Sarcomas are cancers of connective tissue or mesenchyme. While approximately 10,000 sarcomas are diagnosed per year in the United States, greater than 50% of patients will die from the disease. Currently applied diagnostic techniques using morphology, histology, immunohistochemistry and cytogenetics are inadequate. Because diagnostic discrepancy can exceed forty percent, a better system of classification is imperative in order to improve outcomes for these frequently morbid and often deadly tumors. To truly understand the neoplastic process in mesenchymal tissues, one must first improve diagnostic criteria to better classify and stratify the greater than three hundred entities. In the future, molecular phenotyping will become commonplace. Modalities such as cDNA and in situ microarrays, proteomics and hybrid techniques will complement and perhaps replace methods utilized today.

Our long-term goal is to develop a mesenchymal tumor classification system based on cDNA profiling of actual tumors evaluated ex vivo. Our rationale is based on evidence that different profiles of mRNA expression reflect differences in the biological properties of cancer. Identifying signature patterns of gene expression via cDNA profiling may supersede the prognostic ability of histologic subtype, grade, anatomic location, as well as the presence or absence of particular characterized, solitary molecular aberrancies such as translocations. cDNA microarrays facilitate the systematic and comprehensive analysis of transcriptional alterations occurring in diseased tissues. This technique involves quantitative hybridization to a large panel of cloned genes with the total expression complement (cDNA) derived form a particular cell or tumor. We realize that cDNA arrays in isolation may tell us nothing about sarcoma pathogenesis as it does not address protein-driven issues. A given expressed gene does not necessitate that the protein product is integral to the process of sarcomagenesis. Nevertheless, this protein may be specific to the tumor thereby facilitating classification but providing no information as to the mechanisms of pathogenesis. The information generated by microarray projects will complement and incorporate data created via proteomics, in situ microarrays, hybrid and other techniques. Only after learning the relationship of these tumors to normal mesenchyme can we then begin to predict their behavior and elucidate the mechanisms of their pathogenesis.