Introduction
Mild traumatic brain injury (mTBI) led to over 10,000 hospitalisations in Australia in 2021-2022. blood-based biomarkers may provide a tool to determine injury severity, monitor recovery, and inform return-to-play.
Methods
A panel of extracellular vesicle (EV) associated miRNA was generated by treating SIM-A9 microglia with lipopolysaccharide (LPS) and isolating EVs via tangential flow filtration (TFF) and size exclusion chromatography (SEC). miRNAs were sequenced using the Ion Torrent platform. Candidate miRNAs were validated in serum derived EVs from amateur Australia rules football players (n=25) from the BIOREC study with ethics approval by the Monash University Ethics committee (Project ID 27684). Control participants were compared with two timepoints: 24-hour post injury and 2-week post injury. Whole serum GFAP levels were measured using SIMOA. EVs were isolated from serum using qEV single columns, and miRNAs were quantified via qPCR. predictive performance was evaluated using multiple machine learning algorithms including decision trees, SVM, and deep learning.
Results
Two EV-associated miRNAs were found to be predictive at both acute and 2-week post injury. An additional two miRNAs were found to be timepoint specific. Additionally, EV miRNA had a similar predictive capacity to whole serum GFAP. The predictive capacity of EV miRNAs was comparable to serum GFAP. When combined, EV miRNAs and GFAP achieved 89% predictive accuracy using a decision tree model with an AUC of 0.93 and F1 score of 89.78% at the acute time-point.
Conclusions
EV-associated miRNAs offer diagnostic utility for mTBI and could be used in future to inform clinical return-to-play decisions