Purpose: Dialysis is a commonly used method to determine in vitro drug release kinetics of nano-sized particulate drug delivery systems to provide valuable information for the formulation development, quality control, and regulatory filings. One pitfall of the dialysis method, however, is that dialysis membranes may pose significant barrier effects on the drug transport, delaying the appearance of released drug molecules in the sampling compartment. In this case, the apparent drug release results often do not properly reflect the actual drug release kinetics, which can cause misleading conclusions. To address this challenge, we propose a two-step approach which includes an experimental calibration of the drug diffusion across the dialysis membrane and applying a mathematical model to predict the nanocarrier release from the experimental data measured outside the dialysis bag. The model was tested on Doxil® (doxorubicin liposomes), and a good agreement was found between the experimental data and the predicted value. By taking barrier effects of dialysis membranes into consideration, our model can not only enable the proper interpretation of the data from dialysis studies, but also help to evaluate the dialysis methodology applied to in vitro drug release assays.
Methods: A mathematic model was developed based on a series of mass balances and assuming Fickian diffusion across the dialysis membrane. First, in the calibration experiments, the permeation kinetics of doxorubicin (DOX) through dialysis membranes was determined with free DOX solution in dialysis bags. The calibration constant (Kcal) of the dialysis membrane was calculated from the slope of the natural log of retained drug fraction in bags vs. time. Then, Doxil® release assay was performed using a USP-4 apparatus at 45 °C and the flow rate 16 ml/min as previously described (Yuan, Kuai et al. 2017). The DOX concentration outside the membrane was continuously monitored by a UV spectrophotometer at 480 nm. The actual drug release kinetics was calculated by applying the mathematical model to the measured apparent release utilizing the Kcal. To validate the model, the free in-bag DOX concentrations were predicted using the model, and the predicted values were compared with the experimental values.
Results: The barrier effects of different dialysis membranes were determined using calibration experiments (Table 1). Not only the cut-off molecular weight (COMW) but also the material properties were found to affect the permeability of DOX through dialysis membranes. When the model was applied to the Doxil® release data, an excellent prediction of the DOX concentration inside the bag was achieved (Figure 1), validating the model to predict actual drug release. As predicted by the model, an increase of free DOX concentration inside dialysis bags was observed during the first several hours of dialysis, indicating the permeation through the dialysis membrane, not the drug release, was a rate-limiting factor governing the apparent drug release profiles in the early stage of dialysis (Figure 1). As a result, the actual drug release kinetics was underestimated by the apparent drug release profile (Figure 2). Correcting apparent drug release results with our model provided a validated prediction of the actual release of DOX from Doxil®.
Conclusion: We have developed a general mathematical model and 2-step approach to determine the actual drug release kinetics from the apparent drug release data. The model is of practical implication for data interpretation, experimental design, and methodology evaluation.
Wenmin Yuan– University of Michigan
Anna Schwendeman– University of Michigan
Steven Schwendeman– Professor, Department of Pharmaceutical Sciences and the Biointerfaces Institute, University of Michigan, Michigan