The pathogenesis of nonsteroidal anti-inflammatory drug–induced enteropathy is controversial, but it is thought that cyclooxygenase-1 (COX-1) inhibition is of pivotal importance. We compared small intestinal function and morphology in untreated wild-type, COX-1– and COX-2–deficient mice and the effect of indomethacin, selective COX-1 (SC-560), and COX-2 (celecoxib) inhibition.

Intestinal permeability (⁵¹CrEDTA), inflammation (fecal granulocyte marker protein), prostaglandin E₂ (PGE₂) levels, and macroscopic and microscopic appearances were assessed at baseline and after the drugs.

COX-1^−/− animals were normal except for a 97% decrease in intestinal PGE₂ levels. COX-1^+/+ and COX-1^−/− animals reacted in a similar way to indomethacin. However, celecoxib, having caused no damage in COX-1^+/+ animals, caused small bowel ulcers in COX-1^−/− animals. Selective inhibition of COX-1 decreased intestinal PGE₂ levels in COX-2^+/+ and COX-2^−/− animals by 95%–97%, but caused only small bowel ulcers in the latter group. Dual inhibition of COX-1 and COX-2 in wild-type animals resulted in similar small bowel damage. Between 40% and 50% of untreated COX-2^−/− animals had increased intestinal permeability and inflammation. Some had ileal ulcers that were distinctively different from indomethacin-induced ulcers. Furthermore, long-term celecoxib administration in wild-type animals was associated with similar damage as in the COX-2^−/− mice.

COX-1 deficiency or inhibition and short-term COX-2 inhibition are compatible with normal small intestinal integrity. Dual inhibition of the COX enzymes leads to damage similar to that seen with indomethacin. Long-term COX-2 deficiency or inhibition is associated with significant intestinal pathology despite normal intestinal PGE₂ levels, suggesting a role for COX-2 in the maintenance of small intestinal integrity in the mouse.