Sharing Drug Data

William Gibson, the science fiction writer, once wrote that the future is already here-it just isn't very evenly distributed yet.

This is the kind of idea that information technology (IT) companies readily embrace. Mike Svinte, vice president for marketing and business development at IBM Life Sciences, says he likes Gibson's notion because it suggests a handful of isolated IT trendsetters whose efforts give us a window onto the future of the state of the art.

He points to IBM clients, such as the Mayo Clinic and Memorial Sloan-Kettering Cancer Center, that are forging the kind of collaborative research support systems that he says will underpin drug discovery in the era of personalized medicine-the genomics-driven effort to target medications to specific populations or even individual patients.

But if the future of IT for drug and health care organizations is unevenly distributed, the information that needs to be harnessed, vetted, and shared in these sectors is flying around in a state of ever-mounting chaos, according to Svinte and other IT vendors and users. Making matters worse, the drug and health care sectors are, for the most part, living in the past compared with other industries when it comes to data management systems.

The reasons for the lag have been debated for years. Some observers point to an unwillingness by scientists to share information or conform to centrally controlled IT regimens. Others note the complexity of the information that needs to be managed in science-based businesses. Still others say the technology required to manage contemporary drug discovery and development is still being developed.

Everyone agrees, however, that the influx of research methodologies such as genomics, proteomics, microfluidics, and biobanking-the growing effort to collect and catalog tissue samples-increases drug researchers' need for a collaborative data management and research network.

IT in health care is currently advancing on two fronts. One is an effort to develop tools and communication standards that allow researchers to share information within their own organizations and with others in their field. Most of the technology needed to do this is already available. The other is an effort to configure research systems so that chemists and other scientists can mix and match applications on desktop computers without the help of their IT departments or third-party consultants.

Here, new software and systems are being introduced. Tripos, a chemical informatics company that runs its own drug discovery operation in Bude, England, recently introduced a line of desktop research software called Benchware. The product line, including laboratory notebooks, was developed by Tripos at Bude and in collaboration with clients such as Schering-Plough and Pfizer.

According to Tripos Chief Executive Officer John McAllister, Benchware is largely a research variation on the business process management (BPM) software installed in most industries in the 1990s. By "tagging" data with a standard set of attributes, Benchware software allows chemists to move information between applications seamlessly. More important, it establishes a standardized work process that expedites collaborative R&D, he says.

"It is appropriate to view discovery in the same way as other industrial activities that have been more process-focused," McAllister says. "At this point, the activity in laboratories is more serendipitous, akin to academic research." Not that there is a problem with academic research, he points out. "Certainly no one wants to take creativity out of discovery research," he says.

Yike Guo, CEO of software firm InforSense, agrees. InforSense, which has collaborated on system development with Tripos, manufactures workflow software for research based on the BPM concept. The company's Knowledge Discovery Environment software accommodates the frequent reconfigurations that occur in scientific research, Guo says. It also lets the scientist do the reconfiguring.

Guo says the researcher's lack of control over laboratory IT has resulted in much of the information hoarding that he and others see impeding progress in drug discovery. "If the scientific process was represented as a bunch of computer programs, there would be no sharing," he says. "People would code it, and other people would find it unreadable." What is needed, he says, are graphic workflow programs that record the "who, what, where, and when" of data generation and produce sharable diagrams.

With the ubiquitous presence of the Internet, the tools for sharing information are increasingly accessible.

"The technology exists to do it," Guo maintains. "The reason it never caught on in this industry is that scientists don't compose processes on computers. They hire others to do it for them. They need a tool they can use by themselves, rather than relying on IT people to encode their processes. There are very big gaps between the users and providers of information."

Closing that gap will require research labs to reorganize staff and rethink some basic tenets of drug discovery. And there couldn't be a better time to make those changes, according to IBM's Svinte, given the influx of new science in research. With the advent of translational medicine, biobanking, and personalized medicine, the health care industry is ripe for new IT tools and an overhauled infrastructure, he argues.

IBM recently announced an agreement to assist Memorial Sloan-Kettering in the development of a clinical research management system. The deal includes a $3 million grant of technology and services from IBM aimed at linking Sloan-Kettering hospital data with text mining and related analytical capabilities into a unified information management system for researchers and clinicians.

At a recent biobanking conference hosted by IBM in Washington, D.C., Patricia C. Skarulis, Sloan-Kettering's vice president of information management, said the center has a data management system comprised of equipment from several vendors connected via 170 custom links. "We have become expert at doing interfaces," Skarulis said.

Svinte says the fact that institutions such as Sloan-Kettering have developed elegant, but isolated, networks demonstrates the "natural progression" toward a more global data infrastructure for research. "Part one is getting your own institution in order," he says. "Then you think about how you extend from there to reach an integrated delivery network, to what the government is now discussing: a national health information infrastructure."

Such a national system is included in the National Institutes of Health's Roadmap for Medical Research and is getting a big push from the National Cancer Institute. NCI's Cancer Biomedical Informatics Grid (caBIG) is pooling resources from research institutions, drug companies, and IT suppliers to build an IT framework for cancer research, according to Kenneth H. Buetow, who is director of NCI's Center for Bioinformatics.

Buetow says caBIG aims to foster interoperability among data systems in discovery, specimen collection, clinical trials, and in vivo imaging. "It is not only a bench-to-bedside model," Buetow says. "It is about development cycles where the output of clinic research feeds directly the next generation of basic science experiments."

NCI is collaborating with the National Cancer Research Institute, which is developing a similar system in England and is exploring partnerships with organizations elsewhere in Europe and in Asia. The institute is also working with large drug companies and the Food & Drug Administration on a related project called the Clinical Research Information Exchange.

"Our goal is to put information technology tools in the hands of research and clinical practitioners," Buetow says, "so that just as they might pull down a reagent bottle from a lab bench, they can pull down information and use a wide range of applications to synthesize or integrate knowledge."

Until recently, Buetow says, the electronic infrastructure necessary to routinely share information has been difficult to master in the lab. But with the ubiquitous presence of the Internet, the tools for sharing information are increasingly accessible on anybody's desktop, he says. As the technology barriers to sharing come down, Buetow says, so will the cultural barriers on both the individual and institutional level.

"As we move caBIG forward, I'm impressed that some of the large research institutions are the most cooperative in recognizing that the types of questions we face in biomedicine today really are beyond the capacity of even the largest institutions," Buetow says.

Still, many research organizations are focused on intramural IT development. For instance, the Windber Research Institute (WRI) in Windber, Pa., is currently putting into place the first phase of its network, according to Michael N. Liebman, the institute's executive director.

Liebman says the network, which uses InforSense software, facilitates a new level of sharing in the kind of data-rich environment now typical in drug research. WRI, he says, has a tissue bank of 17,000 tissue samples, each with about 600 data points. "We are integrating genomics and proteomics assays, gene sequencing, and protein expressions with mammography, ultrasound, and MRI data," Liebman says. "We are building a data model for researchers to find relationships relative to disease progression."

Although the emerging network is compliant with the broad standards adopted by caBIG, Liebman says WRI needs a far more robust and flexible network than the one being developed by NCI. "Any time you have a large effort by committee, you make the common level of exchange very thin," Liebman says. WRI is, however, committed to broad sharing of not only data, but also bench practices, he says. "We are always interested in methods of exchange. That's what science is about."

A global research network has yet to gel, but laboratory users of new desktop software say they can see clear benefit to imposing business applications such as workflow spreadsheets in the lab. Gale Starkey, a senior scientist at St. Louis-based Apath, an antiviral drug discovery company, says the Tripos Benchware software has increased the speed and volume of data management in the lab and aided communication across disciplines in the company.

"This works great for the chemist," Starkey says of the PC-based system. "We generate and store the data, and we don't need the IT department to extract it. It moves things along faster. I can go to a meeting with a diverse group of people and do characterizations on the fly."

Speed is good. Observers agree that facile data management is critical to breakthroughs in medicine. "Many of the issues we have to solve in the next generation of personalized medicine really are only going to be solvable with the advent and routine use of information technology at the desktop," Buetow says.

Likewise, Svinte says, health care is clearly the new frontier in IT systems development. "We probably have the best jobs at IBM: IT for the future of medicine," he says.