Chapter 12

Can Knowledge Management Systems Help Pfizer?

Pharmaceutical companies are among the most intensive users of knowledge management systems, and you can easily see why. The drug discovery process is long and arduous. Researchers must first identify a biological target such as an enzyme or gene that appears related to a disease; fling hundreds of thousands of compounds at the target to see which interact with it; and conduct animal studies of toxicity, absorption, and the properties of the most promising molecules. If all still looks good, they would then test one of the compounds on humans.

           Only one new chemical entity in 10,000 makes it through the U.S. Food and Drug Administration (FDA) approval process, and only half the drugs approved make it to market. The complete process costs $500 million to $700 million per drug, and each day of delay in a seven-year testing cycle for a hot new drug can cost $2.5 million.

           Today the stakes are higher than ever. There are very few new drugs in the pipelines of major pharmaceutical companies. Despite steadily increasing expenditures on research and development, which now totals more than $25 billion annually in the United States alone, the U.S. FDA statistics show a steady decline in the approval of new drugs, or “new molecular entities.”

           The pharmaceutical companies are doing everything they can to develop new products and come up with new ideas—promoting a more innovative corporate culture, forging collaborative ties with university researchers, and acquiring young pharmaceutical and biotechnology firms to obtain new sources of expertise. Any knowledge from any source that can bring a new drug to market or expedite the drug development process is obviously very valuable.

           Let us look at the role of knowledge management at one of these companies. Pfizer is the world’s largest research-based pharmaceutical firm. Its best-known products include Celebrex, Zoloft, Lipitor, and Viagra. In addition to prescription drugs, the firm makes over-thecounter remedies such as Bengay, Listerine, Benedryl, Visine, and animal health products. Pfizer is divided into three major business segments: pharmaceutical, health care, and animal health, with the pharmaceutical segment accounting for 88 percent of Pfizer’s total revenue.

           Among Pfizer’s 122,000 employees, over 12,500 are scientists who work in research labs around the world. Pfizer Global Research and Development is the industry’s largest pharmaceutical R&D organization, with a $7.1 billion budget for R&D in 2003. Pfizer’s search for new drugs encompasses hundreds of research projects across 18 therapeutic areas more than any other company. The company maintains links with more than 250 partners in academia and industry.

           Like other major pharmaceutical companies, Pfizer relies heavily on knowledge management systems to drive its research and development work. It has systems to manage all of the documents and pieces of data involved in developing a new drug; expertise location systems to identify scientists and knowledge leaders within the company and outside experts who are involved in drug research and development; and searchable databases of information collected during clinical trials. Pfizer has Web-based portals to manage all of the documents and other pieces of knowledge associated with the product life cycle development process, including online discussions. A discussion list capability keeps track of discussion threads.

           Pfizer’s Global Research Division intranet has many dozens of applications organized both geographically and functionally for virtually every area and division of the company. They include an internal telephone directory, access to scientific publications, and sharing of research findings across international borders and time zones. Pfizer linked its intranet with an extranet for managing some 500 strategic alliances so its global teams can access legacy data and collaborate on projects more quickly. Researchers can link from the Pfizer intranet to the U.S. Food and Drug Administration Internet site. A tool called E-sub enables the company to access historical data to expedite preparation of the laborious new drug applications (NDAs) required by the FDA.

           The company is moving toward a global approach to information management. In the past, each R&D library would look first in its own collection to locate requested articles. If the articles were not found there, public libraries and resources would be searched. If a requested article was still not found, an outside firm was commissioned to locate the article. Now Pfizer scientists can search the journal collections of each major Pfizer library from a single master list.

           Pfizer adopted Oracle’s Clinical application, which is designed to help pharmaceutical companies bring products to market faster. The software establishes standards and common working practices. Oracle Clinical has a capability for tracking who accesses each piece of data and how and why changes were made. It includes a subsystem for managing data definitions and can flag any data entered during a study that it cannot validate, so researchers can quickly identify problems with the data or the product under development. Definitions and amendments are automatically propagated to all locations.

           Pfizer was one of the pioneers in using advanced information technology for combinatorial chemistry and high through put screening. Combinatorial chemistry enables companies to design, screen, and test compounds very rapidly by using chemistry, molecular biology, and information technology to create and test thousands of chemical combinations at once. Previously, pharmaceutical companies had to evaluate thousands of compounds individually before finding one possible candidate for further development.

           Combinatorial chemistry and highthroughput screening became popular in the early to mid-1990s as a way to accelerate this process. Rather than have chemists cook up each type of molecule by hand, which could take weeks, machines would create thousands of chemicals in a day by mixing and matching common building blocks. Then robots would drop bits of each chemical into tiny vials containing samples of a bodily substance involved in a disease, such as the protein that triggers cholesterol production. A “hit” occurred when the substance and the chemical produced a desired reaction. (The testing process is called high throughput screening.)

           Virtually all the major pharmaceutical companies embraced combinatorial chemistry and high-throughput screening, spending tens of millions of dollars forming alliances with smaller companies that specialized in this technology. Between 1995 and 2000, Pfizer entered into 36 alliances with 29 different companies in combinatorial chemistry alone, and the number rises to 50 if you include Pfizer’s acquisitions of Warner Lambert and Agouron.

           Intelligent machines churned out chemical after chemical, but almost none produced useful results. Often the machines threw so many ingredients together that the resulting chemicals were too “large” from a molecular standpoint. They would work in a test tube but would get broken down too easily in the human stomach. In one case a drug that prevented infection showed promising results in a test tube, but could not dissolve in water, which is required for intravenous drips. When chemicals were made individually, chemists usually dealt with such issues during the initial stages of development.

           According to Carl Decicco, head of discovery chemistry at Bristol-Myers, many chemists became fixated on creating thousands or millions of chemicals for testing without thinking about whether any of them had any real use. “You end up making things that you can make, rather than what you should make,” he says. Countless combinations of potential druglike chemicals are theoretically possible, but most of these combinations are really useless to humans. Pfizer senior research fellow Carl Lipinski, who retired in 2002, compiled a list of complex technical traits that often make chemicals difficult for humans to absorb and persuaded Pfizer to reprogram its computers so chemists would be warned if chemicals violated the “Lipinski rule.”

           Critics of combinatorial chemistry and high-throughput screening point out that these methods lack human insight, intuition, and intellectual creativity. Opponents believe these methods eliminate opportunities for serendipitous discovery. For example, in 1991 Schering-Plough scientists were looking for a drug to block a certain cholesterol-producing enzyme in the body. During a test on hamsters, they noticed that one molecule failed to block the enzyme but nevertheless lowered cholesterol. Some additional hand-tweaking by chemists turned the molecule into the cholesterol-lowering drug Zetia, which was approved by the FDA in 2002. If a robot had tested the molecule in a test tube, it would have noted the failure but would have missed its serendipitous side effect.

           Because robot screeners can work only with liquids, the huge chemical libraries created by combinatorial chemistry and high-throughput screening are often placed in dimethyl sulfoxide, a standard solution for storing chemicals. In some cases the chemicals settle as a solid at the bottom of the solution or the solution containing the chemical breaks down. The drug-testing robot reaching into such mixtures may only come up with a drop of useless soup. Traditional labs avoid this problem by storing chemicals that might break down in dimethyl sulfoxide as powders, which are put into solution just before screening.

           Pfizer and the other major pharmaceutical companies are trying to rectify these problems. Pfizer spent over $600 million at labs around the world to ensure that the chemicals in its libraries are more druglike and diverse. It is using techniques other than combinatorial chemistry and making sure each chemical can meet Lipinski’s test. Martin Mackay, a senior vice president at Pfizer’s research labs, reports that a higher percentage of compounds at Pfizer are now making it through each stage of testing but that it will take 10 years to tell whether efforts to improve the technology are working. “We’re very confident,” he says.

           Other scientists echo his belief that the industry has solved its early problems with combinatorial chemistry and high-throughput screening and that the pipelines will be filled with new drugs created by these methods a decade from now. “It took a while to learn how to use all these new technologies,” says Richard Gregg, vice president of clinical discovery at Bristol-Myers research labs.

           A study led by David Newman of the National Cancer Institute concluded that combinatorial chemistry and high-throughput screening had failed to create a single FDA-approved drug through the end of 2002. A separate study of 350 cancer drugs now in human trials found only one that had been created with these methods, although the technology did help improve some drugs that were created by more traditional means.

           Some observers believe that pharmaceutical firms’ widespread use of combinatorial chemistry and highthroughput screening is one reason why there is such a dearth of new drugs today. The number of new drugs approved by the FDA each year has declined since 1996. In 2003, the FDA approved only 21 new drugs (of which one was produced by Pfizer and one by Agouron), compared to 56 in 1996.

Sources: Peter Landers, “Drug Industry’s Big Push into Technology Falls Short,” Wall Street Journal, February 24, 2004; Madanmohan Rao, “Leveraging Pharmaceutical Knowledge,” Knowledge Management, March 2003; www.pfizer.com, accessed June 10, 2004; Kim Ann Zimmermann, “In Search of Experts: Pharmaceuticals Enter Next Phase of KM,” KWorld, January 2003; Helene S. Gidley, “Hand in Hand,” PM Network, August 2003; and Stephen S. Hall, “Revitalizing Drug Discovery,” Technology Review, October 2003.

CASE STUDY QUESTIONS