Model-based analysis of the treatment window in passive cutaneous anaphylaxis mouse model applying an optimized study design
Felix Jost (1), Andrew Smith (2), George Qian (2), Yui-Hsi Wang (2)
(1) Sanofi-Aventis Deutschland GmbH, R&D, Frankfurt, Germany, (2) Sanofi, R&D, (Cambridge), MA, USA
Introduction:
Passive cutaneous anaphylaxis (PCA) is a type 1 hypersensitivity reaction induced by an exogenous effect and is used as an acute translational animal model to study immune diseases [1]. In our model variant either saline as a vehicle or a neuropeptide (NP) is administered to a mouse paw to provoke short swelling of the treated tissue defined as the pharmacodynamic (PD) outcome. Saline leads to a paw size increase through the administered volume whereas NP in addition mediates a neurogenic inflammation through the activation of the innate immune system.
The objective of our study was the analysis of the treatment window of paw swelling defined by the administration of saline or NP. An optimized study design was used with multiple sample points to address the large variability in the PD outcome. A challenge model was developed to describe the paw thickness over time to quantitatively study the impact of both challenges.
Methods:
Genetically modified female and male C57/b6 mice were used in the PCA studies. Mice were treated either with subcutaneous administration of saline or NP with a volume of 25uL. Two study designs were applied in which at each time point the paw size was measured three times. Study design 1 consists of two measurement time points, a baseline measurement taken 5 minutes before vehicle administration and a second measurement 30 minutes afterwards. Study design 2 consists of multiple measurement time points including the baseline measurement from design 1 and 6-8 measurements after vehicle administration ranging from 5-300 minutes. For study design 1 in total 20 studies with 124 and 139 samples (mean out of three repeated measurements) pre and post saline and NP challenge were collected, respectively and for study design 2 three studies were performed with 5-10 animals per group.
The data from study design 1 was analyzed by statistical hypothesis testing (one sided t-tests with equal variance) and the longitudinal data was described by a challenge model.
The dynamic of paw size is described by an ordinary differential equation with a baseline paw size B, a paw size reduction rate Kp_c after challenge c and an inflow rate Kin = B Kp_c. Each challenge, effecting the inflow rate Kin via Kin*(1+C(t)), is described by an exponential function C(t) = A Ks t exp(- Ks t) with time t and parameters A and Ks, the swelling amplitude and induction rate constant, respectively. During model fitting, individual parameters Kp_c, A1_c and K1_c were tested for each challenge c to evaluate the challenge-dependent impact on the paw size dynamics [2].
Individual study analysis was performed in PhoenixWinNonlin 8.2.2 using a Naïve pooled approach with a Gauss-Newton differential equation solver. Descriptive statistics was performed within Rstudio 2022.02.1.
Results:
For study design 1, the null hypothesis of a smaller paw size after saline compared to NP challenge could not be rejected in 4 out of 20 tests. The longitudinal data from both challenge groups was adequately described by the model with one induction rate constant and two swelling amplitudes and paw size reduction rates simultaneously. The half-life of the swelling induction was 0.7-1.3 min and the half-lives of paw size recovery after saline and NP were 21-44 min and 82-140 min, respectively. The swelling amplitude ranged from 14-44 and 56-107 for saline and NP, respectively. Thus, the two challenges not only differ in swelling reduction represented by the significant difference at a single time point and half-lives of paw size recovery but also in maximal swelling observed shortly after challenge administration.
The estimated parameter values had good precision but for the study without a 5 minute sample, Ks was not identifiable and needed to be fixed to a prior estimated value. A factor 2 in parameter estimates between studies was pointing to a repeatable reliable PD response.
Conclusions:
Study design 2 provided an in depth understanding of the dynamics of paw thickness after two challenges with differences in maximal paw swelling and duration of paw size recovery. In future studies, additional groups could be included to study the impact of test compounds on the PD outcome. To minimize the variability in PD outcomes with minimal animal numbers, the use of multiple sample points can give a more robust and reliable interpretation of in vivo efficacy and ranking of compounds than a single time point design.
References:
[1] Siebenhaar, F. et al. Mast cell-driven skin inflammation is impaired in the abstence of sensory nerves. J Allergy Clin Immunol 121 (4), 955-961 (2008).
[2] Gabrielsson, J. et al. Modeling and design of challenge tests : Inflammatory and metabolic biomarker study examples. Eur J Pharm Sci 67, 144-159 (2015).