Mechanism of action of injection therapy for bleeding peptic ulcer.

The effects of intramucosal injection of 1:100,000 adrenaline, 5 per cent ethanolamine and normal saline were determined in experimentally created, acutely bleeding gastric mucosal wounds in rabbits. The mean(s.d.) bleeding rate was decreased from 2.3(0.4) to 0.2(0.02) ml/min by adrenaline (P < 0.01), but increased by 1 ml 5 per cent ethanolamine to 4.0(0.6) ml/min (P < 0.05). Normal saline had no haemostatic effect, suggesting that local tamponade is not important. In separate experiments endoscopic injections of 5 per cent ethanolamine, 1:100,000 adrenaline and normal saline were made in the gastric antrum of rabbits. After 48 h the degree of inflammation was greatest with ethanolamine but, despite tissue necrosis and venous thrombosis, neither endarteritis nor arterial thrombosis occurred. Injections of 5 per cent ethanolamine and 80 per cent ethanol placed next to the ear arteries of rabbits caused local ulceration and necrosis, but endarteritis and arterial thrombosis were again absent.