(600a) Avoiding Distribution Confusion: Using the Information in Particle Size Distributions

Particle size distributions of solid active pharmaceutical ingredients (APIs) govern the material properties important to bioavailability such as dissolution. To characterize API properties, rich data sets such as Morphologi G3 or surrogate measures such as laser light scattering are used to measure the full population of particles within a sample. However communicating this distribution often relies on histograms of the distribution or simple summary measures of volume weighted quantiles, e.g D10, D50, D90. These measures cannot adequately capture multi modal distributions and can be insufficient for setting material specifications.

By using concepts from nonparametric statistics and information theory, the information loss incurred by using these summary metrics and histogram binning can be quantified. A method is proposed for compressing a high resolution distribution into a summary measure that best captures the underlying particle characteristics. In addition to a theoretical study of information, an example of different summarization approaches applied to a PBPK model for a model API will be presented. The resulting approach of maximizing the value of particle characterization data is a useful methodology for communicating and modeling the behavior of particulate pharmaceutical compounds.