Ampicillin killing curve patterns of Haemophilus influenzae type b isolates by agar dilution plate count method.

The responses of 20 ampicillin-susceptible Haemophilus influenzae type b clinical isolates to the bactericidal action of ampicillin were studied by using a modified agar dilution plate count method. A well-defined paradoxical effect was observed in each of the 24-h killing curve patterns for 19 of the 20 isolates, the remaining isolate showing a less-well-defined but suggestive paradoxical effect after 48 h of ampicillin exposure. For each isolate, the lowest 24-h persister percentage representing maximum killing (paradoxical trough percentage) occurred over a narrow range of concentrations immediately above the MIC, with such paradoxical trough percentages for the 20 isolates ranging from greater than 0.1 to less than 0.001%. Three isolates selected to represent slow, intermediate, and rapid responses were investigated by repetition of 24-h studies and by determination of expanded killing curve patterns. Resultant agar dilution plate count killing curve patterns were found to be reproducible and strain dependent and served to characterize each isolate. The paradoxical effect became more distinct with the prolongation of ampicillin action. Maximum killing was again evident for a narrow range of ampicillin concentrations immediately above the MIC, with persister percentages rising rapidly over the next few ampicillin concentrations to peak at 1 to 2 log10 increments higher than trough percentages. Based on the broad range of responses observed for the 20 isolates, the consistent presence of the paradoxical effect, and the time-dependent nature of bactericidal action, we suggest that the MBC and MBC/MIC ratios are inadequate indices of bactericidal action and that the all-or-none concept of "antimicrobial tolerance" should be abandoned.