Bounded integer model-based analysis of psoriasis area and severity index in patients with moderate-to-severe plaque psoriasis receiving BI 730357 treatment
Qing Xi Ooi (1), Anders Kristoffersson (1), Julia Korell (2), Mary Flack (2), Elodie L. Plan (1), and Benjamin Weber (2)
(1) Pharmetheus AB, Uppsala, Sweden, (2) Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, CT
Plaque psoriasis is an autoimmune condition causing itching and stinging plaques on the skin . Psoriasis treatments are commonly evaluated using the psoriasis area and severity index (PASI), assessing the extent of skin area affected and the severity of erythema, thickness, and scaling. The resulting composite PASI score can range from 0 to 72 .
Applying modeling to scales like PASI may be powerful, but, depending on the approach, may lead to unrealistic simulations, if treated as continuous and thus violating the nature of the data, or imprecise estimates, when handled as ordered categorical and involving a high number of parameters. The bounded integer model was introduced to offer a parsimonious alternative respecting the discrete and bounded characteristics of the scores .
The objective of this analysis was to predict the expected PASI outcome in the target patient population for different BI 730357 dosage regimens. This was achieved based on population pharmacokinetic (PK) and bounded integer pharmacokinetic-pharmacodynamic (PKPD) models developed to characterize the relationship between BI 730357 exposure and the PASI response.
The data modeled in the PKPD analysis originated from a phase II study in patients with moderate-to-severe plaque psoriasis, the PK model data was from the Phase II study and 3 phase I studies in healthy subjects.
The PK model development explored a wide range of absorption and disposition models. The PKPD model was based on a bounded integer model implemented with improved numerical stability ; assuming a latent function and a scaling parameter for each subject. Parameter estimation was performed in NONMEM using the FOCEI method for the PK model and Monte Carlo importance sampling for the PKPD model.
The exposure and response in patients with moderate-to-severe plaque psoriasis following administration of BI 730357 200 mg once daily, 400 mg once daily, and 200 mg twice daily under fed conditions were simulated. Simulations of 300 data sets, each containing 500 patients, for each dosage regimen, involved the point estimates of the parameters and individual vectors of covariates sampled by non-parametric bootstrap.
The PK dataset included 7686 plasma samples measured for BI 730357 concentrations, and 1554 PASI observations from 274 18 to 75 years old patients (69.3% men) supported the PKPD model.
The PK was described with two compartments and a first-order elimination. The drug was primarily absorbed through a sequential zero- and first-order process after a brief absorption lag, while a smaller fraction followed a zero-order absorption process after a long lag. A dose-subproportionality was attributed to bioavailability decreasing with increasing dose, in a slower fashion for patients, who generally had a lower bioavailability compared with healthy subjects. Clearance (6.72 L/h) was lower for patients and subjects with higher baseline aspartate aminotransferase and C-reactive protein (CRP). Patients’ clearance (5.19 L/h) was lower also after adjusting for their higher CRP.
The PKPD model presented a placebo response not differentiable from 0, and an IC50 and an Imax that could be quantified. The Imax was largest in magnitude for patients with no prior biologic use, smaller for patients with prior use of non-IL-17 inhibitors, and smallest for patients with prior IL-17 inhibitor use. The model allowed precise parameter estimation, and adequate description of the relative change from baseline PASI for each dose level, the reduction in PASI from baseline greater or equal to 75% (PASI75) at 12 weeks, and the Pearson residuals.
The stochastic simulated percentage of patients attaining PASI75 at week 12 were overlapping between the explored dosing regimens (median 23%-26%). The highest simulated PASI75 response (median 32%) was in patients without prior biologic use receiving 200 mg twice daily. The steady-state average concentration associated with a drug effect of 50% change from baseline PASI at week 12 was about 2000 nmol/L. Higher simulated concentrations were linked to minor, plateauing improvements in PASI response.
Our analyses enabled robust predictions of the impact of BI 730357 dosing regimens on the PASI outcome. The simulation of full exposure-response profile addressed key questions such as whether a higher efficacy may be obtained with a higher exposure. The bounded integer model can be used to adequately simulate composite data.
 Rendon, A. & Schäkel, K. Psoriasis Pathogenesis and Treatment. Int. J. Mol. Sci. 20, 1475 (2019).
 Fredriksson, T. & Pettersson, U. Severe Psoriasis – Oral Therapy with a New Retinoid. Dermatology 157, 238–244 (1978).
 Wellhagen, G. J., Kjellsson, M. C. & Karlsson, M. O. A Bounded Integer Model for Rating and Composite Scale Data. AAPS J. 21, 74 (2019).
 Ueckert, S. & Karlsson, M. O. Improved numerical stability for the bounded integer model. J. Pharmacokinet. Pharmacodyn. 48, 241–251 (2021). Q.X.O., A.K., and E.P. are Pharmetheus employees, they designed and performed the research. Support in design of research was supported by Boehringer Ingelheim; J.K., M.F., and B.W. are Boehringer Ingelheim employees.