Seeking Strategic Impact

By Edward Cone  |  Posted 2004-06-18 Print this article Print

Seeking Strategic Impact Seeking Strategic Impact

What Scalet could really use about now is a software package that would make exciting new patented and FDA-approved medicines materialize from thin air.

While that’s not realistic, one way Merck has attacked the issue of research productivity was the $630 million acquisition of Kirkland, Wash.-based Rosetta Inpharmatics in 2001.

Rosetta specializes in the use of "gene chip," or "DNA microarray," technology. Much as microprocessors pack computer circuitry into a small area, gene chips are miniature genetic laboratories that can read the genetic programming of a tissue sample in search of clues to the progression of a disease or probable reactions to a drug. Each chip contains tens of thousands of genetic sensors. Tests are repeated hundreds or thousands of times, so gigabytes of data can result from a single experiment. Rosetta devised software that helps make it manageable.

Certainly, the Rosetta acquisition was a recognition of the potential for computer power to accelerate Merck’s core business, but Merck’s information technology organization had little to do with it. Indeed, Merck doesn’t even call it "information technology."

Instead, such scientific computing is called "informatics." And Rosetta, now part of Merck Research Labs, is pushing the bounds of computing, medicine, genetics and biology, seeking to create new drug formulas. This is "bioinformatics."

What Scalet’s information technology organization is expected to provide is the more basic computer and network platform for the scientist’s work, along with applications to track the administrative aspects of drug development. One area Scalet recognizes as having "huge opportunities" for information systems impact is speeding up clinical trials.

A drug entering clinical trials is the result of decades of experimentation—basic science on the formation of chemicals and their effect on the body, leading to the formulation of a promising compound, initial toxicology studies and animal testing. A few candidate drugs eventually win approval to be tested for safety on a small number of human subjects, in what’s known as Phase I testing. In Phase II, the testing focus is sharpened to determine the drug’s medical effectiveness. Most drug candidates fail before they ever get to Phase III, the large-scale clinical trials aimed at producing medical-efficacy and safety data that will convince the FDA to approve the medicine for sale.

Every day that can be shaved off the clinical trials process potentially brings that drug closer to FDA approval. If Merck could develop a drug with Zocor-like sales and get it to market 30 days sooner, it would gain more than $400 million in additional revenue.

Significant time savings are within reach. A return-on-investment study Merck conducted from October 2002 to May 2003 with DataLabs, a vendor of data collection software for clinical trials, showed a 33-day savings in the time needed to design the data collection forms prior to the start of a trial and, more importantly, a 30-day savings in the time required to conclude the trial and analyze the data. The system also cut information systems costs associated with a trial by about 22%—largely manpower savings, including reduced effort by programmers and database developers. By catching more errors at the time of data entry, the system also reduced the clerical effort devoted to catching errors prior to producing a report on the trial’s results.

Rich Gleeson, vice president of enterprise solutions at DataLabs, says the project showed him how rigorously Merck manages its systems by "methodically measuring the impact—was there a benefit to the software?" To his delight, Merck allowed DataLabs to publish the results, and licensed the software.

Merck continues to evaluate similar products from other vendors and may yet choose a different one as its corporate standard. Still, the DataLabs project shows the potential.

Merck has built a hugely profitable business on the back of blockbuster drugs such as painkiller vioxx. But the pipeline of new products is drying up. Sales growth has slowed to 5% a year and profits are down by 4.5%. To help get through what could be a long dry spell, the company brought in a cio, chris scalet, from the tight-fisted paper industry to help streamline operations and bolster drug research. But scalet faces a tough balancing act: wring efficiencies from i.T. Operations without impeding research initiatives.

When Chris Scalet Arrived at Merck in March 2003, he knew he had a lot to learn. His immediate predecessors at the pharmaceutical giant had backgrounds in scientific computing and intimate knowledge of the information technology needs of the drug maker’s all-important research and development arm—the organization that turned out blockbuster drugs such as anti-cholesterol treatment Zocor and painkiller Vioxx.

"If you can consistently eke out 30 days from the process, that’s worth a lot," says Dr. Louis C. Kirby, president of Pivotal Research, a contract research organization that often conducts clinical trials for Merck. However, the time gain only matters if the drug is ready for FDA review 30 days sooner.

Often, even after the data is available for review, the study sponsors will spend the next 60 to 90 days mulling it over before settling on a design for the next phase of testing. So the trick is not to be caught flat-footed, to have the next phase planned before the last is completed, Kirby says. "You have to be ready to roll right from Phase II to Phase III. If you don’t, that time savings is going to be squandered."

Merck says it has in fact sped up the development and testing process, pointing to another promising drug candidate, a diabetes treatment known as MK-0431. Merck Research Labs president Peter Kim told stockholders at the company’s annual meeting in April that Merck cut nearly in half a key period in the development of that drug. The industry average for the time to get from the first laboratory toxicity tests of a drug candidate through the beginning of Phase III is 4.12 years, he noted. Through a combination of scientific and procedural improvements, Merck reduced that time to 2.25 years for MK-0431, Kim says.

Kim did not try to quantify how much information technology contributed to that gain. Scalet says he believes systems improvements played a role, "but by no means was it I.T. alone; that was I.T. plus a whole lot of hard work." Much of the credit should go to the research organization for how it restructured itself and its decision-making processes, he suggests.

Still, Merck saw the potential to use information technology to compress the clinical trials process a long time ago, and has had supporting information systems in place for 30 years or more, although it hasn’t always been a smooth process.

In the 1990s, Merck struggled to erect a modernized clinical data system known as CRISP (Clinical and Regulatory Information Strategic Program), a project that current and former information systems workers came to regard as a $100 million fiasco.

"This project ran into a lot of problems," says one former manager who was involved near the beginning of the CRISP project. He asked that his name not be used.

Previously, when Merck conducted worldwide clinical trials, its overseas offices would collect the results on paper, which would then be sent to the U.S. for data entry. Merck succeeded in moving from a centralized mainframe database to a distributed database architecture, allowing remote offices to record the results of a trial on a local database, then replicate that data to the home office in New Jersey. By 1995, CRISP was deployed on more than 25 Hewlett-Packard 9000 servers located in more than a dozen countries, and Merck was talking about doubling the deployment over the next two to three years.

The problems started when Merck decided to create a graphical user interface to CRISP, allowing clinical researchers to do their own data entry rather than having Merck personnel transcribe the results from paper forms.

Merck significantly underestimated the challenge, according to a former Merck executive who was involved in the early phases. Merck’s scientists were impressed by the ease of use of the Apple Macintosh computers they were using at the time. They thought creating a graphical data collection tool ought to be an easy task for Merck’s programmers.

The task of creating this software led the programmers into unfamiliar territory. Not only did they have to learn new programming techniques, but they were continually frustrated in their effort to build a single system that would work for all trials. The real problem was that the data to be collected varies significantly with the nature of the medicine being tested and the malady it addresses. That meant both the user interface and the database structure had to be different for each trial. For example, aside from a patient’s vital signs, the data to be collected for a depression drug study might have almost nothing in common with that for a diabetes study.

As a result, developers found themselves continually customizing the system for each new trial. What they really needed was a way of generating new applications automatically. That’s essentially what DataLabs has pulled off, incorporating the Microsoft Visio diagramming tool and a series of templates for visually depicting the design and schedule of a trial, together with the data to be captured. The software then creates a browser-based user interface and the corresponding database structure.

Merck salvaged the best elements of CRISP for what was renamed the Clinical Trials System (CTS), which remains in use. And while Merck is still pursuing the kind of efficiencies CRISP was supposed to produce, today it can take advantage of a new generation of technologies.

CRISP was "too far ahead of its time," says Judy Sromovsky, former director of clinical trial system standards at Merck. "In 2004, when I’m talking to [study] sites, all they want to do is electronic data," says Sromovsky, now study director at The Medicines Co., a smaller New Jersey pharmaceutical firm. "They were more comfortable with paper back then." A large part of the change comes from user familiarity of the Web, which replaces the client/server architecture on which CRISP was based.

Meanwhile, the Clinical Data Interchange Standards Consortium, in which Merck is an active participant, has created an eXtensible Markup Language specification that makes it possible for Merck to create one standard interface to CTS that will work with compliant electronic data collection software from DataLabs and other vendors.

In retrospect, the CRISP project "was too ambitious and involved a lot of process change and new technology," Popper says. It also predates the project methodology he instituted, he says. "When we put the new oversight in place, things started to get better."

But even with those processes in place, one former Merck developer who was laid off in January says he ran into something a lot like CRISP ’s runaway requirements on other projects. The development of one particular database had been under way for years, but the scientists kept piling on new requirements that kept it from being finished. Instead of continuing the endless committee process, the developers finally had to put a freeze on new feature requests so they could complete a working system.

Next Page: Unifying ERP.

Senior Writer and author of the Know It All blog

Ed Cone has worked as a contributing editor at Wired, a staff writer at Forbes, a senior writer for Ziff Davis with Baseline and Interactive Week, and as a freelancer based in Paris and then North Carolina for a wide variety of magazines and papers including the International Herald Tribune, Texas Monthly, and Playboy. He writes an opinion column in his hometown paper, the Greensboro News & Record, and publishes the semi-popular weblog. He lives in North Carolina with his wife, Lisa, two kids, and a dog.

Submit a Comment

Loading Comments...
Manage your Newsletters: Login   Register My Newsletters

Rocket Fuel