OR WAIT null SECS
Robert J. Thomas is a Research Councils UK academic fellow in biomanufacturing at Loughborough University, Loughborough, UK, LE11 3TU, tel. +44 1509 227601
David J. Williams is the professor of healthcare engineering and director of the Research School of Health and Life Sciences at Loughborough University, Loughborough, UK.
Large-scale manufacturing of human stem cells for therapeutic use is a leap in technology and science for the current biotechnology industry.
Regenerative medicine is a relatively new and fast growing industry with a product pipeline that offers a potential step change in the treatment of diverse chronic diseases. Regenerative medicine products are broadly defined as those that use a biological approach to restore, maintain, or improve tissue function. This covers a diverse set of therapies including cellular, biomolecular therapeutics, and tissue-scaffold products. In recent years, pharmaceutical developers have increased their interest in products incorporating a living cellular component because of their potential for dramatic clinical benefits in hard-to-treat and chronic conditions. As a result, approximately 80% of companies in the industry are focused on this class of therapy, with the majority in precommercial stages (1). Most products will use stem cells, or perhaps the differentiated progeny of stem cells, as their cellular component owing to their replicative potential and ability to form different tissue types.
MEDICAL RF/GETTY IMAGES
Background: Regenerative medicine and the need for large-scale stem-cell manufacture
Early "simple tissue" products for skin and cartilage, launched in the mid-1990s, were hampered by high production and distribution costs leading to a wave of insolvencies and consolidation in the industry (2). However, the re-emergence of more successful forms of these products combined with the recent approval of landmark clinical trials for more advanced indications (e.g., spinal cord injury and other neural indications) in both the United Kingdom and United States signal a maturing industry. In spite of this, the manufacturing science in terms of platforms, process control, and product analysis is still immature, necessitating an urgent drive for the development of scalable manufacturing solutions for cellular products. Although regenerative medicine is anticipated to be the major application for stem-cell-derived tissues, other applications such as pharmaceutical screening and disease modeling will also require consistent, high quality, and scalable production of stem cells.
Challenges: Why is large-scale stem-cell manufacture so difficult?
The requirement of the regenerative medicine industry to manufacture commercial therapeutic products incorporating stem cells (or their differentiated progeny) as an active ingredient is a technological and scientific jump beyond conventional biologics production. The key distinction is the use of the cell culture itself as the product in regenerative medicine as opposed to the use of a purified biological molecule derived from a cell culture. Furthermore, the human stem-cell types in question are sensitive to culture conditions and appear less predictable than the microbial and Chinese hamster ovary (CHO) cell cultures with which the biotechnology industry has developed extensive experience (3, 4).
The measurement challenge. The stem cell, as a product, falls under the stringent quality control requirements imposed on a therapeutic product by industry regulators. These include validated measurements of purity, potency, efficacy, and stability. Problematically, there is no current measurement system that can completely define a cell using either an individual or a set of assays; i.e., there exists no measurement tool for cells equivalent to mass spectrometry for small conventional drugs or biomolecules (5). Perhaps the closest "definitive" measurement would be microarray technology. However, although this technology provides a map of the cell's gene expression, it does not indicate cell state at a post-transcriptional level and would require combining with proteomics. As a population assay, it would also be a poor tool for identifying a single unsafe cell in a large therapeutic population. The inability to completely define individual cells or guarantee the homogeneity of a cell population has left researchers, product developers, manufacturers, and regulators relying on functional tests or surrogate indicators of function. Functional tests tend to be time consuming, qualitative, and often destructive. Surrogate functional indicators are hard to validate, particularly when mechanisms of action are incompletely understood.
Large-scale stem cell manufacture is therefore currently being approached with a poor knowledge of product specification. An inability to precisely measure a product hampers validation of the production system, process- or postprocess purification. It also reduces the manufacturer's flexibility to modify the production process after clinical trials if there is low confidence in the ability to detect a consequent change in the product.
The delivery model challenge.The stem-cell manufacturing issue is further confused by an unclear model for delivery of cellular therapies to the patient and poorly defined large-scale manufacturing requirements. There are currently numerous competing cell types for many therapeutic applications. Autologous therapies, in which the stem cell is sourced from the patient, manipulated, and then returned to the patient, may require a type of production facility different from allogeneic therapies, in which a stem cell from a donor is manipulated, expanded, and banked in order to treat a large population. The autologous therapy will require a facility that can handle multiple individual samples without cross-contamination, potentially on a walk-in schedule, and is therefore unlikely to be amenable to the conventional concept of batch manufacture. The allogeneic therapies will require bulk manufacture and cell banking, potentially achievable in a single campaign. The location of the manufacture will influence the specification of the manufacturing equipment. For example, the manufacture of some therapies may have to be partly or completely conducted at small scale on the hospital premises (e.g., to remove cryopreservant, to allow quality control, or for some noncommercial autologous applications). The scale of production is also confused by scarce availability of data from dosing studies. Stem-cell doses for many therapies are not yet defined even to an order of magnitude. It is suggested that cell number requirements will vary between 104 for very specific, small niche applications such as inner-ear hair cell regeneration, and up to 109 for cardiovascular therapies or 1012 for hematological therapies.
The scalable culture environment challenge. Different human stem cells require a plethora of different environments for proliferation and differentiation, potentially including mechanical stimuli or flow conditions, variable gas tensions, chemical gradients, 3D frameworks, and supporting-cell paracrine signaling. This raises an issue for the concept of a generic stem-cell manufacturing solution. These requirements are far more varied between stem-cell types and demanding for given cell types, than the biotechnology industry has had to negotiate to achieve scaled microbial or CHO growth. Although it is possible that a manufacturing solution could be programmable to control some of these criteria, it is unlikely (as with scale of production) to be a one-size-fits-all solution. Interesting questions remain surrounding the possibility of adapting cells to different, more large-scale amenable, culture conditions (similar to CHO adaptation to nonadherent growth) while retaining safety and therapeutic efficacy. Recent attempts to do this with human embryonic stem cell (hES) lines have raised controversy regarding the genetic stability of the cells in the new culture systems (6); it is not yet clear how far cell culture adaptation can be taken.
The state of the manufacturing industry:
Conventional bioprocess versus complex advanced models
Despite these challenges and lack of clarity, the rise of an industry based on the delivery of a viable stem-cell product has created an urgent requirement for scaled bioprocesses to produce stem cells. Addressing the challenges requires an interdisciplinary approach pulling expertise from the conventional bioprocess, manufacturing engineering, and cell-biology communities.
There are three key approaches to the current production of stem cells. Production on tissue culture plastic is commonly carried out in T-flasks or well plates and is the simplest system of growing stem cells in monolayers. It is the usual method employed in small research laboratories, and due to the low cost and ease of use, many therapies have been developed using this technology. However, the flasks suffer limited capacity and poor potential for on-line monitoring and control. Attempts to address the scale issue have been made through the introduction of multilayered triple or hyper flasks and more space efficient systems such as Nunc's (Rochester, NY) "Cell Factory."
Despite the control and monitoring limitations, these technologies seem to perform perfectly adequately for some stem-cell types, and as a result several therapies have reached or are approaching the clinic using T-flask-based cell manufacture. Beyond simple tissue culture plastic systems, there are numerous novel bioreactor approaches of varying degrees of complexity and success. These are often custom-made multifiber or biomimetic material systems that aim to provide a highly controlled niche environment for stem-cell expansion or differentiation.
For many systems, there are question marks over true scalability and potential for CGMP compatibility because of their complexity. Furthermore, the level of control and monitoring for successful scaled production is not established, and these systems will be relatively expensive if simpler systems prove adequate. There may ultimately be an economic trade off between suboptimal production conditions and simpler or more generic production facilities. Finally, there is a surge in interest in applying conventional bioprocess stirred-tank reactors to the scale up of stem-cell production. These systems offer the advantages of decades of development in process control and knowledge of scale-up, much of which would carry over to stem-cell applications. Unfortunately, they are most suitable for nonadherent cell types and most stem cells are adherent cells. Attempts to 'adapt' cells to nonadherent culture using cytokines or cell adhesion molecule ligands, or to use microcarriers to achieve adherent growth in tanks, have had mixed success and are at an early stage.
There may be room for all of these approaches. Stirred tanks would be very good for large batch production of allogeneic therapies where T-flasks might be more appropriate for autologous production of patient-specific therapies. Complex bioreactors may find a role for niche stem-cell types that require highly specific environments, or as closed minimal intervention systems for stem cell manipulation within a clinical environment.
The underpinning manufacturing science:
Automation and process improvement
Although the stem-cell manufacturing requirements are not fully defined, and final solutions are still to be resolved, some spot solutions for the large-scale manufacturing of stem-cell based therapies have begun to emerge. It is clear that, from an economic and regulatory process-control perspective, manual processing of cellular therapies cannot be a long-term solution to cell-therapy production. To address this, commercial automated scalable systems have been marketed, including the Wave Bioreactor Systems technology from GE Healthcare (Piscataway, NJ) and the robotic T-flask handling platforms from The Automation Partnership (Hertfordshire, UK). These systems aim to create an automated robotic parallel to many of the current manual processes. They offer the opportunity to increase process reproducibility by removing operator variation, although they do not move the processes forward in terms of on-line monitoring and advanced environments. Instead, these systems were developed based on the reality that many products are moving toward the clinic using "simple" culture systems, and that these processes will be difficult to change significantly after clinical trials, given the problems of product measurement and importance of process validation. Such "simple" automated solutions will be essential to facilitate the movement of the industry to the next phase of mass market products. However, despite their similarity to manual handling, they face challenges with regard to the sensitivity of the cells to those process changes imposed through automation. They have also faced the skepticism of scientists steeped in the ethos that continual experience- based intervention, as opposed to rigorous process control, is the route to a consistent quality product.
In the Loughborough University Healthcare Engineering group, we have worked with academic and commercial stem-cell therapy researchers to demonstrate the successful and reproducible use of one of these automated cell-culture platforms, The Automation Partnership's CompacT SelecT (see Figure 1), to expand a range of clinically important human stem cell types at large scale. The CompacT SelecT automated cell culture platform can simultaneously manipulate 2 x T175 flasks and house 90 x T175 culture flasks in a robot-accessible incubator. All culture processes are carried out within a sterile class II environment and require no manual intervention. The CompacT SelecT closely mimics the normal manual cell-culture operation with the key exception of a centrifugation step.
Figure 1: The CompacT SelecT automated cell-culture platform can simultaneously manipulate 2 x T175 flasks and house 90 x T175 culture flasks in a robot-accessible incubatabor. All culture processes are carried out within a sterile class II environment and require no manual intervention. The inset shows an enlarged image of the manipulation chamber and pipette head. Major processing components are labeled: A, Robot arm; B, Flask incubator; C, Plate incubator; D, Flask decappers; E, Flask holders; F, Media pumps; G, Pipette head; H, Cedex automated cell counter.
Our objective was to automate commercial near-clinic and publicly available platform stem-cell types to have maximum broad impact on potential regenerative medicine applications. In the initial stage, human Mesenchymal stem cell (hMSC) expansion was automated (7). hMSCs are an adult stem cell, usually derived from bone marrow or umbilical cord blood, that are of clinical interest for mesenchymal tissue repair, including bone, cartilage, and muscle as well as a supporting cell in other tissue-repair scenarios such as post stroke. These target conditions represent a large clinical market. Bone-marrow-derived primary hMSCs that underwent automated expansion replicated at the equivalent rate to cells processed manually and maintained hMSC typical surface marker expression. More advanced automated stem-cell scale-up successes include hES cells and human neural stem cells. hES cells are important because they are pluripotent and therefore have the potential to be a complete platform cell type for any regenerative tissue therapy. They are notoriously difficult to culture. They grow in colonies and show both phenotypic and genotypic sensitivity to culture conditions. In collaboration with colleagues at Nottingham University we demonstrated that it is possible to maintain key cell attributes, including genetic stability, pluripotency markers, and differentiation potential over ten automated passages for two separate cell lines using a novel monolayer culture technique (8). The human neural stem-cell culture was conducted in collaboration with ReNeuron (Surrey, UK). This stem-cell line is entering clinical trials this year for the treatment of stroke, so the established culture conditions are designed to be compatible with CGMP and other regulatory requirements. Neural stem cells expanded in the automated system passed the cell-bank quality control tests including expansion rate, gene expression equivalence to master cell bank, and Nestin expression (9). Data from this work are shown in Figure 2.
Figure 2: Sample data from the successfully adapted automated bioprocessing of human embryonic stem cells (hES) and neural stem cells:A) the growth rate and normal morphology of automation-processed neural stem cells, B) Nestin expression of the neural stem cells C) growth rate and morphology of automation-processed hES cells, and D) normal expression of pluripotency markers by the hES cells.
Automated cell expansion is not only a route to large-scale stem cell production. It provides a platform for the application of manufacturing and process-improvement tools essential to a controlled manufacturing process. We have employed design of experiments (DOE) within a systematic process improvement methodology to improve the efficiency and quality of stem cells from the automated platform (see Figure 3). This work has identified the sensitivity of cells to cell culture factor interactions and reinforced the importance of a systematic approach to process characterization (10).
Figure 3: The concepts of systematic process improvement and quality tools need to be applied to regenerative medicine manufacturing. A typical systematic process improvement cycle as applied to automated stem-cell processes is shown alongside an interaction chart from a designed experiment. The chart shows that a change in cell density significantly affects the impact of serum concentration on the rate of large-scale automated expansion of human mesenchymal stem cells. This type of complex interaction requires the application of sophisticated process optimization techniques. FCS is fetal calf serum.
Robert J. Thomas* is a Research Councils UK academic fellow in biomanufacturing, and David J. Williams is the professor of healthcare engineering and director of the Research School of Health and Life Sciences, both at Loughborough University, Loughborough, UK, LE11 3TU, tel. +44 1509 227601, R.J.Thomas@lboro.ac.uk
*To whom all correspondence should be addressed.
What would you do differently? Submit your comments about this paper in the space below.
1. M.J. Lysaught, A. Jaklenec, and E. Deweerd, "Great Expectations: Private Sector Activity in Tissue Engineering, Regenerative Medicine, and Stem Cell Therapeutics," Tissue Engin., 14 (Part A), 305–315 (2008).
2. R. Archer and D.J. Williams, "Why Tissue Engineering Needs Process Engineering," Nature Biotechnol.,23 (11), 1353–1355 (2005).
3. C. Allegrucci, Y.Z. Wu, and A. Thurston, "Restriction Landmark Genome Scanning Identifies Culture-Induced DNA Methylation Instability in the Human Embryonic Stem Cell Epigenome," Human Molec. Gen.16 (10), 1253–1268 (2007b).
4. D. Brafman et al., "Defining Long-Term Maintenance Conditions of Human Embryonic Stem Cells with Arrayed Cellular Microenvironment Technology," Stem Cells Develop., electronic publication ahead of print (2009).
5. D. Van Hoof, A.J. Heck, J. Krijgsveld, and C.L. Mummery, "Proteomics and Human Embryonic Stem Cells," Stem Cell Res.,1 (3), 169–182 (2008).
6. M.S. Bodnar, J.J. Meneses, R.T. Rodriguez, and M.T. Firpo, Propagation and Maintenance of Undifferentiated Human Embryonic Stem Cells," Stem Cells Develop.,13 (3), 243–253 (2004).
7. R. J. Thomas et al., "Manufacture of a Human Mesenchymal Stem Cell Population Using an Automated Cell Culture Platform," Cytotechnol., 55 (1), 31–39 (2007).
8. R.J. Thomas et al., "Automated, Scalable Culture of Human Embryonic Stem Cells in Feeder-Free Conditions," Biotechnol. Bioengin., 102 (6), 1636–44 (2009).
9. R.J. Thomas et al., "Automated, Serum-Free Production of CTX0E03: A Therapeutic Clinical Grade Human Neural Stem Cell Line," Biotechnolo. Letters, electronic publication ahead of print (2009).
10. R.J. Thomas, P.C. Hourd, and D.J. Williams, "Application of Process Quality Engineering Techniques to Improve the Understanding of the in vitro Processing of Stem Cells for Therapeutic Use," Jrnl. of Biotechnol., 136 (3–4), 148–155 (2008).