Tim Sweeney had been a surgeon for only one year, but he was already fed up with a seemingly unsolvable and dangerous problem. Many patients who underwent a major operation or trauma seemed unwell after, with a fever and a high heart rate. Many looked like they had a postoperative infection, though Sweeney knew the actual rate of those was just 5 to 10 percent. Doctors can give powerful medicine just in case, but that carries its own risks to the patient and contributes to the rise of drug-resistant pathogens. “You can’t just go around treating everyone with antibiotics,” Sweeney told me. “It’s bad for them and it’s bad for the system.”
This conundrum set Sweeney on a mission. In 2011, he joined the lab of Purvesh Khatri, a computational immunologist at Stanford, in hopes of developing a new kind of test that might help surgeons sort out which patients really need treatment. Traditionally, doctors identify infections by looking for dangerous viruses or bacteria that might be circulating in the body—either by culturing a sample to see if the offending microbe is there or, more recently, looking for its genetic markers. But some pathogens can be hard to find in patient samples, or they’re rare enough that standard tests fail to pick them up. Sweeney and Khatri had a different notion, though. In May 2015, they published a paper describing how the genes in our immune system behave in response to sepsis, the life-threatening condition that may result from an infection. A year later, the duo, along with a third partner, founded a company called Inflammatix, which became one of a growing number of start-ups trying to develop diagnostic tools that detect infection by measuring exactly this sort of activity.
As a starting point, Sweeney hopes to offer a test that can discern between bacterial and viral infections. When someone is very sick, doctors can’t always tell which class of pathogen is to blame, at least at first. (One 2015 study in The New England Journal of Medicine found that among patients with confirmed pneumonia, the source of illness remained unknown in 62 percent of cases.) Still, decisions must be made: If the infection is bacterial, then antibiotics can be lifesaving; if the patient has a virus, doctors may consider using antivirals. In August, Inflammatix reported modest success with a test that uses a blood sample to measure the activity level of 29 immune-system genes, then calculates the probability that an infection is bacterial or viral. Other groups have reported similar advances. Last year, a group that included Ephraim Tsalik of the Duke University School of Medicine announced a study showing that it could distinguish with reasonable precision between bacterial and viral infections in less than an hour by looking at a person’s immune-system response.
The idea of assessing the body’s reaction to an illness for diagnostic clues is nothing new—it is, after all, the basis for antibody tests, which may indicate whether a person was infected with a particular microbe, such as SARS-CoV-2. But antibodies are generated only in the days and weeks after an infection. Scientists such as Sweeney and Tsalik are trying something different: They’re looking at the behavior of immune-system genes in order to diagnose an active infection. The approach “has been one of the most exciting areas of research on diagnostic testing over the last decade,” Paul Drain, an infectious-disease physician and associate professor at the University of Washington School of Medicine, told me. If immune-system gene testing works in real-world settings—and accurately informs doctors about which infections are viral versus bacterial—it “would have major implications in primary-care clinics by reducing overuse of antibiotics.”
There might even be a way to extend the value of this approach from early diagnosis of an infection to prognosis. Mahdad Noursadeghi, an infectious-disease professor at University College London, told me that measuring immune-system genes’ activity might one day allow clinicians to predict how patients will fare in response to pathogens, such as whether they’re likely to develop severe disease.
In the meantime, immune-system-gene diagnostics are being tried for many illnesses. Researchers showed in 2013 that the strategy could offer an early-detection method for influenza. A gene-activity signature was detectable as soon as 29 hours after influenza exposure, they found, and more than a day and a half before symptoms peaked.
Khatri, Sweeney’s collaborator, has spent years working on a link between immune-system markers and active tuberculosis, which claims an estimated 1.5 million lives every year. Globally, about one in four individuals has latent tuberculosis. Khatri wants to be able to detect when the pathogen has become reactivated in a host. In 2016, he and his teammates published a paper that described how the activity of three genes, measured via blood draw, could serve as an early-alarm system. A couple of years later, the group reported that the same test could predict who would develop active tuberculosis six months before a traditional test could. It’s a better option for practical reasons, too: Traditional tests require patients to cough up sputum for laboratory analysis—a process that can be challenging for children and may itself help spread the disease.
Other researchers have examined the immune-gene response to pink eye, which is caused by a virus, bacteria, or a fungus. Their test shows promise in detecting which cases are caused by a virus and which are caused by a fungus. (It did not compare infections from bacteria.)
Immune-system-gene tests may even have some value in spotting COVID-19. In February, Tsalik and some of the same researchers who worked on spotting flu signatures reported that the immune-system response to a SARS-CoV-2 infection might have unique characteristics. That gave them hope that a diagnostic test for the coronavirus could be formed based on the host immune response. Tsalik, who co-founded a diagnostic biotech start-up, says that developing a COVID-specific test based on this approach is not an area of active investment, though he and his collaborators did describe their hope, in a paper from last year, for “a new generation of host-based diagnostics to combat this devastating disease.”
Noursadeghi has been on the same track. He and his colleagues studied health-care workers at London’s St. Bartholomew’s Hospital, taking weekly blood samples and nasal swabs for SARS-CoV-2. By comparing the workers’ known COVID status with their immune-gene activity, they were able to zero in on a particular gene, IFI27. By measuring the activity of that gene, which is involved in an inflammatory response against pathogens, they were able to create a COVID test that isn’t that much worse than a PCR test, when given at the same time. More striking was the fact that the test for IFI27 activity was able to pick up a little less than half of SARS-CoV-2 infections at least one day—and possibly up to a week—before the first positive PCR test.
Mala Maini, a co-author on that paper and an immunology professor at University College London, says that their work also hints that the host immune response might reveal infections that a PCR test would never see. The team identified a subset of individuals exposed to SARS-CoV-2 who kept testing negative on PCR tests but showed an increase in IFI27 activity. Further lab tests found that these patients’ T cells, as compared with T cells from people who had never been infected, were more responsive to the virus. According to Maini, that activity may have quashed the infection before SARS-CoV-2 became entrenched enough to show up in the standard tests.
That immune-system signal—the boost in IFI27 activity—is not specific to the coronavirus, though. It also appears in response to other infections, including rhinovirus and influenza, Noursadeghi said, and that lack of specificity makes it much less useful as a diagnostic. On the upside, though, viral variants with new mutations may be less likely to evade an immune-system test than one based on antigens or PCR. Tests that look at a person’s immune response might even catch some brand-new pathogenic species that scientists have never seen before.
Many scientists see tests of the host immune response as a complementary approach to existing testing technologies. Some researchers are bolder in their vision. Khatri is convinced that tests to track the host immune response will become far more widespread. “I think 10 years from now we will not be doing nasal swabs,” he told me. “We will be doing tests like this.” Yet there’s reason to doubt whether they would work for everyone. People who are immunocompromised or have a chronic inflammatory disease might not show the same immune-gene signatures in response to infection as the general population. The same could hold for infants and seniors too. Even so, a new method of testing would be a welcome addition to any infectious-disease doctor’s arsenal, and could provide another way to confirm a diagnosis as well as greater confidence in knowing when you’re really sick. In a pandemic rife with uncertainty, that could help.