New Breath Test Could Help Doctors Track Whether Antibiotics Are Working

Researchers developed a breath test that detects bacterial infections by measuring a carbon-13 signal produced when bacteria consume specially tagged sugar compounds injected into the bloodstream.

In infected mice, the signal tracked bacterial load in real time and dropped sharply after antibiotic treatment, suggesting the test could tell doctors whether a drug is actually working.

The approach successfully detected infections in multiple tissues, including muscle, lungs, bone, and blood, using portable, inexpensive equipment. All findings are currently from animal studies, and significant testing in humans is still needed before the method could reach clinical use. Antibiotic resistance is one of the most serious threats in modern medicine, and much of the problem traces back to a basic diagnostic failure: doctors often can’t confirm whether a bacterial infection is actually present, let alone whether the drugs they’ve prescribed are doing anything. Millions of antibiotic courses are handed out unnecessarily each year because physicians lack fast, reliable tools to tell a bacterial illness apart from a viral one. Now, researchers say a simple breath test could fill that gap, giving doctors a near real-time read on whether bacteria are alive and active inside the body.

Scientists at the University of California, San Francisco, and St. Jude Children’s Research Hospital showed, in a new study published in ACS Central Science, that bacteria living inside a mammal’s body will consume certain sugar compounds injected into the bloodstream and release a detectable form of carbon dioxide in exhaled breath. Healthy animals given the same sugars produce little to no signal. In infected mice, the signal rose with bacterial load and dropped sharply after antibiotic treatment cut live bacteria by a factor of 100,000.

That treatment-tracking ability is where the approach gets genuinely interesting. X-rays, CT scans, and MRI generally can’t show in real time whether antibiotics are working. Tissue damage visible on a scan often lingers long after bacteria have been cleared, and early signs of improvement at the microbial level don’t show up on imaging for days. A breath test tied to living bacterial activity could offer something those tools simply can’t.

Bacteria consume certain sugars that human cells largely leave alone. Researchers took advantage of that divide by using versions of three compounds, maltose, mannitol, and arabinose, enriched with carbon-13, a heavier, naturally occurring form of carbon. These particular sugars were chosen because human cells don’t significantly break them down, meaning most of the signal comes from bacteria, not the patient. Other common sugars, like glucose, produced large background signals in healthy animals, which would make it harder to tell whether bacteria are actually present, and were ruled out for that reason.

Before moving to animals, the team screened the tagged compounds against six species of human-infecting bacteria in lab cultures. Most bacterial species broke down one or more of the compounds and produced detectable signals. Uninfected mice given the same compounds intravenously produced little to no response. Infected mice produced clear, measurable elevations across multiple infection types, including muscle, lung, bone, and bloodstream infections.

In the most clinically relevant experiment, mice infected with E. coli in their leg muscles were given a course of ceftriaxone, a widely used hospital antibiotic. Before treatment, infected animals harbored roughly 10 billion colony-forming units (a standard measure of live bacteria). After 24 hours of treatment, that count fell to around 100,000. Breath signals tracked that drop directly, falling in step with the bacterial counts.

“In clinical practice, a major failure of diagnostic imaging techniques (plain-film, CT and MR) is their inability to determine whether infection is responding appropriately to antibiotic therapy in a timely manner, as radiologic improvements usually lag behind microbiological and clinical improvement,” the authors write.

That gap has real consequences for antibiotic resistance. Physicians often keep patients on antibiotics longer than necessary because there’s no reliable signal that the infection has cleared. Others prescribe them when bacteria aren’t the cause at all. A test that directly measures live bacterial activity could let doctors tighten those decisions, shortening courses when the job is done and escalating treatment when it isn’t.

Bone infections, known as osteomyelitis, illustrate the broader diagnostic problem well. Structural changes on imaging often appear only after significant damage has already occurred, and those changes can look very similar to arthritis, bone death, or age-related wear. In the osteomyelitis model, infected mice showed elevated breath signals at both four and eight days after infection, with stronger signals in animals carrying higher bacterial loads.

Breath testing also showed an unexpected secondary use. Researchers compared four strains of methicillin-resistant Staphylococcus aureus (MRSA) and found that breath signal strength predicted how well each strain would respond to an experimental PET tracer, called [2-18F]maltose, currently being evaluated for use in humans. Strains with high breath signals had high PET uptake; strains with low signals had low uptake. Since no widely used screening method currently exists to predict PET tracer performance across MRSA strains, a breath test could help guide imaging decisions before the procedure is ordered.

All of this still needs to be tested in people. The study was conducted in mice, with groups of four to sixteen animals per experiment, and significant questions remain, including how much bacteria must be present to generate a detectable human signal and whether normal gut and airway bacteria create background interference. On the practical side, all three compounds tested have established safety records in humans when given intravenously for other purposes. Detection equipment is portable and inexpensive, requiring no specialized laboratory. Humans, unlike mice, can exhale deliberately, which should produce stronger and cleaner signals than those recorded in the study’s metabolic chambers.

Most efforts to fight antibiotic resistance focus on developing new drugs. Far less attention goes to fixing the diagnostic failures that lead to unnecessary prescribing in the first place. A bacterial breath test that can confirm infection and track treatment response could give physicians a more precise grip on the drugs already in use, and in the long run, that may matter just as much.

Disclaimer: This article is based on preclinical research conducted in mice. The findings have not yet been tested in humans and should not be interpreted as medical advice. Consult a qualified healthcare provider regarding any concerns about bacterial infections or antibiotic treatment.

This study was conducted entirely in preclinical mouse models, and results will need to be validated in human subjects before any clinical application is possible. Unresolved questions include the bacterial burden required to generate a detectable breath signal in humans, potential background interference from gut and airway bacteria, optimal dosing of the carbon-13-enriched compounds, and how these agents behave in patients with underlying conditions. Whether fasting affects results also requires further study. Infection models were laboratory-induced rather than naturally occurring, and group sizes were small, ranging from four to sixteen animals per cohort.

This work was funded by the National Institutes of Health (R01-EB030897, R01-AI181378, R01-AI161027, R21-AI164684, R01-EB028338, R01-AI192221), the St. Jude Children’s Research Hospital Center for Infectious Disease Research (GR1002458), and the Cystic Fibrosis Foundation (20A0). One author, Marshall D. McCue, is an employee of Sable Systems International, which sells breath testing systems, and of McCue Consulting LLC, where anonymized samples from this study were analyzed. Work described in the paper led to U.S. patent application #63/686,445, titled “Methods of detecting bacteria using 13C-labelled compounds,” with co-inventors M. López-Álvarez, M.A. Ohliger, K.D. Neumann, and D.M. Wilson. All other authors reported no conflicts of interest.

The study was authored by Marina López-Álvarez, Sang Hee Lee, Anju Wadhwa, Mohammad Yaqoob Bhat, Tyler S. Simmons, Jung Min Kim, Anil P. Bidkar, Spenser R. Simpson, Shari Dhaene, Jeffrey D. Steinberg, Joseph Blecha, Robert R. Flavell, Marshall D. McCue, Amanda M. Green, Renuka Sriram, Tom Desmet, Joanne Engel, Jason W. Rosch, Michael A. Ohliger, Kiel D. Neumann, and David M. Wilson. Corresponding authors are Kiel D. Neumann (St. Jude Children’s Research Hospital) and David M. Wilson (University of California, San Francisco). Published in ACS Central Science. DOI: https://doi.org/10.1021/acscentsci.5c01995

Physicians often keep patients on antibiotics longer than necessary because there's no reliable signal that the infection has cleared. Others prescribe them when bacteria aren't the cause at all.

My experience is that they go the other way. They are reluctant to prescribe them because they are hoping that it is a virus and they often try to not prescribe a second round when the first round failed to clear the infection up. Often because the delay allow the infection to settle in nice and deep.

Either way this test could be quite useful.

Wow. You have the opposite doctors of what I have seen. They will prescribe antibiotics for viral infections. There are even studies that urgent care doctors do that even more. This tendency is why we have so much antibiotic resistance.

Doctors, especially younger American trained doctors are very reluctant, I have found, to give you antibiotics.

Older doctors are more likely to give them to you and foreign trained doctors will give them for a hangnail.

Just imagine the role of these lab technicians testing the breath.

If I were the lab tech, I would insist upon Hazard Pay and a guarantee of PTSD treatment later.

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.