When most of us get sick, we can simply call off work for the day, or even two. But for others that's not quite so easy -- take, for example, soldiers out on military operations that depend on each member of the team to perform at their peak. There's no time for a soldier to take a break for the common cold or flu, and certainly no time for an epidemic of illness to take hold and spread throughout the group.
"If you have a battalion of 100 guys and one gets sick as you are ready to ship off, should you go forward or should you hold back?" asked IGSP Member Chris Woods, a physician and associate professor of medicine in the infectious diseases division. "You want to know if others are incubating a virus."
Detecting those bugs before illness becomes obvious would allow more timely treatment with anti-viral drugs, Woods added, noting that earlier treatment in the course of illness is known to result in a better response.
And that's why the Defense Advanced Research Projects Agency, aka DARPA -- the same agency that invented the internet and GPS -- issued a call back in May 2006 for research proposals aimed at developing a tool that could predict the onset of infectious diseases before symptoms appear.
Heeding the Call
At Duke, Woods and Geoff Ginsburg, director of the IGSP's Center for Genomic Medicine, had already been tossing around ideas for collaboration. As an infectious diseases expert, Woods had done some work aimed at applying genomics to tease out which people with fever -- a rather nonspecific symptom -- actually have a severe, even fatal, infection and which are only mildly ill. As Woods likes to say, all fevers are local, and it's not so easy to tell who has malaria, tuberculosis or a run-of-the-mill cold.
Meanwhile, Ginsburg's overall strategy at Duke was to develop molecular signatures that can classify and make predictions about complex disease, including cardiovascular disease and cancer, and apply them in settings where they could be medically useful. Those efforts led to the creation of the Clinical Genomics Studies Unit (CGSU), a group now running several clinical trials to test the use of genomic signatures for guiding cancer therapy.
Ginsburg learned of DARPA's call as part of their "Predicting Health and Disease" program from Larry Carin, an electrical engineer at Duke's Pratt School of Engineering who is now a newly minted member of the IGSP. Carin had done quite a bit of work with DARPA before, developing analytical methods for remote sensing of landmines and other applications.
"We try to represent complicated data in a compact way," Carin explained. "Even if the data are complicated, there's usually a simple explanation for what is going on." They suspected that notion would also apply to the genomic response to viral infection, yielding genomic signatures similar to those that the IGSP team had developed for other forms of disease.
"We had the technology and the infrastructure in place and a long history of developing innovative statistical methods for teasing apart complex data and building useful predictors as we've done in cancer," Ginsburg added. "This was an opportunity to apply those methods in the completely new area of infectious diseases. There was some evidence it could be done, but no one had taken it to the next level."
It was a prime opportunity to bring together the multidisciplinary talent at Duke, under the auspices of the IGSP and its CGSU, toward significant advances in the management of infection among the military and in the health care sector as a whole, he added. Woods, Ginsburg and Carin also enlisted IGSP Member Aimee Zaas, another expert in infectious diseases who could help bridge any statistical findings with their biological relevance, along with several others, to form the interdisciplinary team they'd need to answer DARPA's call.
An Adventure Begins
The Duke team put together what was a bold proposal: over the course of a single year, they would conduct "human challenge studies" with live virus. This meant that they would infect 20 or so healthy volunteers with one of several common cold and flu viruses and collect blood, urine, breath and nasal wash samples from each exposed individual as they developed the illness or not over the course of two weeks. Those samples would then be analyzed for gene activity, along with metabolites and proteins.
Their ultimate goal was to find a "signature" in that expression data that could predict who would later become sick before the first cough or sneeze. Those signatures developed via the challenge study would then be tested in a "real world" setting -- among Duke students during the cold and flu season.
"When you give someone these viruses, you expect to see a spike as gene expression changes," said Joe Lucas, an IGSP Investigator who is working with the team. "Ultimately, you want to be able to ‘see' that spike coming before it happens. You want to push the identification as early as possible."
Carin explains it this way: While there may be 20,000 genes or other markers that change in one way or another following viral exposure, only a tiny fraction of those are likely in play as a direct result of the virus. Their statistical methods coud be used to pull out those "needles in a haystack," he said, sorting those genes that move in an essentially random way with respect to the infection from those that rise or fall in a coherent and meaningful fashion.
DARPA awarded them more than $6 million for two years, and Ginsburg recruited Tim Veldman, now Assistant Director of Operations for the IGSP's Center for Genomic Medicine, to help "glue" the team together and ensure their success.
Within months, the first rhinovirus challenge study was underway with collaborator Ron Turner at the University of Virginia. Under Veldman's leadership, the CGSU team applied sophisticated coding methods and tracking systems to keep all the various biological samples -- collected from each of 20 participants every four hours -- in perfect order for entry into the IGSP's biorepository for storage.
Despite that, at times it still "felt like sheer chaos," Woods said. "It must be what it's like to run a startup company. I can't imagine anything more exciting."
An Illness Foretold
Less than two months later, they had data back from the blood samples they took, showing that they could make a genome-based prediction about the onset of cold symptoms with a very high rate of success .
"The data are just fantastic," Ginsburg says. "I could not have imagined a better scientific outcome. We can predict with a great degree of accuracy who will get a cold days before they develop their first symptoms."
The team had already exceeded DARPA's expectations, and the agency was interested in other viruses as well. They struck a deal with a company in the U.K. called Retroscreen Virology Ltd. to do two more challenge studies, first with respiratory syncytial virus and later with the influenza virus.
Those studies too are now complete and, while the team isn't yet ready to divulge all the details, early indications are that the blood profiles are good predictors of impending illness there as well.
The team is now turning to the other samples they've collected -- the breath condensate, nasal washes and urine -- to see if they can land a useful signature in a medium that might be less trouble than blood to collect. They also plan to validate the signatures in undergraduate living groups at Duke, electronically tracking the onset of illness in several households and then monitoring others in the house for the early signs of sickness. Ultimately, they want to see just how soon after viral exposure their future-telling can go.
"If I can tell you are sick based on the underlying biology but you are already sneezing, that wouldn't be useful," Carin said. "The real challenge is being able to tell you that you will get sick tomorrow, and we think we can do that. The question still is how far back can we push it. We'd like to be able to tell you when you are still feeling good."