Abstract:

Objective: To evaluate the diagnostic accuracy of tongue swab-based PCR fluorescence probe method for pulmonary tuberculosis (PTB), analyze the consistency across various PTB diagnostic methods, and explore its application value in the diagnosis of PTB. Methods: A prospective study was conducted to recruit 76 suspected PTB patients who sought health care in Guangxi Chest Hospital from January to March 2024. Clinical symptoms and laboratory results, including sputum smear microscopy, GeneXpert MTB/RIFXpert, and tuberculous infection T-lymphocyte spot tests, were collected. The tongue swab method utilized a nucleic acid detection kit for Mycobacterium tuberculosis complex. Using sputum culture or microbiological reference standard (MRS) as reference standard, parameters with high diagnostic efficacy were selected through baseline comparison, univariate ROC analysis, variable correlation analysis, Lasso regression, and forest plots to fit a model for PTB diagnosis consistency analysis was conducted to compare the tongue swab method with the existing PTB diagnostic methods including Xpert, sputum culture, and smear. Results: Using sputum culture as the reference, there were statistically significant differences among Xpert (χ²=49.045, P<0.001), tongue swab method (χ²=42.626, P<0.001), and sputum smear (χ²=17.443, P<0.001) between positive and negative groups. AUCs were ranked as follows: Xpert (AUC=0.902), tongue swab method (AUC=0.875), and sputum smear (AUC=0.709). Model fitting indicated that Xpert combined with the tongue swab method was the optimal diagnostic model. Univariate ROC and comparative analysis showed that Xpert outperformed the tongue swab method (AUCs of 0.902 and 0.875, sensitivities of 87.10% and 83.87%, specificities of 93.33% and 91.11%, respectively; P<0.001 for both), although the difference was not statistically significant (Z=0.795, P=0.427). Using MRS as a reference, significant differences were found for the tongue swab method (χ²=46.723, P<0.001) and sputum smear (χ²=23.262, P<0.001) between MRS positive and negative groups, with AUCs of 0.888 for the tongue swab method and 0.738 for sputum smear. Model fitting selected the tongue swab method as the optimal diagnostic model. Univariate ROC analysis indicated that the tongue swab method significantly outperformed sputum smear (AUCs of 0.888 and 0.738, sensitivities of 82.35% and 50.00%, specificities of 95.24% and 97.62%, P<0.001 for both), with a significant difference (Z=2.889, P=0.004). The tongue swab method demonstrated high consistency with MRS (Kappa=0.7847) and Xpert (Kappa=0.8348), with an overall consistency rate of 92.11%. Conclusion: Regardless of whether sputum culture or MRS is used as a reference, the tongue swab method shows a diagnostic accuracy comparable to that of Xpert and significantly superior to sputum smear, suggesting its potential as a valuable diagnostic method for PTB.

Key words: Tongue swab, Mycobacterium tuberculosis, Polymerase chain reaction, Evaluation studies, Data interpretation, statistical