infancy, but it is something the consortium team are excited to explore.
Through the study of big data in this way, the eye can act as a window for research into other conditions which are historically hard to study, which is a fascinating concept.
What are the key achievements of the Eye and Vision Consortium to date and what will be the impact for patients in the future?
It’s hard to quantify or summarise the scale of the success of the UK Biobank. The ongoing work is fascinating and has led to a large number of high-impact publications and research projects all over the world. To give one example, a healthcare research group at Google is taking our retinal images and looking at ways of predicting cardiovascular disease, heart attacks, strokes, etc. The true impact of this work will rely on projects like this, where the data is used more broadly, looking beyond ophthalmology. This principle of facilitating discoveries through big data is what the UK Biobank is built on.
Ultimately, however, as a research tool this dataset will continue generating information that will undoubtedly lead to many research discoveries for the next quarter of a century, and possibly beyond. While there has been an abundance of research that has stemmed from both UK Biobank eye and vision data already, I feel we’ve yet to see the true scale of research it will support in years to come. Within the last two years alone, there are already signs that there'll be an exponential growth of discoveries stemming from these datasets. Furthermore, the value of the data will increase over time, as the participants continue to contribute their longitudinal health data throughout their lives, continuously strengthening the dataset.
From an ophthalmic viewpoint, one of the key achievements that comes to mind was when we received huge media interest from our consortium-led pollution and glaucoma publication in 2019. It was the second most widely reported publication in the history of the leading international eye and vision science journal Investigative Ophthalmology & Visual Science. It shed light on how pollution may increase the risk of glaucoma, which was a unique study of how environmental factors can impact our eye health in a way which hadn’t been seen before.
What will be the next phase of this evolving project?
The value of the Biobank is that the data is integrated. To this end, researchers Anthony Khawaja, Pirro Hysi and Chris Hammond are exploring the genetic data of the participants. This work should be very insightful and shed light on the genetic causes of eye conditions, potentially paving the way for the next generation of ophthalmic gene therapies.
However, we’re also keen to integrate clinical data, in particular noting whether participants have had cataract surgery or not – a gap in our current dataset. Our focus has therefore shifted to linking Biobank data with electronic hospital records – a more efficient way of collecting participants’ data. Of course, this will require quite a considerable and coordinated effort in terms of obtaining the necessary ethical and regulatory approval.
Logistically, integrating clinical records will also require a strategy for navigating the various different ways that hospitals and healthcare systems store their data. A large proportion of NHS eye records are now kept electronically on a programme called Medisoft, a very pro-research platform. We’re collaborating with Medisoft to help support the integration of patient records. We’re very lucky to be working with Parul Desai at Moorfields, who is dedicated to working out how best to do this. She is uniquely positioned to guide us, as she has significant experience in electronic health data (having previously been the Director of the NHS Information Centre).
Can you shed light on the collaborations and funding associated with this Eye and Vision Consortium?
Looking at the collaborations within the consortium itself, the Steering Committee has always strived to make this a UK-wide exercise, as it was acknowledged that country-wide input was needed to make it a success. The Consortium has clinical and academic representation from all four nations in the UK. However, it also needed individuals and groups who have the right skills and who know the value of this endeavour, which I feel we achieved. Within the Consortium itself, research groups have formed to study data on visual acuity, refractive error, intraocular pressure, retinal vascular characteristics and genetics. Drawing on the broad expertise of the Eye and Vision Consortium has made this a very successful venture.
While the infrastructure and administration of the Eye and Vision Consortium are based at Moorfields, the UK Biobank eye dataset has always been viewed as an enabling platform for the rest of the UK. The Consortium is a great illustration of a broad UK-wide collaborative research network that is achieving great results, and of how it is possible to get so many stakeholders engaged for the mutual benefit of all in the Consortium. This collaboration is a template for how the UK can compete strongly on the global stage.
The core funding for the UK Biobank was provided by the Wellcome Trust, The Medical Research Council, The Department of Health, Scottish Government and the Northwest Regional Development Agency. I’ve always felt that the collaboration and contribution of all these funders is a clear indication of how important this work was perceived to be. Supported by the NHS, UK Biobank is run as a non-profit charity, which has initial funding of about £62 million. However, additional support for training, implementation and quality control of the eye examinations came from the NIHR Biomedical Research Centre at Moorfields Eye Hospital and UCL Institute of Ophthalmology – critical funding that provided not only experienced staff with the right expertise, but also the right infrastructure to kick-start this mammoth project.
Across UK Biobank, there has been a huge effort to engage with major research funding organisations and charities. The Eye and Vision Consortium has working relationships with Fight for Sight, the International Glaucoma Association, the Macular Society, Diabetes UK and the British Heart Foundation, who all saw the value of this consortium [see full list of funders here]. Charitable organisations are vital in providing funding for an array of costs that are incurred with such a wide-reaching project. However, the key to the longevity of this project will be making the funding sustainable.
How did UK Biobank get interest from such a large number of volunteers?
All participants are volunteers and are mostly involved for altruistic reasons. When staff at the different sites recruited potential participants, it was important to communicate that the benefits of the study will be seen in the long term rather than the short term. Their own involvement may not benefit them directly, but it will have an impact on the type of healthcare that future generations will receive – which is crucial.
We’re very fortunate in the UK, as the public are both proud and supportive of the NHS – a healthcare system like no other in the world, where we all have a stake in the success of this publicly supported service. Because of this, a lot of people are willing to volunteer for medical research, especially for research that will play a part in the future transformation of the NHS.
How did you address the ethical debate around the storing and sharing of health data?
The aspect of the Biobank that draws the most questions from participants and the public concerns who is given access to the data, and what level of access researchers have.
The Wellcome Trust’s strategy is to fund research that is for the global good. The only way we would have secured funding from them is by allowing this resource to be openly accessible, supporting researchers and healthcare services around the world. But, of course, there are political and ethical implications that come with open access – for instance, what does the NHS get in return for this endeavour? Could we be boosting other countries’ private bio-resource industries at the expense of our own? On the other hand, the study can be seen as a major philanthropic and humanitarian endeavour whose value lies in the open sharing of data that facilitates research globally.
We have to be mindful that as this study evolves, the ethical debates will need to continue in parallel to ensure we maintain the trust of those who volunteered their information in the first place. There is a huge effort in ‘big data’ research to get patients engaged by debating and discussing the ethics around patient data: how it is stored, analysed, and, most importantly, accessed – and by whom. We have a lot to learn from patients and the public, and because the Biobank has received public funding, it's important that the future of this major resource for research is shaped by informed public debate.