Limited-sampling strategy models for estimating the area under the plasma concentration-time curve for amlodipine

Evaluation of product bioequivalence when subject blood volume necessitates limited sampling strategies

In a traditional blood level bioequivalence (BE) study, every subject provides drug concentrations at each blood sampling time. However, this approach is not suitable for animals whose blood volume limits or prohibits multiple sample collections. In our previous research, we presented an approach that can be applied to studies using a destructive sampling design where each animal provides only 1 blood sample that is then incorporated into a composite profile. Another situation we sometimes face is that of when the animals can contribute more than one sample but are still limited in the number of blood draws (e.g., 3) such that a complete profile per animal is not feasible. Unlike the destructive sampling situation, we cannot combine all blood samples into a single "composite" profile and ignore the correlation of values obtained from the same subject. To avoid the complexities associated with needing to include a covariance component among experimental units into the statistical model, we propose an approach whereby study subjects are randomly assigned to housing unit (e.g., cage or pen) and then randomly assigned to a sampling schedule within each housing unit. In doing so, housing unit rather than the individual subject serves as the experimental unit. This article provides an assessment of this alternative approach to assess product BE when only a limited number of samples can be obtained per study subject.

Limited Sampling Modeling for Estimation of Phenotypic Metrics for CYP Enzymes and the ABCB1 Transporter Using a Cocktail Approach

Plasma concentration data points (n = 2,640) from 16 healthy adults were used to develop and validate limited sampling strategies (LSS) for estimation of phenotypic metrics for CYP enzymes and the ABCB1 transporter, using a cocktail of subtherapeutic doses of the selective probes caffeine (CYP1A2), metoprolol (CYP2D6), midazolam (CYP3A), losartan (CYP2C9), omeprazole (CYP2C19), and fexofenadine (ABCB1). All-subsets linear regression modelling was applied to estimate the AUC_0–12h for caffeine, fexofenadine, and midazolam, and the AUC_0–12h ratio of metoprolol: α-OH metoprolol and omeprazole:5-OH omeprazole. LSS-derived metrics were compared with the parameters’ ‘best estimates’ obtained by non-compartmental analysis using all plasma concentration data points. The correlation coefficient (R²) was used to identify the LSS equations that provided the best fit for n timed plasma samples, and the jack-knife statistics was used as an additional validation procedure for the LSS models. Single time-point LSS models provided R² values greater than 0.95 (R² > 0.95) for the AUC_0–12h ratio of metoprolol:α-OH metoprolol and omeprazole:5-OH omeprazole, whereas 2 time-point models were required for R² > 0.95 for the AUC_0–12h of caffeine, fexofenadine, and midazolam. Increasing the number of sampling points to three led to minor increases in R² and/or the bias or prediction of the estimates. In conclusion, the LSS models provided accurate prediction of phenotypic indices for CYP1A2, CYP2C19, CYP2D6, CYP3A, and ABCB1, when using subtherapeutic doses of selective probes for these enzymes and transporter.

S-Warfarin Limited Sampling Models to Estimate Area Under the Concentration Versus Time Curve for Cytochrome P450 2C9 Baseline Activity and After Induction

BACKGROUND

Phenotyping cytochrome P450 (CYP) 2C9 activity using S-warfarin has routinely required extensive blood sampling over at least 96 hours after dose to estimate the area under the concentration time curve from zero to infinity (AUC). Alternatively, S-warfarin limited sampling models (LSMs) using one or 2 concentration timepoints have been proposed to estimate AUC. This study evaluated whether S-warfarin LSMs accurately estimate CYP2C9 baseline and induction conditions in healthy adults and in advanced-stage cancer patients.

METHODS

Plasma S-warfarin concentrations from healthy adults (n = 92) and in advanced-stage cancer patients (n = 22) were obtained from 6 published studies where a single 10 mg dose of oral warfarin was administered at CYP2C9 baseline and induction conditions. S-warfarin observed AUC was determined by noncompartmental analysis, whereas estimated AUC was calculated from the LSMs. Bias and precision were assessed by percent mean prediction error, percent mean absolute error, and percent root mean square error.

RESULTS

Different results were observed for S-warfarin LSMs in estimating CYP2C9 baseline activity, with most studies resulting in unacceptable bias and precision. The percent mean prediction error, percent mean absolute error, and/or percent root mean square error exceeded acceptable limits for LSMs in patients with advanced-stage cancer and during CYP2C9 induction with lopinavir/ritonavir.

CONCLUSIONS

The differing results during CYP2C9 baseline conditions, as well as unacceptable bias and precision in patients with advanced cancer and during CYP2C9 induction, considerably limit the widespread use of previously published S-warfarin LSMs.

Limited sampling strategy for determining metformin area under the plasma concentration-time curve

AIM

The aim was to develop and validate limited sampling strategy (LSS) models to predict the area under the plasma concentration-time curve (AUC) for metformin.

METHODS

Metformin plasma concentrations (n = 627) at 0-24 h after a single 500 mg dose were used for LSS development, based on all subsets linear regression analysis. The LSS-derived AUC(0,24 h) was compared with the parameter 'best estimate' obtained by non-compartmental analysis using all plasma concentration data points. Correlation between the LSS-derived and the best estimated AUC(0,24 h) (r(2) ), bias and precision of the LSS estimates were quantified. The LSS models were validated in independent cohorts.

RESULTS

A two-point (3 h and 10 h) regression equation with no intercept estimated accurately the individual AUC(0,24 h) in the development cohort: r(2)  = 0.927, bias (mean, 95% CI) -0.5, -2.7-1.8% and precision 6.3, 4.9-7.7%. The accuracy of the two point LSS model was verified in study cohorts of individuals receiving single 500 or 1000 mg (r(2)  = -0.933-0.934) or seven 1000 mg daily doses (r(2)  = 0.918), as well as using data from 16 published studies covering a wide range of metformin doses, demographics, clinical and experimental conditions (r(2)  = 0.976). The LSS model reproduced previously reported results for effects of polymorphisms in OCT2 and MATE1 genes on AUC(0,24 h) and renal clearance of metformin.

CONCLUSIONS

The two point LSS algorithm may be used to assess the systemic exposure to metformin under diverse conditions, with reduced costs of sampling and analysis, and saving time for both subjects and investigators.

Evaluation of partial area under the concentration time curve to estimate midazolam apparent oral clearance for cytochrome P450 3A phenotyping

BACKGROUND

Midazolam apparent oral clearance (CLORAL) is used to estimate intestinal and hepatic cytochrome P450 (CYP) 3A activity. A limited sampling approach was performed to access a midazolam partial area under the concentration time curve (AUC) to estimate CLORAL.

METHODS

Midazolam plasma concentrations from healthy adults were obtained during CYP3A baseline (n=116), inhibition (n=75), and induction or activation (n=66) from seven published studies. Observed CLORAL and partial AUCs of AUC0-2, AUC0-4, AUC0-6, AUC1-2, AUC1-4, AUC2-4, and AUC2-6 were determined by noncompartmental analysis. Subject data were randomly divided into a training set and a validation set. Linear regression equations, derived from partial AUCs, were developed from training set data. Predicted CLORAL was determined from these equations from validation set data. Preset criterion was a coefficient of determination (r2) greater than or equal to 0.9. Bias and precision were evaluated by relative percent mean prediction error (%MPE) and relative percent mean absolute error (%MAE).

RESULTS

During CYP3A baseline conditions, all of the evaluated CLORAL equations had unacceptable r2 (range: 0.34-0.86). During CYP3A inhibition, all of the evaluated CLORAL equations had unacceptable %MAE. Acceptable r2, %MPE, and %MAE were observed during CYP3A induction/activation with AUC0-4 (r2=0.99, %MPE=3.9, %MAE=12.5) and AUC1-4 (r2=0.99, %MPE=6%, %MAE=11.1%). The same equations also predicted the extent of CYP3A induction as a lack of equivalence was observed with AUC0-4 and AUC1-4.

CONCLUSIONS

Midazolam partial AUCs were unable to estimate CYP3A activity during the evaluated baseline and inhibitory conditions. Midazolam CLORAL utilizing a partial AUC0-4 and AUC1-4 was able to estimate CYP3A induction with rifampin and Ginkgo biloba extract.

Limited Sampling Strategy of S-Warfarin Concentrations, but Not Warfarin S/R Ratios, Accurately Predicts S-Warfarin AUC during Baseline and Inhibition in CYP2C9 Extensive Metabolizers

S-warfarin area under the concentration-time curve (AUC(0- infinity )) and clearance are used as measures of cytochrome p450 (CYP) 2C9 activity. In addition, warfarin S/R ratios are used to assess CYP2C9 activity. The determination of S-warfarin AUC(0- infinity ) requires multiple blood samples. Limited sampling strategy (LSS) is a validated technique that estimates AUC(0- infinity ) using limited blood samples. The objective of this study was to evaluate LSS of S-warfarin concentrations and warfarin S/R ratios to predict S-warfarin AUC(0- infinity ) during CYP2C9 baseline activity and inhibition with fluvastatin. Fifty-one healthy subjects, genotyped as CYP2C9 extensive metabolizers, were administered oral warfarin 10 mg. Blood samples were collected over 96 hours. S-warfarin AUC(0- infinity ) equations were derived from a training set of 20 subjects using multiple linear regression. Validation of the equations used data from the remaining 31 subjects. All derived equations were within acceptable limits for measures of bias and precision. Single-point and two-point S-warfarin concentrations, but not warfarin S/R ratios, were predictive of S-warfarin AUC(0- infinity ) during CYP2C9 baseline activity and inhibition. No correlation was observed between CYP2C9(*)1/(*)1 and (*)1/(*)2 genotypes and either S-warfarin concentrations or warfarin S/R ratios. The equation using two-point S-warfarin concentrations at 24 and 48 hours was the most accurate predictor of S-warfarin AUC(0- infinity ). LSS using S-warfarin concentrations is an efficient and accurate technique to evaluate S-warfarin AUC(0- infinity ) when using warfarin as a CYP2C9 probe drug.

Random sparse sampling strategy using stochastic simulation and estimation for a population pharmacokinetic study

The purpose of this study was to use the stochastic simulation and estimation method to evaluate the effects of sample size and the number of samples per individual on the model development and evaluation. The pharmacokinetic parameters and inter- and intra-individual variation were obtained from a population pharmacokinetic model of clinical trials of amlodipine. Stochastic simulation and estimation were performed to evaluate the efficiencies of different sparse sampling scenarios to estimate the compartment model. Simulated data were generated a 1000 times and three candidate models were used to fit the 1000 data sets. Fifty-five kinds of sparse sampling scenarios were investigated and compared. The results showed that, 60 samples with three points and 20 samples with five points are recommended, and the quantitative methodology of stochastic simulation and estimation is valuable for efficiently estimating the compartment model and can be used for other similar model development and evaluation approaches.

Omeprazole limited sampling strategies to predict area under the concentration-time curve ratios: implications for cytochrome P450 2C19 and 3A phenotyping

PURPOSE

To develop a limited sampling strategy (LSS) to predict area under the concentration-time curve (AUC) ratios of omeprazole (AUC(OPZ)) to its metabolites 5-hydroxyomeprazole (AUC(5OH)) and omeprazole sulfone (AUC(SUL)) as phenotyping parameters for cytochrome P450 (CYP) 2C19 and 3A.

METHODS

Data were obtained from 37 (4 women) Caucasian, Chinese, and Korean healthy adults from three published studies. The AUC(OPZ), AUC(5OH), and AUC(SUL) were calculated via noncompartmental analysis. Observed AUC(OPZ, OBS)/AUC(5OH, OBS) and AUC(OPZ, OBS)/AUC(SUL, OBS) were determined. Plasma concentrations of omeprazole, 5-hydroxyomeprazole, and omeprazole sulfone at 1, 1.5, 2, 3, 4, 6, and 8 h post-dose were used to generate limited sampling strategy (LSS) models to predict AUC(OPZ,PRE)/AUC(5OH,PRE) and AUC(OPZ,PRE/)AUC(SUL,PRE). Bias and precision were assessed via percentage mean prediction error (%MPE) and percentage mean absolute error (%MAE), with acceptable limits being <15%.

RESULTS

For CYP2C19, the AUC(OPZ,OBS)/AUC(5OH,OBS) was [mean ± standard deviation (SD)] 2.10 ± 1.63. Five LSS models of AUC(OPZ,PRE)/AUC(5OH,PRE) were generated, but none met the bias or precision criteria. Upon stratification by CYP2C19 genotype and ethnicity, a three-timepoint (at 1, 2, and 4 h) LSS model accurately predicted AUC(OPZ)/AUC(5OH) in Caucasian CYP2C19*1/*1 subjects. For CYP3A, AUC(OPZ,OBS)/AUC(SUL,OBS) (mean ± SD) was 1.79 ± 0.67. All LSS models had unacceptable %MAE, even when stratified by CYP2C19 genotype and ethnicity.

CONCLUSIONS

A LSS model to predict AUC(OPZ)/AUC(5OH), and thus CYP2C19 activity, was generated for Caucasian CYP2C19*1/*1 subjects. However, additional model validation is needed prior to general use. LSS models to predict AUC(OPZ)/AUC(SUL), and thus CYP3A activity, were not possible, even upon stratification by CYP2C19 genotype and ethnicity.

Challenges associated with the evaluation of veterinary product bioequivalence: an AAVPT perspective

The Generic Animal Drug Patent Term Restoration Act (GADPTRA) enacted in 1988 provided the same benefits to animal drug products that were granted to human generic products. It has been over 13 years since the GADPTRA was enacted, and veterinary drug sponsors and regulators have gained enormous insight and experience into some of the unique challenges associated with the determination of product bioequivalence for veterinary dosage forms. Moreover, advances in information and technology have opened both new issues that must be addressed and new mechanisms for demonstrating product bioequivalence. While many aspects of the existing Center for Veterinary Medicine Bioequivalence Guidance continue to provide invaluable guidance to the animal drug industry, there are also aspects of this guidance that are being called into question. Therefore, during the 2001 annual meeting of the American Academy of Veterinary Pharmacology and Therapeutics, participants were asked to address issues and concerns associated with the evaluation of veterinary product bioequivalence. This manuscript provides a summary of the concerns and discussions that transpired.

Limited-sampling strategy models for estimating the pharmacokinetic parameters of 4-methylaminoantipyrine, an active metabolite of dipyrone

Bioanalytical data from a bioequivalence study were used to develop limited-sampling strategy (LSS) models for estimating the area under the plasma concentration versus time curve (AUC) and the peak plasma concentration (Cmax) of 4-methylaminoantipyrine (MAA), an active metabolite of dipyrone. Twelve healthy adult male volunteers received single 600 mg oral doses of dipyrone in two formulations at a 7-day interval in a randomized, crossover protocol. Plasma concentrations of MAA (N = 336), measured by HPLC, were used to develop LSS models. Linear regression analysis and a "jack-knife" validation procedure revealed that the AUC(0-infinity) and the Cmax of MAA can be accurately predicted (R2>0.95, bias <1.5%, precision between 3.1 and 8.3%) by LSS models based on two sampling times. Validation tests indicate that the most informative 2-point LSS models developed for one formulation provide good estimates (R2>0.85) of the AUC(0-infinity) or Cmax for the other formulation. LSS models based on three sampling points (1.5, 4 and 24 h), but using different coefficients for AUC(0-infinity) and Cmax, predicted the individual values of both parameters for the enrolled volunteers (R2>0.88, bias = -0.65 and -0.37%, precision = 4.3 and 7.4%) as well as for plasma concentration data sets generated by simulation (R2>0.88, bias = -1.9 and 8.5%, precision = 5.2 and 8.7%). Bioequivalence assessment of the dipyrone formulations based on the 90% confidence interval of log-transformed AUC(0-infinity) and Cmax provided similar results when either the best-estimated or the LSS-derived metrics were used.

Development and validation of limited-sampling strategies for predicting amoxicillin pharmacokinetic and pharmacodynamic parameters

Amoxicillin plasma concentrations (n = 1,152) obtained from 48 healthy subjects in two bioequivalence studies were used to develop limited-sampling strategy (LSS) models for estimating the area under the concentration-time curve (AUC), the maximum concentration of drug in plasma (C(max)), and the time interval of concentration above MIC susceptibility breakpoints in plasma (T>MIC). Each subject received 500-mg amoxicillin, as reference and test capsules or suspensions, and plasma concentrations were measured by a validated microbiological assay. Linear regression analysis and a "jack-knife" procedure revealed that three-point LSS models accurately estimated (R(2), 0.92; precision, <5.8%) the AUC from 0 h to infinity (AUC(0-infinity)) of amoxicillin for the four formulations tested. Validation tests indicated that a three-point LSS model (1, 2, and 5 h) developed for the reference capsule formulation predicts the following accurately (R(2), 0.94 to 0.99): (i) the individual AUC(0-infinity) for the test capsule formulation in the same subjects, (ii) the individual AUC(0-infinity) for both reference and test suspensions in 24 other subjects, and (iii) the average AUC(0-infinity) following single oral doses (250 to 1,000 mg) of various amoxicillin formulations in 11 previously published studies. A linear regression equation was derived, using the same sampling time points of the LSS model for the AUC(0-infinity), but using different coefficients and intercept, for estimating C(max). Bioequivalence assessments based on LSS-derived AUC(0-infinity)'s and C(max)'s provided results similar to those obtained using the original values for these parameters. Finally, two-point LSS models (R(2) = 0.86 to 0.95) were developed for T>MICs of 0.25 or 2.0 microg/ml, which are representative of microorganisms susceptible and resistant to amoxicillin.