Tongue swabs improve TB detection with a no-heat Xpert Ultra method

Diagnosing tuberculosis (TB) usually relies on sputum, which can be hard to produce and test. Tongue swabs (TS) are a novel non-sputum specimen for molecular diagnosis of TB, and researchers led by Rachel C. Wood evaluated whether a widely used clinical test could work well on these swabs. The study compared two ways of preparing TS samples for analysis by Cepheid Xpert MTB/RIF Ultra® (Xpert Ultra): an earlier Heat+TE protocol that requires a heating block and a newly reported 2:1 SR method that does not require heating. Researchers used previously collected TS samples from symptomatic patients in South Africa and ran replicate, paired samples from the same patients through the two protocols. Because TS testing is known to be least sensitive when bacterial load in sputum is low, the team prioritized low-bacillary-load participants to maximize the ability to see differences between methods. The primary comparison focused on N=108 participants, of whom 88 were TB-positive by the sputum microbiological reference standard (MRS) and 20 were TB-negative, enabling a direct head-to-head assessment of the two sample-preparation approaches.

The study measured diagnostic accuracy of both TS processing methods using Xpert Ultra and compared results against the sputum microbiological reference standard (MRS). In the prioritized low-bacillary-load set (N=108; 88 MRS-positive, 20 MRS-negative), the Heat+TE method was 42.5% (38/88; 95CI 33-54%) sensitive, while the 2:1 SR method was 52.3% (46/88; 95CI 41-63%) sensitive relative to sputum MRS. Both methods showed 100% (20/20; 95% CI 83-100%) specificity in that sample set. A secondary analysis used a larger cohort (N=241) that included participants with both high and low bacillary loads and compared single-sample testing to dual-sample testing. In that group, testing a single TS sample was 75.5% (182/241; 95CI 70-81%) sensitive relative to sputum MRS, whereas dual sampling increased sensitivity to 81.3% (196/241; 95CI 76-86%). The comparison used replicate, paired samples and the same Xpert Ultra platform for all tests.

These findings show that testing tongue swabs with Cepheid Xpert MTB/RIF Ultra® can detect a meaningful fraction of TB cases, and that a simple no-heat 2:1 SR protocol recovered more positives than the heat-dependent Heat+TE approach in this low-bacillary-load group. The perfect specificity observed in the prioritized set (100% in 20 negatives) suggests that positive TS results with these protocols are unlikely to be false positives in this study sample. Importantly, dual sampling—running two swabs instead of one—raised overall sensitivity in a broader cohort, indicating an operational way to improve detection without changing the molecular platform. Because Xpert Ultra is already widely used and familiar in many clinical settings, a practical, no-heat TS prep method could make non-sputum testing easier to adopt. At the same time, sensitivities reported here—particularly in low-bacillary-load patients—remain below perfect, so TS testing would complement rather than replace sputum-based diagnosis based on these results.