Bayesian estimation and prediction of clearance in high-dose methotrexate infusions

Much attention has been paid to the problem of estimating the pharmacokinetic parameters of individual patients in order to optimize dosage choices. Individual kinetics determined by test-dose bolus injection are a basis for predicting drug concentrations after high-dose methotrexate infusion and for computing appropriate dosages. Simplifications may be attempted, even allowing the test-dose to be omitted by using Bayesian estimation rather than likelihood estimation. To individualize pharmacokinetic parameters, Bayesian estimation combines information about population characteristics and those of individuals based on few measured plasma levels during high-dose infusion. Application of this procedure to methotrexate reveals interesting predictive performances and ability to handle variation due to intraindividual time variability without using test doses. The methodology promises to be more efficient in computing dosages in order to avoid toxic levels and will be less expensive in routine clinical use.Part of this work constituted a DEA thesis in Pharmacokinetics presented by Mr Bachir-Raho.     

Bayesian estimation and prediction of clearance in high-dose methotrexate infusions

Much attention has been paid to the problem of estimating the pharmacokinetic parameters of individual patients in order to optimize dosage choices. Individual kinetics determined by test-dose bolus injection are a basis for predicting drug concentrations after high-dose methotrexate infusion and for computing appropriate dosages. Simplifications may be attempted, even allowing the test-dose to be omitted by using Bayesian estimation rather than likelihood estimation. To individualize pharmacokinetic parameters, Bayesian estimation combines information about population characteristics and those of individuals based on few measured plasma levels during high-dose infusion. Application of this procedure to methotrexate reveals interesting predictive performances and ability to handle variation due to intraindividual time variability without using test doses. The methodology promises to be more efficient in computing dosages in order to avoid toxic levels and will be less expensive in routine clinical use.     

Bayesian estimation of cyclosporine clearance in bone marrow graft

The dosage regimen of cyclosporine (CsA) can be individualized in patients by means of a test dose (TD) method in the few days before bone marrow graft. To simplify the test dosing protocol, we used a Bayesian estimation (BE) of CsA clearance requiring population data and partial kinetic information for a given patient. In the first part, 42 patients, aged 11-47 years, were given a 2-h infusion of CsA (4 mg.kg-1). CsA concentrations were determined in several blood samples. The obtained concentration-time curves were fitted according to a three-compartment model. Maximum likelihood estimation (MLE) allowed CsA determination of pharmacokinetic parameters. Early population parameters of CsA were determined using 22 TD by a two-stage method. In the remaining 20 patients, individual pharmacokinetic parameters were also computed by BE procedure, taking into account the above determined population parameters and CsA concentration determinations from three blood samples drawn at 5 and 30 min and 3 h after the end of TD infusion. In the second part, 16 patients were used to evaluate performance of BE in directly predicting target concentration values. Finally, early population characteristics were updated on the basis of 42 patients. Statistical comparisons between MLE and BE estimates of CsA clearance and between concentrations predicted after BE and those experimentally obtained showed that three blood CsA determinations allowed accurate clearance estimation and target concentration predictions. The method presented here can be used to calculate an individual CsA dosage regimen in real time, thus improving patient comfort.(ABSTRACT TRUNCATED AT 250 WORDS)     

Two general methods for population pharmacokinetic modeling: non-parametric adaptive grid and non-parametric Bayesian

Population pharmacokinetic (PK) modeling methods can be statistically classified as either parametric or nonparametric (NP). Each classification can be divided into maximum likelihood (ML) or Bayesian (B) approaches. In this paper we discuss the nonparametric case using both maximum likelihood and Bayesian approaches. We present two nonparametric methods for estimating the unknown joint population distribution of model parameter values in a pharmacokinetic/pharmacodynamic (PK/PD) dataset. The first method is the NP Adaptive Grid (NPAG). The second is the NP Bayesian (NPB) algorithm with a stick-breaking process to construct a Dirichlet prior. Our objective is to compare the performance of these two methods using a simulated PK/PD dataset. Our results showed excellent performance of NPAG and NPB in a realistically simulated PK study. This simulation allowed us to have benchmarks in the form of the true population parameters to compare with the estimates produced by the two methods, while incorporating challenges like unbalanced sample times and sample numbers as well as the ability to include the covariate of patient weight. We conclude that both NPML and NPB can be used in realistic PK/PD population analysis problems. The advantages of one versus the other are discussed in the paper. NPAG and NPB are implemented in R and freely available for download within the Pmetrics package from www.lapk.org.     

Maximum a posteriori Bayesian estimation of epirubicin clearance by limited sampling

AIMS: To develop a limited sampling strategy for estimation of epirubicin clearance. METHODS: The data set comprised 1051 concentrations measured in 105 patients with advanced or metastatic breast cancer treated with epirubicin alone. Ten limited sampling designs comprising two or three blood samples were proposed, taken at times identified by D-optimality from population pharmacokinetic parameter estimates. The data set was then truncated to include the sampling times for each of the designs. MAP Bayesian estimates of clearance were generated for each design and compared with clearance estimates obtained using all the data. The limited sampling designs were also validated using a separate data set obtained from 18 patients with either breast cancer or hepatocellular carcinoma. The sensitivity of the best limited sampling designs to sample time recording errors of 0-10% or 10-20% was then assessed using a simulated data set including 200 patients. RESULTS: The optimum sampling times were: end of the injection and 18 min, 40 min, 3 h, 10 h and 48 h after the start of the injection. The best three-sample design included samples at 40 min, 3 h and 48 h and gave unbiased estimates of clearance with an imprecision of 9.1% (95% CI 7.3, 10.5). The best two sample design included samples at 3 and 48 h and gave unbiased estimates of clearance with an imprecision of 12.4% (95% CI 9.6, 14.6). Using the validation data set, these two and three sample designs gave unbiased estimates of clearance with an imprecision of 5.6% (95% CI 3.7, 7.0) and 4.2% (95% CI 2.6, 5.3), respectively. Simulations that included 0-10% or 10-20% errors in the recording of the blood sampling times had negligible effects on the bias and imprecision of clearance estimates. CONCLUSIONS: Limited sampling designs have been identified and validated that estimate epirubicin clearance with adequate precision and without bias from two or three blood samples. These designs also allow flexibility in blood sample collection and are robust with regard to sample time recording errors.     

Nonparametric expectation maximisation (NPEM) population pharmacokinetic analysis of caffeine disposition from sparse data in adult caucasians: systemic caffeine clearance as a biomarker for cytochrome P450 1A2 activity

OBJECTIVE: To explore the ability of the nonparametric expectation maximisation (NPEM) method of population pharmacokinetic modelling to deal with sparse data in estimating systemic caffeine clearance for monitoring and evaluation of cytochrome P450 (CYP) 1A2 activity. DESIGN AND PARTICIPANTS: Nonblind, single-dose clinical investigation in 34 non-related adult Bulgarian Caucasians (18 women and 16 men, aged between 18 and 62 years) with normal and reduced renal function. METHODS: Each participant received oral caffeine 3 mg/kg. Two blood samples per individual were taken according to the protocol for measuring caffeine plasma concentrations. A total of 67 measured concentrations were used to obtain NPEM estimates of caffeine clearance. Paraxanthine/caffeine plasma ratios were calculated and correlated with clearance estimates. Graphical methods and tests for normality were applied and parametric and nonparametric statistical tests were used for comparison. RESULTS: NPEM median estimates of caffeine absorption and elimination rate constants, k(a) = 4.54 h(-1) and k(el) = 0.139 h(-1), as well as of fractional volume of distribution and plasma clearance, V(S1) = 0.58 L/kg and CL(S1) = 0.057 L/h/kg, agreed well with reported values from more 'data rich' studies. Significant correlations were observed between paraxanthine/caffeine ratios at 3, 8 and 10 hours and clearance (Spearman rank correlation coefficients, r(s), >0.74, p 相似文献   

Pharmacoepidemiologic investigation of clonazepam relative clearance by mixed-effect modeling using routine clinical pharmacokinetic data in Japanese patients.

The effects of drug-drug interactions on clonazepam clearance were examined through a retrospective analysis of serum concentration data from pediatric and adult epileptic patients. Patients received clonazepam as monotherapy or in combination with other antiepileptic drugs. A total of 259 serum clonazepam concentrations gathered from 137 patients were used in a population analysis of drug-drug interactions on clonazepam clearance. Data were analyzed using a nonlinear mixed-effects modeling (NONMEM) technique. The final model describing clonazepam clearance was CL = 152 x TBW(-0.181) x DIF, where CL is clearance (ml/kg/h), TBWis total body weight (kg), and DIF (drug interaction factor) is a scaling factor for concomitant medication with a value of 1 for patients on clonazepam monotherapy, 1.18 for those patients receiving concomitant administration of clonazepam and one antiepileptic drug (carbamazepine or valproic acid), and 2.12 x TBW(-0.119) for those patients receiving concomitant administration of clonazepam and more than two antiepileptic drugs. Clonazepam clearance decreased in a weight-related fashion in children, with minimal changes observed in adults. Concomitant administration of clonazepam and carbamazepine resulted in a 22% increase in clonazepam clearance. Concomitant administration of clonazepam and valproic acid resulted in a 12% increase in clonazepam clearance. Concomitant administration of clonazepam with two or more antiepileptic drugs resulted in a 23% to 75% increase in clonazepam clearance.     

Epidemiologic investigation of the relative clearance of haloperidol by mixed-effect modeling using routine clinical pharmacokinetic data in Japanese patients

The steady-state trough concentrations of haloperidol were studied to clarify the role of the characteristics of Japanese patients in estimating haloperidol dosing regimens by using routine therapeutic drug-monitoring data. Nonlinear mixed-effects modeling (NONMEM) was used to estimate the effect of a variety of developmental and demographic factors on haloperidol clearance values using 270 serum level measurements obtained from 191 patients during their clinical course. The final model describing haloperidol's relative clearance was CL = 0.74 x TBW(0.594) x DOSE(0.326) x 1.32CO1 x 0.867AGE, where CL is clearance (measured in liters per hour), TBW is the total body weight (in kilograms), DOSE is the daily dose of haloperidol (in grams per kilogram per day), CO1 = 1 for concomitant administration of antiepileptic drugs (phenobarbital, phenytoin, or carbamazepine) and CO1 = 0 otherwise, and AGE = 1 for patients aged 55 years or older and AGE = 0 otherwise. Concomitant administration of haloperidol and antiepileptic drugs resulted in a 32% increase in haloperidol clearance. Patients aged 55 years or older showed a 13.3% reduction in clearance values compared with the younger population.     

Vancomycin pharmacokinetics and Bayesian estimation in pediatric patients

The vancomycin pharmacokinetic profile was characterized in six pediatric patients and the potential of nonlinear mixed effects modeling and Bayesian forecasting for vancomycin monitoring was explored using NONMEM V (1.1). Based on steady state serial vancomycin concentrations, the estimates of mean t1/2, Vd, and Cl derived by the Sawchuk and Zaske method (1) were 3.52 hours, 0.57 L/kg, and 0.12 L/h per kg, respectively. NONMEM analysis demonstrated that a weight-adjusted two-compartment model described individual patients' data better than a comparable one-compartment model. The two-compartment estimates of mean t1/2alpha, t1/2beta, Vss, and Cl were 0.80 hour, 5.63 hours, 0.63 L/kg, and 0.11 L/h per kg, respectively. The relatively long mean t1/2alpha suggests that peak vancomycin concentrations measured earlier than 4 hours postdose do not reflect postdistributional serum concentrations. NONMEM population modeling revealed that a weight-adjusted two-compartment model provided a better fit than a comparable one-compartment model. The resulting population parameters and variances were fixed in NONMEM to obtain Bayesian predictions of individual vancomycin serum concentrations. Bayesian estimation with either a single midinterval or trough sample has the potential to provide accurate and precise predictions of vancomycin concentrations. This should be evaluated using a vancomycin population pharmacokinetic model based on a larger sample of pediatric patients.     

新铂类药物双环铂与顺铂、卡铂体内毒性的比较研究

目的:比较新铂类药物双环铂与顺铂、卡铂的体内毒性.方法:大鼠重复静脉注射双环铂、顺铂与卡铂,分别于给药结束以及恢复期末收集标本,进行血液学、血尿生化指标及病理切片的检测.结果:与溶剂对照组比较,重复静脉注射后,双环铂的高剂量组(13.89 mg·kg-1)全血网织红细胞以及红细胞明显降低(P＜0.01),骨髓切片示有核细胞显著减少;恢复期末网织红细胞和骨髓切片基本恢复至正常,而红细胞仍维持较低水平(P＜0.05).而双环铂各组尿酶、血尿素氮和肌苷以及肾脏病理切片在给药结束和恢复期均未见明显改变.结论:双环铂对SD大鼠的肾毒性明显低于顺铂,其骨髓抑制作用与卡铂相近,暗示肾毒性可能不会成为限制双环铂临床应用的主要毒副反应,临床应用需密切观察其对造血组织的损伤.     

Bayesian estimation of methotrexate pharmacokinetic parameters and area under the curve in children and young adults with localised osteosarcoma

BACKGROUND: Methotrexate is the most efficient anticancer drug in osteosarcoma. It requires individual exposure monitoring because of the high doses used, its wide interpatient pharmacokinetic variability and the existence of demonstrated relationships between efficacy, toxicity and serum drug concentrations. OBJECTIVE: To develop a maximum a posteriori (MAP) Bayesian estimator able to predict individual pharmacokinetic parameters and exposure indices such as area under the curve (AUC) for methotrexate from a few blood samples, in order to prevent toxicity and facilitate further studies of the relationships between efficacy and exposure. METHODS: Methotrexate population pharmacokinetics were estimated by a retrospective analysis of concentration data from 40 children and young adults by using the nonparametric expectation maximisation method NPEM. A linear two-compartment model with elimination from the central compartment was assumed. Individual pharmacokinetic parameters and AUC were subsequently estimated in 30 other young patients, using MAP Bayesian estimation as implemented in two programs, ADAPT II and an inhouse program Winphar((R)). RESULTS: The pharmacokinetic parameters used in the model were the volume of the central compartment (V(1)) and the transfer constants (k(10), k(12) and k(21)). The mean values (with percentage coefficient of variation) obtained were: 18.24L (54.1%) and 0.41 (42.3%), 0.0168 (68.7%), and 0.1069 (61.3%) h(-1), respectively. Bayesian forecasting enabled nonbiased estimation of AUC and systemic clearance using a schedule with two sampling times (6 and 24 hours after the beginning of the infusion) and either program. Collection of a third sample at 4 hours improved the precision. CONCLUSION: The Bayesian adaptive method developed herein allows accurate estimation of individual exposure to methotrexate and can easily be used in clinical practice.     

A mechanistic approach for the scaling of clearance in children

BACKGROUND AND OBJECTIVE: Clearance is an important pharmacokinetic concept for scaling dosage, understanding the risks of drug-drug interactions and environmental risk assessment in children. Accurate clearance scaling to children requires prior knowledge of adult clearance mechanisms and the age-dependence of physiological and enzymatic development. The objective of this research was to develop and evaluate ontogeny models that would provide an assessment of the age-dependence of clearance. METHODS: Using in vitro data and/or in vivo clearance values for children for eight compounds that are eliminated primarily by one process, models for the ontogeny of renal clearance, cytochrome P450 (CYP) 3A4, CYP2E1, CYP1A2, uridine diphosphate glucuronosyltransferase (UGT) 2B7, UGT1A6, sulfonation and biliary clearance were developed. Resulting ontogeny models were evaluated using six compounds that demonstrated elimination via multiple pathways. The proportion of total clearance attributed to each clearance pathway in adults was delineated. Each pathway was individually scaled to the desired age, inclusive of protein-binding prediction, and summed to generate a total plasma clearance for the child under investigation. The paediatric age range included in the study was premature neonates to sub-adults. RESULTS: There was excellent correlation between observed and predicted clearances for the model development (R2 = 0.979) and test sets (Q2 = 0.927). Clearance in premature neonates could also be well predicted (development R2 = 0.951; test Q2 = 0.899). CONCLUSION: Paediatric clinical trial development could greatly benefit from clearance scaling, particularly in guiding dosing regimens. Furthermore, since the proportion of clearance via different elimination pathways is age-dependent, information could be gained on the developmental extent of drug-drug interactions.     

Predictive performance of Bayesian and nonlinear least-squares regression programs for lidocaine.

The predictive performance of two computer programs for lidocaine dosing were evaluated. Two-compartment Bayesian and nonlinear least-squares regression programs were used in two groups of patients (15 acute arrhythmia patients and 14 chronic arrhythmia patients). Lidocaine was given as a 1.5 mg/kg bolus and a 2.8 mg/min infusion for 48 h. A second bolus (0.5 mg/kg) was given 10 min after the first bolus over 2 min. Serum samples of the patients receiving lidocaine were drawn at 2, 15, 30 min and 1, 2, and 4 h and were used in forecasting the serum concentrations at 6, 8, 12, and 48 h. Predictive performance was assessed by mean error and mean-squared error. The results (mean +/- 95% confidence intervals) demonstrated the Bayesian program predicted a significant (p less than 0.05) difference at 12 h between the two arrhythmia groups (acute 0.52 [-0.95; -0.09] and chronic 0.28 [0.12; 0.44]). The results also demonstrated the Bayesian method was significantly more precise compared to the nonlinear least-squares regression program at 8, 12, and 48 h for the acute group. While caution is warranted, this study demonstrated that the predictive performance by a two-compartment Bayesian model is more accurate in predicting future lidocaine serum concentrations than that by nonlinear least-squares regression.     

Limited sampling strategy using Bayesian estimation for estimating individual exposure of the once-daily prolonged-release formulation of tacrolimus in kidney transplant children

Background

A limited sampling strategy (LSS) for estimating the area under the curve (AUC) of the prolonged-release formulation of tacrolimus (tacrolimus^PR) is not available in pediatric patients, although the method is of real benefit to children. The objective of this study was to develop and validate a reliable and clinically applicable LSS using Bayesian estimation for estimating tacrolimus^PR AUC in pediatric kidney transplant patients

Methods

The original tacrolimus pharmacokinetic dataset consisted of 22 full profiles from 22 pediatric kidney transplant patients. The Bayesian estimation method was used to develop the LSS. External validation was performed in an independent validation group which consisted of 20 full pharmacokinetic profiles from 12 pediatric kidney transplant patients.

Results

Bayesian estimator using C_0h C_2h and C_3h gave the best predictive performance with a mean prediction error of 2.2 % in the external validation dataset. There was no correlation between the prediction error and age. The Bland–Altman analysis showed that the mean difference between the reference and Bayesian-estimated AUC_0-24 was 3.5 (95 % confidence interval ?3.5–10.5) ng h/mL

Conclusions

A reliable and clinically applicable LSS for estimating AUC_0–24 of tacrolimus^PR was determined and validated in children. The prediction was unbiased and precise. It can be used as a routine procedure to perform AUC-based tacrolimus^PR dosage optimization in pediatric renal transplant patients.     

Comparison of iothalamate clearance methods for measuring GFR.

STUDY OBJECTIVE: To evaluate the bias and precision of three methods of measuring glomerular filtration rate (GFR) relative to a standard method. DESIGN: Prospective, outpatient study. SETTING: University-affiliated general clinical research center. PATIENTS: Twenty-six patients with various degrees of renal function (GFR range 25-151 ml/min/1.73 m2). INTERVENTIONS: Each patient received iothalamate twice during the study visit, first as a bolus injection and then as a priming dose followed by a constant-rate infusion for 2.5 hours. MEASUREMENTS AND MAIN RESULTS: Plasma (ClpIVB) and renal clearances (ClrIVB) after bolus injection and plasma clearance during constant-rate infusion (ClpINF) were compared with standard renal clearance during constant-rate infusion (ClrINF). All three measures were highly correlated with ClrINF (r>0.90, p<0.001). The mean ClrIVB was not significantly different from ClrINF (106.3+/-30.4 vs 104.2+/-28.5 ml/min/1.73 m2) and provided a precise (8.8%, 95% CI 6.5-11.1%) and unbiased measure of GFR. Both ClpIVB and ClpINF were positively biased; values exceeded ClrINF by 11.8+/-11.1 (p=0.0001) and 10.5+/-12.5 ml/min/1.73 m2 (p=0.0003), respectively. Use of a nonrenal correction factor of 9.8 and 10.5 ml/min/1.73 m2 for infusion and bolus plasma clearance values, respectively, eliminated bias and improved the precision of these methods. CONCLUSIONS: Iothalamate renal clearance after bolus injection is a simple, accurate, and precise measurement of GFR and may be a useful alternative to the standard infusion method in clinical investigations. The corrected plasma clearance provides a simple index of GFR for clinical practice.     

Determination of drug absorption rate in time-variant disposition by direct deconvolution using beta clearance correction and end-constrained non-parametric regression

A novel numerical deconvolution method is presented that enables the estimation of drug absorption rates under time-variant disposition conditions. The method involves two components. (1) A disposition decomposition-recomposition (DDR) enabling exact changes in the unit impulse response (UIR) to be constructed based on centrally based clearance changes iteratively determined. (2) A non-parametric, end-constrained cubic spline (ECS) input response function estimated by cross-validation. The proposed DDR-ECS method compensates for disposition changes between the test and the reference administrations by using a "beta" clearance correction based on DDR analysis. The representation of the input response by the ECS method takes into consideration the complex absorption process and also ensures physiologically realistic approximations of the response. The stability of the new method to noisy data was evaluated by comprehensive simulations that considered different UIRs, various input functions, clearance changes and a novel scaling of the input function that includes the "flip-flop" absorption phenomena. The simulated input response was also analysed by two other methods and all three methods were compared for their relative performances. The DDR-ECS method provides better estimation of the input profile under significant clearance changes but tends to overestimate the input when there were only small changes in the clearance.