T staging of gastric cancer: role of multi-detector row CT.

PURPOSE: To evaluate retrospectively the accuracy of multi-detector row computed tomography (CT) in the assessment of serosal invasion in patients with gastric cancer.
MATERIALS AND METHODS: The Ethics Committee does not require approval or informed consent for retrospective studies. Forty-one consecutive patients (24 men, 17 women; mean age, 68 years) with gastric cancer were included in this study. All patients were given 600 mL of tap water to drink and were positioned prone or supine on the scanning table. The detector row configuration included four detector rows, a section thickness of 1.25 mm, a pitch of 6, and a reconstruction interval of 0.63 mm. Transverse and multiplanar reconstruction images were simultaneously evaluated by two independent observers to assess the depth of tumor invasion in the gastric wall (ie, T stage). T staging at multi-detector row CT was compared with T staging at histologic evaluation (reference standard), which was performed by means of surgical or histologic examination of the resected specimen. We also calculated the sensitivity, specificity, and accuracy of multi-detector row CT for each observer in the assessment of serosal invasion.
RESULTS: Analysis of interobserver agreement showed substantial or almost perfect agreement (nonweighted kappa value of 0.78 and weighted kappa value of 0.85). Correct assessment of gastric wall invasion was 80% and 85% for observers 1 and 2, respectively. The sensitivity, specificity, and accuracy of multi-detector row CT in the assessment of serosal invasion were 90%, 95%, and 93%, respectively, for observer 1 and 80%, 97%, and 93%, respectively, for observer 2. Overstaging occurred in six patients, and understaging occurred in five patients. All understaged tumors were scirrhous subtype gastric cancer.
CONCLUSION: Multi-detector row CT scanning of patients with gastric cancer gave 93% accuracy in the assessment of serosal invasion in patients with gastric cancer. RSNA, 2005