Leveraging Antibodies

Human antibodies are hot. Three leading companies with patented means for making them are battling for partners and strategic position, anxious to sign alliances before current competitors muscle in, or next-generation technologies come along. Abgenix and Medarex effectively share a duopoly on transgenic mice that make human antibodies. Both firms license their technologies on an antigen-by-antigen basis, but are moving to build more value by deploying the methods on their own behalf, and via 50/50 deals with companies willing to share targets and development costs. Cambridge Antibody Technology is the leading promoter of phage-display technology, a bacterially-based, relatively high throughput method of making human antibody fragments that can be used as reagents to validate targets, or built up into drugs. Owning products is the end-game for all three players. But their approaches to the goal differ, particularly in the way they're structuring deals and the number of products they intend to put into clinical development.

Companies with rights to key antibody technologies are leveraging their holdings and high valuations to fill out into fully-integrated firms. The competition is fierce.

by Deborah Erickson

• Human antibodies are hot. Three leading companies with patented means for making them are battling for partners and strategic position, anxious to sign alliances before current competitors muscle in, or next-generation technologies come along.

• Abgenix and Medarex effectively share a duopoly on transgenic mice that make human antibodies. Both firms license their technologies on an antigen-by-antigen basis, but are moving to build more value by deploying the methods on their own behalf, and via 50/50 deals with companies willing to share targets and development costs.

• Cambridge Antibody Technology (CAT) is the leading promoter of phage-display technology, a bacterially-based, relatively high-throughput method of making human antibody fragments that can be used as reagents to validate targets, or built up into drugs.

• Owning products is the end-game for all three players. But their approaches to the goal differ, particularly in the way they're structuring deals and the number of products they intend to put into clinical development.

By now, everyone knows that antibodies have come of age. After a tumultuous decade of hope, failure, disappointment and renewed enthusiasm for the commercial prospects of antibodies, it's clear these immune-system proteins can be drugs that are good by both therapeutic and financial measures. Idec Pharmaceutical Inc.'s anti-lymphoma antibody Rituxan, co-marketed by Genentech Inc. , had racked up sales of $250 million through the third quarter of 2000. Genentech's own Herceptinfor breast cancer, launched late in 1998, generated sales of nearly $160 million in the same period. Meanwhile ReoPro, the anti-clotting antibody fragment developed by Centocor Inc. , a division of Johnson and Johnson, and marketed by Eli Lilly & Co. , is expected to generate some $500 million in 2000. With success stories like these erasing memories of previous failures—most spectacularly, the Phase III crashes of anti-sepsis antibodies created by Centocor and Xoma Ltd. —it's small wonder that nearly a third of biotechnology products now in development are monoclonal antibodies.

The question is, who stands to gain the most from the basic technologies for making antibodies—and how will the value be extracted? The earliest means of making the drugs earned the inventors Gerard Kohler and Cesar Milstein a Nobel prize, but no money. The duo never patented their method of immunizing mice with human antigens, to provoke the animals' immune systems into making antibodies that could then be harvested and developed as therapeutics. That mouse-based, or murine, method launched the field, but the drugs turned out to be of limited use, because patients' immune systems reacted to the mouse proteins and caused severe side effects.

The people and companies that developed new and improved means of making antibodies did not repeat Kohler and Milstein's business mistake—every method and twist since has been heavily patented, and the patent-holders are wielding those positions for all they're worth. Protein Design Labs Inc. , which holds rights to technology developed by its CSO Cary Queen, PhD, for humanizing murine antibodies by replacing certain components, has commanded hefty licensing fees. So too, Celltech Chiroscience PLC , which patented a method devised by Michael Boss, PhD, for large-scale expression of humanized and chimeric antibodies. Many of the humanized antibodies now in clinical trials, if successful, will end up owing royalties to PDL. There's no telling how many of the drugs will win approval, however, nor how many more humanized antibodies will follow those into the clinic. PDL's licensing deals have dwindled as other technologies have taken the vanguard.

In the past few years, drugmakers have favored companies with methods of making not just humanized but fully human antibodies. Medarex Inc. of Annandale, NJ, and Abgenix Inc. in Fremont, CA, both breed strains of mice genetically altered so that their bodies create what are essentially human antibodies, instead of mouse antibodies. The anticipated high affinity and minimal toxicity of proteins produced this way makes them highly promising drug candidates. The valuations of these two companies peaked in the early months of 2000 and have slipped slightly lately, but are still undeniably robust. At press time in mid-December, Medarex stock was trading for about $47 and its valuation stood at $3.4 billion. Abgenix was trading at $60, with a valuation of nearly $5 billion.

The market's appreciation derives from the fact that antibody-generating technology isn't just a helper tool: it's a direct means of making antibodies, drugs of a class that by their very nature are ready to enter clinical trials faster than small molecules, and do so with lower risk of failure. The major appeal of the British firm Cambridge Antibody Technology Group PLC (CAT), which generates antibody fragments that can then be built up into whole antibodies, is that it offers a high-throughput way of generating binding reagents to large quantities of targets about which nothing is known. The firms all benefit from a central similarity: their technologies are not so much fixed platforms as fountains that the firms and their partners can keep dipping into with different targets—if their deals are structured that way.

Because their technologies are so similar, Abgenix and Medarex are competing directly against each other for alliances with firms interested in making human antibodies. But the antibody arena contains other firms flaunting their own sorts of antibody-generating technology—such as the bacterially-based "phage display" method hawked by CAT; its arch-rival and litigation adversary MorphoSys AG , of Munich, Germany; Biosite Diagnostics Inc. ; Dyax Corp. and others. CAT is the valuation leader among this group. At press time the company's stock was trading on the London Stock Exchange for about $43 (3627 pence) and its valuation stood at approximately $1.73 billion (£1.187 billion).

Other firms are working to develop their own novel methods of making antibodies, hoping to overcome some still-outstanding problems with the present technologies, including the relatively high cost of manufacturing and the inability of current antibodies to penetrate cells, rather than merely bind to their surfaces. Genzyme Transgenics Corp. , for instance, is making antibodies in goats, while Epicyte Pharmaceutical Inc. aims to turn tobacco plants into antibody factories. A host of younger firms, including MacroGenics Inc. , Diatos SA and Raven Biotechnologies Inc. think their technologies will bring significant advantages over current methods—and some big-name venture capitalists are betting they're right.

In short, at a technology level, the competition for turf in the antibody field is fierce, and it's only going to get tougher. Companies now at the cutting edge know they've got to make the most of their technology while they can, before their methods are superseded, and before the firms with the most attractive potential targets ally with a competitor. It's a classic platform/product battle: use the technological platform while still unique to gather in on a financially advantageous basis as many product candidates as possible before the best partners, targets and projects go to a competitor or to the next generation of platform companies.

To increase their share of the downstream revenues from these projects—and so far Medarex, Abgenix and CAT have signed on for dozens—the antibody firms are often foregoing big upfront fees and ongoing research payments for significant shares of the downstream action. Thanks to the hot market in 2000, they've got the cash to do so: Abgenix has $750 million cash in the bank and Medarex, $400 million. CAT does charge upfront fees to help it carry out its resource-intensive process, but thinks doing so helps assure that partners are not just playing around with new technology, but are committed to developing products.

The competition between these three firms and their competitors thus revolves around two basic issues—a marketing battle for partnerships and a strategic positioning battle. The top firms all claim to have the best and fastest technology, and thus the best commercial prospects. CAT trumpets its advantages in the earliest stages of drug discovery—the ability to use its technology on a relatively high-throughput basis for validating the thousands of potential targets tumbling out of genomics programs. Medarex and Abgenix emphasize the quality of their antibodies as potential drugs—and their newly acquired and in-licensed capabilities for doing the same kind of target validation CAT and the other phage-display companies claim they can do.

But for now the field is moving too fast to declare a winner. That's why understanding the companies' strategic positioning is important—it's how they're preparing themselves to create long-term value. Both Medarex and Abgenix are still licensing their technologies to partners that will pay royalties, but their focus is on their own products or on partnerships that will allow them to keep more of the value they help to create. Abgenix aims to test many antigens but to put just two carefully culled antibodies into the clinic each year. Medarex has far more ambitious goals and has been structuring itself, geographically, to achieve them: it aims to put 10-12 candidates into trials each year, both on its own and through its Danish spin-off Genmab A/S. CAT is further behind: its technology has been more frequently used for discovery-stage target-validation programs than for product development. But CAT is more than a mere target validator: four of its antibodies are now in clinical trials and the firm expects to bring the first fully human antibody to market.

Well-Endowed Mice

As recently as October, Abgenix's CSO C. Geoffrey Davis, PhD, made the case for the superiority of his firm's XenoMouse, by emphasizing that, "Our mouse is well-endowed: it has nearly all the human antibody genes." Medarex's HuMAb-Mouse, on the other hand, had only 15% of the genes, which meant, Abgenix argued, that it would have a more limited initial response to antigens than Abgenix's mouse, which contained 85% of human genes, and would therefore produce a smaller repertoire of antibodies.

But by early December, Medarex announced at the Antibody Engineering Conference in San Diego—where companies interested in the field all had representatives on hand to hear the news—that it was introducing "a new and improved transgenic mouse, which will produce the highest affinity, fully human antibodies to test and create new drug therapies for humans." The mouse is a cross-breed between the company's HuMAb-Mouseand the TC Mouse it got through an exclusive partnership [See Deal] with Kirin Pharmaceutical Division of Japan's Kirin Brewery Co. Ltd. Medarex proclaimed, "This is currently the most advanced transgenic mouse on the market, and it will significantly improve the antibody industry. The company can now create the highest-affinity fully human antibodies for any isotype."

Abgenix CEO R. Scott Greer was ready with a quick counter-punch the day of the Medarex announcement: "Their new mouse should contain all the human variable genes, but whether they're all functioning is another question. Not much is known about that mouse yet." Greer suggested that if observers compare the two technologies that gave rise to the new mouse with Abgenix's XenoMouse, in terms of track record, and the number of products based on the technology already in clinical trials, his firm would come out looking better. "Theirs is just off the shelf," he pointed out.

For his part, Medarex CEO Don Drakeman says the introduction of the new mouse was just one more in a series of steps his firm has been taking to build a complete technology platform—not an attempt to counter Abgenix's claims about the technical superiority of the XenoMouse. "We've accessed tools like this and others because they're useful to us and our partners. We're the only company with a complete platform for generating many therapeutic products from genomic targets," he declared.

Despite his assertions to the contrary, some of Drakeman's recent strategic moves do seem calculated responses to specific competitive pressures, because they're moves that put Medarex in position to claim that it can do anything competitors can do. If potential partners are wavering, Drakeman is in better position to sweep them in if he can say, "Sure, we can do all that and more." In a sense, he's following a one-stop-shopping strategy. Why go anywhere else?

An alliance Medarex entered with point-of-care diagnostics company Biosite in June 2000 [See Deal], for instance, lets the firm offer potential customers phage-display technology, just like CAT can. Simply having the method in-house helps Drakeman advance his spin about Medarex's integrated platform in contrast to the offerings of competitors. The way he talks about it, phage display is just a piece in his platform—not a competing technology, and thus nothing that drugmakers, or investors for that matter, should get overly excited about.

"Phage display is a tool we'll use when we need it, when we want to validate a target," Drakeman says. It's not necessary to use when a target is already known to be a good one; and it's also just one of many ways of validating targets, he points out. To access the technology, Medarex agreed to pay Biosite $3 million annually for eight years, plus research fees, milestones and royalties. Drakeman says he's pleased with Biosite's strong patent position, because it gives Medarex freedom to operate, and lets the company avoid lawsuits like the ongoing one currently entangling CAT and MorphoSys in Europe.

The positive buzz about phage-display technology is that it's a fast way of generating antibodies which can be used to quickly validate targets and create a starting point for developing a drug. That's a marketplace perception which CAT's CEO David Chiswell relishes, finds accurate and wishes to maintain—and one that his competitors would like to dispel. Don Drakeman, for instance, says, "there is some confusion in the marketplace about phage display. It generates antibody fragmentsquickly, but it does not generate full antibodies. You can use fragments for target validation, and then when you find a target you like and validate it, you have a choice of how to proceed. You can build up that fragment, but that takes time and lab work—and that's why the method is not quicker than generating human antibodies in mice. Or we can go back and re-immunize our mice with the target and generate whole antibodies via hybridomas in our mice, which takes a similar length of time."

Greer maintains that if phage display were really a good way of producing drugs, rather than merely research reagents, more candidates would have found their way into trials by now. He notes that, "Phage has been around for six, seven years, and there are four antibodies from it in clinical trials. The transgenic mouse arrived on the scene in 1997. Abgenix has put four antibodies into the clinic, and Medarex has put one. In half the time that phage display has been around, we've put more drugs in the clinic."

Those companies who do put a phage-sourced antibody into the clinic face some legal hassles, say phage's competitors. Many firms and institutions, including CAT, MorphoSys, Biosite, Dyax and the Scripps Research Institute, hold some form of intellectual property relating to phage display, and none of it is cross-licensed. "The minimum risk, as we see it," Greer says, "is that you're looking at possible stacking royalties. At worst, there's uncertainty about your ability to market your product. Why would you want to mess with that?"

Moreover, argues Greer, phage-display technology generally results in lower-affinity antibodies. "The fragments and their binding affinities just aren't up to snuff," he says, "so you have to keep playing with them, screening and screening. It's labor-intensive, like the old humanization methods developed by PDL. You can get reagent-quality antibodies in days or weeks with phage display, but getting therapeutic-quality ones takes much longer." That's the reason, he asserts, that CAT has to get big payments from partners upfront: the technology is too complex for people unfamiliar with it to put it to work. CAT has to teach them how.

Chiswell and his phage-display competitors hate this accusation. "It's absolutely not true that phage display yields lower-affinity antibodies," he says. Moreover, phage advocates say, not only has the technology improved to produce more tightly binding fragments, but even unimproved fragments have shown to be quite valuable drugs. ReoPro, for example, is a chimeric antibody fragment (made with a method other than CAT's). Fragments may be sufficient to block a receptor, as ReoProdoes, yet some scientists say whole antibodies are needed to activate complement fixation, or antibody dependent cellular toxicity (ADCC).

Other researchers point out that turning a fragment into a whole antibody takes only six months and is a relatively simple technological matter. "If the clinical setting is appropriate for use of a fragment, for instance when a drug doesn't need to have a long half-life, then it's only a matter of time and licenses, not technology, that makes starting with whole antibodies preferable to starting with fragments," says Walter Blättler, PhD, EVP at ImmunoGen Inc. , a developer of tumor-activated prodrugs, a kind of monoclonal antibody-drug conjugate. Fragments could come into broader use if their half lives can be extended, for instance through the attachment of polyethylene glycol (pegylation), and that could give phage display a powerful push.

Will Timing Help CAT Now?

CAT's CEO David Chiswell says several factors that came together in the late '90s have sparked a surge of interest in phage-display technology. Genomics is clearly a key driver, because the massive sequencing efforts have brought companies so many new proteins that could be novel drug targets. But at least as important is the fact that other mass-scale technologies such as combinatorial chemistry and high-throughput screening have now been integrated into the drug discovery process. Major drugmakers "spent a lot of money on those supposedly more efficient ways of making drugs," Chiswell points out. By now, companies are more accustomed to thinking about large-scale discovery processes and less reticent about ponying up big sums of money for them—especially if their benefits are tangible.

"The systems to handle all this data [from phage display] are already in place, from the high-throughput screening methods," Chiswell declares. He says, "Our concept is, ‘Make antibodies to everything; try them and see. Treat our antibody libraries like a compound library and put them through assays. If it binds to something, take that drug-like molecule and look more closely at what it does in cells." Chiswell thinks pharmaceutical companies ought to be generating antibody fragments to known drug targets, as well as to the novel targets about which little or nothing is known, because it can't hurt and it could help. He notes that, "ReoProbinds to the same target as aspirin, but it still makes money."

Chiswell counters Greer's criticism about what CAT has to show for itself, given the length of time phage-display technology has been on the market, by saying, "The basic technologies are the same age. The mouse was working in 1991. GenPharm was trying to get deals for it in 1990, but didn't—the major deal flow came for them in the late 90s," mostly after Medarex had taken over. CAT's method is poised to take off now, he asserts, and points to the alliance CAT formed with Human Genome Sciences Inc. (HGS) in March 2000 [See Deal]. The terms of that deal bring CAT the right to select up to 24 proprietary HGS antigens for preclinical development; the British firm can develop six of them on its own. HGS will co-develop with CAT, on a 50/50, global basis, the first three of the six that get to an IND.

But is the deal really a watershed for CAT, or evidence that platform technologies like phage display are less important than the target side of the equation? If CAT got some cash out of the deal—$67 million upfront plus research support—HGS got full rights to use all of CAT's technology in any way it wishes, for 10 years. HGS's CEO William Haseltine, PhD, notes that, "We definitely consider ourselves an antibody company, and we think we should be seen as on par with CAT….We also have some things they do not—our own antigens, and a manufacturing facility."

Chiswell maintains that a 10-year partnership with a firm of HGS's caliber is validation of the merits of CAT's technology. Indeed, he says the experience his team is getting through the arrangement is increasing his firm's credibility in the marketplace. To illustrate the point, he explains that HGS tested a promising antigen called BlyS, for B-Lymphocyte Stimulator, in both CAT's phage-display system and Abgenix's mouse, to see which method produced a better antibody. "At the end of the day, they went with us," Chiswell declares triumphantly. The firms announced at the end of October 2000 that HGS would enter an exclusive development agreement with CAT on the target, a naturally occurring protein that seems to stimulate the production of antibodies. Since the discovery and preliminary description of BLyS' activity was announced by HGS in July 1999, lab studies have suggested that elevated levels of the protein may play a role in autoimmune and neoplastic disorders. HGS thus hopes that an antibody to BLyS may block the protein's pathological effects. It's headed for clinical trials in 2001.

Why did CAT win a partnership on this particular antigen? "The issue with BLyS is that it's similar to mouse protein," Chiswell declares, asserting that this flummoxed his competitor's transgenic mouse, because the antigen didn't look sufficiently different to its immune system. He boasts that the phage system had, within a month or so, kicked out some 10,000 antibody fragments that bound to BLyS. CAT and HGS worked together to characterize more than 1,000 distinct human antibodies, then put them through high-throughput functional screening to generate leads and clinical candidates.

"Within the first six months, with HGS, the whole antibody was finished. So it's true to say that we're fast, and yes, we are resource-intensive. We spend our time trying to pick the best fragment, and the best candidates, instead of immunizing a mouse, letting it sit for six months, then getting the hybridomas that produce just 10 or 12 reasonably good antibodies," Chiswell says.

How often does Chiswell anticipate that a transgenic mouse will be unable to raise antibodies to a human protein? "Often enough," he snaps, arguing that even when that is not a concern—most experts in the field say it seldom will be—drug companies ought to appreciate his technology for another fundamental reason. "With phage display, you're in control of the process, and you get what you want at the end," Chiswell says. CAT can, he claims, define with exact specificity at the start of the antibody-development process, the characteristics it or a partner wants to see in a drug candidate.

"The market is now more comfortable that we will deliver," Chiswell declares, adding, "We can confidently say, you willget antibodies if you come to us—talk to HGS or BASF." CAT did a research-based deal with BASF Bioresearch Corp. , the MA-based division of the mid-sized German company BASF AG , back in 1994 [See Deal], that yielded an antibody which the German firm took into clinical trials. The anti-TNF antibody, D2E7, was the first fully-human antibody to enter Phase III trials. It did so in February of 2000, as a treatment for rheumatoid arthritis, triggering a milestone payment to CAT.

Automation's Appeal

It's been hard for the transgenic mouse companies to argue with the appeal of automation that CAT brings to the drug discovery process. For them, the bottleneck has been the time and effort required to generate hybridomas—the hybrid cell lines made by fusing antibody-producing B cells with myelomas. The cellular crosses have been absolutely key to commercialization of monoclonal antibodies, because unless the cancer cells lend their immortality to the antibody-producing cells, the cells eventually die out and stop producing. Yet the process of making hybridomas is tedious and highly inefficient—the cells fuse properly only about once in a hundred tries.

To address this inefficiency the mouse companies are moving upstream, into CAT's territory. The Biosite/Medarex alliance is one example, uniting the speed and diversity of the phage method with the high affinity that comes from transgenic sources.

But Greer argues that Medarex could yet run into snarls from its use of phage-display because Biosite doesn't have license to CAT's technology. Abgenix went towards automation in a different way. For $77 million or so of its stock, Abgenix acquired a private Vancouver-based company called Immgenics Inc.[See Deal], specifically to get possession of its hybridoma-free way of making antibodies. The small firm's method lets a scientist plate out in individual wells all the B cells producing antibodies in an immunized mouse, so none are wasted in the pursuit of fusion. "They have a media that soups up expression, so the cells make enough antibodies that you can do assays right away," Scott Greer enthuses, explaining, "We'll be able to look at hundreds to thousands more antibodies than we could before."

Because the Immgenics technology is a microplate-based system, it's amenable to robotization, he notes. That will come in handy as the alliance Abgenix has with CuraGen Corp. , a genomics-based drug discovery and development corporation, moves forward. In November, the firms extended an agreement they'd made nearly a year earlier [See Deal], to work together on up to 120 targets related to cancer and autoimmune disease. The new terms say that CuraGen will identify, validate and deliver to Abgenix a whopping 250 targets over the next five years.

Collect 'Em All

The technical wrangling over which firm produces the best antibodies shows no signs of abating. But seasoned executives are looking through the hubbub, and just trying to get on with the business of making antibodies. HGS's William Haseltine believes that, "Transgenic mice and phage display are both good technologies, with advantages and disadvantages." Haseltine's experience to date indicates that assessments are best made on a target by target basis.

Thus, HGS hasn't committed to CAT—it's also got a fairly broad deal with Abgenix, though HGS doesn't get from Abgenix anything like the broad technology access that it gets from CAT. The Abgenix/HGS deal is structured so that the companies choose product candidates from a joint pool of Abgenix antibodies that bind to HGS targets. Each firm pays its own way, with milestones and royalties due to the other for successful development.

But with both Abgenix and CAT, says Haseltine, "there's a substantial amount of work that needs to be done to optimize an antibody for human therapeutic use." He observes, "With some antigens, it appears the human antibodies raised in mice work best, while with others, the phage-display technology of Dyax looks better."

"We are agnostic on the technology," Haseltine asserts, noting, "When we have an antigen, our job is to get an effective antibody to it." He says it's not expensive to try out all the methods; the big expense is developing the antibody as a drug. That's why he figures it makes sense to run basic tests early on, so that he can be sure the best candidate is moving ahead. Since many companies will want to and are already plucking targets from public databases, there's risk in not covering all bases. If the target can't be patent protected, the full brunt of the risk rides with the molecule itself. There's no point picking favorites and letting a competitor apply a method that just might yield a stronger clinical candidate.

Likewise, Bill Rohn, Idec's SVP of commercial operations, says his company has not committed to any particular method of generating antibodies, and won't. The firm's big-selling Rituxanis a chimeric antibody, and it's working just fine. So although fully human antibodies are appealing, they're not necessarily a must-have. Going forward, Rohn says, "We'll use the technology that gets us a product fastest, at the least royalty rate."

Scott Greer's response to the pressure to be "cheapest, fastest" is to introduce a third variable for companies' consideration—namely, which product has the highest likelihood of working. "What's the probability that you'll be able to select a properly functioning antibody, in the shortest time at the best cost?" he asks, adding, "We feel that our technology has the highest probability of success. If your goal is a product candidate, no one can beat us in terms of time."

Greer highlights the fact that Abgenix has created multiple mice sold under the XenoMousebrand. This lets the company offer drugmakers mice that make killing antibodies (IgG1-producing animals) and others that create more neutral antibodies to bind or block a target (via IgG2 or IgG4). Greer says Medarex's original HuMab-Mouse could only make IgG1 antibodies. "They were claiming they could make other sorts of antibodies, but if you pushed them, you'd find those proteins were actually mouse proteins, not human proteins suitable as therapeutics," says Greer. He notes that his competitor's new cross-breed mouse will produce all antibody isotypes in one mouse, "but you get a gemish, a whole mixture. You'll have to search through the whole pool to find what you want." Greer will promote the fact that firms choosing Abgenix's technology can pick the isotype they want, as soon as they've got a clinical hypothesis in mind.

Discussions about antibody quality lead on to talk of manufacturing, because the better an antibody binds to its target, the lower the dose of protein needed to effect a therapeutic response. But whatever the ultimate dose requirements, companies still have to make the stuff somehow, or have it made for them. The issue of manufacturing is one that is just beginning to rear its head, as recognition dawns that antibodies really can move through clinical trials quicker than most small molecule drugs.

The problem is the industry-wide lack of capacity. One key reason Elan Corp. PLC chose Biogen Inc. as its partner to develop Antegren (natalizumab), its antibody for multiple sclerosis, is that Biogen had the manufacturing facility to make the drug [See Deal]. Without access to a plant, antibody developers will, in the best case, have to pay manufacturing fees inflated by the scarcity of capacity and, in the worst case, find their launches delayed or even derailed.

Medarex, which hopes to distinguish itself as the fastest developer in the sector, via its so-called "T-12" program for moving from validated target to testable clinical candidate in 12 months—believes that having and controlling GMP manufacturing will be key to the successful execution of its plans. Medarex has manufacturing capacity through Phase III, Don Drakeman asserts, and the firm is building a facility to handle commercial manufacturing. Already it has the most manufacturing capacity among the antibody-generating companies, though HGS is building more infrastructure too. William Haseltine says, "We have found that manufacturing is a core skill that we must master. Just as we make proteins for own clinical trials, we're going to have to make antibodies too."

Scott Greer is still pondering what approach to commercial manufacturing makes the most strategic sense for Abgenix. The firm purchased dedicated capacity at a contract-manufacturing suite for five years, a few years ago, and that has been good enough so far. He says he doesn't want to get involved with contract manufacturing of antibodies—contending that it's a notoriously low-margin business—though Greer says that it could make sense in certain cases. Abgenix has the ability to produce partners' products in its rented manufacturing suite, he points out, "But we wouldn't take something on if it meant bumping our own," he declares.

Debating Opportunity Costs

For all three players, owning products is the end-game. But all are approaching the issue from different points of view—in terms of the kinds of targets they're looking to access, the kinds of deals they drive with partners (charging higher up-front fees or trading them off for higher downstream rewards), and the number of products they intend to put into clinical development, on their own and through licensees.

The BASF deal was typical of the sort CAT has tended to do—collecting a technology-licensing fee, then milestones as they are achieved, while awaiting royalties on products developed by its partners. But Chiswell says he isn't satisfied with deals of that sort any more: he knows the trend is for companies with powerful tools to use them on their own behalf if possible, and in any event to keep more of the value their technologies help create in the form of products. Now that CAT can claim a successful track record of generating antibodies that proceed into clinical trials, he hopes to attract more favorable deals. "We can now get more evolved business terms, because of the cash that we have, and also because of the credibility we have built. We weren't in position to get better terms before," he declares.

Nonetheless, he says he's been more demanding of partners than his competitors, and expects to have more to show for it in the long run. "You don't have to pay to get mice from Abgenix. So, yes, of course, they can do more deals. But maybe those people are just playing with a new strain of mouse," he declares, explaining, "We say, ‘Give us an assay, and specifics of the kind of drug you're looking for, and pay for us and your group to work together. Otherwise, you're not serious.'" Chiswell says that after 12 months working with CAT, the companies that pay up have a drug candidate and an internal group committed to it.

Some 80% of CAT's products will be developed by partners, Chiswell figures. His concern now is keeping more of the value in the products that phage display helps bring into being. To date, most of CAT's deals have transferred phage-display technology rights to companies planning to use the antibody fragments it generates as reagents, to validate targets. Genentech, Pfizer Inc. , and Lilly, for instance, have licensed rights to use the technology for this purpose. Just a scant handful of CAT's deals link it to products, and then generally only through milestone and royalty arrangements. Few companies have agreed to give the firm targets.

That's why the HGS deal is so important to CAT: if it has its own targets, it can keep more downstream value But such deals aren't easy for CAT to come by. In the meantime, CAT is leveraging the high-throughput nature of phage-display technology by joining forces with firms that want its antibody fragments for new sorts of tools, specifically protein chips. In December, the firm agreed to collaborate with Zyomyx Inc. of Hayward, CA to develop protein chips based on microarrays of antibodies [See Deal]. CAT struck a similar agreement with Oxford GlycoSciences PLC . just two months earlier [See Deal].

In the proprietary product game, Abgenix and Medarex are both much further ahead than CAT. Greer's plans for Abgenix call for it to develop 95% of its products on its own. "If anything we do with partners starts detracting us from our own work, that's an opportunity cost," he says.

Abgenix's proprietary strategy is becoming increasingly apparent in the way the firm is arranging access to targets. The firm has signed long-term, access-exchange alliances with companies that promise to be good sources of antigens, such as HGS, CuraGen Corp. , and Lexicon Genetics Inc. and Immunex Corp. , with whom it recently expanded its earlier product-specific relationship to gain access to a variety of oncology targets. "Immunex will bring targets and their preclinical animal models," Greer says. For its part, Abgenix will make antibodies and characterize them. The firms will then work jointly on manufacturing and commercializing the best candidates. Abgenix has also been developing a target-source network among academics, Greer notes: several dozen researchers have been given access to XenoMouse in exchange for access to their work.

The company has been supplementing its broad target-access relationships with deals for specific, validated targets. The firm's most important such deal is its 50-50 arrangement with Immunex on a Phase I antibody targeting the EGF receptor [See Deal]. With its own chimeric antibody, now in phase III trials, ImClone Systems Inc. has already largely validated antibody approaches to the receptor (the target also for programs from Pfizer and OSI Pharmaceuticals Inc. ), so the purely human Abgenix antibody is in fact a somewhat better bet than the usual Phase I project.

Abgenix's strategy is relatively new. Ready at the end of 1999 to license ABX-EGF to a large drug company in a relatively traditional milestone-and-royalty deal, Abgenix pulled the product back following a pair of successful financings which netted the company nearly $600 million [See Deal] [See Deal]. It suddenly had the money, it felt, to hold out for a 50-50 deal in which it would share the development and commercialization risks, responsibilities and profits.

The main reason Abgenix is still licensing basic technology rights, says Greer, is because it's so attractive from a financial perspective. Breeding doesn't cost much; Abgenix does little more than hand over the mice. Greer says he wouldn't mind offering services if Abgenix could charge a premium for them, or if it could get a higher royalty as a result of putting more into the discovery process. "But if you start adding services, the internal rate of return on these deals—which is now extremely high, because we put nothing into them—would start coming down," Greer points out. Besides, he thinks most firms wouldn't want to pay more than a minimal amount to access antibody-generating technology. "If they got their target from a genomics company, they're going to have to pay somebody for that antigen," he notes. Lowering the hurdle on the front end increases the odds that Abgenix will get a royalty later.

Companies seeking to use a XenoMousemust tell Abgenix exactly what they're working on, Greer notes. He says that stipulation has to do, in part, with the terms of his company's cross-license with Medarex: the firms agreed to carve out licenses, antigen by antigen. But the scrutiny serves another purpose, he notes: "We're also protecting our other partners, by making sure there's no duplication. We work on a first-come, first-served basis."

Greer says Abgenix will continue to evaluate potential partners carefully, before licensing technology rights and certainly before entering more committed relationships. He says, "We have walked away from deals before, if we haven't liked the money or the terms and we'll continue to do so. If you're serious about being in the game yourself, you don't want to be giving your key asset away too easily. If you let others use it with targets you might want to work on, you might possibly be enabling a competitor."

Heading for the Clinic

Abenix plans to immunize its XenoMousewith over 100 antigens per year—targets that have been screened and have passed muster—but the firm intends to put just two of the resulting antibodies into clinical trials per year. Greer thinks that's plenty, because the stringent culling will result in candidates of higher quality. He points out that the top six biotechs each put, on average, 1.6 products into the clinic per year over the past three years. "Amgen managed two. Our goal is two," Greer declares.

Don Drakeman has far greater clinical ambitions for Medarex. He plans for the company to put 10-12 antibodies into clinical trials per year—more than any biotech firm has ever attempted to date. "If Scott wants to do only one or two clinical projects a year, that's his business. But Amgen was living in a pre-genomic world. We're not. If you believe in the post-genomic premise that there will be potentially thousands of new antibody targets and at least hundreds of them, it's less important to harbor targets or target areas, than it will be to find a way to take advantage of all opportunities."

Medarex's development plans are daunting. If the cost of getting new targets is decreasing, the downstream costs of filing INDs, clinical trials, and attaining GMP standards are no less—and arguably greater—than they ever have been. Medarex's goals, says Scott Greer, "seem a bit of a stretch." Haseltine agrees: "They can't put 10 or 12 products into the clinic in a year. It's just too much effort, to file the INDs. Even four would be a lot."

Drakeman sticks with his goal. "We have very good partners, and we're generating good targets. We're going to work every bit as hard at picking the best products to put into the clinic as Abgenix or CAT is." By signing up multiple target-providing partners willing to share costs 50/50, Medarex is betting it can extend its reach.

For instance, Medarex won't do a lot of the work that goes into understanding a target and its role in disease: its partners will. "Once you've done that work, the business of developing an antibody and scaling it up is not a process that gets better with time," Drakeman declares, emphasizing, "We think the important thing is moving into humans. Not rushing headlong—but using a development system that's replicable and efficient, so you don't lose time on things you don't have to. Our T-12 development process for moving from validated target to testable drug candidate isan assembly line, standardizing those things that can be."

Whereas Scott Greer wants to avoid the opportunity cost of being distracted from its proprietary work by partners, from Drakeman's perspective, the opportunity cost of not doing a lot of trials quickly would be objectionable. Since no one firm has a monopoly on a good way to generate targets, he wants Medarex to partner with a significant number of companies that claim they can do it well. By structuring deals so that partners put their own money on the table, Drakeman thinks he helps assure that Medarex becomes involved with motivated firms. He admits there's some risk in doing 50/50 deals with people and firms that "haven't done product development work in their present incarnation." But some of Medarex's small-company partners have done it before, including Regeneron Pharmaceuticals Inc. , Oxford GlycoSciences, and Eos Biotechnology Inc. , which is run by David Martin, PhD, who was previously head of research at Genentech.

To Drakeman, the math works out. He thinks Medarex can spend one to two years, and $2-3 million, to go from a validated target to a preclinical study that produces data viable for filing an IND. Getting to Phase IIa and efficacy data from treating 30-50 patients will cost another $1-2 million. "We can get data with which we can make decisions for $4-5 million, and my share of that cost is 50%," argues Drakeman. "We can do that ten times a year without breaking the bank—especially if the other part of the business is generating cash, through milestones and royalties."

The other part of the business is technology licensing and the interest on the $400 million cash in the bank. "That kitty of dollars will allow Medarex to take some products into Phase III," Drakeman declares. He points out that cancer trials don't have to be huge, to get the job done. Genentech's Herceptinwas approved as a treatment for breast cancer on the basis of data gathered from less than 500 patients, he says. Similarly, he notes that Idec's Rituxan was tested in less than 200 patients, as was Taxol, a big-selling non-antibody cancer drug.

"We can parallel process so many antibodies at once, that we don't have to fall in love with any one in particular. If it doesn't look good to us, we don't have to go ahead. We can be thoughtful about the products we put into Phase III," Drakeman declares. Medarex will pay to do a large number of trials in parallel, as fast as possible, because he thinks it's worth the cash to get the speed and efficiency. As the data comes in, Medarex will decide over the next few years whether to opt mostly for contract development work, or to do it in-house.

Drakeman has been establishing a geographically widespread organizational structure for Medarex to match his broad vision for his technology platform. Nils Lonberg, PhD, the inventor of the HuMAb-Mousetechnology, works at Medarex's West Coast subsidiary, GenPharm International Inc. Drakeman himself and other key staffers including Yashwant Deo, PhD, SVP operations, R&D and regulatory compliance, are on the East Coast of the US, in Annandale, NJ. Medarex also has a presence in Asia, via its partner Kirin.

To get into Europe, Medarex effectively spun out its technology by licensing all rights to it to its new Copenhagen, Denmark Genmab affiliate, taking 50% of the founding stock and thus finding a new way of monetizing its technology assets. (Its share in the firm, with a cost basis of $18 million, is now worth $200 million, says Drakeman.) Genmab, run by Lisa Drakeman, PhD, Don's wife and the former head of business development for Medarex, went public on October 11th, 2000—20 months after its founding—on the Copenhagen Stock Exchange and the Neuer Markt of the Frankfurt Stock Exchange, valued at about $666 million. The IPO [See Deal]of six million shares raised 209.34 million Euros, or $183.2 million—making it the biggest European biotech IPO to date.

Genmab is also paying off from an operational standpoint, Drakeman says, because it enables Medarex to work with more partners, more effectively. Collaborations with target-validating companies are close scientific bonding opportunities, he notes, adding that, "One of the attractions of having Genmab in Europe is that it gives Europeans a chance to use our technology and work together with us in their own time zone, to create novel products." In early December, Genmab struck a multi-target alliance with Gemini Genomics PLC of Cambridge, UK [See Deal]. The British company headed by Paul Kelly, MD, pursues disease-related genes by integrating clinical and genetic information from various population groups, including twins.

But Genmab also increases Medarex's development capacity. Typically, Medarex will get half the US rights to any products developed through Genmab's partnerships with European firms, for just a quarter of the development costs. Genmab will pay 25% of the development costs, for half of the European rights, while the target company pays 50% of the development costs for the other half of the worldwide rights.

The firm has just begun a Phase II trial for its first human antibody, HuMax-CD4, to treat patients with rheumatoid arthritis. It's meant to be administered in multiple doses in combination with methotrexate. The drug's target, the CD4 receptor, was in the public domain, "or patents would expire before we came to market," says Lisa Drakeman. Genmab will bring two additional antibodies into the clinic in 2001, one of them an antibody against IL-15, a cytokine apparently involved in a variety of inflammatory conditions. Genmab sub-licensed rights to do the development work from Medarex, which bought them from Immunex. The Danish company already has two other candidates in preclinical development, and will seek to make other antibodies to known targets even as it pursues more European partnerships for novel targets.

Competitive, But not Cutthroat

The wealth of potential targets for antibodies, old as well as new, isn't exactly reducing the competition between the companies holding key technology rights—but it may be keeping the competition from becoming a full-out cutthroat battle. Indeed, it seems likely that awareness of the therapeutic potential in antibodies will continue expanding the market for all the technologies that go into making them.

CAT's David Chiswell says he's finding that companies he wants to do deals with are "more receptive to doing deals for antibody-generating technology with someonethan they used to be, because they see how much potential these sorts of drugs have." He notes that deal structures which match exclusive rights to use a technology to a particular antigen, are good for all the technology holders. "The deals you do may block some people from coming to you, but they don't stop those people from going to someone else that will help them utilize their antigen. The transgenic mouse companies will win enough and we will win enough to become huge companies."

"The antibody opportunity is exceeding everyone's expectations," declares Scott Greer, adding, "Everybody wants to be working on them, because they're a very fast, efficient way to capitalize on the genomics revolution." Still, for all the plenty and promise, the top firms are under pressure—given their extraordinary market valuations—to ensure that target companies at least try their technologies. Somebody's technology is going to be used to generate antibody products against appealing antigens. Somebody is going to own the value in those drugs. Investors expect key technology holders to capture their share of this expanding market.