A Hybrid Machine Learning and Network Analysis Approach Reveals Two Parkinson's Disease Subtypes from 115 RNA-Seq Post-Mortem Brain Samples.

Andrea Termine,Carlo Fabrizio,Raffaella Cascella,Claudia Strafella,Laura Petrosini,Emiliano Giardina,Carlo Caltagirone,Valerio Caputo

doi:10.3390/ijms23052557

Abstract

Precision medicine emphasizes fine-grained diagnostics, taking individual variability into account to enhance treatment effectiveness. Parkinson’s disease (PD) heterogeneity among individuals proves the existence of disease subtypes, so subgrouping patients is vital for better understanding disease mechanisms and designing precise treatment. The purpose of this study was to identify PD subtypes using RNA-Seq data in a combined pipeline including unsupervised machine learning, bioinformatics, and network analysis. Two hundred and ten post mortem brain RNA-Seq samples from PD (n = 115) and normal controls (NCs, n = 95) were obtained with systematic data retrieval following PRISMA statements and a fully data-driven clustering pipeline was performed to identify PD subtypes. Bioinformatics and network analyses were performed to characterize the disease mechanisms of the identified PD subtypes and to identify target genes for drug repurposing. Two PD clusters were identified and 42 DEGs were found (p adjusted ≤ 0.01). PD clusters had significantly different gene network structures (p < 0.0001) and phenotype-specific disease mechanisms, highlighting the differential involvement of the Wnt/β-catenin pathway regulating adult neurogenesis. NEUROD1 was identified as a key regulator of gene networks and ISX9 and PD98059 were identified as NEUROD1-interacting compounds with disease-modifying potential, reducing the effects of dopaminergic neurodegeneration. This hybrid data analysis approach could enable precision medicine applications by providing insights for the identification and characterization of pathological subtypes. This workflow has proven useful on PD brain RNA-Seq, but its application to other neurodegenerative diseases is encouraged.

Highlights

Parkinson’s disease (PD) is the most common age-related motor neurodegenerative disease, affecting more than 6 million people worldwide, with rising incidence and prevalence imposing a mounting socioeconomic burden on society [1,2,3] and currently, no diseasemodifying treatments are available [4,5,6]
Hierarchical K‐means implementing the centroid method with Manhattan distance was Hierarchical K-means implementing the centroid method with Manhattan distance was selected as the best clustering algorithm based on validation metrics
We observed that PDC1 and PDC2 were characterized by specific disease mechanisms when compared with normal control (NC), further confirming that PDC1 and PDC2 represented two distinct subpopulations in PD

Summary

Introduction

Parkinson’s disease (PD) is the most common age-related motor neurodegenerative disease, affecting more than 6 million people worldwide, with rising incidence and prevalence imposing a mounting socioeconomic burden on society [1,2,3] and currently, no diseasemodifying treatments are available [4,5,6]. Defining which PD subtype we are facing is crucial to better understand underlying mechanisms, predict disease course, and eventually design personalized management strategies able to fully consider the genetic or other specific biological features that can be employed in a precision medicine approach addressed to match the patients’ needs [11,13,14,15]. Empirical clustering stratifies patients based on demographic factors, clinical parameters, and genetic factors, making use of expert-based a priori conceptions. Up to now, these applications have shown limited sensitivity in detecting clinically useful classes of PD patients, hindering the development and deployment of better suited treatments [16]

Objectives

Methods

Results

Discussion

Conclusion