Intestinal ischemia/reperfusion injury (I/R) results from reactive oxygen metabolites generated by the xanthine oxidase system and activated neutrophils (PMN). In animal models, removing PMN from initial reperfusate has consistently decreased tissue injury. This experiment was designed to test this potential clinical treatment in human bowel subjected to I/R. The extent of reperfusion injury was assessed by measuring the activity of mucosal alkaline phosphatase (A phi), which is a specific marker of reperfusion injury. Human small intestine (n = 13) obtained at the time of organ harvest for transplantation was perfused for 60 min on an ex vivo perfusion circuit. Reperfusate consisted of autologous blood passed through a leukocyte filter (n = 6) or unfiltered blood (n = 7). Control intestine was sampled at harvest, after transport to the lab on ice (cold ischemia), and after 60 min warm ischemia. Mucosa was homogenized and assayed for A phi activity by cleavage of p-nitrophenyl phosphate. A phi activity (nmole/mg/min) was not decreased after either cold (774 +/- 37) or warm (753 +/- 40) ischemia compared to freshly harvested bowel (770 +/- 51). Both reperfused segments showed a significant decrease in A phi activity compared to controls (P < 0.05); however, reperfusion with leukocyte-filtered blood attenuated the decrease in enzyme activity compared to unfiltered blood (327 +/- 30 vs 506 +/- 25, P < 0.05), constituting an apparent reduction in injury of 35%. The observation that the severity of reperfusion injury was decreased by removal of PMN from the reperfusate demonstrates the efficacy of this strategy in human intestine for the first time.