Effects of human albumin and serum on the in vitro bactericidal activity of cefditoren against penicillin-resistant Streptococcus pneumoniae.

Attempts to interpret antibiotic pharmacodynamics using reported protein binding may underestimate true activity. To elucidate this issue we examined bacterial killing kinetics at cefditoren concentrations equal to C(max) in the presence of 90% human serum or albumin at physiological concentrations. Killing curves (final inocula of approximately 10(7) cfu/mL, cefditoren concentration of 4.2 mg/L) were performed against Streptococcus pneumoniae strains exhibiting cefditoren MICs (mg/L) of 0.12 (strain 1), 0.25 (strain 2) and 0.5 (strain 3) in different media: (i) C(max)-MH, Mueller-Hinton broth plus 5% lysed horse blood (MH), (ii) C(max)-HS, MH broth with a final human serum concentration of 90%; and (iii) C(max)-HAlb, MH broth with 4 g/dL human albumin. Killing curves were also performed with a final cefditoren concentration of 0.5 mg/L (similar to free-drug C(max) considering 88% protein binding) in MH broth (12% C(max)). No significant differences were found between the different media or concentrations with strain 1 (log(10) reductions >or=4 at 24 h). Against strains 2 and 3, we observed significantly higher initial inocula decreases at 24 h for C(max)-HS as compared with C(max)-HAlb or 12% C(max). Bactericidal activity (>or=3 log(10) reduction) was obtained at 24 h against the three strains only with C(max)-HS and C(max)-MH. The presence of physiological concentrations of human albumin, but not 90% human serum, limited bactericidal activity as did the use of concentrations similar to free-drug C(max), suggesting that extrapolation of active drug from total drug by using the protein binding rate is not an accurate method to study antibacterial activity.