The relationship between resveratrol (RES) bioavalability and its effect on tumor growth was investigated. Tissue levels of RES were studied after i.v. and oral administration of trans-resveratrol (t-RES) to rabbits, rats, and mice. Half-life of RES in plasma, after i.v. administration of 20 mg t-RES/kg b.wt., was very short (e.g., 14.4 min in rabbits). The highest concentration of RES in plasma, either after i.v. or oral administration (e.g., 2.6 +/- 1.0 microM in mice 2.5 min after receiving 20 mg t-RES/kg orally), was reached within the first 5 min in all animals studied. Extravascular levels (brain, lung, liver, and kidney) of RES, which paralleled those in plasma, were always < 1 nmol/g fresh tissue. RES measured in plasma or tissues was in the trans form (at least 99%). Hepatocytes metabolized t-RES in a dose-dependent fashion (e.g., 43 nmol of t-RES/g x min in the presence of 20 microM tRES), which means that the liver can remove circulating RES very rapidly. In vitro B16 melanoma (B16M) cell proliferation and generation of reactive oxygen species (ROS) was inhibited by t-RES in a concentration-dependent fashion (100% inhibition of tumor growth was found in the presence of 5 microM t-RES). Addition of 10 microM H(2)O(2) to B16M cells, cultured in the presence of 5 microM t-RES, reactivated cell growth. Oral administration of t-RES (20 mg/kg twice per day; or included in the drinking water at 23 mg/l) did not inhibit growth of B16M inoculated into the footpad of mice (solid growth). However, oral administration of t-RES (as above) decreased hepatic metastatic invasion of B16M cells inoculated intrasplenically. The antimetastatic mechanism involves a t-RES (1 microM)-induced inhibition of vascular adhesion molecule 1 (VCAM-1) expression in the hepatic sinusoidal endothelium (HSE), which consequently decreased in vitro B16M cell adhesion to the endothelium via very late activation antigen 4 (VLA-4).