A model for evaluation of novel on-demand treatment of bleeding events in hemophilia subjects
Alan Faraj (1), Joakim Nyberg (2), Grant E. Blouse (3), Tom Knudsen (3), Ulrika S.H. Simonsson (1)
(1) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden, (2) Department of Pharmacy, Uppsala University, Uppsala, Sweden, (3) Catalyst Biosciences, South San Francisco, CA, USA
Objectives: Marzeptacog alfa (MarzAA) given subcutaneously (SC) was under development for on-demand treatment of episodic bleeds. The current standard-of-care (SoC) is administered intravenously (IV) and clears rapidly after injection. Administered prophylactically, MarzAA significantly lowered the annual bleeding rate by 90% in a phase-2 trial[1]. Across all trials, MarzAA has demonstrated excellent safety[1-3]. Supported by a potency bridging strategy using population pharmacokinetic modeling[4] a phase-3 cross-over trial was launched to evaluate MarzAAs potential for on-demand treatment[2] in comparison to SoC. The data, indicated that MarzAA was both efficacious and safe in subjects with hemophilia A with inhibitors. Rare diseases suffer from small sample sizes which may negatively impact power to detect drug effects. Model-based analysis may improve power to detect drug effects in these settings[5-7]. This work aimed to use a model-based analysis to detect drug effects during novel on-demand treatments of bleeding events in hemophilia A or B, by analysing repeated categorical outcome data. Using the model, outcome scores were compared between SC MarzAA and IV SoC.
Methods: Efficacy data from a randomized multi-center global cross-over phase-3 trial (NCT04489537) were used[2]. Subjects were randomized to receive either SC MarzAA (60 µg/kg) or IV SoC for 5 consecutive bleeds or vice versa before crossing over. SC MarzAA or IV SoC was administered on-demand one, two, or three times (or as indicated by its label) in three-hourly intervals following a bleeding event. Treatment responses were measured using a 4-point clinical scale (poor, fair, good or excellent control) at different time-points after the first dose. A continuous-time Markov model[8] was developed in which the probability of the score was dependent on the previous score and the time since last score[8-9]. Due to a small sample size, outcome data were binarized into treatment failure (poor/fair hemostatic control [TF]) or treatment success (good/excellent hemostatic control [TS]). Probability of each score was modeled using differential equations describing transition between the two scores. For each bleeding event, the system was reset and the patient was initialized with TF with the possibility to stay in the TF state or enter TS and with the possibility of staying or moving back to a TF state at each timepoint. Inter-individual variability (IIV) and inter-bleeding-variability (IBV) was tested on all transfer rates. Age, bodyweight, height, race, diastolic/systolic blood pressure, heart rate, location and severity of bleed were tested as covariates on the transfer rates. Any difference in MarzAA and SoC transition rates was tested to and from both states. Modeling was performed in NONMEM 7.5[10]. Model diagnostics were done in R using xpose4[11] and was guided by objective function value (OFV), parameter uncertainty and simulation properties.
Results: A total of 15 subjects were included in the dataset. No serious adverse events related to the treatments were reported. The number of evaluable bleeds were 71. Each subject experienced 5 bleeds on average (90% inter-percentile range; 1-8 bleeds). The total number of efficacy evaluations were 222 and 265 for SC MarzAA and IV SoC, respectively. A first-order continuous-time Markov model was successfully developed describing scores after both treatments and representing the states of TF or TS (i.e. effective hemostasis). Transfer rates to TS given TF and to TF given TS was estimated to 5.1 and 0.08 day-1, respectively. A high IIV was supported for the transition rate from TF to TS (CV>100%) but not vice versa. IBV was not supported (ΔOFV = -3.6). None of the tested covariates were found to be statistically significant. No statistically significant difference in transition from TF to TS was identified between SC MarzAA and IV SoC, strongly indicating comparable efficacy between the treatments. The model indicated good data fits and parameter values were estimated with acceptable uncertainty (RSE<50%) given the sample size.
Conclusions: The developed model described the clinical outcome data well. No statistically significant difference was found between SC MarzAA and IV SoC in the rate of achieving effective hemostasis. The model can be used for characterization of exposure-response relationships of SC MarzAA and other novel on-demand treatments aiding future clinical trial designs and in the registrational process for treatments for hemophilia.
References:
[1] Mahlangu, J. et al. Subcutaneous engineered factor VIIa marzeptacog alfa (activated) in hemophilia with inhibitors: Phase 2 trial of pharmacokinetics, pharmacodynamics, efficacy, and safety. Res. Pract. Thromb. Haemost. 5, e12576 (2021)
[2] Crimson 1: A Phase 3 study to evaluate the efficacy and safety of subcutaneous marzeptocog alfa (activated) for on-demand treatment of bleed events in subjects with hemophilia A or B, with inhibitors. ISTH Congr. Abstr.
[3] Gruppo, R. A. et al. Phase 1, single-dose escalating study of marzeptacog alfa (activated), a recombinant factor VIIa variant, in patients with severe hemophilia. J. Thromb. Haemost. 16, 1984–1993 (2018)
[4] Faraj, A. et al. Phase-3 Dose Selection of Marzeptacog Alfa (Activated) Informed by Population Pharmacokinetic Modeling: A Novel Hemostatic Drug. CPT Pharmacomet. Syst. Pharmacol. (2022)
[5] Svensson, R. J., Gillespie, S. H. & Simonsson, U. S. H. Improved power for TB Phase IIa trials using a model-based pharmacokinetic-pharmacodynamic approach compared with commonly used analysis methods. J. Antimicrob. Chemother. 72, 2311–2319 (2017)
[6] Rekic, D., Röshammar, D. & Simonsson, U. S. H. Model based design and analysis of phase II HIV-1 trials. J. Pharmacokinet. Pharmacodyn. 40, 487–496 (2013)
[7] Karlsson, K. E., Vong, C., Bergstrand, M., Jonsson, E. N. & Karlsson, M. O. Comparisons of Analysis Methods for Proof-of-Concept Trials. CPT Pharmacomet. Syst. Pharmacol. 2 (2013)
[8] Markov, A. A. An Example of Statistical Investigation of the Text Eugene Onegin Concerning the Connection of Samples in Chains. Sci. Context 19, 591–600 (2006)
[9] Bergstrand, M., Söderlind, E., Weitschies, W. & Karlsson, M. O. Mechanistic modeling of a magnetic marker monitoring study linking gastrointestinal tablet transit, in vivo drug release, and pharmacokinetics. Clin. Pharmacol. Ther. 86, 77–83 (2009)
[10] Beal SL, Sheiner LB, Boeckmann AJ, Bauer RJ 1989-2011. NONMEM Users Guides. Icon Development Solutions, Ellicott City, Maryland, USA
[11] Keizer, R. J., Karlsson, M. O. & Hooker, A. Modeling and Simulation Workbench for NONMEM: Tutorial on Pirana, PsN, and Xpose. CPT Pharmacomet. Syst. Pharmacol. 2 (2013)
