In June, 2000, a conference entitled BioLogic 2000 was held in Geneva, Switzerland, the core topic of which was manufacturing of biologics in Europe. The conference demonstrated that the European Community is keenly aware that resources for the manufacture of therapeutic proteins in Europe are becoming capacity constrained. While considered well behind the U.S. in terms of biotechnology product development, the Europeans are nonetheless ahead of the U.S. in coming to grips with the unavailability of capacity to manufacture biotherapeutics.
As a result of timely recognition of the problem, there is now a movement afoot across the continent to solicit the support of national governments, banks and venture capitalists to underwrite the cost of manufacturing facilities. However, that process could require five years or more to bring supply into balance with demand for manufacturing. Until the new European facilities can be validated, BioLogic conference attendees seemed secure in the belief that European product R&D companies could seek interim protein manufacture from the installed base of North American contract manufacturing organizations (CMOs).
Attending that conference as a representative of the U.S. CMO segment, it was uncomfortably encumbent upon us to deliver the message that there is very little available capacity remaining in the U.S. for the manufacture of biologics. In fact, U.S. and European companies have consumed most of the most recently built capacity brought on board in 1997, including two mammalian and microbial facilities operated in Europe by Boehringer Ingelheim. This message is generally received with disbelief and denial. In fact, the U.S. manufacturing capacity crisis has become the best-kept secret in biotechnology and, to date, disbelievers include most of the product R&D sector as well as the entire financial community. In fact, the CMO segment’s stubborn refusal to disclose the extent of the problem has been a disservice to the industry for two key reasons: Customers are being told at the eleventh hour, when there is insufficient time to arrange for alternative manufacturing solutions; and lenders are being led to believe that overcapacity exists and hence will not underwrite plant expansion projects.
The scarcity of manufacturing resources will become biotechnology’s next growth rate-limiting factor, and will undoubtedly create a slow-down in the rate of product approvals for the next several years. The precise magnitude and duration of the problem cannot even be predicted until resources are brought to bear upon the solution. However, no solution will be implemented until the reality of this market condition becomes globally accepted. We will clarify the historical events that gave rise to the imbalance between manufacturing supply and demand as well as to propose technology-based solutions that not only provide long term scale-up solutions, but may also insulate an individual company’s product pipeline from development delays that are likely to occur globally during the short term.
Make Vs. Buy: A History of Capital Constraints and Regulatory Issues
Anyone who has been affiliated with biotechnology for the past decade will recall protracted debate over the topic of “Make vs. Buy.” In fact, early seminars often dedicated multiple speaker slots to conflicting arguments in support of in-house manufacture or outsourcing. In the 1980s, the biotech “industry” was divided between:
• a few well-capitalized, vertically integrated companies like Amgen, Chiron and Centocor, which were fortunate enough to have early product successes that merited capital investment in manufacturing facilities, and
• a preponderance of small, “virtual” biotechs for whom capital constraints and aggressive product development timelines made in-house manufacture an unaffordable luxury.
To satisfy the disproportionately large “virtual” niche, several major CMOs constructed facilities in the 80s, mainly to produce the kilogram quantities of monoclonal antibodies that were expected to be approved in the near term. By the end of that decade, however, monoclonals were still in protracted development and those newly built CMOs failed for lack of business. They were acquired by product companies and converted to “captive” manufacturing operations.
In addition to capital constraints and negative net present value calculations, the decision to make versus buy was confounded by regulatory issues. The FDA was not predicted to “approve,” or even facilitate the use of CMOs to produce licensed product. In fact, prior to 1997, product approvals required two separate applications, the Establishment License Application (“ELA”) and the Product License Application (“PLA”). In order to outsource the manufacture of licensed (approved) product, a product sponsor was compelled to make the holder of the ELA at least a partial owner of the product.
In 1997, the FDA Modernization Act changed the definition of the manufacturer and created one application, the Biological License Application (“BLA”), which deemed sponsor and manufacturer to be the same entity, thereby fostering the outsourcing of licensed product to CMOs. The FDA reform also eliminated another obstacle to outsourcing by pronouncing certain proteins “Well Characterized Molecules” (now referred to as “specified products”), meaning that the manufacture could be transferred from one site to another with a greater degree of simplicity than before. This facilitated the use of multiple manufacturing sites and vendors, or the transition from in-house to outsourcing and back in-house, if necessary, without repeating clinical trials.
The regulatory reform in 1997 served as the catalyst that caused the biotechnology industry, after years of indecision, to whole-heartedly embrace outsourcing as the “best” (often the “only”) way to accomplish manufacturing. Corporate strategic planners could cast aside their net present value calculations because capital constraints in biotechnology made it nearly impossible for the average company to even consider anything but outsourcing. The regulatory reform, therefore, merely gave credence to a decision that Wall Street had made a foregone conclusion.
CMO Capacity Planning Adequate For Manufacture of Clinical Materials
By 1990, only 18 products had been approved in the eight years since the first approval in 1982. However, 10 more products were approved by 1993 and, by 1996, approvals were averaging 14 per year! In fact, half of the products on the market today were approved in just the last five years. Monoclonals, which had been predicted to be the largest single product type, had lagged behind since their development in the late 1970s, but 10 were approved in 1996 and 1997. The human anti-mouse antibody reaction (HAMA) issues that caused their delay has been resolved by the advent of humanized or fully human cell lines; it is expected that many more antibody successes will come out of the clinical pipeline.
The upsurge in product approval rates and demand for outsourcing caused the contract manufacturing segment to revisit its earlier failed strategy and plan for expansion in the early 1990s. By 1997, a total of 500,000 square feet of new capacity came on-line in North America. Lonza (formerly Celltech), Covance and DSM (formerly BioIntermediair) all validated new space in 1996 and 1997, together with BioScience Contract Production and Goodwin Biotechnology, which had entered the therapeutic production arena as well. This new capacity was dedicated to serve only the market for manufacture of clinical materials, because the FDA Modernization Act was not passed until 1997 and its impact at that time still unforeseen. By 1997, it appeared that the biotech industry was well-equipped with more than ample capacity to produce clinical trial materials. In fact, an overabundance of capacity made it possible for product R&D companies to secure contracts, on a timely basis and at reasonable prices, for Phase I, II and III manufacturing in these new state-of-the-art facilities.
Clinical Manufacturing Capacity Absorbed by Late-Stage Products
Incredible as it may seem, the 500,000 square feet of manufacturing space, which was overabundant for the manufacture of clinical trial material in 1997, will be absorbed entirely by the end of this year. Even today, industry “experts,” including analysts, lenders and equity investors, believe that a glut of capacity still exists and that manufacturing is still an undesirable business. (After all, CMOs went bankrupt in the 1980s due to lack of demand for their services.) In order to accept the reality of the current critical shortfall, it is important to understand a few key ratios. Accepted industry “wisdom” suggests that one square foot of capacity can produce approximately $1,000 of product annually. Therefore, the 500,000 square feet of manufacturing space can be expected to produce only $500 million of product annually. If that space were dedicated to clinical trial materials, then 500,000 square feet of space can produce approximately 300-500 Phase I/II products, or approximately 100 Phase III products. Therefore, the existing space could produce nearly all of the clinical requirements of the 500 products currently in the U.S. biotechnology pipeline.
The CMO capacity shortage arose from the customers’ desire to outsource the manufacture of large volume, approved products. A typical biotechnology product’s market size can range from $50-$500 million at retail and carries a cost of goods of approximately 10% of retail selling price. Therefore, a $500 million market will require $50 million of product, at the CMO “wholesale” price. Since one square foot of space can produce $1,000 of product, then the 500,000 square feet of existing space could be expected to produce only 10 or 20 approved products (at $25-50 million for each product). So, when the 1997 regulatory reform encouraged product sponsors to embrace outsourcing of approved as well as clinical products, not many products were required to consume the 500,000 square feet of space available. Since the manufacture of approved products represents high dollar value and high margin work, CMOs were quick to “cherry pick” their client inventory, opting to allocate precious space to late-stage and licensed products, and to clients with robust pipelines of products and proven clinical success.
Poor Planning
In summary, the imbalance between demand for biologics manufacturing and supply of manufacturing capacity has come about for one key reason: There has been no concerted effort between suppliers and customers to develop cohesive manufacturing strategies for the long term. In the past, abundant space lulled product R&D companies into believing that they did not have to share their product plans with vendors. They could just walk into any CMO and be given a “one-stop shop” with a menu of fully integrated services including pre-clinical through scale-up manufacture.
Because there has been no need to do so, product development companies have been unwilling or unable to make commitments to manufacturers for long-term utilization of space. As a result, CMOs remain reluctant to build such space. Looking at the history of the CMO sector, the visible failures of the 1980s were as a result of the “If We Build It, They Will Come” mentality. Even if the CMO sector were willing to invest in speculative capacity, there have been no ready sources of funding willing to take such a risk. If capital for biotechnology R&D has been cyclically unavailable, capital for bricks and mortar is nonexistent.
Short-Term Stop-Gaps, Long-Term Solutions
As of the fourth quarter of 2000, many Phase I/II customers are finding themselves turned away by large manufacturers. Some will be fortunate enough to find second-tier manufacturers that still have capacity; others will be forced to incur timeline delays even in lead products. There are approximately 300 products in late-stage U.S. clinical trials, 200 in early stages, and more than 1000 in various stages of pre-clinical development. If historical approval rates continue, approximately 30-50 of these products should be approved each year, during the next five to seven years. Since existing manufacturing capacity is utilized and none of the leading CMOs (except Boehringer Ingelheim) has yet broken ground for expansion, there could be a shortfall of 20-30 product approvals per year for the next five years, since building and validating new capacity requires four years or more. Industry analysts estimate that a single Phase III product can add approximately $800 million to the market cap of its sponsor company; accordingly, the biotechnology industry is at risk of losing $9 billion in market cap due to delayed products in the next few years.
Since none of the established CMOs has yet commenced expansion and the construction and validation timeframe is rather lengthy, no short-term fix is yet in sight. Manufacturing capacity can still be found at some of the second-tier CMOs, but this space will be consumed rapidly and customers will find that they must be willing to provide long-term commitments in order to secure such space. In addition, product sponsors must be willing to invest or at least partially guarantee funding for expansion. There are simply no third-party financing options available to CMOs for independent expansion. It will ultimately become necessary for product sponsors to investigate the option of buying excess capacity from biopharmaceutical companies that have already completed expansion for their own product pipelines. This option has been shunned in the past because biopharmaceutical companies will necessarily expand to fill their own space when needed, potentially leaving contract manufacturing customers without alternative space, just when their products are reaching commercial viability.
The biotechnology industry is just now beginning to fulfill the growth and profitability that it has long promised. By the end of the decade, biotechnology products could easily command one-half of the $300 billion pharmaceutical industry. Recent developments in the characterization of the human genome and in proteomics and bioinformatics will foster even more rapid and prolific development of novel targets and drug candidates than ever. Monoclonal antibody products alone could reach double-digit approval rates in the near term. As of October, 2000, $28 billion has been raised in IPOs and follow-on offerings, more than three times the capital raised for biopharma in 1999. All of this activity suggests that traditional manufacturing methods will not be adequate to produce the quantities of protein that will be needed in the future, even if traditional suppliers of manufacturing capacity were to triple or quadruple their facilities.
Finally, any discussion of issues confronting the manufacture of biologics would be incomplete without also mentioning pharmaco-economics. For, while the rapid advances in drug discovery and healthy capital markets will foster growth in approved products that will require scale-up methods not yet commercialized, so will the maturation of the industry require that those scale-up methods enable manufacturers to reduce unit costs of proteins drastically. As the industry matures, it will face increased competition in disease indications, the nearing generic status of some of biotech’s early approvals, and continued patient reimbursement issues posed by managed healthcare. The most viable solution to ensure adequate capacity for protein production (while at the same time achieving desired unit cost targets) may be offered by the field of transgenics.
Given the structural complexity of biologics, the manufacturing unit will continue to be living tissue; therefore, R&D scientists have turned their attention to devising techniques for transferring a gene from one organism to another. Such “transgenes” can now be effectively placed into animals or plants, having the obvious advantage of being able to be scaled efficiently and quickly to virtually any level.
Transgenic Large Animals
In this technique the gene coding for the biologic product is inserted into the egg of an animal such as a rabbit, pig, goat, sheep or cow. When fully grown, the transgenic animal carries specifically altered DNA that is stably integrated into its genome. For biopharmaceuticals, this foreign DNA contains the genetic information that specifically directs secretion of a target recombinant protein into milk, along with the host animal’s own milk proteins, during lactation after the birth of offspring. On a per cell basis, productivity is similar to mammalian cell culture systems, but since mammary gland cells are so much more dense, concentrations of product in milk are considerably higher. From a capital cost and maintenance point of view, animals are far less expensive than stainless steel bioreactors or fermenters; therefore products made in these systems have a much lower upstream cost than traditional systems. Indeed, transgenic animals have been employed to make kilogram levels of biotherapeutic product, with per gram costs on the order of $100 per gram for purified injectable material. This is very competitive with microbial culture and it is possible that costs for injectable material, depending upon the type of animal and protein expression levels, can be driven down further.
There are three companies that specialize in transgenic animal technology, Genzyme Transgenics Corp. (Framingham, MA), PPL Therapeutics, Ltd. (Edinburgh, Scotland and Blacksburg, VA) and Pharming b.v. (Leiden, The Netherlands). Among them, they employ cows, sheep, goats, pigs and rabbits as transgenic production models. However, none of them offer what might be termed “traditional CMO services,” primarily due to the length of time (and high cost) required to produce suitable lactating animals. For a transgenic goat, from first transgene introduction to full lactation, this time varies from 18 to 28 months. For a cow, this period can vary from 33 to 47 months. Also, cost and timing does not permit producing a sufficient number of transgenic animals to screen for high producers and the task of going from one “founder” animal to a herd necessary for large-scale production may require additional years of effort, even with advanced cloning techniques.
Nonetheless, there are a number of biotherapeutic products made in transgenic animals in the clinical pipeline, mostly as a result of joint collaborations between the firms noted above and biopharmaceutical partner companies. In addition, the FDA issued a “Points to Consider” document in 1995 covering the regulatory issues of concern to product developers employing transgenic animal manufacturing of therapeutic proteins. It is very likely that a biopharmaceutical made in transgenic animals will be approved by the FDA within the next two years, thus validating this approach from a regulatory standpoint.
Avian Transgenics
In the past few years, several companies have developed methods for creating transgenic chickens, the target protein being expressed in the white of eggs. Potential advantages of this technique, which has not been commercialized on a large-scale, include the relatively short time to create and breed generations of transgenic animals (compared to larger mammals), and the amount of target protein which can be extracted from a single egg (~ 1 gram). This speed and productivity (on a per egg basis) mean that relatively large amounts of protein can be generated quickly, protein that can be tested for structural integrity and effectiveness compared to the same protein made by mammalian cell culture or microbial fermentation. One of the major technical hurdles has been the difficulty of manipulating the chicken embryo and the inefficiency of gene transfer methods for poultry.
Typically, the frequency of insertion of the gene construct into the zygote is around 1% due to random integration; however, new techniques are being developed employing replication incompetent retroviruses or using transposons that “jump” from one chromosomal location to another. Such procedures may increase the insertion frequency by a factor of 20 or more and bring these systems closer to commercial readiness. Currently, there are at least four companies involved in developing these techniques, AviGenics, Inc. (Athens, GA), TransXenoGen (Shrewsbury, MA), GeneWorks, Inc. (Ann Arbor, MI) and Origin Therapeutics (San Francisco, CA). These companies are working closely with commercial clients to develop large-scale production capabilities in transgenic chickens for their clients’ products.
It should be noted that there are other transgenic animal based technologies in earlier stages of development. For example, AquaGene (Alachua, FL) is pioneering the use of transgenic fish to make therapeutic proteins.
Transgenic Plants
Within the past five years, transgenic plants have emerged as a potentially very effective way to manufacture therapeutic proteins. Expression systems have been developed for many types of crop plants, including corn, tobacco, potato, tomato, canola, rice and also non-crop plants such as lemna (duckweed). Plants are seen by many as having the best potential for truly low-cost, large-scale production of therapeutic proteins. Unlike animal systems, stable introduction of foreign genes is easy and efficient. Biomass production (either seeds or tissue) is very cost-efficient. Plants cannot harbor human infectious agents such as viruses and prions because these agents cannot replicate in plants. Plants also generally lack closely related structural homologs that can pose purification problems; for example, plants do not make their own antibodies. Finally, plants perform many of the complex protein processing steps, such as isoprenylation, oligomerization, disulfide bridge formation and proteolytic cleavage, which are required for bioactive therapeutic proteins.
While many different plant expression systems exist, they basically fall into two categories, protein production in the seed and protein production in the tissue. Corn has been transgenically modified to produce monoclonal antibodies (Integrated Protein Technologies, St. Louis, MO) and recombinant proteins (Meristem Therapeutics, Clermont-Ferrand, France) used in human clinical trials. Human clinical trials have also been conducted with antibodies produced in tobacco (Planet Biotechnology, Mountain View, CA) and vaccines produced in potatoes (Cornell University, Ithaca, NY). Worldwide, there are more than 20 companies that have developed various plant expression systems, but only three companies have adopted a “CMO strategy” to bring their technologies to market. Integrated Protein Technologies and Meristem Therapeutics use both corn and tobacco and are developing, on a fee-for-service basis, manufacturing systems for clients. Meristem has made and purified recombinant therapeutic proteins at the kilogram level. CropTech Corporation (Blacksburg, VA) employs a unique inducible promoter enabling post-harvest expression of proteins in tobacco. All three companies are developing, on a fee-for-service basis, biotherapeutics for client companies. Other companies in this sector, such as SemBioSys Genetics Inc (Calgary, Canada), ProdiGene (College Station, TX), Applied Phytologics Inc. (Sacramento, CA), and Large Scale Biologies (Vacaville, CA) are engaged in development of their own biotherapeutic products in collaboration with other companies.
Assessing Transgenic Technologies
Currently, there are no transgenic technology CMOs with the experience and demonstrated capabilities for cGMP manufacture of biotherapeutics that can readily be found with fermentation and cell culture CMOs. As noted above, there are many companies and technologies to choose from in looking for a CMO or development partner for transgenic production of your protein. So which is the “right one” for a sponsor’s product?
As with any outsourcing decision, the key decision elements and their relative weights are different for each product, so much so that companies often employ different production technologies for different products. We must review each of these key elements:
• Feasibility Study Capacity
• Process Development Experience
• Plans and Commitment to Full-Scale Manufacturing
• Freedom to Operate
• Technology Pros and Cons
• Regulatory Considerations
Feasibility Study Capability
Transgenic technologies are still relatively new and, as such, many potential clients need to be convinced of any particular technology’s ability to cost effectively make a bioactive protein. Unless the sponsor has unlimited funds and time, it will have to select, at most, only a few technologies to compare. Therefore, speed and cost will be important. Also, it is important that the small-scale feasibility study be able to be extrapolated to the large-scale end-point on the same speed and cost terms. Most companies would like to see costs and timing comparable to the cell culture and/or fermentation technologies with which they are already familiar. Thus, it would be desirable to go from cDNA to milligram levels of protein within a three- to six-month period, at a cost comparable to performing the cell culture or fermentation work in-house (say, approximately $50,000). Each product developer may have different cost and timing thresholds, but the above numbers are typical of the industry.
Quick and inexpensive feasibility studies are not possible with large animal systems — the biology of gene transfer and animal growth and development simply cannot be hurried. These companies often employ smaller animals (e.g., mice) that can shorten the costs and timing considerably, but then the question arises as to whether the data can be quantitatively (or even qualitatively) extrapolated across species. Depending upon the method and effectiveness of gene transfer, the avian transgenics companies should be able to meet a shorter timeframe with the added advantage that protein production could reach the gram level. The plant transgenic companies are also faced with basic biology issues; however, growth and development of plants occurs in a much shorter time than with large animals.
Two companies, Meristem Therapeutics and CropTech Corp., advertise short-term feasibility studies in their tobacco-based production systems. Tobacco, among all crop plants, is uniquely suited to these studies because of the existence of transient expression systems (i.e., one doesn’t have to transform the plant to make transgenic proteins), and the fact that codon reoptimization is minimal for human genes (thus reducing the upfront time required to create transformation vectors). The question still arises as to the transferability of expression data to the large-scale (i.e., transformed plant) situation, but at least one is not dealing with cross-species extrapolation.
Process Development Experience
Demonstrating good expression levels and active protein at the milligram level in a feasibility study is one thing — developing a full-scale process is an entirely different matter. When looking at potential partners, ability to move from milligram to kilogram scale will be essential to meeting product development timelines. The transgenic animal companies have demonstrated experience and capabilities to do this. The field narrows somewhat in the transgenic plant area where, in most cases, companies are still engaged in developing and optimizing their expression systems while at the same time developing pilot-scale production capabilities. Of the transgenic plant companies advertising themselves as CMOs, only Meristem and Integrated Protein Technologies have produced therapeutic-grade proteins at the hundreds of grams (or higher) level that have gone into human clinical trials. Both of these companies are exploring options for full-scale facilities with multi-hundred kilogram capabilities. There are certain developed capabilities at other firms; for example, Large Scale Biologies has a processing facility operational in Kentucky, so it is a good idea to check on the capabilities of several companies before making a commitment.
Plans and Commitment To Full-Scale Manufacturing
The obvious corollary to process development capabilities is large-scale manufacturing plans. Analogous to cell culture and fermentation, there are upstream and downstream components to biologics manufacture in transgenic systems. At this time, similar downstream technologies are employed for all transgenic systems, with the sole exception that plant-based systems do not require insertion of viral clearance steps (see “Regulatory Considerations” below).
On the upstream side, both plants and animals are similar in that there are two distinct components. Protein must be “grown” (in the milk or egg of animals; in the seed or tissue of plants) and then must be harvested and processed to unpurified bulk form. These growing and processing components can be (and often are) geographically distinct from each other, which can lead to certain manufacturing inefficiencies. It would not make sense, for example, to have herds of transgenic animals or acres of transgenic plants in the urban area typical of most cGMP-compliant manufacturing facilities. However, processing of plant or animal tissue to extract the protein to a stable unpurified bulk form can occur in a more rural agricultural area and the bulk product shipped to the downstream processing facility.
The transgenic animal companies have constructed large-scale (i.e., tens to hundreds of kilograms of purified product per year) cGMP-compliant facilities that have been used to manufacture product currently in human clinicals. In terms of experience and availability for commercial manufacture, these companies are several years ahead of their plant competitors. There is no large-scale (i.e., 100 kg or higher) cGMP-compliant facility yet for plant-based systems, but many companies are moving in that direction. In assessing a particular firm for collaboration on a product, one should expect that firm to have a clearly defined plan for pilot and large-scale production processes and facilities. There should also be an equally clear understanding of the financial implications of development and construction of such facilities.
Freedom To Operate
Potential issues with intellectual property may be the most difficult to deal with, as the transgenic industry begins to show promise for production of licensed biologics. On the animal side, the issues and landscape are clearer as the technologies are already being commercially employed and the number of companies involved is relatively small. There is still considerable uncertainty with plant-based systems, particularly with respect to transformation techniques. There are many transformation technologies that are being employed successfully (e.g., Agrobacterium tumefaciens, use of viral vectors and particle bombardment), but each technique has its plethora of patents surrounding not only the general technique itself, but also individual elements of the technique.
Each technique may also have on-going or potential litigation that must be considered. For example, the use of Agrobacterium to transform plants will be problematic for those who have not negotiated the appropriate licenses or choose to ignore the current patent landscape. Until recently, use of this organism was under dispute between Monsanto and Washington University of St. Louis. Also, components of this system (e.g., promoters, terminators and binary vectors) are proprietary to other companies, some of which are ready and willing to grant licenses for commercial use, and others of which are not anxious to negotiate any rights.
In view of these and other intellectual property issues, it is vitally important that the product developer clearly and unambiguously understand the position and rights of the prospective CMO in the plant arena. It is also important the strategy not be “wait and see;”