The Antidote: Inside the World of New Pharma (16 page)

“These people don’t know why they were being brought together—and then smack in the audience is this couple, sitting next to each other looking terrified, and I’m up there to say, ‘You’re all being let go,’ ” he recalls. “I had to take it on the chin for the company.”

As Vertex restructured its research division, Thomson, Moore, and those others who were there from the beginning “grew up fast,” Moore says. Thomson learned that Vertex, which prided itself on its humanity, under certain pressure could be as capricious and unscientific as any other company. Vertex thrived on its own exceptionalism, but he found nothing distinctively heartening about the way people who had given themselves to the company—“bled purple,” as others said of him—were being treated. Nor did he always bite his lip. “I was a boat rocker by that stage,” he says. “Boat rockers in the early days were what we wanted. But around this time, reminding people of a need for a conscience was not appreciated.”

Boger gave himself and Sato until the first week in November to decide which two programs Vertex would take forward on its own. That the company needed to advance two projects had become an article of faith and was a blunt admission both of the risk that neither one was likely to get approved and of the fact that they had no presumptive favorite. Vertex couldn’t afford three projects, yet choosing only one would be putting all its eggs in one preposterously risky basket. No other decision in the company’s history was likely to prove as fateful.

As usual, they set in motion a process to maximize their inputs. The central pillar of Boger’s social experiment was the decision-making process itself, which called for as many voices and as much data as possible, but in the end was the opposite of consensus; that is, Boger, reserving the final decision, would leave things indeterminate until all the information was in and he could make the most informed choice, even if it totally reversed his earlier statements and positions. It was this fluidity
that had allowed him, after writing into the company’s charter that Vertex wouldn’t work on HIV, to reverse himself when Murcko, Tung, and their colleagues convinced him that they had a conceptual advantage. “Success in drug development is usually tied to two or three people who are passionate about their opinion beyond explanation,” he noted at the time. “Vertex has a long history of ignoring my opinions.”

In May, just after the layoffs, Vertex senior management met off-site to discuss overall strategy. Joining their ranks was the company’s new chief scientific officer, Peter Mueller, who formerly ran R&D in North and South America and Japan for Boehringer Ingelheim, the German firm developing the HCV protease inhibitor that Vertex hoped to catch—and beat—in human trials.

Son of a Bavarian banker, Mueller, forty-eight, was a polymath, having been sent at age ten to study at a Benedictine monastery outside Munich, an intellectual “boot camp” where four out of five students failed to make it through. After receiving his doctorate from Albert Einstein University in Ulm, he joined its faculty as professor of theoretical organic chemistry and then migrated to pharmaceuticals via astrophysics. Mueller was relentlessly inquisitive, equaling Murcko in his hunger to innovate. He pushed people hard—himself harder. More to the point, he had vital experience in getting drugs across the finish line. Regularly dressed all in black, his curly blond hair greying but not thinning—with a jolly disposition in the lunchroom, an impatient severity in meetings, and always a rambling Bavarian syntax—Mueller was what Vertex most needed. He was a closer.

“Peter had more personal drug discovery and development experience than the rest of us put together,” according to Sato. “I hadn’t developed small-molecule drugs: HIV was the first one I was sitting at the table for. Josh didn’t have any drug development experience at Merck. John Alam was the closest because he had taken Avonex all the way, but it was a biologic. Peter had been in the business. He knew big drugs, he knew good base hits, he knew things that looked promising but ultimately were gonna die. So his experience and judgment at the table were very important.”

Two months later, senior management met again, this time with
program executives and departmental chiefs, to review the company’s portfolio and try to prioritize their opportunities. Their challenge, as described in a Harvard Business School case study, “was to compare drug candidates at different stages of development, with different technical properties and different potential therapeutic applications.” In other words, Boger and Sato initiated a scientific process in which fifty or so managers from across Vertex would develop criteria to determine which of the company’s molecules to gamble their future on. As part of their analysis, participants used “real option valuation” (ROV), a modeling tool that stacks up the costs and risks of a drug’s clinical evaluation against its estimated commercial value.

They focused on four types of risks:
target, mechanism, molecule
, and
market
. Boger believed the goal was to diversify the kinds of risks Vertex would encounter. If you had a choice, you didn’t want to place all your bets, say, on anti-inflammatories; or on unproven targets like p38 MAP kinase and ICE; or on molecules like VX-745 that might be neurotoxic, or, like VX-950, went straight to the liver. You wouldn’t choose two close follow-on compounds whose performance would be wide open to competitive drubbing, as Agenerase had been. You wanted to vary the type and number of land mines, anticipated and unanticipated, you would surely confront during a period of progressive investment stretching up to a decade into the future. “Companies tend to have biases in how they evaluate risks and which risks they are comfortable with,” he told the HBS researchers.

“Some companies systematically underestimate target risk, some underestimate molecule risk, and some underestimate market risk. And the interesting thing is that when you’re inside the company, you’re probably not even aware what your biases are. So, to protect ourselves from these hidden biases, we deliberately want to make sure that we’re taking different kinds of risks in our portfolio. By balancing our risks, we can avoid being blindsided ten years later.”

Four molecules emerged as front-runners. Two held reasonable promise as “oral Enbrels”: VX-765, a second ICE inhibitor chemically distinct from pralnacasan, a so-called fast follower into the huge rheumatoid arthritis and osteoarthritis sweepstakes; and VX-702, a
second-generation p38 inhibitor that didn’t cross the blood-brain barrier, now being tested in a Phase II-a pilot study against another inflammatory response afflicting almost two million people a year in the United States alone, acute coronary syndrome. A third compound, VX-148, was one of a number of inhibitors Vertex had designed against the enzyme called inosine 5'-monophosphate dehydrogenase (IMPDH). It was nearing the end of a midstage study in patients with moderate to severe psoriasis, a scabrous and painful skin disease. Blocking a validated immune system target, IMPDH inhibitors were also being tested in the treatment of multiple sclerosis and cancer, and Vertex’s first IMPDH inhibitor, called merimepodib, was in Phase II trials for treating hepatitis C.

The fourth portfolio candidate was VX-950, which because it cost so much more to make and was last in development, for the least certain of markets, lagged severely by every measure. Mueller’s first reaction to the molecule was negative, and he was initially skeptical that it could become a drug. “We kept going to these planning meetings where people would do the ROV, and hep C routinely came out at the bottom,” Sato says. “It was four standard deviations away from any program that was ever going to deliver any value to Vertex. Josh and I would go, ‘Wrong answer.’ The analytics kept saying that hep C was a disaster. There were a couple of meetings where poor Steve Lyons, who was program executive, I’m sure felt like Saint Sebastian on a bad day.”

Sato and Boger refused to give up on the molecule. So did Mueller, who without their support couldn’t kill the project even if he wished to. Mueller’s departure from Boehringer—and his familiarity with its protease inhibitor—restricted him from asserting himself either way at this point. The process lurched ahead throughout the summer and fall, even as he cautioned against putting too much stock in the company’s assumptions. “ROV models are more valid for late-stage development compounds, when you have a pretty good feeling of the potential market ahead in one or two years,” he advised. “Everything else is pure speculation. I’m not aware of any prediction for early-stage compounds even close to market outcomes.”

The deadline for a decision—the third-quarter earnings call with Wall Street analysts—loomed. Boger began to exhibit his own core
bias, a blend of romance and risk-reduction. As Thomson put it: “HCV is a profoundly important medical area where Vertex can make a difference. We have a locked-down target with a low biological risk.” What Boger had learned about drug development he had learned at Merck, which is that if a drug has the right concept, and there are no physical reasons why it can’t be scaled up—and, above all, there are no toxicology problems—all the rest is capital, execution, and competitive commitment. Sato admired how Ed Scolnick, hell-bent on getting to market at the same time as the other HIV inhibitors, persuaded chairman Raymond Gilmartin to build a $150 million manufacturing plant a year before Merck could expect FDA approval. Vertex would have to commit to as much effort or more
before
it ever made a profit—yet another strike against VX-950.

As Boger and Sato emailed back and forth on managing the portfolio, hepatitis C kept bobbing to the top of the discussion. “Nobody else was winning,” Sato recalls. “I would have felt differently if somebody had come up with a kick-ass molecule, but nobody had.”

Kwong and her group weren’t let go by the layoffs. Her midnight project for years had been to develop an HCV protease animal model, and she had hired an Indian virologist named Raj Kalkeri to work with her on it. They were trying to engineer a mouse that would produce active HCV protease specifically in the liver, so that Vertex could tell if its—and its competitors’—molecules were reaching and blocking the enzyme where it counted. They had stepped up the pace and were closing in on a solution before the cutbacks. Yet with human testing of merimepodib and Boehringer’s protease inhibitor overshadowing VX-950, management showed little enthusiasm. Kalkeri was laid off. “People were saying, ‘What’s the point?’ ” Kwong recalls.

Kwong and Kalkeri had come up with a novel strategy and, having nearly proved the concept, weren’t prepared to drop it. First they had fused the genes for HCV protease with those of another enzyme that they reasoned would be cleaved, broken down, and released into the bloodstream once the protease was activated. Then they inserted the genes for this fusion protein into a type of virus that, as Kwong knew
from earlier gene therapy studies, all went to the liver when injected into the tail veins of mice. Eventually they began to produce animals whose livers churned out both HCV protease and a toxic enzyme that could be measured in the blood as a marker of protease activity. Under the microscope, the mouse livers looked like a ravaged human liver from someone with late-stage hepatitis C.

Neither of them felt they could stop now. “Raj actually begged to come back, on his severance, to work on the proof of concept of this model,” Kwong recalls. “There are two things that are astonishing about this. Number one, that he did this. He came in day and night. We had to kill ourselves that summer. This is not a two-hour-a-day thing. This was really exhausting. Second thing is, Vertex let him do that. What company would allow somebody who was laid off back in the lab to work day and night? Only one that knows that, whether you’re inside the company or outside, you’re consumed by the project and won’t be stopped.”

Kalkeri finished the work in September, and Vertex began feeding mice VX-950 and Boehringer Ingelheim’s protease inhibitor at a range of doses. He had made a version of the mouse in which the toxic enzyme was expressed only if the HCV protease was active, not if it was blocked. Facing the portfolio decision, Boger was enthralled by the data. “What we could do was administer compounds to the mouse and actually see the liver not get damaged, because we were blocking the production of this toxic enzyme,” he says. “We could do whole-animal PK and actually get livers to look better. We’re getting molecules down the mouse’s gullet, into the body, through all the ways in which molecules can get trapped, and the molecule had to get to the liver and shut down HCV protease in the mouse liver, and if it did that, it saved the liver.”