Human Vss Prediction. James Yates AstraZeneca. Disclaimer 10/25/2011

Transcription

1 0/25/20 Human Vss Prediction Part 2: Comparative Assessment of Prediction Methods of Human Volume of Distribution R.D.O JONES,H.M. JONES, M. ROWLAND, C.R. GIBSON, J.W.T. YATES, J.Y. CHIEN, B. J. RING, K. K. ADKISON, M.S. KU, H. HE, R. VUPPUGALLA, P. MARATHE, V. FISCHER, S. DUTTA,V. K. SINHA, T. BJӦRNSSON, T. LAVE, P. POULIN James Yates AstraZeneca Disclaimer The views and opinions expressed in the following PowerPoint slides are those of the individual id presenter and should not be attributed t to Drug Information Association, Inc. ( DIA ), its directors, officers, employees, volunteers, members, chapters, councils, Special Interest Area Communities or affiliates, or any organization with which the presenter is employed or affiliated. These PowerPoint slides are the intellectual property of the individual presenter and are protected under the copyright laws of the United States of America and other countries. Used by permission. i All rights reserved. Drug Information Association, DIA and DIA logo are registered trademarks or trademarks of Drug Information Association Inc. All other trademarks are the property of their respective owners. Drug Information Association 2

2 0/25/20 Introduction What did we do Criteria for model selection What were the results Identify those that methods performed well Literature comparison Outlier analysis Conclusions Criteria for Models Selected Selection of Empirical, semi-mechanistic and mechanistic models Represent range of input data generated typically at different stages of a drug discovery project Drug Information Association 4 2

3 0/25/20 Models Selected: Empirical Pre-clinical Vdss, fu and Fut Drug Information Association 5 Models Selected: Semi- Mechanistic Oie-Tozer QSAR Pre-clinical Vdss, fu and Fut, physiological concepts PBPK based Drug Information Association 6 3

4 0/25/20 Models Selected: Mechanistic Poulin Physchem, Tissue composition, fup Berezhkovskiy Rodgers and Rowland Drug Information Association 7 Results I: Bird s eye view Empirical Semi- Mechanistic Mechanistic 00 Obs Vss 0 0. }2-fold error Pred Vss (OT-QSAR) Some methods very close to -fold error on average. However some of these have very wide range around Drug Information Association 8 4

5 0/25/20 Results II: Empirical 65% <2-fold 7%<3-fold Simple Allometry correcting for fup 59% <2-fold 77%<3-fold Simple Allometry 65%<2-fold 94%<3-fold Drug Information Association 9 Results III: Semi-Mechanistic 78%< 2-fold 89%< 3-fold 6%< 2-fold 83%< 3-fold 65%< 2-fold 82%< 3-fold Drug Information Association 0 5

6 0/25/20 Results IV: Mechanistic 6% <2-fold 72%<3-fold 50% <2-fold 78%<3-fold Drug Information Association Results V: Summary 50% <2-fold 78%<3-fold 65%<2-fold 94%<3-fold 78%< 2-fold 89%< 3-fold 65%< 2-fold 82%< 3-fold Drug Information Association 2 6

7 0/25/20 Outlier Analysis: Two or more methods 0-fold error All methods applied regardless of whether assumptions apply to a compound #36: Modest lipophilicity (logp=.48), yet extensive Vdss. Impacts predictions by mechanistic models #48: Highly bound (<0.8% free) acid with high lipophilicity (logp=5.35). Low in vivo volume is indicated d by rat volume #9: significant differences across species in Vss and Fup. Free volume is conserved. Allometry performed well.why TC thrown out? Drug Information Association 3 Compared to Literature Findings Performance of tissue composition models is lower than reported in the literature. The data set has a number of Lipophilic bases Models that rely upon single species or two species perform well: Consistent with the findings of Fagerholm and Hosea et al. Additional species data doesn t necessarily offer any benefit Oie-Tozer performance in line with other s experience Drug Information Association 4 7

8 /25/20 CASE STUDY I Base, LogP=3.66, LogD=.53 Between species variation in fup and Vdss: Free volume Species es Vdss (L/kg) (/g) fup (L/kg) (/g) Mouse Rat Dog Human Free volume reasonably well conserved points to a reasonably large volume in man 0. Modest lipophilicity it would generate prediction of a moderate volume this 0.0 would be invalidated by inspecting 0.00 pre-clinical data Plasma concentration (um) Plasma concentration (um) Plasma concentration (um) Rat IV and AR PBPK Dog IV Human IV Drug Information Association 5 CASE STUDY II AZD6302 Pre-clinical data Rat Vss 3 to 5 lkg - CL 8 mlmin - kg - CL 2 %LBF F 30 to 50 % T /2 > 0 h Mouse Vss 0.6 lkg - CL 4.7 mlmin - kg - CL 3%LBF F 30 %? T /2 2.4 h Dog Vss 0.2 lkg - CL.5 mlmin - kg - CL 4 %LBF F > 30 % T /2 2 h Rat (BDC) Vss 2 lkg - CL 3 mlmin - kg - CL 43 %LBF F? % T /2 2 h Rat Vss is significantly different to that expected based on mouse and dog and fu differences. Billiary elimination clear (>40% Cltot into bile) can this process be captured in a model? Drug Information Association 6 8

9 0/25/20 Standard AR (Arundel Model) 00 Standard model: Rat IV kinetics are not well captured PBPK model simulations compared against rat in vivo data: (TOPX: AZ ; AZD6302) 0 Plasma concentration (ug/ml) () Rat IV 3.0 mg/kg () Rat PO 0.0 mg/kg PBPK simulation IV (3.0 mg/kg) Vss = 2.6 L/kg; CLh = 8.5 ml/min/kg; KP6 = 5.0 PBPK simulation PO (0.0 mg/kg) Vss = 2.6 L/kg; CLh = 8.5 ml/min/kg; KP6 = 5.0 Drug Information Association 7 Rat IV data intact and BDC Plasma con ncentration (ug/ml) PBPK with EHC: Rat Vdss input into AR adjusted PBPK model simulations compared against rat in vivo data: (TOPX: AZ ; AZD6302) ous Blood Ven Lung Fat Slow Rapid Arterial Blo ood 0.00 Gut () Rat IV 3.0 mg/kg () Rat PO 0.0 mg/kg PBPK model simulation IV Cp PBPK model simulation PO Cp PBPK model simulation IV Cp,u PBPK model simulation PO Cp,u Rat IV 3 mg/kg, BDC 00 0 model simulations compared against rat in vivo data: (TOPX: AZ ; AZD6302) PBPK then simulated without EHC and compared to BDC rats IV Liver Oral dose EHC simulated by return to depot compartment Plasma concentration n (ug/ml) Venous Blood Lung Fat Slow Rapid Arterial Blood () Rat IV 3.0 mg/kg () Rat PO 0.0 mg/kg PBPK model simulation IV Cp PBPK model simulation PO Cp PBPK model simulation IV Cp,u PBPK model simulation PO Cp,u Rat IV 3 mg/kg, BDC Gut Liver Drug Information Association IV 8 9

10 0/25/20 Validation by scaling to other species Plasma concentration (u ug/ml) PBPK model simulations compared against dog in vivo data: (TOPX: AZ ; AZD6302) More conventional modelling in dog 5 % eliminated in bile in the model! Continuous process assumed () Dog IV 3.0 mg/kg () Dog PO 5.0 mg/kg PBPK model simulation IV Cp PBPK model simulation PO Cp PBPK model simulation IV Cp,u PBPK model simulation PO Cp,u Plasma concentration (ug/ml) PBPK model simulations compared against mouse in vivo data: (TOPX: AZ ; AZD6302) () Mouse IV 3.0 mg/kg () Mouse PO 0.0 mg/kg PBPK model simulation IV Cp PBPK model simulation PO Cp PBPK model simulation IV Cp PBPK model simulation PO Cp,u More conventional modelling in mouse (from dog) 5 % eliminated in bile in the model! Continuous process assumed Drug Information Association 9 Conclusions Tissue composition models that rely on physchem information offer opportunity for prediction early in a discovery program Pre-clinical in vivo PK is valuable: confirming mechanistic models Ensuring in vivo understanding of a compound Understanding model assumptions and in vivo behaviour of compound Improve success of prediction Identify uncertainties Drug Information Association