Parametrial invasion in cervical carcinoma: evaluation of detection at MR imaging with fat suppression.

Abstract

To evaluate detection of parametrial invasion at magnetic resonance (MR) imaging with fat suppression. In 35 patients with cancer of the cervix, MR imaging was performed with the following sequences: T2-weighted turbo spin echo (SE), turbo short inversion time inversion recovery (STIR), and unenhanced or gadolinium-enhanced T1-weighted SE with and without excitation-spoiling fat suppression. Images obtained with each sequence were evaluated for parametrial invasion by two blinded radiologists separately, who scored their level of diagnostic confidence from 0 to 10 (no confidence to high confidence). Then, all images were evaluated together and findings were correlated with histopathologic findings. No statistically significant differences were found between staging with T2-weighted turbo SE, turbo STIR, and T1-weighted fat-suppressed gadolinium-enhanced SE images. Staging with T1-weighted SE, T1-weighted gadolinium-enhanced SE, and T1-weighted fat-suppressed SE images was significantly worse (P < .05). Diagnostic confidence was lower (P < .001) with T1-weighted gadolinium-enhanced SE and T1-weighted fat-suppressed gadolinium-enhanced SE images (5.5 and 6.2 points, respectively) compared to staging with T2-weighted turbo SE and turbo STIR images (8.2 and 7.6 points, respectively). No statistically significant improvement in staging accuracy was found when all images were evaluated together. In MR diagnosis of parametrial invasion, the addition of fat-suppressed or gadolinium-enhanced MR images did not improve the accuracy with T2-weighted turbo SE images alone.