AJR:172, February 1999 383 Local Invasion of Gastric Cancer: CT Findings and Pathologic Correlation Using 5-mm Incremental Scanning, Hypotonia, and Water Filling M. Rossi1 L. Broglia1 P. Graziano2 F. Maccioni1 M. Bezzi1 R. Masciangelo3 P. Rossi1 OBJECTIVE. Our purpose was to assess the accuracy of CT with drug-induced hypotonia and water filling in revealing the depth of tumor invasion of the gastric wall, according to the T factor of TNM classification, and to verify the capability of this technique in differentiating diffuse from intestinal gastric cancer. SUBJECTS AND METHODS. Forty patients (age range, 35-78 years) with histologi- cally proven gastric tumors underwent CT, in the prone position, with drug-induced hypoto- nia and water filling. The images were prospectively reviewed by two radiologists who were asked to assess the depth of tumor invasion in the gastric wall. The thickening of the hy- pod ense layer and the contrast enhancement of lesion were measured. RESULTS. CT correctly assessed gastric wall invasion in 77% and 82% of cases for ob- servers A and B, respectively; overstaging was 20% and 15%, respectively; and understaging occurred in 3% of cases for both observers. Diagnostic sensitivity for serosal invasion was 100% for both observers; specificity was 80% and 87%, respectively. Substantial agreement between the observers was obtained (K = .6). Diffuse and intestinal cancers could be differen- tiated by CT in 92% of cases, considering the thickening of the hypoattenuating layer of the gastric wall (diffuse cancer: 7 ± 1.2 mm; intestinal cancer: 1.4 ± 0.4 mm) and contrast en- hancement (diffuse cancer: 85 ± 8.2 H; intestinal cancer: 5 1 ± 3 H). CONCLUSION. CT with patients in a drug-induced hypotonia and in a prone position, and using water filling, is a promising technique for evaluating the depth of tumor invasion and for differentiating intestinal from diffuse gastric cancer. Received October 27, 1997; accepted after revision July9, 1998. 1 Department of Radiology, University of Rome La Sapienza,” Via Giulio Romano 38, 00197 Rome, Italy. Address correspondence to L Broglia. 2Department of Experimental Medicine and Pathology, University of Rome La Sapienza,” 00197 Rome, Italy. 3Oepartment of Experimental Medicine, University of Rome ‘La Sapienza,” 00197 Rome, Italy. AJR 1999:172:383-388 0361-803X/99/1722-383 © American Roentgen Ray Society S tandard protocols for abdominal CT examinations of the gastric wall using a hyperdense oral con- trast agent have many limitations because of overshooting artifacts from intraluminal air- fluid interfaces I I ]. Moreover, movement due to peristalsis, partial filling of the lumen, and incomplete distention of the gastric wall often mimics focal thickening of the gastric wall and further decreases image quality. In the era before helical CT, many attempts were made by several investigators to avoid technical artifacts and to improve the image quality ofconventional Cf studies [2, 31. Beam- hardening artifacts may be reduced by filling the stomach with water, peristalsis may be avoided with drug-induced hypotonia, and opti- mal filling of the gastric antrum can be obtained by placing the patients in the prone position aS- ter water administration. Thin incremental scan- ning and a small field of view improve image resolution: however, slice collimation of less than 5 mm may increase the acquisition time with conventional equipment and reduce the ef- feet ofthe IV contrast material. In examining the gastric wall, helical CT also has limitations caused by peristalsis, beam- hardening artifacts, and suboptimal gastric dis- tention; therefore, hypotonia, the prone position, and water filling are important with any type of equipment. On the other hand, fast scanning al- lows the reduction of slice collimation and pro- vides imaging at different phases after contrast injection, improving both spatial and contrast resolution. Because of the still-low availability of helical CT scanners in many countries, we have improved a technique for conventional CT that can be used with any type of equipment. When adequately distended at 5-mm incre- mental scanning, the gastric wall appears as a thin multilayered structure, with the same thickness in every aspect of the stomach 14. 5]. In such a scenario, in most cases, gastric cancer appears as an area ftxally destroying this multi- layered structure, and it is well demarcated from the adjacent uninvolved gastric wall.