University of Pittsburgh | Pitt Home | Pitt Home | Find People | Contact Us


Photographs by
Ric Evans

Lifesaving software, in development at Pitt, illustrates the tremendous potential of the kind of research called in silico biology. By integrating computer technology with knowledge from the life sciences and mathematics, three Pitt professors have created a computer program that has the potential to identify case by case the best approach for treating deadly conditions such as sepsis.

Anatomy of a Breakthrough

Cindy Gill

 From left to right: Yoram Vodovotz, Gilles Clermont, and Carson Chow
Gilles Clermont, a critical care physician, watches yet another person struggle to live. This time it’s a man in his early 20s. It’s a Wednesday evening in the liver transplant unit of UPMC Presbyterian Hospital, where Clermont sometimes looks after patients. This young man received a new liver days ago, but he’s not responding well. He labors to breathe normally. He has a fever, and his blood pressure is too low.

Clermont’s brow tightens. His blue eyes survey the patient and then the monitoring machines nearby. The man’s heart races, and his blood pressure isn’t rising, even with a fluid drip. Clermont’s years of experience in the intensive care unit (ICU) tell him something more may be happening than what the machines show. If his patient develops septic shock, even the best medical care in the world may not help. Septic shock results from severe sepsis, a baffling condition that amounts to runaway infection. Much about the biology of sepsis remains a mystery.

Within hours, the young man’s fight for life intensifies. He strains to inhale, with the onset of pneumonia. A machine forces air into his lungs, helping him breathe. His kidneys shut down.

Soon, other organs begin to fail. By morning, he succumbs to overwhelming, system-wide infection. “Sepsis is the nemesis of the ICU physician,” Clermont says quietly, with a hint of his French-Canadian roots still there in his soft-spoken voice. The death certificate may indicate pneumonia or heart failure, but Clermont knows more. “Eighty percent of people who die on surgical units will die of sepsis-related conditions,” he says.

For decades, many have sought to understand sepsis, an escalating cycle of infection and tissue damage that, all too often, results in death. In the past 10 years alone, at least nine companies have tried and failed to develop an effective sepsis drug, spending millions of dollars in the process. Severe sepsis ranks among the top 10 causes of death in the United States, and it’s the leading cause of death in the nation’s ICUs. Worldwide, sepsis kills at least 1,400 people every day, and it’s particularly deadly in the critically ill and those with weakened immune systems.

After years filled with long days and nights of watching too many people die, Clermont may be on the verge of having a new medical defense. He and two Pitt colleagues are developing a 21st-century tool with intriguing implications for evaluating new treatments for sepsis.

An assistant professor in Pitt’s Department of Critical Care Medicine, Clermont brings unusual capabilities to the practice of medicine. He graduated at the age of 19 with first-class honors in physics from McGill University in Montreal. Within six years, he not only completed medical school but also earned an MS in physics, with a thesis on complex physical systems. After an internship and residency in Montreal, and several years practicing family medicine, he came to the University of Pittsburgh as a fellow in critical care medicine. He also picked up additional training in epidemiology and biostatistics and won numerous awards and scholarships along the way, including the Leopold Morissette Prize in Hematology/Oncology and a Young Investigator Award from the American College of Chest Physicians. But he never completely abandoned his love of physics. He continued to read scholarly papers and converse with physicists. His background in the mechanics of complex physical systems, coupled with a medical degree, became a perfect starting point for understanding sepsis, even as he continued to watch the illness cut a swath of misery in the ICU.

Clermont’s cue to move beyond theoretical ideas came in the form of an RFP, or request for proposal, from the National Institutes of Health. The RFP announced funding availability for projects that merged medical science and computer technology. Sepsis, a colossal, complex system, was an ideal candidate. Clermont quickly gathered an assortment of physicians, biologists, physicists, and immunologists from Pitt and Carnegie Mellon University to discuss a potential sepsis project. The group included Pitt mathematician Carson Chow.

“I had no idea what was going on,” Chow recalls now about the initial meeting. “I didn’t even know what inflammation was. All I could figure out was that you get infected, you get this thing called sepsis, and you have this runaway cascade of events where your organs fail.” Despite his initial quandary, Chow had ideal credentials to tackle the mysterious sequence of sepsis. He earned a BS degree in engineering science from the University of Toronto and a PhD in theoretical physics from MIT. He later pursued a postdoctoral fellowship in bioengineering and mathematical biology at Boston University, where he used computers to model human posture and brain rhythms. Chow arrived at Pitt as an assistant professor of mathematics in 1998 and has since won an NIH Mentored Research Scientist Development Award and an Alfred P. Sloan Fellowship.

In that first meeting organized by Clermont, Chow began asking questions about the mechanics of sepsis. He soon learned that the condition begins with either bacterial infection or with tissue damage caused by an injury. In either case, bacteria in the tissue break down, producing harmful byproducts called endotoxins. In response, the immune system releases various cytokines, which are protein hormones that usually help destroy invading microorganisms. The cytokines also stimulate the release of other substances, such as nitric oxide, that provoke another wave of heightened immune response. Blood-clotting factors are likely affected by this condition, which hinders oxygen delivery throughout the body and can lead to organ failure. “This becomes a self-sustaining thing,” explains Chow, “where you damage tissue, you induce inflammation, which causes more tissue damage, and it’s this loop. Once I heard this, I could grab onto something.”

Meanwhile, Clermont sent an e-mail to several people around campus. He needed someone who could help decipher the specifics of sepsis, beyond the primitive framework of current understanding. Yoram Vodovotz, an immunologist and associate professor in Pitt’s Department of Surgery, answered the e-mail. Vodovotz graduated with distinction from the University of Wisconsin-Madison with a triple major in biochemistry, molecular biology, and genetics. He earned a PhD in immunology from Cornell University Graduate School of Medical Sciences and then worked in various research labs at the National Cancer Institute in Washington, D.C. He also gained practical and entrepreneurial experience collaborating with drug companies on diseases ranging from cancer to heart disease to arthritis. He, too, has won numerous scholarships and research awards. A professional connection with Timothy Billiar, chair of Pitt’s Department of Surgery, brought him to campus not long before Clermont called that first RFP meeting.

Clermont, Chow, and Vodovotz began to marshal their resources and develop a strategy. Initially, they received lab space, along with sound advice, from Billiar and Mitch Fink, the chair of Pitt’s Department of Critical Care Medicine. Ultimately, they submitted several funding requests and have been awarded more than $1.5 million from NIH and other funding sources to develop aspects of a computer model that would recreate the sepsis-related immune response. With that understanding, the deadly sepsis process might finally be undone.

That can’t happen too quickly. “The incessant growth of the severe sepsis epidemic in this country ought to be cause for grave alarm,” says Derek Angus, an associate professor of critical care medicine at Pitt. During the 32nd Critical Care Congress, held earlier this year in San Antonio, Texas, Angus presented the results of a seven-year Pitt study on severe sepsis, in which he was coinvestigator. The study found that cases of severe sepsis are increasing in number and complexity. “This condition has long been one of medicine’s greatest foes, taking more lives each year than breast, colorectal, pancreatic, and prostate cancer combined,” says Angus.

To create the computer model, they start by reviewing volumes of peer-reviewed data on the immune response in sepsis. Combining their scientific intuition with known scientific findings, they then select the biological parameters that appear to be fundamental in the sepsis process. They feed the data into a computer program consisting of mathematical formulas designed by Chow with help from Rukmini Kumar, a mathematics graduate student. Then they wait to see if the results match what they think will happen. “Math is a language which can be used to describe any phenomenon,” explains Chow. “You don’t need mathematics to say these are all the possibilities, but mathematics can help you catalog these things in a systematic way.”

The trio looks for answers. What happens after the initial tissue damage from bacteria or injury? What are the various levels of endotoxins, cytokines, proteins, and subsequent tissue damage that result over various time periods? Can the computer model predict survival outcome when presented with varying levels and duration of initial infection or injury? These calculations occur within a mathematics environment that simulates factors such as blood pressure and tissue dysfunction. With each successive program run, the three learn more about the inflammatory response in sepsis. Then, in the lab, Vodovotz works with graduate students and others to verify and calibrate the computer model’s findings in mice. This lab work has substantiated the model’s validity not only for sepsis but also for other conditions that provoke severe inflammation.

Since the 1600s and the theories of mathematician Sir Isaac Newton, it has been accepted that all phenomena are subject to constraints that limit the range of possible behaviors. Systematically applying a successive series of constraints enables scientists to define specific behaviors and interactions within a given biological system. In the group’s model, the researchers continually enter new data and learn from trial and error. Failure is a great teacher. In fact, the evolving program fails much more often than it succeeds. The result is an increasingly tighter net of constraints on the possible immune-response outcomes in sepsis.

The idea of using computers to simulate biological systems isn’t new. At the height of the Cold War, in the 1960s, the federal government developed a simple mathematical model to predict radiation damage in humans. As technology advanced, new knowledge in biology and computing flourished. The inception of the Human Genome Project in the early 1990s exponentially increased the feasibility of integrating computers with biomedical data. More recently, this emerging field has been labeled in silico biology, reflecting the melding of silicon-chip-based computer technology with the life sciences. In 2001, the University of Pittsburgh established the Center for Computational Biology and Bioinformatics, directed by Ivet Bahar, a professor of molecular genetics and biochemistry in the School of Medicine. More than 30 Pitt faculty—including Chow and Vodovotz—and nearly 10 graduate students and postdoctoral fellows are affiliated with the center and working on computational biology projects.

The sepsis researchers acknowledge they’re neophytes on the edge of a vast new universe in medical science. Clermont uses the analogy of a master playing chess. The master understands the game and has developed the strategies needed to win. A new player, though, begins only with the basic rules. “We think,” he says, “that we are beginning to understand the simple rules of the game with sepsis—the molecules and some of the basic reactions—but we don’t understand the larger strategies, we don’t understand what makes all the elements work together. We’re still only apprentices.”

Vodovotz, who has a third-degree black belt in the martial art of Aikido—or the “Way of Harmony”—takes this analogy one step farther. He compares the game to the three-dimensional version of chess depicted in the Star Trek television series. “The system is more than just the sum of these parts,” he says. “It’s a game within a game, and you have to keep track of it all, not only what’s happening at the moment with specific elements, but how you have manipulated them and how that has modified the overall process. It goes beyond the simple rules of linear interaction.”

Clermont, Chow, and Vodovotz now have a single computer model that appears to replicate and predict a few elements of the body’s inflammation response in sepsis. The fact that each person’s body is likely to produce different responses, based on an individual’s genetics, only adds to the challenge. “I’m frustrated by this project all the time,” says Chow, who is now an associate professor of mathematics. “The real system is extremely complex. It’s probably got thousands of molecules and cells and friends of cells involved, so how could you possibly model it? And that’s the question we have. Is this possible at all? I still don’t know.”

Sometimes, the three can be found searching for answers in the local Starbucks, at Forbes and Atwood, with a laptop computer dominating a small cafe table. Vodovotz, the optimist, generates ideas and envisions the “big picture.” Chow questions every idea and pushes for specifics. Vodovotz revises his thinking based on Chow’s skepticism. And Clermont referees the process, bringing a quiet, steady approach to the whole affair. Where are they with the model? What has occurred with data and results since they last met? They share information, coalesce their ideas, and determine next steps. That’s the pattern.

No sufficient treatments yet exist for severe sepsis, so any discovery that has some positive effect will likely translate into commercial value. Pitt’s Office of Technology Management (OTM) has helped the trio patent and license a novel computer model. OTM connected the researchers with Pittsburgh-based LaunchCyte, a national biotech development firm that provides seed money and resources for life sciences innovations. After sponsoring expanded research on the model, LaunchCyte embraced the program’s potential and created a company called Immunetrics to commercialize the trio’s work. Pittsburgh entrepreneur Steven Chang is Immunetrics’ CEO.

Chang has a record of success with start-ups, and he’s positive about Immunetrics’ future. Currently, 80 percent of clinical trials for potential new treatments fail and the development of a single new drug can take 15 years and cost $500 million with no guarantee of success. In fact, most candidate drugs never make it to the marketplace. “Drug discovery today is like throwing things at a dart-board when you’re a mile away and you can’t see the dartboard,” says Chang. “Pharmaceutical companies are looking for fail-fast, fail-cheap strategies.”

In silico biology products, such as the team’s computer model, could significantly reduce the cost of new drug development and shorten the treatment evaluation process, getting treatments to the market faster and cheaper.

Currently, clinical trials are the primary method used to evaluate potential new drugs and treatment options in people. A recent trial involving people with Alzheimer’s disease was stopped because 10 percent of the participants developed brain inflammation. What if a computer model could identify that small percentage of people who were likely to develop inflammation and eliminate them as study participants? Might the drug then be successful in treating the other 90 percent of participants who didn’t develop the inflammation—and, therefore, become an effective drug that could help a large number of people?

By summer’s end, Immunetrics hopes to have analyzed the billions of possibilities posed by known immune system responses in sepsis, using super computers and working collaboratively with the trio. This should enable the company to offer a product based on an inflammatory-response model that, if successful, can accurately replicate the immune system’s response in sepsis. Then new drug compounds could be fed into the system for evaluation, and the model could be used to evaluate appropriate patient candidates for clinical trials. If all goes well, Immunetrics will partner with a drug company to use the model in a clinical setting, most likely a late-stage clinical trial, which is where most companies that have developed sepsis drugs have “crashed and burned.” Revenue should be generated initially through licensing and consulting fees and, ultimately, through the sale of a refined, compact model that might eventually be used at the patient’s bedside.

“The risk-reward ratio is high,” says Chang, “and our ability to build a tremendous amount of value in the company in a short time frame is high. We’re not looking at a start-up that’s got a 15-year incubation period. We’re looking to garner revenue within our first year. That’s quite a statement.”

Light filters in from a single tall window above Clermont’s desk on the sixth floor of Scaife Hall. Photographs of each of his four children are arranged in a diamond pattern on one of his blue walls. A photo of his wife, Anne-Marie, an emergency room physician, sits on a shelf. He sometimes burns scented candles in a lantern on his desk, a small measure of comfort in a world often filled with disease and death. For Clermont, success can’t come soon enough. “I see patients die of overwhelming infection all the time,”he says.

Cindy Gill is senior editor of this magazine.

 Top of Page Home | Pitt Home | Past Issues | Contact Us