2012/3/28 Outline Population PK/PD in Taiwan (from Academic - - PDF document

2012/3/28 1

Population PK/PD in Taiwan (from Academic Viewpoint)

林君榮

Chun-Jung Lin School of Pharmacy, College of Medicine National Taiwan University

Brief introduction International population PK/PD groups Population PK/PD in Taiwan

Outline

Population PK/PD in Taiwan Application of population PK/PD in clinical research Dosage recommendation of phenytoin Dose response of rabeprazole Population Approach Group Europe (first meeting in 1992)

Twenty‐first meeting Twenty first meeting Venice, Italy, 5‐8 June, 2012

http://www.page-meeting.org/default.asp

PAGANZ

Population Approach Group in Australia and New Zealand (first meeting in 1999) since 2006

PAGJA

(first meeting in 2008) since 2008 The first World Conference on Pharmacometrics (WCoP) Seoul, Korea, 5-7 September, 2012.

htt // / http://www.go-wcop.org/ Marc Gastonguay, Metrum Research Group, USA Nick Holford, University of Auckland, New Zealand Mats Karlsson, Uppsala University, Sweden Holly Kimko, Janssen Research & Development, USA France Mentre, Unversite Paris Diderot, France Kyungsoo Park, Yonsei University, Korea (延世大學) Goonaseelan Colin Pillai, Novartis, Switzerland Organizing Committee

2012/3/28 2

Academic

1. 陳瑞龍教授、李勇進教授、蔡義弘教授、許明照教授、林其和教授…等 2. Population PK/PD is probably not the major research (1/3 or less) 3. Different approaches in population PK/PD studies

Industry

1. Less experience in population PK/PD? 2. Major in BA/BE studies for local industry

Population PK/PD in Taiwan

3. Some phase1 , phase2 and phase 3 clinical trials

Regulatory (Center for Drug Evaluation; CDE)

1. Based on US FDA experience 2. Population PK training

Previous workshop

Concepts, Methodologies and Applications of Population Pharmacokinetics in the Clinical Studies April 27, 2001, Tsang-Bin Tzeng (曾滄濱博士)

Why do we want to use population PK/PD approach in research?

sparse data/sample available unbalanced design Outpatients cancer patients neonatal patients parkinsonian patients

PK or PK/PD properties ?

Pharmacostatistical Modeling

Pharmacostatistical Models Structure models Statistical models Structure models Statistical models PK(PD) Model Fixed Effects (theta; θ) Interindividual Variability Intraindividual (Residual) Variability

PK (CL/F, V/F, Ka) Bl d L l

Fixed Effects: Genotypes, Gender, Age, Weight, BSA Concomitant drugs, Renal/Liver functions …etc

Blood Levels Clinical Response

Random Effects: Intersubject variation Intrasubject variation

How to approach population PK ?

• The two-stage approach
• Naïve pooled data (NPD) approach
• NONlinear Mixed Effects Modeling approach

(NONMEM, Beal & Sheiner, 1982)

• Nonparametric maximum likelihood method

(Mallet 1986)

• Bayesian methods

(Racine-Pion & Smith, 1990)

• Variants of the NONMEM

(Amisaki & Tatsuhara, 1988) (Lindstrom & Bates, 1990)

Example 1 Dosage Recommendation of Phenytoin for Patients with Epilepsy with Different CYP2C9/CYP2C19 Polymorphisms

Therapeutic Drug Monitoring 2004; 26:534-540

洪靚娟 劉宏輝

(洪靚娟；劉宏輝)

2012/3/28 3

Therapeutic Range: 10 -20 g/ml Low-extraction ratio (Clearance  fu  CLi) Nonlinear Pharmacokinetic Properties Metabolism of phenytoin can be described by the Michaelis-Menten equation： （Vmax）（Cpss） R ＝ Km＋Cpss R：daily dose (mg/day) Cpss：concentrations at steady state (g/mL) Vmax：maximal metabolic rates (mg/day) Km：Michaelis-Menten constant (g/mL)

CYP2C9:

CYP2C9*1 :wild type CYP2C9*2 : exon 3 C430T (Arg144/Cys) CYP2C9*3 : exon 7 A1075C (CIle359/Leu) Phenytoin (S)-p-HPPH (70-80% of dose)

CYP2C19:

Phenytoin (R)-p-HPPH (5-8% of dose) CYP2C19*1: wild type CYP2C19*2: exon 5 G681A (splicing defect in exon5) CYP2C19*3: exon 4 G636A (TGG-->TGA create a premature stop code) EM Extensive Metabolizers PM Poor Metabolizers

Methods

Patients with epilepsy (n = 169; 505 data) CYP2C9/CYP2C19 genotypes Concentrations of phenytoin, (S)-p-HPPH and (R)-p-HPPH Data analysis by NONMEM (version V)

$SUBROUTINE

(define the model by ADVAN, TRANS or user PRED) The PREDPP library includes five pre-programmed subroutines:

ADVAN1

• ne compartment linear model

ADVAN2

• ne compartment linear model with first order absorption

ADVAN3

two compartment linear model

ADVAN4

two compartment linear model with first order absorption

ADVAN10

• ne compartment model with Michaelis-Menten elimination

ADVAN10

• ne compartment model with Michaelis Menten elimination

Other ADVAN routines that need additional info (e.g., differential equations) to be defined in $MODEL and/or $DES

ADVAN5 general linear

ADVAN6 general nonlinear

ADVAN7

general linear with real eigenvalues

ADVAN8

general nonlinear kinetics with stiff equations

ADVAN9

nonlinear kinetics with equilibrium compartments

2012/3/28 4

GI A(1) Body A(2)

k12

ADVAN6 (MODEL and DES required)

V = (Vm x Cpss) / (km+Cpss) ss m ss m

C K C V A k dt dA A k dt dA      

1 12 2 1 12 1

$PROB Css of phenytoin in Taiwanese (PM and EM) $INPUT ID AGE SEX HT WT EVID TIME AMT SS II DV MDV GT $DATA data $SUBR ADVAN6 TOL=5 $MODEL COMP=(DEPOT, DEFDOS), COMP=(CENTRAL, DEFOBS) $PK $DES C2=A(2)/V2 DADT(1)=-K12*A(1) DADT(1)=-K12 A(1) DADT(2)=K12*A(1)-VM*C2/(KM+C2) $ERROR $THETA $OMEGA $SIGMA $EST $TABLE $COVARIANCE $SCATTER

Minimal value of the objective function：

• 2 log likelihood

Covariates inclusion：minimal value of the objective function  3.84 j Covariates exclusion：minimal value of the objective function  7.88

Objective Function Value

OBJ= - 2 LL = - 2 times the maximum log of the likelihood of the data To compare two models (e.g., differ in a parameter), we camper (-2LL). The distribution of -2LL is about chi-squared distribution with n degree

• f freedom (n = difference in # of parameters between models (e.g., n=1 for
• ne parameter difference in two models).

Difference in # of parameters (n) -2LL P=value 1 > 3.84 < 0.05 1 > 6.63 < 0.01 1 > 7.88 < 0.005 2 > 5.99 < 0.05 2 > 9.21 < 0.01 2 > 10.6 < 0.005

AIC = (-2LL)+2p where p = # of parameters in model AIC = (-2LLA)-(-2LLB)+2(PA-PB)

NONMEM Model

Final model： V’maxij= Vmax *(WTij/60)θpw*GTij

Vmax

K’mij = Km * GTij

Km

For intraindividual variation: Rij=R’ ij*(1+εij) For interindividual variation: Vmaxij=V’maxij*(1+ηj

Vmax)

Kmij= K’mij*(1+ηj

Km)

Group CYP2C9 CYP2C19 CYP2C9 and CYP2C19 phenotype G1 *1 / *1 *1 / *1 CYP2C9EM / CYP2C19EM G2 *1 / *1 *1 / *2 CYP2C9EM / CYP2C19EM *1 / *3 CYP2C9EM / CYP2C19EM G3 *1 / *1 *2 / *2 CYP2C9EM / CYP2C19PM G3 1 / 1 2 / 2 CYP2C9EM / CYP2C19PM *2 / *3 CYP2C9EM / CYP2C19PM G4 *1 / *3 *1 / *1 CYP2C9PM / CYP2C19EM *1 / *2 CYP2C9PM / CYP2C19EM *1 / *3 CYP2C9PM / CYP2C19EM G5 *1 / *3 *2 / *3 CYP2C9PM / CYP2C19PM

2012/3/28 5

Group Group Km Km (μg μg/ /mL mL) ) Vmax Vmax (mg/kg/day) (mg/kg/day) Vmax Vmax/Km /Km (mL mL/kg/day) /kg/day) G1 G1 8.15 8.15 (7.99 (7.99-

• 8.31)

8.31) 10.01 10.01 (9.83 (9.83-

• 10.19)

10.19) 1230 123010 10 G2 G2 9.03 9.03 (8.91 (8.91-

• 9.15)

9.15) 9.77 9.77 (9.65 (9.65-

• 9.89)

9.89) 1081 108110 10 G3 G3 9.38 9.38 a (9.20 (9.20-

• 9.56)

9.56) 9.18 9.18a (9.03 (9.03-

• 9.33)

9.33) 976 9767 a G4 G4 10.38 10.38 a,b

a,b

(10.21 (10.21-

• 10.55)

10.55) 6.31 6.31 a,b,c

a,b,c

(6.20 (6.20-

• 6.42)

6.42) 607 6076 a,b,c

a,b,c

G5 G5 15.63 15.63 5.43 5.43 348 348

a Compared to G1, p＜0.05 b Compared to G2, p＜0.05 c Compared to G3, p＜0.05

concentration (mg/L)

20 30 40 50

G5 G4 G3 G2 G1

G4 G3 G2 G1 G5

Daily Dose (mg/kg/day)

1 2 3 4 5 6 7 8 9

Phenytoin c

10

G1 patients would be 5.5-7 mg/kg/day G2 patients would be 5-7 mg/kg/day

The recommend doses of phenytoin for patients with epilepsy

G3 patients would be 5-6 mg/kg/day G4 patients would be 3-4 mg/kg/day G5 patients would be 2-3 mg/kg/day

entration (mg/L)

30 40 50

G5 G4 G3 G2 G1 daily dose (mg/kg/day)

2 4 6 8

phenytoin conce

10 20

Example 2 Pharmacokinetic-Pharmacodynamic analysis

• f the role of CYP2C19 genotypes

in rabeprazole-based triple therapy against Helicobacter pylori

British Journal of Clinical Pharmacology 2009; 67:503-510 (楊喻帆；楊智欽)

Structures of proton pump inhibitors

benzimidazole pyridine methylsulfinyl group

2012/3/28 6

N N N H S O C H3 O OMe

N N N H S N N N H S O N N N H S O O

Rabeprazole CYP2C19 CYP3A4

N C H3 O OMe N C H3 O OH N C H3 O OMe N N N H S C H3 O OH N N N H S C H3 O COOH

Rabeprazole Thioether Demethylated-rabeprazole Rabeprazole Sulfone Demethylated thioether rabeprazole Thioether carboxylic acid rabeprazole

ration (ng/ml) 100 1000 10000 ncentration (ng/ml) 100 1000 10000

Plasma concentrations of rabeprazole and rabeprazole thioether

• n day-1 and day-4 in CYP2C19 PMs (●) and EMs (○)

TIME (hr) 4 8 12 72 76 80 84 Rabeprazole Concentr 0.1 1 10 100 TIME (hr) 4 8 12 72 76 80 84 Rabeprazole Thioether Con 0.1 1 10 100

Rabeprazole Rabeprazole thioether

Rabeprazole 20 mg bid ntration (pg/ml) 200 250 300

Gastrin concentrations on day-1 and day-4 in CYP2C19 PMs (●) and EMs (○)

Time (hour)

5 10 15 70 75 80 85 90 95 100

Plasma Gastrin Concen

50 100 150 Rabeprazole 20 mg bid n (pg/ml) 150 200 n (pg/ml) 140 160 180 200

Correlation between plasma rabeprazole and gastrin concentrations in CYP2C19 PMs (●) and EMs (○)

Rabeprazole Concentration (ng/ml) 500 1000 1500 2000 Gastrin Concentration 50 100 Rabeprazole Concentration (ng/ml) 500 1000 1500 2000 Gastrin Concentration 20 40 60 80 100 120 Rabeprazole 20 mg bid

Effect compartment in direct-link model

1 Effect Ce GI k1e keo Ka k

] ) )( ( ) )( ( ) )( ( [

1

K K K K e K K K K e K K K K e V FD K K C

a eo t K eo eo a Keot a a eo t Ka a eo e

        

    

2012/3/28 7

Pharmacokinetics Time vs. Conc. Time Blood Conc. Effect Ce PK/PD Ce vs. Effect Effect Blood Conc. PK/PD Conc vs. Effect

Time Effect

PK/PD Time vs. Effect ADVAN2

• ne compartment linear model with first order absorption

(1) Population pharmacokinetic analysis

Ka K A (GI) A (1)

) 1 ( ) ( ) 1 ( ) ( ) ( A k GI A ka dt dA GI A ka dt GI dA       

$PROBLEM Population PK Analysis of RABEPRAZOLE day1 $DATA 4.csv IGNORE= # $INPUT ID TIME EVID AMT DV MDV AGE SEX HT WT GT BSA $SUBROUTINES ADVAN2 $PK $ERROR $THETA $OMEGA $SIGMA $ESTIMATION $COVA $TABLE

 

  e e

BSA CL GT

BSA

   



CL

Final Model of PK analysis

  

  e Ka e e Vd

ka BSA Vd

BSA

    



ADVAN6 General Nonlinear Model; The drug is distributed between compartments according to first-order processes

(2) Population PK/PD Analysis (direct model)

A (1) GI A (2) C K12 K23 A (3) E

  

) ( ) ( ) ( ) 3 ( ) 2 ( dA(3) ) 2 ( ) 1 ( ) 2 ( ) 1 ( ) 1 (

50 max 30 23 12 12

E C EC E C E E E E A k A k dt A k A k dt dA A k dt dA                 K20 K30 Effect

$PROB PK & PD Day1 $DATA 4PD112334.csv $INPUT ID AGE SEX HT WT TIME AMT GT MDV GAS=DV V CL KA ALAG K $SUBROUTINE ADVAN6 TOL=3 $MODEL COMP(DEPOT) COMP(CENTRAL) COMP(EFFECT) $PK $DES DADT(1)= -A(1) * K12 DADT(2)= A(1) * K12 - A(2) * K20 DADT(3)= A(2) * K23 - A(3) * K30 $ERROR $ERROR

CON= A(3)/V3; Ce

TY= E0 + (EMAX-E0) * (CON ** GAM)/(C50 ** GAM + CON** GAM)

IPRED = TY IRES = DV- TY Y= TY * (1 + ERR(1)) + ERR(2)

$THETA $OMEGA $SIGMA $ESTIMATION $COVA $TABLE

2012/3/28 8

 

  e EC e

E

   

max

max

E

Final Model of PD analysis (direct model)

   

      e e k e E e EC

keo GT eo E EC

        

50

50

n] (pg/ml)

100 120 140 160 180

Time after dosing (hr)

2 4 6 8 10 12

[Gastrin

20 40 60 80 100 Day-1 EM Day-1 PM Day-4 EM Day-4 PM Day-7 EM Day-7 PM

結論

Population pharmacokinetics 分析可用於評估 影響藥物濃度之參數，其結果可應用於臨床給藥 劑量的調整。 以phenytoin的研究為例，可以參考劑量濃度分 佈，只要知道病人的基因型，即可根據其基因型 Population PK/PD 分析可用於評估影響藥物反 應之參數並找出濃度與反應之關係。 以 PPI 於幽門桿菌感染治療之研究為例， 可知 不同基因型之病人在不同服藥期間之反應，可作 為療程設計之參考。 佈 只要知道病人的基因型 即可根據其基因型 來選擇安全而有效的劑量，避免副作用的產生 Academic Industry Regulatory

敬 請 指 教