Non-operative differentiation between pancreatic cancer and chronic pancreatitis.

Eighty-five of 186 patients investigated for suspected pancreatic cancer had an unequivocal final diagnosis of either pancreatic cancer (58 patients) or chronic pancreatitis (27 patients). They had been studied prospectively using ultrasonography, computerized tomography, radionuclide scanning, endoscopic retrograde cholangiopancreatography (ERCP), selective celiac and superior mesenteric angiography, duodenal drainage studies, cytologic studies, serum carcinoembryonic antigen assay, and pancreatic oncofetal antigen assay, The results were compared to determine which test would most frequently and reliably differentiate between pancreatic cancer and pancreatitis in a patient believed to have one or other disease. Criteria for interpreting results, first for highest rate of correct diagnoses, and second for highest accuracy were derived. Applying these criteria, ultrasonography achieved the highest rate of correct diagnoses (97% of patients diagnosed with 84% accuracy). ERCP, duodenal drainage studies, and cytology were the most accurate tests ((86% accuracy each test) but, with this accuracy, ERCP most frequently gave a diagnosis (diagnosis rate: ERCP--70%, duodenal drainage--32%, cytology--35%). The results suggest that ultrasonography is the best noninvasive test, and that a combination of ERCP, pancreatic juice assay and cytology in a single procedure may prove to be the best discriminating investigation.