Influence of pharmacogenetics on indinavir disposition and short-term response in HIV patients initiating HAART

Pharmacogenetic considerations in the treatment of co-infections with HIV/AIDS, tuberculosis and malaria in Congolese populations of Central Africa

BACKGROUND

HIV-infection, tuberculosis and malaria are the big three communicable diseases that plague sub-Saharan Africa. If these diseases occur as co-morbidities they require polypharmacy, which may lead to severe drug-drug-gene interactions and variation in adverse drug reactions, but also in treatment outcomes. Polymorphisms in genes encoding drug-metabolizing enzymes are the major cause of these variations, but such polymorphisms may support the prediction of drug efficacy and toxicity. There is little information on allele frequencies of pharmacogenetic variants of enzymes involved in the metabolism of drugs used to treat HIV-infection, TB and malaria in the Republic of Congo (ROC). The aim of this study was therefore to investigate the occurrence and allele frequencies of 32 pharmacogenetic variants localized in absorption distribution, metabolism and excretion (ADME) and non-ADME genes and to compare the frequencies with population data of Africans and non-Africans derived from the 1000 Genomes Project.

RESULTS

We found significant differences in the allele frequencies of many of the variants when comparing the findings from ROC with those of non-African populations. On the other hand, only a few variants showed significant differences in their allele frequencies when comparing ROC with other African populations. In addition, considerable differences in the allele frequencies of the pharmacogenetic variants among the African populations were observed.

CONCLUSIONS

The findings contribute to the understanding of pharmacogenetic variants involved in the metabolism of drugs used to treat HIV-infection, TB and malaria in ROC and their diversity in different populations. Such knowledge helps to predict drug efficacy, toxicity and ADRs and to inform individual and population-based decisions.

Influence of CYP3A5 and SLCO1B1 polymorphisms on atazanavir/r concentrations in Thai HIV-infected patients

Aim: To evaluate the influence of genetic polymorphisms on plasma trough concentrations of atazanavir (ATV) and ritonavir (RTV). Patients & methods: The concentration-to-dose ratios were compared between different genotype groups of CYP3A5, ABCB1, SLCO1B1 and NR1I2 in 490 patients. Multiple regression analysis was used to examine the association between genetic and clinical factors and log-transformed concentration-to-dose ratio of ATV and RTV. Results: Higher concentrations of ATV and RTV were significantly associated with CYP3A5 6986 GG and SLCO1B1 521 TC or CC. Female patients had significantly higher ATV plasma concentration than male patients. Conclusion: Genetic polymorphisms and gender are factors affecting the variability of ATV and RTV concentrations in the Thai population. Thus, genetic testing is worth considering when atazanavir + low dose ritonavir is prescribed.

[Nephrotoxicity of antiretrovirals other than tenofovir]

The remarkable improvement of the outcome of HIV infection came with the price of substantial toxicity of some antiretrovirals. The first molecules used to treat HIV included an important nephrotoxicity. Zalcitabine, stavudine and didanosine can induce severe lactic acidosis. Lactate production is enhanced and the renal capacity to regulate pH is overwhelmed. However, this side effect is not due to a direct dysfunction of the kidneys. Zalcitabine was withdrawn from the market because of this risk. Indinavir, a protease inhibitor, is soluble only in very acidic solutions. Consequently, the small fraction that is excreted in the urine precipitates and can be responsible for uro-nephrolithiasis, leukocyturia, cristalluria, obstructive acute kidney failure, and acute or chronic interstitial nephritis. This is the reason why indinavir is almost not prescribed nowadays, even if it is still marketed. In addition to the direct nephrotoxicity of some antiretrovirals, anti-HIV treatment also includes a toxicity which pathophysiology is not completely elucidated. This nephrotoxicity is the consequence of organ accelerated ageing and of an increased vascular risk. Kidney vascularization (from renal arteries to capillaries) is essential to kidney function and all cardiovascular risks are also renal risks. It is now clearly established that combined antiretroviral treatment increases the vascular risk. A better comprehension of the links between HIV infection, its treatment and very long-term kidney risk is needed to improve the complex management of patients who have now cumulated several decades of HIV infection and treatment with various toxicities.

Differences in the Prediction of Area Under the Curve for a Protease Inhibitor Using Trough Versus Peak Concentration: Assessment Using Published Pharmacokinetic Data for Indinavir

In the present day antiretroviral therapy, Ctrough is a key tool for efficacy assessment. The present work explored the feasibility of using Ctrough or Cmax in the area under the concentration-time curve (AUC) prediction of indinavir. A simple unweighted linear regression model was developed to describe the relationship between Cmax versus AUC (r = 0.8101, P < 0.001) and Ctrough versus AUC (r = 0.8127, P < 0.001) for indinavir. The regression lines were used to predict the AUC values from literature Cmax or Ctrough data of indinavir in HIV and healthy subjects. The fold difference, defined as the quotient of the observed and predicted AUC values, was evaluated along with statistical comparison, including root mean square error (RMSE) prediction for the 2 models. The correlation between Cmax versus AUC and Ctrough versus AUC was established. Majority of the predicted values for Cmax versus AUC were within 0.75- to 1.5-fold differences. However, the Ctrough versus AUC model showed larger variability with approximately one-third of the predictions within 0.75- to 1.5-fold differences. The r value and %RMSE for observed versus predicted AUC for Ctrough (r = 0.5925, n = 65, P < 0.001, and RMSE: 67%) were inferior to the Cmax (r = 0.8773, n = 86, P < 0.001, and RMSE: 46%). In conclusion, Cmax versus AUC and Ctrough versus AUC relationships were established for indinavir showing the utility of a single concentration time point for therapeutic drug monitoring purpose. The Cmax model for indinavir may be more relevant for AUC prediction as determined by the statistical criteria.

Pharmacogenomics of antimicrobial agents

Antimicrobial efficacy and toxicity varies between individuals owing to multiple factors. Genetic variants that affect drug-metabolizing enzymes may influence antimicrobial pharmacokinetics and pharmacodynamics, thereby determining efficacy and/or toxicity. In addition, many severe immune-mediated reactions have been associated with HLA class I and class II genes. In the last two decades, understanding of pharmacogenomic factors that influence antimicrobial efficacy and toxicity has rapidly evolved, leading to translational success such as the routine use of HLA-B*57:01 screening to prevent abacavir hypersensitivity reactions. This article examines recent advances in the field of antimicrobial pharmacogenomics that potentially affect treatment efficacy and toxicity, and challenges that exist between pharmacogenomic discovery and translation into clinical use.

Population pharmacokinetics of nefopam in elderly, with or without renal impairment, and its link to treatment response

AIMS

Nefopam is a nonmorphinic central analgesic, for which no recommendation exists concerning adaptation of regimen in aged patients with or without renal impairment. The objective was to describe the pharmacology of nefopam in aged patients to obtain guidelines for practical use.

METHODS

Elderly patients (n = 48), 65-99 years old, with severe or moderate renal impairment or with normal renal function, were recruited. Nefopam (20 mg) was administered as a 30 min infusion postoperatively. Simultaneously, a 1 min intravenous infusion of iohexol was performed, in order to calculate the glomerular filtration rate. Blood samples were drawn to determine nefopam, desmethyl-nefopam and iohexol plasma concentrations. Nefopam and desmethyl-nefopam concentrations were analysed using a nonlinear mixed-effects modelling approach with Monolix version 4.1.3. The association between pharmacokinetic parameters and treatment response was assessed using logistic regression.

RESULTS

A two-compartment open model was selected to describe the pharmacokinetics of nefopam. The typical population estimates (between-subject variability) for clearance, volume of distribution, intercompartmental clearance and peripheral volume were, respectively, 17.3 l h(-1) (53.2%), 114 l (121%), 80.7 l h(-1) (79%) and 208 l (63.6%). Morphine requirement was related to exposure of nefopam. Tachycardia and postoperative nausea and vomiting were best associated with maximal concentration and the rate of increase in nefopam plasma concentration.

CONCLUSIONS

We identified the nefopam pharmacokinetic predictors for morphine requirement and side-effects, such as tachycardia and postoperative nausea and vomiting. In order to maintain morphine sparing and decrease side-effects following a single dose of nefopam (20 mg), simulations suggest an infusion time of >45 min in elderly patients with or without renal impairment.

[Mathematical dynamical models for personalized medicine]

One of the necessary conditions to perform any personalized medicine is to obtain good individual predictions. In addition to the numerous markers available (omics data), the methods used to analyze the data are very important too. We are presenting an example of mathematical dynamical mechanistic model that could be used for adapting the antiretroviral treatment in patients infected by the human immunodeficiency virus. The interest of this type of approach is to build a model based on biological knowledge about the interaction between markers and therefore to allow for a better predictive power.

In vivo assessment of antiretroviral therapy-associated side effects

Antiretroviral therapy has been associated with side effects, either from the drug itself or in conjunction with the effects of human immunodeficiency virus infection. Here, we evaluated the side effects of the protease inhibitor (PI) indinavir in hamsters consuming a normal or high-fat diet. Indinavir treatment increased the hamster death rate and resulted in an increase in triglyceride, cholesterol and glucose serum levels and a reduction in anti-oxLDL auto-antibodies. The treatment led to histopathological alterations of the kidney and the heart. These results suggest that hamsters are an interesting model for the study of the side effects of antiretroviral drugs, such as PIs.

Reduced indinavir exposure during pregnancy

AIM

To describe the pharmacokinetics and safety of indinavir boosted with ritonavir (IDV/r) during the second and third trimesters of pregnancy and in the post-partum period.

METHODS

IMPAACT P1026s is an on-going, prospective, non-blinded study of antiretroviral pharmacokinetics (PK) in HIV-infected pregnant women with a Thai cohort receiving IDV/r 400/100 mg twice daily during pregnancy through to 6-12 weeks post-partum as part of clinical care. Steady-state PK profiles were performed during the second (optional) and third trimesters and at 6-12 weeks post-partum. PK targets were the estimated 10(th) percentile IDV AUC (12.9 μg ml(-1)h) in non-pregnant historical Thai adults and a trough concentration of 0.1 μg ml(-1), the suggested minimum target.

RESULTS

Twenty-six pregnant women were enrolled; thirteen entered during the second trimester. Median (range) age was 29.8 (18.9-40.8) years and weight 60.5 (50.0-85.0) kg at the third trimester PK visit. The 90% confidence limits for the geometric mean ratio of the indinavir AUC(0,12 h) and Cmax during the second trimester and post-partum (ante : post ratios) were 0.58 (0.49, 0.68) and 0.73 (0.59, 0.91), respectively; third trimester/post-partum AUC(0,12 h) and Cmax ratios were 0.60 (0.53, 0.68) and 0.63 (0.55, 0.72), respectively. IDV/r was well tolerated and 21/26 women had a HIV-1 viral load < 40 copies ml(-1) at delivery. All 26 infants were confirmed HIV negative.

CONCLUSION

Indinavir exposure during the second and third trimesters was significantly reduced compared with post-partum and ∼30% of women failed to achieve a target trough concentration. Increasing the dose of IDV/r during pregnancy to 600/100 mg twice daily may be preferable to ensure adequate drug concentrations.

Dynamical models of biomarkers and clinical progression for personalized medicine: the HIV context

Mechanistic models, based on ordinary differential equation systems, can exhibit very good predictive abilities that will be useful to build treatment monitoring strategies. In this review, we present the potential and the limitations of such models for guiding treatment (monitoring and optimizing) in HIV-infected patients. In the context of antiretroviral therapy, several biological processes should be considered in addition to the interaction between viruses and the host immune system: the mechanisms of action of the drugs, their pharmacokinetics and pharmacodynamics, as well as the viral and host characteristics. Another important aspect to take into account is clinical progression, although its implementation in such modelling approaches is not easy. Finally, the control theory and the use of intrinsic properties of mechanistic models make them very relevant for dynamic treatment adaptation. Their implementation would nevertheless require their evaluation through clinical trials.

Clinical pharmacokinetics of antiretroviral drugs in older persons

INTRODUCTION

Combination antiretroviral therapy has enabled HIV-infected persons to reach older ages in high numbers. Hepatic and renal changes that normally occur with advancing age occur earlier and with higher incidence in HIV-infected individuals. A limited number of prospective controlled studies have demonstrated small reductions (17 to 41%) in lopinavir, atazanavir and lamivudine clearance in older versus younger adults. A much larger number of retrospective studies in adults (age range ∼ 20 to 60 years), including all antiretroviral drugs, have evaluated age as a covariate for pharmacokinetics. Most studies did not detect substantial associations between drug exposures and age.

AREAS COVERED

This review summarizes antiretroviral drug pharmacokinetics in older persons. The authors review articles from PubMed (search terms: elderly, antiretroviral, pharmacokinetics) in addition to the bibliographies of those selected.

EXPERT OPINION

The evidence to date does not support major pharmacokinetic changes in adults between ∼ 20 and 60 years of age. However, additional prospective, well-controlled studies are needed in more persons > 60 years, including those with frailty and comorbidities, with assessment of unbound drug clearance, and incorporation of adherence, pharmacogenetics and concomitant medications. Until then, guidelines for drug-drug interactions and dosing in renal and hepatic impairment should be followed in older HIV-infected individuals.

Effect of single nucleotide polymorphisms in cytochrome P450 isoenzyme and N-acetyltransferase 2 genes on the metabolism of artemisinin-based combination therapies in malaria patients from Cambodia and Tanzania

The pharmacogenetics of antimalarial agents are poorly known, although the application of pharmacogenetics might be critical in optimizing treatment. This population pharmacokinetic-pharmacogenetic study aimed at assessing the effects of single nucleotide polymorphisms (SNPs) in cytochrome P450 isoenzyme genes (CYP, namely, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP3A4, and CYP3A5) and the N-acetyltransferase 2 gene (NAT2) on the pharmacokinetics of artemisinin-based combination therapies in 150 Tanzanian patients treated with artemether-lumefantrine, 64 Cambodian patients treated with artesunate-mefloquine, and 61 Cambodian patients treated with dihydroartemisinin-piperaquine. The frequency of SNPs varied with the enzyme and the population. Higher frequencies of mutant alleles were found in Cambodians than Tanzanians for CYP2C9*3, CYP2D6*10 (100C → T), CYP3A5*3, NAT2*6, and NAT2*7. In contrast, higher frequencies of mutant alleles were found in Tanzanians for CYP2D6*17 (1023C → T and 2850C → T), CYP3A4*1B, NAT2*5, and NAT2*14. For 8 SNPs, no significant differences in frequencies were observed. In the genetic-based population pharmacokinetic analyses, none of the SNPs improved model fit. This suggests that pharmacogenetic data need not be included in appropriate first-line treatments with the current artemisinin derivatives and quinolines for uncomplicated malaria in specific populations. However, it cannot be ruled out that our results represent isolated findings, and therefore more studies in different populations, ideally with the same artemisinin-based combination therapies, are needed to evaluate the influence of pharmacogenetic factors on the clearance of antimalarials.

Nuclear receptor-mediated induction of CYP450 by antiretrovirals: functional consequences of NR1I2 (PXR) polymorphisms and differential prevalence in whites and sub-Saharan Africans

BACKGROUND

Antiretroviral therapy including HIV protease inhibitors and nonnucleoside reverse transcriptase inhibitors can both inhibit and induce expression of cytochrome P450s, potentially leading to drug interactions. However, information is lacking on the impact of genetic polymorphism on this interaction.

METHODS

This study examines the prevalence of 33 polymorphisms in NR1I2 (pregnane X receptor [PXR]), CYP3A4, and CYP2B6 in 1013 white and sub-Saharan African patients with HIV; explores the inductive ability of 16 antiretrovirals on CYP3A4 and CYP2B6 promoter activity through nuclear receptors PXR and constitutive androstane receptor (CAR); and evaluates the influence of naturally occurring PXR genetic variants on antiretroviral activation.

RESULTS

Seventeen polymorphisms were present at different frequencies between the two ethnicities. Darunavir, fosamprenavir, lopinavir, nelfinavir, tipranavir, efavirenz, and abacavir increased CYP3A4 and/or CYP2B6 promoter activity, some through constitutive androstane receptor but mainly through PXR. Addition of low-dose ritonavir enhanced levels of CYP promoter activity for several protease inhibitors. Some PXR variants displayed lower fosamprenavir- and lopinavir-induced CYP3A4 promoter activity than the PXR reference sequence, whereas efavirenz and nelfinavir induction was unchanged.

CONCLUSIONS

The presence of NR1I2 polymorphisms can alter the induction of CYP3A4 and CYP2B6 promoter activity, potentially adding to the unpredictable nature of antiretroviral drug interactions. These polymorphisms differ in prevalence between whites and sub-Saharan Africans.

Influence of body weight on achieving indinavir concentrations within its therapeutic window in HIV-infected Thai patients receiving indinavir boosted with ritonavir

Indinavir boosted with ritonavir (IDV/r) dosing with 400/100 mg, twice daily, is preferred in Thai adults, but this dose can lead to concentrations close to the boundaries of its therapeutic window. The objectives of this analysis were to validate a population pharmacokinetic model to describe IDV/r concentrations in HIV-infected Thai patients and to investigate the impact of patient characteristics on achieving adequate IDV concentrations. IDV/r concentration data from 513 plasma samples were available. Population means and variances of pharmacokinetic parameters were estimated using a nonlinear mixed effects regression model (NONMEM Version VI). Monte Carlo simulations were performed to estimate the probability of achieving IDV concentrations within its therapeutic window. IDV/r pharmacokinetics were best described by a one-compartment model coupled with a single transit compartment absorption model. Body weight influenced indinavir apparent oral clearance and volume of distribution and allometric scaling significantly reduced the interindividual variability. Final population estimates (interindividual variability in percentage) of indinavir apparent oral clearance and volume of distribution were 21.3 L/h/70 kg (30%) and 90.7 L/70 kg (22%), respectively. Based on model simulations, the probability of achieving an IDV trough concentration greater than 0.1 mg/L was greater than 99% for 600/100 mg and greater than 98% for 400/100 mg, twice daily, in patients weighing 40 to 80 kg. However, the probability of achieving IDV concentrations associated with an increased risk of drug toxicity (greater than 10.0 mg/L) increased from 1% to 10% with 600/100 mg compared with less than 1% with 400/100 mg when body weight decreased from 80 to 40 kg. The validated model developed predicts that 400/100 mg of IDV/r, twice daily, provides indinavir concentrations within the recommended therapeutic window for the majority of patients. The risk of toxic drug concentrations increases rapidly with IDV/r dose of 600/100 mg for patients less than 50 kg and therapeutic drug monitoring of IDV concentrations would help to reduce the risk of IDV-induced nephrotoxicity.

[Evidence-based therapeutic drug monitoring for indinavir]

The HIV protease inhibitor indinavir presents a wide inter-individual variability related to an intense hepatic metabolism. Published studies were analyzed to establish whether there is evidence that therapeutic drug monitoring of indinavir could improve patient care. It was reported that indinavir virological efficacy in HIV-infected patients with wild-type virus was significantly associated with trough concentrations > 100-150 ng/mL. Concerning the exposure-toxicity relationship, the risk of occurrence of nephrotoxicity was more frequently associated with trough concentrations > 500-1 000 ng/mL. Studies with concentration-controlled indinavir therapy suggest that therapeutic drug monitoring allows to achieve safe and effective concentrations, therefore, the level of evidence of the interest of indinavir therapeutic drug monitoring is highly recommended when indinavir is not associated with ritonavir and recommended when ritonavir is combined with ritonavir.

Joint population pharmacokinetic analysis of zidovudine, lamivudine, and their active intracellular metabolites in HIV patients

The population pharmacokinetic parameters of zidovudine (AZT), lamivudine (3TC), and their active intracellular metabolites in 75 naïve HIV-infected patients receiving an oral combination of AZT and 3TC twice daily as part of their multitherapy treatment in the COPHAR2-ANRS 111 trial are described. Four blood samples per patient were taken after 2 weeks of treatment to measure drug concentrations at steady state. Plasma AZT and 3TC concentrations were measured in 73 patients, and among those, 62 patients had measurable intracellular AZT-TP and 3TC-TP concentrations. For each drug, a joint population pharmacokinetic model was developed and we investigated the influence of different covariates. We then studied correlations between the mean plasma and intracellular concentrations of each drug. A one-compartment model with first-order absorption and elimination best described the plasma AZT concentration, with an additional compartment for intracellular AZT-TP. A similar model but with zero-order absorption was found to adequately described concentrations of 3TC and its metabolite 3TC-TP. The half-lives of AZT and 3TC were 0.81 h (94.8%) and 2.97 h (39.2%), respectively, whereas the intracellular half-lives of AZT-TP and 3TC-TP were 10.73 h (69%) and 21.16 h (44%), respectively. We found particularly a gender effect on the apparent bioavailability of AZT, as well as on the mean plasma and intracellular concentrations of AZT, which were significantly higher in females than in males. Relationships between mean plasma drug and intracellular metabolite concentrations were also highlighted both for AZT and for 3TC. Simulation with the model of plasma and intracellular concentrations for once- versus twice-daily regimens suggested that a daily dosing regimen with double doses could be appropriate.

Human immunodeficiency virus therapeutics and pharmacogenomics

Pharmacogenomics and pharmacogenetics are promising in development of a personalized treatment approach They are of paramount importance for basic immunology, for peptide based vaccine design (vaccinomics) drug monitoring in clinical setting and molecular pathophysiology of multifactorial diseases like cancer, tuberculosis, cardiac disorders, diabetes, asthma, HIV, etc.

Analysis of pharmacogenetic traits in two distinct South African populations

Our knowledge of pharmacogenetic variability in diverse populations is scarce, especially in sub-Saharan Africa. To bridge this gap in knowledge, we characterised population frequencies of clinically relevant pharmacogenetic traits in two distinct South African population groups. We genotyped 211 tagging single nucleotide polymorphisms (tagSNPs) in 12 genes that influence antiretroviral drug disposition, in 176 South African individuals belonging to two distinct population groups residing in the Western Cape: the Xhosa (n = 109) and Cape Mixed Ancestry (CMA) (n = 67) groups. The minor allele frequencies (MAFs) of eight tagSNPs in six genes (those encoding the ATP binding cassette sub-family B, member 1 [ABCB1], four members of the cytochrome P450 family [CYP2A7P1, CYP2C18, CYP3A4, CYP3A5] and UDP-glucuronosyltransferase 1 [UGT1A1]) were significantly different between the Xhosa and CMA populations (Bonferroni p < 0.05). Twenty-seven haplotypes were inferred in four genes (CYP2C18, CYP3A4, the gene encoding solute carrier family 22 member 6 [SLC22A6] and UGT1A1) between the two South African populations. Characterising the Xhosa and CMA population frequencies of variant alleles important for drug transport and metabolism can help to establish the clinical relevance of pharmacogenetic testing in these populations.

The missing linkage: what pharmacogenetic associations are left to find in CYP3A?

IMPORTANCE OF THE FIELD

An enormous amount of drugs and endogenous substrates are metabolized by the enzymes encoded in the CYP3A gene cluster, making variation at this locus of utmost importance in the field of pharmacogenetics. However, the identification of genetic variation that contributes to the wide phenotypic variability at this locus has been elusive. While dozens of studies have investigated the effects of coding variants, none have found the definitive answer to what variant or variants explain the distribution of enzyme activity and clinical effects seen with the drug metabolized by these genes.

AREAS COVERED IN THIS REVIEW

This review highlights the recent pharmacogenetic work at the CYP3A locus, in particular studies on known functional variants in CYP3A4 and CYP3A5. In addition, common pharmacogenetic strategies as well as considerations specific to the CYP3A locus are discussed.

WHAT THE READER WILL GAIN

The reader will gain a greater understanding of the complexities involved in studying the CYP3A locus, population differences that may affect pharmacogenetic studies at this locus and the importance of variation that affect gene regulation.

TAKE HOME MESSAGE

More innovative and comprehensive methods to assay this region are needed, with particular attention paid to the role of gene regulation and non-coding sequence.

Bioequivalence tests based on individual estimates using non-compartmental or model-based analyses: evaluation of estimates of sample means and type I error for different designs

PURPOSE

The main objective of this work is to compare the standard bioequivalence tests based on individual estimates of the area under the curve and the maximal concentration obtained by non-compartmental analysis (NCA) to those based on individual empirical Bayes estimates (EBE) obtained by nonlinear mixed effects models.

METHODS

We evaluate by simulation the precision of sample means estimates and the type I error of bioequivalence tests for both approaches. Crossover trials are simulated under H ( 0 ) using different numbers of subjects (N) and of samples per subject (n). We simulate concentration-time profiles with different variability settings for the between-subject and within-subject variabilities and for the variance of the residual error.

RESULTS

Bioequivalence tests based on NCA show satisfactory properties with low and high variabilities, except when the residual error is high, which leads to a very poor type I error, or when n is small, which leads to biased estimates. Tests based on EBE lead to an increase of the type I error, when the shrinkage is above 20%, which occurs notably when NCA fails.

CONCLUSIONS

For small n or data with high residual error, tests based on a global data analysis should be considered instead of those based on individual estimates.

Pharmacogenetics and population pharmacokinetics: impact of the design on three tests using the SAEM algorithm

Pharmacogenetics is now widely investigated and health institutions acknowledge its place in clinical pharmacokinetics. Our objective is to assess through a simulation study, the impact of design on the statistical performances of three different tests used for analysis of pharmacogenetic information with nonlinear mixed effects models: (i) an ANOVA to test the relationship between the empirical Bayes estimates of the model parameter of interest and the genetic covariate, (ii) a global Wald test to assess whether estimates for the gene effect are significant, and (iii) a likelihood ratio test (LRT) between the model with and without the genetic covariate. We use the stochastic EM algorithm (SAEM) implemented in MONOLIX 2.1 software. The simulation setting is inspired from a real pharmacokinetic study. We investigate four designs with N the number of subjects and n the number of samples per subject: (i) N = 40/n = 4, similar to the original study, (ii) N = 80/n = 2 sorted in 4 groups, a design optimized using the PFIM software, (iii) a combined design, N = 20/n = 4 plus N = 80 with only a trough concentration and (iv) N = 200/n = 4, to approach asymptotic conditions. We find that the ANOVA has a correct type I error estimate regardless of design, however the sparser design was optimized. The type I error of the Wald test and LRT are moderatly inflated in the designs far from the asymptotic (<10%). For each design, the corrected power is analogous for the three tests. Among the three designs with a total of 160 observations, the design N = 80/n = 2 optimized with PFIM provides both the lowest standard error on the effect coefficients and the best power for the Wald test and the LRT while a high shrinkage decreases the power of the ANOVA. In conclusion, a correction method should be used for model-based tests in pharmacogenetic studies with reduced sample size and/or sparse sampling and, for the same amount of samples, some designs have better power than others.