Early Differential Expression of Chemotactic Genes in Peripheral Blood May Predict Poststroke Blood Brain Barrier Disruption.

PhD Complications of endovascular thrombectomy limit its current utility to a subset of patients. Expanding the utility of mechanical thrombectomy is possible with the addition of blood biomarkers that could be used to identify patients who are at risk for complications related to reperfusion injury such as blood brain barrier (BBB) disruption and malignant edema. The objective of this pilot study was to identify an early pattern of gene expression in the peripheral immune system which could be used to predict post-stroke BBB disruption. Peripheral whole blood samples were obtained from acute ischemic stroke patients at emergency department admission, and BBB disruption was assessed via level of hyperintense reperfusion injury marker (HARM) on MRI at 24-hour follow-up. Microarray was used to transcriptionally profile admission blood samples, and a machine-learning technique known as genetic algorithm k-nearest neighbors (GA/kNN) was used to identify a pattern of gene expression which could optimally discriminate between patients who presented with mild HARM (n=8) and severe HARM (n=8) at follow-up imaging. GA/kNN identified ten genes ( LAIR2, IL8, CXCL5, RBP7, CCDC149, LY96, HPSE, DDIT4, E2F3, and ADAM15 ) whose coordinate expression levels could discriminate between groups with 94% accuracy (88% sensitivity, 100% specificity). Furthermore, functional annotation enrichment analysis suggested that top genes identified by GA/kNN were enriched for genes involved in chemotaxis, suggesting that the early differential expression of genes associated with cellular migration may be predictive of post-stroke BBB disruption. The pattern of gene expression identified in this study shows strong potential for helping to guide selection of patients for mechanical thrombectomy, and warrants further evaluation to determine its true clinical efficacy.