Proteomics to Diagnose Human Tumors and Provide Prognostic Information

Much excitement has been generated in the past few years around the potential of "omics technologies" to produce advances in medicine. For example, global profiling using DNA microarrays has uncovered patterns of gene expression that may have clinical utility. However, it has become clear that numerous obstacles must be overcome before findings from these studies have a substantial impact on clinical practice. Clinical Utility
One challenge is to understand, at a mechanistic level, the significance of associations observed between subsets of genes and the clinical features of disease. Another challenge is to identify the smallest but most informative sets of genes associated with specific clinical features, which then could be interrogated using technologies available in clinical laboratories. Still another challenge is to determine how well RNA levels of predictive genes correlate with protein levels. A lack of correlation may imply that the predictive property of the gene(s) is independent of function and, therefore, represents a mere association. For these and other reasons, DNA microarrays will not eliminate a pressing need for other types of profiling technologies that go beyond measuring RNA levels. Additionally, DNA microarrays have limited utility in the analysis of biologic fluids and in uncovering assayable biomarkers directly in the fluid. There has recently been a tremendous interest in the potential of proteomics to address unmet needs in medicine, from developing a better understanding of disease pathogenesis to implementing more effective strategies for early detection and monitoring of disease as well as the development of more effective therapies. Proteomics is particularly well suited for investigating biologic fluids to identify disease-related alterations and to develop molecular signatures for disease processes. A mobilization effort, involving academia, governments, industry, and philanthropy, has been initiated to develop agendas for medical proteomics. This is reflected, for example, in the prominence of proteomics in the National Institutes of Health roadmap. Range of Proteomic Strategies
In this issue of ONCOLOGY, Ornstein and Petricoin review some of the proteomics technologies used to analyze disease and provide some examples of their application to cancer. The potential of this technology is practically unlimited-from detecting cancer early, to imaging tumors based on the unique proteins that may be expressed on their surface, to administering therapy that targets abnormally expressed proteins. The field is clearly in its early phase. Nevertheless, the inventory of proteomics technologies currently available is substantial and perhaps rather bewildering. However, one must keep in mind that unlike DNA microarrays, which provide one measure of gene expression (namely RNA levels), proteomic strategies need to address the many different features of proteins that could be altered in cancer and other diseases. Such features include the determination of protein levels in biologic samples, their modification, and their selective interaction with other biomolecules, such as other proteins, antibodies, drugs, and various small ligands. Limitations and Challenges
With heightened expectations that proteomics will deliver and overcome some of the limitations of other approaches, or at least complement them, it is important to keep in mind that numerous challenges remain. First, there is substantially greater variability at the protein level than at the genomic or RNA level, particularly in clinical samples. The variability results not only from the heterogeneity that characterizes disease states such as cancer but also from numerous others sources such as the sample procurement process itself, which may result in protein breakdown and other types of protein modifications. Another major challenge is to develop a quantitative readout encompassing all the proteins expressed in a cell or tissue, to fulfill expectations of advancing our understanding of disease processes. No single technology currently enables such analyses, and much ongoing effort will be needed to further develop proteomics technologies to expand their reach. Even with the limited scope of current proteomics technologies, numerous cancer-related investigations may be envisaged, as described by Ornstein and Petricoin. Data Access
Technical considerations aside, an issue that has become quite important in this era of genome- and proteomescale investigations is data sharing. The capacity to generate data far exceeds the ability of one group to fully mine such data, for which it is advantageous to have access to multiple sets of data. There is also a compelling need to integrate data generated at multiple levels, from genomics to proteomics and metabolomics. It is crucial that published data be accessible to other investigators. Therefore, investigators, funding agencies, and publishers share a responsibility to facilitate access to data.