Interobserver agreement in staging gastric malt lymphoma by EUS

Background: EUS is highly accurate for staging gastric lymphoma. Because stage correlates to outcome, interobserver agreement is mandatory. However, previous studies suggest that interobserver agreement might vary according to T-stage. A multicenter evaluation of observer agreement with respect to endosonographic staging of gastric mucosa-associated lymphoid tissue (MALT) lymphoma was therefore conducted. Methods: Fifty-four patients were studied; 42 were also evaluated after eradication of Helicobacter pylori infection. EUS was performed at different institutions by 10 experienced endosonographers who collected the best photographs for each examination. Interobserver agreement was estimated with kappa statistics. Results: Overall interobserver agreement for T-stage was fair, both before and after treatment (κ = 0.38 and κ = 0.37, respectively). Overall interobserver agreement for N-stage was substantial before treatment, but only fair after treatment (κ = 0.63 and κ = 0.34, respectively). The lowest values of agreement occurred with T1sm (submucosa) and T2 stage lesions. Conclusions: Interobserver agreement for staging of gastric MALT lymphoma by EUS is suboptimal before as well as after treatment of H pylori infection. This evidence suggests that gastric EUS may be more difficult technically compared with EUS of other organs. Lack of agreement is crucial because it influences the choice of therapy and assessment of response to treatment. Good interobserver agreement would permit better communication concerning the clinical status of patients, comparison of the results of different studies, and stratification of patients within clinical trials. Copyright © 2002 by the American Society for Gastrointestinal Endoscopy.