BACTERICIDAL ACTIVITIES OF TEICOPLANIN, VANCOMYCIN, AND GENTAMICIN ALONE AND IN COMBINATION AGAINST STAPHYLOCOCCUS-AUREUS IN AN IN-VITRO PHARMACODYNAMIC MODEL OF ENDOCARDITIS

We adapted an in vitro pharmacodynamic model of infection to incorpora te simulated endocardial vegetations. The bactericidal activities of t eicoplanin, vancomycin, gentamicin, and various combinations of these drugs were studied against a strain of methicillin-susceptible Staphyl ococcus aureus obtained from a patient being treated for endocarditis at Detroit Receiving Hospital. Bacteria were grown overnight, concentr ated, and added to a mixture of cryoprecipitate (80%) and thrombin (10 %) to achieve approximately 5 x 10(9) CFU/g. Fibrin clots (8 to 10) we re suspended into the model, removed at 24, 48, and 72 h in duplicate, weighed, and homogenized in 1.25% trypsin. Control experiments were c onducted to characterize the growth kinetics. The following antibiotic s were administered to simulate the pharmacokinetics of the drugs in h umans: teicoplanin at 3 and 15 mg/kg of body weight, vancomycin at 15 mg/kg, and gentamicin at 1 mg/kg. Fibrin clot samples used to detect r esistance were plated on antibiotic containing tryptic soy agar plates . For the teicoplanin and vancomycin regimens, protein binding to cryo precipitate, thrombin, and fibrin clot was determined to be 32, 43, an d 50% and 26, 28, and 29%, respectively. In comparison with no treatme nt, vancomycin or teicoplanin at 15 mg/kg or either of these regimens combined with gentamicin significantly reduced bacterial counts (P < 0 .0001). Monotherapy with teicoplanin at 3 mg/kg or gentamicin resulted in no killing activity. Combination treatment with teicoplanin at 3 m g/kg and gentamicin resulted in the killing of approximately 2 log(10) CFU/g by 72 h and the development of resistance to gentamicin. The re sults obtained ,vith the in vitro model of endocarditis are similar to the results reported by several investigators with the rabbit model o f infective endocarditis. This unique infection model is useful for de signing initial drug dosage regimens and may be predictive of drug eff icacy against infective endocarditis.