(M1230-11-74) Mass Transport Analysis of Solute Transport through Lipophilic Membranes

Purpose: Membrane-based permeability assays, such as cell culture models, high-throughput parallel artificial membrane permeability assay, and diffusion cell-based flux measurements, are widely used in early- and mid-development of pharmaceuticals to evaluate drug permeability and formulation bioavailability. However, the membrane permeation rate of a drug is subject to various factors such as receiver buffer condition, the presence of colloidal species in the donor, the geometry of the apparatus used, as well as the physicochemical properties of the drug. The goal of this study is to use mass transport models to understand the membrane permeation of different drugs and formulations, thus enabling accurate prediction of membrane permeability.

Methods: A side-by-side diffusion cell was utilized to evaluate trans-membrane flux. A hydrophilic PVDF filter membrane (450 nm, Millipore Sigma) impregnated with GIT-0-lipid solution (Pion Inc., Billerica, MA) was used (Figure 1). Various poorly soluble drugs that have slow crystallization propensities as well as moderate to high amorphous solubility values were chosen as model compounds. The donor chamber was filled with pH 6.5 phosphate buffer containing the drug with or without amorphous nanoparticles (NP), while the acceptor compartment contained the plain buffer or the same blank buffer with 2% pre-dissolved sodium dodecyl sulphate (SDS) as a sink condition modifier. The system was stirred at a controlled rate of 600 rpm and maintained at 37°C. Both donor and receiver drug concentrations were simultaneously monitored as a function of time. Lucifer Yellow was used as a marker for membrane integrity. The experimental data obtained were fitted into different mass transport models (Figure 1). Briefly, the free drug and blank buffer system was used to obtain mass transfer rate coefficients k₁ and k₂, which depend on the drug’s hydrophobicity (logP) and the thickness of the effective unstirred water layer (UWL) adjacent to the membrane. Subsequently, free drug data collected with SDS were used to calculate ε₄. In the presence of amorphous nanoparticles, ε₁ (describing the contribution of paricles to transmembrane flux) and ε₃ (describing membrane resistance) were simultaneously fitted. Using these parameters fitted from different control experiments, the donor and receiver concentrations for systems containing amorphous nanoparticles with SDS in the receiver were then calculated and compared with experimental values for model validation.

Results: As shown in Figure 2, fitted logP values were consistent with reported logP values. The effective UWL thickness was then determined from the individual k₁ values fitted from data obtained for each drug. A similar effective UWL of 66.87 ± 4.988 µm was obtained for all drugs with the exception of danazol (31.74 µm). The impact of receiver sink condition on membrane flux is also shown in Figure 2. A linear relationship was observed between ε₄ and the degree of crystalline solubility enhancement by the receiver sink buffer, consistent with altered membrane-buffer partition coefficients by SDS solubilization of the drug. In the absence of receiver sink, exceptionally small ε₁ values were obtained for most drugs, suggesting negligible particle drifting effect. This was due to negligible particle dissolution and slow solute removal in the UWL as a result of slow membrane permeation rate, confirmed by membrane drug accumulation calculations. The role of membrane resistance, reflected in ε₃ values, were similar across all drugs. Two model drugs with higher logP values, danazol and felodipine, showed higher ε₁ values, consistent with their higher membrane drug accumulation. Finally, in the presence of both amorphous nanoparticles and receiver sink, significant particle drifting effect was observed. Similar fitted ε₁ values were obtained for most drugs, consistent with rapid particle dissolution. Using ε₃ and ε₄ values obtained from control experiments, the donor and receiver concentrations were then calculated, showing good agreements with experimental data for most drugs evaluated. Representative fitting examples are shown in Figure 3.

Conclusion: The membrane permeation rate of different drugs can be predicted as a function of drug logP, receiver sink condition, and in the absence or presence of amorphous nanoparticles. These results also revealed several key factors affecting the particle drifting effect, which may contribute to better in vitro-in vivo translations and bioavailability predictions.

Acknowledgements: The authors gratefully acknowledge the Center of Pharmaceutical Processing Research (CPPR) for funding support.

Figure 1. Schematic illustration of the side-by-side setup used, where V_1, V_2, and V_m represent the volume of solution present in the donor receiver, and membrane, respectively, while C_1, C_2, and C_m represent the donor, receiver, and membrane solution concentration at time t, whilst ε_2 k_1, ε_3 k_1, ε_1 k_2, and ε_4 k_2 are the overall mass transport rate constants describing solute transfer from donor and receiver bulk solutions to the membrane and from the membrane to bulk solutions.

Figure 2. Fitted logP, effective UWL thickness, and ε_4 values for different model drugs

Figure 3. Representative fittings results showing agreements with experimental data