Today, more than 30 products using a manufacturing process based on mammalian cell culture have been approved for human therapy. Most of these products are currently supplied with stirred tank bioreactors operated in batch or fed-batch mode. However, the bioprocess industry needs to constantly increase the productivity of its cell culture processes in order to supply more material with minimal investments in additional equipments. Therefore, alternative production techniques must also be considered, such as the perfusion culture of mammalian cells immobilized in packed-bed bioreactors (PBRs). PBRs can reach very high cell density and hence very high volumetric productivity; so their potential for bioprocess applications must be further evaluated. A review of current state-of-the-art in PBRs development (Chapter 2) showed that the latest generation of PBRs used for bioprocess applications have achieved very high cell densities (i.e. 107-108 cells per milliliter) leading to outstandingly high volumetric productivity. However, the major bottleneck for bioprocess PBRs is their relatively small volume due to the impossibility to avoid nutrient concentration gradients in PBRs of large volume. The current maximal volume seems to be in the range 10-30 liters, and more than 10-fold scale-up would still be required make the PBRs a competitive production technique. Beside their use for bioprocessing applications, PBRs have proven to be an excellent tool to fulfill the requirements of compact bioartificial organs in biomedical applications: they can reach the cell density and volume of an organ. However, as observed during the development of bioartificial livers, a decrease of metabolic activity is frequently observed after 1-2 weeks of culture. Therefore, the main challenge in this field is to develop cell lines that grow consistently to high cell density in vitro, and that maintain a stable phenotype for a minimum of 1-2 months to make them applicable to the PBR technology and to fulfill the clinical demand. PBRs are characterized by high cell density levels, thus by high metabolic rates, and accurate control of the cultures is required. Furthermore cell number cannot be determined directly in PBRs so an indirect method is needed. An indirect method based on the Glucose Consumption Rate (GCR) was developed to monitor a PBR process using recombinant Chinese Hamster Ovarian (CHO) cells cultured on Fibra-Cel® disk carriers (Chapter 3). A key step in this process was the switch from the cell growth phase to the production phase triggered by a reduction of the temperature. In this system, a GCR-based criterion was defined for the switch, and this control strategy proved to be robust, very simple, and was applied successfuly in routine operations for the monitoring and control of an industrial process at pilot-scale. The process operated with this GCR-based strategy yielded consistent, reproducible process performance across numerous bioreactor runs performed on multiple production sites. Another consequence of the high cell density reached in PBRs is the need for an intensive medium perfusion rate (feed and harvest) that should be used in order to keep the cells viable and productive. It appears that the perfusion rate is one of the central parameters of such a process: it drives the volumetric protein productivity, the protein product quality and has a very strong impact on the overall economics of the process. Therefore at industrial scale the optimal stationary packed-bed bioreactor process should operate with a perfusion rate as low as possible without compromising on quantity and quality of the product. In Chapter 4, the optimal medium perfusion rate to be used for the continuous culture of a recombinant CHO cell line in a packed-bed bioreactor made of Fibra-Cel® disk carriers was determined. A first-generation process had originally been designed with a high perfusion rate (2.6 vvd), in order to rapidly produce material for pre-clinical and early clinical trials. A reduction of the medium perfusion rate by –25% and –50% was investigated. With a –25% reduction of the perfusion rate, the volumetric productivity was maintained compared to the first-generation process, but a 30% loss of productivity was obtained when the medium perfusion rate was further reduced to –50% of its original level. The protein quality under reduced perfusion rate conditions was analysed for purity, N-glycan sialylation level, abundance of dimers or aggregates, and showed that the quality of the final drug substance was comparable to that obtained in reference conditions. Finally, a reduction of –25% medium perfusion was implemented at pilot scale in the second-generation process, which enabled to maintain the same productivity and the same quality of the molecule, while reducing costs of media, material and manpower of the production process. In Chapter 5, another limitation found in high-density PBR cultures was studied: oxygen supply. A model describing the oxygenation in a PBR of Fibra-Cel® disk carriers was developed. With the help of this model, it was possible to identify that the initial PBR system operated in sub-optimal conditions. Using the model two options were proposed, which could improve the performance of the initial system by about 2-fold: by increasing the density of immobilized cells per volume of carrier from 6.1·107 to 1.2·108 cell·mL-1, or by increasing the packed-bed height from 0.2 to 0.4 meters. Both strategies would be rather simple to test and implement in the packed-bed bioreactor system used for this study. As a result it would be possible to achieve a substantial improvement of about 2-fold higher productivity as compared with the basal conditions. Finally, this work compared the performance of two different CHO cell culture processes (PBR and fed-batch) used for the production of a therapeutically active recombinant glycoprotein at industrial pilot-scale (Chapter 6). The 1st-generation process was based on the Fibra-Cel® PBR technology. However, it appeared during the development of the candidate drug that high therapeutic doses were required (>100 milligrams per dose), and that future market demand would be in excess of 100 kilograms per year. This exceeded by far the production capacity of the 1st-generation process, and triggered a change of technology from a PBR process with limited scale-up capabilities to a fed-batch process with scale-up potential to several thousand litres. The volumetric productivity (in Product·m-3·year-1) reached with the fed-batch process was comparable to productivitiy of the PBR process. However, since the PBR system was limited in scale (0.6 m3 max.) compared to the volumes reached in suspension cultures (15 m3), the fed-batch was selected as the 2nd-generation process. In fact, the total product output per year (in Product·year-1) was about 18-fold higher for the fed-batch compared to the PBR. Data from perfusion and fed-batch harvest samples indicated that comparable product quality (relative abundance of monomers, dimers and aggregates; N-glycan sialylation level; isoforms distribution) was obtained in both processes. This illustrates the need to fix the production process early during the development of a new drug product in order to minimize process conversion efforts and to shorten product development timelines.