Early detection of colorectal cancer from patient blood plasma using microRNA-based q-rt-PCR

microRNAs (miRNAs) constitute a recently discovered class of small RNAs (typically 21-23 nt) that function as post-transcriptional regulators of gene expression. Current estimates indicate that more than one third of the cellular transcriptome is regulated by miRNAs, and that each miRNA potentially regulates hundreds of different mRNAs. As a consequence, miRNAs have been proposed to be master regulators of cellular state. This hypothesis has been borne out by a large number of studies demonstrating a causal link between miRNA dys-regulation and numerous disease states, including a diverse array of human cancers. Furthermore, the high stability of miRNA in common clinical source materials (e.g. FFPE blocks, plasma, serum, urine, saliva, etc.) and the ability of miRNA expression profiles to accurately classify discrete tissue types and disease states have positioned miRNA quantification as a promising new tool for a wide range of diagnostic applications.

Colorectal cancer (CRC) ranks 4th in terms of prevalence and second in numbers of deaths among cancers of the western world. Although early detection of CRC leads to a favorable prognosis, and though CRC can easily be detected in the locally restricted state using a variety of diagnostic procedures, frequent late diagnosis means that CRC is still a leading cause of cancer mortality worldwide. Current procedures suffer from one or more disadvantages within areas such as cost, safety, inconvenience to the patient with the consequence of low compliance, lack of trained personnel, sensitivity etc, thereby precluding their adoption as a population screening tool. There is therefore an unmet need for a generally acceptable CRC screening assay.

To facilitate discovery and clinical transfer of miRNA-based diagnostic markers, we developed a genome-wide LNA™-based miRNA q-rt-PCR platform with unparalleled sensitivity and robustness. The platform uses a single RT reaction to profile >700 human miRNAs from 2 predefined 384 well plates and thus allows high-throughput profiling of miRNAs from important clinical sources without the need for pre-amplification. Using this system, we have profiled a large number of plasma samples from localized and regional CRC patients, and from matched healthy controls. An extensive QC system has been implemented in order to secure technical excellence and reveal any unwanted bias in the dataset. We will present our approaches to data normalization and the results of signature development using linear classification methods. We show that we can detect the majority of cancer cases with good specificity, using a cross-validation approach. In summary, our results show that minimally invasive early detection of CRC using a clinically viable approach is feasible.