Brigitte Tranchand

Are population pharmacokinetic and/or pharmacodynamic models adequately evaluated? A survey of the literature from 2002 to 2004.

Model evaluation is an important issue in population analyses. We aimed to perform a systematic review of all population pharmacokinetic and/or pharmacodynamic analyses published between 2002 and 2004 to survey the current methods used to evaluate models and to assess whether those models were adequately evaluated.We selected 324 articles in MEDLINE using defined key words and built a data abstraction form composed of a checklist of items to extract the relevant information from these articles with respect to model evaluation. In the data abstraction form, evaluation methods were divided into three subsections: basic internal methods (goodness-of-fit [GOF] plots, uncertainty in parameter estimates and model sensitivity), advanced internal methods (data splitting, resampling techniques and Monte Carlo simulations) and external model evaluation.Basic internal evaluation was the most frequently described method in the reports: 65% of the models involved GOF evaluation. Standard errors or confidence intervals were reported for 50% of fixed effects but only for 22% of random effects. Advanced internal methods were used in approximately 25% of models: data splitting was more often used than bootstrap and cross-validation; simulations were used in 6% of models to evaluate models by a visual predictive check or by a posterior predictive check. External evaluation was performed in only 7% of models.Using the subjective synthesis of model evaluation for each article, we judged the models to be adequately evaluated in 28% of pharmacokinetic models and 26% of pharmacodynamic models. Basic internal evaluation was preferred to more advanced methods, probably because the former is performed easily with most software. We also noticed that when the aim of modelling was predictive, advanced internal methods or more stringent methods were more often used.

Evaluation of uncertainty parameters estimated by different population PK software and methods.

The uncertainty associated with parameter estimations is essential for population model building, evaluation, and simulation. Summarized by the standard error (SE), its estimation is sometimes questionable. Herein, we evaluate SEs provided by different non linear mixed-effect estimation methods associated with their estimation performances. Methods based on maximum likelihood (FO and FOCE in NONMEMTM, nlme in SplusTM, and SAEM in MONOLIX) and Bayesian theory (WinBUGS) were evaluated on datasets obtained by simulations of a one-compartment PK model using 9 different designs. Bootstrap techniques were applied to FO, FOCE, and nlme. We compared SE estimations, parameter estimations, convergence, and computation time. Regarding SE estimations, methods provided concordant results for fixed effects. On random effects, SAEM and WinBUGS, tended respectively to under or over-estimate them. With sparse data, FO provided biased estimations of SE and discordant results between bootstrapped and original datasets. Regarding parameter estimations, FO showed a systematic bias on fixed and random effects. WinBUGS provided biased estimations, but only with sparse data. SAEM and WinBUGS converged systematically while FOCE failed in half of the cases. Applying bootstrap with FOCE yielded CPU times too large for routine application and bootstrap with nlme resulted in frequent crashes. In conclusion, FO provided bias on parameter estimations and on SE estimations of random effects. Methods like FOCE provided unbiased results but convergence was the biggest issue. Bootstrap did not improve SEs for FOCE methods, except when confidence interval of random effects is needed. WinBUGS gave consistent results but required long computation times. SAEM was in-between, showing few under-estimated SE but unbiased parameter estimations.

Etoposide pharmacokinetics and survival in patients with small cell lung cancer: A multicentre study

PURPOSEnTo investigate the prognostic value of systemic exposure to etoposide (Area Under the concentration Curve (AUC(VP16))) on overall survival (OS) in patients with small cell lung cancer (SCLC).nnnPATIENTS AND METHODSnData from 52 patients with limited stage (n=17) or metastatic (n=35) SCLC were analysed. They received at least two courses of etoposide (120mg/(m(2)day) on 3 days) combined with either doxorubicin-ifosfamide (AVI, n=29) or platinum compounds (carboplatin: n=16; cisplatin: n=7). Population pharmacokinetic-pharmacodynamic (PK-PD) study was performed using NON-linear Mixed Effect Model (NONMEM) and Splus software with univariate and multivariate analyses.nnnRESULTSnEtoposide plasma concentration vs. time was described by a two compartment model. Etoposide clearance (CL) was significantly dependant on serum creatinine (Scr). Ifosfamide (IFO) coadministration increased etoposide clearance by 28% (median CL(VP16): 2.42L/h vs. 1.89L/h, p<0.0005) leading to a reduced systemic exposure (median AUC(VP16): 260mgh/L vs. 339mgh/L). No influence of body surface area (BSA) on CL(VP16) was observed. Median percent decrease of absolute neutrophil count (ANC) after the first chemotherapy course was greater when etoposide 24h concentration was above 0.33mg/L (88% vs. 0%, p=0.028). Median OS was significantly longer in patients treated without ifosfamide (11.0 months vs. 7.0 months, p=0.049) and in patients with CL(VP16)<2.22L/h (14 months vs. 7 months, p=0.013) and AUC(VP16)>254.8mgh/L (11 months vs. 7 months, p=0.048). The independent prognostic factors regarding OS were LDH, CL(VP16) and AUC(VP16).nnnCONCLUSIONnIn this study it was found that CL(VP16) is reduced in patients with elevated serum creatinine, whilst ifosfamide coadministration increases CL(VP16) and reduces AUC(VP16), demonstrating the interaction between VP16 and ifosfamide. CL(VP16) and AUC(VP16) correlate significantly with overall survival of patients with SCLC patients receiving etoposide regimens.

Predictive values of hCG clearance for risk of methotrexate resistance in low-risk gestational trophoblastic neoplasias

BACKGROUNDnEarly identification of patients at high risk for chemoresistance among those treated with methotrexate (MTX) for low-risk gestational trophoblastic neoplasia (GTN) is needed. We modeled human chorionic gonadotropin (hCG) decline during MTX therapy using a kinetic population approach to calculate individual hCG clearance (CL(hCG)) and assessed the predictive value of CL(hCG) for MTX resistance.nnnPATIENTS AND METHODSnA total of 154 patients with low-risk GTN treated with 8-day MTX regimen were retrospectively studied. NONMEM was used to model hCG decrease equations between day 0 and day 40 of chemotherapy. Receiver operating characteristic curve analysis defined the best CL(hCG) threshold. Univariate/multivariate survival analyses determined the predictive value of CL(hCG) and compared it with published predictive factors.nnnRESULTSnA monoexponential equation best modeled hCG decrease: hCG(t) = 3900 x e(-0.149 x t). Median CL(hCG) was 0.57 l/day (quartiles: 0.37-0.74). Only choriocarcinoma pathology [yes versus no: hazard ratio (HR) = 6.01; 95% confidence interval (CI) 2.2-16.6; P < 0.001] and unfavorable CL(hCG) quartile (< or =0.37 versus >0.37 l/day: HR = 6.75; 95% CI 2.7-16.8; P < 0.001) were significant independent predictive factors of MTX resistance risk.nnnCONCLUSIONnIn the second largest cohort of low-risk GTN patients reported to date, choriocarcinoma pathology and CL(hCG) < or =0.37 l/day were major independent predictive factors for MTX resistance risk.

Exposure-effect population model of inolimomab, a monoclonal antibody administered in first-line treatment for acute graft-versus-host disease.

Background and objectiveInolimomab, a monoclonal antibody against interleukin (IL)-2Rα (CD25) has shown promising results in the treatment of corticosteroid-resistant acute graft-versus-host disease (GvHD). The objective of the present study was to characterise the pharmacokinetic and pharmacodynamic properties of inolimomab as first-line treatment in this condition.MethodsThe data came from 21 patients with acute GvHD (8 with an International Bone Marrow Transplant Registry [IBMTR] score of B, 11 with a score of C and 2 with a score of D) following haematopoietic stem cell transplantation after a median delay of 26 days (range 10–127 days). Inolimomab was administered at 0.1, 0.2, 0.3 or 0.4 mg/kg daily in association with methylprednisolone (2 mg/kg) for 8 or 16 days depending on the status at day 9. Then, for responder patients, administrations were continued three times weekly until day 28. Inolimomab concentrations and pharmacodynamic data (acute GvHD scores) were recorded during the study. The pharmacodynamic data were assessed in four grades according to the IBMTR and Glucksberg classification in parallel with Karnofsky scores. A population analysis was developed using a nonlinear mixedeffects model to define the pharmacokinetic model, to test covariates and, when apparent, to model the exposure-effect relationship by a proportional odds model. The modelling was finally qualified by a predictive check.ResultsThe best pharmacokinetic model was two-compartmental. For each score, the most demonstrative exposure-effect graphics linked the cumulative area under the concentration-time curve to cumulated probabilities of observing a specific score. This relationship was identified as a maximum effect model for the skin (with two patient subpopulations: sensitive/less sensitive) and a linear model for the intestinal tract and liver. No covariate was identified as influencing any of these parameters.ConclusionInolimomab exposure-effect relationships as first-line treatment for acute GvHD have been identified and modelled. The discovered dose-effect relationship allows confirmation of the treatment response, thereby establishing the first step towards optimising the inolimomab dosage in future trials.

Predictive Value of Modeled AUCAFP-hCG, a Dynamic Kinetic Parameter Characterizing Serum Tumor Marker Decline in Patients With Nonseminomatous Germ Cell Tumor

OBJECTIVEnThe early decline profile of alpha-fetoprotein (AFP) and human chorionic gonadotropin (hCG) in patients with nonseminomatous germ cell tumors (NSGCT) treated with chemotherapy may be related to the risk of relapse. We assessed the predictive values of areas under the curve of hCG (AUC(hCG)) and AFP (AUC(AFP)) of modeled concentration-time equations on progression-free survival (PFS).nnnMETHODSnSingle-center retrospective analysis of hCG and AFP time-points from 65 patients with IGCCCG intermediate-poor risk NSGCT treated with 4 cycles of bleomycin-etoposide-cisplatin (BEP). To determine AUC(hCG) and AUC(AFP) for D0-D42, AUCs for D0-D7 were calculated using the trapezoid rule and AUCs for D7-D42 were calculated using the mathematic integrals of equations modeled with NONMEM. Combining AUC(AFP) and AUC(hCG) enabled us to define 2 predictive groups: namely, patients with favorable and unfavorable AUC(AFP-hCG). Survival analyses and ROC curves assessed the predictive values of AUC(AFP-hCG) groups regarding progression-free survival (PFS) and compared them with those of half-life (HL) and time-to-normalization (TTN).nnnRESULTSnMono-exponential models best fit the patterns of marker decreases. Patients with a favorable AUC(AFP-hCG) had a significantly better PFS (100% vs 71.5%, P = .014). ROC curves confirmed the encouraging predictive accuracy of AUC(AFP-hCG) against HL or TTN regarding progression risk (ROC AUCs = 79.6 vs 71.9 and 70.2 respectively). Because of the large number of patients with missing data, multivariate analysis could not be performed.nnnCONCLUSIONnAUC(AFP-hCG) is a dynamic parameter characterizing tumor marker decline in patients with NSGCT during BEP treatment. Its value as a promising predictive factor should be validated.

Chemotherapy may be delivered based on an integrated view of tumour dynamics

The aim here was to explore the potential of pharmacokinetic (PK)/pharmacodynamic (PD) and physiopathological parameters in explaining the primary effects of an anti-cancer treatment that targets cells in a specific cell cycle phase. The authors applied a theoretical multi-scale disease model of tumour growth that integrates cancer processes at the cellular and tissue scales. The mathematical model at the cell level relies on a dynamic description of cell cycle regulation while the model at the tissue level is based on fluid mechanics considerations. Simulations show that the number of target cells oscillates as the tumour grows after a first cycle of chemotherapy. Both treatment effect and tumour growth processes drive these oscillations. Nonetheless, results indicate that parameters related to physiopathological processes may have greater relevance than classical drug-related parameters in determining the efficacy of a chemotherapy treatment protocol. Physiopathological parameters, in particular those related to cell cycle regulation, may be integrated in PK/PD models aimed at optimising the delivery of phase-specific cytotoxic treatments.

Advantages of prostate-specific antigen (PSA) clearance model over simple PSA half-life computation to describe PSA decrease after prostate adenomectomy

OBJECTIVESnA population kinetic approach based on PSA clearance (CL(PSA)) may be a more rational strategy to characterize prostate-specific antigen (PSA) decrease profile after prostate surgery than the commonly used method (half-life from mono/bi-exponential models).nnnMETHODSnWe used 182 post-adenomectomy PSA concentrations from 56 benign prostatic hyperplasia patients to build, with NONMEM software, a multi-exponential and a CL(PSA) model for comparison.nnnRESULTSnThe best multi-exponential model was PSA(t)=4.96e(-)(0.269t)+3.10e(-)(0.16t)+0.746e(+)(0.0002t) with a stable median residual PSA at 0.64 ng/mL. The best model parametrized with clearance was CL(PSA)=0.0229()(AGE/69)(3.78). Akaike information criteria and standard errors favored the CL(PSA) model. Median peripheral zone and transitional zone productions were 0.034 ng/mL/cm(3) and 0.136 ng/mL/g. A threshold at 2 ng/mL on day 90 allowed for a diagnostic of biochemical relapse diagnostic.nnnCONCLUSIONSnThe population CL(PSA) model was superior to the multi-exponential approach for investigating individual post-adenomectomy PSA decreases.