Breath tests that detect bacterial infections in minutes - not days - just passed their first animal trials

Diagnosing a bacterial infection today often means waiting. Blood cultures take 24 to 72 hours. Molecular diagnostics are faster but expensive. Imaging shows inflammation but cannot tell you what is causing it. For patients in emergency rooms with suspected sepsis or pneumonia, those hours of diagnostic limbo can be the difference between targeted treatment and a broad-spectrum antibiotic gamble.

A research team led by David Wilson, with Kiel Neumann and Marina Lopez-Alvarez, has now demonstrated that a breath test - conceptually simple, potentially portable, results in minutes - can detect active bacterial infections in animal models. The work, published in ACS Central Science, represents an early but concrete step toward breath-based infectious disease diagnostics.

The H. pylori precedent

Breath testing for infection is not entirely new. The Helicobacter pylori breath test has been in clinical use for years. A patient drinks a liquid containing a carbon-13-labeled substrate that H. pylori bacteria metabolize. If the bacteria are present in the stomach, they convert the substrate to carbon-13-labeled carbon dioxide, which the patient exhales and a detector picks up.

That test works, but only for one specific bacterium in one specific location. Wilson's team wanted to generalize the approach - to detect a broader range of bacterial infections anywhere in the body, from the lungs to the bones to the bloodstream.

Sugars that bacteria eat and human cells ignore

The key insight is substrate selectivity. The researchers tested a panel of sugars and sugar alcohols tagged with carbon-13, a stable (non-radioactive) isotope of carbon. They were looking for compounds that bacteria would readily metabolize into CO2 but that human cells would largely leave alone.

In laboratory experiments, several candidates emerged - compounds that bacteria converted efficiently into carbon-13-labeled carbon dioxide while mammalian cells showed minimal metabolism. The labeled gas was then detected using nondispersive infrared spectroscopy, a technique that is well-established, portable, and relatively inexpensive.

Ten-minute signals in infected mice

The team tested the approach in mice with four different types of infection: pneumonia, bloodstream infections, muscle infections, and bone infections. After intravenous injection of the carbon-13-tagged compounds, breath from infected animals showed elevated levels of labeled carbon dioxide. Healthy control mice showed little to no carbon-13 in their exhaled breath.

The timeline was striking. Although the breath testing protocol was not fully optimized, researchers typically saw elevated signals within the first 10 minutes of administering the tagged substrate and beginning breath sampling. Compare that to the day or more required for blood cultures, and the clinical appeal becomes obvious.

In one experiment modeling E. coli infection, the team tracked breath signals during antibiotic treatment. As bacterial levels dropped in response to antibiotics, the carbon-13 signal in the breath declined correspondingly. This suggests the test could serve double duty - not just diagnosing infection but monitoring whether treatment is working, in something close to real time.

Portable hardware, safe substrates

Two practical features strengthen the approach's clinical potential. The detection hardware - nondispersive infrared spectrometers - already exists as portable, relatively affordable instruments. There is no need for mass spectrometers or specialized laboratory equipment. And the sugar and sugar alcohol substrates used in the test are compounds generally recognized as safe for human consumption, which should simplify the regulatory path.

Wilson framed the motivation in clinical terms: "If a patient visits the Emergency Room or Acute Care clinic, we hope that he or she can be diagnosed with an acute bacterial infection as efficiently as possible."

Mouse lungs are not human lungs

The study tested the concept in animal models only. Mouse respiratory physiology differs from human physiology in ways that matter for breath-based diagnostics - tidal volumes, breathing rates, and airway architecture all differ substantially. Whether the signal-to-noise ratio that worked in mice will hold in human patients breathing through a collection device is an open question.

The substrates were administered intravenously in this study. For clinical use, an oral or inhaled route would be far more practical, but whether the substrates reach infection sites effectively through those routes has not been tested.

Specificity is another concern. The test detects bacterial metabolism broadly. It does not identify which species of bacteria is causing the infection - information that is critical for choosing the right antibiotic. A positive breath test might still need to be followed by conventional culture or molecular diagnostics to guide treatment selection.

The researchers have filed a U.S. patent on the technology, signaling intent to pursue commercial development. But the path from animal proof-of-concept to a validated clinical diagnostic typically takes years and requires human trials, regulatory clearance, and head-to-head comparisons with existing diagnostic methods.

What the study establishes is that the core principle works: bacteria in a living host will metabolize specific tagged substrates into detectable breath markers quickly enough to be clinically useful. That is a foundation worth building on.