PAT SUPPLEMENT


modelling) in Figure 4. Two are the principal component scores (i.e. t1 and t2), as per cross- validation, explaining about 60 per cent of the variability. Each batch is now represented by one single data point of coordinates (t1, t2). As can be seen in Figure 4, all the manu -


facturing operations fall within the 95 per cent confidence ellipsoid, no matter if golden or not. This indicates an overall similarity of behaviour across all the operations in scope of the MVDA. We could therefore conclude that the behaviour during the upstream process does not appear to have an impact on the final DS characteristics.


REFERENCE


FIGURE 3 Golden MSPC model for the main bioreactor. The average or desired process signature is also shown (pale green trace). The red traces are indicative of the variability between reference batches (±3 standard deviations from the average)


In addition, the inoculation density of cells to the main bioreactor appears to be important and the medium feeding rate should be a function of the actual viable cell density in the main bioreactor phase. Generally, it appears important to maintain the aeration rate at a lower level at the beginning of the culture. All the batches have then been summarised by combining all the process data (both seed


1. S. Wold et al. Multi- and Megavariate Data Analysis Part 1, 2nd edition – chapters 12 and 13


BIOGRAPHY


to improve the understanding of the process ”


“MVDA is a valid tool


and main bioreactor) over time and the DS quality attributes at the batch level (i.e. using a Batch-level Statistical Process Control


Dr. Marianna Machin is currently Senior Process Analytical Technology Expert at Novartis. She is responsible for MultiVariate Data Analysis and Process Analytical Technology within Global Technical Operations. Prior to taking her current position, Dr. Machin worked at


GlaxoSmithKline as Quality by Design Champion in the Pharmaceutical Development department and also worked on analysis, interpretation and representation of PharmacoKinetics data in the Clinical Pharmacokinetic Modelling and Simulation group. Marianna is a chemical engineer by background and holds a PhD in bioengineering from Padua University in Italy in collaboration with The Scripps Research Institute (La Jolla, California).


BIOGRAPHY


Antonio Peinado Amores is currently a senior PAT expert in Global Pharma - ceutical Engineering at Novartis. His current position involves implementing NIR and Raman solutions for monitoring pharmaceutical processes and using Multivariate Data Analysis for Process


Understanding and Troubleshooting. Previously, Antonio worked for five years in Pharmaceutical Development at GlaxoSmithKline. Antonio obtained his PhD at the University Autonomas of Barcelona in 2004. He completed a post-doc in chemometrics sponsored by Total-Fina-Elf at CNRS-Lille.


BIOGRAPHY


FIGURE 4Batch-level principal component analysis score scatter plot obtained by Batch-level Statistical Process Control (BSPC) modelling which summarises the cultivation data for each operation into a single data point. Golden batches are represented by blue triangles and non-golden batches by red squares. All the batches lie within a 95 per cent confidence interval


8


European Pharmaceutical Review Volume 16 | Issue 6 | 2011


Lorenz Liesumis currently a senior PAT expert in Global Pharmaceutical Engineering at Novartis and is leading PAT projects within the manufacturing department. Previously he worked in chemical and pharmaceutical development for Roche and Novartis as


an analytical scientist. Lorenz is a chemist by training and obtained his PhD at the ETH Zurich in the field of magnetic resonance spectroscopy.


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17