OBJECTIVETo investigate the therapeutic efficacy of endoscopic variceal ligation (EVL) for treating various types of gastric varices (GV) by reviewing patient cases in order to identify the influencing factors of EVL-related recurrence and rebleeding.

METHODSThe medical records of 101 GV cases treated by EVL in our department between January 2008 and October 2012 were retrospectively reviewed. The cases were grouped according to GV type: type 1 (GOV1, n = 63), type 2 (GOV2, n = 18), GOV1 coexisting with GOV2 (n = 11), isolated GOV1 (IGV1, n = 9), and GOV2 (IGV2, n = 0). Data from follow-up examinations (range: 1.5 - 48.0 months, average: 14.9 +/- 9.1 months) were extracted for analysis and included early (less than 72 h after EVL) and late (72 h to 6 weeks after EVL) rebleeding and recurrence. In addition, data from computed tomography (CT) or CT angiography (CTA), performed in 32 of the patients, were extracted to determine the influence of supplying veins, gastrorenal or splenorenal shunts, and portal vein and/or splenic vein diameters on GV recurrence. Data analysis was carried out by ANOVA, Chi-square, Fisher's exact or rank-sum tests, as appropriate. Kaplan-Meier analysis was used to evaluate the time of first recurrent bleeding, and the log-rank test was used to compare between-group differences.

RESULTSGOV2 and IGV1 varices were more severe than the varices of GOV1 (GOV2 and GOV1: u = -2.960; IGV1 and GOV1: u = -2.871; both P less than 0.05). GOV1 had a significantly lower recurrence rate than all other GV types (x2 = 7.054, P less than 0.05). The CT and CTA data indicated that all GV were supplied by left gastric veins, while 83.3% of IGV1 had blood supplementation by left gastric veins and short gastric or posterior gastric veins, and 100% of IGV1 had gastrorenal or splenorenal shunts. Approximately one-half of the total GV cases (56.3%, 18/32) had gastrorenal or splenorenal shunts, and this parameter was correlated with portal vein diameter (t = -2.766, P less than 0.05). The presence of gastrorenal or splenorenal shunts was correlated with both recurrence and rebleeding (P less than 0.05).

CONCLUSIONEVL can effectively control bleeding and prevent rebleeding for GV; although, the best therapeutic efficacy and lowest rate of recurrence was achieved in GOV1 cases. The presence of gastrorenal or splenorenal shunts increases the risk of GV recurrence.