Abstract
Using whole blood and peripheral blood mononuclear cells, microarray-derived gene expression profiles have shown promise for the detection of acute ischemic stroke. Circulating leukocytes contain multiple cellular subsets of highly specific functions that may provide more powerful and more specific stroke detection than whole blood based expression profiles. The objectives of this study were to determine the cellular sources of gene expression changes in whole blood in ischemic stroke and the utility of leukocyte subset profiles for stroke detection. Using high-throughput reverse transcription real time PCR, the absolute expression of 41 stroke-related transcripts identified in whole blood and peripheral blood mononuclear cells was quantified in 6 major leukocyte subsets – CD15+ granulocytes, CD14+ monocytes, CD4+ T lymphocytes, CD8+ T lymphocytes, γδTCR+ cells and CD20+ B lymphocytes. Hierarchical cluster analyses were used to identify clusters of cell subset specific gene expression patterns for ischemic stroke detection. CD15+ granulocytes and CD8+ T lymphocytes were found to be the major sources of the expression changes in ischemic stroke, with 14 and 16 genes up-regulated respectively. None of the genes were significantly altered in CD14+ monocytes or CD20+ B lymphocytes. Multiple clusters of transcripts were identified that discriminated between ischemic stroke and control, most notably in CD15+ granulocytes (p=2.88e-5) and CD8+T lymphocytes (p=1.71e-5). A CD15+ granulocyte-derived 3 gene cluster (CA4, MMP9, NAIP) showed high accuracy for ischemic stroke detection (AUC=0.813) and was 100% sensitive in a validation cohort. We conclude that transcripts identified in microarray studies in circulating leukocytes in stroke are predominantly expressed in CD15+ granulocytes and CD8+ T lymphocytes. Leukocyte subset specific gene expression clusters show promise for ischemic stroke detection.