A limited sampling method for estimation of the carboplatin area under the curve

Model-Based Evaluation of Linear Limited and Bayesian Sparse Sampling for Therapeutic Monitoring of Recombinant Coagulation Factor IX

Dosing of coagulation factor products is mainly determined based on a patient's body weight; however, several studies have reported high interindividual variability in their pharmacokinetics (PK). The objective of this study was to develop and evaluate 2 sparse sampling methods for the estimation of AUC of recombinant factor IX (BeneFIX) as proof of concept for dose individualization. A population pharmacokinetic model was used to generate the plasma factor IX activity-versus-time data. The linear limited sampling model (LLSM) was developed based on the correlation of factor IX activity versus AUC_0-72 hours following screening of several blood sampling times in adolescent and adult subjects (n = 90 subjects). Factor IX trough concentrations were predicted from a relationship established from AUC versus factor IX activity measured 72 hours postdosing. Using the best selected sampling time, the LLSM and Bayesian model were validated in separate data sets (n = 75 subjects). Using the LLSM and Bayesian analysis, a blood sample at 24 hours predicted AUC with bias and root mean square error < 5% and < 15%, respectively. The predicted trough concentrations were ≥1 IU/dL in 99% and 100% of subjects by the LLSM and Bayesian model, respectively. The average factor IX dose for a target AUC of 800 IU·h/dL was 61, 60, and 63 IU/kg using the extensive (reference), LLSM and Bayesian model, respectively. Overall, the AUC, trough concentrations and individualized dosing of recombinant factor IX could be reasonably predicted using the LLSM and Bayesian model.

Performing and interpreting individual pharmacokinetic profiles in patients with Hemophilia A or B: Rationale and general considerations

OBJECTIVES

In a separate document, we have provided specific guidance on performing individual pharmacokinetic (PK) studies using limited samples in persons with hemophilia with the goal to optimize prophylaxis with clotting factor concentrates. This paper, intended for clinicians, aims to describe how to interpret and apply PK properties obtained in persons with hemophilia.

METHODS

The members of the Working Party on population PK (PopPK) of the ISTH SSC Subcommittee on Factor VIII and IX and rare bleeding disorders, together with additional hemophilia and PK experts, completed a survey and ranking exercise whereby key areas of interest in the field were identified. The group had regular web conferences to refine the manuscript's scope and structure, taking into account comments from the external feedback to the earlier document.

RESULTS

Many clinical decisions in hemophilia are based on some form of explicit or implicit PK assessment. Individual patient PK profiles can be analyzed through traditional or PopPK methods, with the latter providing the advantage of fewer samples needing to be collected on any prophylaxis regimen, and without the need the for a washout period. The most useful presentation of PK results for clinical decision making are a curve of the factor activity level over time, the time to achieve a certain activity level, or related parameters like half-life or exposure (AUC). Software platforms have been developed to deliver this information to clinicians at the point of care. Key characteristics of studies measuring average PK parameters were reviewed, outlining what makes a credible head-to-head comparison among different concentrates. Large data collections of PK and treatment outcomes currently ongoing will advance care in the future.

CONCLUSIONS

Traditionally used to compare different concentrates, PK can support tailoring of hemophilia treatment by individual profiling, which is greatly simplified by adopting a PopPK/Bayesian method and limited sampling protocol.

Basophil Activation Test is a Relevant Biomarker of the Outcome of Rapid Desensitization in Platinum Compounds-Allergy

BACKGROUND

Rapid drug desensitization (RDD) has become a cornerstone in the management of immediate drug hypersensitivity reactions (DHRs) to chemotherapeutic agents. Because of the inherent risk of anaphylaxis during RDD, biomarkers to predict patients at risk of developing such severe reactions are needed. The basophil activation test (BAT) has been used in DHRs as a diagnostic tool.

OBJECTIVE

We evaluated basophil CD63 and CD203c expression (BAT) as a biomarker to assess the safety and effectiveness of RDD in platinum compounds-allergic patients.

METHODS

Patients allergic to platinum compounds (n = 15) undergoing RDD were assessed through clinical history, skin testing, serum tryptase levels, and BAT. BAT was performed immediately before RDD, assessing CD203c and CD63 expression on basophils. BAT was also performed in 6 patients tolerant to platinum compounds and in 6 healthy volunteers.

RESULTS

BAT was positive to CD203c or CD63 in 11 out of 15 patients allergic to platinum compounds (73%), with increased expression of CD203c and CD63 in 11 (73%) and 6 (40%) patients, respectively. Increased CD63 expression tended to be associated with more severe initial reactions. All controls had negative test results. Reactions during RDD were associated with BAT positivity and increased tryptase levels. Only 1 of 4 patients with negative BAT had a mild reaction during RDD. BAT remained positive in multiple sequential RDD.

CONCLUSIONS

BAT identified patients allergic to platinum compounds with an increased risk of reactions during desensitization and higher CD63 expression was observed in severe reactions. Multiple RDDs to platinum compounds did not induce persistent hyporesponsiveness on basophils. BAT is a potential biomarker for RDD.

Therapeutic drug monitoring of cancer chemotherapy

Therapeutic drug monitoring is not routinely used for chemotherapy agents. There are Several reasons, but one major drawback is the lack of established therapeutic Concentration ranges. Combination chemotherapy makes the establishment of Therapeutic ranges for individual drugs difficult, the concentration-effect relationship for a single drug may not be the same as when that drug is used in a drug combination. Pharmacokinetic optimization protocols for many classes of cytotoxic compounds exist in specialized centers, and some of these protocols are now part of large multicentre trials. Nonetheless, TDM clearly has the potential to improve the clinical use of chemotherapy gents, most of which have very narrow therapeutic indices and highly variable pharmacokinetics. A substantial body of literature accumulating during the past 15 years demonstrates relationships between systemic exposure to various chemotherapy agents and their toxic or therapeutic effects. This article reviews TDM concepts in addition to tools based on pharmacokinetic modeling of chemotherapy agents. The administered dose of chemotherapy agents is sometimes adjusted individually using either a priori or a posteriori methods. These models can only be applied by using the same dose and schedule as the original study. Bayesian estimation offers more flexibility in blood sampling times and, owing to its precision and to the amount of information provided is the method of choice for ensuring that a given patient benefits from the desired systemic exposure. Moreover, the role and application of Pharmacogenetics as a tool for individualizing chemotherapy is discussed highlighting the agents and mechanisms that have been well studied and defined and their relevance to clinical practice. Finally, this paper address issues critical to the optimal use of TDM in a clinical setting, and the role of clinical pharmacist in this regard. In addition, it discusses future developments in this field that can contribute to improving cancer chemotherapy In terms of patient outcome and survival. J Oncol Pharm Practice (2007) 13: 207—221.

Therapeutic drug monitoring in cancer chemotherapy

For a select number of drugs, proper management of patients includes monitoring serum or plasma concentrations of the drugs and adjusting the doses accordingly - this practice is referred to as therapeutic drug monitoring (TDM). The need for TDM arises when pharmacokinetic variability of drugs is not easily accounted for by common clinical parameters. Many chemotherapeutic drugs have large interindividual variability, yet TDM is not commonplace in chemotherapy management. This review will discuss pharmacokinetics in the context of chemotherapeutic drugs, examine the few instances where TDM is currently used in the field of oncology and propose other drugs where TDM might be useful for dose adjustments in the management of chemotherapy.

A systematic review of limited sampling strategies for platinum agents used in cancer chemotherapy

Despite evidence in the literature suggesting that a strong correlation exists between the pharmacokinetic parameters and pharmacodynamic effect of anticancer agents, many of these agents are still dosed by body surface area. Therapeutic drug monitoring with the aim of pharmacokinetic-guided dosing would not only maintain target concentrations associated with efficacy but may potentially minimise the likelihood of dose-related systemic toxicities. The pharmacokinetic parameter that displays the best correlation with the pharmacodynamics of anticancer drugs is the area under the plasma concentration-time curve (AUC). However, accurate determination of the AUC requires numerous blood samples over an extended interval, which is not feasible in clinical practice. Therefore, limited sampling strategies (LSSs) have been proposed as a means to accurately and precisely estimate pharmacokinetic parameters with a minimal number of blood samples. LSSs have been developed for many drugs, particularly ciclosporin and other immunosuppressants, as well as for certain anticancer drugs. This systematic review evaluates LSSs developed for the platinum compounds and categorises 18 pertinent citations according to criteria adapted from the US Preventive Services Task Force. Thirteen citations (four level I, six level II-1, three level II-2) pertained to LSSs for carboplatin, four citations (one level II-1, one level II-2, two level III) to cisplatin LSSs, and one citation (level II-2) to nedaplatin. Based on the current evidence, it appears that LSSs may be useful for pharmacokinetic-guided dosage adjustments of carboplatin in both adults and children with cancer. Although some validation studies suggest that LSSs can be extended to different cancer populations or different chemotherapy regimens, other studies dispute this finding. Although the use of LSSs to predict the pharmacokinetic parameters of cisplatin and nedaplatin appear promising, the quality of evidence from published studies does not support routine implementation at this time.LSSs represent one approach in which clinicians can make specific dosage adjustments for individual patients to optimise outcomes. However, the limitations of these strategies must also be taken into consideration. There is also a need for prospective studies to demonstrate that application of LSSs for platinum agents ultimately improves patient response and decreases systemic toxicities.

Sparse Sampling Design for Therapeutic Drug Monitoring of Sequentially Administered Cyclophosphamide, Thiotepa, and Carboplatin (CTC)

The alkylating agents cyclophosphamide, thiotepa, and carboplatin (CTC) are administered simultaneously in high-dose chemotherapy regimens. This regimen is sometimes complicated by severe organ toxicities, which may be caused by interindividual variability in the pharmacokinetics of the agents. Monitoring plasma levels and adapting doses may reduce variability in exposure to the compounds and their metabolites. The aim of this study was to develop and validate a sparse sampling design for routine dose individualization of cyclophosphamide, thiotepa, and carboplatin both during and between courses in the CTC regimen. Models describing the population pharmacokinetics of the prodrug cyclophosphamide (4000 or 6000 mg/m) and its activated metabolite 4-hydroxycylophosphamide, thiotepa (320 or 480 mg/m), and its equipotent metabolite tepa, and carboplatin (1067 or 1600 mg/m) in the 4-day CTC regimen have been developed previously using the program NONMEM. Based on these models, plasma concentrations were calculated in 20 groups of 50 simulated patients in each group during multiple courses of therapy, and the exposure, expressed as area under the plasma concentration-versus-time curve (AUC), was calculated. Subsequently, individual model-predicted AUCs were calculated for all courses, based on selected simulated plasma concentrations during the first course of therapy. Strategies were compared by assessment of their predictive performance of the AUC and their applicability in clinical practice. Withdrawal of 3 samples on the first day of the course at 190, 290, and 400 minutes after start of cyclophosphamide infusion resulted in unbiased and precise first course AUC predictions of 4-hydroxycylophosphamide, thiotepa and tepa, and carboplatin (precision [root mean squared relative prediction error, %RMSE] 20%, 16%, 8.8%, respectively). Applying this same strategy at day 3 (or 4) of the course, with an additional sample at 600 minutes on both days, resulted in unbiased and precise predictions of the AUC of a following course (%RMSE 21%, 18%, 17%, respectively). Prospective validation of the strategies in 23 additional patients yielded comparable results. It can be concluded that a good and useful sparse sampling design was developed for precise and accurate estimation of the AUCs of 4-hydroxycyclophosphamide, thiotepa and tepa, and carboplatin in the CTC regimen. This method is valuable in pharmacokinetically guided dose adaptation both during and between CTC courses.

Development of an optimal sampling strategy for clinical pharmacokinetic studies of the novel anthracycline disaccharide analogue MEN-10755

AIM

MEN-10755 is a novel anthracycline analogue that has shown an improved therapeutic efficacy over doxorubicin in animal models, especially in gynaecological and lung cancers and is currently under clinical development for the treatment of solid tumours. The aim of the project was to develop an optimal sampling strategy for MEN-10755 to provide an efficient basis for future pharmacokinetic/pharmacodynamic investigations.

METHODS

Data from 24 patients who participated in a phase I clinical pharmacokinetic study of MEN-10755 administered as a short i.v. infusion were included. Individual pharmacokinetic values were calculated by fitting the plasma concentration data to a two-compartment model using nonlinear least-squared regression (KINFIT, Ed 3.5). Population pharmacokinetic analysis was carried out using (a) the traditional standard two-stage method (STS) based on all data (KINFIT-ALL), (b) the iterative two-stage Bayesian (IT(2)B) population modelling algorithm (KINPOP), and (c) the STS method using KINFIT and using four optimally timed plasma concentrations (KINFIT-OSS4). Determinant (D) optimal sampling strategy (OSS) was used to evaluate the four most information-rich sampling times. The pharmacokinetic parameters V(c) (l), k(el) (h(-1)), k(12) (h(-1)) and k(21) (h(-1)) calculated using KINPOP served as a model for calculation of four D-optimal sampling times. D-optimal sampling data sets were analysed using KINFIT-OSS4 and compared with the population model obtained by the traditional standard two-stage approach for all data sets (KINFIT-ALL).

RESULTS

The optimal sampling times were: the end of the infusion, and 1.5 h, 3.8 h and 24 h after the start of the infusion. The four-point D-optimal sampling design determined in this study gave individual parameter estimates close to the basic standard estimates using the full data set.

CONCLUSION

Because accurate estimates of pharmacokinetic parameters were achieved, the four-point D-optimal sampling design may be very useful in future studies with MEN-10755.

Pharmacokinetics of high-dose chemotherapy

There is considerable variation in the severity of preparative regimen-related toxicity (RRT) in hematopoietic stem-cell transplantation (HSCT). This variation has been recognized to be due, in part, to the wide variation in the pharmacokinetics (PK) of high-dose chemotherapy (HDC). Consequently, therapeutic drug modeling and pharmacokinetic-directed therapy (PKDT) represents an attractive strategy in this setting. Advances in our understanding of drug metabolism, the nature of the active metabolites, and the ability to measure drug concentrations have led to the point where for some agents it is now possible to treat to a given PK end point with a great deal of reliability. In-depth knowledge of the PK and pharmacodynamics (PD) associations of the agents employed in the high-dose setting will make possible more efficient research into preparative regimen dosing intensity and comparisons of different preparative regimens as well as safer HSCT overall. In this review, we discuss PK and PD studies of high-dose cyclosphamide, melphalan, thiotepa, carmustine, cisplatin, carboplatin, paclitaxel, docetaxel, and busulfan.

Dosing of medications in morbidly obese patients in the intensive care unit setting

OBJECTIVE

To derive recommendations for the dosing of commonly used medications in the morbidly obese patient in the ICU.

DATA SOURCES

Articles were obtained through computerized searches involving MEDLINE. The bibliographies of retrieved publications and textbooks were reviewed for additional references.

STUDY SELECTION

All studies involving the pharmacokinetics or pharmacodynamics of medications in obese subjects or patients.

DATA EXTRACTION

The emphasis was on studies involving morbidly obese patients but, in the absence of such data, investigations involving lesser forms of obesity were extracted.

DATA SYNTHESIS

There is a paucity of data upon which to make recommendations for dosing commonly used medications in the morbidly obese patient in the ICU, although recommendations were provided based on the available information.

CONCLUSIONS

There is clearly a need for more investigations involving dosing regimens of medications in the morbidly obese population. Until such studies are available, the clinician must try to derive the best dosing regimens for medications based on the limited pharmacokinetic data available for some agents and clinical judgement.

Optimal sampling strategies to assess inulin clearance in children by the inulin single-injection method

BACKGROUND

Glomerular filtration rate in patients can be determined by estimating the plasma clearance of inulin with the single-injection method. In this method, a single bolus injection of inulin is administered and several blood samples are collected. For practical and convenient application of this method in children, it is important that a minimal number of samples are drawn. The aim of this study was to develop and validate sampling strategies with fewer samples for reliable prediction of inulin clearance in pediatric patients by the inulin single-bolus-injection method.

METHODS

Complete inulin plasma concentration-time curves of 154 patients were divided into an index (n = 100) and a validation set (n = 54). A population pharmacokinetic model was developed for the index set. Optimal sampling times were selected based on D-optimality theory. For the validation set, Bayesian estimates of clearance were generated using the derived population parameters and concentrations at two to four sampling times. Bayesian estimates of clearance were compared with the individual reference values of clearance.

RESULTS

The strategies with samples taken at 10/30/90/240 min, 10/30/240 min, 10/90/240 min, 30/90/240 min, and 90/240 min allowed accurate prediction of inulin clearance (bias <3% and not significantly different from 0; imprecision <15%).

CONCLUSIONS

Strategies involving two to four samples, including a sample at 240 min after administration of inulin, in the inulin single-injection method allow accurate prediction of inulin clearance in pediatric patients. Even one blood sample at 240 min showed acceptable performance. The proposed strategies are practical and convenient to children, and reduce repetitive blood sampling without compromising accuracy.

Pharmacokinetic-pharmacodynamic guided trial design in oncology

The application of pharmacokinetic (PK) and pharmacodynamic (PD) modeling in drug development has emerged during the past decades and it is has been suggested that the investigation of PK-PD relationships during drug development may facilitate and optimize the design of subsequent clinical development. Especially in oncology, well designed PK-PD modeling could be extremely useful as anticancer agents usually have a very narrow therapeutic index. This paper describes the application of the current insights in the use of PK-PD modeling to the design of clinical trials in oncology. The application of PK-PD modeling in each separate stage of (pre)clinical drug development of anticancer agents is discussed. The implementation of this approach is illustrated with the clinical development of docetaxel.

A Bayesian approach for the estimation of pharmacokinetic parameters in children

A Bayesian method is proposed for the first-time assessment of pharmacokinetic parameters (area under the curve [AUC], maximum concentration [C(max)], half-life [T(1/2)]) in children after a single oral dose. Pharmacokinetic parameters were compared using sparse sampling (three blood samples) with extensive sampling (nine blood samples). Two different therapeutic classes of drugs were evaluated in this study. The population pharmacokinetic parameters and the blood sampling time points were obtained from adult data. The results indicate that the Bayesian approach can be used to estimate pharmacokinetic parameters in children with fair degree of accuracy. The Bayesian method described here can be used to assess pharmacokinetic parameters in children, provided there is a prior knowledge of the pharmacokinetics of a drug in adult population.

Development and validation of limited sampling strategies for prediction of the systemic exposure to the novel anticancer agent E7070 (N-(3-chloro-7-indolyl)-1,4-benzenedisulphonamide)

AIMS

E7070 is a novel, sulphonamide anticancer agent currently under clinical development for the treatment of solid tumours. The aim of this study was to develop and validate limited sampling strategies for the prediction of E7070 exposure in two different treatment schedules for phase II studies using the Bayesian estimation approach.

METHODS

Data from two phase I dose finding studies were used in which E7070 was administered either as a single 1 h infusion or as a daily 1 h infusion for 5 days. Plasma concentration-time data from 75 patients were randomly divided into an index data set, used for the development of the strategies, and a validation data set. Population pharmacokinetic parameters were derived on the basis of the index data set. The D-optimality algorithm was used for the selection of optimal time points for both treatment schedules. The developed strategies were compared by assessment of their predictive performance of exposure, expressed as AUC (area under the plasma concentration vs time curve), in the validation data set.

RESULTS

The developed population pharmacokinetic model comprised three compartments, with saturable distribution to one peripheral compartment and both linear and saturable elimination from the central compartment. For the 1 h infusion, a four sample strategy was selected which resulted in unbiased and accurate predictions of AUC (bias 0.74%, precision 13%). A five sample strategy was generated for the daily times five schedule yielding unbiased (bias 3.2%) and precise (12% precision) predictions of AUC.

CONCLUSIONS

Optimal sampling strategies were developed and validated for estimation of E7070 exposure in two different treatment schedules. Both schedules enabled accurate and unbiased predictions of AUC.

Application of pharmacokinetic modelling to the routine therapeutic drug monitoring of anticancer drugs

Over the last 10 years, proofs of the clinical interest of therapeutic drug monitoring (TDM) of certain anticancer drugs have been established. Numerous studies have shown that TDM is an efficient tool for controlling the toxicity of therapeutic drugs, and a few trials have even demonstrated that it can improve their efficacy. This article critically reviews TDM tools based on pharmacokinetic modelling of anticancer drugs. The administered dose of anticancer drugs is sometimes adjusted individually using either a priori or a posteriori methods. The most frequent clinical application of a priori formulae concerns carboplatin and allows the computation of the first dose based on biometrical and biological data such as weight, age, gender, creatinine clearance and glomerular filtration rate. A posteriori methods use drug plasma concentrations to adjust the subsequent dose(s). Thus, nomograms allowing dose adjustment on the basis of blood concentration are routinely used for 5-fluorouracil given as long continuous infusions. Multilinear regression models have been developed, for example for etoposide, doxorubicin. carboplatin, cyclophosphamide and irinotecan, to predict a single exposure variable [such as area under concentration-time curve (AUC)] from a small number of plasma concentrations obtained at predetermined times after a standard dose. These models can only be applied by using the same dose and schedule as the original study. Bayesian estimation offers more flexibility in blood sampling times and, owing to its precision and to the amount of information provided, is the method of choice for ensuring that a given patient benefits from the desired systemic exposure. Unlike the other a posteriori methods, Bayesian estimation is based on population pharmacokinetic studies and can take into account the effects of different individual factors on the pharmacokinetics of the drug. Bayesian estimators have been used to determine maximum tolerated systemic exposure thresholds (e.g. for topotecan or teniposide) as well as for the routine monitoring of drugs characterized by a very high interindividual pharmacokinetic variability such as methotrexate or carboplatin. The development of these methods has contributed to improving cancer chemotherapy in terms of patient outcome and survival and should be pursued.

Pharmacokinetically guided administration of chemotherapeutic agents

The current practice for the dose calculation of most anticancer agents is based on body surface area in m2, although lower interpatient variation in pharmacokinetic parameters has been reported with pharmacokinetically guided administration. As chemotherapeutic agents have a narrow therapeutic window, pharmacokinetically guided administration may lead to less toxicity and higher efficacy than administration on the basis of body surface area. Pharmacokinetically guided administration, using parameters such as area under the plasma concentration-time curve (AUC), steady-state plasma drug concentration and drug exposure time above a certain plasma concentration, has been studied for many antineoplastic agents. Assessment of pharmacokinetic profiles allows the characterisation of relationships between pharmacokinetic parameters and efficacy and toxicity. AUC appears to be more closely correlated with pharmacodynamics than does the dose per unit of body surface area. In particular, the AUC-guided administration of carboplatin has been extensively studied, based on the close relationship between the renal clearance of the drug and glomerular filtration rate. Several formulae and limited sampling models have been derived to predict the AUC of carboplatin. The relationship between AUC and pharmacodynamics has also been studied for other anticancer agents, for example fluorouracil, topotecan, etoposide, cisplatin and busulfan, but all less extensively than for carboplatin. The pharmacokinetically guided administration of these agents needs to be investigated further before the use of alternative administration formulae can become standard clinical practice. Prospective studies of pharmacokinetically guided versus surface area-based administration should be performed to validate pharmacokinetic-pharmacodynamic relationships and to facilitate optimal dosage of anticancer agents in the clinic.

Limited sampling model for the estimation of pharmacokinetic parameters in children

A limited sampling model (LSM) is proposed for the first-time assessment of pharmacokinetic parameters (area under the concentration-time curve (AUC), Cmax, and T1/2) in children after a single oral dose of drug. Three drugs were evaluated in this study. The LSM was developed for each drug from the data of 10 healthy adult volunteers. The relationship at selected time points between plasma concentration and the AUC or Cmax was evaluated by multiple linear regression. The multiple linear regression that gave the best correlation coefficient (r) for 3 sampling times versus AUC or Cmax was chosen as the LSM. Pharmacokinetic parameters generated using sparse sampling (3 blood samples) were compared with pharmacokinetic parameters generated using extensive sampling (>7 blood samples). The results indicated that a limited sampling model can be developed from adult data to estimate pharmacokinetic parameters in children with fair degree of accuracy.

The clinical pharmacology of alkylating agents in high-dose chemotherapy

Alkylating agents are widely used in high-dose chemotherapy regimens in combination with hematological support. Knowledge about the pharmacokinetics and pharmacodynamics of these agents administered in high doses is critical for the safe and efficient use of these regimens. The aim of this review is to summarize the clinical pharmacology of the alkylating agents (including the platinum compounds) in high-dose chemotherapy. Differences between conventional and high doses will be discussed.

Adaptive control methods for the dose individualisation of anticancer agents

Numerous studies have found a clear relationship between systemic exposure and the toxicity or (more rarely) the efficacy of anticancer agents. Moreover, the clearance of most of these drugs differs widely between patients. These findings, combined with the narrow therapeutic index of anticancer drugs, suggest that patient outcome would be improved if doses were individualised to achieve a target systemic exposure. Bayesian maximum a posteriori probability (MAP) forecasting is an efficient and robust method for the optimisation of drug therapy, but its use for anticancer drugs is not yet extensive. The aim of this paper is to review the application of population pharmacokinetics and MAP to anticancer drugs and to evaluate whether and when MAP Bayesian estimation improves the clinical benefit of anticancer chemotherapy. For each drug, the relationships between pharmacokinetic variables [e.g. plasma concentration or the area under the concentration-time curve] and pharmacodynamic effects are described. Secondly, the methodologies employed are considered and, finally, the results are analysed in terms of predictive performance as well as, where possible, the impact on clinical end-points. Some studies were retrospective and intended only to evaluate individual pharmacokinetic parameter values using very few blood samples. Among the prospective trials, a few studied the pharmacokinetic/pharmacodynamic relationships which provided the basis for routine pharmacokinetic monitoring. Others were performed in clinical context where MAP Bayesian estimation was used to determine maximum tolerated systemic exposure (e.g. for carboplatin, topotecan, teniposide) or for pharmacokinetic monitoring (e.g. for methotrexate or platinum compounds). Indeed, its flexibility in blood sampling times makes this technique much more applicable than other limited sampling strategies. These examples demonstrate that individual dose adjustment helps manage toxicity. The performance of pharmacokinetic monitoring is linked to the methodology used at each step of its design and application. Moreover, a limitation to the use of pharmacokinetic monitoring for certain anticancer drugs has been the difficulty in obtaining pharmacokinetic or pharmacodynamic data. Recent progress in analytical methods, as well as the development of noninvasive methods (such as positron emission tomography) for evaluating the effects of chemotherapy, will help to define pharmacokinetic-pharmacodynamic relationships. Bayesian estimation is the strategy of choice for performing pharmacokinetic studies, as well as ensuring that a given patient benefits from the desired systemic exposure. Together, these methods could contribute to improving cancer chemotherapy in terms of patient outcome and survival.

A limited sampling strategy for determining carboplatin AUC and monitoring drug dosage

There are several convincing reports showing relationships between the area under the curve of ultrafilterable concentration versus time (AUC) and pharmacodynamics of carboplatin. It is advisable, in treated patients, to check the AUC that is effectively delivered as compared with the prescribed AUC. To this end, limited sampling strategy seems to be an adequate approach since it limits the constraints of repeated blood sampling for both patients and nursing staff. A flexible limited sampling method for assessing ultrafilterable carboplatin AUC was developed. This method was based on a Bayesian estimation of carboplatin clearance using the NON linear Mixed Effect Model (NONMEN) program and a large pharmacokinetic and covariates database (103 patients). The optimal sampling design was a two-sample schedule (1 and 4 h after the end of infusion). During a prospective evaluation, it allowed an adequate estimation of carboplatin clearance with a non-significant bias (-4.5%) and a good precision (9%). In a second stage, this method was clinically applied to monitor carboplatin AUC in a group of 5 patients with metastatic germ cell tumours treated with intensified high dose carboplatin-based chemotherapy for 4 days. Dosage adjustments were performed according to daily controls of their AUC in order to obtain a total AUC of 20 mg/ml x min. By using this strategy all patients effectively received a total AUC very close to this targeted AUC, thus proving the clinical usefulness of this limited sampling method.

A comparison of methods of calculation for estimating carboplatin AUC with a retrospective pharmacokinetic-pharmacodynamic analysis in patients with advanced non-small cell lung cancer. European Lung Cancer Working Party

We retrospectively analysed the data obtained in a large randomised trial performed in 505 eligible patients with advanced non-small cell lung cancer. Its purpose had been to compare a combination of carboplatin (200 mg/m2) and cisplatin (60 mg/m2) with or without the addition of ifosfamide. The present retrospective analysis assessed two ways of dosing carboplatin: according to body surface area (mg/m2) or to the estimated targeted area under the concentration versus time curve (AUC). Two different methods were used in the latter calculation: the Calvert formula using the Cockroft approximation to evaluate the glomerular filtration rate and the Chatelut equation. There was an excellent linear correlation between them. With the Chatelut method, the calculated administered AUC were lower. Whichever method was used, carboplatin AUC was not significantly associated with antitumour response rate nor patient survival. A statistically significant increase in haematological toxicity, mainly thrombopenia, was observed with an increase in the AUC. This effect was observed whatever AUC variable was considered, i.e. total dosage at course one, total dosage during the first three chemotherapy courses or dose intensity during the first three courses. The effect remained highly significant after adjustment for treatment arm. We conclude that for a moderate carboplatin dose in non-small cell lung cancer, the therapeutic index could be improved if dosage is calculated according to the AUC.

Development of a limited sampling approach in pharmacokinetic studies: experience with the antiepilepsy drug tiagabine

A sparse sampling method is proposed to assess pharmacokinetic parameters after a single dose of the antiepilepsy drug tiagabine. Pharmacokinetic parameters obtained from two different pharmacokinetic studies were compared using sparse sampling (7 blood samples) with extensive sampling (15 to 16 blood samples). The results indicated that sparse blood samples taken at appropriate times can be used to estimate pharmacokinetic parameters as accurately as extensive blood samples. In addition, a limited sampling model (LSM) was developed using samples from 10 subjects at two time points (6 and 8 hours). The model was validated in 40 subjects and provided good population mean estimates of area under the concentration-time curve (AUC) and maximum concentration (Cmax). The sparse sampling method described here can be used to assess pharmacokinetic parameters in drug development provided a prior knowledge of the pharmacokinetics of a drug has been obtained from extensive sampling. Further, the LSM described here may be useful in estimating AUC and Cmax of tiagabine using two samples in clinical settings. The LSM approach described here can also be used to estimate AUC and Cmax of a drug in preclinical toxicokinetic studies without detailed pharmacokinetic studies.

A limited sampling method for the estimation of AUC and Cmax of carbamazepine and carbamazepine epoxide following a single and multiple dose of a sustained-release product

AIMS

The objectives of this study are to develop a model to predict area under the curve (AUC) and maximum plasma concentration (Cmax) of carbamazepine (CBZ) and its active metabolite carbamazepine epoxide (CBZE) following single and multiple dose of CBZ using one or two samples in volunteers.

METHODS

Limited sampling models (LSM) were developed for CBZ and CBZE following 200-800 mg single oral dose and 400-800 mg twice daily dose for 14 days of a sustained-release product (CBZ-SR) to estimate AUC and Cmax. The LSM was developed from a training data set of 15 subjects using one blood sample taken at 48 h following a single dose. The model was validated on 60 subjects who received different doses of CBZ. Following multiple dosing, the LSM was developed from a training data set of 10 subjects using the steady state Cmin (plasma concentration obtained 5 min before the last CBZ-SR dose).

RESULTS

The model provided good estimates of AUC and Cmax for CBZ and CBZE. The bias and the precision of the predicted AUC and Cmax for CBZ and CBZE were less than 10% and 15%, respectively. Similar results were obtained when CBZ was given as multiple dose.

CONCLUSIONS

The method described here may be used to estimate AUC and Cmax for CBZ and CBZE without detailed pharmacokinetic studies following single or multiple dose of CBZ.

A limited sampling approach in bioequivalence studies

A limited sampling model (LSM) has been developed for an antidepressant immediate-release product (Drug A) and an antiepileptic controlled release product (Drug B) to predict the area under the curve (AUC) and the maximum plasma concentration (Cmax) and to compare the bioequivalence of two formulations of each drug using predicted versus observed AUC and Cmax after a single oral dose. The LSM for drug A was developed using data from 10 healthy people. The correlation between plasma concentration (independent variable) at selected time points with the AUC or Cmax (dependent variable) was evaluated by simple regression analysis. The linear regression that gave the best correlation coefficient (r) for a single sampling time versus AUC or Cmax was chosen as the LSM. The model provided good estimates of AUC and Cmax for drug A. The 90% confidence interval on log transformed observed and predicted AUC and Cmax were as follows: AUC observed = 100% to 118%, AUC predicted = 101% to 117%, Cmax observed = 99% to 125%, and Cmax predicted = 100% to 131%. The LSM for drug B was developed using a similar approach to drug A. The 90% confidence interval on log transformed observed and predicted AUC and Cmax were: AUC observed = 99% to 110%, AUC predicted = 99% to 118%, Cmax observed = 107% to 120%, and Cmax predicted = 99% to 111%. Although the predicted Cmax did not meet the 90% confidence interval for drug A, the method described here may be used to estimate AUC and Cmax for a drug in bioequivalence studies without detailed blood sampling. More research is needed in this direction.

Pharmacokinetic study of carboplatin given on a 5-day intravenous schedule

We investigated whether carboplatin pharmacokinetics is altered when the drug is delivered daily over 5 days, compared to a single-day infusion. Carboplatin was infused in 11 patients with lung cancer, who were randomly assigned to 2 groups. In the first group, the agent was administered on a conventional single-day schedule in the first course and then on a 5-day schedule in the second course. In the second group, the order was reversed (crossover design). The dose was calculated using Calvert's formula with 24 h creatinine clearance (Ccr, ml/min) as a substitute for glomerular filtration rate (GFR): carboplatin (mg) = AUCx(Ccr+25), where AUC denotes the area under the concentration versus time curve (mg ml-1 min). No difference of carboplatin clearance between the single-day and 5-day schedule was observed (94.8 +/- 19.9 versus 96.1 +/- 29.9 ml/min, P = 0.818, paired t test). The formula systematically overestimated the carboplatin clearance: the ratio of estimated clearance/ observed clearance ranged from 1.01 to 1.58 (median 1.28; 95% confidence interval, 1.18 to 1.39). We concluded that the individual dosing strategy based on renal function can be applied with a 5-day schedule as well as a single-day schedule. Carboplatin is overdosed when Ccr is substituted for GFR in Calvert's formula.

A comparative computer simulation study of three different sparse-sampling methods for the estimation of steady-state area under the concentration-time curve (AUC) and maximum concentration (Cmax) in toxicokinetics

A limited-sampling method is proposed to estimate the area under the curve (AUC) of concentration versus time and maximum concentration (Cmax) following single or multiple oral doses of a hypothetical drug. The plasma concentration versus time data sets for 50 animals were generated by simulation. The limited-sampling model (LSM) was developed with samples from 10 animals at a single time point. The model was validated in another 40 animals who received either a 500-mg single dose or multiple doses orally. The model provided good population mean estimates of AUC and Cmax. The proposed method was compared with the existing two methods; they are, naive sampling (five time points) and optimal sampling (three time points). The method described here may be useful in estimating AUC and Cmax with one or two samples in toxicokinetic studies following single or multiple oral dosing without detailed pharmacokinetic studies.

A limited sampling method for the estimation of vigabatrin maximum plasma concentration and area under the curve

A limited sampling model (LSM) was developed to estimate the area under the curve (AUC) and maximum plasma concentration (Cmax) for a 1-g oral dose of vigabatrin. The model was developed using the data from 10 healthy subjects and one time point. The following equations describe the model for AUC and Cmax: AUC(predicted) = 5.4 x C3h + 70 and Cmax(predicted) = 0.18 x AUC(0-infinity) + 9.4. The model was validated in 49 subjects who orally received 1-g vigabatrin. This LSM was also used to predict AUC and Cmax volunteers who received 2- and 4-g vigabatrin doses and in renal failure patients who were given a 0.75-g dose. The model provided good estimates of both AUC and Cmax in all groups of subjects except renal dysfunction patients. The method described here may be used to estimate AUC and Cmax of vigabatrin without detailed pharmacokinetic studies.

Pharmacokinetics and pharmacodynamics of carboplatin administered in a high-dose combination regimen with thiotepa, cyclophosphamide and peripheral stem cell support

The aim of this pharmacokinetic/pharmacodynamic study was to define the relationships of the carboplatin exposure with the toxicity in patients treated with high dose carboplatin (400 mg m-2 day-1), cyclophosphamide (1500 mg m-2 day-1) and thiotepa (120 mg m-2 day-1) for four consecutive days, followed by peripheral stem cell transplantation. Exposure to carboplatin was studied in 200 treatment days by measuring the area under the carboplatin plasma ultrafiltrate (pUF) concentration vs time curve (AUC). The AUC was obtained by using a previously validated limited sampling model. A total of 31 patients was studied who received one, two or three courses of this high-dose chemotherapy regimen. The unbound, plasma ultrafiltrate carboplatin was almost completely cleared from the body before each next treatment day in a course; the day-to-day AUC variation was 3.3%. The mean cumulative AUC over 4 days was 19.6 (range 14.1-27.2) mg ml-1 min-1. In 97 treatment days the carboplatin dose was calculated using the Calvert formula with the creatinine clearance as the measure for the glomerular filtration rate (GFR). For these courses, the inter-patient variability in pharmacokinetics was significantly reduced from 21% to 15% (P = 0.007) in comparison with the schemes where it was given as a fixed dose of 400 mg m-2. There were no relationships found between toxicity and the AUC of carboplatin, which may be due to the influence of overlapping toxicities of cyclophosphamide and thiotepa. However, the ototoxicity was strongly related to the cumulative carboplatin AUC. This toxicity was dose limiting for carboplatin in this schedule. It appeared that the carboplatin pharmacokinetics in these regimens were similar to those reported at conventional dosages. To reduce the inter-patient variation, the carboplatin dose can be calculated using the Calvert-formula with the creatinine clearance as the measure for the GFR.

A single-sample assay for the estimation of the area under the free carboplatin plasma concentration versus time curve

The aim of this study was to develop and validate a simple and rapid method for the estimation of the area under the free carboplatin plasma concentration versus time curve (AUC). The relationship between the carboplatin AUC and the total plasma platinum (Pt) concentration 24 h after treatment was studied using data from 49 patients treated with 20-1600 mg/m2 carboplatin as a 60-100 min infusion (median 60 min). The relationship was confirmed by the in vitro incubation of carboplatin in human plasma and prospectively validated in 13 ovarian cancer patients. Free carboplatin was separated by ultrafiltration (MW cut off 30,000), and free and total Pt measured by atomic absorption spectrophotometry. There was a linear relationship in vivo between the 24 h (median 24.4; range 16.3-27.3 h) total plasma Pt concentration (microM) and free carboplatin AUC (mg/ml.min): AUC=(24 h Pt+0.3)/0.82 (r2=0.93, AUC median 5.8 (0.13-28)mg/ml.min, 24h Pt median 4.4 (0.1-23) microM). A similar relationship was observed in vitro [AUC =(24h Pt +0.1)/0.93 (r2=0.98, AUC median 7.9 (2.0-17) mg/ml.min, 24 h Pt median 7.1 (1.8-15) microM)]. The relationship derived from the in vivo data gave an unbiased and reasonably accurate estimate of the measured carboplatin AUC in 13 patients (AUC =5.1-8.7 mg/ml.min, GFR=59-129 ml/min, infusion time 30-45 min, 24 h sampling time 22.9-24.5 h), giving a percentage mean error of -4.2% and root mean squared percentage error of 11.5%. These results show that the analysis of a single blood sample taken 24 h after carboplatin administration can be used to produce an unbiased and reasonably accurate measure of the free carboplatin AUC. Unlike published limited sampling strategies, this method is not complicated by the need to accurately control the duration of the carboplatin infusion or the time at which the sample is taken.

Comparison between a cisplatin-containing regimen and a carboplatin-containing regimen for recurrent or metastatic bladder cancer patients. A randomized phase II study

BACKGROUND

The aim of this randomized Phase II study was to compare the efficacy and toxicity of a cisplatin-containing regimen with a carboplatin-containing regimen for patients with recurrent or metastatic bladder cancer.

METHODS

Fifty-seven patients with recurrent or metastatic bladder cancer were randomized to receive M-VEC treatment (methotrexate, vinblastine, epirubicin, and cisplatin) (n = 29) or M-VECa treatment (methotrexate, vinblastine, epirubicin, and carboplatin) (n = 28). The chemotherapy was scheduled at 28-day intervals. Recombinant granulocyte-colony stimulating factors were administered daily when the absolute neutrophil count fell below 1000/mm3. The development of ototoxicity was evaluated by measuring auditory brain stem response.

RESULTS

Of the 57 entered patients, 55 were evaluable for response and toxicity. The overall clinical response rate was 71% (with 25% complete responses) in the M-VEC group and 41% (with 11% complete responses) in the M-VECa group (P = 0.04). M-VEC chemotherapy was associated with more pronounced side effects. There was a statistically significant difference between M-VEC and M-VECa in terms of gastrointestinal toxicity (P = 0.04), nephrotoxicity (P = 0.03), and neurotoxicity (P = 0.02) during Cycle 3 of chemotherapy. Leukopenia and neutropenia were worse in the M-VECa arm, but not significantly so (P = 0.4). Ototoxicity was only detected in one of seven examined M-VEC patients after two cycles of chemotherapy.

CONCLUSIONS

M-VECa has a low level of gastrointestinal, renal, neurologic, and otologic toxicity, but is apparently less effective than M-VEC in the treatment of recurrent or metastatic bladder cancer. However, a larger, randomized Phase III trial is needed to confirm these results.

Comparison of methods for the estimation of carboplatin pharmacokinetics in paediatric cancer patients

The antitumour and toxic effects of platinum drugs, in particular carboplatin, have been related to their plasma concentration and this has led to the concept of a target area under the plasma concentration-time curve (AUC) for carboplatin dosing. A formula based on renal function has been successfully applied to carboplatin dosing in adults and modified versions have also been proposed for paediatric patients. In order to monitor carboplatin AUC with maximum efficiency and minimum patient inconvenience, limited sampling strategies are desirable. A population method with Bayesian estimation is described, based on one or two samples taken following a dose of carboplatin. Population data were obtained from 22 paediatric patients treated with 200-1000 mg/m2 carboplatin as a 60-90 min infusion. Ultrafilterable carboplatin was determined by atomic absorption spectrophotometry. A two compartment model was fitted to each data set using the Maximum Likelihood estimator of the ADAPT programme. These parameter estimates provided the prior means and covariance matrix for the Bayesian estimator using a lognormal distribution. The test data sets consisted of ultrafilterable carboplatin concentrations in 23 patients (aged 1 month-18 years) who received similar treatment. The two compartment model was fitted to data sets containing one or two points, using the Bayesian maximum a posteriori (MAP) estimator and an error model derived from the population error model parameters. Results from the Bayesian analysis and other methods for the estimation of AUC, including relating clearance to surface area or to renal function, were evaluated by comparing the AUC estimate with the AUC determined by model-independent analysis. Overall, the optimal sampling strategy performed better than estimates based on renal function, which had a median bias of 5% and precision of 22%. With one data point at 60 min postinfusion, the median bias and precision were 3 and 6%, respectively. Addition of a second data point at 30 min during the infusion improved the estimate slightly (median bias -2%, precision 3%). Bayesian estimation produced more reliable estimates of AUC compared to values based on renal function, which in turn was slightly better than using surface area. A technique, developed in adult patients, for estimating AUC from a measurement of 24 h total plasma platinum was comparable to estimates based on renal function, but was less reliable. The estimation of carboplatin AUC can be performed using only one or two plasma samples and Bayesian analysis. This approach is less biased and more precise than methods based on surface area, renal function or total platinum at 24 h postdose, but is probably best used in combination with dosing based on renal function.

Effects of storage on the binding of carboplatin to plasma proteins

Plasma ultrafiltrates are routinely used in pharmacokinetic studies of carboplatin. Experiments were performed to detect and quantitate artifactual decreases in the platinum concentration of ultrafiltrates prepared from plasma samples stored at -20 degrees C and -70 degrees C. Carboplatin was added to anticoagulated, whole human blood to produce a 20 microgram/ml concentration. Plasma produced from the blood was stored frozen at either -20 degrees C or -70 degrees C. Aliquots from each storage condition were thawed and ultrafiltered once a week for up to 100 days. Platinum concentrations in ultrafiltrates and plasma were determined by flameless atomic absorption spectrometry. There was no loss of ultrafilterable platinum in plasma samples stored at -70 degrees C, whereas there was a steady decrease in free platinum concentration in ultrafiltrates prepared from plasma samples stored at -20 degrees C. These results imply that pharmacokinetic studies of carboplatin should use ultrafiltrates prepared immediately or that plasma for such studies should be stored at -70 degrees C. Storage of carboplatin-containing plasma at -20 degrees C and subsequent ultrafiltration is not acceptable, because measurement of platinum in such ultrafiltrates will be artifactually low.

Limited-sampling models for anticancer agents

Pharmacokinetic parameters of antineoplastic drugs are usually generated from concentration/time profiles obtained after multiple venipunctures. With limited-sampling models (LSM) this number can be reduced to between one and three timed plasma samples. LSMs may facilitate population pharmacokinetic/pharmacodynamic studies, which eventually may lead to a dosing strategy based on the characteristics of the individual patient. In this article, the development, validation and application of several LSMs reported in the literature are reviewed.

Limited sampling models for topotecan pharmacokinetics

BACKGROUND

Limited sampling models for the estimation of the topotecan Area Under the concentration versus time Curve (AUC) and its lactone ring opened form (AUC Tm), from one or more plasma concentration determinations, are desired for further population-kinetic studies.

PATIENTS AND METHODS

The models were developed and validated using 34 pharmacokinetic curves in 19 patients who participated in a phase I study.

RESULTS

A single point model was selected as optimal: AUC (mumol/L.min) = 499 (min).C2h (mumol/L) +0.85 (m2/mg.mumol/L.min).dose(mg/m2), and for topotecan-metabolite (Tm), AUC Tm (mumol/L.min) = 55.1 (min).CTm2h (mumol/L) -0.011 (m2/mg.mg.mumol/L.min).dose (mg/m2), where C2h is the plasma concentration (mumol/L) of topotecan at 2 h after the end of a 30-min infusion, and CTm2h the concentration of the opened form at the same time point. The models are valid for dosages ranging from 0.5 to 1.5 mg/m2/day and proved to be unbiased (MPE% = -1.8% and -9.3%, respectively) and precise (RMSE% = 17.9% and 22.7%, respectively). From the predicted AUCs, the clearance (Cl = dose (mumol)/AUC(mumol/L.min)) could also reliably be predicted, as well as the total AUC (AUC+AUC Tm) RMSE% = 17.1% and MPE% = -0.02%). Half-life values could not be predicted with acceptable precision and accuracy.

CONCLUSION

The limited sampling models presented may useful for future studies focused on pharmacokinetic/pharmacodynamic relationships of topotecan in large populations.

Validation of a limited sampling model for carboplatin in a high-dose chemotherapy combination

A limited sampling model for the estimation of the carboplatin area under the concentration versus time curve (AUC), as developed by Sørensen et al., was validated prospectively for the use in a high-dose combination chemotherapy schedule. The model allows an estimation of the AUC on the basis of only one timed plasma drug concentration, sampled at exactly 2.75 h after a 1-h carboplatin infusion. Pharmacokinetic curves were obtained from nine patients receiving carboplatin (400 mg/m2 per day) combined with cyclophosphamide (1500 mg/m2 per day), thiotepa (120 mg/m2 per day), and mesna (3 g/day) for 4 consecutive days. Peripheral blood stem-cell transplantation (PBSCT) was performed 3 days later to restore hematopoiesis. Using this combination of high doses, the model proved to be unbiased (MPE -3.40%; SE, 1.22%) and highly precise [root mean squared prediction error (RMSE), 5.15%; SE, 0.17%] for estimation of the AUC during 4 consecutive days. The validated limited sampling model provides a starting point for future pharmacokinetic studies in a larger population of patients, which might lead to more insight into the relationships with the pharmacodynamic outcome of carboplatin and may help in achieving more rational dosing of patients on the basis of an AUC determination.

Individual dose adaptation of anticancer drugs

The dose of anticancer drugs is currently adjusted to the patient body surface area, although patients have different abilities to clear anticancer drugs. The dose adjustment to physiological functions permits major toxic accidents to be avoided. The adjustment to tumour drug content is considered, but for ethical or technical reasons, it cannot be used routinely The best criterion for the dose adjustment seems to be drug plasma concentration. The relationship between plasma concentration and efficacy may not be excellent, since it depends on the presence of resistant cells and on the blood flow through the tumour. A relationship between plasma concentration and/or the area under the curve (AUC) with toxicity has been reported with all major anticancer drugs. Different methods of dose adjustment to the drug plasma concentration are reported. In conclusion, dose adjustment to the drug plasma concentration or to the AUC can improve the chemotherapy efficacy, while reducing toxicity.

A limited sampling method for estimation of the etoposide area under the curve

A limited sampling method for estimation of the etoposide area under the curve (AUC) is presented. The method was developed and validated in 23 patients (42 pharmacokinetic studies) with small-cell lung cancer (SCLC), limited disease. The patients received 100 mg/m2 etoposide as a 90-min intravenous infusion in combination with carboplatin, allowing for etoposide dose modification at a following course (25% increase or decrease) due to high or low nadir values for leukocytes or thrombocytes. Of the 42 pharmacokinetic studies, 27 were used in the model development and 15 were used in the model validation. Single regression analyses of the AUC versus the fitted concentrations for the model data set were performed at several time points. The analyses demonstrated high and essentially identical correlation coefficients in the interval between 2 and 21 h, with a maximal value of 0.96 being recorded at 4 h. Multiple regression analysis was then performed using fitted concentrations corresponding to 0.08-21 h. The best model for one sample was AUC = 1.01 x (dose level divided by 100 mg/m2) + 799 x C4 h, that for two samples was AUC = 1.43 x (dose level divided by 100 mg/m2) + 544 x C4 h + 1756 x C21 h, and that for three samples was AUC = 0.07 x (dose level divided by 100 mg/m2) + 110 x C5 min + 474 x C4 h + 1759 x C21 h. Not unexpectedly, the model validation revealed that the one-sample model was less precise than the two- or three-sample model [percentage of root mean squared error (RMSE%) = 11.6%, 7.1%, and 5.4%, respectively]. All models proved to be unbiased in the validation [percentage of mean predictive error (MPE%) +/- SE = 4.2% +/- 11.0%, 7.9% +/- 6.1%, and 6.3% +/- 5.3%, respectively]. The models were subsequently validated in 14 pharmacokinetic studies of patients with metastatic germ-cell tumours who were receiving combination chemotherapy with cisplatin and bleomycin plus 100 mg/m2 etoposide as a 90-min infusion. The RMSE% was 13.4%, 10.8%, and 9.0% and the MPE% +/- SE was -1.0% +/- 11.9%, 1.7% +/- 10.5%, and 2.7% +/- 7.9% for the one-, two-, and three-sample models, respectively. The limited sampling methods presented herein may prove to be a most valuable tool for therapeutic drug monitoring in regimens in which etoposide is given in combination with carboplatin or with cisplatin and bleomycin.