Correlating Powder Flowability to Particle Size Distribution Using Chemometric Techniques

A compound terminated in Phase III clinical trials was being developed as a 140 mg dose.  In order to meet the needs of the patient  population, the tablet must be no larger than 450 mg, resulting in a high drug substance loading.  The drug product was manufactured using a roller compaction process.  The high drug load required that the drug substance have good flow properties as a dry powder.  An understanding of the impact that the particle size distribution of the drug substance on the powder flowability was needed to allow for the establishment of particle size specifications on the drug substance.  Principal Components Analysis (PCA) was applied to the particle size distribution data generated from numerous crystallization conditions and scales to identify clustering.  A predictive model for powder flowability was built based on the particle size distribution data using Partial Least Squares (PLS).  This model was further refined using local weighting techniques to deliver better prediction metrics.  The results of these modeling efforts will be presented.