We compared the performance of two PCR assays, an IS6110-based in-house protocol and the COBAS AMPLICOR MTB PCR (COBAS MTB) system, for the detection of Mycobacterium tuberculosis complex in 43 human lymph node samples from 40 patients. For the in-house PCR and the COBAS MTB assays, respectively, sensitivities were 87.5% versus 45.5% (P < 0.05), specificities were 100.0% versus 91.3% (P > 0.05), and inhibition rates were 4.8% versus 19.5% (P < 0.05). For the COBAS MTB system, additional N-acetyl-L-cysteineNaOH pretreatment of the samples changed neither the inhibition rate nor the sensitivity significantly.In Germany, tuberculous lymphadenitis is the most common form of extrapulmonary tuberculosis, comprising about 7 to 8% of all cases of tuberculosis (19). Rapid and accurate diagnosis is important for the effective treatment of lymph node tuberculosis. Conventional methods for detecting Mycobacterium tuberculosis complex include acid-fast staining and culture. The sensitivity of microscopy in lymph node samples of human immunodeficiency virus noninfected patients is low because of the low numbers of mycobacteria present. Furthermore, M. tuberculosis complex cannot be differentiated from mycobacteria other than tuberculosis (MOTT) by acid-fast staining. Culturing methods, including those in solid and liquid media, detect M. tuberculosis complex with reasonable sensitivity. They are a prerequisite for accurate identification to species level and susceptibility testing. However, culture requires several weeks for completion due to the slow growth of mycobacteria. Nucleic acid amplification assays have been developed for detecting M. tuberculosis complex directly from clinical specimens. However, commercially available automated methods like the COBAS AMPLICOR MTB PCR (COBAS MTB) assay (Roche Diagnostics) have been approved for pulmonary specimens only. Several authors have reported acceptable sensitivity and specificity rates for extrapulmonary specimens by the COBAS MTB assay (8, 18), but so far, lymph node samples have not been specifically examined.The purpose of this study was to evaluate the performance of the COBAS MTB assay for the detection of M. tuberculosis complex in lymph node samples and to compare it with an IS6110 in-house PCR assay. The results of acid-fast smear and culture and the clinical data of the patients were used for reference. Patients and samples.From January 1998 to December 1999, a total of 43 lymph node samples were surgically removed from 40 patients with possible lymph node tuberculosis. The specimens were sent in sterile tubes containing saline to the Hygiene Institute of the University of Heidelberg for the detection of mycobacteria. The patients were 18 females and 22 males with a median age of 25 years (range, 1 to 91 years).Processing of lymph node specimens. The specimens were carefully cut into 1-to 2-mm 3 pieces in sterile Petri dishes with sterile scalpels before being divided into three portions of equal size. The three aliquots were homogenized in 1 ml of sterile saline e...