Evaluation of the Gen-Probe amplified Mycobacterium tuberculosis direct detection test

Mycobacterium tuberculosis direct (MTD) testing was performed on 423 specimens (from 293 patients) submitted for mycobacterial culture, including 121 from 47 patients with current or recent cultures positive for tuberculosis. Resolved sensitivity and specificity were 91.4% and 99.3%, respectively. The assay detected 12 of 14 patients with one specimen and 31 of 32 with > or = 2 specimens. The 10 missed specimens included 2 resembling blood clots and 6 smear negative or 1+ specimens; however, a 2+ and a 4+ specimen were missed, which implies inhibitors. Some smear positive/culture negative and smear negative/culture negative specimens from treated patients gave positive results. The number of smear negative, culture positive specimens from untreated patients was too low to evaluate rigorously. Consideration of these observations and of costs suggests that the assay is appropriate only for untreated patients with smear-positive, nonbloody specimens. They should not be reported as negative for tuberculosis until three smear positive specimens have given negative MTD results. Factors involved in the generation of false positive results, including amplicon contamination, are evaluated.