There are new changes in the results categories in Xpert Ultra assay. The standard Xpert used to report the semi-quantitative categories as high, medium, low and very low detection. For the Ultra assay, the addition of new semi-quantitative category trace was introduced in Ultra that corresponds to the lowest bacillary burden TB detection. In the trace TB detection, one or both of the probes for multi-copy targets are positive with Ct values less than 37 cycles and no more than one rpoB probes have a Ct value less than 40 cycles (WHO Meeting Report of a Technical Expert Consultation: Non-inferiority analysis of Xpert MTB/RIF Ultra compared to Xpert MTB/RIF, 2017).
In addition, among the positive trace call TB detection, it should be considered as a true positive results for people with HIV, children and extra-pulmonary specimen (WHO Meeting Report of a Technical Expert Consultation: Non-inferiority analysis of Xpert MTB/RIF Ultra compared to Xpert MTB/RIF, 2017) . For people that are not at risk for HIV and have a trace call TB results, a fresh specimen from the patient should undergo a repeat testing and the second Ultra test results should be used for clinical decisions and follow-up (WHO Meeting Report of a Technical Expert Consultation: Non-inferiority analysis of Xpert MTB/RIF Ultra compared to Xpert MTB/RIF, 2017). However, if the second repeated trace call samples has a negative results, further clinical and radiological assessment should be made to initial the TB treatment. If the second repeated trace call sample has a positive result, it is sufficient to give a positive TB detection with no recent TB history.For the results of rifampin resistance detection, RIF resistance is not detected if MTB is detected and not in trace call, and all the four rpoB probes have identifiable melting temperature peaks with the wild type profile. RIF resistance is detected if MTB is detected and not in trace call, and at least one of the rpoB probes has a mutant profile. However, if the MTB is detected in trace, RIF resistance could not be interpreted and reported as RIF indeterminate. DISCUSSION In our study population, we found that Xpert Ultra had an overall improved sensitivity over the previous Xpert assay. Xpert Ultra had 96.7% sensitivity in all pulmonary samples, which was a 13.4% increase in sensitivity compared with Xpert. There were also 15.4% and 33.4% increase in sensitivity in extra-pulmonary and CSF samples in this study respectively. Specifically, Xpert Ultra improved MTB detection by 22.2% compared to Xpert in smear-negative, culture positive TB pulmonary samples. Xpert Ultra had even more improvement in MTB detection by 25% and 33.4% in smear-negative culture-positive extra-pulmonary and CSF samples, respectively. Therefore, Xpert Ultra had an overall improved MTB sensitivity in all types of specimens, and even more increased in sensitivity in non-pulmonary specimens. It also demonstrated that Ultra had greater added values in smear-negative culture-positive specimens. According to a published study from Chakravorty et al, it compared the Ultra and Xpert assays on 227 pulmonary samples and showed sensitivity of 87.5% and 81.0% respectively (Glass et al., 2017). In a previous multicenter study from FIND also found that the sensitivity of Ultra was 5% higher than that of Xpert (Diagnostics, n.d.). Therefore, our study result of higher sensitivity in Ultra assay was compatible with the previous few studies, which Ultra had an advanced diagnosis in MTB detection and even better in the smear-negative culture- positive samples. Some previous studies mentioned that Ultra detects non-replicating or non-viable bacilli present in patients with a recent TB history; and therefore, reducing the specificity of Ultra (WHO Meeting Report of a Technical Expert Consultation: Non-inferiority analysis of Xpert MTB/RIF Ultra compared to Xpert MTB/RIF, 2017). However, it was attributed only less than 1% of all results in the study (WHO Meeting Report of a Technical Expert Consultation: Non-inferiority analysis of Xpert MTB/RIF Ultra compared to Xpert MTB/RIF, 2017). The increased yields of the sensitivity with the benefits of detecting MTB smear-negative culture-positive specimens, pediatric specimens and HIV positive patients became the trade-off between increased diagnosis and overtreatment. Moreover, the improved sensitivity of Ultra was due the use of multi-copy of IS1081 and IS6100 genes as a target sequence, which allowed the additional new semi quantification of trace call to be introduced to detect the lowest bacillary load for MTB detection. The trace detection means that only the ISO108 and IS6100 were detected, but not other TB region in the rpoB gene. A trace positive result was sufficient to initiate therapy to the individuals, especially with suspected HIV infected people, children and for extra-pulmonary samples from patients (WHO Meeting Report of a Technical Expert Consultation: Non-inferiority analysis of Xpert MTB/RIF Ultra compared to Xpert MTB/RIF, 2017). It could still help early detect TB infection and give appropriate treatment. For those patients, the radiologic findings and clinical presentations could be added to carefully analyze the results for managing TB. One of the limitations of our study was the relative small sample size compared with the previous studies. Therefore, we did not get any false-positive results and had a lower specificity in Ultra. As the previous multicenter diagnostic study from Dorman et al showed that the specificity was lower in Ultra than in Xpert, and yielded a higher proportion of false positive results. This was due to the detection of the non-viable bacilli and the successful completion of treatment of previous TB history patients in Ultra. More modelling studies were needed to allow more in depth exploration of the trade-offs between increased numbers of patients correctly and falsely diagnosed under different epidemiological settings. Besides the laboratory results, among those retreated patients, a good clinical history and physical exam might also be needed. Possibly, another molecular diagnostic test would be aimed to for TB detection for those patients with previous TB history. It could probably mitigate the loss in specificity and maintain some gains in sensitivity in such case. For the detection of rifampicin resistance, our study showed similar results between Xpert Ultra and Xpert in both sensitivity and specificity. Both Ultra and Xpert correctly detected rifampicin resistance and no false positive results were detected. There was no superiority for the Ultra over the Xpert as we seen in MTB detection. This finding showed a similar result with the study from Dorman et al. The study from Dorman et al found that both Xpert Ultra and Xpert had similar results in rifampicin resistance detection. However, Dorman et al suggested that the limitation of the study was that larger numbers of samples were needed to more precisely characterize Ultra accuracy in rifampicin resistance detection. In addition, a previous study from Wang et al found that Xpert Ultra had a slightly higher sensitivity than Xpert in rifampicin resistance detection, whereas the specificity was 100% in both assays. However, the difference was not significant with a p-value >0.05. Therefore, due to the low prevalence of multidrug-resistance strains, the number of rifampicin resistance isolates tested was limited. A larger study population was also needed to confirm the analytical results in Ultra for rifampicin resistance detection. Moreover, the previous Xpert was not capable of detecting resistance-conferring mutation located outside the 81 bp rifampicin resistance determining region of the rpoB gene. The Ultra was developed to overcome this limitation by adding 24 different rifampicin resistance determining region mutations to cover almost the entire rpoB from condons 510 to 533. Ultra was able to differentiate silent mutation at codons 513 and 514 and detect a hetero-resistance sample that was missed by phenotypic sensitivity test and Xpert. However, Ultra could not detect mutation such as IIe481Phe at this point. This was one of the limitations of Ultra in Rifampicin resistance detection.Another finding from our study was that Ultra had a lower limit of detection per sample in MTB detection than Xpert. Our study showed that Ultra had a limit of detection of 4 bacterial colony forming units per mL in MTB detection. From the package insert of the new Ultra assay, it indicated the LoD for MTB detection was 11.8 CFU/mL. Therefore, our finding was comparable to the package insert results and even better. According to the package insert of Xpert, the limit of detection for MTB is 131 CFU/mL. This was a large improvement in the analytical sensitivity in Ultra. Furthermore, the long turnaround time for culture-positive samples was due to the low numbers of bacilli which where under the limits of detection of the test. Therefore, Ultra improved the MTB detection with lower number of bacilli and a shorter turnaround time (less than 90 minutes). Both the Ultra and Xpert assays showed the correlations between each other in our study. By comparing the performance agreement between Ultra and Xpert, we performed both kappa and percentage agreement calculations. There were pros and cons in both calculations. Kappa was designed to take account of the possibility by chance. Percentage agreement directly interpreted the two assays straightforwardly. However, we found that the agreements between the two assays were compatible in both kappa and percentage agreement calculations. Both Xpert Ultra and Xpert were correlated with each other. Kappa showed a substantial agreement of 76.8% between the two assays. The overall percentage agreement, the positive percentage and the negative percentage agreement were all above 80% with a p-value