Author(s): Rsch T, Dittler HJ, Strobel K, Meining A, Schusdziarra V,
Abstract Share this page
Abstract BACKGROUND: It has been claimed in several prospective studies that endoscopic ultrasonography (EUS) is highly accurate in the locoregional staging of pancreatic cancer. However, the value of the EUS criteria for the diagnosis of vascular involvement is less well established. To totally exclude potential bias introduced by the availability of prior information, a completely blinded analysis of videotapes of patients with cancer of the pancreatic head was therefore conducted. METHODS: Videotape sequences of 75 patients with cancer of the head of the pancreas with surgical confirmation or unequivocally positive angiography demonstrating vascular invasion were reevaluated without any clinical data or information from other imaging studies. Involvement of the vascular system (portal vein with confluence, superior mesenteric vein, celiac axis) was assessed on EUS with special emphasis on EUS parameters of the tumor-vessel relationship. RESULTS: The overall sensitivity and specificity of EUS in the diagnosis of venous invasion were 43\% and 91\%, respectively, when using predetermined parameters (visualization of tumor in the lumen, complete obstruction, or collateral vessels). If the parameter "irregular tumor-vessel relationship" had been added to these criteria, the sensitivity would have risen to 62\%, but the specificity would have fallen to 79\%. The only vascular system that could be properly visualized by EUS was the portal vein/confluence area. The positive and negative predictive values for the single parameters chosen to diagnose portal venous involvement were as follows: 42\% and 33\% for irregular tumor-vessel relationship, 36\% and 34\% for visualization of tumor in the vascular lumen, 80\% and 28\% for complete vascular obstruction, and 88\% and 18\% for collateral vessels. CONCLUSIONS: In a completely blinded evaluation of the EUS diagnosis of vascular invasion by cancer of the head of the pancreas it was not possible to find suitable morphologic parameters with clinically useful sensitivity and specificity values (over 80\%).
This article was published in Gastrointest Endosc
and referenced in Journal of Vascular Medicine & Surgery