Prospective evaluation of angiography in the diagnosis and management of patients suspected of having pancreatic cancer.

One hundred three patients suspected of having pancreatic cancer underwent celiac and superior mesenteric arteriography which was reported 'blind.' Final diagnosis, operative findings, and resectability of pancreatic cancers were correlated with angiography reports. Forty patients had pancreatic cancer. Thirty-seven of these underwent laparotomy, and 14 (37%) had resectable lesions. Sensitivity and specificity of arteriogrphy for pancreatic cancer were 72% (29/40) and 71% (45/63) respectively. Resectability was unrelated to the angiographic diagnosis. Nonresectability was related to an angiographic finding of definite tumor vessel (4/4), large-vessel encasement (14/7), and major venous pathology (8/11). Only the first of these was unique to pancreatic cancer. Hepatic metastases were detected in only six of 13 patients. Small-artery encasement, arterial occlusion, vessel displacement, and gallbladder distention were not indicators of diagnosis or extent of disease. Major anatomic anomaly of foregut vasulature was revealed in 25% of angiographic studies. Prior knowledge of such anomalies has important bearing on the operative approach to pancreatic resection. We conclude that angiography should be performed prior to laparotomy whenever a resectable pancreatic cancer is suspected from other investigations. When used prospectively, angiography is not a reliable diagnostic or prognostic tool for pancreatic cancer.