Abstract 24: Diagnosing small round cell tumors using whole transcriptome expression profiling.

Abstract

Abstract INTRODUCTION: Differential diagnosis of malignant small round cell tumors (SRCTs) can be challenging due to their undifferentiated character. Their primitive histopathological features often render the identification of morphological features difficult and therefore, no definitive diagnosis may be possible. This is crucial in pediatric SRCTs as disparate approaches to therapy are used for distinct tumor types based on a patient's risk. This study sought to discover and implement diagnostic signatures for a variety of SRCTs using whole transcriptome profiling. METHODS: RNAs from 115 prospectively-obtained pediatric SRCTs were profiled on Affymetrix Human Exon microarrays. These comprised of Ewing sarcoma (n=10), neuroblastoma (n=10), PAX-FKHR fusion-negative rhabdomyosarcoma (F-RMS, n=20), PAX-FKHR fusion-positive rhabdomyosarcoma (F+RMS, n=19), lymphoma (n=10), Wilms tumor (n=10), clear cell sarcoma (n=7), rhabdoid tumor (n=9), medulloblastoma (n=10), and osteosarcoma (n=10). Expressions of 1,393,765 probe selection regions (PSRs) were analyzed. Binary screens were performed for each tumor type to identify diagnostic PSRs. Selected PSRs were combined as a signature, and its diagnostic potential was assessed by k-nearest neighbor algorithms. Primers designed to interrogate loci corresponding to the diagnostic PSRs were tested on cell lines and cases representing the various tumor types using the WaferGen SmartChip MyDesign platform. RESULTS: Binary screens identified hundreds of diagnostic PSRs corresponding to annotated and unannotated transcripts for each tumor type. To limit signature size, the top 6 PSRs were chosen for each tumor type. 17%-83% of features in each signature represented non-exonic content. Hierarchical clustering demonstrated the ability of each signature to diagnose the corresponding tumor by binary and multiplex analyses. Discrimination analysis showed that signatures for Ewing sarcoma, neuroblastoma, F+RMS, lymphoma and osteosarcoma were the best performing and most exclusive. Overall, each tumor signature had robust diagnostic potential (binary discrimination, all p<0.0001). To explore the possibility of translating the signatures to a quantitative PCR platform, RNA from 8 cell lines corresponding to 5 tumor types were used: TC32, CHLA10, TC252 (Ewing sarcoma); RD (F-RMS); Rh30, JR (F+RMS); REH (surrogate for lymphoma); and Saos-2 (osteosarcoma). Diagnostic primers sets showed substantial efficacy in distinguishing cell lines and cases corresponding to the tumor type of interest when compared with other samples. CONCLUSIONS: Using pediatric SRCTs as an example, this study illustrates that whole transcriptome analysis can reveal potential transcripts that can robustly diagnose malignancies that pose a histological challenge. Such concise genomic signatures can be translated to quantitative PCR-based assays for routine clinical implementation. Citation Format: Anirban P. Mitra, Dana Haydel, Sheetal A. Mitra, Jude Dunne, Scott Silveria, Larry Wang, Timothy J. Triche. Diagnosing small round cell tumors using whole transcriptome expression profiling. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 24. doi:10.1158/1538-7445.AM2013-24