Modelling of PK/Efficacy/Toxicity in rats to help design a FHD for a Cell Cycle Inhibitor X Celine Pitou, PharmD1; Sophie Glatt, PhD3;Inaki Troconiz, PhD2 1, PK/PD/TS dpt, Eli Lilly and Co, UK; 2, School of Pharmacy, University of Navarra, Spain; 3, former Lilly employee, currently at UCB Celltech Ltd, UK OBJECTIVES The primary objective of PK/PD modelling in preparation of the FHD study was to estimate a pharmacologically effective and safe dose range in humans, based on preclinical data. In addition, doses and administration schedules for this compound were optimized in order to minimise the duration of neutropenia. METHODS The PK model was built using both oral and IV rat data from 12 studies over a total dose range of 0.5 to 60mpk dosed QD. Drug The PK/efficacy model was developed using rat HCT116 xenograft data by means of a modified Gompertz model (1) where compound LY is assumed to act as a cytotoxic agent by increasing the rate of cell death. This model was built using IV rat data over a total dose range of 10 to 30mpk dosed weekly. This model is described by several parameters: TS0, tumour size at the start of the experiment; Kgw1, rate constant controlling the cell proliferation with an exponential rate; Kgw2, rate constant controlling the cell proliferation with a linear rate; Kdth, rate constant controlling the cell death; Ktr, first order rate of transition. Kprol (=kgrw1, kgrw2) + kdth TS feedback=(Circ0/circ) Kprol=(ktr) The PK/Tox model was developed using neutrophil count data obtained in nude rats from 5 pharmacology studies over a total dose range of 9 to 60 dosed QD, Q2D, Q3D or Q7D. The model (2) describes the time course of plasma neutrophil count is assumed to reduce the proliferation rate or induce cell loss, which was modelled as a linear function. A time effect was also added to account for the increase in baseline neutrophil count over time. ktr Prol kdth Damage Cells 1 Trst 1 Trst 2 ktr ktr ktr Trst 3 Neu kCirc(=ktr) Drug Effect MTT Allometric Scaling approach was performed to predict human PK parameters using mouse, rat, dog and monkey IV PK data (2 cpt model was assumed for all species). MODELLING RESULTS Allometric Scaling Results: The estimated median and 90% CI for human Cl, V1 and V2 were 8.50L/h (4.61-15.42), 121.55L (35.86-391.76) and 36.61L (19.4868.41), respectively. Inter-individual variability in humans was assumed to be 30% for CL, V1 and V2. Rat PK data are described adequately with a two cpt model with first order 250 200 150 5000 600 400 100 200 50 0 0 0 4 8 12 16 20 0 24 4 8 12 21.7 9.57 5th 95th median actual data Dose of 0.5mpk IV w 0.784 (88.5%) V3 0.0148 (12.2%) 22 140 34.9 additive error 0.44 Dose = 20 mg 100 80 60 Dose = 30 mg 2000 2500 2000 1500 1000 40 0.155 (12.5%) 2.11 (35.3%) Dose = 25 mg 5000 3000 120 2 14 2 0 14 Time (days) 500 20 multiplicative error additive error w2 Kgw1 0 0 4 8 12 16 20 24 0 4 8 12 time (hr) median 5th 95th 16 20 24 Placebo group time (hr) actual data median 5th 95th actual data Rat PK/Tox model describes the data with a reasonable uncertainty. A time effect for the baseline was added in the model due to the experimental data as shown below. This was also described and modelled in literature (3). It’s important to note that the rats were immunosuppressed which explain the low Circ0 value. Dose of 15mpk weekly 6 Parameters Circ0 (106/mL) MTT (h) GAM slope (mL/ng) time effect Estimates 1.2 49.7 0.469 0.00182 0.00112 %SEE 7.98 2.56 6.12 11.5 27.7 9 8 5 7 neutrophil counts (109/L) 27 actual data 2000 4 3 2 6 5 4 3 2 1 1 0 0 0 168 336 504 672 0 168 5th 95th 0.106 (32.6%) 0.889 (94.3%) additive error multiplicative error 0.46 61.40% 28 60 504 672 504 672 time (hr) actual data median Dose 30mpk weekly w2 Circ0 w2 slope 336 time (hr) median 5th 95th actual data Dose of 45mpk weekly 10 12 9 10 8 7 Neutrophil counts (109/L) 28.5 95th neutrophil counts (109/L) 433 0.109 (33%) 0.599 (77.4%) 2 5th %SEE 5.46 12.4 7.03 11 3500 plasma concentration (ng/mL) 0.0166 (12.9%) w2V2 plasma concentration (ng/mL) w2Ka 24 Estimates 0.196 220 60.9 0.00503 Dose of 10mpk IV 160 w2CL 20 Parameters Kgw1, zero order (/day) Kgw2, first order (mg/day) TS0 (mg) Kdth (mL/h/ng) neutrophil counts (109/L) 0.378 0.465 16 time (hr) time (hr) median Ka (h-1) F Dose = 15 mg 800 Tumor Size (mg) plasma concentration (ng/mL) plasma concentration (ng/mL) % SEE 6.22 18.9 18.5 16.6 14 Dose = 10 mg 1000 300 Estimates 0.613 2.7 2.21 2.19 2 Control 1200 350 Parameters CL (L/h) V2(L) Q (L/h) V3 (L) Rat PK/efficacy model was better define with a modified Gompertz model Dose of 30mpk PO Dose of 10mpk PO absorption 6 5 4 8 6 4 3 2 2 1 0 0 0 168 336 504 672 0 168 336 time (hr) median 5th 95th time (hr) actual data median 5th 95th actual data RAT SIMULATION RESULTS Rat Simulation: exploratory relationship weekly dosing Q7d*3 at 15, 20, 30, 40mpk IV 1hr infusion 1.45 15 1.40 fold decrease from baseline (neutrophils account) 1.35 11 1.30 1.25 9 1.20 7 1.15 5 1.10 fold decrease from placebo on D28 (tumor size) 13 In order to correlate the efficacy and the myelosuppression rat models, some simulations were performed at several doses. The figure below represents in function of AUC(0-168h), the fold decrease of tumour volume and the fold decrease of neutrophil count. In this figure, the curve of the toxicity is following the curve of the tumour growth decrease. Based on this, the starting targeted efficacious exposure in rats is 5700ng*h/mL. 3 1.05 1.00 0 2000 4000 6000 8000 10000 12000 14000 1 16000 predicted AUCweekly (ng*h/mL) PD neutrophils HUMAN SIMULATION RESULTS Efficacy Human PK and PD simulations in terms of neutrophil count were generated after adaptation of the rat PD parameters from literature data. It was performed following one or two injections per 21-day cycle for the same total dose per cycle. Human median Neutrophil account simulations at several doses twice per cycle D1 and D8 over 2 cycles (21days) 5.5 Human median Neutrophil account simulations at several doses once per cycle over 2 cycles (21days) 5.0 6.0 4.5 5.5 Neutrophil count (109/L) 4.5 4.0 3.5 3.0 2.5 2.0 1.5 grade 2 1.0 grade 3 0.5 Compound X given twice per cycle gave a more prolonged nadir compare to a single administration over the cycle. This has been noticed for 5-FU in literature (4). These two sets of simulations suggested that, upon one dose every 21 days, a dose range of A to E should be required to reach efficacy and to have a manageable neutropenia. grade 4 Neutrophil Count (109/L) 4.0 5.0 3.5 3.0 2.5 2.0 1.5 grade 2 1.0 grade 3 0.5 grade 4 0.0 0 0.0 0 168 Amg median 336 Bmg median 504 time (hr) Cmg median Dmg median 672 840 Emg median 168 336 1008 504 672 840 1008 time (hr) Amg median Time of the Dose Bmg median Cmg median Dmg median Emg median CONCLUSIONS This modelling study combined both efficacy and toxicity rat data within one fully integrated mechanistic model. This integrated model described the rat data adequately and helped connect efficacy and toxicity outcomes. The model was translated to human to derive an efficacious and safe dose range for the FHD study. References: 1. Simeoni et al, Cancer Res 2004;64:1094-1101, 2. Friberg LE et al, J Clin Oncol 2002;20: 4713-4721. 3. Segura et al, Pharm Res 2004:21:567-572, 4. Friberg LE et al, J Pharmacol Exp Ther 2000;295:734-740 Time of Dose