Petra Jauslin* (1), Anna Largajolli* (1), Emilie Schindler (1), Tomohisa Saito (2), Luca Loprete (2), Nathalie Wagner (2), Vincent Duval (1). *Equal contribution.
(1) Certara Strategic Consulting, (2) Chugai Pharmaceutical Co., Ltd, (3) Nestlé Skin Health - Galderma Research and Development
Objectives: To characterize, in patients with moderate-to-severe atopic dermatitis (AD), the pharmacokinetic-pharmacodynamic relationship between serum concentrations of nemolizumab (a humanized antibody against interleukin-31 receptor A) and the following clinical outcome variables: eczema area and severity index (EASI, score range 0-72) and weekly average of peak pruritus numeric rating scale (PP-NRS, score range 0-10).
Methods: A total of 3’620 EASI scores from 525 patients and 4’578 weekly PP-NRS scores from 225 patients were analyzed. EASI scores were available from three randomized placebo-controlled clinical studies: a phase 1 SAD study (CIM001PJ: N=36, dose range: 0.3-3 mg/kg) [1], a phase 2a safety and efficacy study (CIM003JG: N=264, dose range: 0.1-2 mg/kg Q4W or 2QW) [2] and a phase 2b dose-range finding study (RD.03.SPR.114322: N=225, dose groups: 20 mg loading dose (LD) + 10 mg, 60 mg LD + 30 mg and 90 mg Q4W) [3]. PP-NRS scores were only available from the phase 2b study RD.03.SPR.114322. Both endpoints were modelled as continuous bounded scales. Logistic functions were used to transform the amount in the indirect response model compartment from [0, +∞] to the bounded scales of EASI ([0, 72]) or PP-NRS ([0, 10]). Longitudinal nemolizumab serum concentrations were described by a one-compartment population pharmacokinetic (popPK) model with linear elimination, first-order absorption with lag time and dose effect on bioavailability. Inter-individual variability (IIV) was included in clearance (CL), volume of distribution (V) and the absorption rate constant. CL and V were correlated at an individual level. Concentrations were log-transformed for analysis. Hence, the residual error was additive in the log domain. Covariate effects of serum albumin on CL and body weight on CL and V were identified. Empirical Bayes post hoc estimates generated from the popPK model were then used to drive turnover models for the respective pharmacodynamic (PD) endpoints. The structure of a model for pruritus visual analogue scale by Saito et al [4] was used as a starting point for both the EASI and the PP-NRS models. The drug effect was described with an inhibiting effect on the zero-order rate constant for production of response (Rin) in both cases. Linear and non-linear link functions and placebo effects were evaluated. For the description of EASI scores, a baseline model was implemented, in which the observed baseline response was included as a covariate acknowledging the residual variability [5]. The effect of covariates (age, body weight, and sex) on model parameters was assessed. All models were developed in NONMEM 7.3 [6].
Results: In both PD models, an inhibiting Imax model was chosen as most appropriate link function. For EASI, IC50 was difficult to estimate with sufficient precision. To stabilize the model, IC50 was therefore replaced by the parameter S0, computed as Imax/IC50, as suggested by Schoemaker et al [7]. A constant placebo effect was introduced as additive to the drug effect in both models. For EASI, this placebo effect was considerably larger in the Phase 2b study than in the earlier trials. It was therefore modeled as a study-specific parameter. IIV was included in S0, the first-order rate constant for loss of response (kout) and the placebo effect (P1). For PP-NRS, IIV was included on estimated baseline PP-NRS, Imax, P1, and kout. Owing to eta correlations, an omega block for these random effects was included in the covariate matrix. The residual error was combined additive and proportional for both models, and no covariates were identified in either case. All model parameters were estimated with acceptable precision (<30% RSE for fixed effects and ≤35% RSE for random effects). Visual predictive checks showed adequate predictive performance of the models (overall, as well as stratified by study [for EASI] and dose [for both models]).
Conclusions: The developed turnover models provide a good description of the EASI and PP-NRS data in AD patients treated with a wide range of nemolizumab doses. These models, together with a continuous-time Markov model for an additional PD endpoint (investigator’s global assessment [IGA] score, described in a separate abstract) allowed for the characterization of nemolizumab effect in patients with moderate-to-severe AD and will further support the development program of nemolizumab.
References:
[1] Nemoto O, et al. The first trial of CIM331, a humanized antihuman interleukin-31 receptor A antibody, in healthy volunteers and patients with atopic dermatitis to evaluate safety, tolerability and pharmacokinetics of a single dose in a randomized, double-blind, placebo-controlled study. Br J Dermatol. 2016;174(2):296-304
[2] Ruzicka T, et al. Anti-interleukin-31 receptor A antibody for atopic dermatitis. N Engl J Med. 2017;376(8):826–834.
[3] https://clinicaltrials.gov/ct2/show/NCT03100344.
[4] Saito T, et al. Dosage Optimization of Nemolizumab Using Population Pharmacokinetic and Pharmacokinetic-Pharmacodynamic Modeling and Simulation. J Clin Pharmacol. 2017;57(12):1564-1572.
[5] Dansirikul C, Silber HE, Karlsson MO. Approaches to handling pharmacodynamic baseline responses. J Pharmacokinet Pharmacodyn. 2008;35(3):269-83
[6] Beal SL et al. NONMEM 7.3.0 users guides. (1989–2013). Hanover: ICON Development Solutions.
[7] Schoemaker RC, van Gerven JM, Cohen AF. Estimating potency for the Emax-model without attaining maximal effects. J Pharmacokinet Biopharm. 1998;26(5):581-93
Reference: PAGE 28 (2019) Abstr 9088 [www.page-meeting.org/?abstract=9088]
Poster: Drug/Disease Modelling - Other Topics