Case Study

In the drug business, being first to market can make you rich. Isis Pharmaceuticals got wired to win

Before Mark Lotz can even gulp down his last swig of morning coffee, a message pops up on his computer screen. It's the sort of instant E-mail "you never want to see," says Lotz, the executive director of regulatory affairs for Isis Pharmaceuticals, a $26-million drug-development company in Carlsbad, Calif. A cancer patient who was participating in one of the company's clinical trials in Germany has just died. Lotz's mind begins to race.

He can almost hear the clock ticking. Lotz may have just 72 hours to fire off an explanation of the death not only to the U.S. Food and Drug Administration but also to numerous regulatory bodies throughout Europe, where the drug is currently in clinical trials. And that's not the worst of it: each of the European governments has its own guidelines for how a death must be reported. If Lotz fails to make the deadline, Isis could face substantial fines--even criminal liability.

Just a year ago, Lotz would have quickly become entangled in a world of frantic phone calls and faxes. It would have taken him days to sort out the details of the death and type them up in the proper format for each country. Today, fortunately, the process is much simpler. He just has to flip from his instant E-mail message to his Web browser, log on to the company's intranet server, and call up a detailed description of the death as reported by Jo Glover, Isis's European medical director, in Surrey, England. Lotz can then click on a button at the bottom of Glover's report and call up a menu that allows him to automatically format and print out the report according to the specifications of the FDA or any of the European agencies. Within 24 hours, regulatory bodies around the world could be notified. "We used to fear that we'd miss the deadline," says Lotz. "Now we can speed through the process in a day."

This morning, thankfully, Lotz is off the hook. From Glover's report it's clear that the death in Germany was neither caused by the company's drug nor unexpected--the patient was in the late stages of cancer even before he entered the study. According to the rules of the FDA and most of the other regulatory agencies, Lotz doesn't have to report the incident until the trials are completed. So now, instead of racing the clock, he scrolls to the bottom of the report and clicks on a button labeled "comments." He then types in his reason for not immediately reporting the death and posts it to an electronic bulletin board that keeps select Isis employees up to date.

This slick deployment of the company's intranet is just one of the many technological tricks that Isis does to meet the rigorous and costly bureaucratic demands of drug development and approval. Developing a single compound takes 15 years on average, according to the Center for the Study of Drug Development at Tufts University, and generally costs about $300 million. In the race to create new drugs, even one misspent day can easily add up to thousands of dollars lost to the bottom line. That's certainly the case for Isis. The small company is on track to be first to market with a revolutionary class of drugs known as antisense, which target and modify the strands of ribonucleic acid (RNA) in the body that produce disease-causing proteins. In the pharmaceutical industry, where physicians are slow to accept new medicines unless they offer dramatic advantages over existing ones, being first to introduce a drug is critical; it's the golden ring that can firmly establish a company as a market leader. "The importance of time can't be underestimated," says A. Paul Boni, a senior research analyst who covers the biotechnology industry for Genesis Merchant Group Securities, an investment banking firm in San Francisco. "Delaying a product's entrance into a new market for even a few months can mean millions lost."

Traditionally, large pharmaceutical companies have spent millions of dollars on data-crunching systems that will help them mitigate the expense and complexity of drug development--systems that have been far beyond the reach of small newcomers like Isis. But within the past three years, less expensive versions of these sophisticated systems have become available, enabling smaller outfits to compete more effectively not just with their counterparts but also with the behemoths who in the past have threatened to acquire them or profit from their discoveries through licensing agreements. Isis, for one, wasted no time in taking advantage of the affordable technology. Since 1995 it has invested $300,000 in computer equipment--everything from a document-management database and a local area network (LAN) to an intranet on the Web. Lotz can't imagine how the company could handle the FDA-mandated clinical trials--in which a new drug must be tested on thousands of patients through at least three clinical phases--without it.

Isis, of course, is not alone in its aggressive use of technology. Even its primary competitor--the $4-million Hybridon, in Cambridge, Mass.--has a similar system. But Isis has an edge: currently in the third phase of clinical testing for its first drug, Isis 2922, which halts the progression of cytomegalovirus (CMV) retinitis, a viral infection common among late-stage AIDS patients, the company may be one of the first pharmaceuticals--large or small--to piece together its clinical findings in the form of a fully electronic new drug application (NDA), which it will present to the FDA on a single CD-ROM. Isis's NDA is short--a mere 40,000 pages printed out, compared with the more standard 200,000--which makes it a perfect pilot case for the sleek electronic format.

By submitting an electronic NDA, Isis believes it can shave at least a few months off the FDA's average 15-month review. "We think that we can get this through the FDA in a year," says B. Lynne Parshall, the company's chief financial officer. Isis's goal, of course, is to beat out its competitors and get Isis 2922 to the $200-million CMV retinitis market by the end of 1998 so that it can quickly begin to gobble up market share. The payoff could be enormous: just a month's worth of sales from Isis 2922 could more than cover the entire cost of the company's $300,000 computer system. More important, being first to market could be a key factor in making the company profitable.

Looking back on the launch of isis, CEO and founder Stanley T. Crooke admits that at first the company's prospects seemed especially bleak. "I don't think even I would have bet on us early on," says Crooke. In 1987 Crooke was the president of research and development for SmithKline Beckman (now Beecham), in Philadelphia. But traditional drug exploration was no longer satisfying Crooke's scientific appetite, and antisense drug technology, though new and untested, offered a potential giant leap forward in the fight against disease. With the help of a young scientist named David Ecker, Crooke began to experiment in the nascent field.

The two men soon discovered that SmithKline couldn't provide the resources needed to conduct a meaningful investigation into an entirely new category of drugs. The antisense experiments required more than just different scientific instruments, says Ecker; they had to be conducted by a different type of scientist. SmithKline had always hired chemists and biomedical engineers, who have a strong knowledge of the proteins that cause diseases. But the designers of antisense drugs need a deep understanding of RNA. "We were focusing on a totally different molecule," says Ecker. "And that meant we needed an entirely different workforce." So in 1989 Crooke and Ecker left the giant company and, with Daniel Kisner, a former vice-president of R&D at Abbott Labs, in Abbott Park, Ill., started Isis.

Crooke thrives on a challenge--though you'd never guess that from his appearance. He often comes to work dressed in a pink polo-style short-sleeved shirt and khaki shorts--proper attire given that it's balmy Southern California. But if you look beneath the surface, there are clear signs of a fierce competitor. For one thing, in spite of his delicate frame and admitted lack of vertical rise, Crooke is an avid basketball player--out on the courts daily. "I live to play the game," he says.

For a start-up in the pharmaceutical industry, however, the challenge isn't only about trouncing competitors; it's about creating a sturdy business and a cutting-edge scientific laboratory under one roof. And that takes as much knowledge of organizations as it does of organisms. Crooke has both. Glance at his bookshelves and you can't help but notice his ambidextrous expertise: Rosabeth Moss Kanter's The Change Masters leans up against archaic-sounding tomes like Advances in Prostaglandin, Thronboxane, and Leukotriene Research.

In the beginning Isis dove into identifying molecules that could become stable antisense drugs. By early 1991 Crooke had raised nearly $21 million in private placements and had already begun to see some optimistic lab results. The early success was enough to satisfy Wall Street, and in May of that year Isis went public, taking in close to $25 million. It was enough cash to begin clinical trials of five new drugs. Isis 2922 was the most promising.

Coordinating clinical trials is no easy task. For starters, test sites can be anywhere in the world, and each site generally handles between 20 and 1,000 patients. The process is long and involved because overwhelming amounts of data must be collected. Every time a patient in a study takes a shot or ingests a pill, which could be once a week or three times a day, the doctors and nurses on-site must record a slew of information about the patient's health. Everything from blood pressure to skin tone is written on a special case-report form, which is then delivered to the pharmaceutical company, where the information is entered into a database. For any given study, there can be as many as 100 or more pages per patient.

At Isis, those case-report forms typically arrive on Stuart Nixon's desk. He's the company's director of clinical-data management and biostatistics and the architect behind the company's data-management system. Nixon seems to embody the stereotype of a laid-back Californian, with his blond hair, startlingly blue eyes, Grateful Dead watch, and Jerry Garcia tie. But he is anything but far-out when it comes to processing data from the company's clinical trials. Nixon has set up a two-tiered system at the company: a massive document-management application called Optix, from Blueridge Technologies, to maintain a graphical record of the completed case-report forms, and a relational-database application called 4D, from ACI, to enable Isis's biostatisticians to analyze the data.

As soon as the case-report forms enter the building, a small army of data-entry operators quickly scans them into a Sun workstation as images and stores them in Optix. (They can't be scanned in as text because of the idiosyncratic nature of the compilers' handwriting.) The data-entry operators then keypunch the data from the images into the 4D database. Now biostatisticians can begin to manipulate and correlate the results from the studies using sophisticated statistical-analysis programs such as SAS, by the SAS Institute, and Axum, by MathSoft.

Nixon estimates that the analysis stage of development has been reduced by 5% to 10%, thanks to the Optix system alone. "But the best savings," he says, "is that material isn't misplaced or left out." Before the company got the Optix and 4D databases, the case-report forms had to be photocopied and distributed to several people for review, including data-entry operators and clinical research scientists. If any of them needed to review a form from a different study, it meant a long cruise through rows of file cabinets. Now the information from the forms is entered once and then reviewed in Optix via the company's LAN, which is made up of Windows NT running on a Compaq server. A field monitor on-site at one of the clinical studies follows a similar protocol: if, say, she wants to review a patient's temperature from the previous week's assessment, she just dials into the LAN, opens the Optix database, and searches for the appropriate form.

Nixon's greatest worry is that data will be entered incorrectly. Even the slightest inconsistencies in clinical findings can raise an eyebrow or two at the FDA. To avoid mistakes, he has designed an elaborate error-control system, complete with stringent checks and balances (see sidebar, below). In the best of all possible worlds, he says, on-site personnel would enter the information remotely (perhaps using pen-based laptops), dial up the LAN, and input the information directly into the company's 4D database. To date, however, Isis's only foray into electronic data collection involves its central laboratory services, which send the company a disk containing blood values from the clinical trials; Nixon then imports the data electronically into the 4D database. But with the popularization of the Internet, he says, that may soon change. "It used to be too expensive and difficult to set up proprietary remote data-entry systems," Nixon says. "But expanding our use of the Internet would be relatively inexpensive, and everyone would be familiar with how to use the system. We should be able to shave weeks off the average time it takes to complete a clinical trial."

Mark Lotz leans forward in his chair in one of Isis's nondescript conference rooms and waves his arm toward the hall. It used to be, he explains, that when he needed a statistic from a clinical trial he'd have to scrounge around in a dusty closet stuffed with the hundreds of three-ring binders that contain a single drug's NDA. Now, however, with the development of Isis 2922's completely electronic version, he'll be able to hold all that information, literally, in the palm of his hand. And finding the most obscure number will require little more than a swift electronic search. "Having this on disk will change the way I work," Lotz says.

Filing an NDA isn't exactly like applying for a new credit card. Lotz and his staff of four will use FrameMaker, a desktop-publishing application from Adobe, to assiduously compile every cell culture, every animal test, every case-report form, every chart, and every graph into one cohesive, cogent electronic document that meets the government's strict guidelines. "At just 40,000 pages, our NDA is really small," says Lotz. "But I still must read every page of it."

For years the FDA has required that pharmaceutical companies submit at least three copies of the NDA in binders that were no more than 3 inches thick. Dr. Carl Peck, who served as the executive director of the FDA's Center for Drug Evaluation and Research (CDER) from 1987 to 1993, recalls workers unloading countless binders from semitrailers and then lugging them into the agency's headquarters in Rockville, Md. Even by the late 1980s, the FDA had done little to automate the NDA review process. "I was astounded at the lack of computers at the agency when I first arrived," says Peck, who also claims that most medical reviewers had little computer expertise and knew only how to do minor tasks on the agency's out-of-date Wang word processors.

The FDA officially started accepting computer-assisted NDAs (CANDAs)--paper applications supported by some background material in an electronic format--in the mid-1980s, but until recently few small companies had bothered submitting them. The CANDA was a fairly expensive proposition because the drug companies typically had to loan computers to the FDA. "A company our size couldn't afford to tie up its computers at the FDA," says Lotz. It wasn't unheard of for medical reviewers to have four or five computers from different drug companies packed into their cramped offices. "I think the record was six," says Robert Bell, director of the CDER's Office of Management from 1987 to 1995. But there was more than just a space problem; the reviewers had no clue how to run the various systems, which meant that despite the background data on disk or on a loaned computer's hard drive, they continued to rely almost exclusively on the paper applications. "We weren't just stacking up paper," says Bell, "we were stacking up computers."

Things are very different today. The FDA upgraded its computer system three years ago and currently has networked PCs running Windows on almost every reviewer's desk. And according to both Peck and Bell, the new breed of medical reviewer is much more technically savvy. The agency now encourages drug companies to present the electronic portions of their NDAs in the Portable Document Format (PDF), a popular computer-file format invented by Adobe that converts printed documents into computer files without losing the original documents' design qualities. The advantages of PDFs are that almost anyone can create a document in the formats and it can contain hyperlinks like those on the Web.

When Isis puts the finishing touches on its first NDA, Lotz will simply save the document in PDF, press it into one CD-ROM, and then ship it to the FDA. Of course, he won't get off without printing at least one paper copy for the agency's archives. But the time and cost savings over printing and hole-punching nine copies will be tremendous. Lotz estimates that it will take 30% less time (a full month less) to prepare and ship the NDA to the FDA than it would if the company weren't automated. That's substantial when you consider that every month saved could mean an additional million or so in the coffer.

Even more important, though, is the fact that the single CD-ROM will allow the FDA to review the application much faster. In the past, NDAs were divided into sections, which were dispersed to the appropriate departments throughout the agency. One master copy was always kept in the agency's library. That meant that if, for example, the FDA's toxicologists needed to review a section of the NDA that the chemists were holding, they had to request a copy from the librarian, who would photocopy the proper section and then deliver it to the reviewers. Satisfying the request could take as long as a week.

But with the electronic version, FDA reviewers can jump from section to section of the NDA via hyperlinks in the text. It's Lotz's job to anticipate the links that the FDA might want to use. For example, he might link raw data from an animal test to a toxicologist's analysis of the same test so that the reviewer can simply flip between the data and the write-up.

There have been unanticipated benefits to the electronic NDA as well. Lotz happily reports that it has actually helped soothe the often adversarial relationship between the pharmaceutical industry and the FDA. In order for him to anticipate all the proper links within the NDA for Isis 2922, he's had to work closely with FDA reviewers to find out exactly what they need and how they want it presented. "I now feel like I'm actually collaborating with the FDA," says Lotz.

The prognosis for Isis looks good. "Antisense has finally gained credibility as a method for drug treatment," says Crooke. Recent articles in the trade and popular press support that assessment. In May Science claimed that antisense drugs had "shed their troubled image" and were beginning to be shown clinically effective. And in June Chemical & Engineering News trumpeted "Antisense Drugs: Finally Fulfilling Their Promise."

The investment community seems to agree. Many analysts predict that Isis will be profitable by 1999. Right now, the company burns about $25 million in cash each year. Still, it has approximately $70 million in the bank, which should keep it operating comfortably until Isis 2922 hits the market.

Between now and then, Isis has lots of work to do: it must complete the clinical trials for Isis 2922 by year's end; analyze the results of its treatments for cancer, HIV, and Crohn's disease; work on its strategic partnership with Novartis; and bulk up its R&D department with an eye toward developing new generations of antisense molecules. "We are just at the end of the beginning," Crooke says.

Joshua Macht is an associate editor at Inc. magazine.

Safeguards: No Mistake About IT
Whether test marketing a fruit drink, beta-testing a software application, or clinically testing an antibiotic, most small companies face the same dilemma: how to gather and analyze data in a way that will enable them to get to market faster. A solid information system can be the linchpin in the process.

But for the system to be worth anything, the data must be entered correctly. To guard against errors, Isis's Stuart Nixon has devised this painstaking five-step program:

1. Manual review
Before a patient's information is even entered into the computer, a data-entry operator reads it and checks for obvious mistakes. For example, a patient's age should never be 150.

2. Double entry
Once the patient's information has been punched into the database, a second data-entry operator reenters the same information into a new file. The computer then automatically compares the two files and prints out a list of inconsistencies, which must be reconciled by the two operators.

3. Visual verification
A data manager compares the data that has been entered against the original handwritten notations.

4. Electronic verification
A data manager runs a series of relational edit-check programs that check the data in selected fields within the database to make sure it's logical and consistent.

5. Final audit
Once an entire clinical trial is completed, Nixon and his team conduct one final audit by checking a representative cross section of data against the original handwritten notations. --J.M.