ABSTRACT
We used a mouse model of pneumococcal pneumonia to assess the bactericidal effect of increasing doses of amoxicillin (AMX) against clinical strains with various susceptibilities to penicillin. Twelve strains that exhibited similar virulence in mice were selected. Three were penicillin susceptible (PS) (penicillin and AMX MICs = 0.01 to 0.03 microgram/ml), three were intermediately resistant (PIR) (penicillin and AMX MICs = 0.5 to 1 microgram/ml), and six were penicillin resistant (PR) (penicillin and AMX MICs = 1 to 8 micrograms/ml). Leukopenic Swiss mice were infected intratracheally with 10(7) CFU of each strain. Treatment was initiated 3 h after infection and consisted of a single subcutaneous injection of AMX at doses ranging from 2.5 to 10 mg/kg (PS strains), 5 to 100 (PIR strains), and 25 to 3,000 (PR strains). Bacterial killing kinetics were recorded in the lungs over 9 h. The maximal log CFU reduction (Emax) was observed 3 h postinjection. The relation between Emax and log10(dose/MIC) showed two populations. With seven strains (the three PS, the three PIR, and one of the six PR [MICs, penicillin/AMX = 4/1]) a good correlation was observed between Emax and log10(dose/MIC) (r = 0.772; P < 0.02). A bactericidal effect equal to 3.5 log10 CFU was observed at a log10(dose/MIC) = 2. At this ratio, with the five other PR strains, Emax varied from 0.4 to 1.6 log10 CFU. In brain heart infusion medium containing AMX at 50 times the relevant MIC, these five PR strains were tolerant in vitro. Treatment failure with AMX was found in vivo, with tolerant, highly resistant strains.
