Nalini Raghavachari 
Clinical biologists feel that microarray-based transcriptome analysis is a frontier technology for the identification of potential biomarkers by application to biological materials that are most relevant to the phenotypes under investigation. These include biopsy materials from fine needle aspirates (FNA), cell sub-populations or enriched isolates from laser capture microdissection (LCM). Although transcriptional profiles in such target disease tissues or cells are ideal for such analyses, our experience in procuring such specimen and the low amount of RNA in them for standard microarray assays suggests that whole blood is a more viable surrogate tissue.
The ready availability of blood, the minimally invasive method of specimen collection and its character as a dynamic storehouse of cellular information on molecular clues of infection, inflammation, and autoimmune diseases make whole blood a more practical and attractive proposition for clinical research 1, 2 . Whole blood has an edge over tissue biopsies or cells in studies measuring disease state or drug response and for the discovery of biomarkers of hematologic diseases as well as a wide range of non-hematologic disorder. Applying microarray technology on peripheral blood may provide new insights of variations in global gene expression specifically associated with normal and disease states.
Although we have previously studied global gene expression on fractionated blood samples such as Peripheral Blood Mononuclear Cells (PBMCs) 3 , successful studies of gene expression profiles in whole-blood total RNA have been limited until now due to heterogeneous cell types and potential ex vivo changes from blood handling and processing. PBMCs with a more uniform cell population, containing lymphocytes and monocytes are the most transcriptionally active cells in blood, making it an ideal study specimen. However, the extra fractionation procedure for PBMCs requires a prolonged period before RNA stabilization, and this has been shown to have significant ex vivo changes in gene expression profiling. In addition, in multi-center clinical trials, we have found that isolation of PBMCs at the time of sample collection is a major shortcoming as skilled technicians are needed for processing the samples at the site and this could also lead to operator induced variability in microarrys.
Traditionally whole blood samples for gene expression were collected in CPT tubes (Becton Dickinson) with an anticoagulant. However, RNA expression profiles have been observed to change over time in such fractionated samples thereby implying the need for stabilisation of RNA between sample collection and isolation to maintain the expression profile of blood cells.
Attempts to overcome these hurdles led to the development of new approaches that would ease the sample collection and RNA stabilisation. In this respect, the PAXgene blood RNA system has now been widely employed in gene expression studies of peripheral blood. Although, this system employs an easy way to collect, store, transport and stabilise RNA from whole blood, many studies have demonstrated that RNA prepared from the PAXgene blood tubes result in significant increase in overall variability and decrease in the transcript detection sensitivity of leukocyte-derived mRNAs. The observed anomalies have largely been attributed to the presence of predominant amounts of reticulocyte-derived globin transcripts that constitute close to 70% of mRNA in whole blood samples. This represents a major problem for the study of vascular diseases such as hemolytic anaemias and sickle cell disease, owing to the high abundance of globin transcripts in nucleated erythrocytes and reticulocytes.
Addressing these limitations, we undertook studies that combine stabilisation of RNA and reduction of globin transcripts in whole blood in order to evaluate the suitability of the globin reduced RNA for microarray-based whole blood transcriptome studies in hematologic disorders. Using samples collected from a globin abundant disease, namely sickle cell anemia 4 , where the globin transcripts constitute a majority of the whole blood RNA, we observed that efficient removal of globin transcripts in PAXgene stabilised whole blood reduced the variability and improved the detection sensitivity of microarray experiments.
Importantly, we also observed an enhanced identification of differentially expressed transcripts from nucleated red cells and reticulocytes that are quite abundant in diseases associated with hemolytic anemia. We would propose that in diseases such as sickle cell anemia, thalassemia, G6PD, pyruvate kinase deficiency and in malaria, these early red blood cell progenitors represent major contributors to pathophysiology. Insights into the transcription profile of these cells may contribute greatly to our understanding of mechanism of disease, prognosis, and responses to therapeutics.
From a biological perspective, the globin depleted PAXgene versus the PBMC difference expression profile provided a window into real time erythrocyte expression profiles in vascular diseases. While isolated PBMCs remain the optimal specimen for microarray based biomarker studies for non hematological disorders, in clinical investigations on hematological diseases and in multi-center clinical trials where isolation of PBMCs is not feasible, application of globin reduction process is highly recommended. The combination of globin mRNA reduction after whole-blood RNA stabilisation represents a robust clinical research methodology for the discovery of biomarkers for hematologic diseases and in multicenter clinical trials investigating a wide range of nonhematologic disorders where fractionation of cell types is impracticable.
The author is the Director of the Genomics Core Facility at the National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, Maryland, USA.
