Dosing Regimen Optimization Of Olanzapine,Quetiapine,and Aripiprazole For Children And Pregnant Women Using Physiologically Based Pharmacokinetic Modeling

Children and pregnant women are special populations for clinical medication,and the rational drug use in such populations has always been a major clinical challenge.Due to ethical and other reasons,the pharmacokinetic(PK)and pharmacodynamic(PD)studies of most drugs in children and pregnant groups are rarely reported,and as a consequence,the practice of rational drug dosing for special groups is often limited.Ontogeny and pregnancy will lead to significant physiological changes,resulting in changes in the in vivo process and bioavailability of drugs;therefore,unreasonable drug dose may lead to failure to reach the treatment demand or excessive blood concentrations.At present,off-label and empirical drug use that are not recognized by the regulatory authorities are very common in clinical practice.Physiologically based pharmacokinetic(PBPK)is a mechanistic method to describe the complex processes of drug absorption,distribution,metabolism and excretion through mathematical modeling and simulation.PBPK model is a multi-compartment model.The compartments correspond to the actual tissues and organs of the human body interconnecting with blood flow.PBPK provides a mechanism-based understanding of drug disposition in vivo.It is becoming a powerful tool to facilitate new drug development and regulatory evaluation.PBPK has unique advantages in prediction of pharmacokinetics in special populations.Over the past two decades,the PBPK method for extrapolation of pediatric pharmacokinetics has made considerable progress and gradually been recognized by regulatory authorities.Schizophrenia and bipolar disorder are mental diseases affecting up to 70 million people worldwide related to a large number of disabilities.The individual drug exposure and response to atypical antipsychotics such as olanzapine,quetiapine and aripiprazole should be closely monitored in clinical treatment.The Arbeitsgemeinschaft fuer Neuropsychopharmakologie und Pharmakopsychiatrie(AGNP)has given olanzapine,quetiapine and aripiprazole a high grade for conducting therapeutic drug monitoring(TDM).Epidemiological studies have found that the use of such drugs in children and pregnant women has increased dramatically over the past decade.The three drugs ranked the top four most commonly prescribed antipsychotics in outpatient children.For women,the diseases have a peak incidence between 25 and 35 years of age,a period during which women are highly likely to experience pregnancy.Pharmacokinetic studies of these drugs in adults have been reported continually,however,such studies in pediatric and pregnant population are limited.Previous studies only showed that there was no significant difference in PK behavior between adults and adolescents over 12 years old.TDM data in pregnant women showed that steady-state plasma concentrations of quetiapine and aripiprazole decreased significantly during pregnancy.At present,general dose adjustment schemes for the two special populations has not been developed.It is of great significance to optimize pharmacotherapy for mental diseases according to the physiological characteristics of pediatric and pregnant populations.This research aims to provide reference information for the rational use of three antipsychotic drugs in children and pregnant population using modeling and simulation techniques.This study will establish PBPK models of quetiapine,olanzapine,aripiprazole and their active metabolites in children and pregnant women,investigate the pharmacokinetic changes in these two groups,and optimize dosing regimens for children of different ages and women of different pregnancy stages.In addition,this study systematically reviewed the adverse extrapyramidal reactions in children or fetal malformations caused by the use of the three drugs in order to evaluate the rationality of the dose adjustment from the perspective of evidence-based medicine.Methods1.Development and verification of PBPK models of quetiapine,olanzapine,aripiprazole,and dehydroaripiprazole.First,PBPK models of the three drugs were developed by using a ‘bottom-up’ and ‘top-down’ combined strategy on the adult 18 compartment physiological model.Published PK studies were searched and divided into test and validation sets.The test set is composed of intravenous PK data and representative single-dose oral PK data,which were used to assist modeling.The chemical parameters of the input model included physical and chemical parameters such as lipid-water partition coefficient,ionization constant,molecular weight,solubility,etc.Absorption parameters such as intestinal permeability,gastrointestinal transit time and drug release properties(described by Weibull equation);Distribution parameters such as plasma unbound drug ratio and whole blood/plasma partition coefficient;Clearance parameters such as intrinsic clearance by metabolic enzymes and renal clearance.Rodgers & Rowland and other chemometric methods were used to calculate the distribution of drugs in vivo.We obtain the contribution of metabolic enzymes through the in vivo mass balance study or in vitro metabolism study.Parameter optimization is based on Monte-Carlo method.The pharmacokinetic parameters such as simulated and actual timeconcentration curves,Cmax and AUC were compared.The goodness of fit of plasma concentration was depicted,and the quantitative evaluation indexes such as geometric mean fold error(GMFE)and mean absolute prediction error(MAPE)were calculated.2.Development and application of pediatric scaling method.The pediatric PBPK model is still based on the physiological structure of adult 18 compartments.Using the large-scale population physiological data provided by the International Committee for Radiation Protection(ICRP),the age-dependent physiological anatomy parameter algorithm is used to create virtual pediatric groups.The adjusted physiological parameters included height,weight,tissue volume,tissue blood flow,cardiac output,hematocrit,total body water,lipid and protein concentrations.The plasma unbound fraction in children was calculated based on the adult value by introducing plasma protein ontogeny factor.Presuming the consistent metabolic pathways between adults and children and well-stirred model conditions,we calculated the intrinsic clearance in children with age-dependent ontogeny factors of metabolic enzymes.The mechanism-based renal clearance extrapolation considers the growth of glomerular filtration and renal tubular secretion with age.We used oxycodone as a test compound to assess the reliability of the above methodology for predicting PK in children.Subsequently,the platform was used to simulate published PK reports of the three antipsychotic drugs in children.In the dose optimization study,four virtual groups were created in the model,i.e.,adults(19–40 years old),adolescents(12–18 years old),school-age children(6 –12 years old)and preschool(2–6 years old).Each group had 500 individuals,half male and half female.The pediatric dose was designed with reference to the blood drug exposure in the adult population and the therapeutic window recommended by the Consensus Guideline for TDM.3.Development and application of pregnant PBPK model.On the basis of 18 compartments for healthy adults,9 compartments unique to pregnant women were added,including breast,myometrium,endometrium,amniotic fluid,fetus,placenta and umbilical cord venous blood belonging to pregnant women and fetuses.The internal structure of the tissue compartment except umbilical vein blood is the same as that of adult solid tissue compartment.The special physiological connection of placenta forms the material exchange between maternal plasma in villi space and fetal trophoblast cells.We introduced changes in age-related anatomical and physiological parameters into the model,and corrected changes in plasma unbound fraction and metabolic enzyme activity during pregnancy.Three pregnancy virtual groups were created by PBPK model,including 1st trimester(1-11 weeks),2nd trimester(12-26 weeks)and 3rd trimester(27-38 weeks)according to gestational weeks,with a non-pregnant population(20-40 years old).Each virtual group contained 1000 individuals.The model was used to predict the steady-state pharmacokinetics of non-pregnant women and pregnant women under multiple dosing with a fixed dose.We optimized the doses for pregnant population referring to the plasma exposure in non-pregnant women and recommended therapeutic window.4.Systematic evaluation of extrapyramidal adverse reactions in children and fetal malformation induced by the three drugs.Literatures were searched in PubMed,Web of Science,Embase,Cochrane library,CNKI,Wanfang and VIP databases according to the predetermined search strategy.The extrapyramidal adverse reactions in children were limited to randomized controlled trials,and teratogenic studies are generally cohort studies or case-control studies.Literature screening criteria were set according to the PICOS principle,and information such as the first author,publication years,number of researchers,age,gender information of patients in the group,dosage of related drugs,and outcome indicators of the study were extracted.Randomized controlled trial evaluation using Cochrane Handbook recommended RCT evaluation tools,divided into low risk of bias,moderate risk of bias,high risk of bias or bias risk unclear;the cohort study or casecontrol study were evaluated by the Newcastle scale,which included the selection of cases,the definition of comparability and exposure between the two groups of cases.The enumeration data were combined in the form of mean difference,and the measurement data were combined in the form of ratio.R 3.6.1 software was used for mesh meta-analysis,and Revman 5.3 software was used for traditional meta-analysis.Results1.Adult PBPK model development and verificationThe adult PBPK models of the three drugs were developed.The olanzapine model well simulated PK profile of a 10 mg single oral dose,and the prediction error for Cmax and AUC was less than 12%.The AUC ratio(AUCR)predicted by the model was 1.88 when co-administered with fluvoxamine,a CYP1A2 inhibitor,while the reported value was 1.76.Nine population PBPK simulations for the validation set showed that more than 95% of the predicted drug concentrations were within 2 times error range of the measured values,and about 68% were within 1.25 times error range.MAPE of the plasma concentrations for all validation sets was less than 27%,while GMFE of Cmax and AUC was 1.13 and 1.07,respectively.The quetiapine model well simulated the PK after a 25 mg single oral dose and multiple administration using incremental doses.Model verification results showed that 92% of the predicted drug concentrations were within 2 times error range of the measured values,and about 60% were within 1.25 times error range.The GMFE of Cmax was 1.21,and that of AUC was 1.05.MAPE of the plasma concentrations for all validation sets was less than 43%.The test set of aripiprazole and dehydroaripiprazole included PK reports after intravenous and oral dose of 3 mg,and the prediction error for Cmax and AUC by the model was less than 4%.Eight clinical studies were used for model validation,and more than 85% and 50% of aripiprazole and dehydroaripiprazole plasma concentrations,respectively,were within 1.25 times error range.The GMFE of Cmax and AUC were 1.11 and 1.07,respectively.Except one study,the MAPE of PBPK predictions was less than 23%.In addition,the PK properties of aripiprazole in different CYP2D6 phenotypes and in combination with itraconazole,a potent CYP3A4 inhibitor,were tested.The prediction error of Cmax R and AUCR was less than 15%.Therefore,we successfully established adult PBPK models of olanzapine,quetiapine and aripiprazole and active aripiprazole metabolite,which can be used for scaling to special populations.2.Prediction of pharmacokinetics in children and pediatric dose adjustment using PBPK modelingThe PBPK model of test compound oxycodone successfully predicted the pharmacokinetics of oxycodone in adult population.The predicted AUC,Cmax and tmax were in the range of 0.7–1.3 times of the reported values,and the GMFE was 1.12,1.16 and 1.23,respectively.After correcting the changes in anatomy and physiology,plasma protein binding,and drug clearance,the model had good prediction performance for PK in children,and the GMFE of AUC was 1.17.According to the application of PBPK model,we found that different intravenous doses should be given in children of different ages compared to a standard 0.1 mg/kg in adults,while a progressive increasing dose with age growth following oral administration should be adopted.The extrapolation platform can be used for pediatric dose design of three antipsychotic drugs.The PBPK model was used to simulate the pharmacokinetics of olanzapine in 8 children aged 12-18 years.The prediction error of Cmax and AUC was less than 25%.Compared with the 10 mg daily dose in adult population,the doses for adolescents and children aged 6–12 should be reduced by 25% and 50%,respectively.The doses for preschool aged 2–6 should be reduced by 50%–75%.In the pediatric simulations of quetiapine,the GMFE of parameters Cmax-ss,AUCτ-ss and tmax were less than 1.50.Compared with 200 mg twice daily dose for adults and adolescents,children aged 6–12 should reduce the dose by 37.5% to 125 mg,and children aged 2–6 should reduce the dose by 62.5% to 75 mg.The GMFE of Cmax-ss and AUCτ-ss of aripiprazole and dehydroaripiprazole were less than 1.40.The target daily dose of 15 mg for adults should be reduced by 1/6,1/2 and 2/3,i.e.,12.5 mg,7.5 mg and 5 mg,for adolescents,school-age and preschool children,respectively.The model may overestimate the plasma concentrations of quetiapine and aripiprazole in children.The individual variation and plasma concentration fluctuation in children were significantly greater than those in adults,and the individual variation grew with the decrease of age.3.Prediction of pharmacokinetics and dose adjustment in pregnant women using PBPK modelingThe PBPK model predicted that under olanzapine 10 mg daily dose,the steadystate trough concentrations of olanzapine in the 1st,2nd and 3rd trimester of pregnancy were decreased by 11.9%,14.9% and 18.7%,respectively,but the steady-state concentrations could attain the effective treatment limit(20 ng/m L).The steady-state Cmax,AUCτ-ss and half-life of olanzapine had little change(less than 17.5%)throughout pregnancy.Therefore,dose adjustment is not needed for olanzapine during pregnancy.The pregnancy PBPK model basically recovered the TDM results of quetiapine and aripiprazole,demonstrating a progressively lower system exposure with gestational age.When quetiapine was administered 200 mg twice a day,the model predicted that the mean maximum steady-state concentration in the 1st,2nd and 3rd trimester decreased by 9.9%,34% and 48%,respectively,and the steady-state AUC decreased by 8.7%,35% and 49%,respectively.During the 3rd trimester of pregnancy,plasma concentrations cannot reach 100 ng/m L for more than half the time of a dosing interval.Compared to 200 mg twice daily for the nonpregnant patients,a 50% increased dose in the 2nd trimester and a 150% increase in the 3rd trimester should be administered to achieve comparable steady-state system exposure.For aripiprazole,the prediction showed that the blood concentrations of the parent drug were significantly decreased in women in the 2nd and 3rd trimesters of pregnancy,and the steady-state concentration cannot reach the lower limit of the therapeutic range.The blood concentrations of active moiety(aripiprazole + dehydroaripiprazole)in late pregnancy could not reached the lower limit of the recommended range(150 ng/m L).The steadystate Cmax and AUC of aripiprazole and dehydroaripiprazole decreased by 33%–43% in the 2nd trimester and 55%–62% in the 3rd trimester.With the daily dose of 15 mg in non-pregnant women as the reference,25 mg per day in the 2nd trimester and 40 mg per day in the 3rd trimester produced approximately the same fluctuation of aripiprazole concentration.The pharmacokinetic parameters were within 15% among the three scenarios.Doubling the dose in the 3rd trimester(30 mg q.d.)brings the mean steady-state trough concentrations exactly to the lower limit of the therapeutic range,while the generated AUCτ-ss is approximately 25% lower than at baseline.4.Systematic evaluation of extrapyramidal responses in children and teratogenicity in pregnant populationsAfter literature retrieval and screening,32 pediatric extrapyramidal reactions and 7 teratogenic reports were included.Quantitative analysis of extrapyramidal reactions in children found that aripiprazole had a higher risk of Parkinson-like symptoms than olanzapine(ln OR = 16.26(2.03,46.87)).For extrapyramidal events,sedentary inability and dystonia,though the risks of corresponding symptoms were different,no statistical difference was found.For the pregnant women using the above drugs,the OR values obtained by calculating the comparison between the exposed group and the non-exposed group showed that only quetiapine slightly increased the risk of neonatal congenital malformations,but the baseline information of the cohort population was not available to correct the final risk ratio with the hybrid factor.The most credible evidence comes from a large sample cohort study,indicating that these three drugs did not increase the risk of neonatal malformations.ConclusionIn this study,we used PBPK modeling and simulation to predict the pharmacokinetics and guide dose adjustment of olanzapine,quetiapine and aripiprazole in children and pregnant women prospectively.The clinical pharmacokinetic studies of the three drugs provide support for model development and evaluation,from which the reliability for prediction was confirmed.The models suggest that the doses need to be reduced in children,depending on drugs and children’s age.Olanzapine dose should need no adjustment during pregnancy,while the doses for quetiapine and aripiprazole should be increased in the latter two trimesters to maintain the therapeutic concentration.This study provides the possibility of using PBPK model to guide dose design in future clinical studies in children and pregnant women and provides evidence for model-guided pharmacotherapy decisions.In addition,this study enriches the evidence for the application of PBPK method in rational drug use,and further verifies the reliability of physiology-based extrapolation method.With the continuous accumulation of such evidence,it is possible for modeling and simulation technology to assist drug decision-making in clinical practice.