The objective of our study was to determine the negative predictive value of MDCT with curved planar reformations for detecting vascular invasion and predicting overall resectability in patients with pancreatic adenocarcinoma. Imaging findings related to vascular invasion and overall tumor resectability in 25 patients who underwent contrast-enhanced biphasic MDCT evaluation were correlated with actual vessel invasion and overall resectability determined at surgery and pathologic examination. The presence of vascular invasion was assessed in 110 major peripancreatic vessels in 22 patients who underwent resection. On MDCT, 23 (92%) of 25 patients were deemed to have resectable pancreatic adenocarcinoma. The tumors in the remaining two (8%) were considered not resectable because of the presence of vascular invasion (which was confirmed in only one patient at surgery). Of those 23 patients deemed to be candidates for curative resection on the basis of MDCT results, 20 were found to have resectable adenocarcinoma at time of surgery, yielding a negative predictive value for MDCT of 87% (20/23 patients) for overall resectability. In the other three patients, adenocarcinoma was deemed to be unresectable because of small metastases to the liver (two patients) or to the peritoneum (one patient) discovered at surgery. For detection of vascular invasion, MDCT yielded a negative predictive value of 100% (108/108 vessels) with no false-negative findings and an accuracy of 99% (109/110 vessels) with 108 true-negative findings, one true-positive finding, and one false-positive finding. Our preliminary data on MDCT show that the technique has excellent negative predictive value for vascular invasion and good negative predictive value for overall tumor resectability in patients with pancreatic adenocarcinoma, suggesting an improvement over previous results reported using single-detector CT. The problem of undetected micrometastases to the liver and peritoneum remains.
Read full abstract