Review 1: "Xpert MTB/RIF Ultra Resistant and MTBDRplus Susceptible Rifampicin Results in People with Tuberculosis: Utility of FluoroType MTBDR and Deep Sequencing"

RR\ID Evidence Scale rating by reviewer:

• Reliable. The main study claims are generally justified by its methods and data. The results and conclusions are likely to be similar to the hypothetical ideal study. There are some minor caveats or limitations, but they would/do not change the major claims of the study. The study provides sufficient strength of evidence on its own that its main claims should be considered actionable, with some room for future revision.

***************************************

Review: The submitted manuscript by a multinational investigative team evaluated sputum results that are rifampicin resistant by Xpert MTB/RIF Ultra (“Ultra”), a low complexity automated nucleic acid amplification test (LC-aNAAT), and rifampicin susceptible by MTBDRplus version 2, a line probe assay. Both tests are endorsed by the World Health Organization for use in high burden settings, Ultra for the detection of TB and the detection of rifampicin resistance, MTBDRplus for the detection of resistance to rifampicin and/or isoniazid.  Both assays have limitations in terms of identifying drug resistance when the mutation is outside the region of variation covered by the test, when there is heteroresistance, and when the bacillary burden is low. Samples were tested through usual programmatic protocols in a South African reference laboratory. For this study, the reference standard was targeted deep sequencing with single molecule-overlapping reads (SMOR). The authors also describe the results from FluoroType MTBDR v 2.0 (FT-MTBDR) a moderate complexity aNAAT.

In people identified programmatically as having discordant rifampicin results (Ultra-resistant, MTBDRplus-susceptible), the authors aimed to evaluate both tests compared to the reference standard in detecting RIF resistance, and MTBDRplus in detecting INH resistance, to identify conditions associated with discordant results, evaluate whether Ultra probe melting temperatures can be used to identify heteroresistance, evaluate MDTBDRplus and describe FT-MTBDR outcomes using discordant samples.

Among Ultra rifampicin-resistant samples that had actionable MTBDRplus results, 8% (109/1347) were discordant. 79% of samples (79/100) were confirmed as rifampicin resistant based on the reference standard, supporting the Ultra result in the majority of samples. Almost one-half of the samples (11/24) that were not confirmed as rifampicin resistant were samples that were semi-quantitative graded as very low by Ultra.

For the detection of isoniazid resistance, MTBDRplus was specific (98%) but not sensitive (53%) when using discordant samples. It is important to keep in mind that this estimated accuracy should not be applied to usual samples as these discordant samples are unusual for heteroresistance, low bacillary burden or perhaps other characteristics.

Compared to Ultra detected rifampicin resistance, FT-MTBDR had a concordant finding in  55% of samples. Among the 45% of discordant results, SMOR demonstrated that 53% (21/40) were in fact rifampicin-resistant.

The authors list the following as key findings: 1) Most 222 discordance (79%) was from MTBDRplus not detecting rifampicin-resistance; 2) 69% of these MTBDRplus-susceptible were detected as FT-MTBDR resistant, indicating FT-MTBDR has higher sensitivity than MTBDRplus; 3) a substantial proportion with sequencing-detected resistance (31%; all of which were heteroresistant) were missed by FT-MTBDR and, in people with heteroresistance, multiple resistant strains were often present; 4) although the Ultra heteroresistance probe pattern was associated with heteroresistance, this pattern had suboptimal sensitivity and specificity for heteroresistance and 5) more than half of the individuals were rifampicin monoresistant, supporting the need for isoniazid DST. While these conclusions are supported by the data, it would be useful for the authors to discuss the utility of follow-up testing when rifampicin-resistance is detected by Ultra as this is, in my opinion, the most important implication of their findings.  As regards to rifampicin-resistance, it seems that MTBDRplus testing is not helpful; when there is a discordant finding, MTBDRplus is only correct in 21% of cases. While data on isoniazid resistance could be additional information, given that in these discordant samples the sensitivity (compared to SMOR) was only 53% it suggests at best a limited role for MTBDRplus testing of discordant samples. While FT-MTBDR had greater accuracy than MTBDRplus and correctly identified nearly all samples that were in fact rifampicin-susceptible, there were 21 of  77 samples (27%) that were identified as susceptible that were rifampicin-resistant by SMOR. The manuscript would benefit from a brief discussion of this finding and implications.