Track: Manufacturing and Analytical Characterization - Chemical - Integrated and Continuous Processing and Manufacturing - For Use in Drug Product Manufacture
Category: Poster Abstract
In-line Assessment of Powder Stream Density Variation and Its Effect on Near Infrared Measurements during Scale-Up of a Simulated Continuous Process
Purpose: The success of continuous pharmaceutical manufacturing often relies on the effective implementation of process analytical technology (PAT) for real-time quality monitoring of the process stream. Near-Infrared (NIR) spectroscopy has been widely recognized as a powerful process analytical technology (PAT) tool for monitoring powder stream chemical composition in continuous pharmaceutical processes. However, quantitative NIR monitoring methods are sensitive to changes in the physical properties of samples, which often decreases the prediction performance. Often, different powder stream flow rates are necessary to match the production requirements (e.g. scale-up) of a continuous process. The physical variation (e.g. density) introduced when NIR quantitative measurements are performed on dynamic powder samples at different flow rates is a driving factor in model complexity and lack of robustness. Collecting NIR measurements across multiple flow conditions is a common approach to counter any physical variation due to flow rate and develop robust NIR quantitative models. Such an approach takes time and a significant amount of materials to collect a relatively large calibration set. Thus, the development of a more efficient strategy to characterize, understand and mitigate the impact of powder physical properties on NIR measurements during scale-up is therefore highly desired and is the focus of this study. Methods: A model blend formulation of acetaminophen (19.5 %w/w), Avicel PH-200 (47.7 %w/w), lactose monohydrate (31.8 %w/w), silica (0.5 %w/w) and magnesium stearate (0.5 %w/w) was run thorough a twin screw loss-in-weight feeder (K-tron, K2MLT20) and quartz tube assembly, delivering continuous powder streams at various design levels of flow rate and tube angle. A three-level full factorial design of two factors was used to vary the process parameters. The flow rate was set at 15 kg/h, 25 kg/h and 30 kg/h, while tube angle was adjusted at 37°, 40° and 43°. The powder stream was monitored simultaneously by in-line diffuse reflectance NIR, live imaging and dynamic weight measurements. NIR spectra were collected through the quartz tube using a NIR spectrometer (model: NIR-256-2, Control Development, Inc.) with a halogen light source and bifurcated fiber optics probe (Ocean Optics, Inc). Live video frames (images) were collected using an Ethernet CCD camera and LED light panel system (JM Canty, Inc.) to calculate the powder stream volume via image analysis (Matlab, R2014b, The Mathworks Inc.). The dynamic measurements of powder mass, along with the calculated volume, were used to estimate powder stream density. A two-way analysis of variance (ANOVA) with interactions model was implemented to assess the effect of process parameters (flow rate and tube angle) on powder stream density. Principal component analysis (PCA) was used to characterize and compare the influence of both process parameters and powder stream density on the NIR spectra. A partial least squares (PLS) regression model was developed to assess the correlation between the spectroscopic response and powder stream density. Results: An experimental setup for the continuous monitoring of apparent powder stream density was developed by enabling in-line live imaging and dynamic weight recording. This in-line method facilitated the assessment of powder stream density variation on the NIR spectra as a function of two process parameters, flow rate and tube angle. At a significance level of 0.01, a two-way main effects model with interactions demonstrated that both tube angle and flow rate have a significant influence on powder stream density with a p-value of less than 0.0001, respectively. A p-value of 0.02 for the interaction term indicated that the effect of tube angle and flow rate on powder density are independent from one another. A PCA model of 3 PCs, which explained more than 99% of the spectral variance, was used to evaluate the influence of both process parameters and powder stream density on NIR spectra. The PC1 (96.41%) vs PC2 (2.14%) scores plot showed spectral data trends with respect flow rate and tube angle. The PC1 (96.41%) vs PC3 (1.06%) scores plot revealed strong patterns in the direction of PC3 associated with the changes in the powder stream density due to changes in the scale. PC1 vs PC3 scores plot showed density and spectral matching between powder streams at different process conditions. PCA results and the strong correlation between powder stream density and NIR spectra further evidenced that powder stream density is a strong source of NIR variability due to flow rate. Conclusion: A PAT method for monitoring in real-time the physical variation of the powder stream that influences NIR spectral variance was developed by incorporating in-line imaging and dynamic weighing. Statistical analysis and multivariate modeling confirmed powder density as a significant source of spectral variability due to flow rate. This effect can be simulated through changes in density from adjusting the tube angle. Besides allowing for a broader process understanding, results elucidated the processing conditions to be considered for building a density-insensitive NIR model with enhanced robustness against scale-up for the prediction of drug content in a simulated continuous process.