Evaluation of the analytical performance of the Xpert MTB/RIF assay

Abstract

We performed the first studies of analytic sensitivity, analytic specificity, and dynamic range for the new Xpert MTB/RIF assay, a nucleic acid amplification-based diagnostic system that detects Mycobacterium tuberculosis and rifampin (RIF) resistance in under 2 h. The sensitivity of the assay was tested with 79 phylogenetically and geographically diverse M. tuberculosis isolates, including 42 drug-susceptible isolates and 37 RIF-resistant isolates containing 13 different rpoB mutations or mutation combinations. The specificity of the assay was tested with 89 nontuberculosis bacteria, fungi, and viruses. The Xpert MTB/RIF assay correctly identified all 79 M. tuberculosis isolates and correctly excluded all 89 nontuberculosis isolates. RIF resistance was correctly identified in all 37 resistant isolates and in none of the 42 susceptible isolates. Dynamic range was assessed by adding 10(2) to 10(7) CFU of M. tuberculosis into M. tuberculosis-negative sputum samples. The assay showed a log-linear relationship between cycle threshold and input CFU over the entire concentration range. Resistance detection in the presence of different mixtures of RIF-resistant and RIF-susceptible DNA was assessed. Resistance detection was dependent on the particular mutation and required between 65% and 100% mutant DNA to be present in the sample for 95% certainty of resistance detection. Finally, we studied whether assay specificity could be affected by cross-contaminating amplicons generated by the GenoType MTBDRplus assay. M. tuberculosis was not detected until at least 10(8) copies of an MTBDRplus amplicon were spiked into 1 ml of sputum, suggesting that false-positive results would be unlikely to occur.

FIG. 1.

Dynamic range studies. Log dilutions of M. tuberculosis H37Rv cells were added to 1 ml of M. tuberculosis-negative sputum to final concentrations ranging from 10² to 10⁷ CFU/ml (n = 5 or 6 per dilution). (A) Average rpoB probe B cycle thresholds (C_Ts) were plotted for each M. tuberculosis concentration tested. Clinically relevant M. tuberculosis concentrations all fall within the linear range of the assay. (B) Average B. globigii internal control probe C_Ts were plotted for each concentration of M. tuberculosis tested. The B. globigii assay gave an average C_T of 29.7 (95% CI, 28.9 to 30.4) for all dilutions and was not influenced by the concentration of M. tuberculosis in the sample. Dotted lines indicate maximum valid C_T values for probe positivity.

FIG. 2.

Minimum detectable fraction of mutant DNA. DNA extracted from RIF-resistant and RIF-susceptible M. tuberculosis isolates were mixed to six final concentrations of mutant DNA. Mutant DNA contained either an rpoB 531ttg mutation (A) or an rpoB 533ccg mutation (B). Sample mixtures equivalent to 20 times the assay limit of detection (LOD) of 4.5 genomes per reaction were added to the chamber receiving eluted DNA in the full cartridge protocol and were processed using a PCR-only protocol. The percentage of samples (n = 20) identified as RIF resistant was plotted at each concentration. As determined by logistic regression, there was a 95% probability of detecting RIF resistance when the DNA mixture contained 61.2% or 100% of the total DNA in the sample mixture for a 531ttg and 533ccg mutant, respectively. The curves from top to bottom indicate the upper, middle, and lower bounds of the 95% CI. Dashed lines indicate limit of detection.

FIG. 3.

Potential risk from amplicon contamination. (A) Alignment demonstrating the overlap between the GenoType MTBDRplus PCR amplicon (determined by sequencing) and the priming regions of the Xpert MTB/RIF assay. (B) MTBDRplus simulated amplicon (10¹⁰ to 10⁵ copies per ml) was spiked into a mixture of sputum and SR prior to transfer to the cartridge sample chamber (n = 3 per dilution). At least 10⁸ copies per ml were required to induce a false-positive result. Sputum spiked with 10³ CFU of Mycobacterium bovis BCG per ml was processed according to the package insert as a control for sputum inhibition (open circle). (C) MTBDRplus amplicon was spiked into TET buffer that also contained B. globigii DNA added directly to the elution receiving chamber of an open cartridge and then processed with a PCR-only protocol (n = 2 per dilution). At least 10⁴ copies per PCR were required to induce a false-positive result. The control (open circle) was 45 copies of M. tuberculosis DNA per reaction. Dotted lines indicate maximum valid C_T values for probe positivity.