| Abstract: | This prospective study evaluates the ability of MRI using T1-weighted fat-suppressed spin-echo (T1FS) and dynamic gadolinium chelate (Gd) enhanced spoiled-gradient echo (SGE) to detect the presence of pancreatic tumor in patients in whom spiral CT findings are inconclusive. Sixteen consecutive patients who underwent spiral CT and had findings that were considered inconclusive for pancreatic tumor underwent MR within 2 weeks of CT. Spiral CT and MR images were interpreted in a prospective fashion by separate individual investigators blinded to the results of the other imaging modality. CT was performed on a spiral CT scanner. MRI was performed on a 1.5-T MR machine. Imaging sequences included T1FS pre-Gd and post-Gd and SGE pre-Gd and immediately post-Gd. Data were analyzed using receiver operating characteristic (ROC) analysis. Confirmation was obtained by pancreatic biopsy (n = 4), surgical resection (n = 1), and clinical imaging (n = 4) or clinical follow-up (n = 7). MRI was superior to spiral CT (P = .027) in this selected patient group at detecting or excluding pancreatic tumor by ROC analysis, with areas under the curve of .982 and .764, respectively, which was significant (P = .041). The greatest advantage of MRI was in patients in whom spiral CT demonstrated enlargement of the pancreatic head without clear definition of tumor, which was significant (P = .033). In 10 patients with this CT appearance, MRI demonstrated a high confidence for presence of tumor in four and a high confidence of absence in six. Association of imaging findings with patient diagnosis was significant for MRI (P = .001) but not significant for CT (P = .148). The results of our study suggest that MRI may add significant diagnostic information in patients in whom spiral CT is inconclusive for the presence of pancreatic tumor. The greatest advantage of MRI was in the evaluation of patients in whom spiral CT findings revealed an indeterminate enlarged pancreatic head. |
