New biomarkers of pancreatic adenocarcinoma are needed to improve the early detection of this deadly disease. We performed surface enhanced laser desorption ionization (SELDI) mass spectrometry using ProteinChip technology (Ciphergen Biosystems, Fremont, CA) to screen for differentially expressed proteins in pancreatic juice. Pancreatic juice samples obtained from patients undergoing pancreatectomy for pancreatic adenocarcinoma were compared with juice samples from patients with other pancreatic diseases. We identified a peak approximately 16,570 daltons present in the pancreatic juice from 10/15 (67%) of the patients with pancreatic adenocarcinoma and in the pancreatic juice from 1/7 (17%) of the patients with other pancreatic diseases. Using a ProteinChip immunoassay, we identified this differentially expressed protein as hepatocarcinoma-intestine-pancreas/pancreatitis-associated-protein I (HIP/PAP-I), a protein released from pancreatic acini during acute pancreatitis and overexpressed in hepatocellular carcinoma. We then quantified by ELISA the pancreatic juice HIP/PAP-I levels in 43 patients (28 with pancreatic adenocarcinoma, 15 with other pancreatic diseases) and the serum HIP/PAP-I levels in 98 patients (53 with pancreatic adenocarcinoma, 45 with other pancreatic diseases or healthy individuals). HIP/PAP-I levels were significantly higher in both the pancreatic juice (P < 0.001) and in the serum (P < 0.001) of patients with pancreatic adenocarcinoma compared with the control group. HIP/PAP-I levels were approximately 1000-fold higher in pancreatic juice compared with serum and the magnitude of the difference between the pancreatic adenocarcinoma group and the control group was greater in the pancreatic juice samples (143.75 +/- 235.52 microg/ml versus 6.04 +/- 7.59 microg/ml) than in the serum samples (99.96 +/- 140.66 ng/ml versus 35.25 +/- 28.44 ng/ml). In our study, patients with pancreatic juice HIP/PAP-I levels > or= 20 microg/ml were 21.9 times (95% confidence interval, 3.5-136.5; P < 0.001) more likely to have pancreatic adenocarcinoma than patients with levels <20 microg/ml. Immunolabeling of tissue sections revealed that the HIP/PAP-I protein was strongly expressed in acini adjacent to the invasive adenocarcinoma, but it was only rarely (1/30; 3%) expressed in the neoplastic epithelium, which suggests that the main source of HIP/PAP-I release in the pancreatic juice is acini. This low level of HIP/PAP-I expression in pancreatic adenocarcinoma was confirmed by reverse transcription-PCR: only 1 (5%) of 19 pancreatic cancer cell lines expressed HIP/PAP-I transcripts. Taken together, these data suggest that pancreatic juice measurement of HIP/PAP-I may help to identify patients with pancreatic adenocarcinoma.