Machine Learning Approach May Offer Future Diagnostic Advances in Parkinson Disease

A machine learning approach that works to identify mitochondrial impairments, a potential mechanism underlying the pathogenesis of Parkinson disease (PD), accurately distinguished between patients with PD and healthy controls, according to study findings published in npj Systems Biology and Applications.

As the second most common neurodegenerative disease, PD is growing at a faster rate than any other neurodegenerative condition among the general population. More troubling, the difficulty in diagnosing PD has greatly impacted the likelihood of misdiagnosis, with more than 1 in 5 (21%) patients indicating in a survey that they had to see their general provider 3 times before being referred to a specialist.

While the detailed molecular and cellular mechanisms underlying the pathogenesis of PD remains unclear, recent evidence suggests that "mitochondrial dysfunction may play a role in the onset of the condition, saying in a statement that “mitochondria–small cellular 'subunits' involved in cell metabolism and energy generation–constantly and dynamically interact with each other, forming perpetually changing networks known as mitochondria interaction networks (MINs).”

Researchers sought to further examine whether PD-related features may exist in topological patterns of MINs, an analysis that has never been performed in general populations or patients with neurodegenerative conditions. They utilized a large 700 gigabyte data set of 3-dimensional mitochondrial images, consisting of colonic neurons collected from patients with PD and healthy controls, as well as dopaminergic neurons from stem cells.

"Since conventional analysis focusing on individual mitochondria has not provided satisfying insights into PD pathogenesis, our pioneering work has gone a step forward by investigating the interaction networks between these organelles", said study author Feng He, PhD, MSc, Group Leader of the Immune Systems Biology Group of the Luxembourg Institute of Health (LIH) Department of Infection and Immunity, in a statement.

In their findings, researchers uncovered particular network structure features within MINs that were altered in patients with PD compared to controls:

• Mitochondria formed connected subnetworks were generally larger in patients with PD than controls
• Efficiency of the energy and information transmission and distribution among the different mitochondria in the MINs of patients with PD was significantly lower than in controls
  

Speaking about these transmission delays among patients with PD, He notes that these different topological patterns in MINs may mean that energy and information are possibly produced, shared, and distributed less competently in the neuronal mitochondria of patients with PD relative to healthy controls. “Suggesting their connection to mitochondrial damage, deficiencies, and fragmentation typical of neurodegenerative disorders,” added He.

After applying the machine learning approach to analyze the MIN characteristics, the use of a combination of these network features allowed researchers to accurately distinguish between patients with PD and healthy controls. Additionally, researchers found these different MIN patterns to be highly correlated with commonly-used PD clinical scores derived from rating scales (ie, Unified PD Rating Scale) on individual patients.

"Our findings bring forward the potential of using particular mitochondrial network features as novel biomarkers for the early diagnosis and classification of patients with PD, which might help develop a new health index,” said study author Rejko Krüger, MD, director of Transversal Translational Medicine at LIH.

“As a next step, we will explore how our results may offer new perspectives for the understanding of various other neurodegenerative diseases characterized by mitochondrial dysregulation, such as Huntington disease and Alzheimer disease, making our work a true instance of translational and transversal research."

Reference

Zanin M, Santos BFR, Antony PMA, et al. Mitochondria interaction networks show altered topological patterns in Parkinson’s disease. NPJ Syst Biol Appl. Published online November 10, 2020. doi:10.1038/s41540-020-00156-4