Researchers from the University of Sydney have used AI to identify potential biological markers linked to a higher risk of developing Parkinson’s disease.
Their study, published 10 May 2023 in PLOS ONE, examined the health measurements of over 300,000 people and found that having higher levels of inflammatory markers such as insulin-like growth factor 1 (IGF-1), lymphocyte count, and neutrophil/lymphocyte ratio (NRL) were associated with an increased risk of Parkinson’s.
A group of factors aligned with frailty also ranked highly, particularly those related to motor dysfunction, such as walking pace and hand grip strength.
The researchers, who co-contributed equally to the work, said that their results supported the idea that PD was linked to inflammatory disease ratio, with inflammatory markers elevated both before and at the time of diagnosis, and noted that the biomarkers could be used to predict future risk, improve early diagnosis, and may even lead to new treatment options.
“To prospectively screen for PD and better elucidate the disease mechanism, there is a need to identify blood-borne biomarkers, as well as environmental and genetic factors that are associated with greater risk or are protective,” they said.
“This is critical because neurodegenerative processes in dopamine neurons of the midbrain start many years before PD diagnosis. Thus, there is also a need to identify future risks, enabling early interventions to be offered, which may take the form of beneficial lifestyle changes or the development of novel neuroprotective agents.”
The study used an integrated machine learning algorithm developed by the team, IDEARS, to explore the impact of 1,753 measured non-genetic variables in 334,062 eligible UK Biobank participants, including 2,719 who developed PD after being recruited.
“IDEARS applies machine learning to health-related questionnaire data, longitudinal inpatient data, blood assays, genetic and neuroimaging data,” the researchers explained.
“Our model demonstrated that gender was the most important feature, with PD being more prevalent in males, which led us to further split subsequent analyses by gender to uncover [an age-independent hierarchy] of gender specific features.
“Our unbiased machine learning approach uncovered a novel set of features most associated with PD, and interestingly, several well-established risk factors thought to have a high association level with PD – such as pesticide exposure, smoking status, traumatic brain injury and caffeine consumption – were not identified in the most important features in our model.”
Smoking and caffeine intake had historically been considered to provide some level of protection against developing PD.
The IDEARS model also identified several features associated with cardiovascular health and body adiposity, with total and LDL cholesterol levels reduced in men with PD nearly 10 years before a diagnosis, while only 5 years before diagnosis in women.
“Cardiovascular and body fat variables appear to impact the risk of PD, with larger waist circumference (ranked 14th) being causative and elevated total cholesterol (16th) being protective,” the authors noted.
However, the most promising biomarkers for PD risk were elevated IGF-1, aspartate transaminase to alanine aminotransferase ratio (AST: ALT), NLR, reduced urate, and total and LDL cholesterol.
“These biomarkers demonstrated a consistent change before PD onset in males, however only IGF-1 and NLR were robustly elevated before diagnosis in females,” the researchers said.
“Given the non-specific nature of some of these biomarkers (e.g., AST: ALT, NLR), we suggest that they would be best used in combination to predict PD risk.”