Anne-Gaelle Dosne(1), Elodie Valade(1), Kim Stuyckens(1), Peter De Porre(1), Loretta Sullivan-Chang(3), Dominique Swerts(1), Daniele Ouellet(3), Lilian Li(3), Ruben Faelens(4), Quentin Leirens(4), Juan Jose Perez Ruixo(1)
(1)Janssen Research & Development, Beerse, Belgium (2)Janssen Research & Development, Raritan, NJ, USA (3)Janssen Research & Development, Spring House, PA, USA (4)SGS Exprimo NV, Mechelen, Belgium
Objectives:
To guide a phosphate-based dosing algorithm for erdafitinib, an orally administered panfibroblast growth factor (FGFR) tyrosine kinase inhibitor, using PKPD modeling and simulation.
Methods:
An integrated PKPD model linking the time course of plasma concentrations of erdafitinib to serum phosphate levels, an on-target PD marker of FGFR engagement, was built using data collected in 373 subjects enrolled in six Phase 1 and Phase 2 studies. Subjects included were healthy volunteers and advanced cancer subjects, including subjects with metastatic or unresectable urothelial cancers with FGFR genomic alterations. Erdafitinib was orally administered either as a solution, capsules, or tablets (reference formulation). Doses investigated were single and continuous once daily (QD) doses from 0.5 to 12 mg or intermittent (7 days on/7 days off) dosing regimens of 10 and 12 mg QD. In the patient studies, dose modifications and dose interruptions were performed according to study guidelines based on phosphate levels and other toxicity biomarkers. Due to high protein binding (>99.5%), both free and total concentrations were measured. Parameter estimation was performed with NONMEM (7.3.0), dose optimisation and evaluation of the dosing algorithm rules was assessed though PKPD model based-simulation with Simulo (6.4.1). Diagnostics were obtained via R (3.2.3).
Results:
A semi-mechanistic PK model, including the interaction with alpha-1-acid glycoprotein (AGP), to which erdafitinib is avidly bound, was able to capture total and free erdafitinib plasma concentration profiles. Erdafitinib pharmacokinetics were described with a three-compartment linear disposition model with a first-order absorption process. The unbound fraction was related to the amount of AGP in a hyperbolic fashion, determined by the dissociation constant, Kd. Both total and free concentrations were modelled simultaneously but the model was expressed in terms of total concentrations. Predefined PK parameters were made dependent of the unbound fraction according to distribution hypotheses specific to the compound, i.e. that only free drug was able to distribute and that nonlinear binding to AGP only occured in the central compartment. Free concentrations were linked to total concentrations using the free fraction. Differences in AGP binding were able to account for PK differences between healthy volunteers and patients. The time course of phosphate was linearly related to erdafitinib unbound plasma concentration at biophase and enabled an accurate description of phosphate levels at the different dosing regimens. Extensive model-based simulations were performed and indicated that a starting dose of 8 mg QD using a continuous regimen, with serum phosphate measurement on Day 14 to guide potential individualized up (9 mg) or down (6 mg) titration, will maximize the proportion of patients achieving the target phosphate level (>5.5 mg/dL) while avoiding dose interruptions due to hyperphosphatemia.
Conclusions:
The PKPD modelling and simulation approach developed was used to guide selection of the optimal erdafitinib dosing regimen. Bladder cancer patients enrolled in the ongoing Phase 2 and Phase 3 studies are currently receiving the optimized daily dosing regimen, with individualization of each patient’s dose based on their measured phosphate levels.
Reference: PAGE 27 (2018) Abstr 8584 [www.page-meeting.org/?abstract=8584]
Poster: Drug/Disease Modelling - Oncology