Predicting Survival And Metastasis In Soft Tissue Sarcoma: A Gene Microarray Study Using Rna Derived From Archival Paraffin-Embedded Tumours.

10074 Background: The genetic determinants of survival and metastasis in soft tissue sarcoma (STS) are poorly understood and progress in this field has been limited by the rare nature of STS and the need for fresh/frozen tissue (FT) for gene microarray analyses. Our objective was to determine whether valid gene microarray data can be obtained from archival formalin-fixed paraffin-embedded tumours (FFPET) in order to retrospectively identify prognostic STS gene signatures and potential molecular targets from patients with known clinical outcome data. Methods: Total RNA was extracted from macrodissected viable tumour tissue for 34 FFPET (14 liposarcomas, 11 leiomyosarcomas and 9 synovial sarcomas) and two paired FFPET and FT primary leiomyosarcomas (modified Optimum kit, Ambion, for FFPET RNA extraction and Trizol, Invitrogen, for FT RNA extraction). One-cycle labelled cRNA was hybridised to Affymetrix U133 Plus 2 microarrays, and strict QC protocols used to identify 19 arrays for further analysis using R and BioConductor. Log rank regression and Kaplan-Meier plots of disease-specific survival were performed to identify genes predictive of survival and/or metastasis. Results: Similar fold changes in gene expression were obtained for paired cRNA samples, although the number of genes called present was lower for cRNA from FFPET. Five hundred genes (selected from FFPET cRNA arrays) were discriminatory for survival (p=0.0006 to 0.0088) across all STS subtypes and correctly assigned 17/19 cases by leave-one-out cross validation. Many are known tumour prognostic genes, significant for survival, metastasis, invasion, angiogenesis and apoptosis. Twenty-five novel and known candidate genes (with 3–10 fold differential expression) were selected for validation on 19 test and 60 independent cases by RT-PCR for relative gene expression and immunohistochemistry for protein detection. Conclusions: We have shown that prognostic information can be derived from archival FFPET, permitting the identification of candidate prognostic genes and therapeutic targets in STS, and opening the way for studies in other tumours where FT is unavailable. No significant financial relationships to disclose.