Dual Energy CT as a Noninvasive Method to Screen for Gastroesophageal Varices

Gastroesophageal Varices

Cirrhosis leads to portal hypertension and development of gastroesophageal varices, which are the most common cause for bleeding in cirrhosis and a major cause of death. The American Association for the Study of Liver Disease (AASLD) recommends screening endoscopy every 2 years to evaluate for gastroesophageal varices, and annual surveillance for those with small varices on endoscopy. Unfortunately, endoscopy is costly, requires sedation, is poorly tolerated, is subject to high inter-observer variability, and is associated with risks that include bleeding, esophageal injury and aspiration. Noninvasive methods for evaluation of gastroesophageal varices are needed. CT is noninvasive, rapid, less expensive than endoscopy, requires no sedation, provides a quantitative measure of the size of the varices, and allows for assessment of para-esophageal varices, varices in other body locations, ascites, other signs of portal hypertension, patency of liver vasculature, and detection, diagnosis and staging of hepatocellular carcinoma. Single-Energy CT (SECT) has relatively high accuracy in prospective studies for detection of any and large varices but is associated with suboptimal contrast opacification of gastroesophageal varices. Dual-Energy CT with the GE scanners with GSI Xtream (DECT) improves the contrast-to-noise ratio by 60% compared to SECT and is currently standard of care at UAB for evaluation of cirrhosis. The primary objective of this study is to determine the accuracy of DECT for detecting any varices and high-risk varices. The study hypothesis is that the accuracy (AUROC) of DECT will be \>0.90 and \>0.95 for detecting any and high-risk varices in a prospective pilot study (N=50) that uses endoscopy as the reference standard. This will be a single-center pilot observational prospective IRB-approved study. A total of 50 adult patients presenting to UAB Endoscopy for surveillance endoscopy to detect and grade gastroesophageal varices will be enrolled.

Cirrhosis leads to portal hypertension and development of gastroesophageal varices, which are the most common cause for bleeding in cirrhosis and a major cause of death. Bleeding varices have a 6-week mortality of 15%-25%. About 50% of patients with cirrhosis have varices, and 30% have large varices (\>5 mm) that are high risk for bleeding. The American Association for the Study of Liver Disease (AASLD) recommends screening endoscopy every 2 years to evaluate for varices, and annual surveillance for those with small varices on endoscopy. Patients at a high risk of bleeding with large varices, small varices and red wale signs (an endoscopic finding), or small varices and decompensated cirrhosis proceed to treatment such as prophylactic band ligation and beta blockers. Conversely, patients with no varices or small varices (≤5 mm) continue surveillance efforts by endoscopy to monitor for development of large varices. Unfortunately, endoscopy is costly, requires sedation, is poorly tolerated, is subject to high inter-observer variability, cannot detect other signs or portal hypertension or para-esophageal varices that are at risk for future bleeding events, and is associated with risks that include bleeding, esophageal injury and aspiration. Many of these factors contribute to poor patient compliance with AASLD recommendations. Noninvasive methods for detecting, grading, and risk stratification of esophageal varices are needed. Imaging tests such as ultrasound elastography to measure liver stiffness have been proposed as a method to predict the presence of varices but have insufficient accuracy to eliminate the need for endoscopy.10 An ideal biomarker to screen for esophageal varices would be part of the routine standard of care of patients with cirrhosis, noninvasive, rapid, less expense than endoscopy, highly accurate, highly reproducible, and would require no sedation, provide a quantitative measure of the size of the varices, provide a mechanisms to assess the risk of future bleeding, allow for an assessment for other signs of portal hypertension, and provide other benefits to the patient (e.g. detect ascites and HCC and assess liver vasculature). Computed tomography (CT) is standard of care to screen for HCC. CT is noninvasive, rapid, less expensive than endoscopy, requires no sedation, provides a quantitative measure of the size of the varices, and allows for assessment of para-esophageal varices, varices in other body locations, ascites, other signs of portal hypertension, patency of liver vasculature, and detection, diagnosis and staging of HCC. Conventional Single-Energy CT (SECT) has relatively high accuracy in prospective studies for detection of any and large varices and has higher inter-observer agreement than endoscopy (kappa 0.56 vs. 0.36, respectively). Major deficiencies in SECT include relatively suboptimal contrast opacification of gastroesophageal varices, inconsistent accuracy that is dependent upon SECT image acquisition technique, and suboptimal stratification of the risk of bleeding (e.g. inability to detect red wale sign) compared to endoscopy. Dual-Energy CT (DECT) improves the contrast-to-noise ratio by 60% compared to SECT. DECT also improves visualization by taking advantage of the markedly increased attenuation of iodine at photon energy levels just above the iodine K edge (33 keV). Using material decomposition techniques, DECT can map the concentration of iodine on a voxel by voxel basis which, combined with higher contrast to noise resolution on these same type of images, improves the conspicuity of enhancing structures. DECT is routinely used to screen for HCC in cirrhotic patients. While DECT has been shown to improve image quality and portal venography compared to SECT, the accuracy of DECT for screening for varices has not been reported. The primary objective is to determine the accuracy of dual energy CT for detecting any varices and high-risk varices in patients with cirrhosis presenting for upper gastrointestinal endoscopy.

No related papers found yet.

United States

Participants by arm