Eight proteins in the blood may help predict the neurodegenerative disorder called Parkinson’s disease (PD) up to seven years before the start of its movement-related symptoms, researchers said in a study released on Tuesday.

Their study that combined artificial intelligence with protein biochemistry has suggested that the eight proteins serve as “biomarkers” for molecular events in the early stages of the disease and could identify individuals at risk for PD.

PD is a progressive disorder that results from the loss of cells in a part of the brain called the substantia nigra and depletion of a key messenger chemical called dopamine. Its symptoms include tremors, slow movements and impaired posture or balance, among others. While patients are currently treated with dopamine replacement therapy, a big challenge in PD has been the lack of any biomarkers that could be used to assess the progression of the disease.

“We are at present shutting the stable door after the horse has bolted,” Kevin Mills, professor at the University College London and the study’s senior author, said in a media release. “We need to start (evaluate) experimental treatments before patients develop symptoms.”

Their study was published in the research journal Nature Communications on Tuesday.

In the study, researchers at the UCL and the University Medical Centre in Goettingen, Germany, analysed blood samples from 99 patients recently diagnosed with PD, 72 patients who had a sleep behaviour disorder but no PD-like movement-related symptoms, and 36 healthy persons.

They applied a branch of artificial intelligence called machine learning to the analysis and found that the concentrations of eight proteins in the blood are consistently altered in patients with PD. Every patient with PD showed specific concentrations of these eight proteins.

They tested the machine learning model to determine whether it could predict which among 72 patients with the sleep behaviour disorder would develop PD. The researchers followed up with the patients for 10 years and found that the machine-learning model predictions matched the onset of symptoms. It has correctly predicted that 16 patients are likely to develop PD, displaying an accuracy of 79 per cent up to seven years before the onset of PD’s symptoms.

The researchers are hoping the eight protein biomarkers could help in the evaluation of candidate therapies that are under development or might emerge as an option to slow or prevent the progression of PD. The biomarkers are directly linked to processes such as inflammation and the degradation of certain proteins.

These markers (also) represent possible targets for new drug development, said Michael Bartl, the study’s co-first author at the University Medical Centre in Goettingen who conducted the research alongside Jenny Hallqvist at the University College London.