Company Rushed
Crixivan To Market
Now It Faces Production
Bottleneck
----
Watching the Patient-Count

Merck & Co. has spent more than $1 billion over the past decade to create Crixivan, one of the most promising drugs ever to attack AIDS. It has paid off.

The drug won the fastest federal approval in regulatory history, is helping to restore health and hope to thousands of patients, and is soon expected to bring Merck a half-billion dollars in yearly revenue. After seven months on the market, it is outselling other new drugs in its class.

There is just one problem: The company has been struggling for months to keep ahead of demand.

One of the most closely watched numbers inside Merck these days is the weekly Crixivan patient count. About 90,000 patients world-wide are already on the drug, but there is only enough supply for about 110,000. Demand could increase sharply as more states approve funding for the expensive therapy and more countries allow its sale.

As a result, Merck has been in a breathless race to rev up production and avoid a public-relations disaster: the prospect of running short of supply and having to turn away AIDS sufferers.

How did one of the world's most powerful drug makers get into such a predicament?

Merck created some of its own problems. In the face of a world-wide abundance of production capacity, it chose to build its own plants rather than contract out more work to outside suppliers. And, upon learning that two rivals might beat it to market with similar therapies, Merck requested and got approval of Crixivan -- the most complex drug it had ever tried to mass-produce -- well before its plants were ready.

Much of the production crunch, however, is an unavoidable pitfall of making ever-more-complex medicines. Merck's struggles illustrate the obstacles other companies will face as they pursue new AIDS therapies.

Competition to develop a new class of AIDS drugs began in 1989, when Merck published its discovery of the three-dimensional structure of the enzyme protease, a promising target for attacking the human immunodeficiency virus that causes AIDS. Protease acts like a scissors, cutting up a key HIV protein and enabling the virus to replicate.

The new drugs -- dubbed protease inhibitors -- would mimic part of the protein's shape, creating a decoy to lure protease away from the real protein and prevent it from fulfilling its duties.

About a dozen companies jumped in, and by the early 1990s a handful emerged with promising candidates: Merck, Roche Holding Ltd., Abbott Laboratories, Upjohn Co. (now Pharmacia & Upjohn Inc.), Vertex Pharmaceuticals Inc., and Agouron Pharmaceuticals Inc.

Merck began testing Crixivan in humans in early 1993, and that summer -- long before even knowing whether the drug would work -- top scientists met for lunch in the company cafeteria to chart how to mass-produce it.

"We gulped. We didn't know how we were going to do this," recalls one of the scientists, Paul J. Reider.

Launching full-scale production of a new drug usually begins only after the drug is in advanced testing and has at least an 80% chance of success. The Crixivan effort, because of the AIDS crisis and competitive pressures, began far earlier -- when the drug had only a 5% to 10% shot.

The effort was further hampered by the huge quantities of Crixivan that would need to be produced. Each patient must take six 400-milligram pills a day, every day, in combination with other AIDS drugs to achieve the desired effect of reducing HIV levels in the blood. Supplying just 90,000 patients with Crixivan is the equivalent of producing enough Vasotec, Merck's popular hypertension drug, for more than 21 million people.

Merck pushed other projects aside, assigned 400 employees to the Crixivan team and found ways to produce small quantities while rushing to build two factories -- even before it knew exactly how the production lines would work.

"We were always trying to get our heads above water, and every time we did, the water kept rising," says Dr. Reider, an ebullient, 45-year-old chemist who oversaw the design of the chemical process. Adds Michael L. King, a 44-year-old Merck senior vice president and chemical engineer who helped implement the production line: "If we had gone through it the traditional way, it wouldn't have happened."

Most Merck pharmaceuticals are made in about four steps over two weeks. Making a batch of Crixivan is a six-week process that requires 15 chemical steps. It takes 77 pounds of 30 raw materials to produce just 2.2 pounds of the drug, enough to supply one patient for a year.

In late 1993, Dr. Reider's team began using a prototype production line at a plant in Rahway, N.J., which was able to turn out bigger batches but took up to four months to do it. Each time, only 15% of the volume of ingredients that went in emerged in the desired shape of the Crixivan molecule. Dr. Reider's team needed to cut the production time down to a month and a half and bring the yield up to 50%. Doing so in mass production -- filling vats two stories high -- would be far more difficult. That is because of Crixivan's terribly complex molecular structure.

A quick chemistry lesson: Many drugs are based on central hubs or "chiral centers," consisting of a carbon atom with four branches of various atoms attached. These hubs usually are readymade in nature; only in the past decade have scientists mastered the ability to create and link together multiple synthetic chiral centers, one of the most difficult feats in modern chemistry.

Most medicines are based on just one or two naturally occurring chiral centers; Vasotec, by contrast, is far more complicated because it has three chiral centers, including a hard-to-create synthetic one. Crixivan goes further still: It has five central hubs, all of them man-made.

One of the big problems with Crixivan's low yield lay in a crucial and especially troublesome step, in which a chemical reaction simultaneously creates two new hubs while fusing those atoms onto an already-joined pair of chiral centers. The resulting molecule kept "overcooking" because of high acidity in the process, ruining a large portion of the output. Four of Merck's best scientists tried to solve the problem but failed.

Then a 28-year-old chemist named Peter Maligres took a shot, adding a common chemical and mixing in some baking soda. He improved the yield from that step by 55%. "Sometimes kids come in and try the simplest things, and they work," Dr. Reider says. "It's a humility lesson." By November 1994, the team had designed the complete Crixivan process.

By now, Merck had to decide whether to pay an outside supplier to produce the drug or to build production from the ground up. Archrival Abbott Labs had confronted the same decision as it rushed to scale up production of its own protease inhibitor, Norvir. Abbott decided to strike deals with suppliers in Japan, Italy and France to handle up to 80% of production.

But Merck didn't entirely trust outsiders to handle its most complex chemistry experiment. Early on, the Crixivan team had hired a supplier to make 200 kilograms of a portion of the molecule for batches to be used in clinical trials. The supplier missed the deadline, threatening to set back the entire project, and Merck had to rush through a batch of its own.

So Merck restricted its outside dependence and relied on just two suppliers, Mitsui & Co. of Japan and DSM NV of the Netherlands, to handle seven early and less-complicated steps of the chemical process. Doing most of the work in-house would require Merck to commit several hundred million dollars to designing and building new production lines -- even though Crixivan wasn't expected to go before regulators for another two years.

It was a big gamble, given that Crixivan thus far had delivered promising results in fewer than a dozen patients, who were taking it in combination with the older drug AZT. HIV's uncanny ability to develop resistance had already nixed one Merck drug candidate and had weakened Crixivan's effectiveness when taken without other AIDS drugs. Clearly, Crixivan might eventually fail in clinical trials.

If Merck didn't proceed immediately, however, its new lines wouldn't be ready in time if Crixivan did work. Edward M. Scolnick, Merck's head of research and manufacturing, believed Crixivan would work, because one patient from an early trial persistently kept the virus at bay, even though all other patients relapsed.

"There was no way that patient could have been a fluke," Dr. Scolnick says now. As later tests combining Crixivan with AZT showed promise, "it provided enough assurance, at least internally, that we should go forward with the program."

In late 1994, Raymond V. Gilmartin, who had just become Merck's chairman, agreed to get production up and running -- provided Dr. Scolnick would pull the plug if the drug began to fail in clinical trials. The Merck board gave the go-ahead in February 1995, and frantic construction at two existing plants in Elkton, Va., and Albany, Ga., began a month later.

At the same time, Merck was getting pummeled by AIDS activists, who accused it of dragging its feet on funneling Crixivan to patients. One group, led by Jules Levin, executive director of the National AIDS Treatment Advocacy Project, demanded that Merck allow an outside drug-manufacturing consultant to evaluate its production efforts.

In January 1995, the consultant met with Merck scientists, who explained in detail their difficulties. But Mr. Levin was unconvinced when the consultant told him Merck was doing all it could. The activist continued to prod Merck to speed up its work, and demanded that it make supplies available on a "compassionate-use" basis, free of charge, to severely ill patients who didn't qualify for clinical trials.

Merck was supplying about 300 patients in trials at the time and needed to produce enough for 2,000 for late-stage trials. Given its manufacturing problems, Merck was hesitant about supplying hundreds more patients.

But the activists' lobbying worked. It "reminded us in the middle of this to step back and think about how important every single life is," Dr. Reider says. The chemist committed to supplying 1,400 patients on a compassionate-use basis, before knowing whether Merck could deliver. "I was risking my career," he says.

By September 1995, competitive pressures were intensifying. Abbott reported that it had compiled impressive clinical data on Norvir. It was clear Abbott's drug would win quick federal approval. The same month, Roche's Hoffmann-La Roche Inc. unit announced it had applied for Food and Drug Administration approval of its protease inhibitor, Invirase; by November, it had won an FDA panel's endorsement and FDA Commissioner David Kessler's promise of quick approval, which came the following month.

Under its original timetable, Merck was still a year away from launching Crixivan. But it was hell-bent on getting to market at the same time as its competitors. So its top researchers waged a pitched lobbying campaign to have Crixivan considered alongside Abbott's drug at a coming meeting of an FDA advisory panel. Never mind that Merck would have to file its FDA application by January 1996, eight months earlier than planned. And that its factories weren't ready.

The FDA agreed, but had a principal concern: Could Merck produce enough Crixivan to meet initial demand? If too many patients started on the drug, they might find their pharmacies out when they went for refills. Patients who stop taking the drug even for a short time risk developing resistance to it.

Merck assured the FDA it could make enough on its prototype line for up to 30,000 U.S. patients. To ensure a steady supply, Merck would funnel most prescriptions through a single major distributor, a mailorder retailer named Stadtlanders Pharmacy in Pittsburgh, which is now majority-owned by Counsel Corp. of Toronto. Merck could then control the number of patients on the drug, and bar new patients when demand matched supply. By contrast, Abbott planned to distribute Norvir to pharmacies nationwide.

Abbott's drug won FDA approval on March 1 of this year, and Merck's entry breezed through just 13 days later.

Soon, however, Merck encountered a torrent of protests over Crixivan's restricted distribution. AIDS activists accused the company of giving Stadtlanders a monopoly that let it price-gouge patients, and retail pharmacists decried getting cut out of the deal. Stadtlanders charges customers a 37% premium above what it pays Merck; however, the pharmacy says most patients pay much less than the list price because of discounts under various insurance, government and other plans.

Mr. Levin, the activist, says that despite his earlier doubts, in retrospect "it's obvious Merck does have a production problem."

Meantime, as Merck churned out small quantities of the drug at its prototype plant, construction proceeded at a furious pace at the 45,000-square-foot main plant in Elkton and the satellite in Albany.

Finally, on May 27, the big operation was ready. The first Crixivan batch entered the system at 9:30 a.m. For six weeks, Merck engineers and chemists crossed their fingers as 30,000 gallons of liquid trickled through 20 miles of pipes, vats, dryers, pressers and capsule- and bottle-fillers. On July 6, the last of the batch's white capsules were in boxes, ready for shipment.

But demand for the drug had begun to pick up after the benefits of protease inhibitors received wide press coverage following an international AIDS conference that week. The Merck crew still had months of work ahead to bring the plants up to full production.

Slowly, more Crixivan "runs" are going on-line; the plants will eventually handle 18 batches at once, each lot producing two million capsules. The question now is whether Merck will get the operation up to full-scale production before demand can overtake it.

Crixivan appears to be outselling Norvir and Invirase by a wide margin, according to IMS America Ltd., a Plymouth Meeting, Pa., market-research firm. Doctors and patients say they favor Crixivan because it has fewer side effects than Norvir and is more potent than Invirase. Demand for the drug may have been slowed by the limited availability of government funds, both in the U.S. and abroad, to pay for the treatment, which when combined with other drugs can cost $14,000 a year per patient.

A big concern is that demand could shoot up at any moment. In recent weeks, for example, the European Union and Canada, two large markets, approved the drug.

Last week, Merck said that the fast pace of international approvals and the higher-than-expected numbers of patients in the U.S. taking the drug have prevented it from building enough inventory to expand distribution to other pharmacies besides Stadtlanders by the end of the year as planned. Instead, distribution through retail pharmacies will begin "sometime in 1997," a spokesman said.

But Merck officials say they are confident they can keep pace. They estimate the new plants will be able to crank out enough Crixivan for 150,000 patients by the end of the year and 250,000 by early 1997.

"It wasn't humanly possible to be ahead of where we are, given the history of the project," Dr. Scolnick says. But he vows: "Merck isn't going to run out of this drug."