Population Pharmacokinetic Model for BI 1015550 in Healthy Volunteers and Patients with Idiopathic Pulmonary Fibrosis
Nieves Velez de Mendizabal (1), Sonoko Kawakatsu (1), Yichao Yu (2), Donald Zoz (2), Julia Korell (2)
(1) Metrum Research Group, USA; (2) Boehringer Ingelheim Pharmaceuticals, Inc., USA
Introduction:
BI 1015550 is a preferential inhibitor of the phosphodiesterase-4B (PDE4B) with broad anti-inflammatory and anti-fibrotic activities. It is under development for the treatment of idiopathic pulmonary fibrosis (IPF), a fibrotic lung disease characterized by worsening dyspnea and progressive loss of lung function [1]. The natural history of IPF is variable and unpredictable [2], and the disease is ultimately fatal, with a median survival time of 2 to 3 years following diagnosis [1, 3]. Nintedanib (Ofev®) and pirfenidone (Esbriet®) are the only drugs approved for the treatment of IPF. However, despite the availability of these drugs, the medical need remains high in this devastating disease. Based on its mode of action, BI 1015550 is hypothesized to have complementary activity to current therapies in IPF, as well as improved tolerability compared to other marketed pan-PDE4 inhibitors due to its lower affinity for the PDE4D subtype.
The objectives of this analysis were to characterize the pharmacokinetics (PK) of BI 1015550 in healthy volunteers and patients with IPF, to determine the effects of demographic and clinical factors on the PK of BI 1015550, and to investigate whether a drug-drug-interaction (DDI) between BI 1015550 and nintedanib or pirfenidone as background medication could be identified.
Methods:
Data from six Phase I studies in healthy volunteers, one Phase 1c study in patients with IPF and a Phase 2 study also in patients with IPF were combined.
A Population PK analysis was conducted via nonlinear mixed effects modeling using NONMEM, Version 7.4. The Stochastic Approximation Expectation Maximization estimation method followed by the Importance Sampling algorithm were employed during the estimation process. Key and final models were run using the full Bayesian estimation method. Assessment of model adequacy and decisions about increasing model complexity were driven by the data and guided by goodness-of-fit criteria, including (1) visual inspection of diagnostic plots, (2) plausibility of parameter estimates, (3) precision of parameter estimates, (4) the objective function value, and (5) inspection of visual predictive checks. A full model approach was used to identify covariate effects. The impact of covariates included in the final model on the exposure of BI 1015550 was assessed through simulations.
Results:
The full PK analysis dataset included 244 subjects, contributing a total of 3952 observations.
The final model was a two-compartment model with a consecutive zero to first-order absorption rate model with lag-time and first-order elimination. Interindividual variability (exponential model) was estimated on all PK parameters using two independent full covariance matrices; one for the parameters related to the absorption process and another for the clearance and volume parameters.
Weight was included on all clearance and volume parameters using allometric scaling with fixed exponents, resulting in decreasing BI 1015550 exposure with increasing weight. Fasting status was included as a covariate on absorption-related parameters, with fed administration resulting in slightly higher trough and lower maximum plasma concentrations compared to fasted administration. Japanese ethnicity was identified as a significant covariate on central and peripheral volume of distribution parameters resulting in slightly higher maximum concentrations and total exposure in Japanese subjects than non-Japanese subjects.
Neither co-administration of nintedanib nor of pirfenidone was found to be a significant covariate on any BI 1015550 PK parameters. However, the number of subjects receiving these background treatments in the available dataset was small (N = 22 each) compared to the overall sample size, and slight trends were observed in the diagnostic plots for the Phase II study.
Conclusions:
The model adequately described the PK of BI 1015550 in healthy volunteers and patients with IPF. The overall impact of covariates on BI 1015550 exposure was found to be minimal, therefore dose adjustments in subsequent Phase III studies are not warranted. Although neither nintedanib nor pirfenidone co-administration was included as covariate in the final model, dedicated DDI studies with these approved IPF treatments are required to conclusively determine their DDI potential with BI 1015550.
References:
[1] Raghu G., et al.; Am. J. Respir. Crit. CareMed. 183 (2011):788–824.
[2] Ley, B., et al.; Am. J. Respir. Crit. CareMed. 183 (2011):431–440
[3] Nalysnyk, L., et al.; Eur. Respir. Rev. 21 (2012):355–361.