Purpose
To evaluate the prevalence of extra-gastric direct connection between afferent and efferent veins of gastric varices (GVs) (i.e., EAEDC) and its clinical significance during balloon-occluded retrograde transvenous obliteration (BRTO).

Materials and methods
57 patients who underwent BRTO for GVs obliteration were retrospectively enrolled in this study. Pre-procedural CT images were reviewed for the presence of EAEDC. Patients were categorized into group A (patients with EAEDC) and group B (Patients with no detectable EAEDC). Intraprocedural images were reviewed to see if EAEDCs could be seen and if additional techniques were used to preserve or occlude them. Post-procedural CT images were reviewed for GVs obliteration, portal/splenic vein thrombosis, EAEDC patency, and ascites exacerbation. Post-procedural esophageal varices aggravation was evaluated by upper endoscopy.

Results
39 EAEDCs were identified in CT images of 35 patients (i.e., group A = 61.4%). Among them, only 20 EAEDCs were visualized during BRTO. In the remaining 22 patients, EAEDC was not identified in CT images or during BRTO (i.e., group B = 38.6%). There was no statistically significant difference between group A and B regarding post-BRTO GVs obliteration and portal/splenic vein thrombosis. Use of additional techniques to preserve EAEDC patency had significantly reduced the incidence of ascites and esophageal varices exacerbation (p = 0.036 and 0.028, respectively). In patients with EAEDC diameter ≥ 5 mm, EAEDC preservation or obliteration by coils or balloon had significantly reduced the injected sclerosant volume (p = 0.003).

Conclusion
CT is very useful for EAEDC detection. EAEDC preservation may decrease the incidence of post-BRTO ascites and esophageal varices exacerbation.