Under The Weather? A Drop Of Blood Can Tell If Antibiotics Are Needed
Updated: Jan 17, 2020
By Judy Stone
Published in Forbes
An exciting new test might help us save antibiotics. Needing only a drop of blood, researchers at Duke have developed a rapid assay that can tell viral infections—which generally can’t be treated—from bacterial ones that may benefit from antibiotics.
While I don’t usually write about drugs or products in development, this test piqued my interest and left me excited about its potential to help contain antibiotic overuse.
Acute respiratory tract infections—a “cold,” bronchitis and pneumonia—caused 71 million patients to visit a healthcare provider (2007). In almost 3/4 of those visits, antibiotics were inappropriately prescribed, accounting for 41% of antibiotics prescribed in ambulatory care settings.
As many people have likely heard, we are facing a crisis in superbugs, with many bacteria becoming resistant to multiple antibiotics. Just last month, the new mcr-1 gene was discovered; this gene confers resistance to the last of our effective antibiotics for serious Gram negative infections like carbapenem-resistant Enterobacter (CRE).
When you go to the doctor now with a sore throat, a culture for Streptococcus takes at least a day. There is a rapid test, but it can miss too many cases, so a culture is done for confirmation. If you seek care for bronchitis, a sinus infection or “walking pneumonia,” no cultures are done and, too often, you will leave with a prescription for antibiotics.
If you are hospitalized with pneumonia, an array of blood and sputum cultures are ordered.
There can be diagnostic delays of a day or more if you can’t cough up a deep sputum sample, and cultures usually take two to three days to tell anything.
With this new test, a drop of blood can distinguish between bacterial or viral infections, or a “systemic inflammatory response” (fever, rapid heart rate, etc.) due to other non-infectious diseases. Instead of relying on time-consuming cultures, this test looks at the person’s response to infection, using a “host gene expression classification.” Bacteria, viruses—and even some underlying diseases— each elicit a unique immune response. The patterns of these gene expression signatures tell if specific genes are turned on or off. Examining 25,000 genes at a time, these gene expression signatures can differentiate between these causes of illness and tell if antibiotics will be useless and shouldn’t be prescribed. Furthermore, it should save many patients from serious allergic reactions or C. difficile infections, which increasingly complicate antibiotic treatment and can be life-threatening.
Even if this assay or currently available rapid tests diagnose influenza, treatment with an antiviral like Tamiflu is currently not recommended unless the individual is immunocompromised, seriously ill or pregnant. Again, just as with antibiotics, overuse promotes resistance; Tamiflu also has very limited efficacy and causes nausea and vomiting.
In their study, Ephraim Tsalik, M.D., Ph.D., of Duke University and his colleagues examined 273 emergency room patients presenting with acute respiratory infections (ARI) in Durham, Chapel Hill and Detroit. A clinical assessment (phenotype) of the cause of the patient’s illness was done by chart review before the gene expression classification was made. I was impressed with the level of care in assessments and the detail provided in the paper, particularly for patients with discordant (where the clinical assessment and the gene assay didn’t match) results.
Validation of Bacterial and Viral ARI Classifiers in GSE6269.
There are some assays, like a blood test called procalcitonin, that can help distinguish between bacterial and viral infections. The gene expression classifier was more accurate than procalcitonin (86% vs. 78%) In my experience, working in community hospitals, the procalcitonin level is not helpful, as results take at least three days to return. (Turnaround times are sometimes faster at a large or university hospital.) The patient is likely to be discharged by then, or dead. The host gene classifier assay currently takes ten hours. Dr. Tsalik believes the turn around time can be reduced to one hour, making it suitable for clinic or emergency room use. He is aiming for the assay to be ready to go to the FDA for approval in about a year.
Besides the obvious, that this study has to be replicated in larger populations and that the assay requires refinement, my experience suggests several other problems. A huge barrier is likely to be the cost and accessibility of the assay, not only the direct costs, but the cost of taking time off and of the visit itself. While this should be more than offset by the direct savings from getting the diagnosis more or less right the first time (fewer return visits, decreased morbidity/mortality) and indirect savings from, one hopes, more rational antibiotic use and decreased resistance, I’m not certain patients or insurers will view it that way.
Offices are often inundated in flu season and too often physicians call in prescriptions for demanding patients (remember that all important patient satisfaction sometimes trumps quality care). Patients also sometimes say they feel too ill to be seen and evaluated. These will all limit the utility of the gene expression assay.
What was particularly exciting to me is that this new assay was able to rapidly distinguish noninfectious illnesses from bacterial or viral infections—the first time this has been done at the molecular level. That such blood signatures can be remarkably accurate for respiratory infections was quite striking. As the authors note, the host responses to bacterial, viral or noninfectious insults (injuries) are unique and quantifiable (in terms of how likely you have the condition your gene expression signature suggests you have). Further, the assay was able to help determine the severity of the disease and can even be used to detect viral illnesses like influenza before symptoms arise.
The Task Force for Combating Antibiotic-Resistant Bacteria was created by President Obama in 2014 because of the urgency of the antibiotic resistance threat, to prioritize the development of new diagnostics. There is also the Antibacterial Resistance Leadership Group (ARLG, an NIH funded consortium that this study was a part of. Funding for this study came from DARPA*, NIH, AHRQ, the VA, and in-kind donations from bioMérieux, Inc.
As I previously reported, a related and important issue are host dependent factors that lead to antibiotic resistance. Dr. Michael Mahan and colleagues at UCSB nicely showed that antibiotic susceptibility is dependent on conditions within an intracellular environment, including pH and iron concentrations—something that is overlooked with the traditional Mueller-Hinton broth testing. He noted, “Our research suggests the need for animal models to be incorporated early during the antibiotic development process and for lab drug sensitivity testing to incorporate media that mimic the specific biochemical environments that trigger resistance in the body.”
If we could combine Tsalik’s new approach to identifying who needs antibiotics with Mahan’s strategy of considering intracellular conditions or microenvironments within the body when doing antibiotic susceptibility testing, we could make huge inroads in improving appropriate antibiotic use. Dr. Tsalik shared his enthusiasm with me, explaining that they ‘took an unanswered clinical question – how do we know whether a patient has the type of infection that needs antibiotic treatment?” then designed a study “that led to incredibly insightful answers about how the patient’s biology responds to the stress of infection.” Now they are “translating those results into something providers can use at the bedside.” I share his excitement.
For more background: Shifting the Paradigm
* U.S. Defense Advanced Research Projects Agency, NIH, the Agency for Healthcare Research and Quality, the U.S. Department of Veterans Affairs Office of Research and Development, and an inkind contribution bioMérieux, Inc.