The scope of global gene expression in peripheral blood for anti-doping testing

Abstract

This study aimed to determine, whether measures of gene expression can detect the effects of autologous blood transfusion, which is used illicitly by athletes to enhance performance. Gene expression data from blood samples collected longitudinally from subjects participating in two autologous blood transfusion Trials were analysed. Four participants in both Trials were transfused with 3U blood. Four additional subjects were transfused with 1U blood in the second Trial. In the first Trial samples were collected at baseline, 21 days post phlebotomy, and at 7, 14 and 28 days post re-infusion. During the second Trial samples were collected at baseline and 14-days post re-infusion. Chapter 2 describes transfusion-induced signatures of differentially expressed genes. Generalized linear modelling with subsequent Bayes moderation and FDR adjustment for multiple testing was used to identify characteristic transfusion-induced signatures of differentially expressed genes for follow-up testing at 7, 14, 21 and 28 days post re-infusion. We detected transfusion effects with 100% accuracy in samples collected 14 days post-transfusion in Trial 1 and reproduced these findings in Trial 2. The transfusion-response signatures were enriched for genes linked to the physiological response to autologous transfusion, including inhibition of erythropoiesis, impairment of cortical cytoskeleton and cell cortex formation and suppression of heme biosynthesis. These responses to transfusion remained detectable for at least 28 days after transfusion. Chapter 3 describes the results of the Weighted Gene Co-Expression Network Analysis, which results in a range of quantitative network descriptors with diagnostic potential (i.e., eigengene score, intra-modular connectivity score, extra-modular connectivity score). We identified conserved patterns of gene-expression network organisation through assignment of genes to distinct modules based on their co-expression similarity across all available sample data. We assessed the impact of transfusion on gene-expression networks by comparing intra- vs. extra- modular connectivities at different stages of the interventions. We also compared the characteristics of the modules with blood markers obtained from the participants. We investigated the functional enrichment of modules and of the most interconnected genes within each module (hubgenes). Finally, we quantified module-module interconnectedness and rated connection strength to assess the effect of transfusion on the transcriptome’s organisational structure. We were able to detect transfusion at 14 days post blood re-infusion with 100% accuracy in Trial 1 and reproduce these findings in Trial 2. The peripheral blood transcriptome is organised into distinct modules with characteristic transfusion-induced changes that reflect relevant physiological processes, such as suppression of erythropoiesis and metabolic response to alterations in blood composition. Peripheral blood transcriptome has great potential as a diagnostic tool for autologous blood transfusion, which remains hard to detect by other methods. It can also potentially detect physiologicaL responses to a wide range interventions. ‘Hubgene’ analysis, in particular, which allows assessment of intrinsic network properties through both quantitative (hubgene connectivities) and qualitative (hubgene function) measures, may be of great value in testing for responses to a wide range of substances and procedures, and in testing for yet unknown doping agents.