Nonsteroidal anti-inflammatory drugs (NSAIDs), such as sulindac, have cancer chemopreventive properties by a mechanism that has been suggested to involve cyclooxygenase inhibition and reduction of prostaglandin (PGE2) levels in the target tissue. To test this hypothesis, we studied the effect of dietary sulindac sulfone (500-2000 ppm), a metabolite of sulindac reported to lack cyclooxygenase inhibitory activity, on tumor formation and PGE2 levels in the azoxymethane model of colon carcinogenesis. Rats treated with sulindac at 400 ppm and piroxicam at 150 ppm were used as positive controls. Rats received two s.c. injections of azoxymethane (15 mg/kg) for 2 weeks and were fed either experimental or control diets until necropsy. After 31 weeks of sulfone treatment, a dose-related increase in sulfone levels in both serum and cecal contents was measured; there was no evidence of metabolic conversion to sulindac or other metabolites. Rats treated with sulfone at 1000 and 2000 ppm, sulindac, and piroxicam had significantly fewer colonic adenomas and carcinomas compared with rats fed control diet as measured by tumor incidence, multiplicity, and tumor burden. Sulfone-treated rats also showed a dose-response relationship for inhibiting all tumor parameters. Colons from rats treated with sulindac or piroxicam contained PGE2 levels that ranged from approximately 16-49% of control levels. PGE2 levels in rats treated with sulfone up to 2000 ppm ranged from 78-118% of control levels. Moreover, the effects of sulindac sulfone on various enzymes responsible for regulating prostaglandin levels were evaluated. No significant inhibitory effects were observed for cyclooxygenase, lipoxygenase, or phospholipase A2. These results suggest that reduction of prostaglandin levels in the target tissue may not be necessary for the chemopreventive properties of sulindac.