1288P Efficacy and safety of palliative procedures for malignant gastric outlet obstruction: A systematic review and network meta-analysis

Abstract

Only head-to-head trials evaluated between gastrojejunostomy (G), endoscopic stenting (S), stomach partitioning gastrojejunostomy (P), and endoscopic ultrasound-guided gastroenterostomy (E) for malignant gastric outlet obstruction (GOO). We conducted a network meta-analysis (NMA) to compare the efficacy and safety among these treatments. We searched for randomized controlled trials (RCTs) and non-RCTs that compared palliative surgical strategies for malignant GOO from Pubmed, Medline, Embase, ClinicalTrial.gov, and Cochrane databases. We included studies that reported at least one outcome of interest (clinical success, mortality, and reintervention rate). Evidence from RCTs and non-RCTs were combined through the frequentist framework, inverse variance model, and naïve combination. We estimated the effective results using relative risks (RR). The probability of best treatment was ranked by P-score (0-1), a higher score indicating better treatment. This NMA included 4 RCTs and 37 non-RCTs (4 prospectives and 33 retrospectives) with 3462 patients. The overall rates of clinical success, mortality, and reintervention were 85.3%, 7.95%, and 17.07%, respectively. P was ranked the best approach for clinical success rate and reintervention rate (P-score: 0.92, 0.80, respectively). E had the highest probability of being the best for mortality rate (P-score: 0.8). Cluster rank combined the probability of the best treatment for safety regarding mortality and efficacy regarding reintervention prevention showed the benefit of P and E (cophenetic correlation coefficient: 0.90, P and E were in the same cluster). Stomach partitioning gastrojejunostomy and endoscopic ultrasound-guided gastroenterostomy should be recommended for malignant GOO. Stomach partitioning gastrojejunostomy could be an alternative approach when endoscopic ultrasound-guided gastroenterostomy is unsuccess or not feasible.