Nucleotide motif-guided selection of plasma microRNA biomarkers for organ injury prediction in trauma

BACKGROUND: Trauma remains a leading cause of morbidity and mortality in part due to secondary organ injury and infection. Yet, our ability to predict the downstream pathophysiologic responses leading to organ injury and adverse outcomes is limited. Extracellular microRNAs (ex-miRNAs) as a DAMP can drive innate immune response and organ injury. Here, we tested plasma miRNAs as predictive biomarkers for organ injury in trauma. METHODS: Twelve miRNAs were selected based on RNAseq and pro-inflammatory nucleotide motifs identified by machine learning. Digital PCR was employed to quantify plasma miRNAs and Luminex to measure trauma injury markers. Multivariate Random Forest models were built to assess the predictive performance of the miRNA biomarkers. RESULTS: We identified a set of five nucleotide motifs that can predict the pro-inflammatory property of plasma miRNAs with a sensitivity of 84% and specificity of 69%. There was a marked and severity-dependent increase in the plasma miRNA biomarkers and numerous trauma injury markers at time of admission. The plasma concentrations of these miRNA biomarkers were highly correlated with the injury markers linked to various trauma endotypes. AUROC analyses indicated that the miRNA biomarkers possess strong diagnostic abilities and prediction in overall severity, organ injury, metabolic acidosis, coagulopathy, and innate inflammation in the trauma (n=48) but not sepsis (n=47) cohort. In a combined cohort, miR-224-5p and miR-145-5p exhibited a superior performance in differential diagnosis of trauma and sepsis with AUROC of 0.90 and 0.91, respectively. CONCLUSION: The panel of plasma miRNAs are specific biomarkers with strong diagnostic and prognostic performance in trauma-induced organ injury.