PROGRAM UPDATE: J UNE 2009 Clinical Proteomic Technologies for Cancer Advancing Protein Science for Personalized Medicine 1 NCI Clinical Proteomic Technologies for Cancer EXECUTIVE SUMMARY The addition of protein biomarker panels to the cancer diagnostic armamentarium is an area of considerable interest in medicine. The discovery that proteins and peptides are “leaked” by tumors into clinically accessible bodily fluids such as blood and urine has led to the possibility of diagnosing cancer at an early stage or monitoring response to treatment simply by collecting these fluids and testing for the presence of cancer-related biomarkers. Prostate-specific antigen (PSA) and cancer antigen 125 (CA-125) are examples of blood-borne cancer protein biomarkers that are currently being used in the clinic. However, the measurement of individual biomarkers has clinical limitations with respect to both sensitivity and specificity. For this reason, combinations of protein/peptide analytes are under intense investigation as biomarker panels can potentially bring greater sensitivity and specificity to cancer screening than any one analyte alone. If the potential of personalized medicine is to be realized, it must include this next generation of molecular diagnostics. We do not suffer a lack of candidate protein biomarkers. As of 2006, there were 1,261 putative cancer protein/peptide biomarkers described in the scientific literature. The sobering reality, however, is that very few of these candidates have been validated, and even fewer have made it into a medical diagnostic product (Figure 1). This discrepancy indicates that the issue lies within the candidate biomarker pipeline. Figure 1. Reality Check Adapted from Ludwig and Weinstein (Nov. 2005). Biomarkers in Cancer Staging, Prognosis and Treatment Selection. Nature Rev Cancer 5, 845-856. Proteomic technologies, which hold great promise for the discovery of novel cancer biomarkers, can help make sense of the complexities inherent in the proteome. In recent Program Update 2 years, however, studies that have applied proteomic technologies to clinical applications—such as mass spectrometry and affinity-based detection methods—have met with some disappointment. There are two major issues at hand: 1) Variability within biomarker discovery. A paucity of standard reagents and methods for protein identification and measurement has led to pervasive problems with reproducibility and comparison of research results among laboratories, posing a significant challenge to the translation of discoveries to clinical applications. 2) Biomarker candidates need to be pre-validated, or verified, prior to costly clinical validation studies. For clinical validation of protein biomarkers, an enzyme- linked immunosorbent assay (ELISA) is developed for each antigen to test large cohorts in clinical trials. However, these tests are very expensive. Each ELISA takes up to one year and $2 million to develop. A more efficient biomarker development pipeline will require coupling biomarker discovery in tissue and proximal fluid to verification in plasma prior to clinical validation (Figure 2). Verification is a rapid way to assess if a given candidate is detectable in blood and changes in a measurable way in relation to the presence or stage of disease. This bridging technology can rapidly triage a lengthy list of candidates prior to investing very large sums of money and time on the development of antibodies suitable for use in an ELISA. Figure 2. A Better Bridge is Needed between Biomarker Discovery and Clinical Validation Traditional approaches have contributed all they can; it is time for new technologies and approaches. To make clinical cancer proteomics a reality, the scientific community must 3 NCI Clinical Proteomic Technologies for Cancer first invest in much needed technologies and infrastructure in order to build a better biomarker development pipeline. The goal of the Clinical Proteomic Technologies for Cancer (CPTC) initiative is to develop a more refined, efficient and reliable biomarker development pipeline. This pipeline is anticipated to produce better credentialed candidate leads, ultimately accelerating the translation of new cancer biomarkers into diagnostic tests. Fixing this pipeline is too great an endeavor for a single investigator or institution. Accordingly, CPTC has brought together the best minds in proteomics to accomplish this goal. Together, the CPTC network is laying the foundation for clinical cancer proteomics by addressing each of the following barriers: Optimizing proteomic technologies and developing appropriate standards. Current and emerging protein measurement technologies must be optimized and calibrated through the use of standard protocols and performance reagents to produce comparable results among laboratories. Standardizing procedures for collecting, processing, and storing biological samples used in proteomics research. The use of high-quality biospecimens is critically important for proteomic research because the output—the data—is only as good as the input. The methods of biological sample preparation must be made more consistent to reduce variability in experimental results. Uniform sample quality, as well as access to large numbers of high-quality samples, will lead to more reliable results. Making high-quality reagents available and accessible. These include capture reagents (e.g., antibodies) that can be used in protein arrays, as well as other techniques that are used to measure proteins. However, a key challenge for proteomic researchers is acquiring high-quality, well-characterized monoclonal antibodies. While numerous commercial reagent suppliers make antibodies available for research, they tend to be expensive and may or may not be extensively characterized. Developing technologies that can quantify proteins across a large dynamic range. Enormous variation in protein concentrations and modifications are found within cells and body fluids, and the development of ultra-sensitive detection methods that have a large dynamic range has been a significant challenge. Capturing and identifying dilute proteins from a complex mixture has proven to be especially difficult. A biomarker present at 10 pg/mL in human plasma is roughly equivalent to trying to locate a softball somewhere between earth and the moon! In addition, many of the best techniques for finding low abundance proteins are not quantitative—that is, they can determine whether the protein is present or not but they cannot measure its concentration. Developing common bioinformatics resources with shared algorithms and standards for processing, analyzing, and storing proteomic data. Proteomic informatics tools that permit data sharing and computation among laboratories are essential for rapid progress in the field. Program Update 4 Implementing a verification step in the protein biomarker pipeline. Protein biomarker candidates should be “pre-qualified” before costly clinical validation studies are initiated. This important step can be accomplished by implementing a verification stage in the biomarker pipeline (Figure 3). Figure 3. Verification: Pre-qualifying biomarker candidates prior to costly clinical validation For clinical validation of protein biomarkers, an ELISA is developed for each antigen in order to test large cohorts in clinical trials, which can take up to one year and millions of dollars to develop. A more efficient biomarker development pipeline will triage candidates before an ELISA is developed, providing greater confidence in candidates prior to costly clinical validation. Using targeted proteomics, CPTC is creating this bridge from discovery to validation, which will allow investigators to run up to 200 assays in a matter of months for a fraction of the cost of a single ELISA. Adopting an interdisciplinary team approach to science. No one laboratory working on its own could possibly examine all of the candidate biomarkers, develop all of the necessary technologies, or assemble all of the pieces of evidence required to 5 NCI Clinical Proteomic Technologies for Cancer understand the molecular mechanisms of disease. It will require many laboratories working together to accomplish these goals. In many ways, the challenges facing the clinical proteomic community are comparable to those that were faced by the genomics community prior to the Human Genome Project (HGP)—the current technology enables the sampling of only a small portion of the proteome and at different levels of quality. Visionaries brought the HGP to life, but it was improvements made in DNA sequencing technologies that made this endeavor possible. The technologies became high-throughput, reliable, and reproducible. Had the project moved forward without these much needed technological improvements, the genomics community would never have been able to live up to its promise and the opportunities afforded by the HGP would never have been realized. Clinical proteomics stands at a juncture: Do we improve the technologies and methodologies first—do we forge ahead or build a modern highway that can routinely deliver biomarkers to the cancer research community? CPTC is building the highway. This document is a report on the first two years of progress in this five-year program. Some high level accomplishments include: Biospecimens: • A large collection of breast cancer cases (expected N=500) and control plasma samples (expected N=1500) have been accrued prior to diagnosis, to avoid bias of "baseline inequality.” Discovery Technologies for