Abstract
Diagnostic tests for tuberculosis (TB) usually require collection of sputum, a viscous material derived from human airways. Sputum can be difficult and hazardous to collect and challenging to process in the laboratory. Oral swabs have been proposed as alternative sample types that are noninvasive and easy to collect. This study evaluated the biological feasibility of oral swab analysis (OSA) for the diagnosis of TB. Swabs were tested from South African adult subjects, including sputum GeneXpert MTB/RIF (GeneXpert)-confirmed TB patients (n = 138), sputum GeneXpert-negative but culture-positive TB patients (n = 10), ill non-TB patients (n = 37), and QuantiFERON-negative controls (n = 34). Swabs were analyzed by using a manual, nonnested quantitative PCR (qPCR) targeting IS6110 Two swab brands and three sites within the oral cavity were compared. Tongue swabbing yielded significantly stronger signals than cheek or gum swabbing. A flocked swab performed better than a more expensive paper swab. In a two-phase study, tongue swabs (two per subject) exhibited a combined sensitivity of 92.8% relative to sputum GeneXpert. Relative to all laboratory-diagnosed TB, the diagnostic yields of sputum GeneXpert (1 sample per subject) and OSA (2 samples per subject) were identical at 49/59 (83.1%) each. The specificity of the OSA was 91.5%. An analysis of "air swabs" suggested that most false-positive results were due to contamination of manual PCRs. With the development of appropriate automated methods, oral swabs could facilitate TB diagnosis in clinical settings and patient populations that are limited by the physical or logistical challenges of sputum collection.