A new in vitro model for the study of bacterial responses to antibacterial agents at exponentially varying concentrations, i. e. at given elimination half-life values simulating in vivo conditions, is described. The initial concentrations and elimination rates of the agents, composition of bacterial milieu, and commencing density of the bacterial population may be altered. The principal responses of key gram-negative and gram-positive species to bactericidal and bacteriostatic agents and experimental variation are described.

We wanted to simulate the conditions within the body where bacteria are exposed to antibiotics at concentrations which diminish exponentially and thus do not remain constant as in the usual test systems for determination of the minimum inhibitory or bactericidal concentrations (MIC, MBC). For this purpose we employed a model with a glass chamber containing a constant volume of culture which was supplied continuously with medium at a constant rate. We studied the pattern of bacterial response using a series of bactericidal and bacteriostatic agents added to the system at various multiples of the MIC of several bacterial strains, and using different rates of elimination of the antibacterial agent. Strains of Escherichia coli, Pseudomonas aeruginosa, Staphylococcus aureus, and Streptococcus pyogenes were exposed to amoxicillin, ampicillin, benzylpenicillin, carbenicillin, cefazolin, chloramphenicol, flucloxacillin, gentamicin, or oxytetracyclin. The bacteria were killed or maintained at their initial density depending on the antibiotic and its concentration. The main findings were: for most drugs drop in bacterial number persisted while the concentration of the chamber remained above MIC, a period of time elapsed before recovery and bacterial multiplication resumed at the same rate as in control cultures, the period of antibacterial postantibiotic effect lasted longer with Gram-positive than with Gram-negative species, the point in time when the bacteria resumed multiplication corresponded to the IC50. We observed only minor differences in the pattern of bacteria during exposure to potentially bactericidal agents compared to antibiotics with mainly bacteriostatic effects.

The responses of bacteria exposed to amoxycillin and ampicillin were studied at continuously decreasing levels with half-life values similar to those which could occur in vivo. For Escherichia coli, the kill-rates were higher with amoxycillin than with ampicillin. The bactericidal response was exponential. With an antibiotic half-life of one hour, the amoxycillin first order inactivation rate was 3.544 h-1 and the viable cell half-life was 0.196 h; the respective values for ampicillin were 2.341 h-1 and 0.296 h. With an antibiotic half-life of five hours, the inactivation rate was 0.704 h-1 corresponding to a viable cell half-life of 0.985 h for amoxycillin compared to 0.358 h-1 and 1.937 h respectively for ampicillin. Comparison of viable counts and photometric monitoring showed that the former is the preferable method for recording the bacterial response to these beta-lactam antibiotics. During the phase of exponential kill, a plateau occurred in the optical density values. This was due in part to an increased biomass per cell. During the recovery phase, the number of viable cells started to increase several hours sooner than did the rise in optical density. For Staphylococcus aureus,the rates of kill were similar with both agents. Amoxycillin had a long bacteriostatic phase which was not seen with ampicillin. This led to a longer lasting antibacterial effect and reduction to a lower total count with amoxycillin. With staphylococci, the viable counts and the photometric responses were parallel.