Treating drug-resistant tuberculosis is a long, punishing job. Patients swallow pills for many months, and the drugs only work if they reach the right concentration in the blood. Too little, and the bacteria survive and grow more resistant. Too much, and the side effects can be brutal. The obvious way to check is to draw blood and measure the drug directly. That means needles, a lab, a cold chain, and money. In the places where drug-resistant TB is most common, all four are in short supply.
A study published this week in the European Respiratory Journal asks whether saliva could do part of that job instead. The answer, at least for one of the two drugs tested, is a cautious yes.
Reading the dose from spit
Researchers ran a community-based study in Vietnam between 2021 and 2023. They collected paired plasma and saliva samples from patients being treated for multidrug-resistant TB, taking them at zero, two, and five hours after a dose. Saliva was measured with a point-of-care device called the NanoPhotometer NP80. The same samples were also run through liquid chromatography-mass spectrometry, the slow and expensive lab method that serves as the gold standard.
The two drugs in question were levofloxacin and linezolid, both mainstays of modern drug-resistant TB regimens. To decide whether a patient was getting enough, the team calculated total drug exposure over 24 hours and compared it against therapeutic targets tied to how much drug it takes to suppress the bacteria.
For levofloxacin, a saliva exposure threshold of 70.3 mg·h·L flagged 16 of the 19 patients whose blood levels were too low. That works out to a sensitivity of 84 percent and a specificity of 78 percent across 60 patients. Not perfect, but useful. A quick spit test that catches most under-dosed patients could tell a clinic which people need a closer look, without sending everyone for blood work.
Where the saliva test fell short
Linezolid was a different story. Only 17 patients had data for that drug, and the numbers were thin. Saliva testing did correctly spot the single patient who was under-dosed and the single patient who was over-dosed. But with so few cases to learn from, the researchers could not pin down a reliable saliva threshold at all. For levofloxacin they had a cutoff. For linezolid they had two lucky hits and not enough to build a rule.
That gap matters, because linezolid is one of the drugs whose side effects most need watching. Nerve damage and blood problems climb when levels run high for too long. A saliva monitor that could catch those cases early would be worth a lot. This study does not deliver that yet.
A screening tool, not a verdict
It is worth being clear about what the work shows and what it does not. This was a feasibility study, cross-sectional, with small numbers, especially for linezolid. It establishes that saliva concentrations track plasma exposure well enough to screen for trouble. It does not prove that saliva-guided dosing changes how patients actually do. The authors themselves call for more research to set a linezolid threshold and to test whether adjusting doses based on saliva improves outcomes or saves money.
Saliva and blood are also not the same fluid. A drug's passage into saliva depends on its chemistry, how much binds to proteins, and the pH of the mouth. That is part of why levofloxacin behaved predictably here and linezolid did not. Any real-world version of this test would need to account for those quirks drug by drug.
Still, the direction is promising. Therapeutic drug monitoring has long been a luxury reserved for well-funded hospitals. A non-invasive test that a health worker could run in a clinic, or eventually a village, points toward a version of that care built for the settings where drug-resistant TB actually lives. Getting the dose right is half the battle against a disease that kills over a million people a year. A little spit might help.
Comments