Plasma Cell-Free RNA as Non-Invasive Biomarker for Parkinson's Disease
Abstract
Parkinson disease (PD) is the most common neurodegenerative disorder, after Alzheimer disease (AD). Many attempts have been made to find a good biomarker, including alpha-synuclein protein levels in the cerebrospinal fluid (CSF). Cell free nucleic acids-based diagnostic tests have revolutionized prenatal screening. They have also been investigated in cancer and fetal development among other traits, including neurodegenerative diseases. We have successfully developed a preliminary predictive model for AD using cell-free plasma RNA sequencing (cfRNASeq) and machine learning techniques. We used an exploratory dataset (10 AD cases and 10 controls) to train a predictive model. We obtained an area under the ROC (AUC) of 0.84 in an independent replication dataset (10 independent AD cases and 10 controls). Moreover, this model provided similar accuracy (AUC=0.86) when tested in four preclinical AD. Using state-of-art deep neural network approaches, the accuracy increased up to 0.94. Overall, these results indicate that we can identify individuals that will progress to dementia. We think this technique can be applied to PD to generate disease-specific predictive model. We hypothesize that there are detectable changes in the plasma free nucleic acid composition due to PD pathogenesis, even in early stages. We will use bioinformatics tools to construct a predictive model for PD, leveraging longitudinal plasma data that will allow the modeling of plasma cfRNA composition changes over the curse of the disease, thus maximizing the power of selecting informative transcripts to construct the predictive model.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 2021
- Accession Number
- AD1158161
Entities
People
- Laura Ibanez
Organizations
- Washington University in St. Louis