Author s response to reviews Title: Bacterial risk factors for treatment failure and relapse among patients with isoniazid resistant tuberculosis Authors: Phan Thai (phanvuongkhacthai@yahoo.com) Dang Ha (hadtm@oucru.org) Nguyen Hanh (annanguyenhanh@gmail.com) jeremy day (jday@oucru.org) sarah dunstan (sarah.dunstan@unimelb.edu.au) nguyen nhu (nhuntq@yahoo.com.vn) vo kiet (kietvs@oucru.org) Nguyan Lan (nguyenhuulan1965@yahoo.com.vn) ngueyn dung (bshuydung@yahoo.com) nguyen lan (ngoclan@hcm.fpt.vn) thuong thuong (thuongntt@oucru.org) nguyen Lan (ngoclan0456@yahoo.com) Pham Lieu (lieuphamthithuy@yahoo.com) nguyen thi hong (hongquang0701@yahoo.com) dao diep (daocongdiep@yahoo.com) Nguyen thanh (kimthanh0459@gmail.com) nguyen hoi (nguyenvanhoi99a@gmail.com) nguyen van nghia (bsnguyennghia@yahoo.com.vn) Truong dai (truongngocdai@gmail.com.vn) hoang minh (bsminhsg@yahoo.com.vn) nguyen thom (pklaoq8@gmail.com) jeremy farrar (jfarrar@oucru.org) Maxine Caws (mcaws@hotmail.com) Version: 1 Date: 13 Aug 2017
Author s response to reviews: BMC Infectious Diseases Response to reviewers. Reviewer 1. This is a good manuscript that adds to the growing literature on the poorer outcome of INHresistant TB, in this study 12.6% by sputum smear criteria during therapy, but an additional ~16% failed during 2 year follow-up using culture. We thank the reviewer for his comments on the value of the contribution of the manuscript to the literature on INH-resistant TB treatment outcomes. 1) How about prior treatment, extensiveness of disease, cavitation, site of disease, malnutrition - all the usual risk factors for poor outcome or relapse that we think about? All Patients recruited to the study were new TB patients (no prior history of disease). This stated on line 160-162. Regarding site of disease, patients were all pulmonary TB cases as stated on line 160-162. Patients were also all HIV-negative as stated on line 160-162. For clarity we have modified the relevant sentence in the methods section of the abstract (line 87-89): Between December 2008 and June 2011 2,090 consecutive HIV-negative adults ( 18 years of age) with new smear positive pulmonary TB presenting at participating district TB units in Ho Chi Minh City were recruited. We do not have data on extent of cavitation or malnutrition which we agree are all factors contributing to poor outcome for TB disease in general. 2) Should we test for Beijing in INH-R disease as a matter of routine and treat them differently? This study is not sufficient evidence to recommend testing for Beijing lineage as a routine measure for patients with INH-R disease but does suggest further investigation is warranted and genotyping should be included in randomized clinical trials for TB treatment to establish if all lineages respond well to any novel regimen. This observational cohort was treated with the 8-
month regimen, with an HE continuation phase which is no longer recommended by WHO and is being phased out by Vietnam, therefore it would not be appropriate to make routine treatment recommendations based on this data, however we believe the data is of value in the literature on INH-R TB. A consideration of lineage in patient management may be warranted if studies with current and novel regimens confirm a higher risk of failure/relapse for Beijing lineage INH-R strains. 3) Additional patients were detected with unfavorable outcome during follow-up - but this additional rate needs to be compared against INH-S controls to interpret it. This is mentioned as a limitation in the discussion but it is significant. If relapse rates in INH-S patients are also high, then one is mainly left with a high failure rate of INH-R during therapy and intensive monitoring during that phase would be targeted. By contrast if relapse of INH-R is particularly high, then perhaps longer durations would be targeted. This is discussed as the reviewer notes in the discussion section, lines 368 to 371. Although two year follow-up data on the INH-S cohort was not available due to funding limitations, the existing literature on relapse in INH-S TB evaluated by culture at 2 years shows that INH-S TB does not have these high rates of unfavourable outcome. We have added a relevant reference reporting the 30-month follow-up of the IUATLD landmark study A which evaluated the 8- month regimen. The rate of unfavourable outcome for fully susceptible TB treated with the 8- month regimen was 10% in this study. Nunn AJ, Jindani A, Enarson DA, Study Ai: Results at 30 months of a randomised trial of two 8-month regimens for the treatment of tuberculosis. Int J Tuberc Lung Dis 2011, 15(6):741-745. Regardless of the comparison with INH-S TB, the extremely high rate of unfavourable outcome in the INH-R cohort clearly shows that this 8-month regimen is unacceptable for the treatment of INH-R TB and that the majority of these failures occur after the treatment completion evaluation at 8-months and so would not be apparent under standard programmatic conditions. 4) It appears there was substantial discordance between MODS and MGIT. Would be helpful to describe this correlation in a paragraph, even if poor. It is understandable to consider MGIT as
the gold-standard for simplicity sake, but this method is not perfect and reporting of discordances is underrepresented in the literature which perpetuates poor gold-standards. The dramatic scale-up of programmatic rapid molecular testing for TB using the genexpert system has reduced the application of MODS testing as a diagnostic for TB. Evaluation of the discrepancy between MODS and MGIT testing is not the aim of this manuscript and would require a separate report conforming to STAARD guidelines to present in full. MODS was the initial rapid test used to identify INH-R patients for enrolment into the study cohort. Due to a change in the international recommendation for the critical concentration for INH_R MODS testing in the middle of the study, we used MGIT testing which is a well-validated reference standard, to confirm INH resistance in patients included for the INH-R analysis at completion of the study. We recognise this is not an ideal situation, however the retrospective MGIT testing was the most appropriate way to ensure consistent classification of the isolate drug susceptibility profile, in light of the issues with MODS classification which became apparent internationally during the study period. 5) Table 4 would be helpful to include the (n) for the risk factors on the left. Some risk factors may have been very rare, leading to imprecision with the OR estimates. We have added the (n) for risk factors for table 4. The 95% CI which is presented gives the precision of the OR estimates. Reviewer two. 1) The definition of "isoniazid-resistance" is very elusive. Isoniazid resistance is defined in the context of this study in the abstract, lines 92-93. Isoniazid resistance is considered any isoniazid resistant isolates that are susceptible to rifampicin. Resistance to isoniazid and rifampicin is classified as multi-drug resistant TB (MDR TB) in line with WHO definitions. The term isoniazid monoresistant is not applicable as the isolates classified as isoniazid resistant here may also be susceptible to streptomycin or ethambutol, or
both these drugs. The susceptiblilty patterns of isolates from patients in the cohort is shown in table one. We have added the definition of INH-R to the background (lines 147-148) and methods sections for clarity (lines 241 to 244). This definition is consistent with the literature on TB and WHO definitions. 2) What would be very useful is a clearly composed manuscript where MODS/MGIT is performed upon diagnosis and cases of isoniazid monoresistance are followed-up and compared to MDR and those without any resistance. Please reformat the manuscript to describe this story. We fully agree with the reviewer it would be ideal if MGIT had also been performed at diagnosis. However, this was not the original study design because MGIT susceptibility results take longer than MODS and we originally planned the study to ensure fast identification of INH- R for recruitment into the INH-R cohort. However, due to the change in international standard MODS critical concentration in the middle of the study, we performed MGIT DST retrospectively to standardize the catagorisation of drug susceptibility profile for all patients analysed in the study. It was obviously not possible to implement MGIT testing at diagnosis retrospectively when the problems with MODS testing for INH-R became apparent after international systematic review of available literature. This is described in methods (lines 211-218). See also response to reviewer one, point (4) above. 3) What would be very useful is a clearly composed manuscript where MODS/MGIT is performed upon diagnosis and cases of isoniazid monoresistance are followed-up and compared to MDR and those without any resistance. MODS was performed upon diagnosis. Isoniazid monoresistance is relatively rare in Vietnam and therefore analysis of an INH monoresistant cohort would be small. Isoniazid resistance usually occurs with streptomycin and/or ethambutol resistance in this setting. Therefore the
analysis of INH monoresistance is less relevant here than the analysis of INH-resistance without rifampicin resistance [isoniazid resistant isolates susceptible to rifampicin +/- resistance to streptomycin/ethambutol]. This is consistent with the literature on TB, where INH resistant TB is considered to be any resistant to INH which is rifampicin susceptible (ie. Not MDR TB). 4) For ease of reviewing please use continuous line numbers. Due to the line numbers restarting on each page I have simply referred to the page below. Line numbers have been changed to continuous. The revised continuous line numbers are referenced in this response. 5) No need to capitalise isoniazid in the title Capitalization of isoniazid in the title has been removed. 6) when was MODS performed? MODS was performed at diagnosis. This is stated in the methods lines 169-171. 7) What was the relationship between MGIT and MODS? Were the 239 people with I-R bacteria also detected via MODS?
As for reviewer one, point (4) above. MODS was performed at diagnosis to identify patients for recruitment to the INH-R cohort. Due to changes in the international standard definition these were later confirmed as INH-R by MGIT testing. The 239 people analysed in the INH-R cohort were therefore INH-R by both MODS and MGIT testing. 8) Are these purely isoniazid monoresistant? The classification of INH-R is addressed in reviewer two response (1) and (4) above. It is not INH monoresistance. 9) what does unfavourable mean? You should not need to delve into the manuscript to understand what criteria you are using. Definition of unfavourable outcome is given in methods section on statistical analysis (lines 237-240). This is a standard WHO definition used in the literature on assessment of TB treatment outcome and clinical trials for TB. We did not consider it necessary to define in the introduction but are happy to add if the editor considers it necessary. Favourable outcomes are the WHO programmatic outcome categories cure and treatment completion. Unfavourable outcomes are the WHO outcome categories death, and treatment failed. For further information see the WHO standard guidelines for the programmatic management of tuberculosis. http://www.who.int/tb/publications/9789241547833/en/ 10) As above - what is standard WHO monitoring This is also defined in the methods lines 203-204. This is standard WHO programmatic TB treatment evaluation used by all national TB Programmes using smear microscopy to assess
response to treatment. For the 8-month regimen, this is sputum smear evaluation at months 2,5 and 8. Again, we are happy to add to the introduction if requested by the editor but consider it more appropriate to the methods section. 11) Ensure WHO is spelled out correctly (organization) We have corrected spelling of WHO and thank the reviewer for noting the typo. 12) Ref 8 link is dead. We have updated the WHO link to TB treatment guidelines: http://www.who.int/tb/publications/9789241547833/en/ 13) Explain what standard WHO monitoring is As above for point (10). 14) what blood testing is done? This has been corrected to more detailed follow-up and testing. We thank the reviewer for noting the typo.
15) What is CRF? We have added the definition of CRF-( Case Report Form) and apologize for the omission. A CRF is used to collect patient information in clinical studies. 16) Does this means MODS was only performed at 5 and 8 months? When on treatment? If baseline MODS was not performed then I don't see what can be concluded from any of this study as any resistance is likely to have been de novo. MODS culture was performed at diagnosis as described in the recruitment section of the methods, lines 169-171. 17) What did the other 4.9% of patients receive? Why were these individualised regimens used? The remaining 4.9% of patients received individualized regimens as described in the subsequent paragraph (lines 257-260). This was determined by the treating clinician. This was an observational cohort study under programmatic conditions and not a treatment trial, therefore the study protocol did not dictate treatment which was determined by the treating clinician, usually in accordance with the Vietnamese NTP guidelines. 18) Again, when was MODS performed? Why was there no culture in such a significant proportion of supposedly smear-positive people? How were MODS -ve and indeterminate managed? Did INH-R and MDR patients receive appropriate treatment following MODS testing?
MODS was performed at diagnosis, as for point (6) above. This is stated in the methods lines 169-171. Culture is not routinely performed for TB patients in Vietnam. This is the case for most high burden countries. Culture is routinely only performed for retreatment patients. MODS was not an approved diagnostic test to determine treatment for the Vietnamese NTP at the time of the study and was used as a research tool here, to identify patients for recruitment to the INH-R cohort for intensified follow-up. Patients who were smear positive for TB were treated and managed according to the Vietnamese NTP guidelines and treatment decisions were made by the treating clinician. Follow-up of MDR cases was beyond the scope of this manuscript. 19) Lower down on the page the 324 MODS cases becomes 328. Which is correct? The number of INH-R MODS cases has been corrected to 324, apologies and we thank the reviewer for noting this error. 20) It is very confusing that MGIT DST is available on 1710 and it is not clear when this was performed, why MODS was performed, how one led to the other, etc. This has already been addressed in reviewer one point (4) and reviewer two point (2). The timing of MODS testing is given in the methods section 169-171. MGIT DST was retrospectively performed on the cultures isolated at diagnosis. A number of cultures failed to revive or were contaminated to subculture and therefore MGIT DST results were available on 1710 cases. 21) Results should actually be presented in the results section, not purely refer to the tables. We have described the key findings in the results section in the interests of manuscript length and referred to the tables for detailed presentation of all the statistical comparisons made. We are happy to extend this text if the editor feels it is necessary.
22) Where does the figure 457 come from? Describe in more detail the comparisons made for the mutation significance tests. Were promoter mutations, etc. present and in what numbers - all of these figures should be provided. How are INDO strains more likely to be wild type when supposedly the comparisons were being made among resistant strains - what then was the resistance mutation? The mutations at katg 315 and inha -15 were analysed for this study, the major mutations conferring ING resistance in Mtb. The occurrence of Mtb isolates phenotypically resistant to INH without any known mutations for INH resistance is well documented, hence wild-type INH-R isolates are well described in the literature. We have rephrased the sentence to clarify that the use of wild-type here refers to the sites tested (katg and inha). We have added the number mutated in katg and inha. 457 is the number of isolates resistant to INH by MGIT with a molecular genotype result available. As this analysis of lineage with molecular genotype did not include outcome, it contained a greater number of patients, not only those recruited into the INH-R intensified follow-up cohort. 23) Pg 16 - The figure 239 then appears. Where did this come from as it doesn't fit anywhere with what is shown in table 1. 239 patients were included in the INH-R cohort with intensified follow-up of outcomes and received the standardized regimens. This is stated at the start of the paragraph (lines 311-313): 239 patients in the extended INH study had confirmed INH resistant, RIF susceptible isolates by MGIT DST, received a standardised regimen and were therefore analysed for risk factors associated with unfavourable outcome. 24) Pg 17 - Don't quote 12.6% as a percentage of all INH-R cases, quote it as a percentage of the ones with unfavourable outcomes.
12.6% here is not a percentage of those with unfavourable outcomes as suggested by the reviewer, it is the % of all INH-R cases with unfavourable outcomes by WHO evaluation- the original sentence is correct. 25) Rather than persistence, the relationship of resistance with Beijing lineage is most likely a hypermutable state. Would be worth exploring this further as it is the only significant finding in this study and therefore needs to be placed within the context of other studies. We agree this finding warrants further exploration. While the hypermutable state has been proposed as a theory for the increased prevalence of resistance among Beijing lineage strains in some settings, the current evidence is weak and it remains one of several possible explanations for the characteristics attributed to the Beijing genotype. We disagree that there is sufficient evidence to say that the higher rates of unfavourable outcome documented here are attributable to a hyper-mutable state of the Beijing lineage strains, but we agree this requires further investigation to understand the underlying mechanisms responsible. The true explanation is likely multi-factorial and a combination of many characteristics which allow greater persistence and emergence of resistance of bacilli despite standard treatment. 26) Pg 18 - You note that katg has previously been associated with higher MICs than the promoter mutation but then have not shown this to not be the case in the results section Yes, the katg315 mutation has been associated with higher MIC in previous studies but was not in this study. 27) Pg 19 - The fact that it is so difficult to know the timeframe of when susceptibility testing is carried out and the (actually high) likelihood that much of this may be reinfection, I believe that this is a very major limitation of the study. MIRU-24 would go some way to help distinguish
whether this is the case or not. The "genotypes" in this study are extremely broad and not based on current gold standards. The timeframe of testing has been addressed in multiple points above. Final classification of resistance was based on the initial diagnostic culture, performed retrospectively by MGIT DST. The lineage classification used here is the current gold standard and consistent with the current understanding of Mtb phylogeny and is appropriate for this study. A more discrepant classification, such as MIRU-24 is appropriate to a transmission study but not here as this study was looking at the association of bacterial lineage with outcome. It is highly unlikely that these unfavourable outcomes are reinfection as this is a low HIV prevalence setting and the annual infection rate is too low. This is in contrast to high HIV prevalence settings in sub-saharan Africa. If reinfection occurred from the original, presumably untreated, index patient within two years, MIRU-24 analysis would not detect this as a reinfection because the genotype would be identical. 28) Table 2: Is INH Resistant TB monoresistant? INH monoresistant TB is Mtb that is only resistant to isoniazid and susceptible to all other TB drugs. This is a standard definition. INH resistant TB in table 2 is not INH monoresistant TB, it is TB resistant to INH. 29) Table 2: What is going on in the District 1 cell that just contains 0.? This has been corrected to [0.68-2.53], we thank the reviewer for noticing this omission of this 95% CI.
30) Table 3: What comparisons are actually being tested in this table The footnote to table 3 (lines 417-418) explains the comparison made. Each lineage is compared with all other isolates to examine the association between lineage and drug resistance. 31) Table 4: Why adjust for resistance to streptomycin and ethambutol? Adjustment has been made for strep and ethambutol resistance to these additional drugs may confound the analysis.