A comparison of seven tests for serological diagnosis of tuberculosis

Seven serological tests, two immunochromatographic tests, ICT Tuberculosis and RAPID TEST TB, and five enzyme-linked immunosorbent assays, TUBERCULOSIS IgA EIA, PATHOZYME-TB complex, PATHOZYME-MYCO IgG, PATHOZYME-MYCO IgA, and PATHOZYME-MYCO IgM, were evaluated simultaneously with 298 serum samples from three groups of individuals: 44 patients with active tuberculosis, 204 controls who had undergone the Mantoux test (89 Mantoux test-positive and 115 Mantoux test-negative controls), and 50 anonymous controls. The sensitivities of the tests with sera from patients with active tuberculosis were poor to modest, ranging from 16 to 57%. All the tests performed equally with sera from subgroups of those with active tuberculosis, those with pulmonary (33 patients) versus extrapulmonary (11 patients) disease, and those who were smear positive (24 patients) versus smear negative (12 patients) (P > 0.05). The specificities of the tests ranged from 80 to 97% with sera from the Mantoux test controls and 62 to 100% with sera from the anonymous controls. The TUBERCULOSIS IgA EIA had the highest sensitivity (57%) with sera from patients with active tuberculosis, with a high specificity of 93% with sera from the Mantoux test controls, but a very poor specificity of 62% with sera from the anonymous controls. Overall, ICT Tuberculosis followed by PATHOZYME-MYCO IgG had the best performance characteristics, with sensitivities of 41 and 55%, respectively, with sera from patients with active tuberculosis and specificities of 96 and 89%, respectively, with sera from the Mantoux test controls and 88 and 90%, respectively, with sera from the anonymous controls. By combining all the test results, a maximum sensitivity of 84% was obtained, with reciprocal drops in specificities to 55 and 42% for the Mantoux test controls and anonymous controls, respectively. The best combination was that of ICT Tuberculosis and PATHOZYME-MYCO IgG, with a sensitivity of 66% and a specificity of 86% for the Mantoux test controls and a sensitivity and specificity of 78% for the anonymous controls. While a negative result by any one of these tests would be useful in helping to exclude disease in a population with a low prevalence of tuberculosis, a positive result may aid in clinical decision making when applied to symptomatic patients being evaluated for active tuberculosis.