(T1530-10-68) In Silico Predictive Modeling of the Pharmacokinetic (PK) Profiles of Nasal Sprays in Adults and Children

Purpose: In silico models can provide a way to evaluate how differences in product- and patient-related characteristics can affect clinical performance. For nasal drug products like suspension-based nasal sprays, these characteristics can include the formulation, device, nasal anatomy, and the region of the depositing particles within the nose. To facilitate a better understanding of how these characteristics may impact in vivo performance metrics like systemic exposure, the objective of this study was to develop a pharmacokinetic (PK) model for predicting the exposure of triamcinolone acetonide (TAA) in adult and pediatric patients following nasal spray administration. A computational fluid dynamics (CFD) model coupled with a dissolution, absorption, and clearance (DAC) methodology was used to predict the PK profiles (CFD-DAC-PK) for nasally-administered TAA (110, 220, and 440 mcg) in the adult and pediatric models [1, 2]

Methods: Figure 1 shows the coupled CFD-DAC-PK approach. This approach included two sets of in vitro 3D nasal cavity geometries (low (L), medium (M) and high (H) drug delivery models) initially developed as adult geometries [3] but extended to pediatric nasal geometries. These geometries were used for CFD models developed with ANSYS FLUENT 2022 R2 that were applied to predict regional spray transport and deposition. The 3D in silico model was then unwrapped to create a 2D ‘planar’ version of the 3D geometry of the airway surface liquid (ASL) and epithelium regions, using MATLAB R2022a. This was the starting point for the post-deposition DAC-PK modeling in ANSYS FLUENT. The DAC mechanistic model simulations of the suspended TAA particles included dissolution, evolution of the drug concentration profile, drug diffusion from ASL to the epithelium, and posterior epithelium absorption along with mucociliary clearance. The DAC model was combined with a compartmental PK model to calculate systemic exposure over time. The model was initialized by specifying mucus and tissue properties in the ASL and epithelium regions. Before injecting the suspended particles, laminar fluid flow equations were solved in the DAC-PK model by implementing a user defined function (UDF) for specifying liquid mass source in the ciliated region (posterior nasal cavity) using an approach as reported by Rygg et al. [2]. Adult PK model parameters were estimated based on in vitro and/or in vivo data. Allometric scaling was done for the pediatric models, where the predicted PK profile data were plotted alongside the in vivo data and analyzed in Python 3.10. The area under the curve (AUC) was estimated using the trapezoidal method from the NumPy library in Python.

Results: Figures 2A and 2B show the plasma concentration profiles for the three adult models (L/M/H) for a nasally-administered TAA dose of 110 mcg and 220 mcg, respectively, compared to publicly available in vivo data [4]. The DAC-PK model appeared to overestimate the maximum plasma concentration (C_max) for all three adult models compared to the in vivo data. However, the predicted AUC values appeared to agree with the in vivo data for the adult L and H models (Figure 2D). Figure 2B shows that the predicted C_max values for the adult M model was comparable with the in vivo values and the predicted AUC for the L and H model appeared to agree with the in vivo data. Figures 3A and 3B show the plasma concentration profiles for the three pediatric models (L/M/H) for a nasally-administered TAA dose of 110 and 440 mcg, respectively. The DAC-PK model appeared to overpredict C_max with a comparable t_max for the pediatric M model; however, it underpredicted the t_max for the pediatric L and H models, while overpredicting the AUC compared to the in vivo data (Figure 3C). In Figure 3B, the DAC-PK model appeared to predict a C_max value that is comparable to the in vivo data for the pediatric H model and a comparable value of AUC to that presented in the in vivo study by Nayak, et al. [5] for the pediatric L model (Figure 3C). However, numerical instability was observed for the pediatric M model for the 440 mcg dose, which remains to be resolved.

Conclusion: Overall, the DAC-PK modeling approach appeared to predict comparable plasma concentration profiles in a model-specific manner. The adult L and H model PK profile predictions showed comparable AUC values to in vivo measurements for the 110 and 220 TAA mcg doses, while the adult M model appeared to underestimate the AUC. Interestingly, the C_max in the adult M model was slightly comparable for the 110 mcg and 220 TAA mcg doses, respectively, suggesting a faster clearance mechanism. However, both the C_max and AUC were overestimated for the 110 mcg TAA dose in the pediatric L and M model while only the AUC was overestimated for the H model, which suggested a slower clearance mechanism. Spray wall interactions, among other factors, will be considered in future models to improve predictability.

References: 1. Dutta R, Kolanjiyil A, Golshahi L, and Longest P. Development of a CFD-PK Nasal Spray Model with In Vivo Human Subject Validation. Respiratory Drug Delivery (RDD) 2022. 1: 483-488.
2. Rygg, A., Hindle, M., & Longest, P. W. (2016). Linking suspension nasal spray drug deposition patterns to pharmacokinetic profiles: A proof-of-concept study using computational fluid dynamics. Journal of pharmaceutical sciences, 105(6), 1995-2004.
3. Alfaifi A, Hosseini S, Esmaeili AR, Walenga R, Babiskin A, Schuman T, Longest W, Hindle M, Golshahi L (2022). Anatomically realistic nasal replicas capturing the range of nasal spray drug delivery in adults. International Journal of Pharmaceutics. 622,121858.
4. Roy P, Qiu W, Tornoe C. (2007). FDA-Nasacort®AQ Clinical Pharmacology Review - NDA 20468 SE05, S-24.
5. Nayak, A. S., Ellis, M. H., Gross, G. N., Mendelson, L. M., Schenkel, E. J., Lanier, B. Q., ... & Smith, J. A. (1998). The effects of triamcinolone acetonide aqueous nasal spray on adrenocortical function in children with allergic rhinitis. Journal of allergy and clinical immunology, 101(2), 157-162.

Acknowledgements: Funding for this work was made possible, in part, by the U.S. Food and Drug Administration through Contract 75F40120C00172 and Broad Agency Announcement (BAA) HHSF223201810144C. Abhinav Mohan was supported by a fellowship program administered by the Oak Ridge Institute for Science and Education (ORISE) through an interagency agreement between the U.S. Department of Energy and the U.S. FDA. The views expressed in this abstract are from the authors only and do not necessarily reflect the official policies of the Department of Health and Human Services, nor does any mention of trade names, commercial practices, or organization imply endorsement by the United States Government.

Figure 1: Translation of the 3D model to a 2D ‘planar’ mesh (top) used in the coupled CFD-DAC-PK approach (bottom) [1].

Figure 2: DAC-PK profiles shown for A) 110 mcg, B) 220 mcg, C) 440 mcg doses of triamcinolone acetonide, and D) data table outlining the tmax, Cmax, and AUC. For the DAC-PK profiles, a single run is shown, while the in vivo data shown is a mean with standard deviation of the plasma concentrations at each time point from all the subjects evaluated (n = 15) [4].

Figure 3: DAC-PK profiles shown for A) 110 mcg, B) 440 mcg doses of triamcinolone acetonide, and C) data table outlining the tmax, Cmax, and AUC. For the DAC-PK profiles, a single run is shown, while the in vivo data for the 110 mcg dose shown is a mean with the standard deviation of the plasma concentrations at each time point from all the subjects evaluated (n = 15) [4]. For the 440 mcg dose, the in vivo data shown is a mean with the standard error of the tmax and Cmax from all the subjects evaluated (n = 25-28) [5].