Using Data under Drug Development Yoshitaka Yano Kyoto Pharm. - PowerPoint PPT Presentation

Prediction of Human Pharmacokinetic Profiles Using Data under Drug Development Yoshitaka Yano Kyoto Pharm. Univ., Japan - WCoP 2012, Seoul, Korea - 1

Purpose of PK Prediction • During drug development, we would like to.. – determine FTIH dose regimens, – by predicting PK (and hopefully PD) parameters in human, – by predicting PK (and hopefully PD) in profiles human. • Several prediction strategies have been proposed – Parameters : Regression of chemical / biological Information – Profiles : Model-based approach 2

Prediction of PK Profiles • Available data for a new compound: – Chemical properties of the compound • Log P, Molecular weight etc. – PK data (PK parameters) in animals • CL, Vss, Cmax, Tmax, etc. • Css-MRT Method to predict PK profile • Normalized Curve: y = Conc. / Css, x = time / MRT where Css = Dose / Vss = Dose / (AUC / MRT) 3

Topics • Review of Css-MRT Method for PK Prediction • Prediction of Oral PK Profiles • Prediction of Tissue Distribution Profiles • Prediction of Pediatric PK from Adults’ PK 4

Css-MRT Method (for i.v. prediction) i.v. PK Profiles Calc. Vss, CL of Animals and Css, MRT Normalized Profile Model Fitting (e.g. exponentials) Predicted Vss, CL .. Back- Any Prediction Methods and Css, MRT Normalization - e.g. QSAR-based Wajima-Method Log (CL man ) = function( log (CL rat ), log(CL dog ), MW, etc.) Log (Vss man ) = function( log (Vss rat ), log(Vss dog ), MW, etc.) Wajima, Yano et al., J.P.S. 93 : 1890- (2004). Wajima, Fukumura et al., J.P.S. 91 : 2489- (2002). Wajima, Fukumura et al., J. Pharm. Pharmacol. 55 : 939- (2003). 5

Examples of Css-MRT Method • In Css-MRT method, predictability of PK profiles depends on accuracy of PK parameter Prediction. a: CL overestimated, b: CL underestimated, c: acceptable CL, Vss estimation, d: CL underestimated, Vss overestimated 6

Topics • Review of Css-MRT Method • Prediction of Oral PK Profiles • Prediction of Tissue Distribution Profiles • Prediction of Pediatric PK from Adults’ PK 7

Convolution Theory for Predicting Human PK Profiles • Function for IV PK profiles predicted by Css-MRT method can be a core module for prediction system. • Under linear PK assumption, convolution theory can be applied. Input Function Weighting Function Output Function Absorption process Intravenous PK Profile Oral PK Profile Intravenous PK Profile Tissue Distribution Process Tissue PK Profile 8

Cmax, Tmax are More Frequently Available • For PO prediction, F (bioavailability) and Ka required – Some approaches are reported; • Predicted F (BA) by regression, averaged Ka in animals. – Futa et. al., Biopharm. Drug Disp. 29 : 455- (2008) – Review by Vuppugalla et al., J.P.S. 100 : 4111- (2011) Cmax • Tmax, Cmax Ka – F, Ka of drugs are not Slope = Cmax / Tmax necessarily reported. – Tmax, Cmax are Tmax reported more frequently. 9 9

Prediction of Human Cmax by Regression • Same Concept as Wajima’s Regression – Use dose normalized Cmax (and Tmax) in rat and dog. • ln(Cmax/Dose|man) = function {ln(Cmax/Dose|rat), ln(Cmax/Dose|dog), MW, cLogP, etc.} Correlation Plot for ln(Cmax/Dose|man) 10

Prediction of Human Cmax and Tmax • Cmax: Prediction using regression results – ln(Cmax/Dose|man) = 2.2 + 0.62 * ln(Cmax/Dose|dog) + .. • Tmax: – Low correlation with animals – Tmax, (and also Cmax) depends on sampling design – Possible ways of prediction • Regression based • Same as those in dog • Arbitrarily choose • … 11

FILT and Convolution Theory • Transfer function – Laplace transform of weighting function • Convolution: f 1 (s) * f 2 (s) f 1 (s) f 2 (s) • FILT (Fast Inverse Laplace Transform) algorithm – for model-based convolution / deconvolution – FORTRAN program available • FILT algorithm: Hosono, Radio Sci. , 16 : 1015- (1981). • Application of FILT to PK Analysis: Yano, Yamaoka et al., Chem. Pharm. Bull. , 37 : 1035- (1989).         1 f t L F s             a   e   a j n 0 . 5      n   f t F F 1 Im  F  n n     t t 12  n 1

Modeling of Oral Absorption Process • Convolution - Transfer function in Laplace domain ( f a (s) , f p (s) ) Absorption, f a (s) Plasma(IV), f p (s) • f a (s) : Laplace transform of f a (t) - for absorption process • f p (s) : Laplace transform of f p (t) - predicted or observed iv plasma profile • Modeling of f a (s) ; e.g. Mono-exponential  F k     A B    a , app f s , f s ,      a p s k s s a , app    F k   A B      a , app C s        p   s k s s a , app      dC t F k   A B         p a , app s C s s        p   dt s k s s a , app 13 13

Simulation of Plasma Profile after Oral Dose • Estimation of F and Ka,app – IV profile assumed to be predicted by Css-MRT method – Simultaneous fitting of an example data (Tmax, Cmax) = (2.0, 2.35), (Tmax, dCp(t)/dt| t=Tmax ) = (2.0, 0) – Estimated F = 0.472, k a,app = 0.562 • M&S with FILT    F k   A B      a , app C s        p   s k s s a , app      dC t F k   A B         p a , app s C s s        p   dt s k s s a , app Predicted PO profile (solid curve). 14

Topics • Review of Css-MRT Method for PK Prediction • Prediction of Oral PK Profiles • Prediction of Tissue Distribution Profiles • Prediction of Pediatric PK from Adults’ PK 15

Tissue Distribution Parameters for Convolution • Convolution - Transfer function in Laplace domain ( f p (s) , f t (s) ) Plasma, f p (s) Tissue, f t (s) • f p (s) : Laplace transform of f p (t) - (predicted or observed) plasma iv profile • f t (s) : Laplace transform of f t (t) - function for tissue distribution kinetics • Modeling of f t (s) ; e.g. Mono-exponential • Kp = AUC tissue / AUC plasma , MTT = MRT tissue – MRT plasma   1 K MTT     A B    p f s , f s ,       t p 1 s MTT s s     1 K MTT   A B      p C s         t 1   s s s MTT 16

Simulations of Tissue PK Profiles by FILT • Modeling for Ft(s): e.g. mono-exponential; – Kp = 2, MTT = 0.1 (rapid equilibrium), 1.0, 5.0 (slow eq.) Bottom panels are the weighting functions. 17

Topics • Review of Css-MRT Method for PK Prediction • Prediction of Oral PK Profiles • Prediction of Tissue Distribution Profiles • Prediction of Pediatric PK from Adults’ PK 18

Pediatric PK Prediction using Adults’ PK Data Plasma PK Profiles in Adults and Pediatrics for a Series of Drugs - Calc. Mean of Vss and CL for each drug Allometric Regression for Vss and CL vs. Body Weight Relationships Using Mixed-Effect Modeling            for j-th drug, k-th subject log CL log a b log BW jk j j jk jk           log a log a b b j a , j j b , j Example in 15 beta-lactam antibiotics b = 0.670 (Vss) b = 0.565 (CL) CL (L/hr) Vss (L) BW (kg) BW (kg) 19 Shimamura, Wajima, Yano, J.P.S. 96 : 3125- (2007).

Pediatric PK Prediction using Adults’ PK Data Allometric Regression for Vss and CL vs. Body Weight Relationships Using Mixed-Effect Modeling Adults PK of New Drug Conc. Empirical Bayes Estimation of a j and b j using Adults Vss and CL for New Drug Time (hr) Estimate Individual Vss and CL for a Pediatric Patient using BW, Predicted Conc. and Prediction by Css-MRT Method Details were presented at, “International symposium in drug development -Modeling & Simulation in Drug Development and Clinical Applications”, 20 20 Yonsei University in Seoul, 15-16 November, 2006. Observed Conc.

Summary (1) • Today’s presentation shows only the ideas of new types of modeling and simulations focused on the use of FILT. • These methods are practical because the data are all available during routine experiments for drug development. • Predictability (reliability) of the methods should be more precisely and systematically evaluated. 21

Summary (2) • Intravenous PK profiles predicted by Css-MRT method can be a core module of the integrated prediction system . • Any methods for prediction of PK parameters can be combined (as a module) with the Css-MRT method. – Flexible Integrated prediction system . • Prediction is never perfect, results should be given with predictive (credible) region. 22

Acknowledgements • Collaborators – Toshihiro Wajima Ph.D., Shionogi & Co. Ltd., Japan – Kenji Shimamura, Shionogi & Co. Ltd., Japan – Shunsuke Kawabe, Kyoto Pharm. Univ., Japan • Special Friend – Atsunori Kaibara Ph.D., Astellas, Japan. • This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (22590153). 23