Preoperative staging of cancer of the pancreas: value of MR angiography versus conventional angiography in detecting portal venous invasion.

Objective: The purpose of this study was to compare contrast-enhanced MR angiography with conventional catheter angiography for detecting portal venous invasion in the preoperative staging of pancreatic cancer, using the surgical confirmation of vascular involvement as the standard of truth.

Subjects And Methods: MR and conventional angiography were performed in 20 patients with pancreatic carcinoma, with surgical confirmation in all cases. MR angiography was performed at 1.5 T, with coronal (2.9 mm) and axial (6.0 mm) contrast-enhanced breath-hold two-dimensional time-of-flight imaging. Data from each imaging technique were collected prospectively and analyzed in a blinded fashion by expert vascular radiologists. Vascular involvement in each patient and in each vessel (main portal vein, confluence, splenic vein, and superior mesenteric vein) determined whether the tumor was resectable (normal, abutment) or nonresectable (encased, occluded). Surgical confirmation of the vascular involvement of the portal venous structures was used as the standard of truth in all patients.

Results: Among the 20 patients, 11 tumors were surgically resectable and seven were nonresectable with performance of a palliative bypass. MR angiography and conventional angiography had an overall concordance in 65% of patients (13/20; seven resectable, four nonresectable, two false-negatives) on the basis of the vascular status in each patient of the portal venous structures and in 84% (47/56) of the individual vessels surgically confirmed. MR angiography correctly identified 11 of 11 resectable patients and five of nine nonresectable patients, with four false-negative cases. Conventional angiography correctly identified seven of 11 resectable patients and six of nine nonresectable patients, with three false-negative cases and four false-positive cases.

Conclusion: The lack of false-positives by MR angiography suggests that MR imaging may provide a noninvasive screen for nonresectability on the basis of vascular involvement, with no patients with potentially resectable tumors being denied surgery by MR angiography in this cohort. However, the presence of false-negatives using MR angiography indicates the procedure would still not fully eliminate unnecessary laparotomies.