Sophisticated computer technology, long used in engineering, is starting to tackle the body's complexities, with the promise of a payoff for drugmakers
PREMIUM SEARCH Search by job title, geography and build a list of executive contacts
What do a Boeing 777 and the human body have in common? Both are complex systems, dependent on millions of complex parts, whether they be a jet-propelled engine or a pumping organ such as the heart. The big difference: Engineers can design and build highly accurate computer models of the way a Boeing 777 will behave in flight. The human heart? Its complexity has long stymied efforts by researchers intent on turning drug development into a predictive science, much like building airplanes.
But that's changing. A handful of companies are developing software that can model single cells, whole organs, cellular metabolism and toxicology, diseases throughout a patient's body, and even an entire clinical trial. These computer models could speed up research and drug development by avoiding unnecessary experiments and dead ends. That, in turn, could save drug companies millions, or even billions. A recent PricewaterhouseCoopers report, "Silicon Rally: The Race to e-R&D," estimates that new computer-based technologies could save the pharmaceutical industry two to three years of effort and $200 million in development costs per drug.
"What modeling allows you to do is get to where the real problem is: understanding how genes impact function," says Thomas Colatsky, executive vice-president of Physiome Sciences, a Princeton (N.J.) company that builds organ-modeling software.
PREDICTIVE INFO. Most pharmaceutical companies already use computers to mine genetic data and identify new drug targets, a technique called bioinformatics. Many also use cheminformatics: processes that tap information on a drug's molecular structure to predict its toxicity or its ability to fit into the active site of an enzyme or a molecular receptor. Pharmaceutical companies have also taken to models that predict a drug's absorption or metabolism, with many of the companies that do modeling working in this area.
But these technologies can't take a big-picture view of the body. For example, a bioinformatics program could find the gene that turns an enzyme regulating the thyroid hormones on or off. Yet the program can't map out the biochemical impacts of turning this gene on or off, for example, or how it might affect appetite in a patient over time.
In that sense, the pharmaceutical industry is 20 years behind the engineering industries, according to Steve Arlington, a partner at PricewaterhouseCoopers. "Today, an engineering company wouldn't dream of doing their development without using computer-aided design and testing facilities," he says. "The pharmaceutical industry is one of the most conservative industries I know." In fact, the Boeing 777 was designed almost entirely by computer simulation with technology available for two decades, Arlington points out. And a computer even flew the prototype aircraft for millions of miles before a human pilot ever entered the cockpit.
THE BIG PICTURE. Truth is, biological data are much harder to organize and collect than the qualitative data that simulated-computer-flight programs use to design configurations of materials such as carbon, steel, or titanium. Take Physiome Sciences. The company needed data from decades of experiments to generate models of the heart ranging from a single cell to the entire organ in three dimensions. Only now can the heart models take into account risk factors such as gender or differences in genetic background.
Other companies are trying to create even bigger-picture models. Entelos -- a Menlo Park (Calif.) company that counts Aventis, Astra-Zeneca, Johnson & Johnson, and Bayer among its clients -- is building a computer model designed to synthesize all the biological systems known to contribute to diseases such as asthma, obesity, and diabetes. The company says it can configure its models for "virtual" patients, taking into account variables such as genetic background or behavior, for example. "This is a way that on the computer, many times a day, you can run virtual clinical trials and experiments on patients that would take weeks, months, years to run in the real world," says Entelos' chief technology officer, Alex Bangs.
The models also can be used to streamline lengthy drug trials. PharSight, based in Mountain View, Calif., says it can take a pharmaceutical company's existing preclinical or clinical data and information on related drugs to design more efficient drug trials. PharSight claims that 18 of the top 20 pharmaceutical companies have used its product.
READY FOR PRIME TIME? Most of these modeling companies are still private, so it's difficult to say what they're worth. And critics wonder whether any of them are ready for prime time. They question whether the models can, in fact, capture enough of the complexities inherent in biological systems. PharSight's Initial public offering last August got a lukewarm response from Wall Street, with the stock only bumping up slightly from its asking price of $10 per share. Currently, PharSight's stock hovers at around $7 a share, nearly 30% below the initial IPO price.
But Jim McCamant, editor of the Medical Technology Stock Letter, thinks PharSight and its ilk have a bright future. "These are some of the most cutting-edge companies," he says. "If investors can bid dot-coms up, they certainly can bid up companies that are providing real tools and real software." It may never be as effective as flight simulation, but the promise is there.
By Alka Agrawal in New York Edited by Alex Salkever
Copyright 2000, by The McGraw-Hill Companies Inc. All rights reserved.
Printer-Friendly Version
