Multicenter evaluation of the Abbott LCx Mycobacterium tuberculosis ligasechain reaction assay

Four Australian hospital laboratories evaluated the performance of the Abbo tt LCx Mycobacterium tuberculosis assay with 2,347 specimens (2,083 respira tory and 264 nonrespiratory specimens) obtained from 1,411 patients. A tota l of 152 specimens (6.5%) were culture positive for Myobacterium tuberculos is complex (MTBC); of these, 79 (52%) were smear positive, After resolution of discrepant data, the overall sensitivity, specificity, and positive and negative predictive values for the LCx assay were 69.7, 99.9, 99.1, and 97 .7% respectively. For smear-positive respiratory specimens that were cultur e positive for MTBC, the values were 98.5, 100, 100, and 98.4%, respectivel y, while the values for smear-negative respiratory specimens were 41.5, 99. 9, 96.4, and 98%, respectively. Relative operating characteristic curves we re constructed to demonstrate the relationship between sensitivity and spec ificity for a range of possible cutoff values in the LCx assay. These graph s suggested that the assay sensitivity for respiratory samples could be inc reased from 70.2 to 78.6%, while the specificity would be reduced from 99.9 to 99.4% by inclusion of a grey zone (i.e,, LCx assay values of between 0. 2 and 0.99). An algorithm is presented for the handling of specimens with L Cx assay values within this grey zone.