摘要:
Introduction: Inhibition of non COX1 dependent platelet function (NCDPF) by aspirin (ASA) is variable, with few insights into underlying biology. NCDPF testing is unavailable in most patient care settings. To overcome these barriers, we hypothesized that a whole blood RNA expression profile - a stable biomarker that includes platelet RNA - would correlate with multiple measures of NCDPF on ASA and highlight underlying biological pathways. Methods: Healthy volunteers were treated with 325mg/day ASA. We assessed post-ASA NCDPF by calculating the area under the aggregometry curve (AUC) induced by ADP 10 uM, Epi 10 uM, and collagen 5mg/ml. After 14 days of ASA, we hybridized whole blood RNA to Affymetrix U133 plus 2.0 microarrays. An unsupervised Bayesian sparse factor regression allowed groups of coexpressed genes to be aggregated into 20 "factors". These factors were used in multivariable linear regression to predict platelet function with each agonist. The performance of the predictive model was assessed using leave-one-out cross validation. We used the KEGG pathways database to understand the biological pathways represented by the most correlative factors. Results: Forty volunteers (20 female) completed the study (100% compliant). There was wide variability (median [IQR]) in the AUC induced by ADP (291 [264,308]), Epi (201 [148, 249]), and collagen (214 [178, 242]). The predicted platelet function using the 20 Bayesian factors correlated strongly and significantly with each original measure before and after leave-one-out cross validation. (Table). Factor 6 appeared as a top correlative factor across all agonists (Table). Factor 14 had the strongest (r = 0.5) individual correlation and represented multiple platelet specific genes. Conclusions: Whole blood RNA expression profiling classifies NCDPF on aspirin, reveals a common, pathway underlying the response to multiple agonists, and highlights platelet specific biological factors.
展开