Tongue swabs miss many TB cases with Trace Xpert Ultra

Diagnosing pulmonary tuberculosis (pTB) usually relies on sputum samples, but collecting good sputum can be difficult for many people. Investigators are testing non-sputum approaches, including molecular amplification of tongue swab samples, as a simpler way to detect TB. An improved manual qPCR method for tongue swabs previously reported more than 90% sensitivity overall compared with a sputum microbiologic reference standard, raising hopes that a quick swab of the mouth could be a useful diagnostic tool. However, it was not known how well tongue swabs perform in people whose sputum tests show very low levels of bacterial DNA. To address that question, a team led by corresponding author Adrienne E. Shapiro recruited adults in South Africa and Uganda who had a sputum Xpert MTB/RIF Ultra Trace (TR+) result. The study enrolled these participants for confirmatory evaluation and follow-up, aiming to see whether tongue swab testing could confirm TB in people with that low-positive molecular finding and whether it might help guide clinical decisions when sputum results are of uncertain significance.

Participants underwent a structured evaluation that included symptom review, a physical examination, chest X-ray, further sputum testing, and collection of two tongue swabs. Further sputum testing included a repeat Xpert Ultra and two solid and liquid mycobacterial cultures. Tongue swabs were tested using qPCR amplification of the IS6110 gene; the study considered a single copy detected on one or more swabs as TB-positive. People with a Trace Xpert Ultra result who were not diagnosed with TB at the initial visit were re-evaluated at one and three months. In total, 225 enrolled TR+ participants provided at least one tongue swab at baseline; 115 (51%) were women, median age was 38 (IQR 30-47), and 130 (58%) were people living with HIV (PWH). Using culture alone as the reference, 45 (20%) were culture-positive at baseline. By a microbiologic reference standard (MRS: any positive result from Xpert Ultra or culture) 58 (26%) were positive, and by a composite reference standard (CRS: a clinical recommendation for TB treatment or any positive culture) 83 (37%) were positive. Against culture, tongue swabs had sensitivity 25% [95% CI 13-40%] and specificity 94% [90-97%]; against MRS sensitivity 25% [95% CI 14-38%] and specificity 96% [91-98%]; and against CRS sensitivity 16% [9-26%] and specificity 94% [89-98%].

The results show that in people with a sputum Xpert Ultra Trace result, tongue swabs tested with the described qPCR method detected only a small fraction of cases that other tests labeled as TB. Sensitivity was low across reference standards, meaning many true infections identified by sputum testing or clinical judgment would be missed by a tongue swab approach in this specific group. Specificity was moderately high, so a positive tongue swab result was usually correct, but the low sensitivity limits the test’s usefulness as a replacement or sole confirmatory tool for people with low-positive molecular signals. The authors conclude that tongue swabs have limited value for diagnosing people with low-positive molecular test results of uncertain clinical significance. In practical terms, the findings suggest tongue swab qPCR is not reliable enough on its own to resolve the ambiguity of Trace Xpert Ultra results and that clinicians and programs should be cautious about using tongue swabs to make treatment decisions in this population.