Putting All the Pieces Together

Most people in the industry are struggling with quality by design and how it relates to the acceleration of process development. Many are confused by the new FDA approach to bioprocess development, unsure of the specific implications of QbD on the CMC section of their marketing applications, and unclear how the risk-based approach applies to their particular operations. Some have trouble understanding the precise link between CQA and CPPs under a life-cycle approach and are stuck considering the exact definitions of such terms as critical and variable. But help is coming from many fronts. FDA reviewers are being specifically trained to expect and assist in the incorporation of QbD principles in regulatory filings. Results from the QbD pilot program are emerging as a valuable resource for guidance. Industry leaders are publishing and providing case studies about their experiences, as seen in this special issue.

Independent associations such as the ISPE, IEEE, and IFPAC are providing tools for comprehensive understanding and practical applications. Other organizations such as BIO and IBC Life Sciences are presenting at conferences the newest developments and solutions in the implementation of operational excellence in the biopharmaceutical industry. Service and equipment vendors such as Thermo Fisher Scientific, Tecan, Seahorse Biosciences, and Stat-Ease are providing direction and examples of how to implement new monitoring, analytics, and control solutions. And publications such as BPI are presenting specific direction from industry leaders on the implementation of the FDA’s 21st-century initiatives.

BAYER HEALTHCARE (WWW.BAYER.COM)

With a Grain of Salt: Where’s all this taking us? William Whitford is a frequent BPI author and bioprocessing market manager at Thermo Fisher Scientific. “Speed and economy are industry goals,” he cautioned, “but really, isn’t the FDA’s overriding goal patient safety? Can OpEx, QbD, PAT, and so on actually save money in a specific process development or operation? It may pay for itself in the long run, but up-front it’s an investment.” He urges readers to consider that by the most credible definitions, the word risk in the “risk-based approach” refers to product safety risk alone — and that the words faster, leaner, and cheaper are not in the PAT guidance, itself. The most successful companies will bear that in mind — and not lose sight of the primary goal of safety while seeking ways to save money and time.

Quality consultant Laureen Little elaborated on the subject. “How do we know that a product that is different within its design space, with a certain amount of heterogeneity, is going to be safe in application?” she asked. “It hasn’t gone to the clinic yet, and the bottom line is the same: You have to have clinical data. Products get approved based on how they work in the clinic. You can have an absolutely gorgeous product design space that allows only 5% variability of your product. Say there are two major product variants, with 90% form A and 5% form B, 2.5% form C and some other minor product-related variants. Unless you put all those variations into the clinic, you can’t get it approved. If there is a ‘sweet spot’ before phase 3, then you have to have a design space between phase 2 and 3, where you scale up. If you’re trying to put the variability of your clinical lot into pre-phase 3, but you don’t have a good scale-up model, and the design space shifts — which does happen — then what do you do? You don’t want to use your pivotal trial to support your design space; I think you may end up doing smaller phase 2–like studies, but I’m not sure that’s a good answer.

“Now, this comes from my background as a bioassay person: The more biomimetics you have for an analytical method, the better will be your understanding of the therapeutic mechanism and structure–function relationships from a potency/efficacy perspective. But you also need to understand it from a safety perspective. Does a given molecular structure elicit more immunogenic response, or not? Go back to efficacy: If you have something that’s extremely biomimetic, and you have a good structure–function relationship, then you have a much better, scientifically sound basis to say, ‘We know our mechanism of action, we’ve probed it, and we’ve seen no difference.’ And that’s what people love: going from descriptive to being more prescriptive.”

So it comes back to the need to know as much about your molecule as early as possible. “And the key is the critical quality attribute,” Little pointed out. “It’s interesting that we use that because the structure–function relationship has been a topic of even academic conferences for decades, and that’s what we’re trying to do now. We’re trying to say, ‘If I have this structure, then I’ll have that function.’ As you move into a QC environment, that’s when it becomes a quality-determining attribute. You know what a CQA is because you link structure to function and back to your mechanism of action — and that’s hard work. DoE may be the greatest thing since sliced bread, but it gives you a lot of information that’s all about your process and how to vary it. All you’re doing is attuning your hammer to what you want to do; it doesn’t actually swing the hammer at the nail.”

In another interview, Jens Vogel talked about his company’s current approach to the question. “QbD is of course very close to our hearts here at Bayer. For a couple years now, we’ve looked very closely into this and how we can optimize process understanding early on. But there’s always going to be a trade-off because of the speed-to-market issue. In the back-loaded model, you want to go into the clinic for a proof of concept before you invest significant resources into the actual design space. Yes, QbD concepts are used from early on, but certainly an IND is not the time to file a design space. That’s what you do for the marketing authorization. Although we are extensively applying QbD approaches, the biggest benefit is in process robustness at the time of licensure. For biologics, we still have to understand that design space may be limited: There’s only so much you can actually learn from characterization because of assay variability, for example, and just the complexity of the molecules. The industry is moving in this direction, and we have certainly made progress here at Bayer in that regard. But I think we are still some time away from what could be considered a true design-space biologic license application.”

Communication

Industry consultant Sally Seaver points out that for biologics, at least, “the FDA is still gaining experience” when it comes to QbD. As the traditional first-adopter of ICH guidances such as Q8, Q9, and Q10, the EMEA may be somewhat ahead — but in general, to become proficient in QbD/QRM, the biopharmaceutical industry and regulators need to work together. The international separation science society CASSS, among others, has facilitated this process through its Well-Characterized Biotechnology Products conference series and the associated CMC forums — which have been reported on extensively in our pages. To get where the biopharmaceutical industry needs to go, working together is more important than ever. This is true from the highest levels right down to project teams in individual companies.

Canping Jiang of Bristol-Myers Squibb explained it this way: “Under QbD approaches, product quality is ensured by well-developed, well-characterized, and well-controlled processes. A cross-functional team is essential for successful development and scale-up implementation of manufacturing processes. At BMS, cross-function project teams that include representatives from process development, analytical development, technology transfer, validation, manufacturing, supply chain, and quality are formed for process development and commercialization.”

CMC consultant Nadine Ritter wrote to us, “In my experience with numerous sponsors — large and small, early and late development, US and international — hands-down, the most effective and efficient product teams are cross-functional both in their technical specialties (e.g., process, analytical, and stability) and across functions (e.g., technical, quality, regulatory, and operations).” But all too often, in her past work in ‘big pharma’ and contract testing and now as a consultant, she’s observed product development activities conducted “in a vacuum.” Many people skip routine communications among relevant groups because they think it’s a waste of time. They say, “We know what we’re doing,” or “We don’t need to involve them yet.”

In too many cases, Ritter explained, “valuable communication — with all sides contributing their perspectives as well as seriously listening to their counterparts — would have spotted a minor derailment long before it occurred, thus preventing a major train wreck that would later cost considerably more time and money to remediate. Sponsors manage innumerable trade-offs from risk assessments of all aspects of product developments. But those decisions should be based on accurate, informed risks derived from all related functional areas. Because our field moves forward so fast, what might have been an acceptable risk just three or five years ago may be unacceptable now. Making decisions in a vacuum, people can miss major emerging concerns that affect current thinking on process/product development.”

“Working with the end in mind seems best to increase speed,” Amos M. Tsai of Amgen confirmed in another interview with BPI consulting editor Lorna McLeod. “Process development is about what eventually happens in either the clinical or commercial manufacturing plant. The more we can anticipate how a process will manifest itself once it is transferred, the higher the quality of the final product.” And quality equals (or should equal) safety as well as efficacy. “So communication and collaboration between those who scale up processes should take place before process development is initiated at the bench. Representatives from clinical and commercial manufacturing should be part of the team to identify equipment capability, manufacturing practices, and plant capacity to handle utilities, raw materials, and waste.” Inevitably, different scenarios will emerge over the life cycle of process development and scale-up because of manufacturing constraints and development findings. “The ability to model those scenarios helps us understand their impact on product quality, potential capital investment, and operation cost long before the actual tech transfer takes place. The emphasis on speed has motivated us to develop a thorough understanding of the interdependence of functional areas, which can be formally documented and used as a planning tool.”

There has always been a need to bring product as quickly to the market as possible, and platforms came along as a response to that. Because of the need for speed, the need for more effective communication between different groups to ensure smooth project transition has increased. Because a platform process has common methods of operations across various projects, implementation of platform technology has ensured that different groups can communicate easily while delivering on project targets. Lorna spoke about teamwork with Ravishankar Vadali and Delphi Krishna at GlaxoSmithKline.

“Whenever a new process needs to be developed,” Vadali added, “the platform in place can be implemented. If that delivers project targets, then it can be transferred immediately to manufacturing. This approach ensures that the product is brought to the market within shorter timelines. The development and implementation of this platform approach is a response to the demands from project teams and manufacturing.”

In the modern environment, the classic “silo” mentality of so many larger pharmaceutical and biopharmaceutical companies will have to go by the wayside. Current circumstances may be converging to bring down some of those walls and make people talk to each other more than before. “In fact,” Krishna told Lorna, “even within GSK, we are evolving into a small biotech organization. The need to effectively communicate, have more fun in an entrepreneurial environment like a small company, is encouraged — instead of just staying within your own group without talking to others. GSK has a big QbD initiative that’s dissolving the boundaries.” She said that a process developer’s job isn’t finished once the process is delivered. “You talk to groups before and after you because you need to see if you have satisfied their criteria. QbD definitely helps dissolve those ‘silo’ boundaries.”

Harald Dinter and Jens Vogel of Bayer elaborated on these concepts in relation to facilities and engineering. “In the old days,” Dinter said, “it was develop your molecule or your process, then call in the engineers and say, ‘This is how to produce the molecule, now design the manufacturing facility. Here are all the process specifications, off you go.’ Clearly that does not work anymore.” Today, those engineers need to set up a facility when a manufacturing process is not yet fully defined. “Now we need to pull in the designers very early on,” Dinter explained. “The project team set-up has changed. While the blueprint matures, the development process matures too. With the back-loaded approach, process optimization and understanding are pushed into much later phases than they were five or 10 years ago.” And the team that works on this together is a different team than it would have been 10 years ago as well. “For example, in Jens’ project team [the CMC development team for a new protein drug] is someone from the production organization.”

Vogel added, “I think we’ll see more of that integration and basically a trend toward new approaches to assess the manufacturability of candidates in early selection. In the past, there’s been no emphasis on that R&D interface, and now I think more companies are looking at how it can be integrated into the decision-making process.”

Dinter confirmed that. “We had internal restructuring to put the R&D parts together and really minimize or prevent any ‘throwing over the walls.’ In biotech, you have a lot of small companies mainly started by researchers. They come up with a molecule, but they don’t have the CMC development capabilities.” So they outsource such work to contract manufacturers. Big organizations such as Bayer operate differently, Dinter explained. “In early research, we need to provide some guidance. A molecule may be nice from the research side, but it can’t be produced, or the cost of goods would be too high. So maybe we need to modify that molecule to manufacture it with a lower cost of goods — which ultimately should also result in lower drug prices and increased competitiveness.”

Vogel added a technical perspective. “It’s much more than just talking to each other; it’s really coming up with technological concepts such as high-throughput and robotic approaches. Maybe you develop an antibody against a specific target, and the research group is selecting among 20–30 different candidates, all with relatively similar binding affinities for the target. So one idea is to subject those antibodies to a series of experiments using your robot screening platform. What you may see is that a few aggregate strongly at the pH you typically use for virus inactivation. So rather than picking those candidates on the research side and pushing them into development, and then having to figure out how to change their processing platform to adapt to a difficult-to-make antibody, you could exclude a subset of molecules and pick one that should be easy and straightforward (and fast) to develop and bring to market.”

Transformation

Clearly, technology is what’s making the difference. But what techniques out there are really new? And what will be the revolutionary, so-called “disruptive” technologies of the near future? Certainly automation is a major transformer of processes and projects — and promises to be even more so as time goes by. From microscale cell line screening assays to chromatographic optimization experiments to automated differential scanning calorimetry used in formulation studies, robots have found their way from discovery laboratories and into process development. Computer modeling, too, is helping companies characterize the “hot spots” on protein molecules — sections that may be vulnerable to deamidation, oxidation, hydrolysis, and other problems that can lead to aggregation — through “in silico” experiments as well.

“People are adding mass spectrometry to even more analytical techniques,” consultant Sally Seaver told us. “You can very quickly figure out what’s coming off columns, for example. High-content, high-throughput analytics are starting to come over a bit into other types of assays, especially for characterization. I also think it’s going to be invaluable for analyzing clinical samples because you are looking at multiple biomarkers in each sample. That can very effective.”

But, she cautions, further extrapolation into process analytical technologies (PATs) will be a challenge. “Part of the problem with taking it into anything more than process development is making sure the equipment and software are GMP Part 11 compliant. Even with PCR, many people who manufacture the machines have no idea what a QC lab requires. People use PCR because it’s so important for rapid detection of potential contaminants, such as viruses. But the manufacturers need to understand that they can’t just arbitrarily replace obsolete instruments and software and update software. (HPLC systems went through the same thing years ago.) As for the reagents you buy, well, you’d better qualify them in your own internal quality control.”

Single-Use Technology: When asked about other transformative technologies, Seaver said, “I think disposable equipment has really been useful. If you looked at some of the original cell culture, disposables were being used. Then they lost popularity. Later, people began looking at disposables again when they began to realize that cleaning was a lot more complicated than they had anticipated.”

Another trend she pointed out is occurring at Genentech and Amgen already, and should be seen at Genzyme soon. “To keep adventitious agents from occasionally being introduced (with absolutely disastrous results), people are looking at ways of sterilizing media in place using short-term high temperatures or UV treatment. That’s something to watch. It’s been happening in some places for at least five years, but I don’t remember when I first heard it discussed. To many people, it’s brand new. At the PDA cell substrate compounds meeting, people were asking, ‘Why are we waiting for our fermentors to crash and our facilities to be contaminated to put in real-time media treatments?’ Real-time PCR can be used to test for certain viruses, such as minute mouse virus (MVM). I think people are trying to be proactive especially for contaminations.”

Bayer’s Harald Dinter elaborated further on single-use technology. “We have our own experience in the antibody area. We compared products made in steel systems with those made in disposables. With the analytical tools we have, we don’t see a difference, which gives us a pretty good confidence that disposable is the way to go. Obviously we’re still limited in volume, if we talk about fermentors.

“At the same, antibody titers are increasing. A few years ago we were around 1–3 g/L; now we’re talking 5–10 g/L. Someone reported recently even 13–15 g/L. So the need for volume could drop from 10,000-L tanks down to 3,000 L, and you can handle that with disposables. If you have smaller products for limited markets, then you want to have a flexible facility where you can quickly switch from product A to product B. Do you want to automate everything? Most likely not. You want to have some people in there who can quickly turn that facility around. So automation is possible, but it depends on meaningful capital investment.”

Günter Jagschies of GE Healthcare described those production advances as a transformational technology in themselves. “Even in the future,” he said, “there will still be some very large-scale biopharmaceuticals, such as insulins and a few MAbs, so some large facilities will be needed to run that type of production. For MAbs, it looks now that most of these facilities already exist or are on the way to being commissioned. In fact, a case study I performed indicates that the 2008 market demand for all approved MAbs could have been satisfied from one fully used 90,000-L facility and another one partly used, assuming 3–5 g/L titers in all processes (which isn’t necessarily the case). For 2010, estimates assume more than 3,000,000 L of available capacity, making annual production capacity 100 tons at 2 g/L already. So this industry has very significant overcapacity.

“Legacy products with less up-to-date processes will use some of the excess capacity, but the situation is still very obvious. It is the legacy products that are most likely to be produced with huge bioreactors (10,000–20,000 L), even in the future. However, most people doubt that a new MAb produced with a modern process at titers >3 g/L could alone use up the capacity of a large plant. It would need to have a demand of 5–10 tons/year, and that’s not in sight today.” So multiproduct facilities are likely to be the norm, with flexibility the key to making them work. “Cross-contamination is and always will be a concern in multiproduct facilities, precautions to prevent it will need to be in place. One way to reduce that risk is to use disposable components. This opportunity gets more real when production scales are reduced, and we look at bioreactors no larger than 1,000–2,000 L.”

Jagschies pointed out that disposables are only one tool for flexible manufacturing. Economic considerations are driving their adoption, but results from real case studies are hard to come by. The use of simple disposable devices such as bags for buffers, intermediate storage, and reactors in seed-train cell culture usually offers economical and operational advantages. But for more complex and costly disposable devices, it’s not always so obvious and will depend, for example, on the number of batches to be run (more batches increases consumables costs). “Another variable is the time to readiness,” said Jagschies. “If the time to start of production is critical, disposable solutions may help prevent delays of clinical trials and lead to earlier revenues. In such cases, higher costs might be far outweighed by prevention of lost or delayed revenues.” Another economical advantage can come when a company eschews fixed installations and gains flexibility through use of disposables. It reduces change-over times between production campaigns, and it can facilitate introducing an entirely new process to a facility without much advance warning or planning. “Indirectly and after some time,” Jagschies said, “this may yield financial benefits.”

Separation and Purification: Marc Bisschops of Tarpon Biosystems brought up another technology. “Some of the more advanced membrane absorbers could be very powerful. The other thing I like very much are new media that allow higher liquid velocities, such as Poros perfusion chromatography media, membrane absorbers, and monoliths. In certain applications, those will show significant improvements over agarose and other media.

“What I’m also trying to keep track of is developments in which people are moving away from three columns and trying to get the same performance from two. I think that in many cases there’s no strict need to develop a robust process in three columns. Purifying monoclonal antibodies, which are well understood, I’m pretty sure that often can be done in two columns. And there are advances related to disposables and anything that makes processes smaller, reduces tank farms and buffer consumption, and so on.

“I think what is needed in process automation is compatibility and user friendliness. I’m no expert in this, but it seems that every supplier has its own platform while people are moving toward common platforms such as protein A, which is still extraordinarily expensive in my opinion. There are a few platforms out there, and every company is using its own. You bring this kit into a large company, and there’s a huge IT department that’s going to overrule your software and develop its own. To me this doesn’t sound like a very sensible approach, and most traditional process industries normally don’t do this. Normally they have all their control systems very well compatible with each other. And people accept new technologies pretty easily as long as it can be implemented on the facility platform in the overall manufacturing process software. That’s something we should learn from even though we always say that we are such a regulated industry with so many hurdles, and so on. I think we have to grow up a little bit there too.”

Prognostication

Ten years ago, biotechnology companies were generally considered small and nimble as a group. But the largest biopharmaceutical companies are not so small and nimble these days. The major emphasis back then was on rapid scaling up of processes rather than flexibility, which is becoming much more important in the modern business climate.

Sally Seaver explained. “People were trying to take even uncloned transfected cell lines or microbes and just quickly make some material for a little bit of process development, a little bit of in vitro toxicology testing. I can’t say that isn’t happening now, but sometimes it worked, and sometimes you fell flat on your face. So I think what’s replacing that idea of just, ‘Let’s go fast and cut corners,’ is this idea of platform technologies. There’s this game of trying to change phase 2 into pivotal trials, depending on your indication.

“I’m concerned about how much you can shorten process development. If all you’re doing is making monoclonal antibodies, then yes, you can have platform technologies. But you have to be careful with the first few batches for a new MAb to evaluate whether its production parameters are going to be within the robustness of the platform. If you’re doing different enzymes for replacement therapy, it is not obvious to me how applicable platform technology will be. You may still have some goals around contaminating DNA or host-cell protein, but your process might be radically different.”

Regulators are encouraging biotech companies to use these technologies. At the meetings we’ve attended, we’ve heard FDA staff say, “Yes, we want you to try these new things, but we expect it to take longer to get approvals at first. We’re going to have to look at these very closely because we don’t fully understand it yet, ourselves.”

Hard work and profound knowledge are the only ways to get there. Once regulatory people understand what companies are trying to do — and once the industry has implemented QbD and is moving along in that direction — can we expect to get faster, safer, and cheaper processes and products? Is that a reasonable outcome to ask for?

Consultant Stanley Deming of Statistical Designs spoke about these questions with Lorna McLeod. “The way I heard it was ‘faster, cheaper, better.’ Statisticians and people who work in the industry know that you’re not going to get all three simultaneously. You can get two, but not the third. NASA had a ‘better, faster, cheaper’ program for their Mars missions in the 1990s, and they ended up wasting a lot of money. Anybody could have told them it wouldn’t work.

“Maybe regulatory agencies really do think you can make it safer, cheaper, and faster — but they’ve got to be a little careful because the faster part is where things break down. If you do use things like experimental design, then you can do some things faster. They’re efficient, they’re effective, they’re very productive techniques. Those are good tools to use as long as you know what your goals are and what you need to achieve. I would guess that probably over the next 5–10 years this will pay off, and people will then be doing it routinely like in other industries such as electronics and small chemicals. But it’s going to take a bit of a learning curve for people to become comfortable with it and just begin to do it as standard operating procedure. And it will take a lot of cooperation between the regulatory agencies and the pharmaceutical companies.”

How Do the Little Guys Compete?“Smaller companies, of course, are a lot more nimble,” said quality consultant Laureen Little. “Once you get really good at DoE, if you can look at scale-down models, it will help you design things faster. But I think smaller companies tend to be made up mostly of research groups who are developing structure–function relationships. So I would think that’s where small companies might be able to move things more quickly from candidate to phase 1 and develop rational designs for these products faster.”

However, said Sally Seaver, smaller companies need “to understand exactly what their strengths are and not try to mimic large companies. Each company needs to know its own strengths and what niche it can fill the best.”

Marc Bisschops told us he thinks disposables will be a key to success for smaller companies in the near future and beyond. “Many of the giants have these large six-pack facilities and designs around ~1 g/L production, so they face significant bottlenecks in their manufacturing processes.” Echoing Günter Jagschies, he said bigger biotech companies are likely to have a hard time using up all their capacity and getting a reasonable cost of goods in the process. “But if you’re small, flexible, and have your own disposable manufacturing facility, you can truly produce at much lower cost than what people are doing right now at Amgen and Biogen and Genentech. I think that’s something that will enable smaller companies to become competitive.”

Bisschops noted that biosimilar products cannot be ignored as a market and “will give some smaller companies a foothold in the marketplace, even though everybody realizes that the follow-on biologic pathway is not trivial. It’s not something that can be done as cheaply as a lot of people hoped, and the cost savings are generally not as big as some expected. It will take a few more decades before that is truly developed, but the smaller companies can do that as well.”

Delfi Krishna told us that indeed, “biosimilars are here. Patents are expiring, and price pressures are across the board, all across the industry. But I think innovation is going to be key. We cannot compromise on that. Just because biosimilars are there, and they are available at one-third the price, somewhere someone has to pay the price of innovation. So the message needs to be communicated to peers and patients that for different kinds of diseases and cancers, time and money will have to go into coming up with new and better therapies, and someone somewhere has to pay for it. It’s not low-hanging fruit.”

Bayer’s Harald Dinter believes partnering is key for both the large and small companies. “It depends on attracting the right partner: the one that fits, with the right mind-set, and most important, one with whom you can establish a very productive interaction.”

Flexibility for the Future: Dinter’s colleague at Bayer, Jens Vogel, emphasized flexibility for all. “Hard-type stainless steel and automated equipment is not very flexible for making process changes. If you require some additional buffers, some additional solutions, you’ll have a hard time trying to fit this into existing facilities. Or you can try to build them with maximum flexibility, but very quickly you’ll find that capital costs can spiral out of control. So there’s still just a practical trade-off. It’s not so much about technological limitation as it is about capital investment, up-front investment.”

“Operational excellence and automation tools are going to contribute to flexible manufacturing in a big way,” said GE Healthcare’s Günter Jagschies. Operational excellence is all about knowing a manufacturing operation very well, constantly improving it, and reducing associated error rates or accidents. It is also related to optimizing supply chains and reduction of costs in all aspects of manufacturing. “Some solutions that are good for classic installations may not be the best in disposable operations,” said Jagschies. “Disposable sensors and different ways of operating and automating a plant may be necessary amendments to currently available product options.”

He pointed out the “plug-and-play” feature as particularly helpful. “Exchange of detectors, addition or removal of a processing device, and many more examples in a flexible manufacturing environment can be facilitated with automation and recognition, verification, qualification, and documentation of changes through an automated system. Such things cause major headaches in a classic hard-piped facility, especially those of the first generation built in the 1980s and early 1990s. Things started to change/improve in more recent facilities.”

On a related subject, we wondered whether the industry has reached its limit in antibody titers. Jagschies believes that, based on detailed case studies of very large-scale manufacturing, titers >3–5 g/L won’t offer much additional savings, especially when higher product quantities won’t necessarily be in demand. “Higher titers have been reported,” he said, “and I believe they are being used in some clinical trial manufacturing already. But I am not alone saying that the higher productivity from such processes will often cost through increased problems further downstream as existing downstream installations need expensive upgrading, purification and filtration get more difficult with more complex impurity patterns to solve, and many other reasons.” As long as companies can make products for the market with good margins, there’s not much reason for managers to push (or even allow) higher-titer developments that could cost time and valuable R&D resources.

“Other priorities would benefit from the focus of those resources,” Jagschies suggested. “There seems to be some agreement out there: For examples, see some recent publications by Brian Kelley, head of process development at Genentech. If titers increase anyway, we will need centrifuges that can handle the higher biomass without compromising the product quality or yield; either much larger-area and more expensive filters or a whole new generation of them; chromatography resins with higher capacity but equal or better resolution (likely that purification problems are more tricky at very high titers); a new generation of virus filters; and a few more things as well. I wonder whether higher titers, with no strong product demand increases, could ever justify all that and related costs across the industry.” Some of those technology improvements he mentioned are needed or are happening anyway, though. “Our R&D is addressing many downstream challenges as we speak,” he admitted.

Some people have touted a future of personalized medicine as further reason that extremely high production titers and massive product quantities won’t be necessary. But Jagschies says, “There are some misunderstandings on personal healthcare. It will be enabled through better understanding of disease and each individual patient’s particular condition in relation to that and any potential treatment.” He says biomarkers are one likely key tool in this development. “In a broader sense, diagnosis will become key — not just to knowing what disease a patient has, but also to predict how each patient will respond to a given treatment.” Although patient populations might shrink for drugs that are involved with such diagnostic tools (because only those patients that respond will get the treatment), some drugs may well become more popular than their competitors if they’re perceived to offer better chances for successful treatment and fewer side effects. Such products would then take higher shares of the market, producing more business for their sponsor companies.

“Over the long term,” Jagschies said, “this combination of therapy and diagnostic tools should increase the overall share of biotherapeutics and widen the market access for personalized medicine. But the one caveat remains the cost of treatment: A partially individualized treatment cannot become so expensive that its market is limited by cost. So this will be most effective for therapies with patient populations that are very large to begin with.” Genentech’s Herceptin antibody is thus a good example.

We talked with Johannes R. Roebers of Elan Pharma International Ltd. about future trends in operations and facilities. Is the day of the blockbuster drug behind us?

“No, I certainly think that biologics can still grow and will,” he said. “It will be more difficult for new biologics to become blockbuster drugs, but there will be a few new ones. Blockbuster biologics will be those that really are new therapies or address new targets. There will be fewer of those than the past, but there will be some.”

We wondered about the nature of those products. Will classic protein molecules continue to be the most common biopharmaceuticals, or would fusion proteins and conjugate products take over? “I do believe that the classic biopharmaceutical will have a strong future,” he said, “specifically monoclonal antibodies. You see maybe some fusion proteins here and there, some PEGylated compounds, and some other things — antibody fragments or modern antibodies, more engineered antibodies. But the current leading antibodies still will have a great future and only a few if any of these new compounds will be as big as leading current biologics, especially the antibodies.”

What about process technologies? Will we see fully disposable process trains or a continued presence of stainless steel in well into the future?

“The world is not black and white,” Roebers said. “Both will always exist. Single-use technology will find increasing use. Especially exciting, to me, are the disposable bioreactors. These are now very well established up to 1,000 L, some go to 2,000. If you combine them with modern, high-titer expression systems, you can get comparable productivity to that of stainless steel bioreactors. So I do think that single-use technology will play an increasing role.

“At the same time, I also think that stainless steel is far from dead. I think it will continue to be the preferred way to make the current leading biologics and antibodies because they’re already in those systems. They’re well-established. I do predict that fewer new large stainless steel facilities will be built because efficiency and titer increase with new products, so we just won’t need those large facilities for new products. You may actually see stainless steel bioreactors going down in size. I’m not a big believer in the idea of personalized medicine driving smaller bioreactors or single-use systems. I just think that in the future there will be fewer blockbusters, which will lower the need for high productivity. Single-use bioreactors could be part of a fully disposable, or fully single-use production train for smaller-volume, smaller-mass biologics.”