Downloaded from http://journals.lww.com/jcge by BhDMf5ePHKbH4TTImqenVHV1elE1c3IGZm/FBm4rC7HfkINra4b9xuxk5mjVhsX0 on 09/20/2018 Gastric Varices in Patients With Portal Hypertension Evaluation With Multidetector Row CT Kangshun Zhu, MD, Xiaochun Meng, MD, Pengfei Pang, MD, Jiesheng Qian, MD, Min Shen, MD, Bin Hu, MD, and Hong Shan, MD Background: Gastric varices (GVs) are a major cause of gastro- intestinal bleeding in patients with portal hypertension. Few studies have evaluated GVs with multidetector row computed tomography (MDCT). Goals: To assess the diagnostic performance of MDCT in detecting GVs and revealing variceal hemodynamic changes in patients with cirrhosis. Study: A total of 127 consecutive cirrhotic patients who underwent both liver MDCT and esophagogastroduodenoscopy (EGD) were analyzed retrospectively. Two independent radiologists reviewed MDCT images for the detection of GVs. The variceal hemody- namic changes were assessed by the 2 radiologists in consensus on MDCT portography. Results: On the basis of EGD, of the 127 patients, 36 had GVs (28.4%), including small GVs in 15 patients and large GVs ( Z5 mm) in 21 patients. In detecting and grading GVs, there were moderate agreements (k value: 0.514 to 0.563) between MDCT and EGD, but in differentiating large varices requiring prophylactic therapy, a substantial agreement (k value: 0.804 for radiologist 1 and 0.796 for radiologist 2) was found. For radiologist 1, the sensitivity, specificity, accuracy, and positive and negative pre- dictive values of MDCT for the identification of large GVs were 85.7%, 96.2%, 94.5%, 81.8%, and 97.1%, respectively; whereas for radiologist 2, they were 81.0%, 97.2%, 94.5%, 85.0%, and 96.3%, respectively. In evaluating the afferent and efferent veins of varices, the sensitivity, specificity, accuracy, and positive predictive value of MDCT portography were more than 80.0%. Conclusions: MDCT is an effective screening tool for differentiating large GVs and revealing the afferent and efferent veins of varices in patients with cirrhosis. Key Words: gastric varices, cirrhosis, portal hypertension, computed tomography (J Clin Gastroenterol 2010;44:e108–e115) G astric varices (GVs) have been increasingly recognized as a major cause of gastrointestinal bleeding in patients with portal hypertension. 1 These varices are present in 5% to 33% of patients with portal hypertension with a reported incidence of bleeding of about 25% in 2 years, with a higher bleeding incidence for fundal varices. 2 Although GVs bleed less frequently than esophageal varices, 1 bleeding tends to be more severe, to require more transfusions, and to have a higher mortality rate than esophageal variceal bleeding. 3 The important predictors of bleeding risk from GVs include the presence of large varices ( Z5 mm diameter), red color spots, fundal location, and advanced Child-Pugh stage. 3–5 The presence of large varices is considered to be an indication of primary prophylaxis against variceal bleeding with nonselective b-blockers. 2,3,6,7 Patients without varices or with small varices (<5 mm diameter) are currently not candidates for prophylactic therapy, but need to undergo endoscopic surveillance to monitor for the development of large varices. 2,7,8 Screening esophagogastroduodenoscopy (EGD) is recommended for patients with portal hyperten- sion for evaluating the presence of gastroesophageal varices (GOV) and identifying patients at high risk of bleed- ing. 2,7–10 However, because of the invasion of EGD and the need of sedation in most cases, some patients are unwilling to accept EGD. 11–13 Therefore, a less invasive test with high sensitivity and specificity for the detection of large varices would allow for better selection of patients to undergo EGD for large varices. Several studies have evaluated possible noninvasive markers of GOV in patients with cirrhosis, such as the platelet count, Fibrotest, spleen size, portal vein diameter, and transient elastography. 14–18 But, the predictive accuracy of such noninvasive markers for high-risk varices is found to be unsatisfactory. 2,13 Multidetector computed tomography (MDCT) is increasingly accepted as a minimally invasive technique for imaging GOV, and the afferent and efferent veins of varices. 19–24 There have been some reports on CT diagnosis of esophageal varices in patients with liver cirrhosis, 19–21 but to our knowledge, few studies have evaluated the performance of MDCT in detecting GVs 21,22 ; however, they did not assess the accuracy of MDCT in grading GVs by comparing with EGD. In addition, the results of MDCT were also seldom compared with direct portography on visualizing the afferent and efferent vessels of varices, 22–24 which is important for deciding alternative therapeutic strategies such as balloon-occluded retrograde transvenous obliteration of varices. 25 Therefore, the purpose of our study was to investigate the accuracy of MDCT in patients with cirrhosis for detecting GVs and differentiating large GVs from small or no GVs, and revealing the afferent and efferent vessels of GVs. These findings were compared with the results of EGD and direct portography, respectively. MATERIALS AND METHODS Patients By querying our institutional database, 2 of the authors retrospectively identified 134 consecutive patients Copyright r 2010 by Lippincott Williams & Wilkins Received for publication April 21, 2009; accepted September 11, 2009. From the Department of Radiology, the Third Affiliated Hospital, Sun Yat-sen University, Guangdong Province, China. Conflict of interest disclosure and funding declaration: None declared. Reprints: Hong Shan, MD, Department of Radiology, the Third Affiliated Hospital, Sun Yat-sen University, 600 Tianhe Road, Guangzhou, Guangdong Province 510630, China (e-mail: shanhong5@gmail.com; zhksh010@163.com). ORIGINAL ARTICLE e108 | www.jcge.com J Clin Gastroenterol  Volume 44, Number 5, May/June 2010