www.els-journal.com Mathieu Streefland 1 Dirk E. Martens 1 E. Coen Beuvery 2 Ren´ e H. Wijffels 1 1 Bioprocess Engineering, Wageningen University, Wageningen, The Netherlands 2 PAT Consultancy, Vianen, The Netherlands Review Process analytical technology (PAT) tools for the cultivation step in biopharmaceutical production The process analytical technology (PAT) initiative is now 10 years old. This has resulted in the development of many tools and software packages dedicated to PAT application on pharmaceutical processes. However, most applications are restricted to small molecule drugs, mainly for the relatively simple process steps like drying or tableting where only a limited number of parameters need to be controlled. A big challenge for PAT still lies in applications for biopharmaceuticals and then especially in the cultivation process step, where the quality of a biopharmaceutical product is largely determined. This review gives an overview of the currently available tools for monitoring and controlling the biopharmaceutical cultivation step and of the main challenges for the most common cell platforms (i.e. Escherichia coli, yeast, and mammalian cells) used in biopharmaceutical manufacturing. The real challenge is to understand how intracellular mechanisms (from synthesis to excretion) influence the quality of biopharmaceuticals and how these mechanisms can be monitored and controlled to yield the desired end product quality. Modern “omics” tools and advanced process analyzers have opened up the way for PAT applications for the biopharmaceutical cultivation process step. Keywords: Biopharmaceutical cultivation / Biopharmaceutical manufacturing / Bioreactor monitoring / Process analytical technology (PAT) / Quality by design (QbD) Received: July 3, 2012; revised: October 25, 2012; accepted: December 24, 2012 DOI: 10.1002/elsc.201200025 1 Process quality in the (bio)pharmaceutical industry The quality of any product or good is dependent on how well and robust it is designed. For mass produced products, it is also dependent on how well the process is designed. Controlling the process towards a desired end point will thus control the quality of the product. This line of thinking arose around the Second World War in United States and Japan and was implemented fully by William Edwards Deming to boost Japans post World War industry in consumer goods [1]. His principles have been Correspondence: Dr. Mathieu Streefland (mathieu.streefland@ wur.nl), Bioprocess Engineering, Wageningen University, P.O. Box 8129, 6700 EV, Wageningen, Netherlands. Abbreviations: CPP, critical process parameter; CQA, critical quality at- tribute; DNA, deoxyribonucleic acid; EMA, European Medicines Agency; FDA, US Food and Drug Administration; ICH, International Confer- ence on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use; NIRS, near infrared spectroscopy; PAT, process analytical technology; PCA, principal component analysis; QbD, quality by design; RTR, real-time release implemented essentially unchanged in all modern manufactur- ing processes where product quality is of the essence. Surpris- ingly, Deming’s principles have only recently become a topic of interest in the highly regulated (bio)pharmaceutical industry. The performance of a process can be well described based on its “sigma level”. The sigma level of a process is defined by the relationship between the SD of average manufacturing perfor- mance and the upper and lower limits of product specification. Any product outside these specification limits is no longer fit for its use and is generally discarded or has to be reworked to in- crease its quality, resulting in economic loss. Therefore, in order to minimize the risk of economic loss, the chance that a product will fall outside the specification limits should be minimized. The sigma level is defined by the number of SDs of average normal manufacturing that fit between the upper and lower specification limits as is illustrated in Fig. 1. Another, and often better, way to describe process perfor- mance is the process capability index, ˆ C pk , see equation 1, where μ is the average manufacturing output and σ is the SD of man- ufacturing output. ˆ C pk = min  USL − ˆ μ 3ˆ σ , LSL − ˆ μ 3ˆ σ  (1) 212 Eng. Life Sci. 2013, 13, 212–223 C  2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim