Performance evaluation of the Xpert MTB/RIF assay according to its clinical application

BACKGROUND The Xpert MTB/RIF assay (Xpert assay; Cepheid, Sunnyvale, CA) is becoming the test of choice for the rapid diagnosis of tuberculosis and rifampin (RIF) resistance. The aim of this study was to evaluate the performance of the Xpert assay with respect to its clinical application at a tertiary care hospital in Korea, a country with an intermediate tuberculosis burden and high-resource. METHODS A total of 303 Xpert assay results from 109 smear-positive and 194 smear-negative respiratory specimens were retrospectively reviewed. Based on patients' medical records, four categories of clinical applications of the Xpert assay were identified: (1) the diagnosis of pulmonary tuberculosis in patients with a high probability of pulmonary tuberculosis according to their clinical and radiological features; (2) the exclusion of tuberculosis in clinically indeterminate patients for pulmonary tuberculosis; (3) the differentiation of Mycobacterium tuberculsosis (MTB) from nontuberculous mycobacteria in a smear-positive specimen; and (4) the diagnosis of RIF resistance. Standard culture and drug susceptibility tests were used as reference methods. RESULTS The sensitivity of the Xpert assay for MTB detection in category 1 was 89.8% (95% confidence interval [CI], 78.5-95.8%), but 66.7% (95% CI, 12.5-98.2%) in category 2. The positive predictive values ranged from 33.3% (95% CI, 6.0-75.9%) in category 2 to 91.3% and 91.7% in categories 1 and 3, respectively. The negative predictive values were over 90% in all categories. The Xpert assay correctly detected RIF resistance in six of the seven (85.7%) isolates tested. CONCLUSIONS The Xpert assay exhibited variable performance according to its clinical application; this finding cautions that careful interpretation for the results of this assay would be needed according to its intended purpose.