By Perry Glasser

Case Study: Eli Lilly and Using IT to Accelerate Research



Focus On | Knowledge management
Business Goal | Discover and move costly new products to market faster
Technology Issue | Key data stored in too many places, being handled by too many people, slowing access, turnaround and discovery
Management Challenge | Silo-thinking, alignment gaps between IT and business executives, and cultural hurdles in convincing top brass that speed doesn't sacrifice quality

A few years ago, if a chemist at Eli Lilly and Co. enjoyed an "ah-ha" moment about a new substance that might lead to a great new miracle drug—and huge new profits—the first thing he or she would have to do is check a database to see if anyone else in the global pharmaceutical company had enjoyed a similar insight. Then he or she would have to check a second database. And then he or she would have to check yet a third—and even then not know for sure. Problem was, the three different Lilly substance databases didn't store information under the same labels or registration numbers, so it wasn't always clear what Lilly knew or didn't: A researcher in Singapore might have registered the substance, but under a different name, slowing the process of discoveries to a walk instead of a run. And that was just one speed bump slowing researchers in the urgent race to get new drugs to market.

But speed bumps are no longer acceptable at Indianapolis-based Lilly. And no wonder. With the new revolution in the life sciences that is following the mapping of the human genome—an explosion of innovation that could raise the number of new substances worthy of investigation from the current 500 to 10,000 per year to 10,000 per day—Lilly can't afford to waste another second getting mired in information-processing headaches. "Speed to market is everything now," says CIO Roy Dunbar, an MBA with a pharmacy degree who has spent most of his career on the business side at Lilly, in product development. Indeed, Dunbar's whole mission since being named CIO in 1999 has been to find new ways to use information technology to shrink the 10 to 15 years it takes to discover new drugs and bring them to market.

Lilly has plenty of reasons to be impatient. Cost pressures are enormous—even on those drugs already in the pipeline. In today's high-paced world of drug development, each day a new drug is not on store shelves can mean $1 million or more in lost revenues. And recent events have raised those stakes even higher. Last April, the U.S. Food and Drug Administration ordered Lilly to delay production of Cialis, a new anti-impotence drug that's expected to rival Pfizer Inc.'s Viagra. The reason: FDA inspectors raised questions over quality problems in Lilly's Indianapolis manufacturing plants.

At press time, for the same reason, the FDA was also delaying the production of as many as four other new drugs in Lilly's otherwise vaunted pipeline—including the potentially lucrative osteoporosis drug Fortéo and the antidepressant Cymbalta.

Lilly is particularly dependent right now on the FDA for new-product approvals to turbocharge its financial results, and the delays have been especially challenging since a 2001 court decision ended Lilly's exclusive rights to Prozac. The antidepressant generated $2.8 billion in revenues for Lilly in 1999 and another $2.6 billion in 2000, and Kenneth Nover, vice president and pharmaceuticals analyst for St. Louis-based A.G. Edwards Inc., estimates Lilly will eventually lose 90 percent of those sales to generic manufacturers—and at least $1 billion in the first year alone.

Adding to those woes are disappointing sales of Xigris, a new drug for sepsis Lilly introduced in late 2001. The lackluster introduction precipitated a 20 percent drop in profits for Lilly in the second quarter of 2002, and third quarter revenues were not expected to be much better. Indeed, said Lilly CEO Sidney Taurel in July, "the next 12 to 18 months will be a decisive period for the company." Particularly disturbing, he says, is the very real possibility that Lilly's new-drug pipeline—among the best in the industry—could be little more than a gleam in the company's eye until well into 2003.

Lilly isn't the only big player in the pharmaceutical industry getting nervous. As R&D costs soar and federal regulations set the bar for drug approval ever higher, the pharmaceutical industry is becoming a horse race with no favorites. Blockbuster drugs like Pfizer's Lipitor, a cholesterol-lowering medicine, can bring in $3 million or $4 million per day. But when investors see a major source of revenue about to expire by judicial fiat, the effects can be devastating. Last August, after a federal judge doubted the efficacy of a new patent on allergy drug Claritin, Schering-Plough's stock valuation took a five percent hit in hours.

In addition, the challenges of the new bioscience revolution, while a potential boon to new drug discovery, is generating terabytes of new data that is testing new knowledge sharing systems as never before—and telegraphing their shortcomings. "One of our big challenges as an industry is to make sure we build data storage to handle it all, and not data landfills," says Daniel Klingler, the senior vice president of information and knowledge management systems for Bristol-Myers Squibb's worldwide medicines group. "In the early 1980s, we sifted through hundreds of new compounds a year. Five years ago, we sifted through 10,000 compounds per year. Today, we're having to find ways to sift through 10,000 new compounds a day. IT's challenge now is finding a needle in a haystack, and in hours—not years."

In many industries—soap, shoes or soft drinks, for example—a company might do well to hedge its bets by seeking safety in numbers with an ever-broader product portfolio. Not so in the highly regulated drug industry. To the uninitiated, the costs and risks of pharmaceutical R&D read like astronomy tables. According to a study by the Tufts University Center for the Study of Drug Development, the average successful drug requires 10 to 15 years of research at an average cost of $802 million—a cost that includes the price tag on the ones that never make it to market. Couple those onerous upfront investments with the fact that patents last only 20 years, and it becomes plain that when a drug might have an earnings window of a mere five or six years, companies are eager to abandon a bad bet early in the race. Another big IT goal in the drug business these days, therefore, is not only to speed winners, but winnow out those costly also-rans as early in the game as possible.

And there's another headache for which aspirin is not enough. Analyst Nover estimates that fewer than one tenth of one percent of substances that are deemed targets for investigation actually pan out as marketable drugs. The challenge to the industry, he says, is to use IT not only to get more information faster, but to get better information. "Money needs to be spent very wisely now," Nover says. "To do that, pharma companies need correct data that is easy to use and at their fingertips. If they have that, they can more quickly make the 'no-go' decision. And the sooner R&D money is saved, the sooner it can be spent elsewhere." The point isn't lost on Dunbar. "The productivity of R&D has not been sufficiently good," he says. "We need to improve it to continue to enjoy our attractive positions in investment portfolios."

Other companies faced with the pressure to move faster in the face of these challenges might be tempted to merge or acquire to pick up speed, but Lilly CEO Taurel—despite the FDA's recent crackdown—reaffirmed in August that's not an option. Taurel prefers partnerships, many of them to help farm out the risk and the cost of gaining speed. On Sept. 20, Lilly said it would pay up to $325 million to Amylin Pharmaceuticals, a biotech company based in San Diego, for the rights to a potentially promising treatment for diabetes.

While a few analysts, including Nover, darkly muse that Lilly just might be exposing itself to a takeover, the antimerger policy, at least for now, is just fine with Dunbar. An M&A strategy, he says, could be another speed bump, and might prove even more confining to his chances of using IT to help bolster the company's bottom line. "At least my legacy systems can talk to each other," he says. "Integrating new systems together would only cause more delays" in the pipeline.

Maybe so, but limiting one's options that way only puts additional pressure on Dunbar and his shop.

Can he deliver? Dunbar doesn't seem fazed by the challenge. Since being sent over to the IT side three years ago, to get information technology aligned more with the business goals of the company, Dunbar has had his work cut out for him. Not only has his mission always been straightforward, but his methods have also been fairly clear.

The Wait for Profits (click image to enlarge)

The Wait for Profits

Net Prophets

Net Prophets

In his first months as CIO, Dunbar brought in IT outsiders, including the chief scientific officer of Sun Microsystems, to preach the value of IT-business alignment to both the business side of the company and the technologists. "The question at the time I took the job was, what value did we get from IT?" Dunbar recalls. "At the time, we seemed not to be able to communicate to our internal business partners how value came about." Bringing in people from outside to "preach the gospel of alignment," Dunbar says, has been helpful: "It's like the story of Jesus, who goes back to his own people and they say, 'Who are you? Get lost.' The prophet from afar is always listened to with much more attention." That's important, he says—IT-business alignment has been and will be critical to shortening time to market and staying in the game in the biotech revolution.

Dunbar has also reached out in other ways, helping to form e.Lilly, a division of Lilly whose mission is to find new business models that can be driven by information technology. To offload research risk, e.Lilly has accessed a worldwide network of independent researchers through InnoCentive, a company partly funded by Lilly that offers bounties for solutions to specific problems in chemistry—all posted on InnoCentive's Web site (www.innocentive.com). Independent chemists willing to accept Lilly's rules about intellectual property rights can register and accept the risk of failure while they pursue financial rewards.

If you have a Bunsen burner in your garage, you may want to check out a recent offer of $90,00 to the first person who offers, for example, "an efficient synthetic strategy for the deazaguanine ester." This molecule, Dunbar explains, has been previously reported in chemical literature, "but the existing known synthetic route may be lengthy, expensive and low yielding." Your time, your risk and Lilly's research needs all add up to a pretty good deal, provided you are the winning researcher. Dr. Sangtae Kim, vice president and information officer at Lilly Research Laboratories, says some of the bigger prizes are in the $50,000 to $100,000 range, and the total of all the prizes on the site is in the millions.

Finally, to meet the stringent requirements of the FDA for lengthy clinical trials—and speed up the lengthy trial process, which can take years, Lilly is reaching beyond its own walls in other ways, making use of e-business partners such as San Francisco-based 1747 Inc. and Phase Forward Inc. in Waltham, Mass. Both companies are outsourcers that work with physician groups and hospitals to gather sufficient suitable subjects for experimental drugs, and then take a drug through the stages of those trials, guarding against statistical anomalies or irregularities in the geographically dispersed sites where the trials go forward. For Lilly, such new alliances are also speed-gainers: Sure, they don't spread risk, but they can drastically cut time to market—by months, if not years.

And that's not all. Dunbar is also trying to improve IT's support of the processes used to discover new drugs. One of his first efforts as CIO was to clear up time-eating confusion over multiple databases for compounds under research. He did this by creating what he calls Lilly's Sample Identification Database, a system that registers and stores all Lilly compounds under a single digital name globally. This means chemists can spend less time calculating and more time doing research. "Using this tool and its algorithms, we have been able to demonstrate an eight- to 100-fold improvement in the quality and quantity of research hits," Dunbar says—referring to a higher success rate in making discoveries that could lead to new blockbuster drugs. Further, he says, "chemists can access our chemical library in seconds," thereby cutting from days to minutes the time it now takes for chemists to act on a good hunch.

Also to boost speed, Dunbar has set up a Molecule Library, a database that lets people access all documentation created for any compound in the pipeline. The timesavings per search is estimated at two to three hours over the previous system. That's huge: It translates to 200 to 420 hours daily, and annual time- and cost-savings in the months and the millions.

Boosting the capacity and ease of storage of all of this new data is also critical, Dunbar says. His Beacon warehousing and data management project takes 20-plus different data storage systems and boils them down to one integrated system, saving hours per search. Dunbar says each two- to three-hour search is reduced to 20 minutes, thanks to simpler and more streamlined data warehousing systems. It's Dunbar's hope that Beacon, now in pilot testing with 2,000 scientists, will also enable Lilly to expand the depth of its research. "Storage is critical to the future of this industry, as the amount of sheer data that will be available to mine is growing exponentially by the day," Dunbar says. "Our ability to store it, mine it and access it all in a heartbeat is at the core of our present and future ability to compete."

Meanwhile, Lilly's GAME—a waggish acronym for Gene Anatomy Made Easy—is a piece of proprietary software that enables researchers to perform sequence analysis on their own without help from bioinformatics experts. This way, Lilly saves on personnel, and better yet, says Dunbar, "GAME is able to compute and generate more information in five minutes than most researchers could in four hours."

Power in Numbers

Power in Numbers

Down the road? Think grid computing—using untapped computing power across Lilly's corporate landscape and elsewhere to solve complex research problems faster. Grid computing also might let organizations and companies inside the life sciences business share computing power, data and storage capacity via private networks or over the Web, and Lilly spokeswoman Joan Todd says the company is in discussions with several research institutions about future grid strategies. Grids could help Lilly add thousands of hours of computing power to any effort, thereby increasing the speed at which new compounds could be screened—or scrapped. Says Dunbar: "We are in the midst of two breathtaking revolutions—one in the biological sciences, and a second in the availability of computing power and associated network capabilities," Dunbar says. "Pharmaceutical IT success in the post-genomic era will be a product of integration of systems, data and the information."

Moving forward will also require more applications of the new science of bioinformatics, which Dunbar defines as "research, development or application of computational tools and approaches for expanding the use of biological, medical, behavioral or health data, including those to acquire, store, organize, archive, analyze or visualize such data." Dunbar says he asks other drug companies this: "Does each chemist in your company have immediate access to every single compound in the company's chemical library with a latency of no more than a few seconds?" Then, he adds, "Ask yourself who owns the vision and the work product for an effort. If there is no clear answer, then worry."

What's ahead for Dunbar? A lot more of the same. Dunbar summarizes the continuing challenge for his IT group and Lilly by saying that "the single biggest contributor [to costs and speed bumps] is poor information visibility in R&D, which in turn leads to poor decision-making, which contributes significantly to the upwardly spiraling costs and the disturbing slide in R&D productivity."

Will the revolution in biosciences, the rise of grid computing and the plummeting costs of data storage save the day for Lilly in the long run? It's hard to say, but Dunbar guesses that R&D productivity at Lilly "has probably doubled in recent years" thanks to some of the company's knowledge sharing initiatives already in place—and should continue to improve significantly.

But Dunbar concedes that aligning the three cultures of business, IT and science will continue to be difficult. Lilly, for all of its IT smarts, has only just begun to integrate itself into a speed demon. "We have made a lot of progress in IT," he says, "but the important part yet to be done is the full integration of all the tools. When we fully integrate, I suspect the results will be breathtaking." Dunbar is counting on it; Lilly, of course, can't afford to see it any other way.

Perry Glasser, the director of the writing program at Wichita State University, writes frequently about business and information technology.


's Speed Gains">

Lilly's Speed Gains

When time is money, a dose of IT can be just what the doctor ordered.

BUSINESS/RESEARCH CHALLENGE Potential drug compounds are tested against many organic samples to determine a potential biological response, but promising avenues of research are hard to discover and take a long time to achieve.

solution Structure-Based Drug Design (SBDD) uses computer-based algorithms to identify the compounds most likely to have drug-candidate properties. Only these compounds are researched further.

payoff Lilly now arrives at a "go" or "no-go" decision for further research quicker, hiking productivity eight- to 100-fold, depending on the type of compounds and screens used.

BUSINESS/RESEARCH CHALLENGE Researchers are slowed by complex retrieval systems and unreliable data: No single database is the repository of records for compounds being researched.

solution The Sample Identification Database, a chemical library, houses data about all Lilly compounds, giving each a single unique identifier and common attributes.

payoff Some 1,000 global users make about 400 new registrations per day. Accurate retrieval of information from the database is now possible in just seconds.

BUSINESS/RESEARCH CHALLENGE Not everyone in Lilly's manufacturing and development staff can access research.

solution The Molecule Library enables key people to access all documentation created for any compound in the pipeline.

payoff Total time savings of 200 to 240 search hours per day across the company.

BUSINESS/RESEARCH CHALLENGE Gene sequence analysis is a complex, specialized skill needed to research new gene therapies under development. The permutations and combinations of genes are not infinite, but they may as well be.

solution GAME (Gene Anatomy Made Easy) is an internally developed software program that lets biologists, toxicologists, physicians and patent attorneys perform their own sequence analysis.

payoff Lilly says GAME is able to generate more information in five minutes than a researcher could perform in four hours.

BUSINESS/RESEARCH CHALLENGE Life sciences data ripe for mining is increasing rapidly in complexity and volume, boosting the amount of data that must be sifted through at a rapid pace.

solution Beacon, a new data warehousing and storage system, replaces sluggish systems and centralizes data.

payoff Helps boost time savings by reducing each data search, on average, from two to three hours down to 20 minutes.




e-Business Intelligence: Turning Information into Knowledge into Profit
By Bernard Liautaud with Mark Hammond
McGraw-Hill, 2000

Managing Enterprise Content: A Unified Content Strategy
By Ann Rockley
New Riders, October 2002

This article was originally published on 10-10-2002