Date of Graduation


Document Type


Degree Type



School of Pharmacy


Pharmaceutical Sciences

Committee Chair

Paul D Chantler

Committee Co-Chair

Taura L Barr

Committee Member

Lori A Hazlehurst

Committee Member

Jason D Huber

Committee Member

Gordon P Meares


Early and accurate diagnosis of stroke improves the probability of positive outcome, however the tools available to clinicians for the identification of stroke are limited. It is well established that the peripheral immune system responds robustly to ischemic brain injury, and thus may be a viable source of stroke biomarkers. The objective of this work was to utilize high throughput transcriptomics in combination with machine learning techniques to identify patterns of gene expression in peripheral whole blood with the potential to diagnose ischemic stroke, and to further determine the role of such gene expression patterns within the context of stroke immunopathology. A discovery cohort was recruited comprised of 39 ischemic stroke patients and 24 controls, and peripheral blood was sampled at emergency department admission. Genome-wide expression profiling was performed via microarray and a machine learning technique known genetic algorithm k-nearest neighbors (GA/kNN) was then used to identify ten genes (ANTXR2, STK3, PDK4, CD163, MAL, GRAP, ID3, CTSZ, KIF1B, and PLXDC2) whose coordinate pattern of expression was able to discriminate between groups with 98.4% accuracy. The diagnostic robustness of this pattern of expression was subsequently tested in an independent validation cohort which included an additional 39 ischemic stroke patients and 50 controls, and was able to discriminate between groups with 95.3% accuracy. Furthermore, leukocyte phenotyping of validation cohort subjects suggested that this pattern of differential expression is a marker of stroke-induced shifts in the cellular composition of the peripheral blood, as many of these ten genes exhibited a strong pattern of lineage-specific expression on isolated leukocyte subpopulations, and whole blood expression levels which were highly correlated with post-stroke white blood cell differential. However, results of in-vitro experiments using primary human leukocyte cultures suggested that one of these ten genes, CD163, which encodes a lymphoinhibitory peptide known as soluble cluster of differentiation 163 (sCD163), may play a role in modulation of stroke-induced peripheral immune phenotype, as monocyte-derived sCD163 present in the serum of ischemic stroke patients demonstrated an ability to inhibit the proliferation of lymphocytes isolated from healthy donors. Due to the robust nature of these findings, the ten transcriptional biomarkers identified by GA/kNN in our analysis warrant further evaluation to determine their true diagnostic efficacy in the clinical setting. Furthermore, future exploration into the role of sCD163 as a post-stroke immunomodulatory agent could provide novel insight regarding the underlying mechanisms which drive stroke immunopathology.