Malignant biliary obstruction: endoscopic ultrasound-guided versus endoscopic retrograde cholangiopancreatography-guided biliary drainage

See also: Endoscopic ultrasound-guided transmural approach versus ERCP-guided transpapillary approach for primary decompression of malignant biliary obstruction: a meta-analysis Endoscopy 2019; 51(10): 950-960 DOI: 10.1055/a-0901-7343 See also: Reply to Jin et al. Endoscopy 2019; 51(12): 1186-1186 DOI: 10.1055/a-1012-1931 We read with interest the meta-analysis by Bishay et al. comparing endoscopic ultrasound-guided biliary drainage (EUS-BD) and endoscopic retrograde cholangiopancreatography (ERCP)-guided drainage in the primary treatment of malignant biliary obstruction [ 1 ]. The authors concluded that EUS-BD showed similar clinical success rates to ERCP. However, we have some questions as follows. First, all included studies were for distal malignant biliary obstruction, but the title suggested comparison for all levels of obstruction (hilar and distal). A more accurate title would have been better. Second, study selection was questionable. It was not clear whether biliary drainage was performed as the primary treatment in the Ridtitid et al. study [ 2 ]; this study should be excluded. Another study [ 3 ] was excluded because patients had indwelling duodenal stents; however, the inclusion/exclusion criteria did not preclude these cases from inclusion. Third, “mixing apples and oranges” – definitions among studies were not congruent: clinical success was defined as a bilirubin decrease within 4 weeks by Kawakubo [ 4 ] and Park [ 5 ], within 2 weeks by Ridtitid [ 2 ] and Bang [ 6 ], and within 1 week by Paik [ 7 ]. For outcome procedure time, Paik used a “one-step” device that expedited the process of EUS-BD. Patients with surgically altered anatomy were included in the Paik study, but excluded in the Park and Bang studies. It is well known that surgically altered anatomy will render ERCP more difficult and prolong the procedure time. For adverse events, stent dysfunction was considered an adverse event by Paik, but not by Kawakubo and Park. Finally, when extracting data from the Paik study, the authors incorrectly reported no stent occlusion in the EUS-BD group and three in the ERCP group; in fact, there were 8 and 23, respectively. In addition, the authors incorrectly reported no stent migration in the EUS-BD group, when in fact, there were two. We think these issues might reduce the strength of the study outcomes and must be recognized.