Blood test and AI could predict Parkinson’s seven years before symptoms

18 Jun 2024, 4 p.m.

A droplet of colourless liquid hangs of the end of a conical tip above a plastic tube. The plastic tip is held in a blue rack that is beginning to go out of focus

A team of researchers, led by scientists at UCL and University Medical Center Goettingen, Germany have developed a simple blood test that uses artificial intelligence (AI) to predict Parkinson’s up to seven years before the onset of symptoms. The lab team have been supported by funding from our NIHR GOSH Biomedical Research Centre.

Parkinson’s disease is the world’s fastest growing neurodegenerative disorder and currently affects nearly 10 million people across the globe.

The condition is a progressive disorder that is caused by the death of nerve cells in the part of the brain called the substantia nigra, which controls movement. These nerve cells die or become impaired, losing the ability to produce an important chemical called dopamine, due to the build-up of a protein alpha-synuclein.

Currently, people with Parkinson’s are treated with dopamine replacement therapy after they have already developed symptoms, such as tremor, slowness of movement and gait, and memory problems. But researchers believe that early prediction and diagnosis would be valuable for finding treatments that could slow or stop Parkinson’s by protecting the dopamine producing brain cells.

Senior author, Professor Kevin Mills, UCL Great Ormond Street Institute of Child Health, said: “As new therapies become available to treat Parkinson’s, we need to diagnose patients before they have developed the symptoms. We cannot regrow our brain cells and therefore we need to protect those that we have.

“At present we are shutting the stable door after the horse has bolted and we need to start experimental treatments before patients develop symptoms. Therefore, we set out to use state-of-the-art technology to find new and better biomarkers for Parkinson’s disease and develop them into a test that we can translate into any large NHS laboratory. With sufficient funding, we hope that this may be possible within two years.”

Predicting Parkinson's

The research, published in Nature Communications, found that when a branch of AI called machine learning, analysed a panel of eight blood based biomarkers whose concentrations are altered in patients with Parkinson’s, it could provide a diagnosis with 100% accuracy.

The team then experimented to see whether the test could predict the likelihood that a person would go on to develop Parkinson’s.

They did this by analysing blood from 72 patients with Rapid Eye Movement Behaviour Disorder (iRBD). This disorder results in patients physically acting out their dreams without knowing it (having vivid or violent dreams). It is now known that about 75-80% of these people with iRBD will go on to develop a synucleinopathy (a type of brain disorder caused by the abnormal buildup of a protein called alpha-synuclein in brain cells) – including Parkinson’s.

When the machine learning tool analysed the blood of these patients it identified that 79% of the iRBD patients had the same profile as someone with Parkinson’s.

The patients were followed up over the course of ten years and the AI predictions have so far matched the clinical conversion rate – with the team correctly predicting 16 patients as going on to develop Parkinson’s and being able to do this up to seven years before the onset of any symptoms. The team are now continuing to follow up on those predicted to develop Parkinson’s, to further verify the accuracy of the test.

The research was funded by an EU Horizon 2020 grant, Parkinson’s UK, the National Institute for Health and Care Research GOSH Biomedical Research Centre (NIHR GOSH BRC), and the Szeben-Peto Foundation.

Professor David Dexter, Director of Research at Parkinson’s UK, said: “This research, co-funded by Parkinson’s UK, represents a major step forward in the search for a definitive and patient friendly diagnostic test for Parkinson’s. Finding biological markers that can be identified and measured in the blood is much less invasive than a lumbar puncture, which is being used more and more in clinical research.

“With more work, it may be possible that this blood based test could distinguish between Parkinson’s and other conditions that have some early similarities, such as Multiple Systems Atrophy or Dementia with Lewy Bodies.

“The findings add to an exciting flurry of recent activity towards finding a simple way to test for and measure Parkinson’s.”

Why does a children's hospital research Parkinson's

At GOSH we lead and collaborate on research in a number of fields. While we are best known for our research and innovation in child health and rare or complex diseases, our expertise and skills can be leveraged to improve and learn across a variety of health conditions.

Working with our closest partner, the UCL Great Ormond Street Institute of Child Health, this is just one of many projects that benefits from core infrastructure funding from our NIHR GOSH Biomedical Research Centre.

Experts across the UK come together to share cutting edge findings in Juvenile Idiopathic Arthritis

The CLUSTER Consortium celebrates over six years of research addressing the needs of children and young people with arthritis.

Astronaut Tim Peake ‘launches’ our new NIHR Clinical Research Facility

We were thrilled to have astronaut Tim Peake join us at the official launch of the new National Institute for Health and Care Research (NIHR) GOSH Clinical Research Facility where he met with patients, families, staff and children from the GOSH nursery.

Young people want AI involved in their care, new GOSH study shows

For the first time, a study has asked children and young people across all four UK nations for their views on how they would like AI to be used to enhance their healthcare

World-first genomic testing scheme ensures every child gets the best cancer treatment for them

Children with cancer are being offered treatment plans tailored to their specific cancer thanks to a single genomic test that reads more than three billion letters of DNA to identify cancer-causing mutations.