Esophageal capsule endoscopy for screening and surveillance of esophageal varices in patients with portal hypertension See Editorial on Page 1434

Abstract

Bleeding from esophageal varices (EV) is a serious consequence of portal hypertension. Current guidelines recommend screening patients with cirrhosis with esophagogastroduodenoscopy (EGD) to detect varices. However, the unpleasantness and need for sedation of EGD may limit adherence to screening programs. Pilot studies have shown good performance of esophageal capsule endoscopy in detecting varices. This multicenter trial was designed to assess the diagnostic performance of capsule endoscopy in comparison with EGD. Patients undergoing EGD for screening or surveillance of EV underwent a capsule study previously. The study was designed as an equivalence study, assuming that a difference of ≤10% between capsule endoscopy and EGD in diagnosing EV would demonstrate equivalence. Two hundred eighty-eight patients were enrolled. Endoscopy was for screening in 195 patients and for surveillance of known EV in 93. Overall agreement for detecting EV between EGD and capsule endoscopy was 85.8%; the kappa score was 0.73. Capsule endoscopy had a sensitivity, specificity, positive predictive value, and negative predictive value of 84%, 88%, 92%, and 77%, respectively. The difference in diagnosing EV was 15.6% in favor of EGD. There was complete agreement on variceal grade in 227 of 288 cases (79%). In differentiating between medium/large varices requiring treatment and small/absent varices requiring surveillance, the sensitivity, specificity, positive predictive value, and negative predictive value for capsule endoscopy were 78%, 96%, 87%, and 92%, respectively. Overall agreement on treatment decisions based on EV size was substantial at 91% (kappa = 0.77). Conclusion: We recommend that EGD be used to screen patients with cirrhosis for large EV. However, the minimal invasiveness, good tolerance, and good agreement of capsule endoscopy with EGD might increase adherence to screening programs. Whether this is the case needs to be determined. (H EPATOLOGY 2008;47:1595–1603.)