Dissertation Date: November 19, 2026
The regional dose of nasal sprays and dry powder inhalers at the site of action is difficult to measure in vivo. As a result, bioequivalence assessment for the development of generic drug products relies on in vitro tests and costly clinical endpoint studies that provide limited information on regional deposition. Computational fluid dynamics (CFD) can estimate regional doses, but these models have traditionally relied on simplified assumptions and limited validation. The objective of this dissertation was to develop and validate CFD methods to optimize nasal spray (Aim 1) and dry powder inhaler (DPI) (Aim 2) drug delivery to treat respiratory diseases.
The nasal valve is a major barrier to nasal spray delivery to posterior structures. In Aim 1, our analysis of experimental data in pediatric nasal cavities revealed a paradoxical negative correlation between posterior dose and nasal valve area. We hypothesized that this discrepancy is due to the assumption that droplets remain where they hit the wall, when, in reality large droplets may splash and move along the walls after deposition. We therefore developed a formulation-aware multiphase CFD framework that couples a Lagrangian wall-film model, representing post-deposition liquid motion, with a non-Newtonian rheology representative of triamcinolone acetonide (Nasacort) nasal spray. The framework was validated against gamma scintigraphy experiments in six airway replicas from healthy and allergic rhinitis subjects, where posterior deposition averaged 46 ± 15%. Conventional deposit-on-touch CFD underpredicted posterior dose (27.5 ± 10.5%), whereas the wall-film model increased it to 40.7 ± 10.6%. Next, we applied this framework to evaluate spray administration technique in another 10 patients with nasal obstruction based on pre- and post-surgery CT scans. The ‘improved use’ strategy (bottle at 45° to the nasal floor and aimed at the nasal valve) increased posterior deposition in all cavities compared to the ‘current use’ strategy (bottle upright) (51.5% vs. 9.5%; p < 0.001). Nozzle direction had a greater impact on the dose delivered beyond the nasal valve than surgery.
The first objective of Aim 2 was to understand why certain DPIs pose a greater risk of steroid inhaler laryngitis, a side effect associated with DPIs intended for lung delivery. We estimated the laryngeal dose of inhaled corticosteroids at 30, 45, and 60 L/min using two turbulence models (large eddy simulation and k−ω SST). For the DPI with large particles (d50 = 50 μm), the laryngeal dose was 1.6-to-3.8-fold higher than the bronchial dose, whereas laryngeal and bronchial doses were similar for the DPI with small particles (d50 = 2 μm), suggesting that DPIs with larger particles pose a greater risk of laryngeal side effects. Building on this understanding of particle-size-dependent laryngeal deposition, the second objective was to determine whether particle size and breathing technique could be optimized to intentionally enhance laryngopharyngeal drug delivery for the treatment of laryngopharyngeal reflux, a condition with no effective treatment, using an airway model of a healthy adult. Laryngopharyngeal doses were 5.3%, 20.3%, and 2.2% of the emitted dose for DPIs with mass median aerodynamic diameters (d50) of 3, 10, and 50 μm, respectively, while inhalation duration and depth had a negligible impact. Therefore, a DPI with d50 = 10 μm was predicted to maximize laryngopharyngeal delivery.
Overall, regional drug delivery was governed by formulation properties, administration technique, and patient anatomy. The wall-film framework improved prediction of site-of-action nasal deposition, while DPI simulations identified particle size as a critical quality attribute of both therapeutic laryngopharyngeal delivery and steroid inhaler laryngitis risk. These in silico tools may support generic drug product development and bioequivalence assessment, although validation in larger cohorts is needed.
