Assessment of EUS for diagnosing, staging, and determining resectability of pancreatic cancer: A review

Abstract

tumor resection found a 5-year survival rate of 3.5%. More recent studies have found improved 5- and 7year survival rates of 19% and 11%, respectively, in highly selected surgical patients.2 This may be due to several factors, including improved surgical technique, the use of adjuvant chemoradiation, earlier diagnosis, and more extensive preoperative staging. Improved preoperative staging theoretically benefits patient selection for surgery. EUS has been one of the most sensitive tests for detecting pancreatic masses, and it is an accurate modality for the evaluation and staging of pancreatic cancer (Fig. 1). In most reports, T-stage accuracy varies from 78% to 94% and N-stage accuracy ranges from 64% to 82%.3-8 A recent retrospective review by Ahmad et al.9 found lower T- and N-staging accuracies of 69% and 54%, respectively. A majority of tumors in this study were of advanced T-stage (T3 or T4), and these investigators hypothesize that peritumoral inflammation and large tumor size may affect US attenuation and thus, T- and N-staging accuracy. This is supported by previous studies, which indicate that EUS is more effective in detecting vascular involvement in tumors less than 3 cm in size.10,11 In their study, Ahmad et al.9 fail to define the number of patients excluded because of the discovery by EUS of significant vascular invasion or metastatic disease, precluding surgery. Factoring this additional cohort of patients into the denominator of T- and Nstaging would have improved their accuracy rates. Sensitivity for detecting vascular invasion and predicting surgical resectability has been greater than 90% in some studies.3,4,12 Vessels typically invaded by cancer in the head of the pancreas are the portal vein and splenoportal confluence, superior mesenteric vein, and splenic vein; and the superior mesenteric artery, splenic artery, hepatic artery, and celiac axis. Generally, it may be difficult to visualize invasion of the superior mesenteric artery and vein with EUS, but portal vein and splenic vein invasion are well seen.12,13 A study by Brugge et al.14 confirmed the inadequacy of EUS in the determination of superior mesenteric vein involvement but found EUS to be superior to angiography in determining portal and splenic vein invasion with tumor, with accuracy ranging from 77% to 85%, depending on criteria for involvement. In a prospective study of 38 patients with pancreatic neoplasm by Snady et al.,15 all 21 patients exhibiting either peripancreatic collaterals, loss of vessel interface, or tumor within the lumen of a major peripancreatic vessel had confirmed vascular invasion at surgery, and the absence of these features in the remaining 17 patients was associated with absence of vascular invasion at surgery. A potential fault of EUS is that accuracy in predicting vascular invasion may be falsely elevated because endosonographers have access to previous image studies, such as standard or helical CT, magnetic resonance imaging (MRI) or magnetic resonance cholangiopancreatography, or angiography. Thus, there may already be suspicion of vascular invasion because of examination bias introduced by availability of other information. This has been studied by review of EUS videotapes of cancers of the pancreatic head.13 In this study by Rosch et al., EUS videotapes of pancreatic head cancers staged from 1991-95 were retrospectively reviewed in 1997 with regard to vascular invasion. Sensitivity and specificity of EUS in the diagnosis of venous invasion were 62% and 79%, respectively. These values were much lower than those of previous studies,