(759a) Simulation Study Comparing Oil Emulsion Particle Size Distributions On a Performance Basis
AIChE Annual Meeting
2013
2013 AIChE Annual Meeting
Pharmaceutical Discovery, Development and Manufacturing Forum
QbD for the Development of Drug Release and Dosage Systems
Thursday, November 7, 2013 - 3:15pm to 3:40pm
Assessments of particle size distribution can be critical in the design and development of multiphasic liquid pharmaceutical formulations and in establishment of the pharmaceutical equivalence of generic emulsion and suspension drug products.
The assessment of particle size distribution similarity is often reduced to comparing mean particle size or span. While these numbers are easy to compare, they fail to capture any skew in the particle size distribution and the physical interpretation of these quantities is limited for multimodal distributions. Particle size distribution similarity needs be quantitatively defined in a way that rationally accounts for these limitations and further investigate if a similarity metric can be used a surrogate for in vitro release performance differences.
In this work, batch particle size distributions were defined with a three parameter log skew normal distribution. Monte Carlo sampling was used to compute the population particle size distribution to account for uncertainty in the three parameters. Systematic differences in test populations were considered to produce several test populations. Each test population’s similarity to the reference population was assessed with two probabilistic metrics of distribution similarity: the overlapping coefficient (OVL) and a related measure that considers the relative location of the probability mass called PROB. Both of these metrics consider any skew in the particle size distribution as well as provide consistent meaning when either the test or reference population is multimodal. We show that the PROB measure of similarity correlates with differences in time to 25, 50 and 75% release as computed by a Higuchi type model under sink conditions and diffusion limited transport. Significance of the differences was assessed with a Wilcoxon rank sum test at 95% confidence.