Objectives: Vancomycin is a vital treatment option for patients suffering from critical infections, and therapeutic drug monitoring is recommended. Bayesian forecasting is reported to improve trough concentration monitoring for dose adjustment. However, the predictive performance of pharmacokinetic models that are utilized for Bayesian forecasting has not been systematically evaluated. Method: Thirty-one published population pharmacokinetic models for vancomycin were encoded in NONMEM ® 7.4. Data from 292 hospitalized patients were used to evaluate the predictive performance (forecasting bias and precision, visual predictive checks) of the models to forecast vancomycin concentrations and area under the curve (AUC) by (a) a priori prediction, i.e., solely by patient characteristics, and (b) also including measured vancomycin concentrations from previous dosing occasions using Bayesian forecasting. Results: A priori prediction varied substantiallydrelative bias (rBias): e122.7e67.96%, relative root mean squared error (rRMSE) 44.3e136.8%, respectivelydand was best for models which included body weight and creatinine clearance as covariates. The model by Goti et al. displayed the best predictive performance with an rBias of e4.41% and an rRMSE of 44.3%, as well as the most accurate visual predictive checks and AUC predictions. Models with less accurate predictive performance provided distorted AUC predictions which may lead to inappropriate dosing decisions. Conclusion: There is a diverse landscape of population pharmacokinetic models for vancomycin with varied predictive performance in Bayesian forecasting. Our study revealed the Goti model as suitable for improving precision dosing in hospitalized patients. Therefore, it should be used to drive vancomycin dosing decisions, and studies to link this finding to clinical outcomes are warranted.

BackgroundTigecycline is a vital antibiotic treatment option for infections caused by multiresistant bacteria in the intensive care unit (ICU). Acute kidney injury (AKI) is a common complication in the ICU requiring continuous renal replacement therapy (CRRT), but pharmacokinetic data for tigecycline in patients receiving CRRT are lacking.MethodsEleven patients mainly with intra-abdominal infections receiving either continuous veno-venous hemodialysis (CVVHD, n = 8) or hemodiafiltration (CVVHDF, n = 3) were enrolled, and plasma as well as effluent samples were collected according to a rich sampling schedule. Total and free tigecycline was determined by ultrafiltration and high-performance liquid chromatography (HPLC)-UV. Population pharmacokinetic modeling using NONMEM® 7.4 was used to determine the pharmacokinetic parameters as well as the clearance of CVVHD and CVVHDF. Pharmacokinetic/pharmacodynamic target attainment analyses were performed to explore the potential need for dose adjustments of tigecycline in CRRT.ResultsA two-compartment population pharmacokinetic (PK) model was suitable to simultaneously describe the plasma PK and effluent measurements of tigecycline. Tigecycline dialysability was high, as indicated by the high mean saturation coefficients of 0.79 and 0.90 for CVVHD and CVVHDF, respectively, and in range of the concentration-dependent unbound fraction of tigecycline (45–94%). However, the contribution of CRRT to tigecycline clearance (CL) was only moderate (CLCVVHD: 1.69 L/h, CLCVVHDF: 2.71 L/h) in comparison with CLbody (physiological part of the total clearance) of 18.3 L/h. Bilirubin was identified as a covariate on CLbody in our collective, reducing the observed interindividual variability on CLbody from 58.6% to 43.6%. The probability of target attainment under CRRT for abdominal infections was ≥ 0.88 for minimal inhibitory concentration (MIC) values ≤ 0.5 mg/L and similar to patients without AKI.ConclusionsDespite high dialysability, dialysis clearance displayed only a minor contribution to tigecycline elimination, being in the range of renal elimination in patients without AKI. No dose adjustment of tigecycline seems necessary in CRRT.Trial registrationEudraCT, 2012–005617-39. Registered on 7 August 2013.Electronic supplementary materialThe online version of this article (10.1186/s13054-018-2278-4) contains supplementary material, which is available to authorized users.

Background: Routine clinical TDM data is often used to develop population pharmacokinetic (PK) models, which are applied in turn for model-informed precision dosing. The impact of uncertainty in documented sampling and infusion times in population PK modeling and model-informed precision dosing have not yet been systematically evaluated. The aim of this study was to investigate uncertain documentation of (i) sampling times and (ii) infusion rate exemplified with two antiinfectives. Methods: A stochastic simulation and estimation study was performed in NONMEM ® using previously published population PK models of meropenem and caspofungin. Uncertainties, i.e. deviation between accurate and planned sampling and infusion times (standard deviation (SD) ± 5 min to ± 30 min) were added randomly in R before carrying out the simulation step. The estimation step was then performed with the accurate or planned times (replacing real time points by scheduled study values). Relative bias (rBias) and root mean squared error (rRMSE) were calculated to determine accuracy and precision of the primary and secondary PK parameters on the population and individual level. The accurate and the misspecified (using planned sampling times) model were used for Bayesian forecasting of meropenem to assess the impact on PK/PD target calculations relevant to dosing decisions. Results: On the population level, the estimates of the proportional residual error (prop.err.) and the interindividual variability (IIV) on the central volume of distribution (V1) were most affected by erroneous records in the sampling and infusion time (e.g. rBias of prop.err.: 75.5% vs. 183% (meropenem) and 10.1% vs. 109% (caspofungin) for ± 5 vs. ± 30 min, respectively). On the individual level, the rBias of the planned scenario for the typical values V1, Q and V2 increased with increasing uncertainty in time, while CL, AUC and elimination half-life were least affected. Meropenem as a short half-life drug (~1 h) was

Assessing parameter uncertainty is a crucial step in pharmacometric workflows. Small datasets with ten or fewer subjects appear regularly in drug development and therapeutic use, but it is unclear which method to assess parameter uncertainty is preferable in such situations. The aim of this study was to (i) systematically evaluate the performance of standard error (SE), bootstrap (BS), log-likelihood profiling (LLP), Bayesian approaches (BAY) and sampling importance resampling (SIR) to assess parameter uncertainty in small datasets and (ii) to evaluate methods to provide proposal distributions for the SIR. A simulation study was conducted and the 0-95% confidence interval (CI) and coverage for each parameter was evaluated and compared to reference CIs derived by stochastic simulation and estimation (SSE). A newly proposed LLP-SIR, combining the proposal distribution provided by LLP with SIR, was included in addition to conventional SE-SIR and BS-SIR. Additionally, the methods were applied to a clinical dataset. The determined CIs differed substantially across the methods. The CIs of SE, BS, LLP and BAY were not in line with the reference in datasets with B 10 subjects. The best alignment was found for the LLP-SIR, which also provided the best coverage results among the SIR methods. The best overall results regarding the coverage were provided by LLP and BAY across all parameters and dataset sizes. To conclude, the popular SE and BS methods are not suitable to derive parameter uncertainty in small datasets containing B 10 subjects, while best performances were observed with LLP, BAY and LLP-SIR.

Objectives: Pharmacokineticepharmacodynamic (PK-PD) considerations are at the heart of defining susceptibility breakpoints for antibiotic therapy. However, current approaches follow a fragmented workflow. The aim of this study was to develop an integrative pharmacometric approach to define MICbased breakpoints for killing and suppression of resistance development for plasma and tissue sites, integrating clinical microdialysis data as well as in vitro timeekill curves and heteroresistance information, exemplified by moxifloxacin against Staphylococcus aureus and Escherichia coli. Methods: Plasma and target site samples were collected from ten patients receiving 400 mg moxifloxacin/day. In vitro timeekill studies with three S. aureus and two E. coli strains were performed and resistant subpopulations were quantified. Using these data, a hybrid physiologically based (PB) PK model and a PK-PD model were developed, and utilized to predict site-specific breakpoints. Results: For both bacterial species, the predicted MIC breakpoint for stasis at 400 mg/day was 0.25 mg/L. Less reliable killing was predicted for E. coli in subcutaneous tissues where the breakpoint was 0.125 mg/ L. The breakpoint for resistance suppression was 0.06 mg/L. Notably, amplification of resistant subpopulations was highest at the clinical breakpoint of 0.25 mg/L. High-dose moxifloxacin (800 mg/day) increased all breakpoints by one MIC tier. Conclusions: An efficient pharmacometric approach to define susceptibility breakpoints was developed; this has the potential to streamline the process of breakpoint determination. Thereby, the approach provided additional insight into target site PK-PD and resistance development for moxifloxacin. Application of the approach to further drugs is warranted.

Purpose Clearance via renal replacement therapy (RRT) can significantly alter the pharmacokinetic profile of drugs. The aim of this study was (i) to improve the use of clinical trial data and (ii) to provide a model that allows quantification of all aspects of drug elimination via RRT including adsorption to dialysis membranes and/or degradation of the drug in the dialysate. Methods An integrated dialysis pharmacometric (IDP) model was developed to simultaneously incorporate all available RRT information. The sensitivity, accuracy and precision of the IDP model was compared to conventional approaches in clinical trial simulations and applied to clinical datasets of teicoplanin and doripenem. Results The IDP model was more accurate, precise and sensitive than conventional plasma-concentration-based approaches when estimating the clearance RRT (relative bias <1%). In contrast to conventional approaches, adsorption and degradation were quantifiable using the IDP model (relative bias: −1.1% and − 1.9%, respectively). Applied to clinical data, clearance RRT , drug degradation (effluent-half-life doripenem : 13.5 h −1 ) and adsorption (polysulphone adsorption capacity teicoplanin : 31.2 mg) were assessed. Conclusion The IDP model allows accurate, precise and sensitive characterization of clearance RRT , adsorption and degradation. Successful quantification of all aspects of clearance RRT in clinical data demonstrated the benefit of the IDP model as compared to conventional approaches. KEY WORDS adsorption . doripenem . pharmacokinetics . renal replacement therapy . teicoplanin ABBREVIATIONS CVVHD Continuous veno-venous hemodialysis CVVH Continuous veno-venous hemofiltration CVVHDF continuous veno-venous hemodiafiltration dOFV Drop in objective function value Eq. Equation IDP model Integrated dialysis pharmacometric model LLP-SIR log-likelihood-profiling based samplingimportance-resampling rBias relative Bias Astrid Broeker and Matthias G. Vossen contributed equally to this work.

This study investigated tigecycline exposure in critically ill patients from a population pharmacokinetic perspective to support rational dosing in intensive care unit (ICU) patients with acute and chronic liver impairment. A clinical dataset of 39 patients served as the basis for the development of a population pharmacokinetic model. The typical tigecycline clearance was strongly reduced (8.6 L/h) as compared to other populations. Different models were developed based on liver and kidney function-related covariates. Monte Carlo simulations were used to guide dose adjustments with the most predictive covariates: Child–Pugh score, total bilirubin, and MELD score. The best performing covariate, guiding a dose reduction to 25 mg q12h, was Child–Pugh score C, whereas patients with Child–Pugh score A/B received the standard dose of 50 mg q12h. Of note, the obtained 24 h steady-state area under the concentration vs. time curve (AUCss) range using this dosing strategy was predicted to be equivalent to high-dose tigecycline exposure (100 mg q12h) in non-ICU patients. In addition, 26/39 study participants died, and therapy failure was most correlated with chronic liver disease and renal failure, but no correlation between drug exposure and survival was observed. However, tigecycline in special patient populations needs further investigations to enhance clinical outcome.

Background Inaccurate documentation of sampling and infusion times is a potential source of error in personalizing busulfan doses using therapeutic drug monitoring (TDM). Planned times rather than the actual times for sampling and infusion time are often documented. Therefore, this study aimed to evaluate the robustness of a limited sampling TDM of busulfan with regard to inaccurate documentation. Methods A pharmacometric analysis was conducted in NONMEM® 7.4.3 and “R” by performing stochastic simulation and estimation with four, two and one sample(s) per patient on the basis of a one-compartment- (1CMT) and two-compartment (2CMT) population pharmacokinetic model. The dosing regimens consisted of i.v. busulfan (0.8 mg/kg) every 6 h (Q6H) or 3.2 mg/kg every 24 h (Q24H) with a 2 h- and 3 h infusion time, respectively. The relative prediction error (rPE) and relative root-mean-square error (rRmse) were calculated in order to determine the accuracy and precision of the individual AUC estimation. Results A noticeable impact on the estimated AUC based on a 1CMT-model was only observed if uncertain documentation reached ± 30 min (1.60% for Q24H and 2.19% for Q6H). Calculated rPEs and rRmse for Q6H indicate a slightly lower level of accuracy and precision when compared to Q24H. Spread of rPE’s and rRmse for the 2CMT-model were wider and higher compared to estimations based on a 1CMT-model. Conclusions The estimated AUC was not affected substantially by inaccurate documentation of sampling and infusion time. The calculated rPEs and rRmses of estimated AUC indicate robustness and reliability for TDM of busulfan, even in presence of erroneous records.