A Population PKPD Model Assessing the Pharmacodynamics of

A Population PK/PD Model Assessing the Pharmacodynamics of a Rapid-acting Recombinant FVIIa Analogue, NN 1731, in Healthy Male Subjects. Andreas Groth 1, Judi Møss 2, Tine Møller 3, Steen H. Ingwersen 1 1 Biomodelling, 2 Medical and Science, Novo. Seven Key Projects, 3 Biostatistics, Novo Nordisk A/S, Copenhagen, Denmark *avg@novonordisk. com Conclusions Aims To explore the potential of prothrombin fragments 1&2 (F 1+2) as a biomarker for haemostatic agents with a model of the effects of the FVIIa analogue NN 1731. Background Haemophilic patients suffer a defect in blood coagulation due to lack of either coagulation factor VIII or IX. In some patients, replacement therapy with the lacking coagulation factor eventually results in the formation of antibodies (inhibitors). Inhibitor patients may be treated with by-passing agents such as activated human coagulation factor FVII (FVIIa, Novo. Seven ). NN 1731 is a FVIIa analogue that in vitro has shown increased activity in stimulating the 1 cleavage of prothrombin to thrombin and F 1+2, a key step in the coagulatory pathway. F 1+2 was measured in the first clinical trial with NN 1731, a dose escalation trial with 4 dosing arms. NN 1731 and F 1+2 plasma concentrations were related to NN 1731 doses to establish a population PK/PD model treating F 1+2 as a PD biomarker. An appreciable dose-concentration-response relationship between NN 1731 and F 1+2 was expressed in a population PK/PD model. Since F 1+2 appearance traces the formation of thrombin, this relationship supports the possibility of using F 1+2 as a biomarker for haemostatic agents. References 1 E. Persson et al, Proc Natl Acad Sci U S A, 96; 13583, 2001 A sketch of the pro-coagulatory actions of thrombin, also known as coagulation factor IIa. Thrombin activates several coagulatory proteins and these actions cascade down eventually leading to the formation of cross-linked fibrin (CLIa) which forms the actual blood clot. Methods • i. ii. • • Strategy: Develop PK-model from NN 1731 plasma concentration data. Develop PD-model from individual post-hoc PK model parameters and F 1+2 plasma concentration data. Data source: (3 pre- and 10 post-dose PK samples + 1 pre- and 5 post-dose PD samples) 6 healthy subjects 4 active (non-zero) dose levels. Dose range 5 µg/kg-30 µg/kg Modelling: PK and PD in man was modelled sequentially using NONMEM V with FOCE. Regarding inter-individual variability (i. i. v. ) on model parameters, log-normal distributions were tested for significance against the hypotheses of zero i. i. v on that parameter (which is why the geometric, rather than the arithmetic, post-hoc estimate means are displayed on fig. 4). Results Figure 1. Structure of PK/PD model Figure 2. Fit of PK/PD model parameters V 2 Model parameter V 1 V 2 CL Q B kout E Unit ml/kg/h p. M 1/h p. M/h/IU/ml) Estimate 59. 6 78. 0 (14%) 120 (14%) 38. 6 132 (29%) 0. 346 0. 164 (46%) S. E. of estimate 2. 2 4. 0 4. 5 2. 0 11 0. 048 0. 042 V 1, V 2: central & peripheral volumes of distribution, CL: clearance, Q: intercompartmental clearance, B: baseline F 1+2 level, kout: rate constant for F 1+2, E: NN 1731 efficacy Values in parenthesis: Coefficients of Variation (CV’s) regarding i. i. v. for each parameter. Q V 1 Cp CL NN 1731 effect The PK/PD model predictions of the F 1+2 time profiles for each trial subject as well as for the typical subject are shown along with the observations and their means for each dose level. F 1+2 kout The resulting PK model was a standard two-compartment model with inter-individual variability (i. i. v. ) on CL and V 2. The PD model was a linear indirect response model with the plasma concentration of NN 1731 affecting the formation of F 1+2 , incorporating i. i. v. on baseline F 1+2 levels (B) and the efficacy parameter (E). The F 1+2 formation rate at the baseline state Cp=0 equals B kout. , , . . Model predictions (individual) Model predictions (typical subject) o, o, . . Observations + Mean of observations at time point Figure 3. Dose-independence check of PK model E p. M/h/(IU/ml) Figure 4. Dose-independence check of PD model V 2 (ml/kg) CL ml/kg//h NN 1731 dose (µg/kg) 30 g/kg 20 g/kg o Individual post-hoc estimates + Geometric mean of individual post-hoc estimates NN 1731 dose (µg/kg) Post-hoc parameter estimates of CL and V 2 were checked for dose independence. Such a dependence appears to be absent for both parameters, indicating that the PK model is valid over the studied dose-range. B (p. M) NN 1731 dose B kout (1+ E Cp) 10 g/kg 5 g/kg NN 1731 dose (µg/kg) o Individual post-hoc estimates + Geometric mean of individual post-hoc estimates NN 1731 dose (µg/kg) The individual post-hoc parameter estimates in the PD model were also checked for dependence on NN 1731 dose. The result is less clear -cut than that of the PK parameters (fig. 3), but since the ranges of values for the lowest and the highest dose are quite similar for the efficacy parameter E, it is concluded that the NN 1731 concentration-PD response relationship is well described.