Cross-priming amplification (CPA) technology for tuberculosis diagnosis from sputum specimens was evaluated. The sensitivity of CPA from smear-and liquid culture-positive specimens was 96.9%, and that from smear-negative and liquid culture-positive specimens was 87.5%. The specificity of CPA in culture-negative specimens was 98.8%.The diagnosis of tuberculosis (TB) is based on phenotypic and genotypic methods. Light microscopy for the detection of acid-fast bacilli (AFB) in sputum smears is a rapid and specific tool commonly used all over the world, but it only detects 30% to 40% of TB patients (10). While the sensitivity of Löwen-stein-Jensen (L-J) solid or liquid culture methods is higher than that of AFB microscopy, culture-based methods require several weeks of incubation time. Several genotypic assays, mostly based on nucleic acid amplification tests, have been developed for rapid TB diagnosis, including the GenProbe amplified Mycobacterium tuberculosis direct test, Roche Amplicor MTB test, Cobas Amplicor test, Abbott LCx test, and the BD-ProbeTec (strand displacement amplification) test (1,2,5,8,12). However, the cost of equipment needed for these methods is a barrier to their widespread use, especially in developing countries. A sensitive, accurate, rapid, and affordable diagnostic tool that will work in resource-limited settings is urgently needed.Cross-priming amplification (CPA) technology is a recent invention from an isothermal DNA amplification system by Ustar Biotechnologies Co., Ltd. Using multiple cross-linked primers (six to eight primers), a DNA target sequence can be amplified at a constant temperature (Fig. 1). The detection of amplified products is performed on a lateral flow strip housed in an enclosed, sealed plastic device to prevent the leakage of amplicons (4, 7). In this study, we evaluated the CPA isothermal amplification and detection kit (Ustar Biotech, Hangzhou, China) to determine whether it could accurately detect M. tuberculosis in clinical sputum specimens from a variety of different types of patients. Sputum specimens from 180 persons who were suspected of having TB, based on their clinical symptoms and presentation, were obtained from three hospitals in Shanghai. Sputum specimens were also collected from 98 non-TB patients (with lung cancer, lung edema, and other pulmonary illnesses) presenting at the same hospitals. For comparison with the CPA assay results, all of the clinical specimens in our study were also processed for AFB smear microscopy, L-J solid medium, and BacT/Alert 3D liquid culture following standard protocols (11). Bacterial genomic DNA was extracted by boiling, as previously described (9). The CPA method was performed according to the manufacturer's instructions, using primers targeting the gyrB gene of M. tuberculosis ( Fig. 2A). The result of each CPA assay was determined by observing the presence (positive result) or absence (negative result) of visible bands on the test strips (Fig. 2B).The detection rates of mycobacteria in the 180 sputum specimens used ...