Staging urinary bladder cancer: value of T1-weighted three-dimensional magnetization prepared-rapid gradient-echo and two-dimensional spin-echo sequences.

Abstract
The purpose of this study was to evaluate a magnetization prepared-rapid gradient-echo (MP-RAGE) sequence as a three-dimensional T1-weighted MR imaging technique for staging urinary bladder cancer and to compare this technique with a commonly used two-dimensional T1-weighted spin-echo sequence technique. For 28 consecutive patients with urinary bladder cancer, MR findings and staging results were compared with histopathologic findings after surgery or autopsy. MR imaging was performed at 1.5 T with a Helmholtz double-surface coil. Conventional T1-weighted spin-echo, three-dimensional MP-RAGE, T2-weighted spin-echo or turbo-spin-echo, and dynamic T1-weighted fast gradient-echo sequences were used. Signal difference-to-noise ratios and T1 contrast were calculated by use of operator-defined regions of interest. The signal difference-to-noise ratios for fluid-tumor and fat-tumor were, respectively, factors of 1.6 and 2.7 better with T1-weighted spin-echo sequences. T1 contrast of fluid-tumor was a factor of 2.6 better with three-dimensional MP-RAGE sequences, resulting in better recognition of small tumors, ascites, and dilated ureters. T1 contrast for fat-tumor was a factor of 2.0 better with T1-weighted spin-echo sequences. With the MP-RAGE sequence, motion artifacts were fewer in number than those noted with the T1-weighted spin-echo sequence, and susceptibility artifacts were equal in number to those noted with the T1-weighted spin-echo sequence. Using the three-dimensional technique, we performed off-line reconstruction of 1- to 2-mm high-resolution images in every desired plane. Because of higher spatial resolution, the availability of multiplanar reconstructions, and better fluid-tumor contrast and despite lower signal difference-to-noise ratios, three-dimensional MP-RAGE images resulted in better recognition of local tumor extension (n = 11), adhesions and bowel wall invasion (n = 5), lymph node metastases (n = 2), and bone marrow metastases (n = 2). The staging accuracy for the combination of three-dimensional MP-RAGE, T2-weighted, and dynamic sequences was 93%; that for the combination of two-dimensional T1-weighted spin-echo, T2-weighted, and dynamic sequences was 78%. Nodal staging was also more accurate with MP-RAGE sequences (accuracy of 93% vs 86% for T1-weighted spin-echo sequences). Compared with two-dimensional T1-weighted spin-echo imaging, three-dimensional MP-RAGE imaging resulted in a 15% improvement in staging. Our findings suggest that optimal staging of urinary bladder carcinoma requires three-dimensional imaging techniques.