11 months ago

2004 Scientific Report

would better account for

would better account for protein heterogeneity and would produce fewer false positive and false negative results. In collaboration with researchers at Evanston Northwestern Healthcare, Yale, and the Fred Hutchinson Cancer Research Center, we are measuring many proteins in the sera of pancreatic cancer patients and controls that are most likely to contribute to diagnostic signatures specific for particular disease states. The proteins chosen are those 1) with serum levels associated with pancreatic cancer; 2) with genes having altered expression in the tumor environment; and 3) with genes belonging to classes that are altered in cancerous or precancerous lesions. We have uncovered protein patterns in the serum of cancer patients that differ from those of healthy individuals or individuals with benign disease (Fig. 1). We want to define the precise patterns that best distinguish the groups and to obtain reproducible measurements from as many proteins as may contribute to a diagnostic signature. The protein profiles from clinical samples also allow additional levels of biological interpretation. For example, we are interested in characterizing the variation of particular classes of proteins, such as inflammation-related proteins, in relation to both benign and malignant disease. Such analyses could give valuable insights into disease development and progression mechanisms. Prostate cancer studies Prostate cancer is the second leading cause of cancer death in U.S. men (over 28,000 deaths annually in the United States). The prostate-specific antigen (PSA) test has improved the outlook for prostate cancer patients, but the limitations of the test are significant. Our laboratory is taking multiple approaches to better characterize the protein alterations associated with prostate cancer and to develop tests that will meet the need for improved diagnostics. An approach similar to that described above for pancreatic cancer is being used in a collaborative study to identify the protein profiles that best characterize the patient groups. Mass spectrometry–based methods will be used to validate the relevant antibody microarray measurements, and candidate markers will be used in further clinical studies with our collaborators. In another collaboration, a complementary approach uses protein microarrays to characterize the immune responses to prostate tumors. We are using microarrays of tumor-derived proteins, prepared from prostate cancer cell lines and separated by two-dimensional liquid chromatography, to measure and characterize these tumorreactive antibodies. In this way, tumor antigens that commonly elicit immune responses in cancer patients can be identified and characterized. These studies could result in new diagnostic tests and could give insight into the nature of the molecular alterations in the tumor. An additional complementary study aims to characterize changes in the disease-associated protein composition of prostatic fluid. These changes give a direct look at the metabolic activity of the secretory epithelial cells of the prostate, which are the main precursor cells of prostate cancer. The concentration of proteins secreted by prostate epithelial cells will be higher in the prostatic fluid than in the serum, making their detection easier. We are measuring the levels of cytokines, proteases, protease inhibitors, and other proteins in samples from prostate cancer patients, prostatitis patients, and controls. Of particular interest is the association of inflammation-related proteins with certain types of pancreatitis and prostate cancer, as these proteins are implicated in the pathologies of both diseases. External Collaborators Phil Andrews, University of Michigan, Ann Arbor Randall Brand, Evanston Northwestern Healthcare, Illinois Carlos Cordon-Costa, Memorial Sloan Kettering Cancer Center, New York City Jose Costa and Paul Lizardi, Yale University School of Medicine, New Haven, Connecticut Ziding Feng, Fred Hutchinson Cancer Research Center, Seattle, Washington Samir Hanash and Gil Omenn, University of Michigan, Ann Arbor Jorge Marrero, University of Michigan Hospital, Ann Arbor Alan Partin, Johns Hopkins University, Baltimore, Maryland 30

Figure 1. Identification of pancreatic cancer–specific serum protein profiles. The levels of many proteins in the sera of pancreatic cancer patients and controls were measured by antibody microarrays detected by RCA. The antibody measurements that together best distinguish the sample groups were clustered. Each column contains measurements from a single sample, with the labels color-coded according to patient class: blue represents pancreatitis; red, pancreatic cancer; pink, other gastrointestinal cancer; green, normal; and tan, other benign gastrointestinal disease. Each row contains measurements from a single antibody, and the color of each square indicates the level of protein binding: red = high, black = medium, and green = low. The columns and rows are ordered by similarity, the most similar being adjacent. The pancreatic cancer cases clearly segregate from both the normal and the pancreatitis samples, giving an early indication of the ability to specifically identify pancreatic cancer based on a serum protein profile. 31

Publications by Year