A Hybrid Computational Framework for Gastroretentive Drug Delivery: Integrating PBPK Modelling and Design of Experiments for Systematic Formulation Optimisation

Introduction: In resource‑limited settings, extensive experimental formulation screening is impractical. Cimetidine, an H₂‑receptor antagonist, suffers from low oral bioavailability (30–40%) and a short half‑life, demanding gastroretentive delivery. We hypothesised that a hybrid framework integrating Physiologically Based Pharmacokinetic (PBPK) modelling with Design of Experiments–Response Surface Methodology (DoE‑RSM) could efficiently optimise an in‑situ gelling system, providing a cost‑effective alternative to extensive pharmacokinetic studies.
Methods: PBPK modelling (PK‑Sim® v11.3), using a proxy population with parameter adjustments representative of Sub‑Saharan African physiology, predicted how the pH‑responsive buoyant system modulates cimetidine absorption, including gastric residence time and viscosity targets. A 10‑run Central Composite Design then optimised sodium alginate (2–6% w/v) and calcium carbonate (0.5–1.5% w/v), targeting buoyancy lag time (<5 min), floating duration (>8 h) and pH compatibility. Model performance was assessed via R², F‑statistics and residual diagnostics.
Results: PBPK predicted a mean residence time of 32.08 h and AUC_inf of 27,106 µmol.min/L (comparative validation pending). DoE‑RSM identified sodium alginate as the dominant factor (buoyancy lag coefficient +1.16, p=3.29×10⁻⁵, R²=0.813). The model‑optimised formulation (4% alginate, 1% CaCO₃) predicted a buoyancy lag time of 4.71 min and viscosity of 555.4 cP; experimental confirmation is required. R² ranged from 0.65 to 0.813 (p<0.01).
Conclusions: This proof‑of‑concept computational framework demonstrates the feasibility of combining mechanistic PBPK prediction with statistical DoE‑RSM for gastroretentive formulation development. The PBPK approach, using a proxy African population and parameter adjustments, offers a cost‑effective alternative to extensive PK studies, is valuable in resource‑limited settings, and provides a methodology applicable to other acid‑labile drugs with narrow absorption windows. Key limitations include unvalidated predictions, the limited 10‑run design, and the absence of experimental and in vivo corroboration.