Microbiology and ethics statement
This study was performed at the University Hospital Clementino Fraga Filho, a public tertiary teaching hospital in Rio de Janeiro, Brazil, with about 70,000 patients/day per year during the study period. It was approved by Human Research Ethics Committee of the University Hospital Clementino Fraga Filho (number 008/15).
A retrospective analysis was conducted evaluating all episodes of S. aureus BSI occurring in adults submitted to treatment between February 2011 and December 2013. Only the first S. aureus isolate associated with a confirmed BSI episode, according to hospital control policies, was included in the analysis. An episode was defined as an isolation of S. aureus in a blood culture with subsequent initiation of anti-staphylococcal therapy. Only single and consecutive episodes were considered for the present study (i.e. one single patient with first S. aureus isolate submitted to specific treatment).
All blood cultures were routinely processed using BacT/ALERT® (BioMerieux Inc., Durham, NC, USA) during the referred period. Bacterial identification was carried out by the automated VITEK® 2 system (BioMerieux, Durham, NC, USA). Identification of bacteria was confirmed using Gram staining, catalase and coagulase production, and evaluation of 0.04U bacitracin resistance by disk-diffusion (13).
All isolates selected for the current study were submitted to susceptibility to oxacillin was determined by cefoxitin disk (CECON, São Paulo, Brazil) diffusion test according to the CLSI (14). Determination of MICs was performed by broth microdilution, using fresh cation-adjusted Mueller-Hinton broth (CAMHB) for vancomycin, teicoplanin, linezolid, daptomycin (Sigma-Aldrich Chemical Company, St Louis, MO, USA) and ceftaroline (donated by AstraZeneca Pharmaceuticals, Schaumburg, IL, USA) . CAMHB was supplemented with 50 µg/mL calcium for daptomycin assay . Linezolid MIC was determined considering growth inhibition of 100% (linezolid-100) and also according to CLSI M100-S26 document (designated linezolid-80).. The CLSI interpretative breakpoints were used for all antimicrobials (14). The ATCC strains 25923 and 29213 were used as controls for the disk diffusion and MIC tests, respectively. The mecA gene detection was performed as previously described (15) for isolates resistant to cefoxitin by disk diffusion.
Antibiotic regimens
Steady-state exposure was assessed for the following antibiotic regimens by the methodology described below: daptomycin 6 mg/kg q24h; daptomycin 10 mg/kg q24h; linezolid 600 mg q12h; teicoplanin three 400 mg q12h as loading dose, followed by 400 mg q24h; teicoplanin 400 mg q12h; vancomycin 1000 mg q8h; vancomycin 1000 mg q12h; ceftaroline 600 mg q12h.
Pharmacokinetic / pharmacodynamic models
Mean pharmacokinetic parameters and their distributions were extrapolated from published patient studies for each antibiotic. For studies to be considered, they had to be conducted in at least 10 actual patients (defined as a clinical study in the presence of an infection), to have described the assay used to determine drug concentrations and presented mean and standard deviation results for the total body clearance in liters per hour, volume of distribution of the central compartment and other pertinent PK parameters. Mean PK parameters and distribution were extrapolated from selected published studies for each antibiotic (16–19).
The PK/PD parameters (fAUC, total AUC or fT > MIC) were chosen based on pharmacodynamic exposure-response relationship for each agent (19–23). AUC0-24h was calculated by dividing Dose24hours/Clearance and then this value was divided by each MIC dilution between 0.625 mg/L to 16 mg/L to provide the total AUC/MIC (for linezolid and vancomycin) or fAUC/MIC (for daptomycin) calculation. Teicoplanin AUC0–24 was calculated using the trapezoidal rule and divided by each MIC dilution as previously described. Each antimicrobial and their respective PK published studies and PD targets adopted are described in specific subsections (described in more detail below). Table 1 summarizes the pharmacokinetic (PK) parameters derived from published studies, their respective references, and the pharmacodynamic (PD) targets chosen for each antimicrobial used in the Monte Carlo simulation.
Daptomycin
The PK parameters of daptomycin were derived from a study with 58 subjects treated for severe Gram-positive infections (16). Daptomycin PK model considered an 80 kg weight individual, with total body clearance of 0.8 ± 0.14 L/h, protein binding of 90% to 93%, linear and dose-proportional PK over dose range studied (6 mg/kg/dose and 10 mg/kg/dose) (16,24). Daptomycin PD target of fAUC⁄MIC >40 was chosen, previously associated with bacteriostasis in thigh murine infection model (20).
Linezolid
Linezolid PK data was obtained from a study with 318 adults with Gram-positive infections (community-acquired pneumonia and skin and soft tissue infections) treated under the compassionate-use protocol (17). The Linezolid PK model then assumed a total body clearance of 6.85 ± 3.45 L/h (17) and a one-compartment model (11,25). Linezolid PD target was total AUC⁄MIC > 82.9, exposure required for a bacteriostatic response in neutropenic murine thigh infection model (21).
Teicoplanin
Teicoplanin PK were derived from a population study with 30 febrile and severely neutropenic patients. Teicoplanin PK then assumed a total body clearance of 1.15 ± 0.56 L⁄h, volume of the central compartment (6.56 ± 4.01 L), k12 (1.29 ± 0.62 h)-1) and k21 (0.18 ± 0.08 h)-1), and a two-compartment model until steady state, to account for its long half-life (11,18). Teicoplanin PD target was a total AUC⁄ MIC ≥ 900, exposure correlated with bacteriological response in patients with documented MRSA infection (22).
Vancomycin
Vancomycin PK data were derived from a populational study of patients receiving treatment for S. aureus lower respiratory tract infection (19). Vancomycin total body clearance was estimated as a function of creatinine clearance (CrCl, mL/min); drug clearance (L / h) = [(CrCl · 0.79) + 15.4] · 0.06 (19). CrCl was assumed to follow a triangular distribution, simulated as a range between 50 mL⁄min and 120 mL⁄min (11). Vancomycin PD target was total AUC⁄ MIC ≥ 350, which was the exposure associated with clinical success for lower respiratory tract infections, approximately corresponding to a trough vancomycin concentration of 15–20 mg/mL (19).
Ceftaroline
Ceftaroline probability of target attainment (PTA) was obtained from literature (23,26). The model was derived from a three-compartment model developed from plasma concentration from Phase 1, 2, 3 studies, the two latter in patients with complicated skin infection and community-acquired pneumonia (27). Ceftaroline PD target used was 51% fT>MIC, which is an exposure associated with 2-log10 CFU reduction from baseline for S. aureus on murine thigh and lung infection models (23).
Monte Carlo simulation
A 5000-patient Monte Carlo simulation (Crystal Ball 2000; Decisioneering Inc., Denver, CO, USA) was performed to calculate a population of total AUC/MIC (free or total) or T>MIC exposures for each antibiotic regimen at each MIC dilution. Clearance, volume of the central compartment, k12 and k21 were each assumed to follow log-Gaussian distributions during simulations for teicoplanin. For the vancomycin simulation, creatinine clearance was assumed to follow a triangular distribution as previously described (11). The number of simulated patients achieving the target pharmacodynamic exposure at each MIC was counted and reported as the PTA at that specific MIC (values in percentages). The cumulative fractions of response (CFR) was calculated as previously described for each drug (28), multiplying the PTA at each MIC by the percentage of isolates with that specific MIC. Final CFR results were obtained as the sum of each PTA per MIC and a CFR ≥ 90% was considered optimal.
Sensitivity analysis
A sensitivity analysis was conducted to explore the robustness of the CFR against entire BSI S. aureus isolates. In order to perform the analysis, different PD targets were applied as a way of comparison with targets originally applied, as follows. For vancomycin sensitivity analysis, an alternative PD target of AUC⁄MIC > 400 was used, which was the exposure required for bacterial eradication in lower respiratory tract infections (19). For daptomycin, PD targets of fAUC/MIC >12 (the minimum value providing static effect with MRSA) and >171 (the minimum ratio which provided 99% kill) were obtained from a thigh murine infection model (20). For linezolid, the alternative PD targets of total AUC⁄ MICs >51.85 (minimum breakpoint associated with clinical cure) and >128 (median AUC/MIC associated to bacterial eradication in the blood of adult patients enrolled in the compassionate use program of linezolid) were also explored (29). For ceftaroline, fT > MIC targets of 26% and 36% were also analyzed, which are targets associated to bacterial reduction endpoints of net bacterial stasis and 1-log10 CFU reductions from baseline for S. aureus based on murine infection models, respectively (23). Finally, teicoplanin trough value of > 13 mg⁄L and > 20 mg⁄L were investigated, they are currently used in clinical practice (30,31).