Aventis's Global Management Challenge

Through the integration of Hoechst Marion Roussel and Rhone-Poulenc Rorer, Aventis is creating a truly global, inexorably interlocked network of drug discovery and development capabilities. The need to improve sales performance and increase capitalization--a defense against takeover--drove the merger. But the merger also gave it both more R&D resources and an opportunity to retool its process. As this question and answer session with Aventis's Frank Douglas, head of R&D, notes, the company is emphasizing the creation of capabilities through internal and external investments, rather than scale. The reconfiguration draws on Douglas's experience in HMR's previous, and far less successful mergers. Aventis has set clear project priorities and created interlocking R&D functions in the US, France, and Germany, with platform technology resources-wherever they're located-directed to priority projects. The result, Douglas believes, is a system that focuses on optimizing the global portfolio, not the site portfolio. The company's challenge now, and perhaps its biggest opportunity, comes from being a truly multinational organization. Aventis has core drug discovery and development sites in the US, France, and Germany, and has chosen to distribute decision-making equally among them. That facilitates outreach to academia and biotechs, but it's not the most efficient way to organize and requires extra time and work for communication.

Through the integration of HMR and RPR, management is creating a truly global, inexorably interlocked network of drug discovery and development capabilities.

by Mark L. Ratner

• The need to improve sales performance and increase capitalization—a defense against takeover—drove the Aventis merger of HMR and RPR.

• But the merger also gave it both more R&D resources and an opportunity to retool its process. As this interview with Frank Douglas, head of R&D, notes, the company is emphasizing the creation of capabilities through internal and external investments, rather than scale.

• The reconfiguration draws on Douglas's experience in HMR's previous, and far less successful, mergers. By clarifying the chain of command down several levels at the outset, unlike post-merger HMR, Aventis's senior R&D managers attempted to focus the organization externally, avoiding turf wars.

• To further the process, Aventis has set clear project priorities and created interlocking R&D functions in the US, France, and Germany, with platform technology resources—wherever they're located—directed to priority projects. The result, Douglas believes, is a system that focuses on optimizing the global portfolio, not the site portfolio.

Among the never-ending consolidations in the pharmaceutical industry, Aventis SA —formed a year ago through the merger of Hoechst Marion Roussel (HMR) and Rhone-Poulenc Rorer (RPR) [See Deal]—is notable for the number and frequency of its conjugations. HMR's acquisition lineage began with Richardson-Merrell, which merged with Dow's pharmaceutical business, and later with Marion Laboratories, forming Marion Merrell Dow (MMD). Hoechst bought MMD [See Deal] and then consolidated the portion of Roussel-Uclaf it didn't already own [See Deal]. On the other side of the aisle, Rorer in 1986 bought Revlon's ethical pharmaceutical business as part of a complex defense against a hostile takeover by CooperVision, merged with Rhone-Poulenc's pharmaceutical business in 1990, carried off Fisons PLC, which had been romancing Medeva PLC, in 1995 [See Deal], and eventually tied up with HMR.

With all the attendant furniture shuffling, redecoration, and rebudgeting this family had already gone through, much of it in the last five years, industry watchers were understandably leery about Aventis's potential before an extended settling-in period and a demonstration of achievement. At the start, Aventis was dogged by worries that management would repeat the HMR experience, which was viewed as a merger of weak sisters that failed in part because of an inability to manage cross-cultural issues. HMR also made aggressive statements about meeting specific margin targets, which it failed to do, and had to weather the Seldaneexperience: the drug was pulled from the market in 1998 due to cardiac toxicity.

By contrast, Aventis's management now talks in terms of net cost savings and net reinvestment in R&D, not margin goals. And its three top drugs are adding handsomely to the top line, beating expectations. HMR successfully converted Seldane(terfenadine) to its metabolite form, Allegra (fexofenadine), which will top €1 billion in 2000 sales. Revenues from low-molecular-weight heparin, Lovenox (enoxaparin sodium), were €782 million in 1999 and may touch €1 billion this year, and the cancer drug Taxotere (docetaxel), which hit €500 million last year, is expected to reach €800 million in 2000. Plus, the impact on sales of the patent expiration of Cardizem (diltiazem), a hypertension and angina drug, is now behind it. Aventis also met its internal approval goals for 2000, registering all but one drug it had slated for market introduction—its compound for unstable angina, Refludan, which the company withdrew after a split negative decision in the US FDA (see Exhibit 1).

As a result, in the year since the merger was approved, Aventis has gained credibility as it has begun to show that the execution risk is lower than previously expected. More comfortable with its ability to manage cross-cultural issues, sell-side analysts, who initially didn't factor into their models the full projected cost savings from consolidation—lower headquarter costs, R&D synergies, and the separation of the agri-business—are doing so now. As perceptions about Aventis evolved, and partly because the initial expectations were low, it has become one of the best-performing stocks in the European drug sector.

The company's challenge now, and perhaps its biggest opportunity, comes from being a truly multinational organization. Aventis has core drug discovery and development sites in the US, France, and Germany, and has chosen to distribute decision-making equally among them. That facilitates outreach to academia and biotechs, but it's not the most efficient way to organize and requires extra time and work for communication. Its choice to headquarter in Strasbourg also stemmed less from strategic considerations than the desire to avoid having to make a choice between Paris and Frankfurt.

The decision to centralize US operations in New Jersey, closing operations in Kansas City (HMR's site), and Collegeville, PA (RPR's site), has also led to a brain drain. Many of the organization's most talented people decided against making the move.

Clearly, moving people in and out is much easier in the US than in France and Germany, where workforce regulations make plant closings and layoffs difficult (see "Hoechst Hurts,"IN VIVO, April 1998 [A#1998800072). Aventis did close an RPR site in the UK, but chose to hold onto its Romainville, France, facility after a deal to sell it to EI DuPont de Nemours & Co. Inc. fell through [See Deal]. Even now, wary of union issues in Germany and France, the company hasn't revealed its plans for plant consolidation and closings. Aventis is actively recruiting in the US whereas, if anything, it has too many people in France and Germany, and the less efficient business climate in those countries still makes it harder to obtain some of the synergies expected in a merger of large companies. Nonetheless, by all accounts, its management of the process has exceeded expectations.

In the Q&A session that follows, the head of Drug Innovation and Approval for Aventis, Frank Douglas, MD, PhD, sheds light on the global integration process under way at Aventis from the R&D perspective, including the lessons he's learned from prior merger experiences at Marion, MMD, and HMR.

Douglas's most innovative concept is the establishment of interlocking R&D functions among Aventis's drug discovery sites. No discovery research site is given the full complement of drug discovery tools—nor could it be, as a practical matter, given the increasing specialization and proliferation of toolkit technologies. Each one shares resources with the others, depending on a project's overall priority within the global organization and where and when a specific capability is needed.

The role of the site heads is to manage a portfolio of projects through Phase IIa. A unit at a site preferentially supports the projects at that site, but they compete on a global basis for resources from global functions—Lead Generation, Lead Optimization, and Product Realization, plus regulatory and marketing support—that are distributed among the sites and supply the best people or expertise, and the best technical solutions, into the teams. The result, Douglas believes, is a communal system that focuses on optimizing the global portfolio, as opposed to optimizing a particular site portfolio. Indeed, Douglas emphasizes that the way to retool research is to focus on maximizing capabilities globally, with a combination of internal resources and external collaborations, as opposed to building new technology capabilities in-house at commercial scale.

Emblematic of this approach is Aventis's June 2000 deal with Millennium Pharmaceuticals Inc. [See Deal]. The $450 million agreement has four distinct components: a joint development collaboration for drugs to treat inflammatory disorders (rheumatoid arthritis, asthma and chronic obstructive pulmonary diseases, multiple sclerosis, and inflammatory bowel disease); a joint technology development agreement; a technology transfer agreement, for which Millennium will receive up to $40 million per year for five years; and a purchase by Aventis of $250 million of equity ($150 million at deal signing and $50 million in January and July 2001).

The partners expect to build nothing less than the world's premier franchise in inflammation, which Douglas expects will yield product candidates to fill his pipeline gaps and, as importantly, will give Aventis a better grasp of how to integrate technologies in-house that will improve productivity, in particular those relevant to target identification and validation.

Q:What were the goals in bringing HMR and RPR together to create Aventis?

Douglas:First, to improve sales performance. Also, in this time of continuing mergers and acquisitions, both companies found themselves with relatively low capitalizations and therefore were potential prey.

So we had similar objectives and were in similar situations. Both were European-based companies with global outlooks and visions of becoming pure life-sciences companies. We needed critical mass across the value chain. For example, there were things we wanted to do in discovery. We both wanted to move into the area of new technologies, and the challenge for both of us was finding a way to do that rapidly.

Getting bigger helps because the best way to gauge your spend in R&D is to look at it as a percentage of sales. The higher your sales number, the more you can put into R&D. We will continue to spend 16-17% of sales in R&D. That's the number we think we have to be at to stay competitive.

From an R&D perspective, which we call Drug Innovation and Approval, RPR and HMR were pretty well aligned, philosophically. I'll always remember that, at the first meeting between François Meyer, who was then the head of RPR research, and myself, I was surprised at the similarities in our points of view. First and foremost, we both believed that the issue in R&D is productivity. And that to improve productivity you needed to have an integrated technological base in discovery. The issue is less how much you spend in any particular technology than how well you integrate technologies to get productivity.

The second thing we had in common—again, it was a productivity theme—was on the clinical side: the need to find ways to identify your best candidates in Phase I and Phase IIa so that you can go fast when you get into Phase IIb and Phase III, with a very low attrition rate. And we both were pursuing strategies to do that.

Q:Did you anticipate any major stumbling blocks?

Douglas:The only significant issue we had to face was where we would put the major discovery sites.

We decided not to discuss this up front. Instead, with the help of an external panel, we looked at the projects we were doing in each of the disease groups, and within each disease group we ranked projects high, medium, or low, and agreed that we would stop doing the low projects. This was done prior to the merger, with all the appropriate safeguards for competition, of course.

We had very clear criteria for evaluating a project, including where it stood with respect to the competition, patent issues (how long the clock had been running), and its progress from the last milestone to now.

Q:Were these criteria applied by the people working on the project?

Douglas:We gave the criteria to each of the disease group heads and they evaluated and ranked their projects internally. They brought those judgments to further meetings with the external experts and some of our key people who also had information on which to evaluate these projects. So right off the bat, we were able to focus our discussions on the differences we had between their assessments and the internal assessments.

A site head would show how he'd ranked his portfolio. Presumably, he and his people had already made the tough decisions. And if the difference between medium and low was perhaps not so great, we'd defer a decision until we understood that project better, when we could really get into it.

Q:What about disease areas as opposed to specific drug projects? Did you decide to stop research in any of those?

Douglas:The only disease area we decided to cut was bone. Otherwise, we decided to maintain all the disease areas, but to reconfigure—recluster—some of them because of where we were going to site them.

Now, once we had gone through the portfolio ranking process, we turned to our external experts and asked them: Given this, how would you put together a discovery program? They made a proposal, we discussed it, and out of that came the disease groups in which we should continue to focus. Then we went the next step and did an analysis to determine where best to site these disease groups, regardless of where they happened to be at the moment. Then we developed criteria: making sure we had critical mass at the site, making sure that there would be good outreach to academia and biotech companies in that country, and that there were scientific synergies within the disease groups—particularly when you think of common mechanisms of disease that you could find by clustering these disease groups together.

Indeed, out of that process came our decision to develop what we call an immunology platform, since we found that we were working in many diseases that had an immunological basis. And with that focus on common mechanisms of disease, we said it made sense to have a core group that would develop and become expert in immunological toolkit technologies, to help us identify targets and validate those targets that in turn would support different disease groups. This core group supports immunology across the board—asthma, oncology, rheumatology—regardless of site.

Q:What were the site decisions?

Douglas:In the US, we moved from Collegeville, Pennsylvania, and concentrated in Bridgewater, New Jersey. In Europe, we decided to move out of our Romainville facility in France and concentrate in Vitry, Croix de Berny, which is the development area, and Frankfurt—our three Strategic Discovery Sites. We tried to sell the Romainville site, but that didn't work out, and so we are now integrating it. We also have here in Bridgewater a Global Drug Development Center, where we manage development from Phase IIb through life-cycle management. And because Japan has special needs, we have a significant development effort there, but no research.

Q:Is there a firm cutoff when projects are handed off from discovery to development, say at Phase IIb?

Douglas:We changed the organization in that respect. We do not have a traditional break between research and development, which is why we call it Drug Innovation and Approval. Within DI&A, we divide the organization into Lead Generation, Lead Optimization, Product Realization, and Global Regulatory Approvals and Marketing Support (GRAMS). In Lead Generation we've clustered all of those disciplines that have to do with identifying a target, validating a target, and finding the first lead compound: genomics, bioinformatics, high-throughput screening, chemistry—the platform technologies. This organizational strategy also helps us break down the cycles and better analyze how to speed cycle time.

We came at this, incidentally, by simply saying: Here's the value chain, here are the things we do from conceiving an idea, understanding a disease, finding a target, validating a target, finding a hit, optimizing a hit, Phase I-II, etc. These are the things we do, so why not organize to support that?

We also asked ourselves what the levers were that improve productivity. For example, the ability to identify and validate a target, and find a lead that is druggable, very fast—that's a lever. Also, because in our view, a compound is not really optimized until you've seen it in a disease state, another lever is the ability to optimize that lead and take it through to Phase IIa—into man. So Lead Optimization has within it the ADME group, toxicology, and Phase I-Phase IIa. It's a bridge between early discovery and Phase IIb-III—late stage, which we call Product Realization.

Q:This was true at HMR as well, yes?

Douglas:Yes, we introduced it in October 1997 at HMR. But it's evolved. We said up front when we introduced this in HMR that it would be refined as we go forward and understand it better. I think we are much better, now, at organizing our discovery groups and supporting them to improve productivity.

We have multidisciplinary teams—we call them units of innovation—distributed among the sites. Supporting them are the global functions: Lead Generation, Lead Optimization, and Product Realization, on the discovery side, plus GRAMS. The role of these global functions is to supply the best people or expertise, and the best technical solutions, into the teams.

The third part of this matrix is the sites—Frankfurt, Bridgewater, and Paris [Vitry and Romainville]. The role of the site heads is to manage a portfolio of projects through Phase IIa, and they compete on a global basis for resources from the global functions. The only thing the site heads have under their direct control is the biology—the pharmacologists, microbiologists, biochemists. The rest of the capabilities are in the units, and the units report to global functions.

Q:How do the units compete?

Douglas:We make it relatively easy. We say that a unit at a site preferentially supports the projects at that site. First principle. But at the beginning of the year, we look at the projects at each site and we prioritize them—high, medium, low. Second principle. So the global head, when he is in a resource crunch, has the right, based on the prioritization of the projects, to move support of a project from one site to another.

Let's say that Bridgewater has more high-priority projects than Frankfurt and it needs more chemists. The global head of chemistry may assign chemists in Frankfurt to support the project in Bridgewater regardless of what's going on Frankfurt—assuming, of course, that Frankfurt does not have enough high-priority projects to use all of those chemists. Or, when a project dies, the global functions have the right to determine how to use those freed-up resources.

If a site head controlled everything at that site, it would be difficult for him to give up resources. But because a site head only controls the biologists—the core, in the early discovery part, of the project teams—he's got to apply to a global person whose job is to optimize the global portfolio. The result is that we focus on optimizing the global portfolio, as opposed to optimizing the site portfolio.

Q:How has this worked, so far?

Douglas:The company has a model—we call it the New York model—that we developed in 1999, to get us to our goal of two or three NDAs per year. It tells us how many compounds we need to put into Phase IIb and Phase III, how many into Phase I and Phase IIa, and all the way back to the number of new targets we need. And through the New York model we have identified where we had gaps: we had a gap in Phase I-IIa, for example, which we filled in a year. And we've closed other gaps with respect to our early development candidates. And we've done it not by optimizing the projects in each of the sites, but by focusing resources on the most promising projects throughout the organization.

Q:Do you have a sense that the sites that are less productive are becoming more productive based on the competition?

Douglas:They are. And they understand their role as part of the global whole. In fact, when one site head looked at the productivity of the three sites, he said directly to his people that his site would lag in terms of relative productivity improvement this year. I really raise my hat to him. Sites weren't accustomed to comparing productivity one to the other and so the lag was not obvious to the organization. It was a bit of a cultural shock for them, and from that standpoint it took courage to acknowledge the lag openly.

Also keep in mind that all these improvements came in year 2000, because we really got going only in January. We had the benefit, of course, of having a lot of discussion and doing a lot of preparation in the time before the actual merger was completed. But we couldn't implement anything because we were still two separate companies and had to be run separately.

But we hit the ground running. In fact, we basically said in March that integration is over. We had planned enough in advance to be able to focus on implementation and forget about integration. When I compare this with integrating HMR, this went very smoothly. HMR was a constant battle for two or three years.

Q:Based on the experience at HMR and RPR, had you flagged any specific integration issues in the 1999 timeframe that you knew you'd have to address?

Douglas:We sat the two teams together—I'm talking at the CEO level—and talked about what lessons were learned in RPR and HMR, and Marion Merrell Dow before that.

We wanted, right off the bat, to identify who the leaders would be down three, four levels. We'd done that by the end of February 2000: people knew who was going to head what.

And we started thinking about that very early on. Back in 1999, the leaders of the executive committee were known and announced. By March of that year we were already informally working as a designated executive committee. So in January 2000, we had all the processes in place for each of the departments to go forward and actually select and announce very rapidly the next two, three levels.

Q:You decided to close Romainville, and then you decided to keep it. Was that politics or strategy?

Douglas:For a while, we had an exclusive agreement to sell the facility to DuPont. And after that deal fell apart, three or four companies expressed an interest in Romainville, including Pierre Fabre, who pushed very hard. But they all wanted something—co-promotion rights or products—to cover the burn rate for the facility. We agreed, in principle, since they were going to take on our employees. But when we got down to the details, and looked at the potential dilution to our operating profit, it made better sense to integrate Romainville rather than to sell it.

And we kept the value of the projects we had there. For example, Ketek(telithromycin), our major antibiotic project, is supported in Romainville, where it was discovered. The hormone replacement therapy for osteoporosis, Ondeva (trimegestone), is another compound out of Romainville: we've obtained marketing approval in Sweden and have just done a deal with Wyeth-Ayerst Laboratories [a division of American Home Products Corp. ] [See Deal] for the drug.

Q:You did a similar thing with Quintiles Transnational Corp. —selling your Kansas City development operations to them, though guaranteeing them significant supporting business while keeping your own projects on track [See Deal] (see "Outsourcing: Hoechst's New Approach to Cost Cutting," IN VIVO, February 1999 [A#1999800032). Did that work?

Douglas:Yes—although the drug, 907, did not make it to market. The joint HMR/Quintiles team had data showing that 907, which is a 5HT2a receptor antagonist, works in schizophrenia. But it was nowhere near as good as Ortho's Haldol [haloperidol, from Ortho-McNeil Pharmaceutical Corp., a division of Johnson & Johnson ]. Therefore the drug failed. But it was a success because in a very short timeframe we were able to get that answer and reduce the amount of money we would otherwise have spent in Phase III.

Q:So it validated the model.

Douglas:Yes, both in terms of a fast no-go decision and the model with Quintiles. This is an important point. We want people to focus on decisions, whether to go or no-go, making them fast, with quality data, so when everybody sees the data it's clear what the decision should be and nothing lingers. Nobody has revisited 907; in fact, closing down the clinical sites went extremely well because the investigators who saw the data understood that continuing made no sense. People weren't coming back and saying "let's do one more study using a different design."

Q:Are you still studying that receptor?

Douglas:Yes, in a couple of other indications, and in a couple of Phase IIa studies in CNS. Gastrointestinal disorder, such as irritable bowel syndrome, where 5HT antagonists have also been shown to play a role, is not a disease area for us. Because it's a large market and we have a good compound, however, I'm sure that issue will come up at our clinical compound portfolio review in March.

Q:In 1998, you co-authored an article for IN VIVO in which you broke down drug companies into two archetypes: scale-based and capabilities-based (see "How to Spend a Billion Dollars," IN VIVO, July/August 1998 [A#1998800141). Have your views changed since then?

Douglas:I'm still a capabilities-based man—in fact, even more so. We talk now about being a network-centric organization. Internally, the discussion is about the value net, rather than the value chain.

Glaxo is the model for the scale-based business. Analysts used to beat me over the head and say, "Look. Glaxo has this big high-throughput screening machine and you guys at HMR are no place." And I used to shrug my shoulders and say, "So what? They can screen 200,000 compounds in a day and produce a ton of data. But have they shown that it's going to help them find a drug?" Because the bottleneck is finding a lead compound that is truly druggable.

We spend a lot of time at Aventis in the chemical space. We make all sorts of libraries, we look at the diversity of compounds, and we're now building in ADME properties as part of the criteria for selecting compounds. But guess what? The biology determines which compound is likely to work. It's the intersection of the biology and the chemical space that will really improve your productivity. And as a result, you begin to realize very quickly that the role of a large pharma company is becoming that of an integrator of knowledge.

No one company is going to have—at scale—all of the capabilities you need. For example, no large company is going to have at scale the ability to rapidly do crystallization. However, a lot of small companies are now beginning to do it. And so our notion—again, it's capabilities-based—is you have to have within your organization core groups that understand the technology, and you have to create networks that work together around a particular project. Some of the networks will be internal and some of them will be external. But either way, at some level, you must have virtual organizations.

So, to me, the question is: What are the core technology platforms we need to have in-house to be very good integrators?

Q:What core technology platforms do you need?

Douglas:For us it's very clear. We need to be able to take advantage of the information coming out of genomics; we have to be able to identify and validate targets. We need some abilities in genetics, because that helps you to understand polymorphisms. We need to have chemi-informatics. And we need to be able to integrate the bioinformatics information and the chemi-informatics information.

But that doesn't mean we need to have all the pieces; we need to have some proteomics in-house, but we don't have to be at scale in mass spec and in 2D gels. What we do need is the capability to successfully manage collaborations and alliances with companies focused in those capabilities. Similarly, we have within our chemistry group a crystallography group, but we don't have to be getting into high-throughput crystallography. We need the talent, the expertise, to collaborate with the high-throughput crystallography guys.

The difference now is that we see capabilities as dynamic; two years ago, when we wrote the article, we saw capabilities in a more static way.

Q:Would this change if the industry became more consolidated—if Glaxo and SmithKline come together, or if Bristol-Myers Squibb does a deal? These gigantic companies, almost by their nature, have to become scale-based in your terms. Does that begin to change your thoughts when someone can spend $4 billion on R&D?

Douglas:If those companies can trim down their infrastructure and really become productive, they'll be able to do R&D better and faster than I'll be able to do it. I do believe the guys who are getting bigger will get better. Not because they have more stuff in-house, but because if they manage their resources well, they'll be able to leverage more outside.

Q:Does that imply you'll need to do another deal, yourself?

Douglas:Let's face the reality: when HMR was formed we were number three. But with other mergers and the like, we quickly dropped to number nine or so.

Q:But you don't need to be in the top five or six for its own sake, you need to because of what you're arguing, that if you learn how to leverage all of this stuff you can skinny down your infrastructure and you can be a lot more productive?

Douglas:Yes. Because we'll be bigger, our 16-17% R&D spend will take us farther.

Q:Do you set parameters for spending those R&D dollars in terms of the portion devoted to internal R&D versus in-licensing?

Douglas:On the early part of the value chain, the discovery side, you've got to commit to X dollars for X timeframe. You can't go in and out of programs. That's because you have to have the expertise in-house to really know if the compound you're getting is worth anything. So if we're not in an area, we wouldn't in-license anything. And that decision would come from the sites, the disease groups.

On the other hand, on the development side, you have a base, but, apart from that, your spend should go up or down as you address specific opportunities. If we were confronted tomorrow with a Lipitor, for example, we should not, as a company, say DI&A doesn't have the budget to go after something like that and develop it. So we are opportunistic at the late stage, and adopt a targeted approach at the early stage.

Q:What about out-licensing? You've been fairly aggressive out-licensors.

Douglas:Because we have a lot of compounds in the early phase, we get many requests for early-phase out-licensing. Our view is that it is better to give some of the early compounds out and pay a premium to get them back in after Phase IIa. The problem we're finding, though, is that too many companies also want financial support. That requirement defeats the purpose; I may as well do it myself.

That's why we haven't done more in the early phases: we did do a deal with Inflazyme Pharmaceuticals Inc. in asthma in which we bundled, with their product, some products we had, and that's been working pretty well [See Deal]. But by and large the reluctance of a licensee to finance development becomes the barrier.

Q: But out-licensing has its downside. You had an embarrassing situation withAltace, which King Pharmaceuticals Inc. licensed from HMR [See Deal].

Douglas:I'd distinguish Altace from other out-licensing opportunities because it was a product in the marketplace. Try as we could, sales just weren't moving. King could do things with pricing that we couldn't. Plus, something unexpected happened [the results of a five-year study on the drug showing that it prevented heart-related deaths]. In retrospect, what we should have done, since we had the trial ongoing, was waited for the results.

We had another embarrassing situation with iloperidone, a Phase II compound we'd out-licensed to a small company [Titan Pharmaceuticals Inc. ] [See Deal], which then out-licensed it to Novartis AG [See Deal]. We were pretty sure iloperidone would work, but we were betting on a different compound—907, the 5HT2a-RA we were developing with Quintiles. We felt that if that drug hit, it would hit big. As it turned out, we out-licensed the wrong compound.

Q:This gets back to the issue of whether, in Phase II, you can really see whether a product is going to work.

Douglas:That's true. But when I went back and looked at the Phase II data on 907, there were some hints that it would not be able to treat some negative symptoms. So I changed the program and we did a very large study to answer that question, which we should have answered in Phase IIa with a smaller study. But I was already geared up for Phase III, the centers were already set up, and I would have had to get the investigators to agree to run a slightly different trial.

Q:The two most important deals in 2000 may be your deal with Millennium and Novartis's deal with Vertex Pharmaceuticals Inc. [See Deal]. They're structurally equivalent except for two basic things: Millennium is teaching you, via the inflammation program, exactly how to transfer the technology; that is to say, they're only going to get 50% of the product if the product succeeds, and the product will only happen if you are part of that team. And Novartis/Vertex is a high royalty deal: Novartis isn't getting any of Vertex's chemogenomics, all it really cares about is getting Phase IIb-ready compounds.

Douglas:Our technology gap analysis led to the Millennium deal. We looked at several companies and asked which one could best help us fill that gap.

But I am also concerned with developing organizational capability. It was for me pointless to do a deal with a biotech company if I didn't improve the capability within our company. The Millennium deal with Bayer AG , for example, gives Bayer hundreds of targets, but Bayer isn't getting any capability [See Deal]. Now, Bayer did that deal because they didn't have much biotechnology inside. Great concept—they stand to make a lot of money on the stock exchange [Bayer paid $96.6 million for 4.96 million shares at $19.48 per share prior to a 4-1 stock split, and its investment is now worth roughly $1 billion]—but they didn't get the benefit of developing the organizational capability.

Millennium and Aventis have something in common: we both feel passionately that the issue in R&D is productivity, how you industrialize parts of discovery to make the whole process more productive. We can help each other accomplish this. Millennium has stated they want to become the biopharmaceutical company of the future, and being a pharmaceutical company—integrating the chemistry and so on—is something we know how to do. On the other hand, Millennium knows how to integrate the toolkit technologies. So we are enabling each other to develop capabilities.

The only way to do that successfully is to become interdependent and interlocking: that's why we settled on the inflammation project. This assures me that Millennium's technology will be transferred. This assures Millennium that they are going to get access to the capabilities they want. There's also a technology development component to the deal, so that, particularly in the area of ADME, we can learn how to utilize the genomics capabilities, which we have already started doing with toxicogenomics, for example. So that we can now operate, hopefully at industrial scale, together with Millennium.

Q:How are you distributing the Millennium technology within Aventis?

Douglas:We currently have a tunnel approach to discovery: we select a number of targets and run them through the tunnel of identification, validation, etc. Millennium enables us to change to a funnel approach, in that we are going to be able to generate lots and lots of targets, and then select the ones for validation. We're improving our choices for targets, and in this game it's all about choices. If you have more opportunities than you can handle, and you've got paradigms that enable you to select the best ones, then you're better off.

So we've decided that, in Cambridge, where we have our genomics center, we are going to develop a global data center and a gene discovery center, with expression, DNA arrays, that sort of thing. In terms of target validation, our concept, which we're now working through, is to have two sites, one in the US and one in Europe, in which we cluster all the technologies that will focus on validation, including pathology and transgenics. And they will support the global organization.

I've explained that we have global groups that can serve all three sites. And part of what we want to do is to make sure that in this Millennium/Aventis discovery chain we create, we have a specialist factory on each of the three discovery sites: target identification in the US, target validation in one site in Europe, high-throughput screening in another site in Europe. But no one site has all the pieces. Again, the sites will have no choice but to coordinate to optimize the global portfolio.

Q:What would your criteria be for buying a company rather than collaborating?

Douglas:The best model is still Roche/Genentech. You purchase a certain percentage of a company, depending on whether you want to have control or not, and you let it be. You agree on the projects, the quality that you need, milestones, you transfer technology, but you don't fiddle with the culture. That's because in biotech companies there are lots of smart people but usually it's three or four guys who really fire up that organization. So you've got to be careful you don't end up with a shell and the brains have left.

Q:Except for some small parts, like Gencell, your venture into gene therapy, Aventis is really a small-molecule company. Do you share pharma's prejudice against large molecules?

Douglas:We've looked really hard at this—should we do something in the area of large molecules? Antibody products are the most attractive area. With humanization technology, antibodies have really demonstrated their worth. What we've found with other proteins, however—excluding the replacement hormones, which have been largely done—is that they have the same problems as small molecules. Demonstrating safety and efficacy is not easy. And people still underestimate the manufacturing issues.

My view is if we want to go into that area, we can always buy something, but if we say we want to get into it, we're looking at a significant change in our personnel and expertise.

Q:Theoretically, you could go out and buy a protein biotech company.

Douglas:Yes. It would have to be profitable in its own right, or the pipeline would have to be advanced enough to convince us of the likelihood it is going to get products out. And I would keep it at arms' length. This is not a simple business. It's not something you can say tomorrow—and this is what I get asked—"let's get into large molecules." How do we do that?

I don't think we can convert Aventis into a large-molecule player. I'd probably make a strategic alliance because I do believe that these are going to have some degree of success.

Q:You've done MMD, you've done HMR, now Aventis. Any merger involves a trade-off, between slowing development of late-stage products and building up these other advantages—getting critical mass in discovery and improving your ability to move faster with late-stage products. With the Aventis merger, some product development slowed. Were these trade-offs worth it, particularly when you compare what happened at Aventis with what happened when Hoechst bought Marion Merrell Dow and when Marion and Merrell Dow merged?

Douglas:Everybody is expecting these two companies [HMR and RPR] not to succeed. How will they judge that? By whether we can deliver on late-stage projects and whether we can deliver on time. They will judge it on promise: whether we've got things coming into the Phase I-IIa area. And they will judge it on whether we have the technology platform that says we will continue to generate leads and move new products into the market.