To explore if the time inside the mutant selection window (T) is a reliable predictor of emergence of bacterial resistance to linezolid, mixed inocula of each of three methicillin-resistant Staphylococcus aureus strains (MIC of linezolid 2 μg ml) and their previously selected resistant mutants (MIC 8 μg ml) were exposed to linezolid pharmacokinetics using an in vitro dynamic model. In five-day treatments simulated over a wide range of the 24-h area under the concentration-time curve (AUC) to the MIC ratio, mutants resistant to 4 × MIC of antibiotic were enriched in a T-dependent manner. With each strain, T relationships with the area under the bacterial mutant concentration-time curve (AUBC) exhibited a hysteresis loop, with the upper portion corresponding to the time above the mutant prevention concentration (MPC; T) of 0 and the lower portion-to the T > 0. Using AUBC related to the maximal value observed with a given strain (normalized AUBC) at T > 0, a strain-independent sigmoid relationship was established between AUBC and T, as well as T (r 0.99 for both). AUC/MIC and AUC/MPC relationships with normalized AUBC for combined data on the three studied S. aureus strains were bell-shaped (r 0.85 and 0.80, respectively). These findings suggest that T at T > 0, T, AUC/MIC and AUC/MPC are useful bacterial strain-independent predictors of the emergence of staphylococcal resistance to linezolid.

Bacterial resistance studies using in vitro dynamic models are highly dependent on the starting inoculum that might or might not contain spontaneously resistant mutants (RMs). To delineate concentration-resistance relationships with linezolid-exposed Staphylococcus aureus, a mixed inoculum containing both susceptible cells and RMs was used. An RM selected after the 9th passage of the parent strain (MIC, 2 g/ml) on antibiotic-containing media (RM9; MIC, 8 g/ml) was chosen for the pharmacodynamic studies, because the mutant prevention concentration (MPC) of linezolid against the parent strain in the presence of RM9 at 10 2 (but not at 10 4 ) CFU/ml did not differ from the MPC value determined in the absence of the RMs. Five-day treatments with twice-daily linezolid doses were simulated at concentrations either between the MIC and MPC or above the MPC. A specific mechanism of oxazolidinone action blocks functional initiation complexes in bacterial translation systems. Given the lack of analogues among available antimicrobials, a low probability of preexisting, naturally occurring resistance mechanisms has been hypothesized (1, 2). For example, resistance of Staphylococcus aureus rarely occurred through spontaneous mutations at frequencies of only 10 Ϫ9 to 10 Ϫ11 (3-5). However, the first report on the detection of a clinical isolate of S. aureus with a linezolid MIC of Ͼ32 g/ml was published as early as 2001 (6). Analysis of 23S rRNA encoding DNA sequences showed that linezolid-resistant S. aureus had a G-to-T mutation at position 2576 (Escherichia coli numbering). The same point mutations also were observed in S. aureus isolated in other clinical studies (7-14). Later, amino acid substitutions in ribosomal protein L3 (50S large-subunit ribosomal protein) were associated with oxazolidinone resistance (linezolid MIC, 8 g/ml) in a clinical S. aureus strain (15). Thus, clinical data indicate that resistance to linezolid mediated by mutations in ribosomal genes may emerge more readily than was initially predicted by routine in vitro studies performed early in the development of oxazolidinones, because they were not specifically designed to simulate antibiotic exposures that would allow the enrichment of resistant mutants.The resistance studies with linezolid cited above were not supported by concomitant pharmacokinetic studies to relate antibiotic concentrations to the selection of resistant mutants. To be sure, some in vitro model studies of the enrichment of resistant S. aureus simulated linezolid pharmacokinetics (16, 17).However, these attempts were unsuccessful because resistant mutants were not enriched, at least in simulations of oscillating antibiotic concentrations that mimic the usual linezolid dosing in humans. Linezolid-resistant staphylococci also were not enriched in other in vitro studies that were not designed to establish concentration-resistance relationships (18)(19)(20), probably because of the lack of spontaneous mutants in the starting inocula.To more clearly delineate concentration-resistance rel...

To explore if combinations of linezolid (L) with rifampicin (R) are able to restrict Staphylococcus aureus resistance, the enrichment of L- and R-resistant mutants was studied in an in vitro dynamic model. L- and R-resistant mutants were enriched in all single drug treatments. In contrast, L-resistant mutants were not enriched and R-resistant mutants were similar to baseline amounts with only minimal regrowth at the end of the combination treatments. These effects appear to be explained by lowering the mutant prevention concentration (MPC) for L+R combinations (MPC) compared to the MPCs of L and R alone (MPC and MPC) and thereby the longer times above MPC (73-100% of the dosing interval for L and 42-58% for R) compared to the times above MPC (0-44%) and MPC (0%). These findings provide an opportunity to predict the selection of S. aureus resistance in L+R treatments using MPCs.

The bell-shaped pattern of AUC24/MIC and AUC24/MPC relationships with S. aureus resistance to linezolid is consistent with the MSW hypothesis. 'Antimutant' AUC24/MIC ratios were predicted based on the AUC24/MIC relationship with AUBCM.

No abstract