Using machine learning to optimize antibiotic combinations: dosing strategies for meropenem and polymyxin B against carbapenem-resistant Acinetobacter baumannii.

OBJECTIVES: Increased rates of carbapenem-resistant strains of Acinetobacter baumannii have forced clinicians to rely upon last-line agents, such as the polymyxins, or empirical, unoptimized combination therapy. Therefore, the objectives of this study were: (a) to evaluate the in vitro pharmacodynamics of meropenem and polymyxin B (PMB) combinations against A. baumannii; (b) to utilize a mechanism-based mathematical model to quantify bacterial killing; and (c) to develop a genetic algorithm (GA) to define optimal dosing strategies for meropenem and PMB. METHODS: A. baumannii (N16870; MICmeropenem = 16 mg/L, MICPMB = 0.5 mg/L) was studied in the hollow-fibre infection model (initial inoculum 108 cfu/mL) over 14 days against meropenem and PMB combinations. A mechanism-based model of the data and population pharmacokinetics of each drug were used to develop a GA to define the optimal regimen parameters. RESULTS: Monotherapies resulted in regrowth to ~1010 cfu/mL by 24 h, while combination regimens employing high-intensity PMB exposure achieved complete bacterial eradication (0 cfu/mL) by 336 h. The mechanism-based model demonstrated an SC50 (PMB concentration for 50% of maximum synergy on meropenem killing) of 0.0927 mg/L for PMB-susceptible subpopulations versus 3.40 mg/L for PMB-resistant subpopulations. The GA had a preference for meropenem regimens that improved the %T > MIC via longer infusion times and shorter dosing intervals. The GA predicted that treating 90% of simulated subjects harbouring a 108 cfu/mL starting inoculum to a point of 100 cfu/mL would require a regimen of meropenem 19.6 g/day 2 h prolonged infusion (2 hPI) q5h + PMB 5.17 mg/kg/day 2 hPI q6h (where the 0 h meropenem and PMB doses should be 'loaded' with 80.5% and 42.2% of the daily dose, respectively). CONCLUSION: This study provides a methodology leveraging in vitro experimental data, a mathematical pharmacodynamic model, and population pharmacokinetics provide a possible avenue to optimize treatment regimens beyond the use of the 'traditional' indices of antibiotic action.

[Pharmacokinetics and pharmacodynamics of antibiotics in intensive care]. / Pharmakokinetik und Pharmakodynamik von Antibiotika in der Intensivmedizin.

Optimized dosage regimens of antibiotics have remained obscure since their introduction. During the last two decades pharmacokinetic(PK)-pharmacodynamic(PD) relationships, originally established in animal experiments, have been increasingly used in patients. The action of betalactams is believed to be governed by the time the plasma concentration is above the minimum inhibitory concentration (MIC). Aminoglycosides act as planned when the peak concentration is a multiple of the MIC and vancomycin seems to work best when the area under the plasma vs. time curve (AUC) to MIC has a certain ratio. Clinicians should be aware that these relationships can only be an indication in which direction dosing should go. Larger studies with sufficiently high numbers of patients and particularly severely sick patients are needed to prove the concepts. In times where all antibiotics can be measured with new technologies, the introduction of therapeutic drug monitoring (TDM) is suggested for ICUs (Intensive Care Unit). The idea of a central lab for TDM of antibiotics such as PEAK (Paul Ehrlich Antibiotika Konzentrationsmessung) is supported.

Evaluation of enrofloxacin use in koalas (Phascolarctos cinereus) via population pharmacokinetics and Monte Carlo simulation.

Clinically normal koalas (n = 6) received a single dose of intravenous enrofloxacin (10 mg/kg). Serial plasma samples were collected over 24 h, and enrofloxacin concentrations were determined via high-performance liquid chromatography. Population pharmacokinetic modeling was performed in S-ADAPT. The probability of target attainment (PTA) was predicted via Monte Carlo simulations (MCS) using relevant target values (30-300) based on the unbound area under the curve over 24 h divided by the minimum inhibitory concentration (MIC) (fAUC0-24 /MIC), and published subcutaneous data were incorporated (Griffith et al., 2010). A two-compartment disposition model with allometrically scaled clearances (exponent: 0.75) and volumes of distribution (exponent: 1.0) adequately described the disposition of enrofloxacin. For 5.4 kg koalas (average weight), point estimates for total clearance (SE%) were 2.58 L/h (15%), central volume of distribution 0.249 L (14%), and peripheral volume 2.77 L (20%). MCS using a target fAUC0-24 /MIC of 40 predicted highest treatable MICs of 0.0625 mg/L for intravenous dosing and 0.0313 mg/L for subcutaneous dosing of 10 mg/kg enrofloxacin every 24 h. Thus, the frequently used dosage of 10 mg/kg enrofloxacin every 24 h subcutaneously may be appropriate against gram-positive bacteria with MICs ≤ 0.03 mg/L (PTA > 90%), but appears inadequate against gram-negative bacteria and Chlamydiae in koalas.

Comparable population pharmacokinetics and pharmacodynamic breakpoints of cefpirome in cystic fibrosis patients and healthy volunteers.

Cystic fibrosis (CF) patients are often reported to have higher clearances and larger volumes of distribution per kilogram of total body weight (WT) for beta-lactams than healthy volunteers. As pharmacokinetic (PK) data on cefpirome from studies of CF patients are lacking, we systematically compared its population PK and pharmacodynamic breakpoints for CF patients and healthy volunteers of similar body size. Twelve adult CF patients (median lean body mass [LBM] = 45.7 kg) and 12 healthy volunteers (LBM = 50.0 kg) received a single 10-min intravenous infusion of 2 g cefpirome. Plasma and urine concentrations were determined by high-performance liquid chromatography (HPLC). Population PK and Monte Carlo simulations were performed using NONMEM and S-ADAPT and a duration of an unbound plasma concentration above the MIC ≥ 65% of the dosing interval as a pharmacodynamic target. Unscaled clearances for CF patients were similar to those seen with healthy volunteers, and the volume of distribution was 6% lower for CF patients. Linear scaling of total clearance by WT resulted in clearance that was 20% higher (P ≤ 0.001 [nonparametric bootstrap]) in CF patients. Allometric scaling by LBM explained the differences between the two subject groups with respect to average clearance and volume of distribution and reduced the unexplained between-subject variability of renal and nonrenal clearance by 10 to 14%. For the CF patients, robust (>90%) probabilities of target attainment (PTA) were achieved by the administration of a standard dose of 2 g/70 kg WT every 12 h (Q12h) given as 30-min infusions for MICs ≤ 1.5 mg/liter. As alternative dosage regimens, a 5-h infusion of 1.33 g/70 kg WT Q8h achieved robust PTAs for MICs ≤ 8 to 12 mg/liter and a continuous infusion of 4 g/day for MICs ≤ 12 mg/liter. Prolonged infusion of cefpirome is expected to be superior to short-term infusions for MICs between 2 and 12 mg/liter.

Population pharmacokinetic comparison and pharmacodynamic breakpoints of ceftazidime in cystic fibrosis patients and healthy volunteers.

Despite the promising activity of ceftazidime against Pseudomonas aeruginosa and Burkholderia cepacia, there has not yet been a study that directly compared the pharmacokinetics (PK) of ceftazidime in cystic fibrosis (CF) patients and healthy volunteers by population PK methodology. We assessed the population PK and PK/pharmacodynamic (PD) breakpoints of ceftazidime in CF patients and healthy volunteers. Eight CF patients (total body weight [WT] [average +/- standard deviation] = 42.9 +/- 18.4 kg) and seven healthy volunteers (WT = 66.2 +/- 4.9 kg) received 2 g ceftazidime as a 5-min intravenous infusion. High-performance liquid chromatography (HPLC) was used for drug analysis, and NONMEM (results reported), S-ADAPT, and NPAG were used for parametric and nonparametric population PK modeling. We considered linear and allometric body size models to scale clearance and volume of distribution. Monte Carlo simulations were based on a target time of non-protein-bound plasma concentration of ceftazidime above MIC of > or =65%, which represents near-maximal killing. Unscaled total clearance was 19% lower in CF patients, and volume of distribution was 36% lower. Total clearance was 7.82 liters/h for CF patients and 6.68 liters/h for healthy volunteers with 53 kg fat-free mass. Allometric scaling by fat-free mass reduced the between-subject variability by 32% for clearance and by 18 to 26% for volume of both peripheral compartments compared to linear scaling by WT. A 30-min ceftazidime infusion of 2 g/70 kg WT every 8 h (q8h) achieved robust (> or =90%) probabilities of target attainment (PTAs) for MICs of < or =1 mg/liter in CF patients and < or =3 mg/liter in healthy volunteers. Alternative modes of administration achieved robust PTAs up to markedly higher MICs of < or =8 to 12 mg/liter in CF patients for 5-h infusions of 2 g/70 kg WT q8h and < or =12 mg/liter for continuous infusion of 6 g/70 kg WT daily.

New semiphysiological absorption model to assess the pharmacodynamic profile of cefuroxime axetil using nonparametric and parametric population pharmacokinetics.

Cefuroxime axetil is widely used to treat respiratory tract infections. We are not aware of a population pharmacokinetic (PK) model for cefuroxime axetil. Our objectives were to develop a semiphysiological population PK model and evaluate the pharmacodynamic profile for cefuroxime axetil. Twenty-four healthy volunteers received 250 mg oral cefuroxime as a suspension after a standardized breakfast. Liquid chromatography-tandem mass spectrometry was used for drug analysis, NONMEM and S-ADAPT (results reported) were used for parametric population PK modeling, and NPAG was used for nonparametric population PK modeling. Monte Carlo simulations were used to predict the duration for which the non-protein-bound-plasma concentration was above the MIC (fT(>MIC)). A model with one disposition compartment, a saturable and time-dependent drug release from the stomach, and fast drug absorption from the intestine yielded precise (r > 0.992) and unbiased curve fits and an excellent predictive performance. The apparent clearance was 21.7 liters/h (19.8% coefficient of variation [CV]) and the volume of distribution 38.7 liters (18.3% CV). Robust (>or=90%) probabilities of target attainment (PTAs) were achieved by 250 mg cefuroxime given every 12 h (q12h) or q8h for MICs of MIC) of >or=40% and for MICs of MIC) of >or=65%. For the >or=40% fT(>MIC) target, the PTAs for 250 mg cefuroxime q12h were >or=97.8% for Streptococcus pyogenes and penicillin-susceptible Streptococcus pneumoniae. Cefuroxime at 250 mg q12h or q8h achieved PTAs below 73% or 92%, respectively, for Haemophilus influenzae, Moraxella catarrhalis, and penicillin-intermediate S. pneumoniae for susceptibility data from various countries. Depending on the MIC distribution, 250 mg oral cefuroxime q8h instead of q12h should be considered, especially for more-severe infections that require near-maximal killing by cefuroxime.