A population approach to the forecasting of amikacin plasma and urinary levels using a prescribed dosage regimen

A simulation study on model-informed precision dosing of amikacin for achieving target area under the concentration-time curve

AIMS

Amikacin requires therapeutic drug monitoring for optimum efficacy; however, the optimal model-informed precision dosing strategy for the area under the concentration-time curve (AUC) of amikacin is uncertain. This simulation study aimed to determine the efficient blood sampling points using the Bayesian forecasting approach for early achievement of the target AUC range for amikacin in critically ill patients.

METHODS

We generated a virtual population of 3000 individuals using 2 validated population pharmacokinetic models identified using a systematic literature search. AUC for each blood sampling point was evaluated using the probability of achieving a ratio of estimated/reference AUC at steady state in the 0.8-1.2 range.

RESULTS

On day 1, the 1-point samplings for population pharmacokinetic models showed a priori probabilities of 26.3 and 45.6%, which increased to 47.3 and 94.4% at 23 and 15 h, respectively. Using 2-point sampling at the peak (3 and 4 h) and trough (24 h) on day 1, these probabilities further increased to 72.3 and 99.5%, respectively. These probabilities were comparable on days 2 and 3, regardless of 3 and 6 sampling points or estimated glomerular filtration rate. These results indicated the higher predictive accuracy of 2-point sampling than 1-point sampling on day 1 for amikacin AUC estimation. Moreover, 2-point sampling was a more reasonable approach than rich sampling.

CONCLUSIONS

This study contributes to the development of an efficient model-informed precision dosing strategy for early targeting of amikacin AUC in critically ill patients.

Population Pharmacokinetics of Amikacin in Patients on Veno-Arterial Extracorporeal Membrane Oxygenation

Veno-arterial extracorporeal membrane oxygenation (V-A ECMO) support leads to complex pharmacokinetic alterations, whereas adequate drug dosing is paramount for efficacy and absence of toxicity in critically ill patients. Amikacin is a major antibiotic used in nosocomial sepsis, especially for these patients. We aimed to describe amikacin pharmacokinetics on V-A ECMO support and to determine relevant variables to improve its dosing. All critically ill patients requiring empirical antimicrobial therapy, including amikacin for nosocomial sepsis supported or not by V-A ECMO, were included in a prospective population pharmacokinetic study. This population pharmacokinetic analysis was built with a dedicated software, and Monte Carlo simulations were performed to identify doses achieving therapeutic plasma concentrations. Thirty-nine patients were included (control n = 15, V-A ECMO n = 24); 215 plasma assays were performed and used for the modeling process. Patients received 29 (24–33) and 32 (30–35) mg/kg of amikacin in control and ECMO groups, respectively. Data were best described by a two-compartment model with first-order elimination. Inter-individual variabilities were observed on clearance, central compartment volume (V₁)_, and peripherical compartment volume (V₂). Three significant covariates explained these variabilities: Kidney Disease Improving Global Outcomes (KDIGO) stage on amikacin clearance, total body weight on V_1, and ECMO support on V₂. Our simulations showed that the adequate dosage of amikacin was 40 mg/kg in KDIGO stage 0 patients, while 25 mg/kg in KDIGO stage 3 patients was relevant. V-A ECMO support had only a secondary impact on amikacin pharmacokinetics, as compared to acute kidney injury.

Evaluation of Current Amikacin Dosing Recommendations and Development of an Interactive Nomogram: The Role of Albumin

This study aimed to evaluate the potential efficacy and safety of the amikacin dosage proposed by the main guidelines and to develop an interactive nomogram, especially focused on the potential impact of albumin on initial dosage recommendation. The probability of target attainment (PTA) for each of the different dosing recommendations was calculated through stochastic simulations based on pharmacokinetic/pharmacodynamic (PKPD) criteria. Large efficacy and safety differences were observed for the evaluated amikacin dosing guidelines together with a significant impact of albumin concentrations on efficacy and safety. For all recommended dosages evaluated, efficacy and safety criteria of amikacin dosage proposed were not achieved simultaneously in most of the clinical scenarios evaluated. Furthermore, a significant impact of albumin was identified: The higher is the albumin, (i) the higher will be the PTA for maximum concentration/minimum inhibitory concentration (Cmax/MIC), (ii) the lower will be the PTA for the time period with drug concentration exceeding MIC (T_>MIC) and (iii) the lower will be the PTA for toxicity (minimum concentration). Thus, accounting for albumin effect might be of interest for future amikacin dosing guidelines updates. In addition, AMKnom, an amikacin nomogram builder based on PKPD criteria, has been developed and is freely available to help evaluating dosing recommendations.

Evaluation of amikacin use and comparison of the models implemented in two Bayesian forecasting software packages to guide dosing

AIMS

Bayesian forecasting software can assist in guiding therapeutic drug monitoring (TDM)-based dose adjustments for amikacin to achieve therapeutic targets. This study aimed to evaluate amikacin prescribing and TDM practices, and to determine the suitability of the amikacin model incorporated into the DoseMeRx® software as a replacement for the previously available software (Abbottbase®).

METHODS

Patient demographics, pathology, amikacin dosing history, amikacin concentrations and Abbottbase® predicted TDM targets (area under the curve up to 24 hours, maximum concentration and trough concentration) were collected for adults receiving intravenous amikacin (2012-2017). Concordance with the Australian Therapeutic Guidelines was assessed. Observed and predicted amikacin concentrations were compared to determine the predictive performance (bias and precision) of DoseMeRx®. Amikacin TDM targets were predicted by DoseMeRx® and compared to those predicted by Abbottbase®.

RESULTS

Overall, guideline compliance for 63 courses of amikacin in 47 patients was suboptimal. Doses were often lower than recommended. For therapy >48 h, TDM sample collection timing was commonly discordant with recommendations, therapeutic target attainment low and 34% of dose adjustments inappropriate. DoseMeRx® under-predicted amikacin concentrations by 0.9 mg/L (95% confidence interval [CI] -1.4 to -0.5) compared with observed concentrations. However, maximum concentration values (n = 19) were unbiased (-1.7 mg/L 95%CI -5.8 to 0.8) and precise (8.6% 95%CI 5.4-18.1). Predicted trough concentration values (n = 7) were, at most, 1 mg/L higher than observed. Amikacin area under the curve values estimated using Abbottbase® (181 mg h/L 95%CI 161-202) and DoseMeRx® (176 mg h/L 95%CI 152-199) were similar (P = .59).

CONCLUSION

Amikacin dosing and TDM practice was suboptimal compared with guidelines. The model implemented by DoseMeRx® is satisfactory to guide amikacin dosing.

Population Pharmacokinetic Study of the Suitability of Standard Dosing Regimens of Amikacin in Critically Ill Patients with Open-Abdomen and Negative-Pressure Wound Therapy

The aim was to assess the appropriateness of recommended regimens for empirical MIC coverage in critically ill patients with open-abdomen and negative-pressure therapy (OA/NPT). Over a 5-year period, every critically ill patient who received amikacin and who underwent therapeutic drug monitoring (TDM) while being treated by OA/NPT was retrospectively included. A population pharmacokinetic (PK) modeling was performed considering the effect of 10 covariates (age, sex, total body weight [TBW], adapted body weight [ABW], body surface area [BSA], modified sepsis-related organ failure assessment [SOFA] score, vasopressor use, creatinine clearance [CL_CR], fluid balance, and amount of fluids collected by the NPT over the sampling day) in patients who underwent continuous renal replacement therapy (CRRT) or did not receive CRRT. Monte Carlo simulations were employed to determine the fractional target attainment (FTA) for the PK/pharmacodynamic [PD] targets (maximum concentration of drug [C _max]/MIC ratio of ≥8 and a ratio of the area under the concentration-time curve from 0 to 24 h [AUC_0-24]/MIC of ≥75). Seventy critically ill patients treated by OA/NPT (contributing 179 concentration values) were included. Amikacin PK concentrations were best described by a two-compartment model with linear elimination and proportional residual error, with CL_CR and ABW as significant covariates for volume of distribution (V) and CL_CR for CL. The reported V) in non-CRRT and CRRT patients was 35.8 and 40.2 liters, respectively. In Monte Carlo simulations, ABW-adjusted doses between 25 and 35 mg/kg were needed to reach an FTA of >85% for various renal functions. Despite an increased V and a wide interindividual variability, desirable PK/PD targets may be achieved using an ABW-based loading dose of 25 to 30 mg/kg. When less susceptible pathogens are targeted, higher dosing regimens are probably needed in patients with augmented renal clearance (ARC). Further studies are needed to assess the effect of OA/NPT on the PK parameters of antimicrobial agents.

Population Pharmacokinetics of Amikacin in Adult Patients with Cystic Fibrosis

Practitioners commonly use amikacin in patients with cystic fibrosis. Establishment of the pharmacokinetics of amikacin in adults with cystic fibrosis may increase the efficacy and safety of therapy. This study was aimed to establish the population pharmacokinetics of amikacin in adults with cystic fibrosis. We used serum concentration data obtained during routine therapeutic drug monitoring and explored the influence of patient covariates on drug disposition. We performed a retrospective chart review to collect the amikacin dosing regimens, serum amikacin concentrations, blood sampling times, and patient characteristics for adults with cystic fibrosis admitted for treatment of acute pulmonary exacerbations. Amikacin concentrations were retrospectively collected for 49 adults with cystic fibrosis, and 192 serum concentrations were available for analysis. A population pharmacokinetic model was developed using nonlinear mixed-effects modeling with the first-order conditional estimation method. A two-compartment model with first-order elimination best described amikacin pharmacokinetics. Creatinine clearance and weight were identified as significant covariates for clearance and the volume of distribution, respectively, in the final model. Residual variability was modeled using a proportional error model. Typical estimates for clearance, central and peripheral volumes of distribution, and intercompartmental clearance were 3.06 liters/h, 14.4 liters, 17.1 liters, and 0.925 liters/h, respectively. The pharmacokinetics of amikacin in individuals with cystic fibrosis seems to differ from those in individuals without cystic fibrosis. However, further investigations are needed to confirm these results and, thus, the need for variations in amikacin dosing. Future pharmacodynamic studies will potentially establish the optimal amikacin dosing regimens for the treatment of acute pulmonary exacerbations in adult patients with CF.

Evaluation of population pharmacokinetic models for amikacin dosage individualization in critically ill patients

Objectives

The aim of this study was to evaluate the reliability for dosage individualization and Bayesian adaptive control of several literature-retrieved amikacin population pharmacokinetic models in patients who were critically ill.

Methods

Four population pharmacokinetic models, three of them customized for critically-ill patients, were applied using pharmacokinetic software to fifty-one adult patients on conventional amikacin therapy admitted to the intensive care unit. An estimation of patient-specific pharmacokinetic parameters for each model was obtained by retrospective analysis of the amikacin serum concentrations measured (n = 162) and different clinical covariates. The model performance for a priori estimation of the area under the serum concentration-time curve (AUC) and maximum serum drug concentration (Cmax) targets was obtained.

Key findings

Our results provided valuable confirmation of the clinical importance of the choice of population pharmacokinetic models when selecting amikacin dosages for patients who are critically ill. Significant differences in model performance were especially evident when only information concerning clinical covariates was used for dosage individualization and over the two most critical determinants of clinical efficacy of amikacin i.e. the AUC and Cmax values.

Conclusions

Only a single amikacin serum level seemed necessary to diminish the influence of population model on dosage individualization.

Renal elimination of amikacin and the aging process

OBJECTIVE

Although amikacin is primarily eliminated via glomerular filtration, drug concentrations are not consistently predicted in all patients. To better describe the relationship between amikacin clearance and both age and renal function, we used a new heuristic approach involving statistical analysis of dependence.

DESIGN AND SETTING

Retrospective pharmacokinetic study using data from seven centres in France.

PARTICIPANTS

634 patients with sepsis aged between 18 and 98 years of age who received intravenous amikacin.

METHODS

Clearance of amikacin was modelled using the NonParametric EM algorithm for a two-compartment model (NPEM2) with intravenous infusion.

RESULTS

A total of 2499 serum amikacin determinations was available for analysis. The relationship between the clearance of amikacin and age was weak. Interestingly, the Z method, which filters data based on dependence criteria, selected data that were best fitted by a polynomial function (r = 0.90; p < 0.001). This representation of the polynomial function was similar to a previously proposed theoretical model describing covariations between the clearance of amikacin and age. However, the polynomial function applied to only 33% of the patients that were selected by the Z method. The correlation between the clearance of amikacin and renal function was also relatively low (r = 0.39). The Z method exhibited a continuous and strong dependence pattern between the clearance of amikacin and age for 49% of the patients.

CONCLUSIONS

The Z methodology, which filters data using dependence criteria, confirms that age, renal function and amikacin clearance are strongly related, but only in less than half of a large sample of patients with sepsis without renal pathology. These results suggest that other variables should be taken into account in order to improve the description of the behaviour of amikacin. The Z methodology improved the classical description of relationships between variables, and should be applied to better select pertinent variables in pharmacokinetic studies.