Christian Laveille (1), Kathy de Graaf (2), Maria Ballabio (2), Cristina de Min (2), Philippe Jacqmin (3)
(1) Calvagone Sarl, Liergues, France, (2) Swedish Orphan Biovitrum AG (Sobi), Basel, Switzerland, (3) MnS Modelling and Simulation, Dinant, Belgium
Introduction: Primary hemophagocytic lymphohistiocytosis (pHLH) is a rare genetic disease. The known causative mutations lead to continuous overactivation of CD8+ T cells and macrophages and to cytokine storms damaging the body’s own organs, including the liver, brain, and bone marrow. pHLH typically presents during infancy and is invariably fatal if untreated. Data from animal models of pHLH and clinical data [1] have indicated that IFNγ plays an essential role in developing and maintaining immune cell overactivation in HLH. Neutralization of IFNγ could help control the disease until a donor is identified for hematopoietic stem cell transplantation, the only curative treatment.
Objectives: To streamline the development of emapalumab, a first-in-class anti-interferon (IFN)-γ monoclonal antibody, in the treatment of pHLH using population (POP) PK/PD modelling and simulation.
Methods: Emapalumab was approved by the FDA [2] based on data from 27 pediatric patients in a single clinical study. This was partly possible due to continuous support of PK/PD modelling. During the initial phase of the study, intensive real-time PK monitoring with model-informed dose adaptation was implemented. Interim population modelling of the data allowed an early understanding of the PK and PD of emapalumab and its interaction with the disease, which facilitated transition to the pivotal phase of the study.
For PK, an adapted target-mediated drug disposition model was used.
For PD, the evolution of CXCL9, a chemokine specifically induced by IFNγ, was modelled using an indirect response model (IRM) with inhibition by emapalumab of the stimulation by IFNγ of the production of CXCL9.
Exposure-effect relationship between the PD biomarker CXCL9 and overall clinical response (OCR) was evaluated by logistic regression and receiver operating characteristic (ROC) analyses.
Results: The results indicated a high inter-subject and unpredictable high intra-subject variability in IFNγ production leading to target-mediated drug disposition impacting the PK and PD of emapalumab.
The PK of emapalumab was best described by a two-compartment model. In pHLH patients, central and peripheral volumes of distribution were 0.059 and 0.079 L/kg, respectively. Clearance (CL) had 2 components: a slow linear CL of 0.00112 L/h for a body weight (BW) of 5 kg, and a target-mediated CL dependent on IFNγ production. For the target-mediated CL, total IFNγ concentration (free + bound), an indicator of IFNγ production, was used as a time-varying covariate. The higher the IFNγ production, the higher the target-mediated CL. At values of total IFNγ from 103 to 106 pg/mL, the emapalumab total CL (linear + target-mediated) ranged from 0.00119 to 0.0140 L/h for a BW of 5 kg, with corresponding terminal half-lives from 17.5 to 2.29 days. This wide range of CLs and half-lives partly explained the dose adaptations that were required in pHLH patients.
Emapalumab significantly decreased the concentration of CXCL9. In the IRM, neutralization by emapalumab of time-varying IFNγ-dependent production of CXCL9 was described using an inhibitory EMAX model. IC50 of emapalumab (~25 ng/mL) was markedly lower than the circulating concentrations of emapalumab after multiple administrations (e.g., median trough of an effective dose of 1 mg/kg every 3 days was >10,000 ng/mL), indicating that neutralization of IFNγ of at least 99% was required in HLH patients. The apparent half-life of CXCL9 was ~1.2 days and determined the delay in the measured response of CXCL9 to emapalumab.
For exposure-response, logistic regression analyses indicated a statistically significant association between CXCL9 concentrations and OCR. The ROC analysis determined the threshold concentration of CXCL9 below which OCR has a good probability to be observed.
The PK, PK/PD, logistic and ROC models have been jointly used in simulations, which included the variability in IFNγ production, to support the proposed dosing scheme of emapalumab in pHLH patients.
Conclusions: In a rare pediatric disease, the development of a first-in-class compound has been greatly facilitated using continuous model-informed PK monitoring. Interim POP PK and PK/PD analyses of data have enabled a rational transition to the pivotal phase. A dosing strategy based on clinical and laboratory disease parameters integrating the impact of target variability on the PK and PD of the compound has been proposed to optimize overall clinical response in the treated population.
References:
[1] Locatelli F. et al. Safety and Efficacy of Emapalumab in Pediatric Patients with Primary Hemophagocytic Lymphohistiocytosis. Blood 2018;132(Suppl 1):LBA-6.
[2] Gamifant US PI (2018)
https://www.accessdata.fda.gov/drugsatfda_docs/label/2018/761107lbl.pdf
Reference: PAGE () Abstr 9323 [www.page-meeting.org/?abstract=9323]
Poster: Oral: Drug/Disease Modelling