2018 - Montreux - Switzerland

PAGE 2018: Drug/Disease modelling
Joćo Abrantes

Integrated modelling of factor VIII activity kinetics, occurrence of bleeds and individual characteristics in haemophilia A patients using a full random effects modelling (FREM) approach

Joćo A. Abrantes (1), Alexander Solms (2), Dirk Garmann (3), Elisabet I. Nielsen (1), Siv Jönsson (1), Mats O. Karlsson (1)

(1) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden, (2) Bayer, Berlin, Germany, (3) Bayer, Wuppertal, Germany

Objectives: Haemophilia A (HA) is a bleeding disorder caused by a deficiency of coagulation factor VIII (FVIII). Although model-based TDM of FVIII products has been encouraged [1], there is still a lack of knowledge on the exposure-response relationship, and therefore on the individual FVIII activity level to target.

In this study, we aimed to develop an integrated pharmacometric model to characterize the relationship between FVIII activity and occurrence of bleeding episodes in HA patients receiving prophylactic treatment, accounting for all available patient and study specific characteristics.

Methods: Pooled pharmacokinetic (PK) and bleeding data during prophylactic treatment with BAY 81-8973 (octocog alfa, Kovaltry®) were obtained from the three LEOPOLD trials [2-4]. The studies had a duration of 6-12 months and included previously treated patients aged 1-65 years. Available patient characteristics were age, weight, body mass index, lean body weight, race, von Willebrand factor levels, number of bleeds in 12 months pre-study (NBL), previous therapy history (on-demand/prophylaxis) and number of target joints at study start.

Initially we evaluated previously developed popPK [5] and parametric repeated-to-event (RTTE) models [6]. The RTTE model was re-estimated including bleeding data from the LEOPOLD kids trial (age ≤12 years), and alternative baseline hazard [h0(t)] parameterizations and different inter-individual variability (IIV) model structures were explored. In addition, the inclusion of a time-dependency between consecutive bleeds was tested with a Markov hazard rate accounting for the time since the last event (TSE), implemented as an exponential term, λmarkov · e-γ markov·TSE. At study start, 1/NBL was used as an estimate of TSE. The PK model was qualified through GOF plots and stratified pcVPCs, and the RTTE model with stratified VPCs of the Kaplan-Meier (KM) curves and KM mean covariate plots. Parameter uncertainty was estimated with SIR.

The updated models were converted to a FREM model [7,8], with all patient characteristics available treated as observations, and all parameter-covariate relationships estimated simultaneously using exponential and power relations. Modelling was performed in NONMEM 7.3 employing the IMPMAP method, assisted by PsN and graphical and statistical analyses by R.

Results: The final FREM model included 1535 FVIII activity observations (N=183 patients; more details on PK sampling design [5]), 663 bleeds (N=172), and 11 individual characteristics.

The previous popPK model was appropriate, however IIV on the residual error further improved the fit (p<0.001) and increased parameter precision, and was therefore included. The drug effect was included on the hazard (p<0.001) described by h(t) = λ·eγ·(t-1)·(1–FVIII/(FVIII+IF50)), with log-normal IIV on the scale parameter (λ). Both λ and IF50 (FVIII activity resulting in half-maximum inhibition) were parametrized in terms of λ0.5 and λ20 at 1 year after study start, representing the hazard when plasma FVIII activity was 0.5 IU/dL (severe HA) and 20 IU/dL (mild HA), respectively. There was no evidence that a bleeding episode transiently changed the hazard of a new bleed (p>0.05 for markov component). The final parameter estimates were λ0.5=2.9 year-1 [95%CI 1.9,3.9], λ20=1.1 year-1 [0.72,1.4], γ=-0.56 year-1 [-0.85,-0.28], and IIV on λ 167%CV [110,225], and the derived IF50 was 11 IU/dL.

The full covariance matrix included the interaction between CL, V, hazard (λ), residual error magnitude and 14 covariates. The parameter-covariate relationship showing the largest effect size was NBL on hazard of bleeding; a patient with 1 bleed one year pre-study had a 24% [-35,-12] lower hazard compared to a patient who had 21 bleeds (mean), and a patient with 84 bleeds had a 144% [51,300] higher hazard. The inclusion of all covariates in the model resulted in a maximum IIV (CV%) drop of 5.7 CL, 1.5 V and 11 λ.

Conclusions: We developed a model describing FVIII activity and occurrence of bleeds over time in adult and paediatric HA patients during prophylaxis, accounting for all available individual characteristics. Plasma FVIII activity and the number of previous bleeds were found to be the main factors predicting the bleeding risk. The developed model may lead to a more effective and cost-efficient dosing in haemophilia A.

[1] McEneny-King A, Iorio A, Foster G, et al. The use of pharmacokinetics in dose individualization of factor VIII in the treatment of hemophilia A. Expert Opin Drug Metab Toxicol. 2016;19:1-9.
[2] Saxena K, Lalezari S, Oldenburg J, et al. Efficacy and safety of BAY 81-8973, a full-length recombinant factor VIII: results from the LEOPOLD I trial. Haemophilia. 2016;22(5):706-12.
[3] Kavakli K, Yang R, Rusen L, et al. Prophylaxis vs. on-demand treatment with BAY 81-8973, a full-length plasma protein-free recombinant factor VIII product: results from a randomized trial (LEOPOLD II). J Thromb Haemost. 2015;13(3):360-9.
[4] Ljung R, Kenet G, Mancuso ME, et al. BAY 81-8973 safety and efficacy for prophylaxis and treatment of bleeds in previously treated children with severe haemophilia A: results of the LEOPOLD Kids Trial. Haemophilia. 2016;22(3):354-60.
[5] Garmann D, McLeay S, Shah A, et al. Population pharmacokinetic characterization of BAY 81-8973, a full-length recombinant factor VIII: lessons learned - importance of including samples with factor VIII levels below the quantitation limit. Haemophilia. 2017;23(4):528-537.
[6] Garmann D, Frede M, Shah A, et al. PAGE 24 (2015) Abstr 3683 [www.page-meeting.org/?abstract=3683].
[7] Yngman G, Nyberg J, Jonsson EN, et al. PAGE 26 (2017) Abstr 7365 [www.page-meeting.org/?abstract=7365].
[8] Karlsson MO. PAGE 21 (2012) Abstr 2455 [www.page-meeting.org/?abstract=2455].

Reference: PAGE 27 (2018) Abstr 8646 [www.page-meeting.org/?abstract=8646]
Oral: Drug/Disease modelling
Click to open PDF poster/presentation (click to open)