A Comprehensive Identification of the Microrna Transcriptome and Its Application in B Cell Malignancies.

Abstract 2403 Poster Board II-380 Background MicroRNAs are 18-22 nucleotide-long RNA molecules that regulate expression of genes. We and others have previously demonstrated a role for microRNAs in the pathogenesis of B cell malignancies. Computational predictions suggest that the human genome encodes several thousand microRNAs. Thus far, about 700 microRNAs have been discovered in humans, including over 200 new microRNAs in the past year alone. The ongoing discovery of microRNAs makes it difficult to comprehensively study their role in a disease group. The advent of high throughput sequencing allows the simultaneous identification of millions of transcripts, thereby providing a sensitivity that is several orders of magnitude higher than conventional methods. We hypothesized that high throughput sequencing would be an effective tool to comprehensively identify microRNAs in normal and malignant B cells. While there is an overlap between diffuse large B-cell lymphoma (DLBCL) and Burkitt lymphoma (BL) in morphology, immunophenotype and cytogenetics, distinguishing between BL and DLBCL is critical because there are important differences in their clinical management. We investigated whether microRNA expression could be used to reliably distinguish BL from DLBCL. Methods and Results We carefully chose 31 human samples to represent the spectrum of normal and malignant B cells including FACS-sorted naive, germinal center, memory, plasma cells, EBV transformed and activated B cells. Samples derived from B cell malignancies included B-lymphoblastic lymphoma, chronic lymphocytic leukemia (immunoglobulin gene mutated and unmutated), mantle cell lymphoma, marginal zone lymphomas, HIV-related lymphoma, BL, DLBCL (activated and germinal center type), primary mediastinal B cell lymphoma, Hodgkin lymphoma, and multiple myeloma. We applied massively parallel, high-throughput sequencing of the 18-22 nt RNAs from these cases and generated a total of 255,624,785 sequences (∼5 billion bases). Using a computational approach that we have previously validated with normal B cells, we identified the expression of 429 known microRNAs in normal and malignant B cells, a number that is over three times higher than previously recognized in any tissue type. We also identified the expression of 302 novel microRNAs in normal and malignant B cells. The vast majority of these microRNAs were highly conserved in multiple species. As a proof of principle, we generated a custom microarray that included all the known human, and viral microRNAs, as well as 302 novel microRNAs identified by sequencing, and applied it to the clinically important distinction of BL from DLBCL. Biopsy samples were collected from 104 patients (BL, N=25, DLBCL, N=79) treated at 9 institutions that comprise an international consortium. All cases were reviewed for pathology diagnosis and profiled for microRNA expression. We constructed a Bayesian predictor to distinguish BL from DLBCL based on the microRNA expression. The predictor performance was tested using leave-one-out cross-validation. We also applied gene expression profiling to the cases of DLBCL to identify the molecular subsets of DLBCL: activated B cell like and germinal center B cell like DLBCL. The microRNA profiles of these cases were equally efficacious in distinguishing the DLBCL subsets. The predictor constructed based on microRNA expression was over 90% accurate in distinguishing BL from DLBCL, using pathology diagnosis as the gold standard. Further, microRNA-based predictor was also over 90% accurate in the distinction of the molecular subsets of DLBCL, compared to the gold standard of gene expression-profiling. As additional validation, we performed in situ hybridization of selected microRNAs to directly visualize their expression using methods that are easily accessible in conventional pathology laboratories. We found excellent concordance between the expression results derived from microarrays and in situ hybridization suggesting a ready path to clinical translation. Conclusion Our study represents the first comprehensive delineation of microRNA expression in B cell malignancies using high throughput sequencing. Our data suggest that microRNAs are a promising marker for the distinction of aggressive lymphomas. Disclosures: No relevant conflicts of interest to declare.