PURPOSEThis study was conducted to develop and validate a dog colon model that predicts colon permeability in humans.METHODSThe following compounds were studied: Class 1 highly soluble (HS)/highly permeable (HP): aminophylline, propranolol, CP-409092; Class 2 LS/HP: nifedipine; trovafloxacin, sertraline; Class 3 HS/LP: azithromycin, atenolol, CP-331684, CP-424391; Class 4 LS/LP: CJ-13610. Administration to dogs was made 30 cm cranial to the anal sphincter with a lubricated Schott Model VFS-5 flexible endoscope. The bioavailability of the compound following the colon administration in dogs, relative to the same formulation administered orally (relative bioavailability), was determined.RESULTSExcept for atenolol, a small hydrophillic molecule, the relative bioavailability from administration to the colon of the dog correlated well with the following compound properties: high solubility and high, passive permeability > high solubility, low permeability > low solubility, high, passive permeability approximately low solubility, low permeability.CONCLUSIONThe dog colon model is proposed as a surrogate for human intubation studies when the controlled release candidate falls in BCS Classes 2 (LS/HP), 3 (HS/LP), and 4 (LS/LP). However, no human intubation or dog colon studies are required for Class 1 (HS/HP), as these compounds are likely to be well absorbed from the colon.