Use of automation, dynamic image analysis, and process analytical technologies to enable data rich particle engineering efforts at the Drug Substance / Drug Product interface: A case study using Lovastatin

Abstract

Automation plays a vital role in reducing development time in the pharmaceutical industry, both for drug substance (DS) and drug product (DP). Many aspects of crystallization development (solubility, kinetics, solid form characterization, concentration) have been positively impacted by automation; both in hardware and data processing, in the last two decades. However, wet milling, a critical unit operation used to manipulate particle properties in a crystallization process, has been non-conducive for automation. This study discusses efforts towards automation of wet milling, coupled with off-the-shelf automation tools for faster execution of physical property design workflows. The hardware and software configuration used, along with challenges and prospects to achieve milling automation are discussed. Lovastatin was used as a model compound to demonstrate the capabilities of the automated platform to achieve morphology and size manipulation. Five different batches of lovastatin were produced using different combinations of milling and thermal cycling. Milling followed by thermal cycling led to formation of rougher, fused particles with a lower aspect ratio compared to those produced by thermal cycling followed by milling. In-situ particle video microscopy was used to decipher that the initial increase in particle size during thermal cycling was driven by a fusion of primary particles followed by growth. Key material property descriptors such as bulk/tap density, flow function coefficient, surface area, and particle size by laser diffraction were collected using traditional means while particle size, aspect ratio and smoothness were also measured using image analysis. Lastly, principal component analysis (dimensionality reduction) was performed to determine the key property relationships impacting flowability. It was found that fines as measured by d10 (10th percentile of the volumetric size distribution) and Hausner ratio had a high positive and negative correlation, respectively with flowability. The on demand milling platform combined with data rich trials using PAT is estimated to accelerate process development for particle engineering and reduce the drug substance delivery time to drug product team.