An eye test for Parkinson's

One day, could it be your optician that is the first to spot that you have Parkinson’s? A major research collaboration between Moorfields Eye Hospital and UCL has found that the eye scanners now used by thousands of high street opticians could potentially be used to detect signs of Parkinson’s seven years before someone is diagnosed by a hospital neurologist.

A Moorfields patient with Parkinson’s having an eye scan. Picture: Moorfields Hospital

The finding demonstrates just what can be achieved by a combination of access to the rich patient data records of the NHS, advances in artificial intelligence, and collaboration between the health service and universities. What is not so clear is how. soon the research might be put to practical use.

When I spoke to Pearse Keane from Moorfields, who along with Siegfried Wagner of the UCL Institute of Opthalmology led the project, I asked why they had even thought there might be some link between the eye and Parkinson’s.

“As ophthalmologists we've known for more than 100 years that we can use the eye as the window to the rest of the body,” he said, pointing out that signs of diabetes or high blood pressure can be detected in the eye. Now, he says they have “supercharged” this 100 year old idea”by using big data, by using advanced imaging techniques for the eye, and the latest advances in artificial intelligence.”

Key to the project have. been three rich sources of data. First, the Moorfields database of 20 million retinal scans, partly developed during Pearse Keane’s work with Google’s DeepMind AI division to create an algorithm to triage the millions of scans generated by high street opticians. This was then linked with HES, the national database of hospital episode statistics,  which enabled the researchers to identify 700 people diagnosed with Parkinson’s who had also had retinal scans..

These were then compared with the 100,000 people in the Moorfields database who did not have Parkinson’s. “We were able to find very, very subtle changes in their inner retina and in particular their inner nuclear layer, which is the layer that has dopaminergic neurons” So, they had discovered a biomarker, a signal in the eye that indicated the presence of Parkinson’s.

The final step involved UK Biobank, which tracks the health of half a million people who agree to have a whole range of tests and scans over a long period. “What we are able to show,” explains Pearse Keane “is that those changes in the inner retina were present in 53 patients who were in UK Biobank, who went on to develop Parkinson's.”

My eye scans could be part of the research

So a brilliant piece of detective work which has - in theory - led to an early diagnostic tool for Parkinson’s. But Professor Keane is quick to point out that what they have detected are very subtle changes in the eye - between 1 and 2 micrometres difference in thickness - and we are unlikely to see tests used to diagnose individuals in the near future. He does however see potential for this technique to be used as a way of assessing participants in clinical trials of Parkinson’s drugs or other forms of treatment.

And this story excites me - partly because as a man diagnosed with Parkinson’s in 2019 who has been treated for an eye condition by Moorfields since 2005, I reckon my data may have been used in the research. More evidence then that despite all of the negative stories about the sharing of patient data, it has a vital role to play in health research.

And Pearse Keane thinks a government that believes the UK can be a leader in AI needs to learn an important lesson from this project:

“I think if the universities could just have better access to NHS data at scale across all diseases then I bet you that would be the single thing that would make the UK a superpower for AI and healthcare.”

World class universities and an unrivalled collection of patient data generated by the health service - it sounds like a winning combination. It is just linking up the researchers with the data which is the challenge.