PHILADELPHIA — Budget constraints and cultural barriers can present formidable hurdles for informatics managers in the biopharmaceutical industry, but those firms that have closely aligned their IT strategies with their research goals have a better chance at seeing a return on their informatics investment, according to biopharma executives who spoke at a conference here last week.
John Reynders, information officer for discovery and development informatics at Eli Lilly's Lilly Research Labs, had some advice for informatics managers trying to keep pace with shifting research goals, namely, "don't play the organizational game," and "eliminate bureaucracy."
Reynders, who kicked off Bridging Discovery and IT, a conference organized by Bio-IT World and Cambridge Healthtech Institute, said that "it's not the technology or the science that's the challenge" for discovery informatics, "it's the dynamics, it's the people." Rather than build a "bridge between 'us' and 'them,'" as the conference title suggests, Reynders said, decision-makers in discovery and informatics would be better served by "blurring the line" between the two disciplines.
The first step, he said, is for managers to clearly define the company's discovery-wide goals, and then make sure that the goals of the informatics group align with that strategy. Next, he said, "configure the organization by capabilities — not by project or platform." This ensures that informatics staffers remain flexible, and can cut across the infamous research "silos" within pharma.
"It's important to make the IT people feel like discoverers, and make the discovery research staff feel like they are part of the success of the business."
Accountability is a "key aspect" to the system that Reynders has implemented at Lilly. By aligning informatics projects with the company's research objectives, and then assigning staff to those projects, "it empowers people to know they contribute to the strategy," he said.
Another recipe for agility, Reynders said, is to "push the major decision-making as close to the work as possible." At Lilly, high-level management sets budgetary guidelines, but the local informatics project teams have the authority to decide how that money is spent. The result, he said, is quicker — and better — decisions.
Reed Vickerman, executive director of IT at Amylin Pharmaceuticals, echoed Reynders' comments about keeping organizational boundaries fluid. "The org charts are going away," he said of his company's informatics strategy. Rather than the traditional hierarchical organizational model, which can isolate informatics teams under the IT umbrella, Amylin's informatics and IT groups are arranged in a "matrix" structure based on "clusters of activity around business processes," he said.
Vickerman said that informatics teams are often perceived as support staff within drug-discovery firms, while the researchers can sometimes lose sight of the company's business goals. Therefore, "it's important to make the IT people feel like discoverers, and make the discovery research staff feel like they are part of the success of the business." One way that Amylin has addressed this challenge is by placing developers and users side-by-side on informatics development projects, Vickerman said.
The cultural and organizational difficulties of discovery IT are nothing new. "Many of these issues have been coming up for years," said Joseph Cerro, head of informatics research at Bayer Pharmaceuticals. Cerro provided a historical account of his tenure at Bayer that indicated some progress has been made, however. Partly due to the "near-death experience" the firm experienced in 2001 when it withdrew a potential blockbuster and embarked upon an "unprecedented restructuring" to a mid-sized pharma model, the company's informatics organization is now streamlined, efficient, and responsive, Cerro said.
One byproduct of the restructuring effort, Cerro explained, was a renewed focus on "IT governance" as a means of aligning projects with corporate goals. Projects are now prioritized according to necessity, feasibility, and the resources required to accomplish them. In addition, he said, the company has moved from an "allocation-based" budgeting model to an "activity-based" one, in which the discovery group is essentially a "customer" of the informatics group and pays for the costs of the services it uses. This model gives research teams the flexibility to request new tools and systems, but ensures they are aware of the costs involved so that everyone stays on budget.
"A modest informatics contribution that helps advance a research project beats building the next unused 'killer app.'"
Placing Safer Bets
In addition to the organizational challenges of running a productive discovery informatics group, several speakers and attendees noted that shrinking pharma R&D budgets have placed more pressure on return-on-investment analysis, which is notoriously difficult to calculate for informatics.
Rainer Fuchs, vice president of research informatics and operations at Biogen Idec, drove that point home when he asked for a show of hands from attendees who have seen "significant cost savings" in drug discovery from their investment in informatics over the last several years.
Nobody raised a hand.
One reason for that response could be that many firms went on spending sprees at the height of the bioinformatics boom a few years ago, when market research firm Frontline projected that bioinformatics could reduce the cost of drug discovery and development by 33 percent by 2004, and Boston Consulting Group estimated in 2002 that informatics could shave $280 million off the $800 million cost of developing a typical drug.
The industry's disappointment in discovery IT is partly due to the structure of the ROI calculation: If the informatics investment is inflated, as it was during the Human Genome Project, the likelihood of generating a positive return is greatly diminished.
On the other hand, the long development timelines of the pharmaceutical industry make it very difficult — if not impossible — to prove that a new technology has indeed led to a better drug. It will take at least 10 years to prove or disprove Boston Consulting Group's projection, for example.
In the near term, as biopharmaceutical firms are feeling the squeeze of belt-tightening, Fuchs said that the emphasis has shifted to smaller informatics projects with simple ROI calculations.
"It's easy to measure improvements in efficiency," Fuchs said, citing "anecdotal evidence" that most biopharma companies have shifted their informatics investments away from cutting-edge technologies and toward "tried and true" methods. "There's a move away from investing in new technology toward improving the use of current technologies," he said.
As an example, Fuchs cited the "changing mix" of his own informatics budget between 2001 and 2005. The shift has been toward projects that fall under the heading of "operational efficiency" — lab automation, workflow management, LIMS, and other projects with "easily demonstrable benefits" — and away from "strategic" investments that carry more risk and prove much more difficult to calculate ROI.
"A modest informatics contribution that helps advance a research project beats building the next unused 'killer app,'" he said.
During a roundtable discussion on the topic of calculating ROI, attendees agreed with Fuchs. "It's easy to quantify the return when it's about productivity," said Ingrid Akerblom, executive director of research information services at Merck. As an example, she cited a recent project at Merck to install an electronic lab notebook across the chemistry group, for which her team was able to calculate the return on the investment based on time savings, paper savings, and other tangible parameters. However, she said, "Anything beyond productivity is difficult to quantify."
Walt Woltosz, chairman and CEO of Simulations Plus, said that even in cases where software does offer a measurable improvement in productivity, the overall cost of development is unlikely to decrease. As an example, he cited a case where a pharmaceutical firm used his company's GastroPlus simulation software to calculate that the maximum exposure level in toxicity testing in dogs was much lower than the firm had assumed — a discovery that should have reduced the number of dogs it used in its experiments. It turned out that the company broke even in the long run because it kept using the same number of dogs but tested more compounds.
Woltosz said that his company used to pitch clients based on the idea that they would save money by reducing the number of experiments they conduct. "It didn't happen," he said. Now, he said, the selling point for his firm is convincing pharma not that it will run fewer tests, but that it will be able to run "more focused tests" that will generate more valuable data.
Rather than calculate ROI on a per-project or per-tool basis, Stan Letovsky, senior director of computational biology at Millennium Pharmaceuticals, said that his group has "identified points in the pipeline where improved decision-making will have the most impact," and then made sure that it had the right tools in place to expedite those steps.
As one example, Letovsky said, Millennium has "added better statistical methods to all its decision-making steps."
Letovsky described this approach, which he called "knowledge retailing," as an alternative to "knowledge wholesaling" — essentially, the massive data-integration projects that many biopharmaceutical firms have dumped money into over the last several years. These projects, he said, can successfully gather disparate data together on a web page, "but they don't address what's done with the data."
The primary role of discovery IT is that of a "decision-support system," Letovsky said, so a more focused, "retail" application "of appropriate IT tools can deliver far more value than a big IT deployment."
The shift toward a lower-risk informatics strategy doesn't mean that discovery firms aren't trying out new technologies, however. Biogen's Fuchs recommended that informatics managers reserve a modest amount of "play money" within their budgets to invest in unproven technologies like pathway informatics and biosimulation tools.
Even though these methods could be "the next land of unfulfilled promise," he said, some risk will always be necessary to remain competitive in the drug -discovery industry. The difference between 2001 and 2005, however, is that "our expectations have become more reasonable," Fuchs said.
— Bernadette Toner ([email protected]eb.com)