Abstract LB-476: A universal method for elimination of haemolyzed plasma samples that improves miRNA signature performance for early detection of colorectal cancer

Abstract

Abstract microRNAs (miRNAs) are short (∼21 bp) non-coding RNAs that regulate gene expression through post-transcriptional interactions with target messenger RNAs. miRNAs are exported to extracellular fluids by both malignant cells and cells of the immune system. Blood plasma and serum miRNAs are stable under standard clinical sampling protocols and have therefore attracted considerable interest for their potential as minimally invasive biomarkers for diverse pathological conditions, including various cancers. Although blood plasma contains as little as 1-20 ng of RNA per mL, plasma miRNAs are easily and reliably quantified using the highly sensitive and specific miRCURY LNA™ Universal RT microRNA PCR platform. However, the relative scarcity of miRNA in plasma creates a potential for contamination of the plasma miRNA expression profile by miRNAs from the cellular constituents in blood. One common cause of plasma sample rejection in clinical chemistry is haemolysis. If not identified, haemolysis can lead to erroneous results in a number of standard clinical laboratory tests, including blood potassium and lactate dehydrogenase (LDH) levels. Here we investigate the effect of sample haemolysis on plasma miRNA profiles. Using isolated red blood cells (RBCs) and genome-wide miRNA PCR profiling, we define the miRNome of RBCs. By spiking RBC extract into a non-haemolyzed plasma sample, we demonstrate that as little as 0.05% contamination of plasma can be detected as an increased expression level of a subset of miRNAs. We also define a set of plasma miRNAs that are not affected by haemolysis. Finally, we discover a miRNA qPCR QC signature that can be used to eliminate haemolyzed plasma samples. In a project aimed at defining a signature for the early detection of colorectal cancer (CRC) from a plasma sample, we screened 325 plasma samples from CRC patients and colonoscopy-verified CRC-negative controls. The samples were part of a clinical trial conducted in 7 different Danish hospitals, and were examined for the expression of 378 miRNAs previously detected in plasma. We show that haemolysis in this sample set correlates with hospital ID, and with the utilization of specific blood sample collection vials. Using our haemolysis signature, we eliminated haemolyzed samples and demonstrated that this step leads to a major improvement of CRC detection (ROC AUC increase from 0.67 to >0.80). We conclude that haemolysis can be a cause of plasma miRNA profile contamination, and that elimination of haemolyzed samples using our miRNA haemolysis QC signature overcomes this clinical problem and leads to an increase in plasma miRNA biomarker performance. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr LB-476. doi:1538-7445.AM2012-LB-476