(W1230-07-38) Ethanol Solubility of Rate Controlling Polymer in Modified Release Formulation Can Impact Alcohol Dose Dumping: An Assessment Through Principal Component Analysis

Purpose: In vitro alcohol dose dumping (ADD) is a quality tool used to predict the potentiation of accidental dosing dumping from modified release (MR) formulations when mixed with alcoholic beverages. Two key factors that can influence ADD – active pharmaceutical ingredient (API) solubility in alcohol and formulation designs with different rate controlling (RC) polymers- were explored in this study. Three formulation designs - osmotic tablets, hydroxypropyl methyl cellulose (HPMC) matrix tablets and ethyl cellulose coated pellets in capsules (ECCP capsules) were investigated. To understand the driving force for ADD in MR formulation, we developed a sequential analysis using principal component analysis (PCA). PCA was employed to identify differences among products with similar dissolution profiles. PCA allows visualization of dissolution profile differences by reducing the multi-timepoint data to select principal components (PCs).

Methods: R-Studio was used to build two sequential PCA models to evaluate the relation of ADD with the alcohol solubility of; (1) API (Model-1) and (2) RC polymer solubility (Model-2). For Model-1, the categorical alcohol solubility of 10 APIs and their ADD dissolution data from 40 abbreviated new drug applications (ANDAs) were collected. APIs were labeled as either class 1 (slightly soluble in alcohol defined as 1-30 mg/ ml) or class 2 (soluble in alcohol defined as more than 30 mg/ml). For Model-2, a different data set including 53 ANDAs was used. MR formulations are divided into three groups based on their formulation design: 1) ECCP capsules, 2) HPMC matrix tablets, and 3) osmotic tablets. The selection of formulation designs is based on the distinct drug release mechanisms. For both Model-1 and 2, percent drug release from the highest strength at 30 min and 120 min in 0%, 20% and 40% alcohol media were collected. PCA is applied to identify similarities and dissimilarities among the independent factors (API solubility in alcohol and formulation design) of data. Percent release was used as numerical data which was analyzed in PCA. PCA score plots were used to evaluate the data structure and identify clusters, outliers, and trends. PCA loading plots determines which variables demonstrated the most impact on each component. Both plots were produced using the package ggplot2 in R-Studio. The prompt “prcomp()” is a function of R programming language which scales and centers the data set. The resulting data is then used to generate the PC scores, which subsequently is plotted as loading and score plots.

Results: Model-1 contains two classes of categorical data based on solubility. It appears PC2 represents the variability in data based on the alcohol concentration in the dissolution media. There was no apparent trend observed for PC1. The score plot for Model-1 did not reveal any distinct clusters with respect to API solubility (figure 1). Model-2 is separated into three groups of categorical data based on the identified formulation design. PC2 and PC3 were selected for the score plot based on the loading plot and the final score plot ability to differentiate ADD data based on the formulation design. PC2 appears to also represent the variability in data based on the alcohol concentration in the dissolution media while PC3 appears to separate the data based on dissolution time. The score plot and loading plot for Model-2 (figure 2) shows that HPMC matrix data points are clustered around the center of PC3 with little variability in the quadrant represented by low alcohol concentration (0% and 20%) at 30 min. This suggest that the HPMC matrix tablets are largely unaffected by higher concentration of alcohol 40% and may preferentially release the maximum percent of drug in 20% alcohol. The cluster for ECCP capsules is spread across and predominantly located in the right half of the PC space which represents data collected in 40% alcohol media. This suggests that the ECCP capsules experience maximum drug release in 40% alcohol. As expected, release from osmotic tablets does not appear to be influenced by alcoholic media or time as evidenced by the centered clustering of this group in the PCA space. This aligns with our current understanding that osmotic tablets release drug at a rate that is independent of dissolution hydrodynamics. HPMC is insoluble in alcohol, ethyl cellulose is soluble in alcohol and therefore, ADD of these formulations appears to be dependent on RC polymer solubility. The score plot of Models 1 and 2 showed that some data sets are outside the clusters. This deviation can be explained by other factors affecting ADD apart from the independent variables and factors selected for this PCA studies.

Conclusion: Model-1 results suggest that the API solubility in alcohol appears not to impact ADD. Model-2 results suggest that formulation design and solubility of RC polymer is a critical factor affecting ADD in MR formulations.

Acknowledgements: Disclaimer: This work reflects the views of the authors and should not be construed to represent FDA’s views or policies.

a) PCA score plot and b) PCA loading plot from Model-1 – Impact of API solubility on ADD using alcohol concentrations as variables and two dissolution time point i.e., 30 minutes and 120 minutes.

a) PCA score plot and b) PCA loading plot from Model-2 – Impact of formulation design on ADD using alcohol concentrations as variables and two dissolution time point i.e., 30 minutes and 120 minutes.