Editor,—Spahr and colleagues recently published a case series which described the poor response of a haemorrhage from gastric antral vascular ectasia (GAVE) to portal decompression by insertion of a transjugular intrahepatic portosystemic shunt (TIPS) (Gut1999;44:739–742). However, the authors' claim that this indicates the absence of a relation between GAVE and portal hypertension is seriously flawed. The failure of this condition to respond to portal decompression cannot exclude a primary role for portal hypertension in the pathogenesis of the disorder. Furthermore, this study would have been more informative if aortoportography—that is, superior mesenteric and splenic angiography with venous phase imaging, had been performed in order to determine the pattern of portosystemic shunting before and after TIPS insertion. The authors have not addressed the question of whether portosystemic shunts supply the vascular lesions of GAVE; it would have been helpful if …