Physiologically based pharmacokinetic modeling of CYP3A4

induction by rifampicin in human: influence of time between

substrate and inducer administration.

Guillaume Baneyxa, Neil Parrotta, Christophe Meillea, Athanassios Iliadisb and Thierry Lavéa

a F.Hoffmann-La Roche AG, pRED, Pharma Research and Early Development, Non-Clinical

Safety, Basel, Switzerland. b Aix Marseille University, Inserm, CRO2, UMR_S 911, 13385

Marseille, France.

Supplementary materials

S1 – Methodology for model construction

Figure S1 - Methodology used for the development of substrate models. After integration of drug and system parameters into the structural model (1) the intravenous (2) and oral administration with (3) and without (4) grapefruit juice were simulated. The key parameters of each step were adjusted in order to obtain concentration-time profiles and PK parameters compatible with those observed from selected clinical studies.

S2 - Physiological values used in the distribution model for a human of 70 Kg – standard values from GastroPlus

Tissue Blood flow

(L/min)

Volume

(L)

Density

(g/mL)

Vnt Vpht Vwt Fvec Capt

(mg/g)

Lung 6.48 0.914 1.050 0.003 0.008 0.811 0.336 3.91Spleen 0.200 0.151 1.054 0.020 0.0198 0.788 0.207 3.18Liver 1.62 1.44 1.07 0.0348 0.0252 0.751 0.161 4.56Hepatic artery 0.519ACAT gut 0.904 *Adipose 0.627 23.8 0.915 0.79 0.002 0.18 0.135 0.40Muscle 0.673 17.0 1.041 0.0238 0.0072 0.76 0.118 1.53Heart 0.249 0.259 1.03 0.0115 0.0565 0.77 0.32 2.25Brain 0.949 1.41 0.97 0.051 0.0565 0.77 0.162 0.40Kidney 1.10 0.227 1.05 0.0207 0.0162 0.783 0.273 5.03Skin 0.254 1.61 1.184 0.0284 0.0111 0.718 0.382 1.32Testes 0.007 0.026 1 0.0207 0.0162 0.783 0.273 5.03Red marrow 0.382 0.965 1.028 0.074 0.0011 0.439 0.100 0.67Yellow marrow 0.106 2.68 0.8 0.79 0.002 0.18 0.135 0.40Rest of body 0.508 11.0 1 0.0201 0.0198 0.788 0.207 3.18Arterial blood 6.48 1.81Venous blood 6.48 3.62

* More details available in Heikkinen et al., 2012.All the values for blood flow, volume, density, fractional neutral lipid content of wet tissue weight (vnt), fractional polar lipid content of wet tissue weight (vpht), fractional water content of wet tissue weight (vwt), volume fraction of interstitial space (Fvec) and concentration of acidic phospholipids in tissue (Capt) are for a human of 70 Kg.

S3 – Comparison of observed and predicted PK profiles for alfentanil, nifedipine and triazolam

A. After intravenous administration

Figure S3A – Observed and predicted plasma concentration-time profiles of ALF and NIF before and after RIF treatment following an intravenous bolus of 1 mg (A, B) and an infusion of 1.46 mg (C). Symbols are observed concentrations ± SD when administered before (closed circles) or after (open circles) RIF treatment. Lines are predicted concentrations when administered before (dashed) or after RIF treatment using Emax = 9 fold (dotted) or Emax = 14.6 fold (solid black). All simulations used K deg , G = 0.03 h-1 and Kdeg , L = 0.0096 h-1. ALF was administered 13 h (A) and 12 h (B) after the last RIF dose whereas NIF was administered after 24 h (C).

B. After oral administration

Figure S3B – Observed and predicted plasma concentration-time profiles of ALF, NIF and TRZ before and after RIF treatment following oral doses of 4 mg (A, B), 20 mg (C) and 0.5 mg (D). Symbols are observed concentrations ± SD when administered before (closed circles) or after (open circles) RIF treatment. Lines are predicted concentrations when administered before (dashed) or after RIF treatment using Emax = 9 fold (dotted) or Emax = 14.6 fold (solid black). All simulations used Kdeg , G =

0.03 h-1 and Kdeg , L = 0.0096 h-1. ALF was administered 13 h (A) and 15 h (B) after the last RIF dose whereas NIF and TRZ were administered after 24 h (C) and 17 h (D), respectively.

S4 - Assessment of model predictability

A. Analysis of weighted residuals of concentrations

WRESi(d )=

y (d ) (t i , x )− yi(d )

y i(d ) (1)

Here, y ( d) (t i , x ) and y i(d ) are the predicted and the observed concentrations for drug d at time

t i in the plasma, respectively. x represents all parameters involved in the PBPK model. The

individual WRESi( d ) were plotted as a function of time and box plots were constructed to

compare dispersion between the experimental conditions.

Figure S4A1 - Weighted residuals of concentrations (WRESi) function of time before (A, B) and after (C-F) RIF treatment. Symbols are MDZ (circle), TRZ (triangle down), ALF (square) and NIF (triangle up). DDI predictions were done with Emax = 9 fold (C, E) or Emax = 14.6 fold (D, F). Horizontal dotted line indicates a theoretical WRES value of zero. Grey area represents a 2 fold error range in predicted concentrations.

Figure S4A2 – Distribution of weighted residuals of concentration (WRESi) before and after RIF treatment for intravenous (A) and oral (B) administrations with in vitro Emax = 9 fold and in vivo Emax = 14.6 fold. The box includes the median, 1st and 3rd quartiles. The whiskers indicate the 5th and 95th

percentiles and closed circles represent outliers outside this range. Horizontal dotted lines indicate a theoretical median of zero. Grey area represents a 2 fold error range in predicted concentrations.

Median 5-95th percentiles p-valueIntravenousalone -0.07 [-0.5; 0.4] 0.01 (*)Emax = 9 fold 0.12 [-0.3; 1.1] 0.003 (**)

Emax = 14.6 fold -0.06 [-0.4; 0.9] 0.9 (ns)

Oralalone 0.06 [-0.4; 0.6] 0.04 (*)Emax = 9 fold 0.55 [-0.3; 3.6] <0.0001 (***)

Emax = 14.6 fold 0.02 [-0.6; 2.1] 0.4 (ns)

Table S4A2 – Median and 5-95th percentiles of weighted residuals of concentration (WRESi) before and after RIF treatment for intravenous and oral administrations with in vitro Emax = 9 fold and in vivo Emax = 14.6 fold. A statistical analysis (Wilcoxon signed rank test) was performed to test if median values were significantly different to zero. Results are shown with several statistical significance levels: no significant for p-value ≥ 0.05 (ns) or significant with p-value 0.01-0.05 (*), 0.001-0.01 (**) and < 0.001 (***).

Alone in vitro E max in vivo E max-1

0

1

2

3

4

5

Wei

ghte

d re

sidu

als

of c

once

ntra

tion

Alone in vitro E max in vivo E max-1

0

1

2

3

4

5W

eigh

ted

resi

dual

s of

con

cent

ratio

n

A B

B. Impact of hepatic CYP3A4 turnover rate on the recovery of midazolam kinetics

Figure S4B - Simulation of MDZ pharmacokinetic recovery after RIF treatment discontinuation. MDZ 2 mg was orally administered on day 21, 28 and 42 corresponding to 1, 2 and 4 weeks after the last RIF dose. The plasma concentration-time profiles have been predicted using Kdeg , L = 0.0096 h-1 (solid black curve) or 0.03 h-1 (dashed curve). Observed concentrations ± SD (open circle) are from Reitman et al., 2011.

AUC∞ (ng.h/mL) Cmax (ng/mL) T 1/2 (h)

observed predicted observed predicted observed predicted

K deg , L0.03 h-1

Kdeg , L0.0096 h-1

Kdeg , L0.03 h-1

Kdeg , L0.0096 h-1

K deg , L0.03 h-1

Kdeg , L0.0096 h-1

after 1 week 8.2 ± 2.5 19.9 11.9 3.3 ± 1.5 7.1 4.8 1.8 ± 0.7 3.3 2.8

after 2 weeks

17.4 ± 6.5 21.6 18.9 6.0 ± 1.9 7.4 6.5 2.7 ± 1.5 3.3 3.2

after 4 weeks

21.4 ± 7.2 21.7 21.5 8.2 ± 2.3 7.4 7.4 2.5 ± 1.0 3.3 3.3

Table S4B - Observed (mean ± SD) and predicted MDZ AUC∞, Cmax and T 1/ 2 on 1, 2 and 4 weeks after the last RIF dose obtained with K deg , L of 0.03 and 0.0096 h-1. Values of predicted parameters are calculated by non-compartmental analysis.