Assessing parameter uncertainty in small-n pharmacometric analyses: value of the log-likelihood profiling-based sampling importance resampling (LLP-SIR) technique

Effect of albumin substitution on pharmacokinetics of piperacillin/tazobactam in patients with severe burn injury admitted to the ICU

BACKGROUND

Pathophysiological changes in severely burned patients alter the pharmacokinetics (PK) of anti-infective agents, potentially leading to subtherapeutic concentrations at the target site. Albumin supplementation, to support fluid resuscitation, may affect pharmacokinetic properties by binding drugs. This study aimed to investigate the PK of piperacillin/tazobactam in burn patients admitted to the ICU before and after albumin substitution as total and unbound concentrations in plasma.

PATIENTS AND METHODS

Patients admitted to the ICU and scheduled for 4.5 g piperacillin/tazobactam administration and 200 mL of 20% albumin substitution as part of clinical routine were included. Patients underwent IV microdialysis, and simultaneous arterial plasma sampling, at baseline and multiple timepoints after drug administration. PK analysis of total and unbound drug concentrations under steady-state conditions was performed before and after albumin supplementation.

RESULTS

A total of seven patients with second- to third-degree burns involving 20%-60% of the total body surface were enrolled. Mean (SD) AUC0-8 (h·mg/L) of total piperacillin/tazobactam before and after albumin substitution were 402.1 (242)/53.2 (27) and 521.8 (363)/59.7 (32), respectively. Unbound mean AUC0-8 before and after albumin supplementation were 398.9 (204)/54.5 (25) and 456.4 (439)/64.5 (82), respectively.

CONCLUSIONS

Albumin supplementation had little impact on the PK of piperacillin/tazobactam. After albumin supplementation, there was a numerical increase in mean AUC0-8 of total and unbound piperacillin/tazobactam, whereas similar Cmax values were observed. Future studies may investigate the effect of albumin supplementation on drugs with a higher plasma protein binding.

Bayesian estimation in NONMEM

Bayesian estimation is a powerful but underutilized tool for answering drug development questions. In this tutorial, the principles of Bayesian model development, assessment, and prior selection will be outlined. An example pharmacokinetic (PK) model will be used to demonstrate the implementation of Bayesian modeling using the nonlinear mixed-effects modeling software NONMEM.

Impact of Key Components of Intensified Ceftaroline Dosing on Pharmacokinetic/Pharmacodynamic Target Attainment

BACKGROUND AND OBJECTIVE

Ceftaroline fosamil is a β-lactam antibiotic approved as a 600 mg twice daily dose (≤1 h infusion, 'standard dosing') or a 600 mg thrice daily dose (2 h infusion) to treat complicated skin and soft tissue infections caused by Staphylococcus aureus (minimum inhibitory concentration [MIC] 2-4 mg/L). We sought to systematically evaluate the relative impact of the three key components of the intensified dosing regimen (i.e. shortened dosing interval, prolonged infusion duration and increased total daily dose [TDD]) on the pharmacokinetic/pharmacodynamic (PK/PD) target attainment given different grades of bacterial susceptibility.

METHODS

A population PK model was developed using data from 12 healthy volunteers (EudraCT-2012-005134-11) receiving standard or intensified dosing. PK/PD target attainment (ƒT>MIC = 35% and 100%) after 24 h was compared following systematically varied combinations of the (1) dosing interval (every 12 h [q12h]→ every 8 h [q8h]); (2) infusion duration (1 h→2 h); and (3) individual and total daily dose (400→900 mg, i.e. TDD 1200→1800 mg), as well as for varying susceptibility of S. aureus (MIC 0.032-8 mg/L).

RESULTS

A two-compartment model with linear elimination adequately described ceftaroline concentrations (n = 274). The relevance of the dosing components dosing interval/infusion duration/TDD for ƒT>MIC systematically changed with pathogen susceptibility. For susceptible pathogens with MIC ≤1 mg/L, shortened dosing intervals appeared as the main driver of the improved target attainment associated with the intensified dosing regimen, followed by increased TDD and infusion duration. For less susceptible pathogens, the advantage of q8h dosing and 2 h infusions declined, and increased TDD improved ƒT>MIC the most.

CONCLUSION

The analysis calls to mind consideration of dose increases when prolonging the infusion duration in the case of low bacterial susceptibility.

Operational characteristics of full random effects modelling ('frem') compared to stepwise covariate modelling ('scm')

An adequate covariate selection is a key step in population pharmacokinetic modelling. In this study, the automated stepwise covariate modelling technique ('scm') was compared to full random effects modelling ('frem'). We evaluated the power to identify a 'true' covariate (covariate with highest correlation to the pharmacokinetic parameter), precision, and accuracy of the parameter-covariate estimates. Furthermore, the predictive performance of the final models was assessed. The scenarios varied in covariate effect sizes, number of individuals (n = 20-500) and covariate correlations (0-90% cov-corr). The PsN 'frem' routine provides a 90% confidence intervals around the covariate effects. This was used to evaluate its operational characteristics for a statistical backward elimination procedure, defined as 'frem_posthoc' and to facilitate the comparison to 'scm'. 'Frem_posthoc' had a higher power to detect the true covariate with lower bias in small n studies compared to 'scm', applied with commonly used settings (forward p < 0.05, backward p < 0.01). This finding was vice versa in a statistically similar setting. For 'frem_posthoc', power, precision and accuracy of the covariate coefficient increased with higher number of individuals and covariate effect magnitudes. Without a backward elimination step 'frem' models provided unbiased coefficients with highly imprecise coefficients in small n datasets. Yet, precision was superior to final 'scm' model precision obtained using common settings. We conclude that 'frem_posthoc' is also a suitable method to guide covariate selection, although intended to serve as a full model approach. However, a deliberated selection of automated methods is essential for the modeller and using those methods in small datasets needs to be taken with caution.

Comparison of ultrafiltration and microdialysis for ceftriaxone protein-binding determination

BACKGROUND

High protein binding (PB) of antibiotics has an impact on their antimicrobial activity. It has been questioned whether in vitro PB determination can capture the dynamic and concentration-dependent PB of highly bound antibiotics.

OBJECTIVES

This clinical study compared in vitro ultrafiltration (UF) and in vivo IV microdialysis (MD) methods to determine ceftriaxone PB.

METHODS

Six healthy male volunteers received a single IV 2 g dose of ceftriaxone. Unbound ceftriaxone plasma concentrations were measured with MD and venous plasma sampling with subsequent UF. Pharmacokinetic parameters were determined using non-compartmental pharmacokinetic analysis. Non-linear mixed-effects modelling was used to quantify the PB. The PTA was estimated.

RESULTS

The Cmax of ceftriaxone total plasma concentration (297.42 ± 21.0 mg/L) was approximately 5.5-fold higher than for free concentrations obtained with UF (52.83 ± 5.07 mg/L), and only 3.5-fold higher than for free concentrations obtained with MD (81.37 ± 26.93 mg/L). Non-linear, saturable PB binding was confirmed for both UF and MD. Significantly different dissociation constants (Kd) for the albumin/ceftriaxone complex were quantified: in UF it was 23.7 mg/L (95% CI 21.3-26.2) versus 15.9 mg/L (95% CI 13.6-18.6) in MD. Moreover, the estimated number of binding sites (95% CI) per albumin molecule was 0.916 (0.86-0.97) in UF versus 0.548 in MD (0.51-0.59). The PTA obtained with MD was at most 27% higher than with UF.

CONCLUSIONS

In vitro UF versus in vivo intravasal MD revealed significantly different PB, especially during the distribution phase. The method of PB determination could have an impact on the breakpoint determination and dose optimisation of antibiotics.

Pilot Pharmacokinetic Study in Healthy Adults Using Intravascular Microdialysis Catheters Modified for Use in Paediatric Patients to Assess Vancomycin Blood Levels

BACKGROUND AND OBJECTIVE

Exhaustive pharmacokinetic (PK) studies in paediatric patients are unavailable for most antibiotics and feasibility of PK studies is limited by challenges, such as low blood volume and venipuncture-related pain. Microdialysis (MD) represents a promising method to overcome these obstacles. The aim of this proof-of-concept study was to develop and validate modified MD catheters that can be used to obtain concentration-time profiles of antibiotics in paediatric patients.

METHODS

Following extensive in vitro MD experiments, a prospective open-labelled study in ten healthy adult volunteers (HVs) was conducted. Subjects received a single intravenous dose of 1000 mg vancomycin, then plasma and intravascular microdialysate were sampled over 24 h. In vivo MD probe calibration was conducted using the retrodialysis technique. Plasma protein binding was measured using ultrafiltration. Confirmation of the measurements was performed using a Bland-Altman plot, relevant PK parameters were calculated, and a pharmacometric model was established.

RESULTS

No safety issues were encountered. The concentration-time curves of microdialysate and plasma measurements showed good alignment. The Bland-Altman plot yielded a mean bias of 0.19 mg/L and 95% limits of agreement of - 9.34 to 9.71 mg/L. A two-compartment model best described plasma PK, model-based estimates for recovery of the MD probes being in high agreement with the observed values. Quantified estimates of fraction unbound were comparable between plasma and microdialysate (p = 0.56).

CONCLUSIONS

An innovative MD catheter that can be inserted into small intravenous lines was successfully developed and applied in HV. This proof-of-concept study is encouraging and opens the way to further experiments leading towards future use of MD in paediatric patients.

Pharmacokinetics of Ribavirin in the Treatment of Lassa Fever: An Observational Clinical Study at the Irrua Specialist Teaching Hospital, Edo State, Nigeria

BACKGROUND

Lassa fever is endemic in large parts of West Africa. The recommended antiviral treatment is ribavirin. Two treatment regimens are currently endorsed in Nigeria: the "McCormick regimen" based on a study published in 1986 and the "Irrua regimen" constituting a simplified schedule developed at the Irrua Specialist Teaching Hospital, Nigeria. Evidence for the safety and efficacy of ribavirin in Lassa fever patients is poor and pharmacokinetic data for both regimens are lacking.

METHODS

Polymerase chain reaction-confirmed Lassa fever patients with mild to moderate disease severity were invited to participate in this prospective, observational pharmacokinetic study. Pharmacokinetics of ribavirin, clinical, virologic, and clinical laboratory parameters were assessed.

RESULTS

Using a population pharmacokinetic approach, plasma concentrations of ribavirin were best described by a 3-compartment model. Drug exposure was remarkably consistent between participants. Overall, drug clearance was 28.5% lower in female compared with male participants. Median (5th-95th percentile) time above half maximal inhibitory concentration (IC50) was 37.3% (16.9%-73.1%), 16.7% (8.2%-58.5%), and 9.6% (4.9%-38.4%) on days 1, 7, and 8, respectively. Clinical laboratory parameters indicated reduction of cell damage and development of hemolytic anemia in the course of the treatment period.

CONCLUSIONS

This observational study characterizes the pharmacokinetics of ribavirin in the treatment of Lassa fever indicating consistent exposure across patients. Whereas only a short time interval of concentrations above the IC50 implies rather low antiviral efficacy in vivo, the prominent reduction of cell damage markers might point to indirect-potentially anti-inflammatory-effects of ribavirin. The role of ribavirin in the treatment of Lassa fever requires further scrutiny.

Ciprofloxacin in Patients Undergoing Extracorporeal Membrane Oxygenation (ECMO): A Population Pharmacokinetic Study

Extracorporeal membrane oxygenation (ECMO) is utilized to temporarily sustain respiratory and/or cardiac function in critically ill patients. Ciprofloxacin is used to treat nosocomial infections, but data describing the effect of ECMO on its pharmacokinetics is lacking. Therefore, a prospective, observational trial including critically ill adults (n = 17), treated with ciprofloxacin (400 mg 8–12 hourly) during ECMO, was performed. Serial blood samples were collected to determine ciprofloxacin concentrations to assess their pharmacokinetics. The pharmacometric modeling was performed (NONMEM^®) and utilized for simulations to evaluate the probability of target attainment (PTA) to achieve an AUC_0–24/MIC of 125 mg·h/L for ciprofloxacin. A two-compartment model most adequately described the concentration-time data of ciprofloxacin. Significant covariates on ciprofloxacin clearance (CL) were plasma bicarbonate and the estimated glomerular filtration rate (eGFR). For pathogens with an MIC of ≤0.25 mg/L, a PTA of ≥90% was attained. However, for pathogens with an MIC of ≥0.5 mg/L, plasma bicarbonate ≥ 22 mmol/L or eGFR ≥ 10 mL/min PTA decreased below 90%, steadily declining to 7.3% (plasma bicarbonate 39 mmol/L) and 21.4% (eGFR 150 mL/min), respectively. To reach PTAs of ≥90% for pathogens with MICs ≥ 0.5 mg/L, optimized dosing regimens may be required.

Tigecycline Dosing Strategies in Critically Ill Liver-Impaired Patients

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.

Bayesian Nonlinear Models for Repeated Measurement Data: An Overview, Implementation, and Applications

Nonlinear mixed effects models have become a standard platform for analysis when data is in the form of continuous and repeated measurements of subjects from a population of interest, while temporal profiles of subjects commonly follow a nonlinear tendency. While frequentist analysis of nonlinear mixed effects models has a long history, Bayesian analysis of the models has received comparatively little attention until the late 1980s, primarily due to the time-consuming nature of Bayesian computation. Since the early 1990s, Bayesian approaches for the models began to emerge to leverage rapid developments in computing power, and have recently received significant attention due to (1) superiority to quantify the uncertainty of parameter estimation; (2) utility to incorporate prior knowledge into the models; and (3) flexibility to match exactly the increasing complexity of scientific research arising from diverse industrial and academic fields. This review article presents an overview of modeling strategies to implement Bayesian approaches for the nonlinear mixed effects models, ranging from designing a scientific question out of real-life problems to practical computations.

Voriconazole Pharmacokinetics Are Not Altered in Critically Ill Patients with Acute-on-Chronic Liver Failure and Continuous Renal Replacement Therapy: An Observational Study

Infection and sepsis are a main cause of acute-on-chronic liver failure (ACLF). Besides bacteria, molds play a role. Voriconazole (VRC) is recommended but its pharmacokinetics (PK) may be altered by ACLF. Because ACLF patients often suffer from concomitant acute renal failure, we studied the PK of VRC in patients receiving continuous renal replacement therapy (RRT) with ACLF and compared it to PK of VRC in critically ill patients with RRT without concomitant liver failure (NLF). In this prospective cohort study, patients received weight-based VRC. Pre- and post-dialysis membrane, and dialysate samples obtained at different time points were analyzed by high-performance liquid chromatography. An integrated dialysis pharmacometric model was used to model the available PK data. The recommended, 50% lower, and 50% higher doses were analyzed by Monte-Carlo simulation (MCS) for day 1 and at steady-state with a target trough concentration (TC) of 0.5–3mg/L. Fifteen patients were included in this study. Of these, 6 patients suffered from ACLF. A two-compartment model with linear clearance described VRC PK. No difference for central (V1) or peripheral (V2) volumes of distribution or clearance could be demonstrated between the groups. V1 was 80.6L (95% confidence interval: 62.6–104) and V2 106L (65–166) with a body clearance of 4.7L/h (2.87–7.81) and RRT clearance of 1.46L/h (1.29–1.64). MCS showed TC below/within/above target of 10/74/16% on day 1 and 9/39/52% at steady-state for the recommended dose. A 50% lower dose resulted in 26/72/1% (day 1) and 17/64/19% at steady-state and 7/57/37% and 7/27/67% for a 50% higher dose. VRC pharmacokinetics are not significantly influenced by ACLF in critically ill patients who receive RRT. Maintenance dose should be adjusted in both groups. Due to the high interindividual variability, therapeutic drug monitoring seems inevitable.

Protein binding of clindamycin in vivo by means of intravascular microdialysis in healthy volunteers

OBJECTIVES

The efficacy of an anti-infective drug is influenced by its protein binding (PB), since only the free fraction is active. We hypothesized that PB may vary in vitro and in vivo, and used clindamycin, a drug with high and concentration-dependent PB to investigate this hypothesis.

METHODS

Six healthy volunteers received a single intravenous infusion of clindamycin 900 mg. Antibiotic plasma concentrations were obtained by blood sampling and unbound drug concentrations were determined by means of in vivo intravascular microdialysis (MD) or in vitro ultrafiltration (UF) for up to 8 h post dosing. Clindamycin was assayed in plasma and MD fluid using a validated HPLC-UV (ultraviolet) method. Non-linear mixed effects modelling in NONMEM® was used to quantify the PB in vivo and in vitro.

RESULTS

C max was 14.95, 3.39 and 2.32 mg/L and AUC0-8h was 41.78, 5.80 and 6.14 mg·h/L for plasma, ultrafiltrate and microdialysate, respectively. Calculated ratio of AUCunbound/AUCtotal showed values of 13.9%±1.8% and 14.7%±3.1% for UF and microdialysate, respectively. Modelling confirmed non-linear, saturable PB for clindamycin with slightly different median (95% CI) dissociation constants (Kd) for the alpha-1 acid glycoprotein (AAG)-clindamycin complex of 1.16 mg/L (0.91-1.37) in vitro versus 0.85 mg/L (0.58-1.01) in vivo. Moreover, the estimated number of binding sites per AAG molecule was 2.07 (1.79-2.25) in vitro versus 1.66 in vivo (1.41-1.79).

CONCLUSIONS

Concentration-dependent PB was observed for both investigated methods with slightly lower levels of unbound drug fractions in vitro as compared with in vivo.

Conditional Non-parametric Bootstrap for Non-linear Mixed Effect Models

PURPOSE

Non-linear mixed effect models are widely used and increasingly integrated into decision-making processes. Propagating uncertainty is an important element of this process, and while standard errors (SE) on pa- rameters are most often computed using asymptotic approaches, alternative methods such as the bootstrap are also available. In this article, we propose a modified residual parametric bootstrap taking into account the different levels of variability involved in these models.

METHODS

The proposed approach uses samples from the individual conditional distribution, and was implemented in R using the saemix algorithm. We performed a simulation study to assess its performance in different scenarios, comparing it to the asymptotic approximation and to standard bootstraps in terms of coverage, also looking at bias in the parameters and their SE.

RESULTS

Simulations with an Emax model with different designs and sigmoidicity factors showed a similar coverage rate to the parametric bootstrap, while requiring less hypotheses. Bootstrap improved coverage in several scenarios compared to the asymptotic method especially for the variance param-eters. However, all bootstraps were sensitive to estimation bias in the original datasets.

CONCLUSIONS

The conditional bootstrap provided better coverage rate than the traditional residual bootstrap, while preserving the structure of the data generating process.

A Model Averaging/Selection Approach Improves the Predictive Performance of Model-Informed Precision Dosing: Vancomycin as a Case Study

Many important drugs exhibit substantial variability in pharmacokinetics and pharmacodynamics leading to a loss of the desired clinical outcomes or significant adverse effects. Forecasting drug exposures using pharmacometric models can improve individual target attainment when compared with conventional therapeutic drug monitoring (TDM). However, selecting the "correct" model for this model-informed precision dosing (MIPD) is challenging. We derived and evaluated a model selection algorithm (MSA) and a model averaging algorithm (MAA), which automates model selection and finds the best model or combination of models for each patient using vancomycin as a case study, and implemented both algorithms in the MIPD software "TDMx." The predictive performance (based on accuracy and precision) of the two algorithms was assessed in (i) a simulation study of six distinct populations and (ii) a clinical dataset of 180 patients undergoing TDM during vancomycin treatment and compared with the performance obtained using a single model. Throughout the six virtual populations the MSA and MAA (imprecision: 9.9-24.2%, inaccuracy: less than ± 8.2%) displayed more accurate predictions than the single models (imprecision: 8.9-51.1%; inaccuracy: up to 28.9%). In the clinical dataset, the predictive performance of the single models applying at least one plasma concentration varied substantially (imprecision: 28-62%, inaccuracy: -16 to 25%), whereas the MSA or MAA utilizing these models simultaneously resulted in unbiased and precise predictions (imprecision: 29% and 30%, inaccuracy: -5% and 0%, respectively). MSA and MAA approaches implemented in TDMx might thereby lower the burden of fit-for-purpose validation of individual models and streamline MIPD.

An Integrated Dialysis Pharmacometric (IDP) Model to Evaluate the Pharmacokinetics in Patients Undergoing Renal Replacement Therapy

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 clearanceRRT (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, clearanceRRT, drug degradation (effluent-half-lifedoripenem: 13.5 h-1) and adsorption (polysulphone adsorption capacityteicoplanin: 31.2 mg) were assessed.

CONCLUSION

The IDP model allows accurate, precise and sensitive characterization of clearanceRRT, adsorption and degradation. Successful quantification of all aspects of clearanceRRT in clinical data demonstrated the benefit of the IDP model as compared to conventional approaches.