Abstract
BACKGROUND: Diagnosing subclinical atherosclerosis is often difficult since patients are asymptomatic. In order to alleviate this limitation, we have developed a molecular prediction technique for predicting patients with atherogenic risks using multi-gene expression biomarkers on leukocytes. METHODS: We first discovered 356 expression biomarkers which showed significant differential expression between genome-wide microarray data of monocytes from patients with familial hyperlipidemia and increased risk of atherosclerosis compared to normal controls. These biomarkers were further triaged with 56 biomarkers known to be directly related to atherogenic risks. We also applied a COXEN algorithm to identify concordantly expressed biomarkers between monocytes and each of three different cell types of leukocytes. We then developed a multi-gene predictor using all or three subsets of these 56 biomarkers on the monocyte patient data. These predictors were then applied to multiple independent patient sets from three cell types of leukocytes (macrophages, circulating T cells, or whole white blood cells) to predict patients with atherogenic risks. RESULTS: When the 56 predictor was applied to the three patient sets from different cell types of leukocytes, all significantly stratified patients with atherogenic risks from healthy people in these independent cohorts. Concordantly expressed biomarkers identified by the COXEN algorithm provided slightly better prediction results. CONCLUSION: These results demonstrated the potential of molecular prediction of atherogenic risks across different cell types of leukocytes.